{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.011776, "eval_steps": 500, "global_step": 387500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 2.56e-06, "grad_norm": 7.336174488067627, "learning_rate": 0.0, "loss": 12.0804, "step": 1 }, { "epoch": 2.56e-05, "grad_norm": 10.891112327575684, "learning_rate": 3.6e-07, "loss": 12.0327, "step": 10 }, { "epoch": 5.12e-05, "grad_norm": 12.334525108337402, "learning_rate": 7.6e-07, "loss": 11.9695, "step": 20 }, { "epoch": 7.68e-05, "grad_norm": 14.853639602661133, "learning_rate": 1.1600000000000001e-06, "loss": 11.8623, "step": 30 }, { "epoch": 0.0001024, "grad_norm": 11.221489906311035, "learning_rate": 1.56e-06, "loss": 11.5693, "step": 40 }, { "epoch": 0.000128, "grad_norm": 12.630057334899902, "learning_rate": 1.9600000000000003e-06, "loss": 11.4733, "step": 50 }, { "epoch": 0.0001536, "grad_norm": 16.161535263061523, "learning_rate": 2.3600000000000003e-06, "loss": 10.7893, "step": 60 }, { "epoch": 0.0001792, "grad_norm": 7.414032459259033, "learning_rate": 2.7600000000000003e-06, "loss": 10.6538, "step": 70 }, { "epoch": 0.0002048, "grad_norm": 8.312623977661133, "learning_rate": 3.1600000000000002e-06, "loss": 10.1741, "step": 80 }, { "epoch": 0.0002304, "grad_norm": 2.9599626064300537, "learning_rate": 3.5600000000000002e-06, "loss": 9.5481, "step": 90 }, { "epoch": 0.000256, "grad_norm": 4.948571681976318, "learning_rate": 3.96e-06, "loss": 9.6164, "step": 100 }, { "epoch": 0.0002816, "grad_norm": 3.8786556720733643, "learning_rate": 4.360000000000001e-06, "loss": 9.1946, "step": 110 }, { "epoch": 0.0003072, "grad_norm": 3.805713653564453, "learning_rate": 4.76e-06, "loss": 9.0166, "step": 120 }, { "epoch": 0.0003328, "grad_norm": 3.35520076751709, "learning_rate": 5.1600000000000006e-06, "loss": 8.8522, "step": 130 }, { "epoch": 0.0003584, "grad_norm": 2.514050245285034, "learning_rate": 5.560000000000001e-06, "loss": 8.7168, "step": 140 }, { "epoch": 0.000384, "grad_norm": 2.411886215209961, "learning_rate": 5.9600000000000005e-06, "loss": 8.6696, "step": 150 }, { "epoch": 0.0004096, "grad_norm": 3.288315773010254, "learning_rate": 6.360000000000001e-06, "loss": 8.7924, "step": 160 }, { "epoch": 0.0004352, "grad_norm": 2.572204113006592, "learning_rate": 6.760000000000001e-06, "loss": 8.4052, "step": 170 }, { "epoch": 0.0004608, "grad_norm": 2.493095874786377, "learning_rate": 7.16e-06, "loss": 8.5792, "step": 180 }, { "epoch": 0.0004864, "grad_norm": 5.325606346130371, "learning_rate": 7.5600000000000005e-06, "loss": 8.0018, "step": 190 }, { "epoch": 0.000512, "grad_norm": 4.776216506958008, "learning_rate": 7.960000000000002e-06, "loss": 8.2712, "step": 200 }, { "epoch": 0.0005376, "grad_norm": 11.784114837646484, "learning_rate": 8.36e-06, "loss": 8.185, "step": 210 }, { "epoch": 0.0005632, "grad_norm": 2.567786455154419, "learning_rate": 8.76e-06, "loss": 7.4673, "step": 220 }, { "epoch": 0.0005888, "grad_norm": 3.3478000164031982, "learning_rate": 9.16e-06, "loss": 7.777, "step": 230 }, { "epoch": 0.0006144, "grad_norm": 1.7570945024490356, "learning_rate": 9.56e-06, "loss": 7.6437, "step": 240 }, { "epoch": 0.00064, "grad_norm": 2.358156442642212, "learning_rate": 9.960000000000001e-06, "loss": 7.5815, "step": 250 }, { "epoch": 0.0006656, "grad_norm": 2.986248731613159, "learning_rate": 1.036e-05, "loss": 7.5875, "step": 260 }, { "epoch": 0.0006912, "grad_norm": 2.4573705196380615, "learning_rate": 1.0760000000000002e-05, "loss": 7.6307, "step": 270 }, { "epoch": 0.0007168, "grad_norm": 1.936881422996521, "learning_rate": 1.1160000000000002e-05, "loss": 7.446, "step": 280 }, { "epoch": 0.0007424, "grad_norm": 2.2101516723632812, "learning_rate": 1.156e-05, "loss": 7.3573, "step": 290 }, { "epoch": 0.000768, "grad_norm": 2.6309401988983154, "learning_rate": 1.196e-05, "loss": 6.9877, "step": 300 }, { "epoch": 0.0007936, "grad_norm": 2.1232411861419678, "learning_rate": 1.236e-05, "loss": 7.5944, "step": 310 }, { "epoch": 0.0008192, "grad_norm": 1.6259442567825317, "learning_rate": 1.2760000000000001e-05, "loss": 7.141, "step": 320 }, { "epoch": 0.0008448, "grad_norm": 2.6265251636505127, "learning_rate": 1.3160000000000001e-05, "loss": 6.5941, "step": 330 }, { "epoch": 0.0008704, "grad_norm": 1.523422122001648, "learning_rate": 1.3560000000000002e-05, "loss": 6.8383, "step": 340 }, { "epoch": 0.000896, "grad_norm": 1.9008511304855347, "learning_rate": 1.396e-05, "loss": 7.053, "step": 350 }, { "epoch": 0.0009216, "grad_norm": 2.304110050201416, "learning_rate": 1.4360000000000001e-05, "loss": 6.7334, "step": 360 }, { "epoch": 0.0009472, "grad_norm": 2.462344169616699, "learning_rate": 1.4760000000000001e-05, "loss": 6.9322, "step": 370 }, { "epoch": 0.0009728, "grad_norm": 3.7496135234832764, "learning_rate": 1.516e-05, "loss": 6.7633, "step": 380 }, { "epoch": 0.0009984, "grad_norm": 2.0797691345214844, "learning_rate": 1.556e-05, "loss": 6.9415, "step": 390 }, { "epoch": 0.001024, "grad_norm": 2.621248245239258, "learning_rate": 1.5960000000000003e-05, "loss": 6.7713, "step": 400 }, { "epoch": 0.0010496, "grad_norm": 1.6956846714019775, "learning_rate": 1.636e-05, "loss": 6.2648, "step": 410 }, { "epoch": 0.0010752, "grad_norm": 1.7431864738464355, "learning_rate": 1.6760000000000002e-05, "loss": 6.1968, "step": 420 }, { "epoch": 0.0011008, "grad_norm": 2.1460072994232178, "learning_rate": 1.7160000000000002e-05, "loss": 6.7863, "step": 430 }, { "epoch": 0.0011264, "grad_norm": 1.2538201808929443, "learning_rate": 1.756e-05, "loss": 6.3821, "step": 440 }, { "epoch": 0.001152, "grad_norm": 1.8675072193145752, "learning_rate": 1.796e-05, "loss": 6.3693, "step": 450 }, { "epoch": 0.0011776, "grad_norm": 2.3658409118652344, "learning_rate": 1.8360000000000004e-05, "loss": 6.3214, "step": 460 }, { "epoch": 0.0012032, "grad_norm": 2.3485188484191895, "learning_rate": 1.876e-05, "loss": 6.6553, "step": 470 }, { "epoch": 0.0012288, "grad_norm": 2.5190188884735107, "learning_rate": 1.916e-05, "loss": 6.1138, "step": 480 }, { "epoch": 0.0012544, "grad_norm": 2.0157082080841064, "learning_rate": 1.9560000000000002e-05, "loss": 6.7692, "step": 490 }, { "epoch": 0.00128, "grad_norm": 3.1594302654266357, "learning_rate": 1.9960000000000002e-05, "loss": 6.2979, "step": 500 }, { "epoch": 0.0013056, "grad_norm": 1.7643921375274658, "learning_rate": 1.9999999973736827e-05, "loss": 6.6535, "step": 510 }, { "epoch": 0.0013312, "grad_norm": 2.412630796432495, "learning_rate": 1.9999999882950548e-05, "loss": 6.4441, "step": 520 }, { "epoch": 0.0013568, "grad_norm": 1.5962061882019043, "learning_rate": 1.9999999727316928e-05, "loss": 6.5944, "step": 530 }, { "epoch": 0.0013824, "grad_norm": 2.3681535720825195, "learning_rate": 1.9999999506835966e-05, "loss": 6.4439, "step": 540 }, { "epoch": 0.001408, "grad_norm": 2.09466552734375, "learning_rate": 1.9999999221507666e-05, "loss": 6.5572, "step": 550 }, { "epoch": 0.0014336, "grad_norm": 1.6325868368148804, "learning_rate": 1.9999998871332028e-05, "loss": 6.397, "step": 560 }, { "epoch": 0.0014592, "grad_norm": 1.886602759361267, "learning_rate": 1.999999845630906e-05, "loss": 6.5305, "step": 570 }, { "epoch": 0.0014848, "grad_norm": 1.9617581367492676, "learning_rate": 1.999999797643875e-05, "loss": 6.6598, "step": 580 }, { "epoch": 0.0015104, "grad_norm": 1.7467939853668213, "learning_rate": 1.999999743172112e-05, "loss": 6.2774, "step": 590 }, { "epoch": 0.001536, "grad_norm": 1.594149112701416, "learning_rate": 1.999999682215616e-05, "loss": 6.1308, "step": 600 }, { "epoch": 0.0015616, "grad_norm": 1.3802576065063477, "learning_rate": 1.9999996147743875e-05, "loss": 6.5974, "step": 610 }, { "epoch": 0.0015872, "grad_norm": 1.8881837129592896, "learning_rate": 1.999999540848428e-05, "loss": 6.5475, "step": 620 }, { "epoch": 0.0016128, "grad_norm": 3.050750732421875, "learning_rate": 1.9999994604377366e-05, "loss": 6.6901, "step": 630 }, { "epoch": 0.0016384, "grad_norm": 2.124467134475708, "learning_rate": 1.999999373542315e-05, "loss": 6.3138, "step": 640 }, { "epoch": 0.001664, "grad_norm": 1.666005253791809, "learning_rate": 1.9999992801621627e-05, "loss": 6.3876, "step": 650 }, { "epoch": 0.0016896, "grad_norm": 2.954392910003662, "learning_rate": 1.999999180297281e-05, "loss": 6.6558, "step": 660 }, { "epoch": 0.0017152, "grad_norm": 3.0646634101867676, "learning_rate": 1.9999990739476708e-05, "loss": 6.6547, "step": 670 }, { "epoch": 0.0017408, "grad_norm": 1.276208519935608, "learning_rate": 1.9999989611133318e-05, "loss": 6.5955, "step": 680 }, { "epoch": 0.0017664, "grad_norm": 1.3130295276641846, "learning_rate": 1.9999988417942657e-05, "loss": 6.385, "step": 690 }, { "epoch": 0.001792, "grad_norm": 8.404816627502441, "learning_rate": 1.9999987159904726e-05, "loss": 6.5981, "step": 700 }, { "epoch": 0.0018176, "grad_norm": 1.5832183361053467, "learning_rate": 1.9999985837019538e-05, "loss": 6.461, "step": 710 }, { "epoch": 0.0018432, "grad_norm": 2.497919797897339, "learning_rate": 1.9999984449287096e-05, "loss": 6.1184, "step": 720 }, { "epoch": 0.0018688, "grad_norm": 2.4530482292175293, "learning_rate": 1.999998299670742e-05, "loss": 6.1912, "step": 730 }, { "epoch": 0.0018944, "grad_norm": 2.3070473670959473, "learning_rate": 1.9999981479280504e-05, "loss": 6.2376, "step": 740 }, { "epoch": 0.00192, "grad_norm": 1.8637981414794922, "learning_rate": 1.999997989700637e-05, "loss": 5.9341, "step": 750 }, { "epoch": 0.0019456, "grad_norm": 1.9861186742782593, "learning_rate": 1.9999978249885024e-05, "loss": 6.19, "step": 760 }, { "epoch": 0.0019712, "grad_norm": 2.07912540435791, "learning_rate": 1.9999976537916475e-05, "loss": 6.081, "step": 770 }, { "epoch": 0.0019968, "grad_norm": 1.3683607578277588, "learning_rate": 1.9999974761100737e-05, "loss": 6.3569, "step": 780 }, { "epoch": 0.0020224, "grad_norm": 1.966201901435852, "learning_rate": 1.9999972919437823e-05, "loss": 6.1415, "step": 790 }, { "epoch": 0.002048, "grad_norm": 1.1430785655975342, "learning_rate": 1.9999971012927738e-05, "loss": 5.7843, "step": 800 }, { "epoch": 0.0020736, "grad_norm": 2.7832627296447754, "learning_rate": 1.99999690415705e-05, "loss": 6.3354, "step": 810 }, { "epoch": 0.0020992, "grad_norm": 1.5076984167099, "learning_rate": 1.9999967005366118e-05, "loss": 6.1938, "step": 820 }, { "epoch": 0.0021248, "grad_norm": 1.3043354749679565, "learning_rate": 1.9999964904314614e-05, "loss": 6.0403, "step": 830 }, { "epoch": 0.0021504, "grad_norm": 1.2553290128707886, "learning_rate": 1.999996273841599e-05, "loss": 6.1904, "step": 840 }, { "epoch": 0.002176, "grad_norm": 1.3072381019592285, "learning_rate": 1.9999960507670266e-05, "loss": 5.9433, "step": 850 }, { "epoch": 0.0022016, "grad_norm": 2.2758209705352783, "learning_rate": 1.9999958212077456e-05, "loss": 6.0169, "step": 860 }, { "epoch": 0.0022272, "grad_norm": 1.7772406339645386, "learning_rate": 1.9999955851637576e-05, "loss": 6.0897, "step": 870 }, { "epoch": 0.0022528, "grad_norm": 1.1076576709747314, "learning_rate": 1.999995342635064e-05, "loss": 6.1365, "step": 880 }, { "epoch": 0.0022784, "grad_norm": 1.3440334796905518, "learning_rate": 1.999995093621666e-05, "loss": 6.0494, "step": 890 }, { "epoch": 0.002304, "grad_norm": 1.070548415184021, "learning_rate": 1.9999948381235658e-05, "loss": 5.9456, "step": 900 }, { "epoch": 0.0023296, "grad_norm": 1.193376064300537, "learning_rate": 1.999994576140765e-05, "loss": 5.9418, "step": 910 }, { "epoch": 0.0023552, "grad_norm": 1.8697278499603271, "learning_rate": 1.9999943076732653e-05, "loss": 6.3103, "step": 920 }, { "epoch": 0.0023808, "grad_norm": 1.1766046285629272, "learning_rate": 1.9999940327210678e-05, "loss": 5.9778, "step": 930 }, { "epoch": 0.0024064, "grad_norm": 1.5622739791870117, "learning_rate": 1.999993751284175e-05, "loss": 6.0242, "step": 940 }, { "epoch": 0.002432, "grad_norm": 1.2968382835388184, "learning_rate": 1.9999934633625887e-05, "loss": 6.0547, "step": 950 }, { "epoch": 0.0024576, "grad_norm": 1.8076188564300537, "learning_rate": 1.99999316895631e-05, "loss": 5.7021, "step": 960 }, { "epoch": 0.0024832, "grad_norm": 1.9092563390731812, "learning_rate": 1.9999928680653418e-05, "loss": 5.8899, "step": 970 }, { "epoch": 0.0025088, "grad_norm": 0.955628514289856, "learning_rate": 1.9999925606896857e-05, "loss": 6.0355, "step": 980 }, { "epoch": 0.0025344, "grad_norm": 1.1215755939483643, "learning_rate": 1.9999922468293432e-05, "loss": 5.8085, "step": 990 }, { "epoch": 0.00256, "grad_norm": 1.6272351741790771, "learning_rate": 1.999991926484317e-05, "loss": 5.8684, "step": 1000 }, { "epoch": 0.0025856, "grad_norm": 1.720639944076538, "learning_rate": 1.999991599654609e-05, "loss": 5.8023, "step": 1010 }, { "epoch": 0.0026112, "grad_norm": 1.4074501991271973, "learning_rate": 1.9999912663402212e-05, "loss": 6.045, "step": 1020 }, { "epoch": 0.0026368, "grad_norm": 1.4439060688018799, "learning_rate": 1.9999909265411556e-05, "loss": 5.9941, "step": 1030 }, { "epoch": 0.0026624, "grad_norm": 1.5217341184616089, "learning_rate": 1.9999905802574148e-05, "loss": 5.9146, "step": 1040 }, { "epoch": 0.002688, "grad_norm": 1.0741924047470093, "learning_rate": 1.9999902274890008e-05, "loss": 5.7828, "step": 1050 }, { "epoch": 0.0027136, "grad_norm": 1.4397602081298828, "learning_rate": 1.999989868235916e-05, "loss": 6.1489, "step": 1060 }, { "epoch": 0.0027392, "grad_norm": 1.7903014421463013, "learning_rate": 1.9999895024981625e-05, "loss": 5.8572, "step": 1070 }, { "epoch": 0.0027648, "grad_norm": 1.5168747901916504, "learning_rate": 1.999989130275743e-05, "loss": 5.7568, "step": 1080 }, { "epoch": 0.0027904, "grad_norm": 1.0270602703094482, "learning_rate": 1.9999887515686597e-05, "loss": 5.8548, "step": 1090 }, { "epoch": 0.002816, "grad_norm": 2.744337320327759, "learning_rate": 1.9999883663769154e-05, "loss": 5.9099, "step": 1100 }, { "epoch": 0.0028416, "grad_norm": 1.0366538763046265, "learning_rate": 1.999987974700512e-05, "loss": 5.6983, "step": 1110 }, { "epoch": 0.0028672, "grad_norm": 1.7717337608337402, "learning_rate": 1.9999875765394522e-05, "loss": 5.477, "step": 1120 }, { "epoch": 0.0028928, "grad_norm": 1.077410101890564, "learning_rate": 1.999987171893739e-05, "loss": 5.7727, "step": 1130 }, { "epoch": 0.0029184, "grad_norm": 1.2565157413482666, "learning_rate": 1.9999867607633746e-05, "loss": 5.9747, "step": 1140 }, { "epoch": 0.002944, "grad_norm": 1.4370874166488647, "learning_rate": 1.9999863431483618e-05, "loss": 5.7386, "step": 1150 }, { "epoch": 0.0029696, "grad_norm": 1.41145658493042, "learning_rate": 1.9999859190487036e-05, "loss": 5.8224, "step": 1160 }, { "epoch": 0.0029952, "grad_norm": 1.4419355392456055, "learning_rate": 1.999985488464402e-05, "loss": 5.5307, "step": 1170 }, { "epoch": 0.0030208, "grad_norm": 1.7412972450256348, "learning_rate": 1.9999850513954606e-05, "loss": 5.9301, "step": 1180 }, { "epoch": 0.0030464, "grad_norm": 0.828038215637207, "learning_rate": 1.9999846078418822e-05, "loss": 6.0618, "step": 1190 }, { "epoch": 0.003072, "grad_norm": 1.1441972255706787, "learning_rate": 1.999984157803669e-05, "loss": 5.9062, "step": 1200 }, { "epoch": 0.0030976, "grad_norm": 1.0216693878173828, "learning_rate": 1.9999837012808242e-05, "loss": 5.8025, "step": 1210 }, { "epoch": 0.0031232, "grad_norm": 1.2097969055175781, "learning_rate": 1.999983238273351e-05, "loss": 5.62, "step": 1220 }, { "epoch": 0.0031488, "grad_norm": 0.9571455121040344, "learning_rate": 1.9999827687812524e-05, "loss": 5.8942, "step": 1230 }, { "epoch": 0.0031744, "grad_norm": 1.268023133277893, "learning_rate": 1.999982292804531e-05, "loss": 5.6707, "step": 1240 }, { "epoch": 0.0032, "grad_norm": 1.5482635498046875, "learning_rate": 1.9999818103431905e-05, "loss": 5.8222, "step": 1250 }, { "epoch": 0.0032256, "grad_norm": 1.55221688747406, "learning_rate": 1.9999813213972337e-05, "loss": 5.5191, "step": 1260 }, { "epoch": 0.0032512, "grad_norm": 2.706651210784912, "learning_rate": 1.9999808259666637e-05, "loss": 5.7871, "step": 1270 }, { "epoch": 0.0032768, "grad_norm": 1.4277355670928955, "learning_rate": 1.999980324051484e-05, "loss": 5.5517, "step": 1280 }, { "epoch": 0.0033024, "grad_norm": 0.9675559997558594, "learning_rate": 1.9999798156516973e-05, "loss": 5.7751, "step": 1290 }, { "epoch": 0.003328, "grad_norm": 1.102674126625061, "learning_rate": 1.9999793007673074e-05, "loss": 5.6572, "step": 1300 }, { "epoch": 0.0033536, "grad_norm": 1.2172574996948242, "learning_rate": 1.9999787793983178e-05, "loss": 5.5903, "step": 1310 }, { "epoch": 0.0033792, "grad_norm": 0.8666584491729736, "learning_rate": 1.999978251544731e-05, "loss": 5.5911, "step": 1320 }, { "epoch": 0.0034048, "grad_norm": 1.182845115661621, "learning_rate": 1.9999777172065512e-05, "loss": 5.6375, "step": 1330 }, { "epoch": 0.0034304, "grad_norm": 1.308653473854065, "learning_rate": 1.999977176383782e-05, "loss": 5.4098, "step": 1340 }, { "epoch": 0.003456, "grad_norm": 1.3621068000793457, "learning_rate": 1.999976629076426e-05, "loss": 5.806, "step": 1350 }, { "epoch": 0.0034816, "grad_norm": 1.0402072668075562, "learning_rate": 1.9999760752844878e-05, "loss": 5.6501, "step": 1360 }, { "epoch": 0.0035072, "grad_norm": 1.4109150171279907, "learning_rate": 1.9999755150079705e-05, "loss": 6.0396, "step": 1370 }, { "epoch": 0.0035328, "grad_norm": 0.8739264607429504, "learning_rate": 1.9999749482468776e-05, "loss": 5.448, "step": 1380 }, { "epoch": 0.0035584, "grad_norm": 1.5565178394317627, "learning_rate": 1.999974375001213e-05, "loss": 5.4235, "step": 1390 }, { "epoch": 0.003584, "grad_norm": 1.7832409143447876, "learning_rate": 1.99997379527098e-05, "loss": 5.2964, "step": 1400 }, { "epoch": 0.0036096, "grad_norm": 2.273591995239258, "learning_rate": 1.999973209056183e-05, "loss": 5.8723, "step": 1410 }, { "epoch": 0.0036352, "grad_norm": 1.610538125038147, "learning_rate": 1.9999726163568258e-05, "loss": 5.7438, "step": 1420 }, { "epoch": 0.0036608, "grad_norm": 1.661357045173645, "learning_rate": 1.9999720171729116e-05, "loss": 5.5721, "step": 1430 }, { "epoch": 0.0036864, "grad_norm": 1.402848482131958, "learning_rate": 1.9999714115044448e-05, "loss": 5.8535, "step": 1440 }, { "epoch": 0.003712, "grad_norm": 1.4308555126190186, "learning_rate": 1.999970799351429e-05, "loss": 5.4384, "step": 1450 }, { "epoch": 0.0037376, "grad_norm": 0.9641668200492859, "learning_rate": 1.9999701807138682e-05, "loss": 5.5599, "step": 1460 }, { "epoch": 0.0037632, "grad_norm": 1.1585301160812378, "learning_rate": 1.9999695555917668e-05, "loss": 5.6747, "step": 1470 }, { "epoch": 0.0037888, "grad_norm": 1.942600965499878, "learning_rate": 1.9999689239851285e-05, "loss": 5.6015, "step": 1480 }, { "epoch": 0.0038144, "grad_norm": 1.2827640771865845, "learning_rate": 1.9999682858939573e-05, "loss": 5.5687, "step": 1490 }, { "epoch": 0.00384, "grad_norm": 1.1010409593582153, "learning_rate": 1.9999676413182575e-05, "loss": 5.6572, "step": 1500 }, { "epoch": 0.0038656, "grad_norm": 1.5150930881500244, "learning_rate": 1.9999669902580337e-05, "loss": 5.2422, "step": 1510 }, { "epoch": 0.0038912, "grad_norm": 1.1007570028305054, "learning_rate": 1.9999663327132896e-05, "loss": 5.3456, "step": 1520 }, { "epoch": 0.0039168, "grad_norm": 1.3473690748214722, "learning_rate": 1.9999656686840295e-05, "loss": 5.6155, "step": 1530 }, { "epoch": 0.0039424, "grad_norm": 1.8166215419769287, "learning_rate": 1.9999649981702575e-05, "loss": 5.5613, "step": 1540 }, { "epoch": 0.003968, "grad_norm": 0.8992148041725159, "learning_rate": 1.9999643211719787e-05, "loss": 5.5823, "step": 1550 }, { "epoch": 0.0039936, "grad_norm": 1.0961189270019531, "learning_rate": 1.9999636376891968e-05, "loss": 5.719, "step": 1560 }, { "epoch": 0.0040192, "grad_norm": 10.277132034301758, "learning_rate": 1.9999629477219166e-05, "loss": 5.3953, "step": 1570 }, { "epoch": 0.0040448, "grad_norm": 1.6544764041900635, "learning_rate": 1.9999622512701427e-05, "loss": 5.8668, "step": 1580 }, { "epoch": 0.0040704, "grad_norm": 1.422712802886963, "learning_rate": 1.9999615483338787e-05, "loss": 5.4772, "step": 1590 }, { "epoch": 0.004096, "grad_norm": 0.9532713294029236, "learning_rate": 1.9999608389131303e-05, "loss": 5.5024, "step": 1600 }, { "epoch": 0.0041216, "grad_norm": 1.172243595123291, "learning_rate": 1.9999601230079014e-05, "loss": 5.5249, "step": 1610 }, { "epoch": 0.0041472, "grad_norm": 1.061806321144104, "learning_rate": 1.999959400618197e-05, "loss": 5.5022, "step": 1620 }, { "epoch": 0.0041728, "grad_norm": 1.0643103122711182, "learning_rate": 1.999958671744022e-05, "loss": 5.4151, "step": 1630 }, { "epoch": 0.0041984, "grad_norm": 1.6815601587295532, "learning_rate": 1.99995793638538e-05, "loss": 5.6107, "step": 1640 }, { "epoch": 0.004224, "grad_norm": 1.4263744354248047, "learning_rate": 1.9999571945422768e-05, "loss": 5.277, "step": 1650 }, { "epoch": 0.0042496, "grad_norm": 1.160820722579956, "learning_rate": 1.9999564462147172e-05, "loss": 5.2309, "step": 1660 }, { "epoch": 0.0042752, "grad_norm": 2.2263715267181396, "learning_rate": 1.9999556914027057e-05, "loss": 5.4689, "step": 1670 }, { "epoch": 0.0043008, "grad_norm": 0.9493284821510315, "learning_rate": 1.999954930106247e-05, "loss": 5.6167, "step": 1680 }, { "epoch": 0.0043264, "grad_norm": 1.1031053066253662, "learning_rate": 1.999954162325347e-05, "loss": 5.2641, "step": 1690 }, { "epoch": 0.004352, "grad_norm": 1.1815071105957031, "learning_rate": 1.9999533880600095e-05, "loss": 5.4139, "step": 1700 }, { "epoch": 0.0043776, "grad_norm": 1.1066168546676636, "learning_rate": 1.99995260731024e-05, "loss": 5.5856, "step": 1710 }, { "epoch": 0.0044032, "grad_norm": 1.4823415279388428, "learning_rate": 1.999951820076044e-05, "loss": 5.2031, "step": 1720 }, { "epoch": 0.0044288, "grad_norm": 0.9160330891609192, "learning_rate": 1.999951026357426e-05, "loss": 5.0836, "step": 1730 }, { "epoch": 0.0044544, "grad_norm": 1.2306625843048096, "learning_rate": 1.999950226154391e-05, "loss": 5.1821, "step": 1740 }, { "epoch": 0.00448, "grad_norm": 1.0849703550338745, "learning_rate": 1.999949419466945e-05, "loss": 5.2386, "step": 1750 }, { "epoch": 0.0045056, "grad_norm": 1.1892255544662476, "learning_rate": 1.9999486062950925e-05, "loss": 5.44, "step": 1760 }, { "epoch": 0.0045312, "grad_norm": 1.2639262676239014, "learning_rate": 1.9999477866388393e-05, "loss": 5.293, "step": 1770 }, { "epoch": 0.0045568, "grad_norm": 1.5859870910644531, "learning_rate": 1.99994696049819e-05, "loss": 5.5419, "step": 1780 }, { "epoch": 0.0045824, "grad_norm": 0.8624865412712097, "learning_rate": 1.999946127873151e-05, "loss": 5.338, "step": 1790 }, { "epoch": 0.004608, "grad_norm": 1.5570334196090698, "learning_rate": 1.9999452887637267e-05, "loss": 5.2724, "step": 1800 }, { "epoch": 0.0046336, "grad_norm": 1.4420344829559326, "learning_rate": 1.999944443169923e-05, "loss": 5.3453, "step": 1810 }, { "epoch": 0.0046592, "grad_norm": 1.4619839191436768, "learning_rate": 1.9999435910917456e-05, "loss": 5.4057, "step": 1820 }, { "epoch": 0.0046848, "grad_norm": 1.3439611196517944, "learning_rate": 1.9999427325291996e-05, "loss": 5.4885, "step": 1830 }, { "epoch": 0.0047104, "grad_norm": 1.2357532978057861, "learning_rate": 1.9999418674822905e-05, "loss": 5.2183, "step": 1840 }, { "epoch": 0.004736, "grad_norm": 1.319114089012146, "learning_rate": 1.9999409959510244e-05, "loss": 5.4054, "step": 1850 }, { "epoch": 0.0047616, "grad_norm": 1.0827070474624634, "learning_rate": 1.9999401179354066e-05, "loss": 5.1188, "step": 1860 }, { "epoch": 0.0047872, "grad_norm": 1.2646232843399048, "learning_rate": 1.999939233435443e-05, "loss": 5.2184, "step": 1870 }, { "epoch": 0.0048128, "grad_norm": 1.4798544645309448, "learning_rate": 1.9999383424511392e-05, "loss": 5.3913, "step": 1880 }, { "epoch": 0.0048384, "grad_norm": 1.1672282218933105, "learning_rate": 1.9999374449825008e-05, "loss": 5.2034, "step": 1890 }, { "epoch": 0.004864, "grad_norm": 1.4302424192428589, "learning_rate": 1.999936541029534e-05, "loss": 5.2952, "step": 1900 }, { "epoch": 0.0048896, "grad_norm": 1.2623285055160522, "learning_rate": 1.9999356305922446e-05, "loss": 5.2645, "step": 1910 }, { "epoch": 0.0049152, "grad_norm": 1.2618943452835083, "learning_rate": 1.9999347136706383e-05, "loss": 5.1107, "step": 1920 }, { "epoch": 0.0049408, "grad_norm": 1.0024371147155762, "learning_rate": 1.999933790264721e-05, "loss": 5.6697, "step": 1930 }, { "epoch": 0.0049664, "grad_norm": 1.4545331001281738, "learning_rate": 1.9999328603744993e-05, "loss": 5.5134, "step": 1940 }, { "epoch": 0.004992, "grad_norm": 1.2373703718185425, "learning_rate": 1.999931923999978e-05, "loss": 5.4624, "step": 1950 }, { "epoch": 0.0050176, "grad_norm": 2.0064969062805176, "learning_rate": 1.9999309811411646e-05, "loss": 5.1221, "step": 1960 }, { "epoch": 0.0050432, "grad_norm": 1.031869649887085, "learning_rate": 1.9999300317980637e-05, "loss": 5.2469, "step": 1970 }, { "epoch": 0.0050688, "grad_norm": 1.0486795902252197, "learning_rate": 1.999929075970683e-05, "loss": 5.434, "step": 1980 }, { "epoch": 0.0050944, "grad_norm": 1.9943357706069946, "learning_rate": 1.9999281136590278e-05, "loss": 4.9929, "step": 1990 }, { "epoch": 0.00512, "grad_norm": 0.9484997987747192, "learning_rate": 1.999927144863105e-05, "loss": 5.0785, "step": 2000 }, { "epoch": 0.0051456, "grad_norm": 3.7481136322021484, "learning_rate": 1.9999261695829198e-05, "loss": 5.3983, "step": 2010 }, { "epoch": 0.0051712, "grad_norm": 1.1690911054611206, "learning_rate": 1.9999251878184792e-05, "loss": 5.6057, "step": 2020 }, { "epoch": 0.0051968, "grad_norm": 1.0068960189819336, "learning_rate": 1.9999241995697894e-05, "loss": 5.1813, "step": 2030 }, { "epoch": 0.0052224, "grad_norm": 1.0808910131454468, "learning_rate": 1.9999232048368574e-05, "loss": 5.3849, "step": 2040 }, { "epoch": 0.005248, "grad_norm": 0.8831846714019775, "learning_rate": 1.9999222036196886e-05, "loss": 5.3101, "step": 2050 }, { "epoch": 0.0052736, "grad_norm": 0.9496194124221802, "learning_rate": 1.9999211959182905e-05, "loss": 5.1761, "step": 2060 }, { "epoch": 0.0052992, "grad_norm": 4.495302200317383, "learning_rate": 1.999920181732669e-05, "loss": 5.2384, "step": 2070 }, { "epoch": 0.0053248, "grad_norm": 1.6834667921066284, "learning_rate": 1.999919161062831e-05, "loss": 5.2951, "step": 2080 }, { "epoch": 0.0053504, "grad_norm": 0.8777990937232971, "learning_rate": 1.999918133908783e-05, "loss": 5.0922, "step": 2090 }, { "epoch": 0.005376, "grad_norm": 1.2641106843948364, "learning_rate": 1.9999171002705312e-05, "loss": 5.0065, "step": 2100 }, { "epoch": 0.0054016, "grad_norm": 1.740434169769287, "learning_rate": 1.999916060148083e-05, "loss": 5.2589, "step": 2110 }, { "epoch": 0.0054272, "grad_norm": 1.4229390621185303, "learning_rate": 1.9999150135414452e-05, "loss": 5.2258, "step": 2120 }, { "epoch": 0.0054528, "grad_norm": 0.9691534042358398, "learning_rate": 1.999913960450624e-05, "loss": 5.109, "step": 2130 }, { "epoch": 0.0054784, "grad_norm": 1.2083088159561157, "learning_rate": 1.9999129008756263e-05, "loss": 5.0464, "step": 2140 }, { "epoch": 0.005504, "grad_norm": 1.1759923696517944, "learning_rate": 1.9999118348164598e-05, "loss": 5.0785, "step": 2150 }, { "epoch": 0.0055296, "grad_norm": 1.5178769826889038, "learning_rate": 1.9999107622731303e-05, "loss": 5.2087, "step": 2160 }, { "epoch": 0.0055552, "grad_norm": 1.186970829963684, "learning_rate": 1.9999096832456454e-05, "loss": 5.2252, "step": 2170 }, { "epoch": 0.0055808, "grad_norm": 1.447716236114502, "learning_rate": 1.9999085977340122e-05, "loss": 5.3581, "step": 2180 }, { "epoch": 0.0056064, "grad_norm": 1.4647458791732788, "learning_rate": 1.999907505738237e-05, "loss": 5.1507, "step": 2190 }, { "epoch": 0.005632, "grad_norm": 1.1427520513534546, "learning_rate": 1.9999064072583278e-05, "loss": 5.3081, "step": 2200 }, { "epoch": 0.0056576, "grad_norm": 1.2448432445526123, "learning_rate": 1.999905302294291e-05, "loss": 5.2025, "step": 2210 }, { "epoch": 0.0056832, "grad_norm": 1.059588074684143, "learning_rate": 1.999904190846134e-05, "loss": 5.0258, "step": 2220 }, { "epoch": 0.0057088, "grad_norm": 1.2828288078308105, "learning_rate": 1.999903072913864e-05, "loss": 5.0003, "step": 2230 }, { "epoch": 0.0057344, "grad_norm": 1.218349814414978, "learning_rate": 1.999901948497489e-05, "loss": 4.9938, "step": 2240 }, { "epoch": 0.00576, "grad_norm": 0.9153841733932495, "learning_rate": 1.999900817597015e-05, "loss": 4.9048, "step": 2250 }, { "epoch": 0.0057856, "grad_norm": 1.0404701232910156, "learning_rate": 1.9998996802124502e-05, "loss": 5.4883, "step": 2260 }, { "epoch": 0.0058112, "grad_norm": 0.8994712233543396, "learning_rate": 1.9998985363438014e-05, "loss": 5.3878, "step": 2270 }, { "epoch": 0.0058368, "grad_norm": 1.3665951490402222, "learning_rate": 1.9998973859910768e-05, "loss": 5.3014, "step": 2280 }, { "epoch": 0.0058624, "grad_norm": 1.2929768562316895, "learning_rate": 1.999896229154283e-05, "loss": 5.2579, "step": 2290 }, { "epoch": 0.005888, "grad_norm": 1.5854967832565308, "learning_rate": 1.999895065833428e-05, "loss": 4.7782, "step": 2300 }, { "epoch": 0.0059136, "grad_norm": 1.1636348962783813, "learning_rate": 1.9998938960285193e-05, "loss": 5.6358, "step": 2310 }, { "epoch": 0.0059392, "grad_norm": 1.8489307165145874, "learning_rate": 1.999892719739564e-05, "loss": 5.4466, "step": 2320 }, { "epoch": 0.0059648, "grad_norm": 2.749114513397217, "learning_rate": 1.9998915369665707e-05, "loss": 5.5928, "step": 2330 }, { "epoch": 0.0059904, "grad_norm": 1.3073753118515015, "learning_rate": 1.9998903477095462e-05, "loss": 5.1601, "step": 2340 }, { "epoch": 0.006016, "grad_norm": 1.6820282936096191, "learning_rate": 1.9998891519684987e-05, "loss": 5.435, "step": 2350 }, { "epoch": 0.0060416, "grad_norm": 1.0388097763061523, "learning_rate": 1.999887949743436e-05, "loss": 5.1982, "step": 2360 }, { "epoch": 0.0060672, "grad_norm": 1.2904067039489746, "learning_rate": 1.9998867410343653e-05, "loss": 5.1784, "step": 2370 }, { "epoch": 0.0060928, "grad_norm": 0.879833996295929, "learning_rate": 1.999885525841295e-05, "loss": 5.025, "step": 2380 }, { "epoch": 0.0061184, "grad_norm": 0.7601300477981567, "learning_rate": 1.999884304164233e-05, "loss": 5.4035, "step": 2390 }, { "epoch": 0.006144, "grad_norm": 1.2355035543441772, "learning_rate": 1.9998830760031867e-05, "loss": 5.0208, "step": 2400 }, { "epoch": 0.0061696, "grad_norm": 0.9268048405647278, "learning_rate": 1.9998818413581643e-05, "loss": 5.3942, "step": 2410 }, { "epoch": 0.0061952, "grad_norm": 1.2550307512283325, "learning_rate": 1.999880600229174e-05, "loss": 4.9986, "step": 2420 }, { "epoch": 0.0062208, "grad_norm": 0.7791248559951782, "learning_rate": 1.999879352616224e-05, "loss": 5.1191, "step": 2430 }, { "epoch": 0.0062464, "grad_norm": 0.8772755265235901, "learning_rate": 1.999878098519322e-05, "loss": 4.956, "step": 2440 }, { "epoch": 0.006272, "grad_norm": 0.8500017523765564, "learning_rate": 1.9998768379384764e-05, "loss": 5.0994, "step": 2450 }, { "epoch": 0.0062976, "grad_norm": 0.8597416877746582, "learning_rate": 1.999875570873695e-05, "loss": 5.2537, "step": 2460 }, { "epoch": 0.0063232, "grad_norm": 1.6287099123001099, "learning_rate": 1.9998742973249862e-05, "loss": 5.2363, "step": 2470 }, { "epoch": 0.0063488, "grad_norm": 1.783626914024353, "learning_rate": 1.9998730172923586e-05, "loss": 5.3444, "step": 2480 }, { "epoch": 0.0063744, "grad_norm": 1.047737717628479, "learning_rate": 1.9998717307758202e-05, "loss": 5.0443, "step": 2490 }, { "epoch": 0.0064, "grad_norm": 0.8462600111961365, "learning_rate": 1.9998704377753794e-05, "loss": 5.1304, "step": 2500 }, { "epoch": 0.0064256, "grad_norm": 1.0595868825912476, "learning_rate": 1.999869138291044e-05, "loss": 4.9513, "step": 2510 }, { "epoch": 0.0064512, "grad_norm": 1.3633850812911987, "learning_rate": 1.9998678323228232e-05, "loss": 4.8032, "step": 2520 }, { "epoch": 0.0064768, "grad_norm": 1.2226295471191406, "learning_rate": 1.9998665198707258e-05, "loss": 5.0707, "step": 2530 }, { "epoch": 0.0065024, "grad_norm": 1.2296866178512573, "learning_rate": 1.9998652009347592e-05, "loss": 4.9368, "step": 2540 }, { "epoch": 0.006528, "grad_norm": 1.1788098812103271, "learning_rate": 1.9998638755149326e-05, "loss": 4.9732, "step": 2550 }, { "epoch": 0.0065536, "grad_norm": 0.7831915616989136, "learning_rate": 1.9998625436112545e-05, "loss": 4.8201, "step": 2560 }, { "epoch": 0.0065792, "grad_norm": 2.436612844467163, "learning_rate": 1.999861205223734e-05, "loss": 5.2748, "step": 2570 }, { "epoch": 0.0066048, "grad_norm": 0.8487124443054199, "learning_rate": 1.9998598603523785e-05, "loss": 4.7744, "step": 2580 }, { "epoch": 0.0066304, "grad_norm": 1.0266036987304688, "learning_rate": 1.999858508997198e-05, "loss": 5.0938, "step": 2590 }, { "epoch": 0.006656, "grad_norm": 1.483925461769104, "learning_rate": 1.9998571511582007e-05, "loss": 5.2804, "step": 2600 }, { "epoch": 0.0066816, "grad_norm": 3.0643556118011475, "learning_rate": 1.9998557868353957e-05, "loss": 4.963, "step": 2610 }, { "epoch": 0.0067072, "grad_norm": 1.5542901754379272, "learning_rate": 1.9998544160287913e-05, "loss": 5.1845, "step": 2620 }, { "epoch": 0.0067328, "grad_norm": 0.87354975938797, "learning_rate": 1.999853038738397e-05, "loss": 4.926, "step": 2630 }, { "epoch": 0.0067584, "grad_norm": 1.269756555557251, "learning_rate": 1.9998516549642213e-05, "loss": 5.0175, "step": 2640 }, { "epoch": 0.006784, "grad_norm": 2.710296392440796, "learning_rate": 1.9998502647062733e-05, "loss": 5.5496, "step": 2650 }, { "epoch": 0.0068096, "grad_norm": 1.0240976810455322, "learning_rate": 1.9998488679645626e-05, "loss": 4.8579, "step": 2660 }, { "epoch": 0.0068352, "grad_norm": 1.2339860200881958, "learning_rate": 1.9998474647390972e-05, "loss": 4.9351, "step": 2670 }, { "epoch": 0.0068608, "grad_norm": 0.7906849384307861, "learning_rate": 1.9998460550298867e-05, "loss": 5.3167, "step": 2680 }, { "epoch": 0.0068864, "grad_norm": 1.959855318069458, "learning_rate": 1.9998446388369405e-05, "loss": 4.9242, "step": 2690 }, { "epoch": 0.006912, "grad_norm": 2.422492504119873, "learning_rate": 1.9998432161602674e-05, "loss": 5.38, "step": 2700 }, { "epoch": 0.0069376, "grad_norm": 1.100089192390442, "learning_rate": 1.999841786999877e-05, "loss": 5.0317, "step": 2710 }, { "epoch": 0.0069632, "grad_norm": 1.265731692314148, "learning_rate": 1.9998403513557783e-05, "loss": 5.1019, "step": 2720 }, { "epoch": 0.0069888, "grad_norm": 1.0354787111282349, "learning_rate": 1.9998389092279803e-05, "loss": 5.1273, "step": 2730 }, { "epoch": 0.0070144, "grad_norm": 1.2718696594238281, "learning_rate": 1.999837460616493e-05, "loss": 5.0652, "step": 2740 }, { "epoch": 0.00704, "grad_norm": 1.1323562860488892, "learning_rate": 1.9998360055213252e-05, "loss": 5.0701, "step": 2750 }, { "epoch": 0.0070656, "grad_norm": 1.452679991722107, "learning_rate": 1.999834543942487e-05, "loss": 5.1904, "step": 2760 }, { "epoch": 0.0070912, "grad_norm": 1.2068257331848145, "learning_rate": 1.9998330758799873e-05, "loss": 5.1342, "step": 2770 }, { "epoch": 0.0071168, "grad_norm": 3.0662405490875244, "learning_rate": 1.999831601333836e-05, "loss": 5.2011, "step": 2780 }, { "epoch": 0.0071424, "grad_norm": 0.8274203538894653, "learning_rate": 1.9998301203040422e-05, "loss": 5.2621, "step": 2790 }, { "epoch": 0.007168, "grad_norm": 1.6426676511764526, "learning_rate": 1.999828632790616e-05, "loss": 4.9259, "step": 2800 }, { "epoch": 0.0071936, "grad_norm": 1.787283182144165, "learning_rate": 1.9998271387935667e-05, "loss": 4.869, "step": 2810 }, { "epoch": 0.0072192, "grad_norm": 2.6628530025482178, "learning_rate": 1.9998256383129042e-05, "loss": 5.5381, "step": 2820 }, { "epoch": 0.0072448, "grad_norm": 1.397439956665039, "learning_rate": 1.9998241313486382e-05, "loss": 4.9047, "step": 2830 }, { "epoch": 0.0072704, "grad_norm": 0.9861513376235962, "learning_rate": 1.999822617900778e-05, "loss": 4.9418, "step": 2840 }, { "epoch": 0.007296, "grad_norm": 1.120434284210205, "learning_rate": 1.9998210979693345e-05, "loss": 4.8135, "step": 2850 }, { "epoch": 0.0073216, "grad_norm": 1.078794002532959, "learning_rate": 1.999819571554317e-05, "loss": 5.2729, "step": 2860 }, { "epoch": 0.0073472, "grad_norm": 2.8487656116485596, "learning_rate": 1.9998180386557346e-05, "loss": 5.7077, "step": 2870 }, { "epoch": 0.0073728, "grad_norm": 1.069906234741211, "learning_rate": 1.9998164992735983e-05, "loss": 5.2518, "step": 2880 }, { "epoch": 0.0073984, "grad_norm": 1.2204437255859375, "learning_rate": 1.9998149534079178e-05, "loss": 5.2067, "step": 2890 }, { "epoch": 0.007424, "grad_norm": 6.030945301055908, "learning_rate": 1.9998134010587028e-05, "loss": 5.2442, "step": 2900 }, { "epoch": 0.0074496, "grad_norm": 0.9002209305763245, "learning_rate": 1.999811842225964e-05, "loss": 5.055, "step": 2910 }, { "epoch": 0.0074752, "grad_norm": 1.2694863080978394, "learning_rate": 1.9998102769097107e-05, "loss": 4.8359, "step": 2920 }, { "epoch": 0.0075008, "grad_norm": 0.8420307636260986, "learning_rate": 1.9998087051099538e-05, "loss": 4.9573, "step": 2930 }, { "epoch": 0.0075264, "grad_norm": 1.0173224210739136, "learning_rate": 1.999807126826703e-05, "loss": 5.0084, "step": 2940 }, { "epoch": 0.007552, "grad_norm": 2.927436590194702, "learning_rate": 1.9998055420599686e-05, "loss": 5.3252, "step": 2950 }, { "epoch": 0.0075776, "grad_norm": 1.0222910642623901, "learning_rate": 1.9998039508097613e-05, "loss": 5.0906, "step": 2960 }, { "epoch": 0.0076032, "grad_norm": 0.8986455798149109, "learning_rate": 1.9998023530760908e-05, "loss": 4.8713, "step": 2970 }, { "epoch": 0.0076288, "grad_norm": 0.8419018387794495, "learning_rate": 1.9998007488589678e-05, "loss": 4.8399, "step": 2980 }, { "epoch": 0.0076544, "grad_norm": 0.9209607243537903, "learning_rate": 1.9997991381584026e-05, "loss": 5.0833, "step": 2990 }, { "epoch": 0.00768, "grad_norm": 0.7017521858215332, "learning_rate": 1.999797520974406e-05, "loss": 4.8055, "step": 3000 }, { "epoch": 0.0077056, "grad_norm": 0.9618639945983887, "learning_rate": 1.9997958973069877e-05, "loss": 4.9048, "step": 3010 }, { "epoch": 0.0077312, "grad_norm": 0.8212770819664001, "learning_rate": 1.9997942671561594e-05, "loss": 4.9143, "step": 3020 }, { "epoch": 0.0077568, "grad_norm": 0.8915886282920837, "learning_rate": 1.9997926305219305e-05, "loss": 4.9437, "step": 3030 }, { "epoch": 0.0077824, "grad_norm": 4.646025657653809, "learning_rate": 1.9997909874043122e-05, "loss": 6.2859, "step": 3040 }, { "epoch": 0.007808, "grad_norm": 6.148813247680664, "learning_rate": 1.999789337803315e-05, "loss": 5.6107, "step": 3050 }, { "epoch": 0.0078336, "grad_norm": 1.0593239068984985, "learning_rate": 1.9997876817189498e-05, "loss": 4.8368, "step": 3060 }, { "epoch": 0.0078592, "grad_norm": 0.9626774191856384, "learning_rate": 1.999786019151227e-05, "loss": 4.8696, "step": 3070 }, { "epoch": 0.0078848, "grad_norm": 1.3050861358642578, "learning_rate": 1.999784350100158e-05, "loss": 4.8045, "step": 3080 }, { "epoch": 0.0079104, "grad_norm": 1.1973904371261597, "learning_rate": 1.9997826745657528e-05, "loss": 4.9524, "step": 3090 }, { "epoch": 0.007936, "grad_norm": 1.2012720108032227, "learning_rate": 1.9997809925480224e-05, "loss": 4.9361, "step": 3100 }, { "epoch": 0.0079616, "grad_norm": 1.1393170356750488, "learning_rate": 1.9997793040469785e-05, "loss": 4.9956, "step": 3110 }, { "epoch": 0.0079872, "grad_norm": 1.0409966707229614, "learning_rate": 1.9997776090626316e-05, "loss": 4.9308, "step": 3120 }, { "epoch": 0.0080128, "grad_norm": 1.200215458869934, "learning_rate": 1.9997759075949922e-05, "loss": 4.9048, "step": 3130 }, { "epoch": 0.0080384, "grad_norm": 1.8657424449920654, "learning_rate": 1.999774199644072e-05, "loss": 5.0014, "step": 3140 }, { "epoch": 0.008064, "grad_norm": 0.985859751701355, "learning_rate": 1.9997724852098816e-05, "loss": 4.8156, "step": 3150 }, { "epoch": 0.0080896, "grad_norm": 1.1623003482818604, "learning_rate": 1.9997707642924325e-05, "loss": 4.8108, "step": 3160 }, { "epoch": 0.0081152, "grad_norm": 2.014660358428955, "learning_rate": 1.9997690368917356e-05, "loss": 4.9923, "step": 3170 }, { "epoch": 0.0081408, "grad_norm": 1.0094740390777588, "learning_rate": 1.9997673030078024e-05, "loss": 4.8997, "step": 3180 }, { "epoch": 0.0081664, "grad_norm": 1.386367678642273, "learning_rate": 1.9997655626406442e-05, "loss": 5.1232, "step": 3190 }, { "epoch": 0.008192, "grad_norm": 0.7553184032440186, "learning_rate": 1.9997638157902714e-05, "loss": 5.1143, "step": 3200 }, { "epoch": 0.0082176, "grad_norm": 1.1478990316390991, "learning_rate": 1.999762062456696e-05, "loss": 4.8301, "step": 3210 }, { "epoch": 0.0082432, "grad_norm": 3.192265748977661, "learning_rate": 1.9997603026399298e-05, "loss": 5.0312, "step": 3220 }, { "epoch": 0.0082688, "grad_norm": 0.9990370273590088, "learning_rate": 1.9997585363399834e-05, "loss": 5.044, "step": 3230 }, { "epoch": 0.0082944, "grad_norm": 0.9352939128875732, "learning_rate": 1.9997567635568688e-05, "loss": 4.9125, "step": 3240 }, { "epoch": 0.00832, "grad_norm": 0.8116731643676758, "learning_rate": 1.999754984290597e-05, "loss": 4.3436, "step": 3250 }, { "epoch": 0.0083456, "grad_norm": 1.3531445264816284, "learning_rate": 1.9997531985411803e-05, "loss": 5.3226, "step": 3260 }, { "epoch": 0.0083712, "grad_norm": 0.9775050282478333, "learning_rate": 1.9997514063086293e-05, "loss": 4.8879, "step": 3270 }, { "epoch": 0.0083968, "grad_norm": 1.2834057807922363, "learning_rate": 1.9997496075929567e-05, "loss": 4.9687, "step": 3280 }, { "epoch": 0.0084224, "grad_norm": 1.1513357162475586, "learning_rate": 1.999747802394173e-05, "loss": 4.8045, "step": 3290 }, { "epoch": 0.008448, "grad_norm": 1.3775718212127686, "learning_rate": 1.999745990712291e-05, "loss": 4.4742, "step": 3300 }, { "epoch": 0.0084736, "grad_norm": 0.945850133895874, "learning_rate": 1.9997441725473214e-05, "loss": 4.9858, "step": 3310 }, { "epoch": 0.0084992, "grad_norm": 1.3380793333053589, "learning_rate": 1.999742347899277e-05, "loss": 4.8658, "step": 3320 }, { "epoch": 0.0085248, "grad_norm": 1.4593005180358887, "learning_rate": 1.9997405167681688e-05, "loss": 4.7992, "step": 3330 }, { "epoch": 0.0085504, "grad_norm": 3.0540707111358643, "learning_rate": 1.9997386791540093e-05, "loss": 4.9285, "step": 3340 }, { "epoch": 0.008576, "grad_norm": 1.467592716217041, "learning_rate": 1.9997368350568104e-05, "loss": 5.0254, "step": 3350 }, { "epoch": 0.0086016, "grad_norm": 0.8184484243392944, "learning_rate": 1.9997349844765833e-05, "loss": 4.9115, "step": 3360 }, { "epoch": 0.0086272, "grad_norm": 1.7319018840789795, "learning_rate": 1.9997331274133408e-05, "loss": 4.7239, "step": 3370 }, { "epoch": 0.0086528, "grad_norm": 1.1393505334854126, "learning_rate": 1.999731263867095e-05, "loss": 4.7954, "step": 3380 }, { "epoch": 0.0086784, "grad_norm": 1.071773648262024, "learning_rate": 1.999729393837857e-05, "loss": 5.0876, "step": 3390 }, { "epoch": 0.008704, "grad_norm": 1.5409640073776245, "learning_rate": 1.99972751732564e-05, "loss": 4.7579, "step": 3400 }, { "epoch": 0.0087296, "grad_norm": 1.0954293012619019, "learning_rate": 1.9997256343304556e-05, "loss": 5.2699, "step": 3410 }, { "epoch": 0.0087552, "grad_norm": 1.0712913274765015, "learning_rate": 1.999723744852316e-05, "loss": 4.8589, "step": 3420 }, { "epoch": 0.0087808, "grad_norm": 0.9777956008911133, "learning_rate": 1.999721848891234e-05, "loss": 4.8648, "step": 3430 }, { "epoch": 0.0088064, "grad_norm": 1.2113597393035889, "learning_rate": 1.9997199464472214e-05, "loss": 5.0639, "step": 3440 }, { "epoch": 0.008832, "grad_norm": 1.1152938604354858, "learning_rate": 1.99971803752029e-05, "loss": 4.9621, "step": 3450 }, { "epoch": 0.0088576, "grad_norm": 0.9489766955375671, "learning_rate": 1.9997161221104536e-05, "loss": 4.9847, "step": 3460 }, { "epoch": 0.0088832, "grad_norm": 1.517730474472046, "learning_rate": 1.9997142002177237e-05, "loss": 4.9234, "step": 3470 }, { "epoch": 0.0089088, "grad_norm": 0.9284108281135559, "learning_rate": 1.999712271842113e-05, "loss": 4.8743, "step": 3480 }, { "epoch": 0.0089344, "grad_norm": 0.9875884056091309, "learning_rate": 1.9997103369836333e-05, "loss": 4.9713, "step": 3490 }, { "epoch": 0.00896, "grad_norm": 1.2524380683898926, "learning_rate": 1.9997083956422986e-05, "loss": 4.8707, "step": 3500 }, { "epoch": 0.0089856, "grad_norm": 4.456395626068115, "learning_rate": 1.9997064478181202e-05, "loss": 4.8321, "step": 3510 }, { "epoch": 0.0090112, "grad_norm": 1.3802707195281982, "learning_rate": 1.999704493511111e-05, "loss": 4.6251, "step": 3520 }, { "epoch": 0.0090368, "grad_norm": 1.3686059713363647, "learning_rate": 1.9997025327212843e-05, "loss": 4.7973, "step": 3530 }, { "epoch": 0.0090624, "grad_norm": 1.3320056200027466, "learning_rate": 1.9997005654486518e-05, "loss": 5.0673, "step": 3540 }, { "epoch": 0.009088, "grad_norm": 1.3500491380691528, "learning_rate": 1.9996985916932272e-05, "loss": 4.6485, "step": 3550 }, { "epoch": 0.0091136, "grad_norm": 0.9931584596633911, "learning_rate": 1.999696611455023e-05, "loss": 4.8128, "step": 3560 }, { "epoch": 0.0091392, "grad_norm": 1.0735350847244263, "learning_rate": 1.9996946247340516e-05, "loss": 5.0192, "step": 3570 }, { "epoch": 0.0091648, "grad_norm": 0.8469030857086182, "learning_rate": 1.9996926315303264e-05, "loss": 5.0007, "step": 3580 }, { "epoch": 0.0091904, "grad_norm": 1.561397671699524, "learning_rate": 1.9996906318438602e-05, "loss": 4.9795, "step": 3590 }, { "epoch": 0.009216, "grad_norm": 0.9925048351287842, "learning_rate": 1.999688625674666e-05, "loss": 5.1025, "step": 3600 }, { "epoch": 0.0092416, "grad_norm": 1.0562442541122437, "learning_rate": 1.9996866130227567e-05, "loss": 4.9264, "step": 3610 }, { "epoch": 0.0092672, "grad_norm": 0.9895694851875305, "learning_rate": 1.9996845938881454e-05, "loss": 4.9, "step": 3620 }, { "epoch": 1.00000256, "grad_norm": 1.6928483247756958, "learning_rate": 1.999682568270845e-05, "loss": 5.3216, "step": 3630 }, { "epoch": 1.00002816, "grad_norm": 0.9945420622825623, "learning_rate": 1.9996805361708695e-05, "loss": 5.0095, "step": 3640 }, { "epoch": 1.00005376, "grad_norm": 1.4631975889205933, "learning_rate": 1.9996784975882306e-05, "loss": 4.5125, "step": 3650 }, { "epoch": 1.00007936, "grad_norm": 0.9631759524345398, "learning_rate": 1.999676452522943e-05, "loss": 4.4227, "step": 3660 }, { "epoch": 1.00010496, "grad_norm": 0.9101148843765259, "learning_rate": 1.9996744009750188e-05, "loss": 4.7133, "step": 3670 }, { "epoch": 1.00013056, "grad_norm": 1.7530477046966553, "learning_rate": 1.999672342944472e-05, "loss": 5.2923, "step": 3680 }, { "epoch": 1.00015616, "grad_norm": 2.1923341751098633, "learning_rate": 1.9996702784313156e-05, "loss": 4.6775, "step": 3690 }, { "epoch": 1.00018176, "grad_norm": 2.13374662399292, "learning_rate": 1.9996682074355634e-05, "loss": 5.1835, "step": 3700 }, { "epoch": 1.00020736, "grad_norm": 1.2379436492919922, "learning_rate": 1.999666129957228e-05, "loss": 4.9551, "step": 3710 }, { "epoch": 1.00023296, "grad_norm": 0.9113930463790894, "learning_rate": 1.9996640459963238e-05, "loss": 4.8727, "step": 3720 }, { "epoch": 1.00025856, "grad_norm": 1.3093386888504028, "learning_rate": 1.9996619555528642e-05, "loss": 5.1026, "step": 3730 }, { "epoch": 1.00028416, "grad_norm": 1.5994341373443604, "learning_rate": 1.9996598586268622e-05, "loss": 4.5239, "step": 3740 }, { "epoch": 1.00030976, "grad_norm": 1.6066362857818604, "learning_rate": 1.9996577552183317e-05, "loss": 4.2641, "step": 3750 }, { "epoch": 1.00033536, "grad_norm": 0.933667004108429, "learning_rate": 1.999655645327286e-05, "loss": 4.4226, "step": 3760 }, { "epoch": 1.00036096, "grad_norm": 1.6100062131881714, "learning_rate": 1.9996535289537396e-05, "loss": 4.483, "step": 3770 }, { "epoch": 1.00038656, "grad_norm": 0.996505618095398, "learning_rate": 1.9996514060977057e-05, "loss": 4.4112, "step": 3780 }, { "epoch": 1.00041216, "grad_norm": 1.2887256145477295, "learning_rate": 1.9996492767591982e-05, "loss": 4.7381, "step": 3790 }, { "epoch": 1.00043776, "grad_norm": 0.9070744514465332, "learning_rate": 1.9996471409382304e-05, "loss": 4.6169, "step": 3800 }, { "epoch": 1.00046336, "grad_norm": 0.881377100944519, "learning_rate": 1.9996449986348167e-05, "loss": 4.7789, "step": 3810 }, { "epoch": 1.00048896, "grad_norm": 2.719639301300049, "learning_rate": 1.999642849848971e-05, "loss": 4.6181, "step": 3820 }, { "epoch": 1.00051456, "grad_norm": 2.3179168701171875, "learning_rate": 1.9996406945807072e-05, "loss": 4.7101, "step": 3830 }, { "epoch": 1.00054016, "grad_norm": 1.0221277475357056, "learning_rate": 1.9996385328300388e-05, "loss": 4.7761, "step": 3840 }, { "epoch": 1.00056576, "grad_norm": 0.9702327251434326, "learning_rate": 1.9996363645969805e-05, "loss": 4.194, "step": 3850 }, { "epoch": 1.00059136, "grad_norm": 1.6270419359207153, "learning_rate": 1.9996341898815462e-05, "loss": 4.6784, "step": 3860 }, { "epoch": 1.00061696, "grad_norm": 0.8938523530960083, "learning_rate": 1.9996320086837494e-05, "loss": 4.4748, "step": 3870 }, { "epoch": 1.00064256, "grad_norm": 1.2363131046295166, "learning_rate": 1.9996298210036048e-05, "loss": 4.4481, "step": 3880 }, { "epoch": 1.00066816, "grad_norm": 1.2837870121002197, "learning_rate": 1.9996276268411264e-05, "loss": 4.5902, "step": 3890 }, { "epoch": 1.00069376, "grad_norm": 1.2830886840820312, "learning_rate": 1.999625426196329e-05, "loss": 4.7893, "step": 3900 }, { "epoch": 1.00071936, "grad_norm": 0.8279328942298889, "learning_rate": 1.999623219069226e-05, "loss": 4.8765, "step": 3910 }, { "epoch": 1.00074496, "grad_norm": 0.8628543019294739, "learning_rate": 1.999621005459832e-05, "loss": 4.6053, "step": 3920 }, { "epoch": 1.00077056, "grad_norm": 1.2594965696334839, "learning_rate": 1.999618785368162e-05, "loss": 4.5198, "step": 3930 }, { "epoch": 1.00079616, "grad_norm": 1.5882312059402466, "learning_rate": 1.9996165587942294e-05, "loss": 4.9772, "step": 3940 }, { "epoch": 1.00082176, "grad_norm": 0.8213825821876526, "learning_rate": 1.999614325738049e-05, "loss": 4.6267, "step": 3950 }, { "epoch": 1.00084736, "grad_norm": 0.9538202881813049, "learning_rate": 1.9996120861996355e-05, "loss": 4.3159, "step": 3960 }, { "epoch": 1.00087296, "grad_norm": 1.0800148248672485, "learning_rate": 1.9996098401790032e-05, "loss": 4.4744, "step": 3970 }, { "epoch": 1.00089856, "grad_norm": 1.3683934211730957, "learning_rate": 1.9996075876761667e-05, "loss": 4.7694, "step": 3980 }, { "epoch": 1.00092416, "grad_norm": 1.4272164106369019, "learning_rate": 1.999605328691141e-05, "loss": 4.4641, "step": 3990 }, { "epoch": 1.00094976, "grad_norm": 0.9858846664428711, "learning_rate": 1.99960306322394e-05, "loss": 4.5447, "step": 4000 }, { "epoch": 1.00097536, "grad_norm": 1.0774524211883545, "learning_rate": 1.999600791274579e-05, "loss": 4.6062, "step": 4010 }, { "epoch": 1.00100096, "grad_norm": 1.9114875793457031, "learning_rate": 1.9995985128430725e-05, "loss": 4.7021, "step": 4020 }, { "epoch": 1.00102656, "grad_norm": 1.0697044134140015, "learning_rate": 1.9995962279294355e-05, "loss": 4.5993, "step": 4030 }, { "epoch": 1.00105216, "grad_norm": 0.7484090924263, "learning_rate": 1.9995939365336822e-05, "loss": 4.2903, "step": 4040 }, { "epoch": 1.00107776, "grad_norm": 1.4322474002838135, "learning_rate": 1.999591638655828e-05, "loss": 4.2918, "step": 4050 }, { "epoch": 1.00110336, "grad_norm": 0.9084761142730713, "learning_rate": 1.999589334295888e-05, "loss": 4.5186, "step": 4060 }, { "epoch": 1.00112896, "grad_norm": 0.8006147146224976, "learning_rate": 1.999587023453877e-05, "loss": 4.4256, "step": 4070 }, { "epoch": 1.00115456, "grad_norm": 1.0194430351257324, "learning_rate": 1.999584706129809e-05, "loss": 4.3829, "step": 4080 }, { "epoch": 1.00118016, "grad_norm": 0.8401071429252625, "learning_rate": 1.9995823823237006e-05, "loss": 4.4982, "step": 4090 }, { "epoch": 1.00120576, "grad_norm": 1.6061104536056519, "learning_rate": 1.9995800520355656e-05, "loss": 4.606, "step": 4100 }, { "epoch": 1.00123136, "grad_norm": 1.2562611103057861, "learning_rate": 1.99957771526542e-05, "loss": 4.2998, "step": 4110 }, { "epoch": 1.00125696, "grad_norm": 1.2375258207321167, "learning_rate": 1.9995753720132782e-05, "loss": 4.6775, "step": 4120 }, { "epoch": 1.00128256, "grad_norm": 1.2595109939575195, "learning_rate": 1.999573022279156e-05, "loss": 4.4837, "step": 4130 }, { "epoch": 1.00130816, "grad_norm": 0.967567503452301, "learning_rate": 1.999570666063068e-05, "loss": 4.6424, "step": 4140 }, { "epoch": 1.00133376, "grad_norm": 0.7728231549263, "learning_rate": 1.9995683033650307e-05, "loss": 4.5338, "step": 4150 }, { "epoch": 1.00135936, "grad_norm": 2.280169725418091, "learning_rate": 1.9995659341850578e-05, "loss": 4.7466, "step": 4160 }, { "epoch": 1.00138496, "grad_norm": 0.9471111297607422, "learning_rate": 1.999563558523166e-05, "loss": 4.4411, "step": 4170 }, { "epoch": 1.00141056, "grad_norm": 2.3587961196899414, "learning_rate": 1.99956117637937e-05, "loss": 4.7618, "step": 4180 }, { "epoch": 1.00143616, "grad_norm": 0.7838448882102966, "learning_rate": 1.9995587877536857e-05, "loss": 4.6152, "step": 4190 }, { "epoch": 1.00146176, "grad_norm": 0.9601026773452759, "learning_rate": 1.9995563926461277e-05, "loss": 4.8242, "step": 4200 }, { "epoch": 1.00148736, "grad_norm": 1.5960084199905396, "learning_rate": 1.9995539910567125e-05, "loss": 4.9147, "step": 4210 }, { "epoch": 1.00151296, "grad_norm": 0.9353383779525757, "learning_rate": 1.9995515829854556e-05, "loss": 4.5132, "step": 4220 }, { "epoch": 1.00153856, "grad_norm": 0.8761973977088928, "learning_rate": 1.999549168432372e-05, "loss": 4.4413, "step": 4230 }, { "epoch": 1.00156416, "grad_norm": 1.1045242547988892, "learning_rate": 1.9995467473974773e-05, "loss": 4.8944, "step": 4240 }, { "epoch": 1.00158976, "grad_norm": 1.6035380363464355, "learning_rate": 1.999544319880788e-05, "loss": 4.8039, "step": 4250 }, { "epoch": 1.00161536, "grad_norm": 4.047190189361572, "learning_rate": 1.9995418858823196e-05, "loss": 5.1564, "step": 4260 }, { "epoch": 1.00164096, "grad_norm": 1.4208452701568604, "learning_rate": 1.9995394454020876e-05, "loss": 4.5838, "step": 4270 }, { "epoch": 1.00166656, "grad_norm": 1.0237945318222046, "learning_rate": 1.999536998440108e-05, "loss": 4.7487, "step": 4280 }, { "epoch": 1.00169216, "grad_norm": 1.1041886806488037, "learning_rate": 1.9995345449963964e-05, "loss": 4.9218, "step": 4290 }, { "epoch": 1.00171776, "grad_norm": 1.1015987396240234, "learning_rate": 1.999532085070969e-05, "loss": 5.0031, "step": 4300 }, { "epoch": 1.00174336, "grad_norm": 0.802054226398468, "learning_rate": 1.9995296186638418e-05, "loss": 4.769, "step": 4310 }, { "epoch": 1.00176896, "grad_norm": 0.8228839635848999, "learning_rate": 1.9995271457750305e-05, "loss": 4.8657, "step": 4320 }, { "epoch": 1.00179456, "grad_norm": 1.5645803213119507, "learning_rate": 1.9995246664045513e-05, "loss": 5.1326, "step": 4330 }, { "epoch": 1.00182016, "grad_norm": 1.0472170114517212, "learning_rate": 1.9995221805524205e-05, "loss": 4.8146, "step": 4340 }, { "epoch": 1.00184576, "grad_norm": 0.7949263453483582, "learning_rate": 1.999519688218654e-05, "loss": 4.5395, "step": 4350 }, { "epoch": 1.00187136, "grad_norm": 1.5860822200775146, "learning_rate": 1.9995171894032676e-05, "loss": 4.6883, "step": 4360 }, { "epoch": 1.00189696, "grad_norm": 1.082730770111084, "learning_rate": 1.999514684106278e-05, "loss": 4.6088, "step": 4370 }, { "epoch": 1.00192256, "grad_norm": 1.1569290161132812, "learning_rate": 1.9995121723277014e-05, "loss": 4.4121, "step": 4380 }, { "epoch": 1.00194816, "grad_norm": 0.7283101677894592, "learning_rate": 1.999509654067554e-05, "loss": 4.5522, "step": 4390 }, { "epoch": 1.00197376, "grad_norm": 0.7884061932563782, "learning_rate": 1.9995071293258522e-05, "loss": 4.6177, "step": 4400 }, { "epoch": 1.00199936, "grad_norm": 1.7358251810073853, "learning_rate": 1.9995045981026123e-05, "loss": 4.874, "step": 4410 }, { "epoch": 1.00202496, "grad_norm": 0.892126202583313, "learning_rate": 1.9995020603978505e-05, "loss": 4.4744, "step": 4420 }, { "epoch": 1.00205056, "grad_norm": 1.165029764175415, "learning_rate": 1.9994995162115835e-05, "loss": 4.3824, "step": 4430 }, { "epoch": 1.00207616, "grad_norm": 0.919543981552124, "learning_rate": 1.999496965543828e-05, "loss": 4.8346, "step": 4440 }, { "epoch": 1.00210176, "grad_norm": 0.8126477003097534, "learning_rate": 1.9994944083946004e-05, "loss": 4.7453, "step": 4450 }, { "epoch": 1.00212736, "grad_norm": 1.0275040864944458, "learning_rate": 1.999491844763917e-05, "loss": 4.5692, "step": 4460 }, { "epoch": 1.00215296, "grad_norm": 1.0434828996658325, "learning_rate": 1.9994892746517945e-05, "loss": 4.7376, "step": 4470 }, { "epoch": 1.00217856, "grad_norm": 1.2648792266845703, "learning_rate": 1.99948669805825e-05, "loss": 4.5056, "step": 4480 }, { "epoch": 1.00220416, "grad_norm": 2.2645530700683594, "learning_rate": 1.9994841149832996e-05, "loss": 4.5382, "step": 4490 }, { "epoch": 1.00222976, "grad_norm": 0.8589903712272644, "learning_rate": 1.9994815254269606e-05, "loss": 4.6108, "step": 4500 }, { "epoch": 1.00225536, "grad_norm": 0.7617235779762268, "learning_rate": 1.999478929389249e-05, "loss": 4.7914, "step": 4510 }, { "epoch": 1.00228096, "grad_norm": 1.1121118068695068, "learning_rate": 1.999476326870183e-05, "loss": 4.6375, "step": 4520 }, { "epoch": 1.00230656, "grad_norm": 1.0980494022369385, "learning_rate": 1.999473717869778e-05, "loss": 4.4686, "step": 4530 }, { "epoch": 1.00233216, "grad_norm": 1.0947949886322021, "learning_rate": 1.999471102388052e-05, "loss": 4.5681, "step": 4540 }, { "epoch": 1.00235776, "grad_norm": 1.2418389320373535, "learning_rate": 1.999468480425021e-05, "loss": 4.985, "step": 4550 }, { "epoch": 1.00238336, "grad_norm": 1.3359341621398926, "learning_rate": 1.999465851980703e-05, "loss": 4.639, "step": 4560 }, { "epoch": 1.00240896, "grad_norm": 1.0675448179244995, "learning_rate": 1.9994632170551142e-05, "loss": 4.6527, "step": 4570 }, { "epoch": 1.00243456, "grad_norm": 1.2809854745864868, "learning_rate": 1.9994605756482723e-05, "loss": 4.7216, "step": 4580 }, { "epoch": 1.00246016, "grad_norm": 1.0470207929611206, "learning_rate": 1.9994579277601942e-05, "loss": 4.4552, "step": 4590 }, { "epoch": 1.00248576, "grad_norm": 1.293008804321289, "learning_rate": 1.999455273390897e-05, "loss": 4.5425, "step": 4600 }, { "epoch": 1.00251136, "grad_norm": 1.2648515701293945, "learning_rate": 1.9994526125403977e-05, "loss": 4.8097, "step": 4610 }, { "epoch": 1.00253696, "grad_norm": 0.8044378757476807, "learning_rate": 1.9994499452087143e-05, "loss": 4.5041, "step": 4620 }, { "epoch": 1.00256256, "grad_norm": 1.0595693588256836, "learning_rate": 1.9994472713958636e-05, "loss": 4.5968, "step": 4630 }, { "epoch": 1.00258816, "grad_norm": 1.1323935985565186, "learning_rate": 1.999444591101863e-05, "loss": 4.375, "step": 4640 }, { "epoch": 1.00261376, "grad_norm": 0.7541035413742065, "learning_rate": 1.9994419043267295e-05, "loss": 4.6897, "step": 4650 }, { "epoch": 1.00263936, "grad_norm": 0.9236681461334229, "learning_rate": 1.999439211070481e-05, "loss": 4.6592, "step": 4660 }, { "epoch": 1.00266496, "grad_norm": 0.8407584428787231, "learning_rate": 1.999436511333135e-05, "loss": 4.6318, "step": 4670 }, { "epoch": 1.00269056, "grad_norm": 1.1560733318328857, "learning_rate": 1.9994338051147085e-05, "loss": 4.6108, "step": 4680 }, { "epoch": 1.00271616, "grad_norm": 1.1888973712921143, "learning_rate": 1.9994310924152197e-05, "loss": 4.9865, "step": 4690 }, { "epoch": 1.00274176, "grad_norm": 1.2526220083236694, "learning_rate": 1.999428373234686e-05, "loss": 4.676, "step": 4700 }, { "epoch": 1.00276736, "grad_norm": 0.8376574516296387, "learning_rate": 1.9994256475731246e-05, "loss": 4.4854, "step": 4710 }, { "epoch": 1.00279296, "grad_norm": 0.8941925764083862, "learning_rate": 1.9994229154305535e-05, "loss": 4.6954, "step": 4720 }, { "epoch": 1.00281856, "grad_norm": 1.235892415046692, "learning_rate": 1.9994201768069905e-05, "loss": 4.54, "step": 4730 }, { "epoch": 1.00284416, "grad_norm": 0.7553697228431702, "learning_rate": 1.999417431702453e-05, "loss": 4.4362, "step": 4740 }, { "epoch": 1.00286976, "grad_norm": 1.3084759712219238, "learning_rate": 1.9994146801169598e-05, "loss": 4.3568, "step": 4750 }, { "epoch": 1.00289536, "grad_norm": 0.7878323197364807, "learning_rate": 1.9994119220505276e-05, "loss": 4.5053, "step": 4760 }, { "epoch": 1.00292096, "grad_norm": 0.8006386756896973, "learning_rate": 1.9994091575031743e-05, "loss": 4.758, "step": 4770 }, { "epoch": 1.00294656, "grad_norm": 0.9987592101097107, "learning_rate": 1.9994063864749188e-05, "loss": 4.5845, "step": 4780 }, { "epoch": 1.00297216, "grad_norm": 0.9917601943016052, "learning_rate": 1.999403608965778e-05, "loss": 4.5822, "step": 4790 }, { "epoch": 1.00299776, "grad_norm": 0.7223289608955383, "learning_rate": 1.999400824975771e-05, "loss": 4.3767, "step": 4800 }, { "epoch": 1.00302336, "grad_norm": 0.8904316425323486, "learning_rate": 1.9993980345049148e-05, "loss": 4.5895, "step": 4810 }, { "epoch": 1.00304896, "grad_norm": 1.6941293478012085, "learning_rate": 1.999395237553228e-05, "loss": 4.8904, "step": 4820 }, { "epoch": 1.00307456, "grad_norm": 0.8682137131690979, "learning_rate": 1.9993924341207288e-05, "loss": 4.6393, "step": 4830 }, { "epoch": 1.00310016, "grad_norm": 0.9696330428123474, "learning_rate": 1.9993896242074357e-05, "loss": 4.7452, "step": 4840 }, { "epoch": 1.00312576, "grad_norm": 0.8014997839927673, "learning_rate": 1.9993868078133657e-05, "loss": 4.5244, "step": 4850 }, { "epoch": 1.00315136, "grad_norm": 0.998539388179779, "learning_rate": 1.9993839849385383e-05, "loss": 4.7278, "step": 4860 }, { "epoch": 1.00317696, "grad_norm": 0.8955872654914856, "learning_rate": 1.9993811555829713e-05, "loss": 4.5805, "step": 4870 }, { "epoch": 1.00320256, "grad_norm": 0.9971116781234741, "learning_rate": 1.999378319746683e-05, "loss": 4.6653, "step": 4880 }, { "epoch": 1.00322816, "grad_norm": 1.03840172290802, "learning_rate": 1.999375477429692e-05, "loss": 4.3998, "step": 4890 }, { "epoch": 1.00325376, "grad_norm": 1.0849061012268066, "learning_rate": 1.9993726286320164e-05, "loss": 4.6389, "step": 4900 }, { "epoch": 1.00327936, "grad_norm": 1.2694908380508423, "learning_rate": 1.9993697733536753e-05, "loss": 4.4542, "step": 4910 }, { "epoch": 1.00330496, "grad_norm": 1.08487868309021, "learning_rate": 1.9993669115946867e-05, "loss": 4.6797, "step": 4920 }, { "epoch": 1.00333056, "grad_norm": 1.092878818511963, "learning_rate": 1.999364043355069e-05, "loss": 4.5831, "step": 4930 }, { "epoch": 1.00335616, "grad_norm": 0.8073432445526123, "learning_rate": 1.9993611686348413e-05, "loss": 4.4793, "step": 4940 }, { "epoch": 1.00338176, "grad_norm": 0.811126708984375, "learning_rate": 1.999358287434022e-05, "loss": 4.5157, "step": 4950 }, { "epoch": 1.00340736, "grad_norm": 0.8381230235099792, "learning_rate": 1.99935539975263e-05, "loss": 4.574, "step": 4960 }, { "epoch": 1.00343296, "grad_norm": 0.9403958320617676, "learning_rate": 1.9993525055906837e-05, "loss": 4.4221, "step": 4970 }, { "epoch": 1.00345856, "grad_norm": 1.1248832941055298, "learning_rate": 1.9993496049482018e-05, "loss": 4.723, "step": 4980 }, { "epoch": 1.00348416, "grad_norm": 0.8181705474853516, "learning_rate": 1.9993466978252037e-05, "loss": 4.5942, "step": 4990 }, { "epoch": 1.00350976, "grad_norm": 0.9075210094451904, "learning_rate": 1.9993437842217072e-05, "loss": 5.0197, "step": 5000 }, { "epoch": 1.00353536, "grad_norm": 0.8368251323699951, "learning_rate": 1.9993408641377325e-05, "loss": 4.4313, "step": 5010 }, { "epoch": 1.00356096, "grad_norm": 1.1866947412490845, "learning_rate": 1.9993379375732977e-05, "loss": 4.3519, "step": 5020 }, { "epoch": 1.00358656, "grad_norm": 1.4483141899108887, "learning_rate": 1.9993350045284215e-05, "loss": 4.3549, "step": 5030 }, { "epoch": 1.00361216, "grad_norm": 3.889697313308716, "learning_rate": 1.999332065003124e-05, "loss": 4.8578, "step": 5040 }, { "epoch": 1.00363776, "grad_norm": 1.0499746799468994, "learning_rate": 1.9993291189974236e-05, "loss": 4.6308, "step": 5050 }, { "epoch": 1.00366336, "grad_norm": 0.8271541595458984, "learning_rate": 1.999326166511339e-05, "loss": 4.4909, "step": 5060 }, { "epoch": 1.00368896, "grad_norm": 1.1068551540374756, "learning_rate": 1.9993232075448902e-05, "loss": 4.8955, "step": 5070 }, { "epoch": 1.00371456, "grad_norm": 1.2383699417114258, "learning_rate": 1.999320242098096e-05, "loss": 4.3981, "step": 5080 }, { "epoch": 1.00374016, "grad_norm": 0.8991289734840393, "learning_rate": 1.9993172701709755e-05, "loss": 4.6194, "step": 5090 }, { "epoch": 1.00376576, "grad_norm": 0.9343798160552979, "learning_rate": 1.9993142917635483e-05, "loss": 4.6803, "step": 5100 }, { "epoch": 1.00379136, "grad_norm": 1.1916584968566895, "learning_rate": 1.9993113068758332e-05, "loss": 4.5235, "step": 5110 }, { "epoch": 1.00381696, "grad_norm": 0.963057279586792, "learning_rate": 1.99930831550785e-05, "loss": 4.6122, "step": 5120 }, { "epoch": 1.00384256, "grad_norm": 0.9004610776901245, "learning_rate": 1.999305317659618e-05, "loss": 4.6251, "step": 5130 }, { "epoch": 1.00386816, "grad_norm": 1.8004659414291382, "learning_rate": 1.9993023133311564e-05, "loss": 4.258, "step": 5140 }, { "epoch": 1.00389376, "grad_norm": 1.1570026874542236, "learning_rate": 1.999299302522485e-05, "loss": 4.3878, "step": 5150 }, { "epoch": 1.00391936, "grad_norm": 1.2939517498016357, "learning_rate": 1.9992962852336233e-05, "loss": 4.6654, "step": 5160 }, { "epoch": 1.00394496, "grad_norm": 1.0024744272232056, "learning_rate": 1.9992932614645905e-05, "loss": 4.538, "step": 5170 }, { "epoch": 1.00397056, "grad_norm": 2.359822988510132, "learning_rate": 1.999290231215407e-05, "loss": 4.6784, "step": 5180 }, { "epoch": 1.00399616, "grad_norm": 0.8537594676017761, "learning_rate": 1.9992871944860913e-05, "loss": 4.609, "step": 5190 }, { "epoch": 1.00402176, "grad_norm": 1.2468141317367554, "learning_rate": 1.999284151276664e-05, "loss": 4.4047, "step": 5200 }, { "epoch": 1.00404736, "grad_norm": 1.1876736879348755, "learning_rate": 1.9992811015871445e-05, "loss": 4.7953, "step": 5210 }, { "epoch": 1.00407296, "grad_norm": 0.866868257522583, "learning_rate": 1.999278045417553e-05, "loss": 4.5649, "step": 5220 }, { "epoch": 1.00409856, "grad_norm": 1.1048564910888672, "learning_rate": 1.9992749827679085e-05, "loss": 4.5746, "step": 5230 }, { "epoch": 1.00412416, "grad_norm": 1.3614780902862549, "learning_rate": 1.999271913638231e-05, "loss": 4.553, "step": 5240 }, { "epoch": 1.00414976, "grad_norm": 0.8974482417106628, "learning_rate": 1.9992688380285416e-05, "loss": 4.5887, "step": 5250 }, { "epoch": 1.00417536, "grad_norm": 0.8084264993667603, "learning_rate": 1.999265755938859e-05, "loss": 4.4177, "step": 5260 }, { "epoch": 1.00420096, "grad_norm": 0.8335790038108826, "learning_rate": 1.9992626673692036e-05, "loss": 4.6637, "step": 5270 }, { "epoch": 1.00422656, "grad_norm": 0.988978922367096, "learning_rate": 1.999259572319595e-05, "loss": 4.4288, "step": 5280 }, { "epoch": 1.00425216, "grad_norm": 1.0539895296096802, "learning_rate": 1.999256470790054e-05, "loss": 4.3513, "step": 5290 }, { "epoch": 1.00427776, "grad_norm": 1.1676182746887207, "learning_rate": 1.9992533627806003e-05, "loss": 4.512, "step": 5300 }, { "epoch": 1.00430336, "grad_norm": 0.8142640590667725, "learning_rate": 1.999250248291254e-05, "loss": 4.7388, "step": 5310 }, { "epoch": 1.00432896, "grad_norm": 0.8920853137969971, "learning_rate": 1.9992471273220353e-05, "loss": 4.3642, "step": 5320 }, { "epoch": 1.00435456, "grad_norm": 1.02957284450531, "learning_rate": 1.999243999872965e-05, "loss": 4.6511, "step": 5330 }, { "epoch": 1.00438016, "grad_norm": 0.7047339081764221, "learning_rate": 1.9992408659440625e-05, "loss": 4.6572, "step": 5340 }, { "epoch": 1.00440576, "grad_norm": 1.0498435497283936, "learning_rate": 1.9992377255353487e-05, "loss": 4.3059, "step": 5350 }, { "epoch": 1.00443136, "grad_norm": 0.706390917301178, "learning_rate": 1.999234578646844e-05, "loss": 4.1973, "step": 5360 }, { "epoch": 1.00445696, "grad_norm": 0.7760142683982849, "learning_rate": 1.9992314252785684e-05, "loss": 4.3614, "step": 5370 }, { "epoch": 1.00448256, "grad_norm": 0.9118815660476685, "learning_rate": 1.9992282654305426e-05, "loss": 4.5067, "step": 5380 }, { "epoch": 1.00450816, "grad_norm": 0.9968594312667847, "learning_rate": 1.9992250991027872e-05, "loss": 4.5142, "step": 5390 }, { "epoch": 1.00453376, "grad_norm": 0.8651778101921082, "learning_rate": 1.9992219262953223e-05, "loss": 4.4799, "step": 5400 }, { "epoch": 1.00455936, "grad_norm": 1.873427152633667, "learning_rate": 1.999218747008169e-05, "loss": 4.6526, "step": 5410 }, { "epoch": 1.00458496, "grad_norm": 0.7413127422332764, "learning_rate": 1.9992155612413476e-05, "loss": 4.5483, "step": 5420 }, { "epoch": 1.00461056, "grad_norm": 1.3293665647506714, "learning_rate": 1.9992123689948787e-05, "loss": 4.4536, "step": 5430 }, { "epoch": 1.00463616, "grad_norm": 1.1047865152359009, "learning_rate": 1.9992091702687833e-05, "loss": 4.491, "step": 5440 }, { "epoch": 1.00466176, "grad_norm": 1.2431012392044067, "learning_rate": 1.999205965063082e-05, "loss": 4.6202, "step": 5450 }, { "epoch": 1.00468736, "grad_norm": 0.9716372489929199, "learning_rate": 1.999202753377796e-05, "loss": 4.5822, "step": 5460 }, { "epoch": 1.00471296, "grad_norm": 0.933987557888031, "learning_rate": 1.999199535212945e-05, "loss": 4.3412, "step": 5470 }, { "epoch": 1.00473856, "grad_norm": 1.5045756101608276, "learning_rate": 1.9991963105685507e-05, "loss": 4.6161, "step": 5480 }, { "epoch": 1.00476416, "grad_norm": 1.0772348642349243, "learning_rate": 1.9991930794446343e-05, "loss": 4.4097, "step": 5490 }, { "epoch": 1.00478976, "grad_norm": 0.7776811122894287, "learning_rate": 1.999189841841216e-05, "loss": 4.404, "step": 5500 }, { "epoch": 1.00481536, "grad_norm": 0.9654356837272644, "learning_rate": 1.9991865977583173e-05, "loss": 4.6212, "step": 5510 }, { "epoch": 1.00484096, "grad_norm": 1.0811593532562256, "learning_rate": 1.9991833471959586e-05, "loss": 4.4082, "step": 5520 }, { "epoch": 1.00486656, "grad_norm": 1.0986762046813965, "learning_rate": 1.999180090154162e-05, "loss": 4.5513, "step": 5530 }, { "epoch": 1.00489216, "grad_norm": 1.0186855792999268, "learning_rate": 1.999176826632948e-05, "loss": 4.4069, "step": 5540 }, { "epoch": 1.00491776, "grad_norm": 1.2182389497756958, "learning_rate": 1.9991735566323375e-05, "loss": 4.4066, "step": 5550 }, { "epoch": 1.00494336, "grad_norm": 1.297755241394043, "learning_rate": 1.9991702801523523e-05, "loss": 4.8746, "step": 5560 }, { "epoch": 1.00496896, "grad_norm": 0.8209927082061768, "learning_rate": 1.9991669971930133e-05, "loss": 4.6083, "step": 5570 }, { "epoch": 1.00499456, "grad_norm": 0.9837113618850708, "learning_rate": 1.9991637077543418e-05, "loss": 4.6449, "step": 5580 }, { "epoch": 1.00502016, "grad_norm": 1.1151083707809448, "learning_rate": 1.9991604118363592e-05, "loss": 4.3076, "step": 5590 }, { "epoch": 1.00504576, "grad_norm": 0.7711336612701416, "learning_rate": 1.9991571094390872e-05, "loss": 4.4955, "step": 5600 }, { "epoch": 1.00507136, "grad_norm": 1.0519589185714722, "learning_rate": 1.9991538005625465e-05, "loss": 4.529, "step": 5610 }, { "epoch": 1.00509696, "grad_norm": 0.9757224321365356, "learning_rate": 1.999150485206759e-05, "loss": 4.2816, "step": 5620 }, { "epoch": 1.00512256, "grad_norm": 0.8334323167800903, "learning_rate": 1.9991471633717465e-05, "loss": 4.2987, "step": 5630 }, { "epoch": 1.00514816, "grad_norm": 2.1014552116394043, "learning_rate": 1.99914383505753e-05, "loss": 4.6554, "step": 5640 }, { "epoch": 1.00517376, "grad_norm": 0.9792777299880981, "learning_rate": 1.999140500264131e-05, "loss": 4.7995, "step": 5650 }, { "epoch": 1.00519936, "grad_norm": 1.6977711915969849, "learning_rate": 1.9991371589915717e-05, "loss": 4.4796, "step": 5660 }, { "epoch": 1.00522496, "grad_norm": 1.1179450750350952, "learning_rate": 1.999133811239873e-05, "loss": 4.6041, "step": 5670 }, { "epoch": 1.00525056, "grad_norm": 0.8541746139526367, "learning_rate": 1.9991304570090576e-05, "loss": 4.5046, "step": 5680 }, { "epoch": 1.00527616, "grad_norm": 1.2766104936599731, "learning_rate": 1.9991270962991467e-05, "loss": 4.4416, "step": 5690 }, { "epoch": 1.00530176, "grad_norm": 1.2780088186264038, "learning_rate": 1.9991237291101616e-05, "loss": 4.5527, "step": 5700 }, { "epoch": 1.00532736, "grad_norm": 1.0048567056655884, "learning_rate": 1.9991203554421252e-05, "loss": 4.4593, "step": 5710 }, { "epoch": 1.00535296, "grad_norm": 0.7469809651374817, "learning_rate": 1.9991169752950587e-05, "loss": 4.429, "step": 5720 }, { "epoch": 1.00537856, "grad_norm": 1.014831781387329, "learning_rate": 1.9991135886689842e-05, "loss": 4.2689, "step": 5730 }, { "epoch": 1.00540416, "grad_norm": 1.0854898691177368, "learning_rate": 1.9991101955639232e-05, "loss": 4.5638, "step": 5740 }, { "epoch": 1.00542976, "grad_norm": 1.0214128494262695, "learning_rate": 1.9991067959798988e-05, "loss": 4.433, "step": 5750 }, { "epoch": 1.00545536, "grad_norm": 1.1730411052703857, "learning_rate": 1.999103389916932e-05, "loss": 4.3145, "step": 5760 }, { "epoch": 1.00548096, "grad_norm": 1.0520044565200806, "learning_rate": 1.999099977375045e-05, "loss": 4.3603, "step": 5770 }, { "epoch": 1.00550656, "grad_norm": 1.200411081314087, "learning_rate": 1.9990965583542607e-05, "loss": 4.3937, "step": 5780 }, { "epoch": 1.00553216, "grad_norm": 1.0911498069763184, "learning_rate": 1.9990931328546e-05, "loss": 4.5355, "step": 5790 }, { "epoch": 1.00555776, "grad_norm": 1.2814102172851562, "learning_rate": 1.9990897008760866e-05, "loss": 4.4932, "step": 5800 }, { "epoch": 1.00558336, "grad_norm": 1.1820791959762573, "learning_rate": 1.9990862624187416e-05, "loss": 4.6686, "step": 5810 }, { "epoch": 1.00560896, "grad_norm": 2.530543327331543, "learning_rate": 1.999082817482588e-05, "loss": 4.4249, "step": 5820 }, { "epoch": 1.00563456, "grad_norm": 0.9983447790145874, "learning_rate": 1.9990793660676476e-05, "loss": 4.5255, "step": 5830 }, { "epoch": 1.00566016, "grad_norm": 1.169055700302124, "learning_rate": 1.999075908173943e-05, "loss": 4.5412, "step": 5840 }, { "epoch": 1.00568576, "grad_norm": 0.8479063510894775, "learning_rate": 1.999072443801497e-05, "loss": 4.3549, "step": 5850 }, { "epoch": 1.00571136, "grad_norm": 0.7902194857597351, "learning_rate": 1.9990689729503316e-05, "loss": 4.2892, "step": 5860 }, { "epoch": 1.00573696, "grad_norm": 1.0000401735305786, "learning_rate": 1.9990654956204692e-05, "loss": 4.2669, "step": 5870 }, { "epoch": 1.00576256, "grad_norm": 1.211412787437439, "learning_rate": 1.9990620118119327e-05, "loss": 4.3258, "step": 5880 }, { "epoch": 1.00578816, "grad_norm": 0.793580174446106, "learning_rate": 1.9990585215247447e-05, "loss": 4.7652, "step": 5890 }, { "epoch": 1.00581376, "grad_norm": 0.697577714920044, "learning_rate": 1.9990550247589273e-05, "loss": 4.6448, "step": 5900 }, { "epoch": 1.00583936, "grad_norm": 0.878109335899353, "learning_rate": 1.999051521514504e-05, "loss": 4.5936, "step": 5910 }, { "epoch": 1.00586496, "grad_norm": 0.9900004267692566, "learning_rate": 1.999048011791497e-05, "loss": 4.5052, "step": 5920 }, { "epoch": 1.00589056, "grad_norm": 1.6249399185180664, "learning_rate": 1.999044495589929e-05, "loss": 4.1319, "step": 5930 }, { "epoch": 1.00591616, "grad_norm": 1.2339216470718384, "learning_rate": 1.999040972909823e-05, "loss": 4.914, "step": 5940 }, { "epoch": 1.00594176, "grad_norm": 1.219236135482788, "learning_rate": 1.9990374437512023e-05, "loss": 4.7992, "step": 5950 }, { "epoch": 1.00596736, "grad_norm": 0.9855391383171082, "learning_rate": 1.999033908114089e-05, "loss": 4.7684, "step": 5960 }, { "epoch": 1.00599296, "grad_norm": 1.3320118188858032, "learning_rate": 1.999030365998506e-05, "loss": 4.5363, "step": 5970 }, { "epoch": 1.00601856, "grad_norm": 0.9950410723686218, "learning_rate": 1.9990268174044765e-05, "loss": 4.738, "step": 5980 }, { "epoch": 1.00604416, "grad_norm": 1.1994644403457642, "learning_rate": 1.999023262332024e-05, "loss": 4.5051, "step": 5990 }, { "epoch": 1.00606976, "grad_norm": 1.0698190927505493, "learning_rate": 1.9990197007811713e-05, "loss": 4.5067, "step": 6000 }, { "epoch": 1.00609536, "grad_norm": 2.2934226989746094, "learning_rate": 1.9990161327519408e-05, "loss": 4.4329, "step": 6010 }, { "epoch": 1.00612096, "grad_norm": 0.8799678087234497, "learning_rate": 1.9990125582443567e-05, "loss": 4.6102, "step": 6020 }, { "epoch": 1.00614656, "grad_norm": 1.051976203918457, "learning_rate": 1.9990089772584413e-05, "loss": 4.4438, "step": 6030 }, { "epoch": 1.00617216, "grad_norm": 0.994381844997406, "learning_rate": 1.9990053897942184e-05, "loss": 4.5601, "step": 6040 }, { "epoch": 1.00619776, "grad_norm": 0.8386251926422119, "learning_rate": 1.999001795851711e-05, "loss": 4.3895, "step": 6050 }, { "epoch": 1.00622336, "grad_norm": 0.7214205265045166, "learning_rate": 1.9989981954309424e-05, "loss": 4.4852, "step": 6060 }, { "epoch": 1.00624896, "grad_norm": 0.9492271542549133, "learning_rate": 1.998994588531936e-05, "loss": 4.3191, "step": 6070 }, { "epoch": 1.00627456, "grad_norm": 0.8908170461654663, "learning_rate": 1.9989909751547155e-05, "loss": 4.4577, "step": 6080 }, { "epoch": 1.00630016, "grad_norm": 0.8931869268417358, "learning_rate": 1.9989873552993035e-05, "loss": 4.6115, "step": 6090 }, { "epoch": 1.00632576, "grad_norm": 0.7954609990119934, "learning_rate": 1.9989837289657242e-05, "loss": 4.5279, "step": 6100 }, { "epoch": 1.00635136, "grad_norm": 1.1079214811325073, "learning_rate": 1.998980096154001e-05, "loss": 4.6024, "step": 6110 }, { "epoch": 1.00637696, "grad_norm": 0.8716193437576294, "learning_rate": 1.9989764568641574e-05, "loss": 4.3811, "step": 6120 }, { "epoch": 1.00640256, "grad_norm": 0.8351333141326904, "learning_rate": 1.998972811096217e-05, "loss": 4.479, "step": 6130 }, { "epoch": 1.00642816, "grad_norm": 0.929793655872345, "learning_rate": 1.9989691588502034e-05, "loss": 4.3169, "step": 6140 }, { "epoch": 1.00645376, "grad_norm": 0.9576134085655212, "learning_rate": 1.9989655001261404e-05, "loss": 4.1931, "step": 6150 }, { "epoch": 1.00647936, "grad_norm": 0.8567198514938354, "learning_rate": 1.9989618349240515e-05, "loss": 4.4652, "step": 6160 }, { "epoch": 1.00650496, "grad_norm": 0.9067171216011047, "learning_rate": 1.9989581632439604e-05, "loss": 4.2993, "step": 6170 }, { "epoch": 1.00653056, "grad_norm": 1.3253611326217651, "learning_rate": 1.998954485085891e-05, "loss": 4.3886, "step": 6180 }, { "epoch": 1.00655616, "grad_norm": 0.7077346444129944, "learning_rate": 1.9989508004498678e-05, "loss": 4.204, "step": 6190 }, { "epoch": 1.00658176, "grad_norm": 0.9722407460212708, "learning_rate": 1.9989471093359138e-05, "loss": 4.6174, "step": 6200 }, { "epoch": 1.00660736, "grad_norm": 0.751762330532074, "learning_rate": 1.9989434117440534e-05, "loss": 4.1477, "step": 6210 }, { "epoch": 1.00663296, "grad_norm": 0.7841206789016724, "learning_rate": 1.9989397076743103e-05, "loss": 4.4866, "step": 6220 }, { "epoch": 1.00665856, "grad_norm": 1.0946553945541382, "learning_rate": 1.9989359971267087e-05, "loss": 4.6363, "step": 6230 }, { "epoch": 1.00668416, "grad_norm": 1.87294340133667, "learning_rate": 1.9989322801012728e-05, "loss": 4.2527, "step": 6240 }, { "epoch": 1.00670976, "grad_norm": 0.9677264094352722, "learning_rate": 1.9989285565980264e-05, "loss": 4.4953, "step": 6250 }, { "epoch": 1.00673536, "grad_norm": 0.9817032814025879, "learning_rate": 1.998924826616994e-05, "loss": 4.3668, "step": 6260 }, { "epoch": 1.00676096, "grad_norm": 0.8607648015022278, "learning_rate": 1.9989210901581993e-05, "loss": 4.4211, "step": 6270 }, { "epoch": 1.00678656, "grad_norm": 1.084511637687683, "learning_rate": 1.998917347221667e-05, "loss": 4.8297, "step": 6280 }, { "epoch": 1.00681216, "grad_norm": 1.0060220956802368, "learning_rate": 1.9989135978074213e-05, "loss": 4.2632, "step": 6290 }, { "epoch": 1.00683776, "grad_norm": 1.342543125152588, "learning_rate": 1.9989098419154863e-05, "loss": 4.4179, "step": 6300 }, { "epoch": 1.00686336, "grad_norm": 0.7813605666160583, "learning_rate": 1.9989060795458865e-05, "loss": 4.6169, "step": 6310 }, { "epoch": 1.00688896, "grad_norm": 1.2504239082336426, "learning_rate": 1.998902310698646e-05, "loss": 4.3764, "step": 6320 }, { "epoch": 1.00691456, "grad_norm": 0.9180771708488464, "learning_rate": 1.9988985353737896e-05, "loss": 4.7573, "step": 6330 }, { "epoch": 1.00694016, "grad_norm": 0.9580790996551514, "learning_rate": 1.9988947535713417e-05, "loss": 4.4672, "step": 6340 }, { "epoch": 1.00696576, "grad_norm": 1.2181296348571777, "learning_rate": 1.998890965291327e-05, "loss": 4.5422, "step": 6350 }, { "epoch": 1.00699136, "grad_norm": 0.7607697248458862, "learning_rate": 1.9988871705337696e-05, "loss": 4.5345, "step": 6360 }, { "epoch": 1.00701696, "grad_norm": 0.9992573857307434, "learning_rate": 1.9988833692986946e-05, "loss": 4.4763, "step": 6370 }, { "epoch": 1.00704256, "grad_norm": 0.9015842080116272, "learning_rate": 1.998879561586126e-05, "loss": 4.4717, "step": 6380 }, { "epoch": 1.00706816, "grad_norm": 0.9455610513687134, "learning_rate": 1.9988757473960894e-05, "loss": 4.5244, "step": 6390 }, { "epoch": 1.00709376, "grad_norm": 0.9999937415122986, "learning_rate": 1.9988719267286087e-05, "loss": 4.636, "step": 6400 }, { "epoch": 1.00711936, "grad_norm": 1.2374941110610962, "learning_rate": 1.9988680995837093e-05, "loss": 4.5165, "step": 6410 }, { "epoch": 1.00714496, "grad_norm": 0.7801550030708313, "learning_rate": 1.9988642659614153e-05, "loss": 4.6069, "step": 6420 }, { "epoch": 1.00717056, "grad_norm": 1.1897332668304443, "learning_rate": 1.9988604258617527e-05, "loss": 4.3243, "step": 6430 }, { "epoch": 1.00719616, "grad_norm": 2.699148178100586, "learning_rate": 1.998856579284745e-05, "loss": 4.3632, "step": 6440 }, { "epoch": 1.00722176, "grad_norm": 1.4052400588989258, "learning_rate": 1.9988527262304183e-05, "loss": 4.6799, "step": 6450 }, { "epoch": 1.00724736, "grad_norm": 0.8730829358100891, "learning_rate": 1.998848866698797e-05, "loss": 4.266, "step": 6460 }, { "epoch": 1.00727296, "grad_norm": 0.8871892094612122, "learning_rate": 1.9988450006899063e-05, "loss": 4.3097, "step": 6470 }, { "epoch": 1.00729856, "grad_norm": 1.213413119316101, "learning_rate": 1.998841128203771e-05, "loss": 4.3552, "step": 6480 }, { "epoch": 1.00732416, "grad_norm": 0.8081543445587158, "learning_rate": 1.998837249240417e-05, "loss": 4.5983, "step": 6490 }, { "epoch": 1.00734976, "grad_norm": 2.23248028755188, "learning_rate": 1.9988333637998684e-05, "loss": 5.2131, "step": 6500 }, { "epoch": 1.00737536, "grad_norm": 1.0069133043289185, "learning_rate": 1.998829471882151e-05, "loss": 4.5067, "step": 6510 }, { "epoch": 1.00740096, "grad_norm": 1.0242412090301514, "learning_rate": 1.99882557348729e-05, "loss": 4.5999, "step": 6520 }, { "epoch": 1.00742656, "grad_norm": 1.2586722373962402, "learning_rate": 1.9988216686153103e-05, "loss": 4.568, "step": 6530 }, { "epoch": 1.00745216, "grad_norm": 0.671323299407959, "learning_rate": 1.998817757266238e-05, "loss": 4.4175, "step": 6540 }, { "epoch": 1.00747776, "grad_norm": 1.2012953758239746, "learning_rate": 1.9988138394400976e-05, "loss": 4.3005, "step": 6550 }, { "epoch": 1.00750336, "grad_norm": 0.6713989973068237, "learning_rate": 1.998809915136915e-05, "loss": 4.3697, "step": 6560 }, { "epoch": 1.00752896, "grad_norm": 1.0421216487884521, "learning_rate": 1.9988059843567157e-05, "loss": 4.4536, "step": 6570 }, { "epoch": 1.00755456, "grad_norm": 1.2932026386260986, "learning_rate": 1.9988020470995253e-05, "loss": 4.7417, "step": 6580 }, { "epoch": 1.00758016, "grad_norm": 0.7396079897880554, "learning_rate": 1.9987981033653687e-05, "loss": 4.4932, "step": 6590 }, { "epoch": 1.00760576, "grad_norm": 0.840097963809967, "learning_rate": 1.998794153154272e-05, "loss": 4.3187, "step": 6600 }, { "epoch": 1.00763136, "grad_norm": 0.6930157542228699, "learning_rate": 1.9987901964662603e-05, "loss": 4.3214, "step": 6610 }, { "epoch": 1.00765696, "grad_norm": 1.166900634765625, "learning_rate": 1.99878623330136e-05, "loss": 4.5006, "step": 6620 }, { "epoch": 1.00768256, "grad_norm": 2.06022047996521, "learning_rate": 1.9987822636595965e-05, "loss": 4.2758, "step": 6630 }, { "epoch": 1.00770816, "grad_norm": 0.8305878043174744, "learning_rate": 1.998778287540995e-05, "loss": 4.281, "step": 6640 }, { "epoch": 1.00773376, "grad_norm": 0.943668007850647, "learning_rate": 1.998774304945582e-05, "loss": 4.4526, "step": 6650 }, { "epoch": 1.00775936, "grad_norm": 0.7768296003341675, "learning_rate": 1.9987703158733832e-05, "loss": 4.3985, "step": 6660 }, { "epoch": 1.00778496, "grad_norm": 2.50935959815979, "learning_rate": 1.998766320324424e-05, "loss": 5.4989, "step": 6670 }, { "epoch": 1.00781056, "grad_norm": 1.1615458726882935, "learning_rate": 1.9987623182987314e-05, "loss": 4.654, "step": 6680 }, { "epoch": 1.00783616, "grad_norm": 0.7816067934036255, "learning_rate": 1.9987583097963302e-05, "loss": 4.3275, "step": 6690 }, { "epoch": 1.00786176, "grad_norm": 0.9985303282737732, "learning_rate": 1.9987542948172465e-05, "loss": 4.3498, "step": 6700 }, { "epoch": 1.00788736, "grad_norm": 0.8550955057144165, "learning_rate": 1.9987502733615066e-05, "loss": 4.2293, "step": 6710 }, { "epoch": 1.00791296, "grad_norm": 1.0033475160598755, "learning_rate": 1.998746245429137e-05, "loss": 4.4409, "step": 6720 }, { "epoch": 1.00793856, "grad_norm": 1.0471199750900269, "learning_rate": 1.9987422110201634e-05, "loss": 4.4085, "step": 6730 }, { "epoch": 1.00796416, "grad_norm": 0.8463475108146667, "learning_rate": 1.998738170134612e-05, "loss": 4.4131, "step": 6740 }, { "epoch": 1.00798976, "grad_norm": 0.8140661716461182, "learning_rate": 1.998734122772509e-05, "loss": 4.3659, "step": 6750 }, { "epoch": 1.00801536, "grad_norm": 0.9503121972084045, "learning_rate": 1.99873006893388e-05, "loss": 4.3385, "step": 6760 }, { "epoch": 1.00804096, "grad_norm": 1.234089970588684, "learning_rate": 1.998726008618753e-05, "loss": 4.4841, "step": 6770 }, { "epoch": 1.00806656, "grad_norm": 1.096703290939331, "learning_rate": 1.9987219418271526e-05, "loss": 4.28, "step": 6780 }, { "epoch": 1.00809216, "grad_norm": 1.0292378664016724, "learning_rate": 1.9987178685591057e-05, "loss": 4.2871, "step": 6790 }, { "epoch": 1.00811776, "grad_norm": 1.1790441274642944, "learning_rate": 1.9987137888146393e-05, "loss": 4.3688, "step": 6800 }, { "epoch": 1.00814336, "grad_norm": 0.9309192895889282, "learning_rate": 1.998709702593779e-05, "loss": 4.3454, "step": 6810 }, { "epoch": 1.00816896, "grad_norm": 1.0261359214782715, "learning_rate": 1.9987056098965524e-05, "loss": 4.5807, "step": 6820 }, { "epoch": 1.00819456, "grad_norm": 0.8403862118721008, "learning_rate": 1.9987015107229846e-05, "loss": 4.5976, "step": 6830 }, { "epoch": 1.00822016, "grad_norm": 0.9921002984046936, "learning_rate": 1.9986974050731032e-05, "loss": 4.3672, "step": 6840 }, { "epoch": 1.00824576, "grad_norm": 1.0381537675857544, "learning_rate": 1.9986932929469346e-05, "loss": 4.4732, "step": 6850 }, { "epoch": 1.00827136, "grad_norm": 1.0835886001586914, "learning_rate": 1.9986891743445055e-05, "loss": 4.556, "step": 6860 }, { "epoch": 1.00829696, "grad_norm": 1.023321270942688, "learning_rate": 1.998685049265842e-05, "loss": 4.2985, "step": 6870 }, { "epoch": 1.00832256, "grad_norm": 1.6784833669662476, "learning_rate": 1.998680917710972e-05, "loss": 3.9281, "step": 6880 }, { "epoch": 1.00834816, "grad_norm": 1.040408730506897, "learning_rate": 1.9986767796799214e-05, "loss": 4.6948, "step": 6890 }, { "epoch": 1.00837376, "grad_norm": 1.027689814567566, "learning_rate": 1.9986726351727173e-05, "loss": 4.3723, "step": 6900 }, { "epoch": 1.00839936, "grad_norm": 1.5035030841827393, "learning_rate": 1.9986684841893865e-05, "loss": 4.531, "step": 6910 }, { "epoch": 1.00842496, "grad_norm": 0.8746511340141296, "learning_rate": 1.9986643267299563e-05, "loss": 4.1646, "step": 6920 }, { "epoch": 1.00845056, "grad_norm": 1.1881660223007202, "learning_rate": 1.998660162794453e-05, "loss": 4.0619, "step": 6930 }, { "epoch": 1.00847616, "grad_norm": 1.4814471006393433, "learning_rate": 1.9986559923829044e-05, "loss": 4.4414, "step": 6940 }, { "epoch": 1.00850176, "grad_norm": 1.362776756286621, "learning_rate": 1.9986518154953366e-05, "loss": 4.3263, "step": 6950 }, { "epoch": 1.00852736, "grad_norm": 1.3475852012634277, "learning_rate": 1.9986476321317775e-05, "loss": 4.3063, "step": 6960 }, { "epoch": 1.00855296, "grad_norm": 1.3463118076324463, "learning_rate": 1.9986434422922536e-05, "loss": 4.5235, "step": 6970 }, { "epoch": 1.00857856, "grad_norm": 0.9577042460441589, "learning_rate": 1.9986392459767926e-05, "loss": 4.3819, "step": 6980 }, { "epoch": 1.00860416, "grad_norm": 0.7713137269020081, "learning_rate": 1.9986350431854218e-05, "loss": 4.435, "step": 6990 }, { "epoch": 1.00862976, "grad_norm": 1.0285364389419556, "learning_rate": 1.9986308339181676e-05, "loss": 4.2959, "step": 7000 }, { "epoch": 1.0086553600000001, "grad_norm": 0.8230820298194885, "learning_rate": 1.9986266181750583e-05, "loss": 4.2962, "step": 7010 }, { "epoch": 1.00868096, "grad_norm": 0.8931657075881958, "learning_rate": 1.9986223959561205e-05, "loss": 4.5507, "step": 7020 }, { "epoch": 1.00870656, "grad_norm": 1.030134677886963, "learning_rate": 1.9986181672613817e-05, "loss": 4.2964, "step": 7030 }, { "epoch": 1.00873216, "grad_norm": 0.9255492687225342, "learning_rate": 1.9986139320908696e-05, "loss": 4.7301, "step": 7040 }, { "epoch": 1.00875776, "grad_norm": 0.9143009185791016, "learning_rate": 1.998609690444612e-05, "loss": 4.3271, "step": 7050 }, { "epoch": 1.00878336, "grad_norm": 0.90647953748703, "learning_rate": 1.9986054423226354e-05, "loss": 4.3803, "step": 7060 }, { "epoch": 1.00880896, "grad_norm": 0.9590080976486206, "learning_rate": 1.9986011877249684e-05, "loss": 4.5566, "step": 7070 }, { "epoch": 1.00883456, "grad_norm": 1.3221570253372192, "learning_rate": 1.998596926651638e-05, "loss": 4.4343, "step": 7080 }, { "epoch": 1.00886016, "grad_norm": 1.077786922454834, "learning_rate": 1.9985926591026718e-05, "loss": 4.469, "step": 7090 }, { "epoch": 1.00888576, "grad_norm": 1.0441595315933228, "learning_rate": 1.9985883850780975e-05, "loss": 4.3383, "step": 7100 }, { "epoch": 1.00891136, "grad_norm": 0.735280454158783, "learning_rate": 1.9985841045779427e-05, "loss": 4.3862, "step": 7110 }, { "epoch": 1.00893696, "grad_norm": 0.8314303755760193, "learning_rate": 1.9985798176022356e-05, "loss": 4.4411, "step": 7120 }, { "epoch": 1.00896256, "grad_norm": 0.9541341066360474, "learning_rate": 1.998575524151004e-05, "loss": 4.3825, "step": 7130 }, { "epoch": 1.00898816, "grad_norm": 0.872765064239502, "learning_rate": 1.9985712242242753e-05, "loss": 4.2537, "step": 7140 }, { "epoch": 1.00901376, "grad_norm": 1.3491343259811401, "learning_rate": 1.9985669178220773e-05, "loss": 4.1164, "step": 7150 }, { "epoch": 1.00903936, "grad_norm": 1.9295649528503418, "learning_rate": 1.9985626049444388e-05, "loss": 4.2561, "step": 7160 }, { "epoch": 1.00906496, "grad_norm": 0.9115093946456909, "learning_rate": 1.998558285591387e-05, "loss": 4.4996, "step": 7170 }, { "epoch": 1.00909056, "grad_norm": 0.9787130951881409, "learning_rate": 1.99855395976295e-05, "loss": 4.1878, "step": 7180 }, { "epoch": 1.00911616, "grad_norm": 0.7562068104743958, "learning_rate": 1.998549627459156e-05, "loss": 4.3348, "step": 7190 }, { "epoch": 1.00914176, "grad_norm": 0.8864599466323853, "learning_rate": 1.998545288680033e-05, "loss": 4.5566, "step": 7200 }, { "epoch": 1.00916736, "grad_norm": 0.8424506187438965, "learning_rate": 1.9985409434256094e-05, "loss": 4.4996, "step": 7210 }, { "epoch": 1.00919296, "grad_norm": 1.0961555242538452, "learning_rate": 1.998536591695913e-05, "loss": 4.5485, "step": 7220 }, { "epoch": 1.00921856, "grad_norm": 0.7525200247764587, "learning_rate": 1.998532233490972e-05, "loss": 4.4706, "step": 7230 }, { "epoch": 1.00924416, "grad_norm": 1.1241905689239502, "learning_rate": 1.9985278688108153e-05, "loss": 4.3878, "step": 7240 }, { "epoch": 1.00926976, "grad_norm": 1.0862901210784912, "learning_rate": 1.9985234976554703e-05, "loss": 4.3726, "step": 7250 }, { "epoch": 2.00000512, "grad_norm": 1.3648544549942017, "learning_rate": 1.998519120024966e-05, "loss": 4.7453, "step": 7260 }, { "epoch": 2.00003072, "grad_norm": 1.0640249252319336, "learning_rate": 1.9985147359193308e-05, "loss": 4.457, "step": 7270 }, { "epoch": 2.00005632, "grad_norm": 0.9620466828346252, "learning_rate": 1.9985103453385926e-05, "loss": 4.0173, "step": 7280 }, { "epoch": 2.00008192, "grad_norm": 0.7976989150047302, "learning_rate": 1.99850594828278e-05, "loss": 4.0512, "step": 7290 }, { "epoch": 2.00010752, "grad_norm": 0.870926558971405, "learning_rate": 1.998501544751922e-05, "loss": 4.2253, "step": 7300 }, { "epoch": 2.00013312, "grad_norm": 1.0115814208984375, "learning_rate": 1.998497134746047e-05, "loss": 4.5343, "step": 7310 }, { "epoch": 2.00015872, "grad_norm": 1.1059685945510864, "learning_rate": 1.9984927182651833e-05, "loss": 4.1634, "step": 7320 }, { "epoch": 2.00018432, "grad_norm": 1.3716661930084229, "learning_rate": 1.9984882953093595e-05, "loss": 4.708, "step": 7330 }, { "epoch": 2.00020992, "grad_norm": 1.3265491724014282, "learning_rate": 1.9984838658786044e-05, "loss": 4.315, "step": 7340 }, { "epoch": 2.00023552, "grad_norm": 0.863372802734375, "learning_rate": 1.9984794299729472e-05, "loss": 4.4899, "step": 7350 }, { "epoch": 2.00026112, "grad_norm": 1.0331696271896362, "learning_rate": 1.998474987592416e-05, "loss": 4.4835, "step": 7360 }, { "epoch": 2.00028672, "grad_norm": 0.9645871520042419, "learning_rate": 1.99847053873704e-05, "loss": 4.0963, "step": 7370 }, { "epoch": 2.00031232, "grad_norm": 1.0477344989776611, "learning_rate": 1.998466083406848e-05, "loss": 3.8145, "step": 7380 }, { "epoch": 2.00033792, "grad_norm": 0.8219043016433716, "learning_rate": 1.9984616216018683e-05, "loss": 3.9677, "step": 7390 }, { "epoch": 2.00036352, "grad_norm": 1.0276628732681274, "learning_rate": 1.998457153322131e-05, "loss": 4.0779, "step": 7400 }, { "epoch": 2.00038912, "grad_norm": 1.3640694618225098, "learning_rate": 1.9984526785676638e-05, "loss": 4.0452, "step": 7410 }, { "epoch": 2.00041472, "grad_norm": 0.9110736846923828, "learning_rate": 1.9984481973384967e-05, "loss": 4.2445, "step": 7420 }, { "epoch": 2.00044032, "grad_norm": 0.8008779883384705, "learning_rate": 1.998443709634658e-05, "loss": 4.215, "step": 7430 }, { "epoch": 2.00046592, "grad_norm": 1.087391972541809, "learning_rate": 1.9984392154561777e-05, "loss": 4.3337, "step": 7440 }, { "epoch": 2.00049152, "grad_norm": 1.1045631170272827, "learning_rate": 1.9984347148030845e-05, "loss": 4.2123, "step": 7450 }, { "epoch": 2.00051712, "grad_norm": 1.4872243404388428, "learning_rate": 1.9984302076754073e-05, "loss": 4.1675, "step": 7460 }, { "epoch": 2.00054272, "grad_norm": 1.1923632621765137, "learning_rate": 1.9984256940731756e-05, "loss": 4.2463, "step": 7470 }, { "epoch": 2.00056832, "grad_norm": 0.969049334526062, "learning_rate": 1.998421173996419e-05, "loss": 3.7535, "step": 7480 }, { "epoch": 2.00059392, "grad_norm": 0.8816612362861633, "learning_rate": 1.9984166474451656e-05, "loss": 4.189, "step": 7490 }, { "epoch": 2.00061952, "grad_norm": 0.9043347239494324, "learning_rate": 1.9984121144194462e-05, "loss": 4.0122, "step": 7500 }, { "epoch": 2.00064512, "grad_norm": 0.9656010866165161, "learning_rate": 1.9984075749192892e-05, "loss": 4.0098, "step": 7510 }, { "epoch": 2.00067072, "grad_norm": 0.9335502982139587, "learning_rate": 1.9984030289447247e-05, "loss": 4.1603, "step": 7520 }, { "epoch": 2.00069632, "grad_norm": 1.1019089221954346, "learning_rate": 1.9983984764957816e-05, "loss": 4.5195, "step": 7530 }, { "epoch": 2.00072192, "grad_norm": 1.0434668064117432, "learning_rate": 1.9983939175724903e-05, "loss": 4.362, "step": 7540 }, { "epoch": 2.00074752, "grad_norm": 0.931082010269165, "learning_rate": 1.9983893521748793e-05, "loss": 4.1645, "step": 7550 }, { "epoch": 2.00077312, "grad_norm": 1.0240827798843384, "learning_rate": 1.9983847803029787e-05, "loss": 4.0746, "step": 7560 }, { "epoch": 2.00079872, "grad_norm": 0.9319478869438171, "learning_rate": 1.9983802019568184e-05, "loss": 4.4649, "step": 7570 }, { "epoch": 2.00082432, "grad_norm": 0.7853296399116516, "learning_rate": 1.9983756171364275e-05, "loss": 4.1593, "step": 7580 }, { "epoch": 2.00084992, "grad_norm": 0.8263872861862183, "learning_rate": 1.9983710258418365e-05, "loss": 3.9873, "step": 7590 }, { "epoch": 2.00087552, "grad_norm": 0.8079833388328552, "learning_rate": 1.9983664280730742e-05, "loss": 4.0355, "step": 7600 }, { "epoch": 2.00090112, "grad_norm": 1.050952672958374, "learning_rate": 1.9983618238301714e-05, "loss": 4.2954, "step": 7610 }, { "epoch": 2.00092672, "grad_norm": 1.0942180156707764, "learning_rate": 1.9983572131131576e-05, "loss": 4.0482, "step": 7620 }, { "epoch": 2.00095232, "grad_norm": 0.9671728610992432, "learning_rate": 1.9983525959220622e-05, "loss": 4.1485, "step": 7630 }, { "epoch": 2.00097792, "grad_norm": 0.9232612252235413, "learning_rate": 1.9983479722569155e-05, "loss": 4.216, "step": 7640 }, { "epoch": 2.00100352, "grad_norm": 1.0362857580184937, "learning_rate": 1.998343342117748e-05, "loss": 4.2076, "step": 7650 }, { "epoch": 2.00102912, "grad_norm": 0.9159532189369202, "learning_rate": 1.9983387055045886e-05, "loss": 4.1166, "step": 7660 }, { "epoch": 2.00105472, "grad_norm": 0.7133485078811646, "learning_rate": 1.9983340624174684e-05, "loss": 3.9325, "step": 7670 }, { "epoch": 2.00108032, "grad_norm": 1.0674208402633667, "learning_rate": 1.998329412856417e-05, "loss": 3.9857, "step": 7680 }, { "epoch": 2.00110592, "grad_norm": 1.1789263486862183, "learning_rate": 1.998324756821465e-05, "loss": 4.0304, "step": 7690 }, { "epoch": 2.00113152, "grad_norm": 1.2212523221969604, "learning_rate": 1.998320094312642e-05, "loss": 4.0876, "step": 7700 }, { "epoch": 2.00115712, "grad_norm": 0.7125314474105835, "learning_rate": 1.9983154253299788e-05, "loss": 3.9799, "step": 7710 }, { "epoch": 2.00118272, "grad_norm": 0.6893883347511292, "learning_rate": 1.9983107498735054e-05, "loss": 4.1433, "step": 7720 }, { "epoch": 2.00120832, "grad_norm": 0.8026241064071655, "learning_rate": 1.9983060679432516e-05, "loss": 4.0578, "step": 7730 }, { "epoch": 2.00123392, "grad_norm": 1.41885507106781, "learning_rate": 1.9983013795392486e-05, "loss": 4.0257, "step": 7740 }, { "epoch": 2.00125952, "grad_norm": 0.9326103329658508, "learning_rate": 1.9982966846615264e-05, "loss": 4.1761, "step": 7750 }, { "epoch": 2.00128512, "grad_norm": 1.8622570037841797, "learning_rate": 1.9982919833101156e-05, "loss": 4.1342, "step": 7760 }, { "epoch": 2.00131072, "grad_norm": 0.7854796051979065, "learning_rate": 1.9982872754850465e-05, "loss": 4.1465, "step": 7770 }, { "epoch": 2.00133632, "grad_norm": 0.720877468585968, "learning_rate": 1.9982825611863495e-05, "loss": 4.1071, "step": 7780 }, { "epoch": 2.00136192, "grad_norm": 1.0500295162200928, "learning_rate": 1.998277840414056e-05, "loss": 4.2828, "step": 7790 }, { "epoch": 2.00138752, "grad_norm": 1.298775553703308, "learning_rate": 1.9982731131681957e-05, "loss": 4.0031, "step": 7800 }, { "epoch": 2.00141312, "grad_norm": 5.724785804748535, "learning_rate": 1.9982683794487995e-05, "loss": 4.2647, "step": 7810 }, { "epoch": 2.00143872, "grad_norm": 0.8756346106529236, "learning_rate": 1.9982636392558984e-05, "loss": 4.1912, "step": 7820 }, { "epoch": 2.00146432, "grad_norm": 0.9736866354942322, "learning_rate": 1.9982588925895226e-05, "loss": 4.4145, "step": 7830 }, { "epoch": 2.00148992, "grad_norm": 1.0966382026672363, "learning_rate": 1.9982541394497033e-05, "loss": 4.3915, "step": 7840 }, { "epoch": 2.00151552, "grad_norm": 0.9307703971862793, "learning_rate": 1.9982493798364712e-05, "loss": 4.0866, "step": 7850 }, { "epoch": 2.00154112, "grad_norm": 0.8000803589820862, "learning_rate": 1.998244613749857e-05, "loss": 4.093, "step": 7860 }, { "epoch": 2.00156672, "grad_norm": 0.7607565522193909, "learning_rate": 1.998239841189892e-05, "loss": 4.4898, "step": 7870 }, { "epoch": 2.00159232, "grad_norm": 1.2087078094482422, "learning_rate": 1.9982350621566072e-05, "loss": 4.3911, "step": 7880 }, { "epoch": 2.00161792, "grad_norm": 1.3394256830215454, "learning_rate": 1.9982302766500326e-05, "loss": 4.4973, "step": 7890 }, { "epoch": 2.00164352, "grad_norm": 1.0659620761871338, "learning_rate": 1.9982254846702006e-05, "loss": 4.1719, "step": 7900 }, { "epoch": 2.00166912, "grad_norm": 0.9185815453529358, "learning_rate": 1.9982206862171414e-05, "loss": 4.3763, "step": 7910 }, { "epoch": 2.00169472, "grad_norm": 0.742888867855072, "learning_rate": 1.9982158812908865e-05, "loss": 4.4476, "step": 7920 }, { "epoch": 2.00172032, "grad_norm": 1.512462854385376, "learning_rate": 1.9982110698914668e-05, "loss": 4.4777, "step": 7930 }, { "epoch": 2.00174592, "grad_norm": 1.1700961589813232, "learning_rate": 1.9982062520189136e-05, "loss": 4.3577, "step": 7940 }, { "epoch": 2.00177152, "grad_norm": 1.2580339908599854, "learning_rate": 1.998201427673258e-05, "loss": 4.4605, "step": 7950 }, { "epoch": 2.00179712, "grad_norm": 1.2516535520553589, "learning_rate": 1.9981965968545316e-05, "loss": 4.6982, "step": 7960 }, { "epoch": 2.00182272, "grad_norm": 0.9092661142349243, "learning_rate": 1.9981917595627655e-05, "loss": 4.2985, "step": 7970 }, { "epoch": 2.00184832, "grad_norm": 0.7993289232254028, "learning_rate": 1.998186915797991e-05, "loss": 4.1573, "step": 7980 }, { "epoch": 2.00187392, "grad_norm": 1.4319984912872314, "learning_rate": 1.99818206556024e-05, "loss": 4.2278, "step": 7990 }, { "epoch": 2.00189952, "grad_norm": 1.075914978981018, "learning_rate": 1.9981772088495433e-05, "loss": 4.1816, "step": 8000 }, { "epoch": 2.00192512, "grad_norm": 0.9132004976272583, "learning_rate": 1.9981723456659326e-05, "loss": 3.9893, "step": 8010 }, { "epoch": 2.00195072, "grad_norm": 1.0604510307312012, "learning_rate": 1.9981674760094395e-05, "loss": 4.1469, "step": 8020 }, { "epoch": 2.00197632, "grad_norm": 0.8625972867012024, "learning_rate": 1.998162599880096e-05, "loss": 4.2264, "step": 8030 }, { "epoch": 2.00200192, "grad_norm": 0.961938738822937, "learning_rate": 1.998157717277933e-05, "loss": 4.4342, "step": 8040 }, { "epoch": 2.00202752, "grad_norm": 0.6944153904914856, "learning_rate": 1.9981528282029826e-05, "loss": 4.0052, "step": 8050 }, { "epoch": 2.00205312, "grad_norm": 1.0147637128829956, "learning_rate": 1.9981479326552764e-05, "loss": 4.0567, "step": 8060 }, { "epoch": 2.00207872, "grad_norm": 1.2505512237548828, "learning_rate": 1.9981430306348463e-05, "loss": 4.414, "step": 8070 }, { "epoch": 2.00210432, "grad_norm": 0.8136240839958191, "learning_rate": 1.9981381221417234e-05, "loss": 4.2813, "step": 8080 }, { "epoch": 2.00212992, "grad_norm": 1.6670968532562256, "learning_rate": 1.9981332071759404e-05, "loss": 4.263, "step": 8090 }, { "epoch": 2.00215552, "grad_norm": 1.1315457820892334, "learning_rate": 1.998128285737529e-05, "loss": 4.238, "step": 8100 }, { "epoch": 2.00218112, "grad_norm": 0.9976080656051636, "learning_rate": 1.9981233578265207e-05, "loss": 4.0838, "step": 8110 }, { "epoch": 2.00220672, "grad_norm": 0.7738621234893799, "learning_rate": 1.9981184234429477e-05, "loss": 4.1345, "step": 8120 }, { "epoch": 2.00223232, "grad_norm": 0.8759828805923462, "learning_rate": 1.9981134825868417e-05, "loss": 4.2085, "step": 8130 }, { "epoch": 2.00225792, "grad_norm": 0.8971572518348694, "learning_rate": 1.9981085352582353e-05, "loss": 4.4092, "step": 8140 }, { "epoch": 2.00228352, "grad_norm": 0.8223668336868286, "learning_rate": 1.9981035814571605e-05, "loss": 4.1961, "step": 8150 }, { "epoch": 2.00230912, "grad_norm": 0.9101478457450867, "learning_rate": 1.998098621183649e-05, "loss": 4.1069, "step": 8160 }, { "epoch": 2.00233472, "grad_norm": 1.3463071584701538, "learning_rate": 1.9980936544377334e-05, "loss": 4.1883, "step": 8170 }, { "epoch": 2.00236032, "grad_norm": 0.8911945223808289, "learning_rate": 1.9980886812194454e-05, "loss": 4.5113, "step": 8180 }, { "epoch": 2.00238592, "grad_norm": 1.096731424331665, "learning_rate": 1.998083701528818e-05, "loss": 4.234, "step": 8190 }, { "epoch": 2.00241152, "grad_norm": 1.5708775520324707, "learning_rate": 1.998078715365883e-05, "loss": 4.2143, "step": 8200 }, { "epoch": 2.00243712, "grad_norm": 1.0660840272903442, "learning_rate": 1.9980737227306727e-05, "loss": 4.2788, "step": 8210 }, { "epoch": 2.00246272, "grad_norm": 1.2567903995513916, "learning_rate": 1.998068723623219e-05, "loss": 4.1287, "step": 8220 }, { "epoch": 2.00248832, "grad_norm": 0.8121759295463562, "learning_rate": 1.9980637180435556e-05, "loss": 4.1094, "step": 8230 }, { "epoch": 2.00251392, "grad_norm": 1.2891674041748047, "learning_rate": 1.998058705991714e-05, "loss": 4.4021, "step": 8240 }, { "epoch": 2.00253952, "grad_norm": 0.7682804465293884, "learning_rate": 1.998053687467727e-05, "loss": 4.0989, "step": 8250 }, { "epoch": 2.00256512, "grad_norm": 0.9782118797302246, "learning_rate": 1.998048662471627e-05, "loss": 4.1477, "step": 8260 }, { "epoch": 2.00259072, "grad_norm": 1.2429983615875244, "learning_rate": 1.9980436310034467e-05, "loss": 4.0151, "step": 8270 }, { "epoch": 2.00261632, "grad_norm": 0.7055304646492004, "learning_rate": 1.9980385930632187e-05, "loss": 4.3153, "step": 8280 }, { "epoch": 2.00264192, "grad_norm": 1.0966885089874268, "learning_rate": 1.9980335486509756e-05, "loss": 4.3024, "step": 8290 }, { "epoch": 2.00266752, "grad_norm": 0.9165486097335815, "learning_rate": 1.9980284977667503e-05, "loss": 4.1978, "step": 8300 }, { "epoch": 2.00269312, "grad_norm": 0.9313655495643616, "learning_rate": 1.9980234404105757e-05, "loss": 4.2604, "step": 8310 }, { "epoch": 2.00271872, "grad_norm": 1.0306317806243896, "learning_rate": 1.9980183765824838e-05, "loss": 4.5823, "step": 8320 }, { "epoch": 2.00274432, "grad_norm": 1.0874146223068237, "learning_rate": 1.9980133062825083e-05, "loss": 4.2919, "step": 8330 }, { "epoch": 2.00276992, "grad_norm": 0.9323046803474426, "learning_rate": 1.9980082295106814e-05, "loss": 4.1004, "step": 8340 }, { "epoch": 2.00279552, "grad_norm": 0.8742963671684265, "learning_rate": 1.9980031462670365e-05, "loss": 4.2879, "step": 8350 }, { "epoch": 2.00282112, "grad_norm": 0.9659060835838318, "learning_rate": 1.997998056551607e-05, "loss": 4.1355, "step": 8360 }, { "epoch": 2.00284672, "grad_norm": 0.8565601110458374, "learning_rate": 1.9979929603644245e-05, "loss": 3.9748, "step": 8370 }, { "epoch": 2.00287232, "grad_norm": 0.9844052791595459, "learning_rate": 1.9979878577055234e-05, "loss": 4.0076, "step": 8380 }, { "epoch": 2.00289792, "grad_norm": 0.8697795271873474, "learning_rate": 1.9979827485749365e-05, "loss": 4.1287, "step": 8390 }, { "epoch": 2.00292352, "grad_norm": 0.7768765687942505, "learning_rate": 1.997977632972696e-05, "loss": 4.3747, "step": 8400 }, { "epoch": 2.00294912, "grad_norm": 1.050646185874939, "learning_rate": 1.9979725108988362e-05, "loss": 4.2499, "step": 8410 }, { "epoch": 2.00297472, "grad_norm": 0.9772445559501648, "learning_rate": 1.99796738235339e-05, "loss": 4.1632, "step": 8420 }, { "epoch": 2.00300032, "grad_norm": 0.9330029487609863, "learning_rate": 1.9979622473363903e-05, "loss": 4.0034, "step": 8430 }, { "epoch": 2.00302592, "grad_norm": 0.8316556215286255, "learning_rate": 1.9979571058478706e-05, "loss": 4.2008, "step": 8440 }, { "epoch": 2.00305152, "grad_norm": 0.7735928297042847, "learning_rate": 1.9979519578878646e-05, "loss": 4.5022, "step": 8450 }, { "epoch": 2.00307712, "grad_norm": 0.9463450908660889, "learning_rate": 1.997946803456405e-05, "loss": 4.234, "step": 8460 }, { "epoch": 2.00310272, "grad_norm": 0.92378830909729, "learning_rate": 1.9979416425535262e-05, "loss": 4.3387, "step": 8470 }, { "epoch": 2.00312832, "grad_norm": 0.8165313601493835, "learning_rate": 1.9979364751792605e-05, "loss": 4.1874, "step": 8480 }, { "epoch": 2.00315392, "grad_norm": 0.7870264053344727, "learning_rate": 1.997931301333642e-05, "loss": 4.3258, "step": 8490 }, { "epoch": 2.00317952, "grad_norm": 0.979377031326294, "learning_rate": 1.9979261210167045e-05, "loss": 4.2272, "step": 8500 }, { "epoch": 2.00320512, "grad_norm": 1.1069118976593018, "learning_rate": 1.9979209342284813e-05, "loss": 4.2972, "step": 8510 }, { "epoch": 2.00323072, "grad_norm": 1.0581104755401611, "learning_rate": 1.997915740969006e-05, "loss": 4.0158, "step": 8520 }, { "epoch": 2.00325632, "grad_norm": 1.226859211921692, "learning_rate": 1.997910541238312e-05, "loss": 4.3031, "step": 8530 }, { "epoch": 2.00328192, "grad_norm": 1.0758758783340454, "learning_rate": 1.9979053350364338e-05, "loss": 4.0618, "step": 8540 }, { "epoch": 2.00330752, "grad_norm": 0.9141461253166199, "learning_rate": 1.9979001223634044e-05, "loss": 4.2976, "step": 8550 }, { "epoch": 2.00333312, "grad_norm": 0.9700636267662048, "learning_rate": 1.997894903219258e-05, "loss": 4.2236, "step": 8560 }, { "epoch": 2.00335872, "grad_norm": 0.7860197424888611, "learning_rate": 1.997889677604029e-05, "loss": 4.0706, "step": 8570 }, { "epoch": 2.00338432, "grad_norm": 1.0289201736450195, "learning_rate": 1.99788444551775e-05, "loss": 4.1846, "step": 8580 }, { "epoch": 2.00340992, "grad_norm": 1.1600922346115112, "learning_rate": 1.9978792069604557e-05, "loss": 4.1581, "step": 8590 }, { "epoch": 2.00343552, "grad_norm": 0.930946409702301, "learning_rate": 1.9978739619321796e-05, "loss": 4.0932, "step": 8600 }, { "epoch": 2.00346112, "grad_norm": 0.8978426456451416, "learning_rate": 1.9978687104329567e-05, "loss": 4.328, "step": 8610 }, { "epoch": 2.00348672, "grad_norm": 0.8358144164085388, "learning_rate": 1.99786345246282e-05, "loss": 4.1909, "step": 8620 }, { "epoch": 2.00351232, "grad_norm": 1.393670678138733, "learning_rate": 1.997858188021804e-05, "loss": 4.6101, "step": 8630 }, { "epoch": 2.00353792, "grad_norm": 0.9508007764816284, "learning_rate": 1.997852917109943e-05, "loss": 4.0513, "step": 8640 }, { "epoch": 2.00356352, "grad_norm": 0.7932361364364624, "learning_rate": 1.997847639727271e-05, "loss": 3.9948, "step": 8650 }, { "epoch": 2.00358912, "grad_norm": 0.9441336989402771, "learning_rate": 1.9978423558738224e-05, "loss": 4.13, "step": 8660 }, { "epoch": 2.00361472, "grad_norm": 1.4568302631378174, "learning_rate": 1.9978370655496307e-05, "loss": 4.2791, "step": 8670 }, { "epoch": 2.00364032, "grad_norm": 1.0119704008102417, "learning_rate": 1.9978317687547317e-05, "loss": 4.2021, "step": 8680 }, { "epoch": 2.00366592, "grad_norm": 0.7525993585586548, "learning_rate": 1.9978264654891583e-05, "loss": 4.1387, "step": 8690 }, { "epoch": 2.00369152, "grad_norm": 0.9581050872802734, "learning_rate": 1.9978211557529457e-05, "loss": 4.4613, "step": 8700 }, { "epoch": 2.00371712, "grad_norm": 1.1338543891906738, "learning_rate": 1.997815839546128e-05, "loss": 4.0003, "step": 8710 }, { "epoch": 2.00374272, "grad_norm": 0.8970514535903931, "learning_rate": 1.99781051686874e-05, "loss": 4.3188, "step": 8720 }, { "epoch": 2.00376832, "grad_norm": 0.779575526714325, "learning_rate": 1.9978051877208154e-05, "loss": 4.3718, "step": 8730 }, { "epoch": 2.00379392, "grad_norm": 0.8526417016983032, "learning_rate": 1.9977998521023897e-05, "loss": 4.1371, "step": 8740 }, { "epoch": 2.00381952, "grad_norm": 0.8712013959884644, "learning_rate": 1.997794510013497e-05, "loss": 4.2811, "step": 8750 }, { "epoch": 2.00384512, "grad_norm": 0.8560386896133423, "learning_rate": 1.9977891614541725e-05, "loss": 4.2448, "step": 8760 }, { "epoch": 2.00387072, "grad_norm": 0.9115575551986694, "learning_rate": 1.9977838064244502e-05, "loss": 3.8801, "step": 8770 }, { "epoch": 2.00389632, "grad_norm": 0.8435284495353699, "learning_rate": 1.9977784449243653e-05, "loss": 4.0549, "step": 8780 }, { "epoch": 2.00392192, "grad_norm": 1.0543479919433594, "learning_rate": 1.9977730769539518e-05, "loss": 4.345, "step": 8790 }, { "epoch": 2.00394752, "grad_norm": 1.0367687940597534, "learning_rate": 1.9977677025132457e-05, "loss": 4.1664, "step": 8800 }, { "epoch": 2.00397312, "grad_norm": 1.1625527143478394, "learning_rate": 1.9977623216022807e-05, "loss": 4.32, "step": 8810 }, { "epoch": 2.00399872, "grad_norm": 0.7649177312850952, "learning_rate": 1.9977569342210925e-05, "loss": 4.1713, "step": 8820 }, { "epoch": 2.00402432, "grad_norm": 1.0930129289627075, "learning_rate": 1.9977515403697157e-05, "loss": 4.0808, "step": 8830 }, { "epoch": 2.00404992, "grad_norm": 1.1666914224624634, "learning_rate": 1.9977461400481856e-05, "loss": 4.351, "step": 8840 }, { "epoch": 2.00407552, "grad_norm": 1.217238426208496, "learning_rate": 1.9977407332565368e-05, "loss": 4.1713, "step": 8850 }, { "epoch": 2.00410112, "grad_norm": 0.8453453183174133, "learning_rate": 1.997735319994804e-05, "loss": 4.1864, "step": 8860 }, { "epoch": 2.00412672, "grad_norm": 1.184369683265686, "learning_rate": 1.9977299002630234e-05, "loss": 4.1913, "step": 8870 }, { "epoch": 2.00415232, "grad_norm": 0.762507975101471, "learning_rate": 1.9977244740612293e-05, "loss": 4.2549, "step": 8880 }, { "epoch": 2.00417792, "grad_norm": 1.3659471273422241, "learning_rate": 1.9977190413894574e-05, "loss": 4.0908, "step": 8890 }, { "epoch": 2.00420352, "grad_norm": 0.9552938938140869, "learning_rate": 1.9977136022477426e-05, "loss": 4.1758, "step": 8900 }, { "epoch": 2.00422912, "grad_norm": 0.9437279105186462, "learning_rate": 1.99770815663612e-05, "loss": 4.1374, "step": 8910 }, { "epoch": 2.00425472, "grad_norm": 0.9137247204780579, "learning_rate": 1.9977027045546255e-05, "loss": 3.9854, "step": 8920 }, { "epoch": 2.00428032, "grad_norm": 0.7300878167152405, "learning_rate": 1.997697246003294e-05, "loss": 4.2243, "step": 8930 }, { "epoch": 2.00430592, "grad_norm": 0.8297160267829895, "learning_rate": 1.997691780982161e-05, "loss": 4.375, "step": 8940 }, { "epoch": 2.00433152, "grad_norm": 0.77110755443573, "learning_rate": 1.9976863094912617e-05, "loss": 4.0164, "step": 8950 }, { "epoch": 2.00435712, "grad_norm": 0.9243825078010559, "learning_rate": 1.997680831530632e-05, "loss": 4.3949, "step": 8960 }, { "epoch": 2.00438272, "grad_norm": 0.7175285220146179, "learning_rate": 1.9976753471003075e-05, "loss": 4.2435, "step": 8970 }, { "epoch": 2.00440832, "grad_norm": 0.7563964128494263, "learning_rate": 1.9976698562003232e-05, "loss": 3.9945, "step": 8980 }, { "epoch": 2.00443392, "grad_norm": 1.094982624053955, "learning_rate": 1.9976643588307156e-05, "loss": 3.8757, "step": 8990 }, { "epoch": 2.00445952, "grad_norm": 1.0552613735198975, "learning_rate": 1.997658854991519e-05, "loss": 4.0619, "step": 9000 }, { "epoch": 2.00448512, "grad_norm": 0.9235352873802185, "learning_rate": 1.9976533446827702e-05, "loss": 4.1693, "step": 9010 }, { "epoch": 2.00451072, "grad_norm": 1.1952928304672241, "learning_rate": 1.9976478279045047e-05, "loss": 4.1416, "step": 9020 }, { "epoch": 2.00453632, "grad_norm": 0.9836117625236511, "learning_rate": 1.997642304656758e-05, "loss": 4.1428, "step": 9030 }, { "epoch": 2.00456192, "grad_norm": 0.7912419438362122, "learning_rate": 1.9976367749395663e-05, "loss": 4.2873, "step": 9040 }, { "epoch": 2.00458752, "grad_norm": 0.6819174885749817, "learning_rate": 1.9976312387529652e-05, "loss": 4.2218, "step": 9050 }, { "epoch": 2.00461312, "grad_norm": 1.2510157823562622, "learning_rate": 1.9976256960969907e-05, "loss": 4.125, "step": 9060 }, { "epoch": 2.00463872, "grad_norm": 0.9596886038780212, "learning_rate": 1.9976201469716784e-05, "loss": 4.1153, "step": 9070 }, { "epoch": 2.00466432, "grad_norm": 1.2094708681106567, "learning_rate": 1.997614591377065e-05, "loss": 4.3048, "step": 9080 }, { "epoch": 2.00468992, "grad_norm": 0.8769728541374207, "learning_rate": 1.9976090293131858e-05, "loss": 4.2305, "step": 9090 }, { "epoch": 2.00471552, "grad_norm": 0.8249374628067017, "learning_rate": 1.9976034607800773e-05, "loss": 3.9742, "step": 9100 }, { "epoch": 2.00474112, "grad_norm": 1.4094483852386475, "learning_rate": 1.9975978857777756e-05, "loss": 4.2703, "step": 9110 }, { "epoch": 2.00476672, "grad_norm": 0.7302181124687195, "learning_rate": 1.9975923043063164e-05, "loss": 4.161, "step": 9120 }, { "epoch": 2.00479232, "grad_norm": 0.7611973285675049, "learning_rate": 1.9975867163657366e-05, "loss": 4.057, "step": 9130 }, { "epoch": 2.00481792, "grad_norm": 0.8909644484519958, "learning_rate": 1.9975811219560718e-05, "loss": 4.3114, "step": 9140 }, { "epoch": 2.00484352, "grad_norm": 0.9371961951255798, "learning_rate": 1.9975755210773588e-05, "loss": 4.0532, "step": 9150 }, { "epoch": 2.00486912, "grad_norm": 0.996842086315155, "learning_rate": 1.9975699137296333e-05, "loss": 4.2542, "step": 9160 }, { "epoch": 2.00489472, "grad_norm": 0.7347186803817749, "learning_rate": 1.9975642999129325e-05, "loss": 4.0657, "step": 9170 }, { "epoch": 2.00492032, "grad_norm": 0.7845838069915771, "learning_rate": 1.997558679627292e-05, "loss": 4.1624, "step": 9180 }, { "epoch": 2.00494592, "grad_norm": 1.0822752714157104, "learning_rate": 1.9975530528727487e-05, "loss": 4.4736, "step": 9190 }, { "epoch": 2.00497152, "grad_norm": 0.9469752907752991, "learning_rate": 1.997547419649339e-05, "loss": 4.305, "step": 9200 }, { "epoch": 2.00499712, "grad_norm": 1.0822087526321411, "learning_rate": 1.9975417799570996e-05, "loss": 4.2129, "step": 9210 }, { "epoch": 2.00502272, "grad_norm": 0.8804514408111572, "learning_rate": 1.9975361337960662e-05, "loss": 4.0295, "step": 9220 }, { "epoch": 2.00504832, "grad_norm": 0.7256231307983398, "learning_rate": 1.9975304811662765e-05, "loss": 4.106, "step": 9230 }, { "epoch": 2.00507392, "grad_norm": 0.7280685901641846, "learning_rate": 1.997524822067767e-05, "loss": 4.1505, "step": 9240 }, { "epoch": 2.00509952, "grad_norm": 0.8750420212745667, "learning_rate": 1.9975191565005738e-05, "loss": 3.9901, "step": 9250 }, { "epoch": 2.00512512, "grad_norm": 0.8384281992912292, "learning_rate": 1.997513484464734e-05, "loss": 3.9999, "step": 9260 }, { "epoch": 2.00515072, "grad_norm": 1.907199501991272, "learning_rate": 1.9975078059602847e-05, "loss": 4.3207, "step": 9270 }, { "epoch": 2.00517632, "grad_norm": 1.1201456785202026, "learning_rate": 1.997502120987262e-05, "loss": 4.4333, "step": 9280 }, { "epoch": 2.00520192, "grad_norm": 1.0531461238861084, "learning_rate": 1.9974964295457034e-05, "loss": 4.1603, "step": 9290 }, { "epoch": 2.00522752, "grad_norm": 0.9387350082397461, "learning_rate": 1.997490731635645e-05, "loss": 4.2355, "step": 9300 }, { "epoch": 2.00525312, "grad_norm": 0.9628258943557739, "learning_rate": 1.997485027257125e-05, "loss": 4.0995, "step": 9310 }, { "epoch": 2.00527872, "grad_norm": 0.9128039479255676, "learning_rate": 1.9974793164101793e-05, "loss": 4.138, "step": 9320 }, { "epoch": 2.00530432, "grad_norm": 1.1353585720062256, "learning_rate": 1.9974735990948456e-05, "loss": 4.2466, "step": 9330 }, { "epoch": 2.00532992, "grad_norm": 0.8574288487434387, "learning_rate": 1.9974678753111603e-05, "loss": 4.1257, "step": 9340 }, { "epoch": 2.00535552, "grad_norm": 1.0307817459106445, "learning_rate": 1.9974621450591615e-05, "loss": 4.1431, "step": 9350 }, { "epoch": 2.00538112, "grad_norm": 0.7848602533340454, "learning_rate": 1.9974564083388854e-05, "loss": 3.9513, "step": 9360 }, { "epoch": 2.00540672, "grad_norm": 0.8429060578346252, "learning_rate": 1.99745066515037e-05, "loss": 4.252, "step": 9370 }, { "epoch": 2.00543232, "grad_norm": 0.7707403898239136, "learning_rate": 1.9974449154936516e-05, "loss": 4.1419, "step": 9380 }, { "epoch": 2.00545792, "grad_norm": 0.844947338104248, "learning_rate": 1.997439159368768e-05, "loss": 3.9708, "step": 9390 }, { "epoch": 2.00548352, "grad_norm": 0.8117473721504211, "learning_rate": 1.9974333967757568e-05, "loss": 4.0828, "step": 9400 }, { "epoch": 2.00550912, "grad_norm": 1.1338813304901123, "learning_rate": 1.997427627714655e-05, "loss": 4.0685, "step": 9410 }, { "epoch": 2.00553472, "grad_norm": 0.854607880115509, "learning_rate": 1.9974218521855002e-05, "loss": 4.1929, "step": 9420 }, { "epoch": 2.00556032, "grad_norm": 1.053183674812317, "learning_rate": 1.9974160701883298e-05, "loss": 4.228, "step": 9430 }, { "epoch": 2.00558592, "grad_norm": 0.9906231164932251, "learning_rate": 1.9974102817231812e-05, "loss": 4.3392, "step": 9440 }, { "epoch": 2.00561152, "grad_norm": 1.1916577816009521, "learning_rate": 1.9974044867900918e-05, "loss": 4.0463, "step": 9450 }, { "epoch": 2.00563712, "grad_norm": 0.8246939778327942, "learning_rate": 1.9973986853890998e-05, "loss": 4.1546, "step": 9460 }, { "epoch": 2.00566272, "grad_norm": 0.9053595066070557, "learning_rate": 1.9973928775202424e-05, "loss": 4.2294, "step": 9470 }, { "epoch": 2.00568832, "grad_norm": 0.7311368584632874, "learning_rate": 1.997387063183557e-05, "loss": 4.0886, "step": 9480 }, { "epoch": 2.00571392, "grad_norm": 0.7691469788551331, "learning_rate": 1.9973812423790815e-05, "loss": 3.9663, "step": 9490 }, { "epoch": 2.00573952, "grad_norm": 1.0023876428604126, "learning_rate": 1.997375415106854e-05, "loss": 3.9132, "step": 9500 }, { "epoch": 2.00576512, "grad_norm": 1.2291269302368164, "learning_rate": 1.9973695813669117e-05, "loss": 4.1649, "step": 9510 }, { "epoch": 2.00579072, "grad_norm": 0.7517366409301758, "learning_rate": 1.997363741159293e-05, "loss": 4.3292, "step": 9520 }, { "epoch": 2.00581632, "grad_norm": 0.7110762596130371, "learning_rate": 1.9973578944840354e-05, "loss": 4.2861, "step": 9530 }, { "epoch": 2.00584192, "grad_norm": 1.0420037508010864, "learning_rate": 1.9973520413411766e-05, "loss": 4.2599, "step": 9540 }, { "epoch": 2.00586752, "grad_norm": 0.8100886344909668, "learning_rate": 1.9973461817307554e-05, "loss": 4.137, "step": 9550 }, { "epoch": 2.00589312, "grad_norm": 0.8461206555366516, "learning_rate": 1.997340315652809e-05, "loss": 3.8705, "step": 9560 }, { "epoch": 2.00591872, "grad_norm": 0.9138414859771729, "learning_rate": 1.9973344431073757e-05, "loss": 4.5924, "step": 9570 }, { "epoch": 2.00594432, "grad_norm": 1.491059422492981, "learning_rate": 1.997328564094494e-05, "loss": 4.4804, "step": 9580 }, { "epoch": 2.00596992, "grad_norm": 0.8831546902656555, "learning_rate": 1.997322678614201e-05, "loss": 4.3702, "step": 9590 }, { "epoch": 2.00599552, "grad_norm": 1.142309546470642, "learning_rate": 1.997316786666536e-05, "loss": 4.2087, "step": 9600 }, { "epoch": 2.00602112, "grad_norm": 0.8865702152252197, "learning_rate": 1.9973108882515363e-05, "loss": 4.4154, "step": 9610 }, { "epoch": 2.00604672, "grad_norm": 0.9662391543388367, "learning_rate": 1.9973049833692406e-05, "loss": 4.1539, "step": 9620 }, { "epoch": 2.00607232, "grad_norm": 0.8170217871665955, "learning_rate": 1.997299072019687e-05, "loss": 4.194, "step": 9630 }, { "epoch": 2.00609792, "grad_norm": 0.8559330701828003, "learning_rate": 1.9972931542029147e-05, "loss": 4.1268, "step": 9640 }, { "epoch": 2.00612352, "grad_norm": 0.7955312728881836, "learning_rate": 1.9972872299189604e-05, "loss": 4.249, "step": 9650 }, { "epoch": 2.00614912, "grad_norm": 0.7657216191291809, "learning_rate": 1.9972812991678642e-05, "loss": 4.1891, "step": 9660 }, { "epoch": 2.00617472, "grad_norm": 0.8594947457313538, "learning_rate": 1.9972753619496636e-05, "loss": 4.1674, "step": 9670 }, { "epoch": 2.00620032, "grad_norm": 0.8087748289108276, "learning_rate": 1.9972694182643973e-05, "loss": 4.0984, "step": 9680 }, { "epoch": 2.00622592, "grad_norm": 1.1114177703857422, "learning_rate": 1.997263468112104e-05, "loss": 4.1906, "step": 9690 }, { "epoch": 2.00625152, "grad_norm": 0.8222482204437256, "learning_rate": 1.9972575114928216e-05, "loss": 4.0252, "step": 9700 }, { "epoch": 2.00627712, "grad_norm": 0.8693839311599731, "learning_rate": 1.9972515484065895e-05, "loss": 4.1321, "step": 9710 }, { "epoch": 2.00630272, "grad_norm": 0.798491895198822, "learning_rate": 1.9972455788534466e-05, "loss": 4.3168, "step": 9720 }, { "epoch": 2.00632832, "grad_norm": 0.8165501952171326, "learning_rate": 1.997239602833431e-05, "loss": 4.1687, "step": 9730 }, { "epoch": 2.00635392, "grad_norm": 2.2570810317993164, "learning_rate": 1.9972336203465812e-05, "loss": 4.2822, "step": 9740 }, { "epoch": 2.00637952, "grad_norm": 0.8480322360992432, "learning_rate": 1.9972276313929366e-05, "loss": 4.0306, "step": 9750 }, { "epoch": 2.00640512, "grad_norm": 0.8219184875488281, "learning_rate": 1.9972216359725357e-05, "loss": 4.1527, "step": 9760 }, { "epoch": 2.00643072, "grad_norm": 0.8206943273544312, "learning_rate": 1.9972156340854177e-05, "loss": 4.0264, "step": 9770 }, { "epoch": 2.00645632, "grad_norm": 0.8840165734291077, "learning_rate": 1.9972096257316216e-05, "loss": 3.9316, "step": 9780 }, { "epoch": 2.00648192, "grad_norm": 0.847308337688446, "learning_rate": 1.997203610911186e-05, "loss": 4.1454, "step": 9790 }, { "epoch": 2.00650752, "grad_norm": 0.8466686606407166, "learning_rate": 1.99719758962415e-05, "loss": 4.011, "step": 9800 }, { "epoch": 2.00653312, "grad_norm": 1.0305739641189575, "learning_rate": 1.9971915618705525e-05, "loss": 4.0571, "step": 9810 }, { "epoch": 2.00655872, "grad_norm": 1.0655031204223633, "learning_rate": 1.9971855276504325e-05, "loss": 3.9496, "step": 9820 }, { "epoch": 2.00658432, "grad_norm": 1.0302146673202515, "learning_rate": 1.99717948696383e-05, "loss": 4.2564, "step": 9830 }, { "epoch": 2.00660992, "grad_norm": 1.7510535717010498, "learning_rate": 1.997173439810783e-05, "loss": 3.9237, "step": 9840 }, { "epoch": 2.00663552, "grad_norm": 0.8828943967819214, "learning_rate": 1.9971673861913314e-05, "loss": 4.1267, "step": 9850 }, { "epoch": 2.00666112, "grad_norm": 1.0165622234344482, "learning_rate": 1.9971613261055144e-05, "loss": 4.2737, "step": 9860 }, { "epoch": 2.00668672, "grad_norm": 1.9563114643096924, "learning_rate": 1.9971552595533713e-05, "loss": 3.8883, "step": 9870 }, { "epoch": 2.00671232, "grad_norm": 0.9084106087684631, "learning_rate": 1.9971491865349414e-05, "loss": 4.1378, "step": 9880 }, { "epoch": 2.00673792, "grad_norm": 0.9330739974975586, "learning_rate": 1.997143107050264e-05, "loss": 4.1452, "step": 9890 }, { "epoch": 2.00676352, "grad_norm": 0.8768689632415771, "learning_rate": 1.997137021099378e-05, "loss": 4.1135, "step": 9900 }, { "epoch": 2.00678912, "grad_norm": 0.8698244690895081, "learning_rate": 1.997130928682324e-05, "loss": 4.4101, "step": 9910 }, { "epoch": 2.00681472, "grad_norm": 0.9686182737350464, "learning_rate": 1.997124829799141e-05, "loss": 4.0491, "step": 9920 }, { "epoch": 2.00684032, "grad_norm": 0.8998849391937256, "learning_rate": 1.997118724449868e-05, "loss": 4.0842, "step": 9930 }, { "epoch": 2.00686592, "grad_norm": 1.1184145212173462, "learning_rate": 1.9971126126345454e-05, "loss": 4.3487, "step": 9940 }, { "epoch": 2.00689152, "grad_norm": 0.8291346430778503, "learning_rate": 1.9971064943532122e-05, "loss": 4.0837, "step": 9950 }, { "epoch": 2.00691712, "grad_norm": 0.8065507411956787, "learning_rate": 1.997100369605909e-05, "loss": 4.4025, "step": 9960 }, { "epoch": 2.00694272, "grad_norm": 1.0717626810073853, "learning_rate": 1.9970942383926744e-05, "loss": 4.1725, "step": 9970 }, { "epoch": 2.00696832, "grad_norm": 0.7763336896896362, "learning_rate": 1.997088100713549e-05, "loss": 4.2501, "step": 9980 }, { "epoch": 2.00699392, "grad_norm": 1.1209830045700073, "learning_rate": 1.997081956568572e-05, "loss": 4.2186, "step": 9990 }, { "epoch": 2.00701952, "grad_norm": 1.0551810264587402, "learning_rate": 1.9970758059577835e-05, "loss": 4.2044, "step": 10000 }, { "epoch": 2.00704512, "grad_norm": 0.9816724061965942, "learning_rate": 1.9970696488812232e-05, "loss": 4.1447, "step": 10010 }, { "epoch": 2.00707072, "grad_norm": 0.8161169290542603, "learning_rate": 1.9970634853389317e-05, "loss": 4.1938, "step": 10020 }, { "epoch": 2.00709632, "grad_norm": 1.0698343515396118, "learning_rate": 1.997057315330948e-05, "loss": 4.3932, "step": 10030 }, { "epoch": 2.00712192, "grad_norm": 1.0084092617034912, "learning_rate": 1.9970511388573134e-05, "loss": 4.1596, "step": 10040 }, { "epoch": 2.00714752, "grad_norm": 0.7414066195487976, "learning_rate": 1.9970449559180664e-05, "loss": 4.2708, "step": 10050 }, { "epoch": 2.00717312, "grad_norm": 0.9537082314491272, "learning_rate": 1.9970387665132482e-05, "loss": 4.0117, "step": 10060 }, { "epoch": 2.00719872, "grad_norm": 1.0447728633880615, "learning_rate": 1.9970325706428984e-05, "loss": 4.1215, "step": 10070 }, { "epoch": 2.00722432, "grad_norm": 1.0344699621200562, "learning_rate": 1.9970263683070574e-05, "loss": 4.291, "step": 10080 }, { "epoch": 2.00724992, "grad_norm": 0.7592358589172363, "learning_rate": 1.997020159505765e-05, "loss": 3.935, "step": 10090 }, { "epoch": 2.00727552, "grad_norm": 0.7955338954925537, "learning_rate": 1.9970139442390625e-05, "loss": 4.0002, "step": 10100 }, { "epoch": 2.00730112, "grad_norm": 1.0195538997650146, "learning_rate": 1.997007722506989e-05, "loss": 4.1172, "step": 10110 }, { "epoch": 2.00732672, "grad_norm": 0.7559549808502197, "learning_rate": 1.997001494309586e-05, "loss": 4.2658, "step": 10120 }, { "epoch": 2.00735232, "grad_norm": 1.450440526008606, "learning_rate": 1.9969952596468928e-05, "loss": 4.8222, "step": 10130 }, { "epoch": 2.00737792, "grad_norm": 0.9424483776092529, "learning_rate": 1.9969890185189503e-05, "loss": 4.1856, "step": 10140 }, { "epoch": 2.00740352, "grad_norm": 0.8894896507263184, "learning_rate": 1.996982770925799e-05, "loss": 4.2574, "step": 10150 }, { "epoch": 2.00742912, "grad_norm": 0.9892232418060303, "learning_rate": 1.9969765168674798e-05, "loss": 4.1617, "step": 10160 }, { "epoch": 2.00745472, "grad_norm": 0.8544788360595703, "learning_rate": 1.9969702563440325e-05, "loss": 4.1287, "step": 10170 }, { "epoch": 2.00748032, "grad_norm": 1.0434564352035522, "learning_rate": 1.996963989355498e-05, "loss": 4.0217, "step": 10180 }, { "epoch": 2.00750592, "grad_norm": 0.9343261122703552, "learning_rate": 1.996957715901917e-05, "loss": 4.1076, "step": 10190 }, { "epoch": 2.00753152, "grad_norm": 1.2503633499145508, "learning_rate": 1.9969514359833305e-05, "loss": 4.1021, "step": 10200 }, { "epoch": 2.00755712, "grad_norm": 1.0596421957015991, "learning_rate": 1.9969451495997786e-05, "loss": 4.4207, "step": 10210 }, { "epoch": 2.00758272, "grad_norm": 1.0239379405975342, "learning_rate": 1.9969388567513026e-05, "loss": 4.1517, "step": 10220 }, { "epoch": 2.00760832, "grad_norm": 0.9479130506515503, "learning_rate": 1.996932557437943e-05, "loss": 4.0311, "step": 10230 }, { "epoch": 2.00763392, "grad_norm": 0.8942089676856995, "learning_rate": 1.9969262516597407e-05, "loss": 4.089, "step": 10240 }, { "epoch": 2.00765952, "grad_norm": 0.827092707157135, "learning_rate": 1.9969199394167364e-05, "loss": 4.1662, "step": 10250 }, { "epoch": 2.00768512, "grad_norm": 1.6542904376983643, "learning_rate": 1.9969136207089713e-05, "loss": 4.0382, "step": 10260 }, { "epoch": 2.00771072, "grad_norm": 0.8976579308509827, "learning_rate": 1.9969072955364865e-05, "loss": 3.9129, "step": 10270 }, { "epoch": 2.00773632, "grad_norm": 0.9620733261108398, "learning_rate": 1.9969009638993225e-05, "loss": 4.1662, "step": 10280 }, { "epoch": 2.00776192, "grad_norm": 1.4754512310028076, "learning_rate": 1.996894625797521e-05, "loss": 4.2555, "step": 10290 }, { "epoch": 2.00778752, "grad_norm": 3.137934923171997, "learning_rate": 1.996888281231123e-05, "loss": 4.9938, "step": 10300 }, { "epoch": 2.00781312, "grad_norm": 1.0039336681365967, "learning_rate": 1.996881930200169e-05, "loss": 4.1259, "step": 10310 }, { "epoch": 2.00783872, "grad_norm": 1.0484122037887573, "learning_rate": 1.9968755727047008e-05, "loss": 4.0792, "step": 10320 }, { "epoch": 2.00786432, "grad_norm": 0.8485099077224731, "learning_rate": 1.9968692087447596e-05, "loss": 4.0817, "step": 10330 }, { "epoch": 2.00788992, "grad_norm": 1.0408656597137451, "learning_rate": 1.9968628383203862e-05, "loss": 3.9278, "step": 10340 }, { "epoch": 2.00791552, "grad_norm": 1.1267815828323364, "learning_rate": 1.996856461431622e-05, "loss": 4.1579, "step": 10350 }, { "epoch": 2.00794112, "grad_norm": 1.3298590183258057, "learning_rate": 1.9968500780785095e-05, "loss": 4.1459, "step": 10360 }, { "epoch": 2.00796672, "grad_norm": 0.7518244385719299, "learning_rate": 1.9968436882610884e-05, "loss": 4.0856, "step": 10370 }, { "epoch": 2.00799232, "grad_norm": 0.9340631365776062, "learning_rate": 1.996837291979401e-05, "loss": 4.0567, "step": 10380 }, { "epoch": 2.00801792, "grad_norm": 0.9780095219612122, "learning_rate": 1.996830889233489e-05, "loss": 4.0872, "step": 10390 }, { "epoch": 2.00804352, "grad_norm": 1.1848087310791016, "learning_rate": 1.9968244800233933e-05, "loss": 4.0821, "step": 10400 }, { "epoch": 2.00806912, "grad_norm": 0.96471107006073, "learning_rate": 1.9968180643491562e-05, "loss": 4.01, "step": 10410 }, { "epoch": 2.00809472, "grad_norm": 0.921994686126709, "learning_rate": 1.9968116422108186e-05, "loss": 4.0424, "step": 10420 }, { "epoch": 2.00812032, "grad_norm": 1.0680731534957886, "learning_rate": 1.9968052136084223e-05, "loss": 4.0327, "step": 10430 }, { "epoch": 2.00814592, "grad_norm": 0.8588264584541321, "learning_rate": 1.996798778542009e-05, "loss": 4.0497, "step": 10440 }, { "epoch": 2.00817152, "grad_norm": 0.9364487528800964, "learning_rate": 1.996792337011621e-05, "loss": 4.2422, "step": 10450 }, { "epoch": 2.00819712, "grad_norm": 0.8687959313392639, "learning_rate": 1.9967858890172993e-05, "loss": 4.3266, "step": 10460 }, { "epoch": 2.00822272, "grad_norm": 0.8969164490699768, "learning_rate": 1.996779434559086e-05, "loss": 4.098, "step": 10470 }, { "epoch": 2.00824832, "grad_norm": 0.9349988102912903, "learning_rate": 1.9967729736370232e-05, "loss": 4.1939, "step": 10480 }, { "epoch": 2.00827392, "grad_norm": 0.8732438087463379, "learning_rate": 1.9967665062511522e-05, "loss": 4.2378, "step": 10490 }, { "epoch": 2.00829952, "grad_norm": 0.9509188532829285, "learning_rate": 1.9967600324015156e-05, "loss": 3.9402, "step": 10500 }, { "epoch": 2.00832512, "grad_norm": 1.0611366033554077, "learning_rate": 1.9967535520881553e-05, "loss": 3.796, "step": 10510 }, { "epoch": 2.00835072, "grad_norm": 0.8599652647972107, "learning_rate": 1.9967470653111126e-05, "loss": 4.2913, "step": 10520 }, { "epoch": 2.00837632, "grad_norm": 0.9801068902015686, "learning_rate": 1.9967405720704304e-05, "loss": 4.0933, "step": 10530 }, { "epoch": 2.00840192, "grad_norm": 0.9441190361976624, "learning_rate": 1.9967340723661505e-05, "loss": 4.2188, "step": 10540 }, { "epoch": 2.00842752, "grad_norm": 0.8499354124069214, "learning_rate": 1.996727566198315e-05, "loss": 3.8329, "step": 10550 }, { "epoch": 2.00845312, "grad_norm": 0.9687700271606445, "learning_rate": 1.996721053566966e-05, "loss": 3.8423, "step": 10560 }, { "epoch": 2.00847872, "grad_norm": 1.0491853952407837, "learning_rate": 1.996714534472146e-05, "loss": 4.1519, "step": 10570 }, { "epoch": 2.00850432, "grad_norm": 0.861030638217926, "learning_rate": 1.9967080089138972e-05, "loss": 3.9895, "step": 10580 }, { "epoch": 2.00852992, "grad_norm": 0.9878951907157898, "learning_rate": 1.996701476892262e-05, "loss": 4.0579, "step": 10590 }, { "epoch": 2.00855552, "grad_norm": 0.8958234786987305, "learning_rate": 1.9966949384072823e-05, "loss": 4.2718, "step": 10600 }, { "epoch": 2.00858112, "grad_norm": 0.7951473593711853, "learning_rate": 1.9966883934590008e-05, "loss": 4.0282, "step": 10610 }, { "epoch": 2.00860672, "grad_norm": 0.7249894738197327, "learning_rate": 1.99668184204746e-05, "loss": 4.1675, "step": 10620 }, { "epoch": 2.00863232, "grad_norm": 1.0034501552581787, "learning_rate": 1.9966752841727027e-05, "loss": 4.0264, "step": 10630 }, { "epoch": 2.00865792, "grad_norm": 0.9867568612098694, "learning_rate": 1.9966687198347707e-05, "loss": 4.0583, "step": 10640 }, { "epoch": 2.00868352, "grad_norm": 1.0023306608200073, "learning_rate": 1.996662149033707e-05, "loss": 4.2303, "step": 10650 }, { "epoch": 2.00870912, "grad_norm": 0.914930522441864, "learning_rate": 1.9966555717695543e-05, "loss": 4.0031, "step": 10660 }, { "epoch": 2.00873472, "grad_norm": 1.0251400470733643, "learning_rate": 1.9966489880423548e-05, "loss": 4.4435, "step": 10670 }, { "epoch": 2.00876032, "grad_norm": 0.8222590088844299, "learning_rate": 1.9966423978521516e-05, "loss": 4.0228, "step": 10680 }, { "epoch": 2.00878592, "grad_norm": 1.8695420026779175, "learning_rate": 1.996635801198987e-05, "loss": 4.102, "step": 10690 }, { "epoch": 2.00881152, "grad_norm": 1.2847833633422852, "learning_rate": 1.996629198082905e-05, "loss": 4.2693, "step": 10700 }, { "epoch": 2.00883712, "grad_norm": 1.6089913845062256, "learning_rate": 1.996622588503947e-05, "loss": 4.1498, "step": 10710 }, { "epoch": 2.00886272, "grad_norm": 1.6209092140197754, "learning_rate": 1.996615972462156e-05, "loss": 4.18, "step": 10720 }, { "epoch": 2.00888832, "grad_norm": 1.3051663637161255, "learning_rate": 1.996609349957576e-05, "loss": 3.979, "step": 10730 }, { "epoch": 2.00891392, "grad_norm": 0.839655339717865, "learning_rate": 1.9966027209902488e-05, "loss": 4.1068, "step": 10740 }, { "epoch": 2.00893952, "grad_norm": 0.7786843180656433, "learning_rate": 1.996596085560218e-05, "loss": 4.1357, "step": 10750 }, { "epoch": 2.00896512, "grad_norm": 0.9722468256950378, "learning_rate": 1.996589443667526e-05, "loss": 4.1198, "step": 10760 }, { "epoch": 2.00899072, "grad_norm": 0.8076518177986145, "learning_rate": 1.996582795312217e-05, "loss": 3.8958, "step": 10770 }, { "epoch": 2.00901632, "grad_norm": 0.9390507936477661, "learning_rate": 1.9965761404943333e-05, "loss": 3.8234, "step": 10780 }, { "epoch": 2.00904192, "grad_norm": 1.1396501064300537, "learning_rate": 1.996569479213918e-05, "loss": 3.9372, "step": 10790 }, { "epoch": 2.00906752, "grad_norm": 0.8464909791946411, "learning_rate": 1.9965628114710142e-05, "loss": 4.1842, "step": 10800 }, { "epoch": 2.00909312, "grad_norm": 0.8743751049041748, "learning_rate": 1.9965561372656655e-05, "loss": 3.9574, "step": 10810 }, { "epoch": 2.00911872, "grad_norm": 0.9993516206741333, "learning_rate": 1.9965494565979156e-05, "loss": 4.0492, "step": 10820 }, { "epoch": 2.00914432, "grad_norm": 0.8175132870674133, "learning_rate": 1.9965427694678074e-05, "loss": 4.2937, "step": 10830 }, { "epoch": 2.00916992, "grad_norm": 1.1465157270431519, "learning_rate": 1.9965360758753833e-05, "loss": 4.2329, "step": 10840 }, { "epoch": 2.00919552, "grad_norm": 1.1288162469863892, "learning_rate": 1.9965293758206883e-05, "loss": 4.2925, "step": 10850 }, { "epoch": 2.00922112, "grad_norm": 0.7611621618270874, "learning_rate": 1.9965226693037652e-05, "loss": 4.145, "step": 10860 }, { "epoch": 2.00924672, "grad_norm": 1.574174404144287, "learning_rate": 1.996515956324657e-05, "loss": 4.0833, "step": 10870 }, { "epoch": 2.00927232, "grad_norm": 0.8303804397583008, "learning_rate": 1.9965092368834077e-05, "loss": 4.0549, "step": 10880 }, { "epoch": 3.00000768, "grad_norm": 1.1264575719833374, "learning_rate": 1.996502510980061e-05, "loss": 4.4512, "step": 10890 }, { "epoch": 3.00003328, "grad_norm": 0.9430621266365051, "learning_rate": 1.9964957786146605e-05, "loss": 4.1497, "step": 10900 }, { "epoch": 3.00005888, "grad_norm": 0.9136727452278137, "learning_rate": 1.9964890397872495e-05, "loss": 3.7134, "step": 10910 }, { "epoch": 3.00008448, "grad_norm": 0.9106637835502625, "learning_rate": 1.996482294497872e-05, "loss": 3.8708, "step": 10920 }, { "epoch": 3.00011008, "grad_norm": 0.6765013337135315, "learning_rate": 1.9964755427465717e-05, "loss": 3.9518, "step": 10930 }, { "epoch": 3.00013568, "grad_norm": 0.9841207265853882, "learning_rate": 1.9964687845333924e-05, "loss": 4.07, "step": 10940 }, { "epoch": 3.00016128, "grad_norm": 1.1735217571258545, "learning_rate": 1.9964620198583774e-05, "loss": 3.9392, "step": 10950 }, { "epoch": 3.00018688, "grad_norm": 1.235886812210083, "learning_rate": 1.9964552487215717e-05, "loss": 4.4593, "step": 10960 }, { "epoch": 3.00021248, "grad_norm": 1.1486914157867432, "learning_rate": 1.9964484711230182e-05, "loss": 3.9738, "step": 10970 }, { "epoch": 3.00023808, "grad_norm": 0.8859565258026123, "learning_rate": 1.996441687062761e-05, "loss": 4.317, "step": 10980 }, { "epoch": 3.00026368, "grad_norm": 0.8344873785972595, "learning_rate": 1.9964348965408447e-05, "loss": 4.1309, "step": 10990 }, { "epoch": 3.00028928, "grad_norm": 0.8829224705696106, "learning_rate": 1.9964280995573127e-05, "loss": 3.8494, "step": 11000 }, { "epoch": 3.00031488, "grad_norm": 0.8447806239128113, "learning_rate": 1.9964212961122092e-05, "loss": 3.5642, "step": 11010 }, { "epoch": 3.00034048, "grad_norm": 0.8701715469360352, "learning_rate": 1.9964144862055783e-05, "loss": 3.7007, "step": 11020 }, { "epoch": 3.00036608, "grad_norm": 0.9130391478538513, "learning_rate": 1.9964076698374645e-05, "loss": 3.844, "step": 11030 }, { "epoch": 3.00039168, "grad_norm": 1.0086982250213623, "learning_rate": 1.996400847007912e-05, "loss": 3.843, "step": 11040 }, { "epoch": 3.00041728, "grad_norm": 0.8818647265434265, "learning_rate": 1.9963940177169646e-05, "loss": 3.9826, "step": 11050 }, { "epoch": 3.00044288, "grad_norm": 0.8197019100189209, "learning_rate": 1.9963871819646666e-05, "loss": 3.9866, "step": 11060 }, { "epoch": 3.00046848, "grad_norm": 0.7435846328735352, "learning_rate": 1.996380339751063e-05, "loss": 4.0321, "step": 11070 }, { "epoch": 3.00049408, "grad_norm": 1.1282312870025635, "learning_rate": 1.996373491076197e-05, "loss": 3.9552, "step": 11080 }, { "epoch": 3.00051968, "grad_norm": 1.4573811292648315, "learning_rate": 1.996366635940114e-05, "loss": 3.8595, "step": 11090 }, { "epoch": 3.00054528, "grad_norm": 0.7791318297386169, "learning_rate": 1.9963597743428585e-05, "loss": 3.9486, "step": 11100 }, { "epoch": 3.00057088, "grad_norm": 0.9729819893836975, "learning_rate": 1.9963529062844748e-05, "loss": 3.517, "step": 11110 }, { "epoch": 3.00059648, "grad_norm": 0.8734391331672668, "learning_rate": 1.9963460317650068e-05, "loss": 3.8995, "step": 11120 }, { "epoch": 3.00062208, "grad_norm": 0.9908273816108704, "learning_rate": 1.9963391507844995e-05, "loss": 3.8078, "step": 11130 }, { "epoch": 3.00064768, "grad_norm": 0.7673224210739136, "learning_rate": 1.996332263342998e-05, "loss": 3.7338, "step": 11140 }, { "epoch": 3.00067328, "grad_norm": 0.7337746024131775, "learning_rate": 1.9963253694405464e-05, "loss": 3.9432, "step": 11150 }, { "epoch": 3.00069888, "grad_norm": 0.9757028222084045, "learning_rate": 1.9963184690771894e-05, "loss": 4.3973, "step": 11160 }, { "epoch": 3.00072448, "grad_norm": 1.215811848640442, "learning_rate": 1.9963115622529724e-05, "loss": 4.0307, "step": 11170 }, { "epoch": 3.00075008, "grad_norm": 0.8426336050033569, "learning_rate": 1.996304648967939e-05, "loss": 3.9268, "step": 11180 }, { "epoch": 3.00077568, "grad_norm": 0.9539929032325745, "learning_rate": 1.996297729222135e-05, "loss": 3.8281, "step": 11190 }, { "epoch": 3.00080128, "grad_norm": 0.9296333193778992, "learning_rate": 1.9962908030156054e-05, "loss": 4.1736, "step": 11200 }, { "epoch": 3.00082688, "grad_norm": 0.7510484457015991, "learning_rate": 1.9962838703483945e-05, "loss": 3.891, "step": 11210 }, { "epoch": 3.00085248, "grad_norm": 1.1817710399627686, "learning_rate": 1.9962769312205475e-05, "loss": 3.8379, "step": 11220 }, { "epoch": 3.00087808, "grad_norm": 0.9675653576850891, "learning_rate": 1.9962699856321092e-05, "loss": 3.7636, "step": 11230 }, { "epoch": 3.00090368, "grad_norm": 1.1056550741195679, "learning_rate": 1.996263033583125e-05, "loss": 3.9442, "step": 11240 }, { "epoch": 3.00092928, "grad_norm": 1.0159446001052856, "learning_rate": 1.99625607507364e-05, "loss": 3.8971, "step": 11250 }, { "epoch": 3.00095488, "grad_norm": 1.0050737857818604, "learning_rate": 1.9962491101036986e-05, "loss": 3.9106, "step": 11260 }, { "epoch": 3.00098048, "grad_norm": 0.8968580961227417, "learning_rate": 1.996242138673347e-05, "loss": 4.0229, "step": 11270 }, { "epoch": 3.00100608, "grad_norm": 1.157130241394043, "learning_rate": 1.9962351607826296e-05, "loss": 3.9336, "step": 11280 }, { "epoch": 3.00103168, "grad_norm": 0.8371566534042358, "learning_rate": 1.996228176431592e-05, "loss": 3.804, "step": 11290 }, { "epoch": 3.00105728, "grad_norm": 0.7773609161376953, "learning_rate": 1.9962211856202794e-05, "loss": 3.7351, "step": 11300 }, { "epoch": 3.00108288, "grad_norm": 0.8657756447792053, "learning_rate": 1.9962141883487374e-05, "loss": 3.7877, "step": 11310 }, { "epoch": 3.00110848, "grad_norm": 0.8422288298606873, "learning_rate": 1.996207184617011e-05, "loss": 3.7609, "step": 11320 }, { "epoch": 3.00113408, "grad_norm": 0.9385292530059814, "learning_rate": 1.9962001744251454e-05, "loss": 3.9279, "step": 11330 }, { "epoch": 3.00115968, "grad_norm": 0.9071043729782104, "learning_rate": 1.996193157773187e-05, "loss": 3.727, "step": 11340 }, { "epoch": 3.00118528, "grad_norm": 0.9645744562149048, "learning_rate": 1.9961861346611804e-05, "loss": 3.9643, "step": 11350 }, { "epoch": 3.00121088, "grad_norm": 0.7123860120773315, "learning_rate": 1.9961791050891716e-05, "loss": 3.6856, "step": 11360 }, { "epoch": 3.00123648, "grad_norm": 1.0869303941726685, "learning_rate": 1.996172069057206e-05, "loss": 3.8746, "step": 11370 }, { "epoch": 3.00126208, "grad_norm": 1.0264387130737305, "learning_rate": 1.9961650265653295e-05, "loss": 3.8967, "step": 11380 }, { "epoch": 3.00128768, "grad_norm": 0.8086841106414795, "learning_rate": 1.9961579776135874e-05, "loss": 3.9434, "step": 11390 }, { "epoch": 3.00131328, "grad_norm": 0.8274354934692383, "learning_rate": 1.9961509222020254e-05, "loss": 3.8607, "step": 11400 }, { "epoch": 3.00133888, "grad_norm": 0.7695878148078918, "learning_rate": 1.9961438603306897e-05, "loss": 3.8933, "step": 11410 }, { "epoch": 3.00136448, "grad_norm": 1.436249852180481, "learning_rate": 1.9961367919996255e-05, "loss": 3.9845, "step": 11420 }, { "epoch": 3.00139008, "grad_norm": 1.0147266387939453, "learning_rate": 1.996129717208879e-05, "loss": 3.7502, "step": 11430 }, { "epoch": 3.00141568, "grad_norm": 0.9594259858131409, "learning_rate": 1.9961226359584963e-05, "loss": 4.0295, "step": 11440 }, { "epoch": 3.00144128, "grad_norm": 0.8903603553771973, "learning_rate": 1.996115548248523e-05, "loss": 3.9278, "step": 11450 }, { "epoch": 3.00146688, "grad_norm": 1.0091084241867065, "learning_rate": 1.996108454079005e-05, "loss": 4.1576, "step": 11460 }, { "epoch": 3.00149248, "grad_norm": 0.9298844933509827, "learning_rate": 1.9961013534499883e-05, "loss": 4.0557, "step": 11470 }, { "epoch": 3.00151808, "grad_norm": 0.8303804397583008, "learning_rate": 1.9960942463615192e-05, "loss": 3.8613, "step": 11480 }, { "epoch": 3.00154368, "grad_norm": 0.8433350920677185, "learning_rate": 1.9960871328136435e-05, "loss": 3.9161, "step": 11490 }, { "epoch": 3.00156928, "grad_norm": 0.7748315930366516, "learning_rate": 1.9960800128064074e-05, "loss": 4.2418, "step": 11500 }, { "epoch": 3.00159488, "grad_norm": 0.9766426682472229, "learning_rate": 1.9960728863398576e-05, "loss": 4.1079, "step": 11510 }, { "epoch": 3.00162048, "grad_norm": 1.2121572494506836, "learning_rate": 1.9960657534140395e-05, "loss": 4.152, "step": 11520 }, { "epoch": 3.00164608, "grad_norm": 1.1307786703109741, "learning_rate": 1.9960586140289996e-05, "loss": 3.9425, "step": 11530 }, { "epoch": 3.00167168, "grad_norm": 0.8795030117034912, "learning_rate": 1.9960514681847846e-05, "loss": 4.1451, "step": 11540 }, { "epoch": 3.00169728, "grad_norm": 0.83788001537323, "learning_rate": 1.9960443158814402e-05, "loss": 4.1634, "step": 11550 }, { "epoch": 3.00172288, "grad_norm": 1.0129342079162598, "learning_rate": 1.9960371571190132e-05, "loss": 4.1301, "step": 11560 }, { "epoch": 3.00174848, "grad_norm": 0.8425036072731018, "learning_rate": 1.9960299918975504e-05, "loss": 4.1058, "step": 11570 }, { "epoch": 3.00177408, "grad_norm": 1.1863882541656494, "learning_rate": 1.9960228202170975e-05, "loss": 4.2026, "step": 11580 }, { "epoch": 3.00179968, "grad_norm": 1.2003675699234009, "learning_rate": 1.9960156420777012e-05, "loss": 4.3716, "step": 11590 }, { "epoch": 3.00182528, "grad_norm": 0.9542918801307678, "learning_rate": 1.9960084574794086e-05, "loss": 4.0271, "step": 11600 }, { "epoch": 3.00185088, "grad_norm": 1.1054574251174927, "learning_rate": 1.9960012664222656e-05, "loss": 3.921, "step": 11610 }, { "epoch": 3.00187648, "grad_norm": 1.2468104362487793, "learning_rate": 1.9959940689063188e-05, "loss": 3.8903, "step": 11620 }, { "epoch": 3.00190208, "grad_norm": 1.185952067375183, "learning_rate": 1.9959868649316156e-05, "loss": 3.9562, "step": 11630 }, { "epoch": 3.00192768, "grad_norm": 0.9449431896209717, "learning_rate": 1.9959796544982022e-05, "loss": 3.7367, "step": 11640 }, { "epoch": 3.00195328, "grad_norm": 0.8541404604911804, "learning_rate": 1.9959724376061253e-05, "loss": 3.9017, "step": 11650 }, { "epoch": 3.00197888, "grad_norm": 1.2723145484924316, "learning_rate": 1.9959652142554317e-05, "loss": 4.0108, "step": 11660 }, { "epoch": 3.00200448, "grad_norm": 0.8668487071990967, "learning_rate": 1.9959579844461684e-05, "loss": 4.1414, "step": 11670 }, { "epoch": 3.00203008, "grad_norm": 0.7450346350669861, "learning_rate": 1.9959507481783826e-05, "loss": 3.7108, "step": 11680 }, { "epoch": 3.00205568, "grad_norm": 1.0057414770126343, "learning_rate": 1.9959435054521207e-05, "loss": 3.8642, "step": 11690 }, { "epoch": 3.00208128, "grad_norm": 1.0977948904037476, "learning_rate": 1.99593625626743e-05, "loss": 4.1624, "step": 11700 }, { "epoch": 3.00210688, "grad_norm": 0.7821947932243347, "learning_rate": 1.995929000624357e-05, "loss": 4.0261, "step": 11710 }, { "epoch": 3.00213248, "grad_norm": 0.9264775514602661, "learning_rate": 1.9959217385229497e-05, "loss": 4.0041, "step": 11720 }, { "epoch": 3.00215808, "grad_norm": 0.8575800657272339, "learning_rate": 1.9959144699632542e-05, "loss": 4.0052, "step": 11730 }, { "epoch": 3.00218368, "grad_norm": 0.9569883942604065, "learning_rate": 1.995907194945318e-05, "loss": 3.811, "step": 11740 }, { "epoch": 3.00220928, "grad_norm": 1.0128107070922852, "learning_rate": 1.9958999134691886e-05, "loss": 3.8885, "step": 11750 }, { "epoch": 3.00223488, "grad_norm": 1.1528242826461792, "learning_rate": 1.9958926255349128e-05, "loss": 3.9575, "step": 11760 }, { "epoch": 3.00226048, "grad_norm": 1.0161973237991333, "learning_rate": 1.995885331142538e-05, "loss": 4.1533, "step": 11770 }, { "epoch": 3.00228608, "grad_norm": 0.7657971978187561, "learning_rate": 1.9958780302921115e-05, "loss": 3.9656, "step": 11780 }, { "epoch": 3.00231168, "grad_norm": 0.861304759979248, "learning_rate": 1.9958707229836808e-05, "loss": 3.9261, "step": 11790 }, { "epoch": 3.00233728, "grad_norm": 1.0185269117355347, "learning_rate": 1.9958634092172928e-05, "loss": 3.9505, "step": 11800 }, { "epoch": 3.00236288, "grad_norm": 0.8617456555366516, "learning_rate": 1.9958560889929955e-05, "loss": 4.2117, "step": 11810 }, { "epoch": 3.00238848, "grad_norm": 1.0911873579025269, "learning_rate": 1.995848762310836e-05, "loss": 3.9766, "step": 11820 }, { "epoch": 3.00241408, "grad_norm": 0.8752987384796143, "learning_rate": 1.995841429170862e-05, "loss": 3.9728, "step": 11830 }, { "epoch": 3.00243968, "grad_norm": 1.4432207345962524, "learning_rate": 1.995834089573121e-05, "loss": 4.0073, "step": 11840 }, { "epoch": 3.00246528, "grad_norm": 1.2683109045028687, "learning_rate": 1.9958267435176605e-05, "loss": 3.9082, "step": 11850 }, { "epoch": 3.00249088, "grad_norm": 0.9003625512123108, "learning_rate": 1.9958193910045286e-05, "loss": 3.8843, "step": 11860 }, { "epoch": 3.00251648, "grad_norm": 0.9356957077980042, "learning_rate": 1.9958120320337723e-05, "loss": 4.1422, "step": 11870 }, { "epoch": 3.00254208, "grad_norm": 0.8941216468811035, "learning_rate": 1.9958046666054394e-05, "loss": 3.8772, "step": 11880 }, { "epoch": 3.00256768, "grad_norm": 0.7995363473892212, "learning_rate": 1.9957972947195784e-05, "loss": 3.9042, "step": 11890 }, { "epoch": 3.00259328, "grad_norm": 1.344509243965149, "learning_rate": 1.9957899163762363e-05, "loss": 3.8499, "step": 11900 }, { "epoch": 3.00261888, "grad_norm": 0.7531570196151733, "learning_rate": 1.995782531575461e-05, "loss": 4.0137, "step": 11910 }, { "epoch": 3.00264448, "grad_norm": 1.0523091554641724, "learning_rate": 1.995775140317301e-05, "loss": 4.0619, "step": 11920 }, { "epoch": 3.00267008, "grad_norm": 0.881079912185669, "learning_rate": 1.9957677426018037e-05, "loss": 3.9599, "step": 11930 }, { "epoch": 3.00269568, "grad_norm": 0.8269854187965393, "learning_rate": 1.9957603384290173e-05, "loss": 4.0455, "step": 11940 }, { "epoch": 3.00272128, "grad_norm": 0.8793949484825134, "learning_rate": 1.9957529277989897e-05, "loss": 4.33, "step": 11950 }, { "epoch": 3.00274688, "grad_norm": 1.1198636293411255, "learning_rate": 1.995745510711769e-05, "loss": 4.0846, "step": 11960 }, { "epoch": 3.00277248, "grad_norm": 0.978986382484436, "learning_rate": 1.9957380871674035e-05, "loss": 3.8366, "step": 11970 }, { "epoch": 3.00279808, "grad_norm": 0.791567862033844, "learning_rate": 1.9957306571659408e-05, "loss": 4.0529, "step": 11980 }, { "epoch": 3.00282368, "grad_norm": 0.9781219959259033, "learning_rate": 1.9957232207074297e-05, "loss": 3.8511, "step": 11990 }, { "epoch": 3.00284928, "grad_norm": 0.7860384583473206, "learning_rate": 1.995715777791918e-05, "loss": 3.7269, "step": 12000 }, { "epoch": 3.00287488, "grad_norm": 0.8838666081428528, "learning_rate": 1.9957083284194542e-05, "loss": 3.8108, "step": 12010 }, { "epoch": 3.00290048, "grad_norm": 0.9202473163604736, "learning_rate": 1.9957008725900863e-05, "loss": 3.8959, "step": 12020 }, { "epoch": 3.00292608, "grad_norm": 0.737223207950592, "learning_rate": 1.9956934103038632e-05, "loss": 4.1533, "step": 12030 }, { "epoch": 3.00295168, "grad_norm": 0.9737456440925598, "learning_rate": 1.995685941560833e-05, "loss": 4.0005, "step": 12040 }, { "epoch": 3.00297728, "grad_norm": 0.9494034051895142, "learning_rate": 1.9956784663610437e-05, "loss": 3.9239, "step": 12050 }, { "epoch": 3.00300288, "grad_norm": 1.000467300415039, "learning_rate": 1.9956709847045442e-05, "loss": 3.7447, "step": 12060 }, { "epoch": 3.0030284800000002, "grad_norm": 1.1412839889526367, "learning_rate": 1.9956634965913833e-05, "loss": 3.9879, "step": 12070 }, { "epoch": 3.00305408, "grad_norm": 0.9798611998558044, "learning_rate": 1.995656002021609e-05, "loss": 4.2152, "step": 12080 }, { "epoch": 3.00307968, "grad_norm": 1.3698606491088867, "learning_rate": 1.9956485009952705e-05, "loss": 4.0441, "step": 12090 }, { "epoch": 3.00310528, "grad_norm": 0.7935123443603516, "learning_rate": 1.9956409935124156e-05, "loss": 4.0168, "step": 12100 }, { "epoch": 3.00313088, "grad_norm": 0.8677695989608765, "learning_rate": 1.9956334795730937e-05, "loss": 4.0084, "step": 12110 }, { "epoch": 3.00315648, "grad_norm": 0.7981150150299072, "learning_rate": 1.9956259591773533e-05, "loss": 4.0633, "step": 12120 }, { "epoch": 3.00318208, "grad_norm": 1.2528266906738281, "learning_rate": 1.995618432325243e-05, "loss": 4.0183, "step": 12130 }, { "epoch": 3.00320768, "grad_norm": 1.0359960794448853, "learning_rate": 1.995610899016812e-05, "loss": 4.021, "step": 12140 }, { "epoch": 3.00323328, "grad_norm": 0.9183046817779541, "learning_rate": 1.9956033592521085e-05, "loss": 3.8136, "step": 12150 }, { "epoch": 3.00325888, "grad_norm": 0.8830884695053101, "learning_rate": 1.9955958130311823e-05, "loss": 4.0401, "step": 12160 }, { "epoch": 3.00328448, "grad_norm": 0.9395894408226013, "learning_rate": 1.9955882603540818e-05, "loss": 3.8726, "step": 12170 }, { "epoch": 3.00331008, "grad_norm": 0.8060921430587769, "learning_rate": 1.995580701220856e-05, "loss": 4.0349, "step": 12180 }, { "epoch": 3.00333568, "grad_norm": 0.9239911437034607, "learning_rate": 1.9955731356315537e-05, "loss": 3.9648, "step": 12190 }, { "epoch": 3.00336128, "grad_norm": 0.8534404039382935, "learning_rate": 1.9955655635862246e-05, "loss": 3.8682, "step": 12200 }, { "epoch": 3.00338688, "grad_norm": 0.9079722166061401, "learning_rate": 1.995557985084917e-05, "loss": 3.9257, "step": 12210 }, { "epoch": 3.00341248, "grad_norm": 1.0810356140136719, "learning_rate": 1.9955504001276807e-05, "loss": 3.8941, "step": 12220 }, { "epoch": 3.00343808, "grad_norm": 0.7915281653404236, "learning_rate": 1.9955428087145645e-05, "loss": 3.9112, "step": 12230 }, { "epoch": 3.00346368, "grad_norm": 0.8903862833976746, "learning_rate": 1.995535210845618e-05, "loss": 4.0683, "step": 12240 }, { "epoch": 3.00348928, "grad_norm": 0.76449054479599, "learning_rate": 1.99552760652089e-05, "loss": 3.9681, "step": 12250 }, { "epoch": 3.00351488, "grad_norm": 1.167258858680725, "learning_rate": 1.99551999574043e-05, "loss": 4.3086, "step": 12260 }, { "epoch": 3.00354048, "grad_norm": 0.9345973134040833, "learning_rate": 1.9955123785042876e-05, "loss": 3.8102, "step": 12270 }, { "epoch": 3.00356608, "grad_norm": 0.7033010125160217, "learning_rate": 1.995504754812512e-05, "loss": 3.7888, "step": 12280 }, { "epoch": 3.00359168, "grad_norm": 0.9747347235679626, "learning_rate": 1.9954971246651524e-05, "loss": 4.0228, "step": 12290 }, { "epoch": 3.00361728, "grad_norm": 0.8873440027236938, "learning_rate": 1.995489488062259e-05, "loss": 3.9597, "step": 12300 }, { "epoch": 3.00364288, "grad_norm": 0.9335889220237732, "learning_rate": 1.9954818450038803e-05, "loss": 3.9932, "step": 12310 }, { "epoch": 3.00366848, "grad_norm": 2.156494379043579, "learning_rate": 1.9954741954900668e-05, "loss": 3.9874, "step": 12320 }, { "epoch": 3.00369408, "grad_norm": 0.9449342489242554, "learning_rate": 1.995466539520867e-05, "loss": 4.0576, "step": 12330 }, { "epoch": 3.00371968, "grad_norm": 0.9418455958366394, "learning_rate": 1.995458877096332e-05, "loss": 3.7712, "step": 12340 }, { "epoch": 3.00374528, "grad_norm": 0.8273712992668152, "learning_rate": 1.9954512082165106e-05, "loss": 4.1615, "step": 12350 }, { "epoch": 3.00377088, "grad_norm": 0.744415283203125, "learning_rate": 1.9954435328814527e-05, "loss": 4.1666, "step": 12360 }, { "epoch": 3.00379648, "grad_norm": 0.8427621722221375, "learning_rate": 1.995435851091208e-05, "loss": 3.8832, "step": 12370 }, { "epoch": 3.00382208, "grad_norm": 1.1089874505996704, "learning_rate": 1.9954281628458265e-05, "loss": 4.0693, "step": 12380 }, { "epoch": 3.00384768, "grad_norm": 0.9293838739395142, "learning_rate": 1.9954204681453575e-05, "loss": 3.9592, "step": 12390 }, { "epoch": 3.00387328, "grad_norm": 0.8952776789665222, "learning_rate": 1.9954127669898517e-05, "loss": 3.6737, "step": 12400 }, { "epoch": 3.00389888, "grad_norm": 0.8744497895240784, "learning_rate": 1.9954050593793585e-05, "loss": 3.8611, "step": 12410 }, { "epoch": 3.00392448, "grad_norm": 0.9772109389305115, "learning_rate": 1.9953973453139278e-05, "loss": 4.1061, "step": 12420 }, { "epoch": 3.00395008, "grad_norm": 1.2433018684387207, "learning_rate": 1.9953896247936103e-05, "loss": 3.9867, "step": 12430 }, { "epoch": 3.00397568, "grad_norm": 1.1897910833358765, "learning_rate": 1.9953818978184555e-05, "loss": 4.027, "step": 12440 }, { "epoch": 3.00400128, "grad_norm": 0.77093905210495, "learning_rate": 1.9953741643885133e-05, "loss": 3.9177, "step": 12450 }, { "epoch": 3.00402688, "grad_norm": 2.185033082962036, "learning_rate": 1.9953664245038348e-05, "loss": 3.876, "step": 12460 }, { "epoch": 3.00405248, "grad_norm": 0.7008191347122192, "learning_rate": 1.995358678164469e-05, "loss": 4.0418, "step": 12470 }, { "epoch": 3.00407808, "grad_norm": 1.0183168649673462, "learning_rate": 1.995350925370467e-05, "loss": 3.9212, "step": 12480 }, { "epoch": 3.00410368, "grad_norm": 0.8238866329193115, "learning_rate": 1.9953431661218787e-05, "loss": 3.9591, "step": 12490 }, { "epoch": 3.00412928, "grad_norm": 0.8508768677711487, "learning_rate": 1.9953354004187543e-05, "loss": 4.0029, "step": 12500 }, { "epoch": 3.00415488, "grad_norm": 0.8225436210632324, "learning_rate": 1.9953276282611447e-05, "loss": 4.0006, "step": 12510 }, { "epoch": 3.00418048, "grad_norm": 1.5737806558609009, "learning_rate": 1.9953198496490996e-05, "loss": 3.862, "step": 12520 }, { "epoch": 3.00420608, "grad_norm": 0.9079727530479431, "learning_rate": 1.9953120645826703e-05, "loss": 3.8501, "step": 12530 }, { "epoch": 3.00423168, "grad_norm": 0.7727810740470886, "learning_rate": 1.9953042730619065e-05, "loss": 3.9302, "step": 12540 }, { "epoch": 3.00425728, "grad_norm": 0.8380152583122253, "learning_rate": 1.9952964750868587e-05, "loss": 3.7655, "step": 12550 }, { "epoch": 3.00428288, "grad_norm": 0.8117395639419556, "learning_rate": 1.9952886706575784e-05, "loss": 4.0629, "step": 12560 }, { "epoch": 3.00430848, "grad_norm": 0.8674809336662292, "learning_rate": 1.995280859774115e-05, "loss": 4.1244, "step": 12570 }, { "epoch": 3.00433408, "grad_norm": 0.7271130084991455, "learning_rate": 1.9952730424365202e-05, "loss": 3.7911, "step": 12580 }, { "epoch": 3.00435968, "grad_norm": 0.8289948105812073, "learning_rate": 1.995265218644844e-05, "loss": 4.1881, "step": 12590 }, { "epoch": 3.00438528, "grad_norm": 0.8445754647254944, "learning_rate": 1.9952573883991375e-05, "loss": 4.0464, "step": 12600 }, { "epoch": 3.00441088, "grad_norm": 0.7838475108146667, "learning_rate": 1.9952495516994513e-05, "loss": 3.7667, "step": 12610 }, { "epoch": 3.00443648, "grad_norm": 1.2723032236099243, "learning_rate": 1.995241708545836e-05, "loss": 3.6603, "step": 12620 }, { "epoch": 3.00446208, "grad_norm": 1.2298706769943237, "learning_rate": 1.995233858938343e-05, "loss": 3.8549, "step": 12630 }, { "epoch": 3.00448768, "grad_norm": 0.9978689551353455, "learning_rate": 1.995226002877023e-05, "loss": 3.9559, "step": 12640 }, { "epoch": 3.00451328, "grad_norm": 0.8929462432861328, "learning_rate": 1.9952181403619266e-05, "loss": 3.9371, "step": 12650 }, { "epoch": 3.00453888, "grad_norm": 0.9111811518669128, "learning_rate": 1.995210271393105e-05, "loss": 3.888, "step": 12660 }, { "epoch": 3.00456448, "grad_norm": 0.8788439035415649, "learning_rate": 1.9952023959706098e-05, "loss": 4.0524, "step": 12670 }, { "epoch": 3.00459008, "grad_norm": 0.7061552405357361, "learning_rate": 1.995194514094491e-05, "loss": 4.0271, "step": 12680 }, { "epoch": 3.00461568, "grad_norm": 0.7755682468414307, "learning_rate": 1.9951866257648007e-05, "loss": 3.855, "step": 12690 }, { "epoch": 3.00464128, "grad_norm": 0.8610135912895203, "learning_rate": 1.9951787309815892e-05, "loss": 3.9052, "step": 12700 }, { "epoch": 3.00466688, "grad_norm": 1.4150333404541016, "learning_rate": 1.9951708297449083e-05, "loss": 4.0863, "step": 12710 }, { "epoch": 3.00469248, "grad_norm": 0.8540485501289368, "learning_rate": 1.995162922054809e-05, "loss": 3.9594, "step": 12720 }, { "epoch": 3.00471808, "grad_norm": 0.7379932403564453, "learning_rate": 1.995155007911343e-05, "loss": 3.7689, "step": 12730 }, { "epoch": 3.00474368, "grad_norm": 0.9036224484443665, "learning_rate": 1.995147087314561e-05, "loss": 4.0375, "step": 12740 }, { "epoch": 3.00476928, "grad_norm": 0.9185813665390015, "learning_rate": 1.9951391602645147e-05, "loss": 4.0037, "step": 12750 }, { "epoch": 3.00479488, "grad_norm": 0.7582453489303589, "learning_rate": 1.995131226761255e-05, "loss": 3.8232, "step": 12760 }, { "epoch": 3.00482048, "grad_norm": 0.9126418232917786, "learning_rate": 1.9951232868048343e-05, "loss": 4.1025, "step": 12770 }, { "epoch": 3.00484608, "grad_norm": 0.8733500242233276, "learning_rate": 1.9951153403953032e-05, "loss": 3.8191, "step": 12780 }, { "epoch": 3.00487168, "grad_norm": 0.9422423839569092, "learning_rate": 1.995107387532714e-05, "loss": 4.0346, "step": 12790 }, { "epoch": 3.00489728, "grad_norm": 0.759638786315918, "learning_rate": 1.9950994282171177e-05, "loss": 3.9067, "step": 12800 }, { "epoch": 3.00492288, "grad_norm": 0.8426761627197266, "learning_rate": 1.995091462448566e-05, "loss": 3.9691, "step": 12810 }, { "epoch": 3.00494848, "grad_norm": 0.817086935043335, "learning_rate": 1.9950834902271105e-05, "loss": 4.2048, "step": 12820 }, { "epoch": 3.00497408, "grad_norm": 0.8225027918815613, "learning_rate": 1.995075511552803e-05, "loss": 4.084, "step": 12830 }, { "epoch": 3.00499968, "grad_norm": 0.8323299884796143, "learning_rate": 1.9950675264256953e-05, "loss": 3.9406, "step": 12840 }, { "epoch": 3.00502528, "grad_norm": 0.8804003000259399, "learning_rate": 1.9950595348458392e-05, "loss": 3.8404, "step": 12850 }, { "epoch": 3.00505088, "grad_norm": 0.741224467754364, "learning_rate": 1.9950515368132864e-05, "loss": 3.8685, "step": 12860 }, { "epoch": 3.00507648, "grad_norm": 0.7436128854751587, "learning_rate": 1.9950435323280892e-05, "loss": 3.8737, "step": 12870 }, { "epoch": 3.00510208, "grad_norm": 0.9320249557495117, "learning_rate": 1.9950355213902987e-05, "loss": 3.7712, "step": 12880 }, { "epoch": 3.00512768, "grad_norm": 0.9468958973884583, "learning_rate": 1.9950275039999677e-05, "loss": 3.8527, "step": 12890 }, { "epoch": 3.00515328, "grad_norm": 0.8870536684989929, "learning_rate": 1.9950194801571477e-05, "loss": 4.0715, "step": 12900 }, { "epoch": 3.00517888, "grad_norm": 0.8759522438049316, "learning_rate": 1.9950114498618903e-05, "loss": 4.1955, "step": 12910 }, { "epoch": 3.00520448, "grad_norm": 0.8924689292907715, "learning_rate": 1.995003413114249e-05, "loss": 3.9137, "step": 12920 }, { "epoch": 3.00523008, "grad_norm": 0.8912957906723022, "learning_rate": 1.9949953699142745e-05, "loss": 4.0356, "step": 12930 }, { "epoch": 3.00525568, "grad_norm": 0.8212257623672485, "learning_rate": 1.994987320262019e-05, "loss": 3.876, "step": 12940 }, { "epoch": 3.00528128, "grad_norm": 0.8865357041358948, "learning_rate": 1.994979264157536e-05, "loss": 3.9137, "step": 12950 }, { "epoch": 3.00530688, "grad_norm": 1.277769923210144, "learning_rate": 1.9949712016008763e-05, "loss": 4.0957, "step": 12960 }, { "epoch": 3.00533248, "grad_norm": 0.9642605781555176, "learning_rate": 1.994963132592093e-05, "loss": 3.8613, "step": 12970 }, { "epoch": 3.00535808, "grad_norm": 1.0290827751159668, "learning_rate": 1.9949550571312382e-05, "loss": 3.9346, "step": 12980 }, { "epoch": 3.00538368, "grad_norm": 0.7831140160560608, "learning_rate": 1.9949469752183642e-05, "loss": 3.8045, "step": 12990 }, { "epoch": 3.00540928, "grad_norm": 0.7689014077186584, "learning_rate": 1.9949388868535234e-05, "loss": 4.025, "step": 13000 }, { "epoch": 3.00543488, "grad_norm": 0.7501690983772278, "learning_rate": 1.994930792036769e-05, "loss": 3.9171, "step": 13010 }, { "epoch": 3.00546048, "grad_norm": 0.8133417963981628, "learning_rate": 1.994922690768152e-05, "loss": 3.7457, "step": 13020 }, { "epoch": 3.00548608, "grad_norm": 0.7648096680641174, "learning_rate": 1.9949145830477263e-05, "loss": 3.9124, "step": 13030 }, { "epoch": 3.00551168, "grad_norm": 0.9320484399795532, "learning_rate": 1.994906468875544e-05, "loss": 3.8452, "step": 13040 }, { "epoch": 3.00553728, "grad_norm": 1.0445493459701538, "learning_rate": 1.9948983482516572e-05, "loss": 3.9679, "step": 13050 }, { "epoch": 3.00556288, "grad_norm": 0.9582853317260742, "learning_rate": 1.994890221176119e-05, "loss": 4.0515, "step": 13060 }, { "epoch": 3.00558848, "grad_norm": 1.0535918474197388, "learning_rate": 1.9948820876489825e-05, "loss": 4.0868, "step": 13070 }, { "epoch": 3.00561408, "grad_norm": 1.1952611207962036, "learning_rate": 1.9948739476703e-05, "loss": 3.8033, "step": 13080 }, { "epoch": 3.00563968, "grad_norm": 0.8020148277282715, "learning_rate": 1.9948658012401246e-05, "loss": 3.9178, "step": 13090 }, { "epoch": 3.00566528, "grad_norm": 0.8438043594360352, "learning_rate": 1.9948576483585084e-05, "loss": 4.014, "step": 13100 }, { "epoch": 3.00569088, "grad_norm": 0.7540832161903381, "learning_rate": 1.994849489025505e-05, "loss": 3.9498, "step": 13110 }, { "epoch": 3.00571648, "grad_norm": 0.9377344846725464, "learning_rate": 1.994841323241167e-05, "loss": 3.6952, "step": 13120 }, { "epoch": 3.00574208, "grad_norm": 0.7926346063613892, "learning_rate": 1.9948331510055473e-05, "loss": 3.6905, "step": 13130 }, { "epoch": 3.00576768, "grad_norm": 1.2715641260147095, "learning_rate": 1.9948249723186993e-05, "loss": 4.0537, "step": 13140 }, { "epoch": 3.00579328, "grad_norm": 0.7604348659515381, "learning_rate": 1.9948167871806756e-05, "loss": 4.0431, "step": 13150 }, { "epoch": 3.00581888, "grad_norm": 0.8621121644973755, "learning_rate": 1.9948085955915294e-05, "loss": 4.0745, "step": 13160 }, { "epoch": 3.00584448, "grad_norm": 1.0033868551254272, "learning_rate": 1.994800397551314e-05, "loss": 4.0271, "step": 13170 }, { "epoch": 3.00587008, "grad_norm": 0.939855694770813, "learning_rate": 1.994792193060082e-05, "loss": 3.8356, "step": 13180 }, { "epoch": 3.00589568, "grad_norm": 1.0410797595977783, "learning_rate": 1.9947839821178875e-05, "loss": 3.781, "step": 13190 }, { "epoch": 3.00592128, "grad_norm": 0.9087957739830017, "learning_rate": 1.994775764724783e-05, "loss": 4.2866, "step": 13200 }, { "epoch": 3.00594688, "grad_norm": 0.8811240196228027, "learning_rate": 1.994767540880822e-05, "loss": 4.2553, "step": 13210 }, { "epoch": 3.00597248, "grad_norm": 0.7887601256370544, "learning_rate": 1.994759310586058e-05, "loss": 4.0997, "step": 13220 }, { "epoch": 3.00599808, "grad_norm": 1.1138241291046143, "learning_rate": 1.9947510738405443e-05, "loss": 4.0086, "step": 13230 }, { "epoch": 3.00602368, "grad_norm": 0.793352484703064, "learning_rate": 1.994742830644334e-05, "loss": 4.1477, "step": 13240 }, { "epoch": 3.00604928, "grad_norm": 0.8231594562530518, "learning_rate": 1.9947345809974813e-05, "loss": 3.934, "step": 13250 }, { "epoch": 3.00607488, "grad_norm": 0.7643166780471802, "learning_rate": 1.9947263249000385e-05, "loss": 3.9715, "step": 13260 }, { "epoch": 3.00610048, "grad_norm": 0.8650283813476562, "learning_rate": 1.9947180623520603e-05, "loss": 3.9268, "step": 13270 }, { "epoch": 3.00612608, "grad_norm": 0.8343472480773926, "learning_rate": 1.9947097933535997e-05, "loss": 4.0096, "step": 13280 }, { "epoch": 3.00615168, "grad_norm": 0.9637627601623535, "learning_rate": 1.9947015179047107e-05, "loss": 3.9988, "step": 13290 }, { "epoch": 3.00617728, "grad_norm": 0.8404972553253174, "learning_rate": 1.9946932360054466e-05, "loss": 3.9459, "step": 13300 }, { "epoch": 3.00620288, "grad_norm": 0.7746430039405823, "learning_rate": 1.994684947655861e-05, "loss": 3.8718, "step": 13310 }, { "epoch": 3.00622848, "grad_norm": 0.9885214567184448, "learning_rate": 1.9946766528560083e-05, "loss": 3.9469, "step": 13320 }, { "epoch": 3.00625408, "grad_norm": 1.1547094583511353, "learning_rate": 1.9946683516059414e-05, "loss": 3.8823, "step": 13330 }, { "epoch": 3.00627968, "grad_norm": 0.8344938158988953, "learning_rate": 1.994660043905715e-05, "loss": 3.8867, "step": 13340 }, { "epoch": 3.00630528, "grad_norm": 0.8415545225143433, "learning_rate": 1.9946517297553824e-05, "loss": 4.1147, "step": 13350 }, { "epoch": 3.00633088, "grad_norm": 1.0128268003463745, "learning_rate": 1.9946434091549973e-05, "loss": 3.9744, "step": 13360 }, { "epoch": 3.00635648, "grad_norm": 1.0607918500900269, "learning_rate": 1.9946350821046145e-05, "loss": 4.0129, "step": 13370 }, { "epoch": 3.00638208, "grad_norm": 1.2268115282058716, "learning_rate": 1.9946267486042873e-05, "loss": 3.834, "step": 13380 }, { "epoch": 3.00640768, "grad_norm": 0.8229510188102722, "learning_rate": 1.99461840865407e-05, "loss": 3.9121, "step": 13390 }, { "epoch": 3.00643328, "grad_norm": 0.8959113955497742, "learning_rate": 1.994610062254017e-05, "loss": 3.8147, "step": 13400 }, { "epoch": 3.00645888, "grad_norm": 0.8292179107666016, "learning_rate": 1.9946017094041818e-05, "loss": 3.7634, "step": 13410 }, { "epoch": 3.00648448, "grad_norm": 0.8688445687294006, "learning_rate": 1.9945933501046187e-05, "loss": 3.9048, "step": 13420 }, { "epoch": 3.00651008, "grad_norm": 1.1779332160949707, "learning_rate": 1.9945849843553823e-05, "loss": 3.8244, "step": 13430 }, { "epoch": 3.00653568, "grad_norm": 0.8814988136291504, "learning_rate": 1.9945766121565266e-05, "loss": 3.8447, "step": 13440 }, { "epoch": 3.00656128, "grad_norm": 1.1298606395721436, "learning_rate": 1.9945682335081055e-05, "loss": 3.804, "step": 13450 }, { "epoch": 3.00658688, "grad_norm": 0.8655868172645569, "learning_rate": 1.9945598484101744e-05, "loss": 3.9787, "step": 13460 }, { "epoch": 3.00661248, "grad_norm": 1.0341802835464478, "learning_rate": 1.9945514568627863e-05, "loss": 3.7166, "step": 13470 }, { "epoch": 3.00663808, "grad_norm": 0.8763182163238525, "learning_rate": 1.994543058865997e-05, "loss": 3.9869, "step": 13480 }, { "epoch": 3.00666368, "grad_norm": 0.9803792834281921, "learning_rate": 1.9945346544198597e-05, "loss": 3.9839, "step": 13490 }, { "epoch": 3.00668928, "grad_norm": 1.397795557975769, "learning_rate": 1.99452624352443e-05, "loss": 3.6997, "step": 13500 }, { "epoch": 3.00671488, "grad_norm": 1.0454304218292236, "learning_rate": 1.9945178261797616e-05, "loss": 3.882, "step": 13510 }, { "epoch": 3.00674048, "grad_norm": 0.8988711833953857, "learning_rate": 1.9945094023859095e-05, "loss": 3.9745, "step": 13520 }, { "epoch": 3.00676608, "grad_norm": 1.6108880043029785, "learning_rate": 1.9945009721429282e-05, "loss": 3.8669, "step": 13530 }, { "epoch": 3.00679168, "grad_norm": 1.0021897554397583, "learning_rate": 1.9944925354508724e-05, "loss": 4.1199, "step": 13540 }, { "epoch": 3.00681728, "grad_norm": 0.8454704880714417, "learning_rate": 1.9944840923097968e-05, "loss": 3.9017, "step": 13550 }, { "epoch": 3.00684288, "grad_norm": 0.8177556395530701, "learning_rate": 1.9944756427197562e-05, "loss": 3.8461, "step": 13560 }, { "epoch": 3.00686848, "grad_norm": 0.8716814517974854, "learning_rate": 1.9944671866808053e-05, "loss": 4.1461, "step": 13570 }, { "epoch": 3.00689408, "grad_norm": 0.7222035527229309, "learning_rate": 1.994458724192999e-05, "loss": 3.8909, "step": 13580 }, { "epoch": 3.00691968, "grad_norm": 0.8993253707885742, "learning_rate": 1.9944502552563923e-05, "loss": 4.1658, "step": 13590 }, { "epoch": 3.00694528, "grad_norm": 1.0112097263336182, "learning_rate": 1.9944417798710397e-05, "loss": 3.9633, "step": 13600 }, { "epoch": 3.00697088, "grad_norm": 0.7353940606117249, "learning_rate": 1.9944332980369966e-05, "loss": 4.0304, "step": 13610 }, { "epoch": 3.00699648, "grad_norm": 0.9274687767028809, "learning_rate": 1.994424809754318e-05, "loss": 4.0108, "step": 13620 }, { "epoch": 3.00702208, "grad_norm": 1.0154720544815063, "learning_rate": 1.9944163150230587e-05, "loss": 3.9954, "step": 13630 }, { "epoch": 3.00704768, "grad_norm": 1.0266696214675903, "learning_rate": 1.994407813843274e-05, "loss": 3.9174, "step": 13640 }, { "epoch": 3.00707328, "grad_norm": 1.1561099290847778, "learning_rate": 1.9943993062150187e-05, "loss": 3.9796, "step": 13650 }, { "epoch": 3.00709888, "grad_norm": 1.2091315984725952, "learning_rate": 1.994390792138348e-05, "loss": 4.1946, "step": 13660 }, { "epoch": 3.00712448, "grad_norm": 0.8918331861495972, "learning_rate": 1.9943822716133177e-05, "loss": 3.8967, "step": 13670 }, { "epoch": 3.00715008, "grad_norm": 0.7603816986083984, "learning_rate": 1.9943737446399826e-05, "loss": 4.0311, "step": 13680 }, { "epoch": 3.00717568, "grad_norm": 0.7881000638008118, "learning_rate": 1.9943652112183976e-05, "loss": 3.8038, "step": 13690 }, { "epoch": 3.00720128, "grad_norm": 0.9960216283798218, "learning_rate": 1.9943566713486192e-05, "loss": 3.9338, "step": 13700 }, { "epoch": 3.00722688, "grad_norm": 1.0993598699569702, "learning_rate": 1.9943481250307016e-05, "loss": 4.0105, "step": 13710 }, { "epoch": 3.00725248, "grad_norm": 1.007058024406433, "learning_rate": 1.994339572264701e-05, "loss": 3.7574, "step": 13720 }, { "epoch": 3.00727808, "grad_norm": 0.7340037822723389, "learning_rate": 1.994331013050672e-05, "loss": 3.7779, "step": 13730 }, { "epoch": 3.00730368, "grad_norm": 0.9182974100112915, "learning_rate": 1.994322447388671e-05, "loss": 3.9492, "step": 13740 }, { "epoch": 3.00732928, "grad_norm": 0.8984625935554504, "learning_rate": 1.994313875278753e-05, "loss": 4.0475, "step": 13750 }, { "epoch": 3.00735488, "grad_norm": 1.4739511013031006, "learning_rate": 1.994305296720974e-05, "loss": 4.4892, "step": 13760 }, { "epoch": 3.00738048, "grad_norm": 1.082923173904419, "learning_rate": 1.994296711715389e-05, "loss": 4.0043, "step": 13770 }, { "epoch": 3.00740608, "grad_norm": 0.9179307818412781, "learning_rate": 1.9942881202620543e-05, "loss": 4.0, "step": 13780 }, { "epoch": 3.00743168, "grad_norm": 0.8886575102806091, "learning_rate": 1.9942795223610255e-05, "loss": 3.8836, "step": 13790 }, { "epoch": 3.00745728, "grad_norm": 1.0669819116592407, "learning_rate": 1.994270918012358e-05, "loss": 3.9383, "step": 13800 }, { "epoch": 3.00748288, "grad_norm": 0.973188579082489, "learning_rate": 1.9942623072161082e-05, "loss": 3.8177, "step": 13810 }, { "epoch": 3.00750848, "grad_norm": 0.9746291637420654, "learning_rate": 1.9942536899723316e-05, "loss": 3.8904, "step": 13820 }, { "epoch": 3.00753408, "grad_norm": 1.146968960762024, "learning_rate": 1.9942450662810838e-05, "loss": 3.8746, "step": 13830 }, { "epoch": 3.00755968, "grad_norm": 1.2456474304199219, "learning_rate": 1.994236436142421e-05, "loss": 4.1738, "step": 13840 }, { "epoch": 3.00758528, "grad_norm": 0.9917517900466919, "learning_rate": 1.9942277995563994e-05, "loss": 3.9398, "step": 13850 }, { "epoch": 3.00761088, "grad_norm": 0.9358296990394592, "learning_rate": 1.9942191565230743e-05, "loss": 3.8201, "step": 13860 }, { "epoch": 3.00763648, "grad_norm": 1.0368461608886719, "learning_rate": 1.9942105070425027e-05, "loss": 3.9154, "step": 13870 }, { "epoch": 3.00766208, "grad_norm": 0.764778733253479, "learning_rate": 1.9942018511147398e-05, "loss": 3.9288, "step": 13880 }, { "epoch": 3.00768768, "grad_norm": 1.0407153367996216, "learning_rate": 1.9941931887398422e-05, "loss": 3.8356, "step": 13890 }, { "epoch": 3.00771328, "grad_norm": 0.768505871295929, "learning_rate": 1.994184519917866e-05, "loss": 3.7064, "step": 13900 }, { "epoch": 3.00773888, "grad_norm": 0.9166963696479797, "learning_rate": 1.9941758446488676e-05, "loss": 3.9565, "step": 13910 }, { "epoch": 3.00776448, "grad_norm": 3.2882044315338135, "learning_rate": 1.994167162932903e-05, "loss": 4.1911, "step": 13920 }, { "epoch": 3.00779008, "grad_norm": 3.275174379348755, "learning_rate": 1.9941584747700284e-05, "loss": 4.6021, "step": 13930 }, { "epoch": 3.00781568, "grad_norm": 0.8753647208213806, "learning_rate": 1.9941497801603007e-05, "loss": 3.7908, "step": 13940 }, { "epoch": 3.00784128, "grad_norm": 0.9839869737625122, "learning_rate": 1.9941410791037757e-05, "loss": 3.8953, "step": 13950 }, { "epoch": 3.00786688, "grad_norm": 1.0875120162963867, "learning_rate": 1.9941323716005097e-05, "loss": 3.9359, "step": 13960 }, { "epoch": 3.00789248, "grad_norm": 0.8477191925048828, "learning_rate": 1.99412365765056e-05, "loss": 3.7093, "step": 13970 }, { "epoch": 3.00791808, "grad_norm": 0.9057764410972595, "learning_rate": 1.994114937253982e-05, "loss": 3.9395, "step": 13980 }, { "epoch": 3.00794368, "grad_norm": 0.9378921389579773, "learning_rate": 1.9941062104108332e-05, "loss": 3.9334, "step": 13990 }, { "epoch": 3.00796928, "grad_norm": 0.8099644184112549, "learning_rate": 1.9940974771211697e-05, "loss": 3.8359, "step": 14000 }, { "epoch": 3.00799488, "grad_norm": 1.0592557191848755, "learning_rate": 1.9940887373850485e-05, "loss": 3.8654, "step": 14010 }, { "epoch": 3.00802048, "grad_norm": 0.9617739915847778, "learning_rate": 1.9940799912025258e-05, "loss": 3.8672, "step": 14020 }, { "epoch": 3.00804608, "grad_norm": 0.8320448398590088, "learning_rate": 1.9940712385736588e-05, "loss": 3.8848, "step": 14030 }, { "epoch": 3.00807168, "grad_norm": 0.9518365859985352, "learning_rate": 1.994062479498504e-05, "loss": 3.7708, "step": 14040 }, { "epoch": 3.00809728, "grad_norm": 0.8380952477455139, "learning_rate": 1.9940537139771178e-05, "loss": 3.8875, "step": 14050 }, { "epoch": 3.00812288, "grad_norm": 0.8762813210487366, "learning_rate": 1.9940449420095572e-05, "loss": 3.7734, "step": 14060 }, { "epoch": 3.00814848, "grad_norm": 0.87605220079422, "learning_rate": 1.99403616359588e-05, "loss": 3.8586, "step": 14070 }, { "epoch": 3.00817408, "grad_norm": 1.1163524389266968, "learning_rate": 1.9940273787361423e-05, "loss": 4.0457, "step": 14080 }, { "epoch": 3.00819968, "grad_norm": 0.8146365284919739, "learning_rate": 1.9940185874304015e-05, "loss": 4.0776, "step": 14090 }, { "epoch": 3.00822528, "grad_norm": 0.9406104683876038, "learning_rate": 1.9940097896787138e-05, "loss": 3.9224, "step": 14100 }, { "epoch": 3.00825088, "grad_norm": 1.3043768405914307, "learning_rate": 1.994000985481137e-05, "loss": 3.9783, "step": 14110 }, { "epoch": 3.00827648, "grad_norm": 0.8118541836738586, "learning_rate": 1.993992174837728e-05, "loss": 4.0119, "step": 14120 }, { "epoch": 3.00830208, "grad_norm": 0.8373180627822876, "learning_rate": 1.993983357748544e-05, "loss": 3.7397, "step": 14130 }, { "epoch": 3.00832768, "grad_norm": 1.2428216934204102, "learning_rate": 1.993974534213642e-05, "loss": 3.7262, "step": 14140 }, { "epoch": 3.00835328, "grad_norm": 0.7707980871200562, "learning_rate": 1.9939657042330794e-05, "loss": 3.8887, "step": 14150 }, { "epoch": 3.00837888, "grad_norm": 0.9030358791351318, "learning_rate": 1.9939568678069133e-05, "loss": 3.944, "step": 14160 }, { "epoch": 3.00840448, "grad_norm": 0.8897598385810852, "learning_rate": 1.9939480249352012e-05, "loss": 3.9626, "step": 14170 }, { "epoch": 3.00843008, "grad_norm": 0.7431142926216125, "learning_rate": 1.993939175618e-05, "loss": 3.6093, "step": 14180 }, { "epoch": 3.00845568, "grad_norm": 1.0501712560653687, "learning_rate": 1.993930319855368e-05, "loss": 3.7041, "step": 14190 }, { "epoch": 3.00848128, "grad_norm": 0.9289107322692871, "learning_rate": 1.9939214576473614e-05, "loss": 3.9505, "step": 14200 }, { "epoch": 3.00850688, "grad_norm": 0.9242643117904663, "learning_rate": 1.9939125889940386e-05, "loss": 3.7841, "step": 14210 }, { "epoch": 3.00853248, "grad_norm": 1.1005101203918457, "learning_rate": 1.9939037138954567e-05, "loss": 3.8228, "step": 14220 }, { "epoch": 3.00855808, "grad_norm": 0.9050759673118591, "learning_rate": 1.9938948323516734e-05, "loss": 4.11, "step": 14230 }, { "epoch": 3.00858368, "grad_norm": 0.8433274030685425, "learning_rate": 1.9938859443627463e-05, "loss": 3.823, "step": 14240 }, { "epoch": 3.00860928, "grad_norm": 0.8111004829406738, "learning_rate": 1.993877049928733e-05, "loss": 3.9553, "step": 14250 }, { "epoch": 3.00863488, "grad_norm": 0.9447992444038391, "learning_rate": 1.993868149049691e-05, "loss": 3.8088, "step": 14260 }, { "epoch": 3.00866048, "grad_norm": 1.2321971654891968, "learning_rate": 1.9938592417256782e-05, "loss": 3.9214, "step": 14270 }, { "epoch": 3.00868608, "grad_norm": 1.1073188781738281, "learning_rate": 1.9938503279567528e-05, "loss": 3.981, "step": 14280 }, { "epoch": 3.00871168, "grad_norm": 0.8650345206260681, "learning_rate": 1.9938414077429718e-05, "loss": 3.8029, "step": 14290 }, { "epoch": 3.00873728, "grad_norm": 1.0680748224258423, "learning_rate": 1.9938324810843933e-05, "loss": 4.2086, "step": 14300 }, { "epoch": 3.00876288, "grad_norm": 0.9072484970092773, "learning_rate": 1.993823547981076e-05, "loss": 3.812, "step": 14310 }, { "epoch": 3.00878848, "grad_norm": 0.9974015355110168, "learning_rate": 1.993814608433076e-05, "loss": 3.9389, "step": 14320 }, { "epoch": 3.00881408, "grad_norm": 0.8987821936607361, "learning_rate": 1.9938056624404528e-05, "loss": 4.0285, "step": 14330 }, { "epoch": 3.00883968, "grad_norm": 3.129868268966675, "learning_rate": 1.9937967100032644e-05, "loss": 4.0036, "step": 14340 }, { "epoch": 3.00886528, "grad_norm": 1.0388574600219727, "learning_rate": 1.993787751121568e-05, "loss": 3.8612, "step": 14350 }, { "epoch": 3.00889088, "grad_norm": 1.334452509880066, "learning_rate": 1.9937787857954226e-05, "loss": 3.7545, "step": 14360 }, { "epoch": 3.00891648, "grad_norm": 0.8655916452407837, "learning_rate": 1.9937698140248854e-05, "loss": 3.9158, "step": 14370 }, { "epoch": 3.00894208, "grad_norm": 0.7818614840507507, "learning_rate": 1.9937608358100155e-05, "loss": 3.9112, "step": 14380 }, { "epoch": 3.00896768, "grad_norm": 0.9325932264328003, "learning_rate": 1.9937518511508705e-05, "loss": 3.9413, "step": 14390 }, { "epoch": 3.00899328, "grad_norm": 0.9388124942779541, "learning_rate": 1.9937428600475087e-05, "loss": 3.6356, "step": 14400 }, { "epoch": 3.00901888, "grad_norm": 0.9659877419471741, "learning_rate": 1.9937338624999886e-05, "loss": 3.5921, "step": 14410 }, { "epoch": 3.00904448, "grad_norm": 0.8678719401359558, "learning_rate": 1.993724858508369e-05, "loss": 3.7528, "step": 14420 }, { "epoch": 3.00907008, "grad_norm": 0.7166104912757874, "learning_rate": 1.9937158480727072e-05, "loss": 3.9341, "step": 14430 }, { "epoch": 3.00909568, "grad_norm": 0.767078161239624, "learning_rate": 1.9937068311930625e-05, "loss": 3.8086, "step": 14440 }, { "epoch": 3.00912128, "grad_norm": 1.2450157403945923, "learning_rate": 1.993697807869493e-05, "loss": 3.843, "step": 14450 }, { "epoch": 3.00914688, "grad_norm": 1.5469813346862793, "learning_rate": 1.9936887781020573e-05, "loss": 4.0941, "step": 14460 }, { "epoch": 3.00917248, "grad_norm": 0.8008288741111755, "learning_rate": 1.993679741890814e-05, "loss": 3.991, "step": 14470 }, { "epoch": 3.00919808, "grad_norm": 1.4189101457595825, "learning_rate": 1.9936706992358215e-05, "loss": 4.1414, "step": 14480 }, { "epoch": 3.00922368, "grad_norm": 0.9491394758224487, "learning_rate": 1.9936616501371387e-05, "loss": 3.9098, "step": 14490 }, { "epoch": 3.00924928, "grad_norm": 0.9335387349128723, "learning_rate": 1.9936525945948245e-05, "loss": 3.879, "step": 14500 }, { "epoch": 3.00927488, "grad_norm": 0.9478955268859863, "learning_rate": 1.9936435326089372e-05, "loss": 3.8441, "step": 14510 }, { "epoch": 4.00001024, "grad_norm": 0.989231526851654, "learning_rate": 1.9936344641795354e-05, "loss": 4.2201, "step": 14520 }, { "epoch": 4.00003584, "grad_norm": 1.1565313339233398, "learning_rate": 1.9936253893066785e-05, "loss": 3.8615, "step": 14530 }, { "epoch": 4.00006144, "grad_norm": 0.6981263756752014, "learning_rate": 1.9936163079904246e-05, "loss": 3.5279, "step": 14540 }, { "epoch": 4.00008704, "grad_norm": 1.181421160697937, "learning_rate": 1.993607220230833e-05, "loss": 3.763, "step": 14550 }, { "epoch": 4.00011264, "grad_norm": 0.9807522892951965, "learning_rate": 1.993598126027963e-05, "loss": 3.7614, "step": 14560 }, { "epoch": 4.00013824, "grad_norm": 1.082722783088684, "learning_rate": 1.993589025381873e-05, "loss": 3.7783, "step": 14570 }, { "epoch": 4.00016384, "grad_norm": 1.660782814025879, "learning_rate": 1.9935799182926225e-05, "loss": 3.859, "step": 14580 }, { "epoch": 4.00018944, "grad_norm": 1.1368836164474487, "learning_rate": 1.99357080476027e-05, "loss": 4.2416, "step": 14590 }, { "epoch": 4.00021504, "grad_norm": 0.948991060256958, "learning_rate": 1.993561684784875e-05, "loss": 3.7135, "step": 14600 }, { "epoch": 4.00024064, "grad_norm": 0.9039499759674072, "learning_rate": 1.9935525583664967e-05, "loss": 4.1677, "step": 14610 }, { "epoch": 4.00026624, "grad_norm": 0.793133020401001, "learning_rate": 1.993543425505194e-05, "loss": 3.8685, "step": 14620 }, { "epoch": 4.00029184, "grad_norm": 0.9962975382804871, "learning_rate": 1.9935342862010262e-05, "loss": 3.665, "step": 14630 }, { "epoch": 4.00031744, "grad_norm": 0.8613654375076294, "learning_rate": 1.9935251404540525e-05, "loss": 3.4029, "step": 14640 }, { "epoch": 4.00034304, "grad_norm": 0.880939245223999, "learning_rate": 1.9935159882643323e-05, "loss": 3.5781, "step": 14650 }, { "epoch": 4.00036864, "grad_norm": 0.9121741056442261, "learning_rate": 1.993506829631925e-05, "loss": 3.7041, "step": 14660 }, { "epoch": 4.00039424, "grad_norm": 0.9895405769348145, "learning_rate": 1.99349766455689e-05, "loss": 3.7147, "step": 14670 }, { "epoch": 4.00041984, "grad_norm": 0.9325006604194641, "learning_rate": 1.993488493039287e-05, "loss": 3.7836, "step": 14680 }, { "epoch": 4.00044544, "grad_norm": 0.7821844816207886, "learning_rate": 1.9934793150791748e-05, "loss": 3.8381, "step": 14690 }, { "epoch": 4.00047104, "grad_norm": 0.7940314412117004, "learning_rate": 1.9934701306766135e-05, "loss": 3.7967, "step": 14700 }, { "epoch": 4.00049664, "grad_norm": 0.8940387964248657, "learning_rate": 1.9934609398316622e-05, "loss": 3.7653, "step": 14710 }, { "epoch": 4.00052224, "grad_norm": 1.4773434400558472, "learning_rate": 1.9934517425443806e-05, "loss": 3.6599, "step": 14720 }, { "epoch": 4.00054784, "grad_norm": 0.7439368367195129, "learning_rate": 1.993442538814829e-05, "loss": 3.6924, "step": 14730 }, { "epoch": 4.00057344, "grad_norm": 1.4104794263839722, "learning_rate": 1.9934333286430663e-05, "loss": 3.4321, "step": 14740 }, { "epoch": 4.00059904, "grad_norm": 0.9235467314720154, "learning_rate": 1.9934241120291525e-05, "loss": 3.5827, "step": 14750 }, { "epoch": 4.00062464, "grad_norm": 0.8314234018325806, "learning_rate": 1.993414888973147e-05, "loss": 3.6817, "step": 14760 }, { "epoch": 4.00065024, "grad_norm": 1.3974039554595947, "learning_rate": 1.9934056594751105e-05, "loss": 3.5559, "step": 14770 }, { "epoch": 4.00067584, "grad_norm": 0.8946942687034607, "learning_rate": 1.993396423535102e-05, "loss": 3.7759, "step": 14780 }, { "epoch": 4.00070144, "grad_norm": 1.089188575744629, "learning_rate": 1.9933871811531817e-05, "loss": 4.3403, "step": 14790 }, { "epoch": 4.00072704, "grad_norm": 0.9839617013931274, "learning_rate": 1.9933779323294097e-05, "loss": 3.7384, "step": 14800 }, { "epoch": 4.00075264, "grad_norm": 0.8155914545059204, "learning_rate": 1.993368677063846e-05, "loss": 3.7561, "step": 14810 }, { "epoch": 4.00077824, "grad_norm": 0.7826376557350159, "learning_rate": 1.99335941535655e-05, "loss": 3.6454, "step": 14820 }, { "epoch": 4.00080384, "grad_norm": 0.9269455671310425, "learning_rate": 1.9933501472075823e-05, "loss": 3.9696, "step": 14830 }, { "epoch": 4.00082944, "grad_norm": 0.6918569803237915, "learning_rate": 1.993340872617003e-05, "loss": 3.6713, "step": 14840 }, { "epoch": 4.00085504, "grad_norm": 2.130439519882202, "learning_rate": 1.9933315915848718e-05, "loss": 3.6858, "step": 14850 }, { "epoch": 4.00088064, "grad_norm": 0.9658183455467224, "learning_rate": 1.9933223041112496e-05, "loss": 3.6264, "step": 14860 }, { "epoch": 4.00090624, "grad_norm": 1.1685101985931396, "learning_rate": 1.9933130101961963e-05, "loss": 3.7373, "step": 14870 }, { "epoch": 4.00093184, "grad_norm": 0.9176197648048401, "learning_rate": 1.9933037098397722e-05, "loss": 3.7402, "step": 14880 }, { "epoch": 4.00095744, "grad_norm": 0.97166907787323, "learning_rate": 1.9932944030420372e-05, "loss": 3.7664, "step": 14890 }, { "epoch": 4.00098304, "grad_norm": 0.8075897097587585, "learning_rate": 1.993285089803052e-05, "loss": 3.8401, "step": 14900 }, { "epoch": 4.00100864, "grad_norm": 1.0056222677230835, "learning_rate": 1.9932757701228774e-05, "loss": 3.7545, "step": 14910 }, { "epoch": 4.00103424, "grad_norm": 0.744836688041687, "learning_rate": 1.9932664440015728e-05, "loss": 3.5837, "step": 14920 }, { "epoch": 4.00105984, "grad_norm": 0.9551469683647156, "learning_rate": 1.9932571114391996e-05, "loss": 3.5793, "step": 14930 }, { "epoch": 4.00108544, "grad_norm": 0.72246915102005, "learning_rate": 1.9932477724358178e-05, "loss": 3.6575, "step": 14940 }, { "epoch": 4.00111104, "grad_norm": 0.8650622963905334, "learning_rate": 1.9932384269914887e-05, "loss": 3.625, "step": 14950 }, { "epoch": 4.00113664, "grad_norm": 0.8650433421134949, "learning_rate": 1.993229075106272e-05, "loss": 3.7409, "step": 14960 }, { "epoch": 4.00116224, "grad_norm": 0.9449290037155151, "learning_rate": 1.9932197167802286e-05, "loss": 3.5156, "step": 14970 }, { "epoch": 4.00118784, "grad_norm": 1.114302396774292, "learning_rate": 1.9932103520134195e-05, "loss": 3.8322, "step": 14980 }, { "epoch": 4.00121344, "grad_norm": 0.7581325769424438, "learning_rate": 1.9932009808059053e-05, "loss": 3.4309, "step": 14990 }, { "epoch": 4.00123904, "grad_norm": 1.188231110572815, "learning_rate": 1.9931916031577465e-05, "loss": 3.7864, "step": 15000 }, { "epoch": 4.00126464, "grad_norm": 0.8249455094337463, "learning_rate": 1.9931822190690042e-05, "loss": 3.693, "step": 15010 }, { "epoch": 4.00129024, "grad_norm": 0.92374187707901, "learning_rate": 1.993172828539739e-05, "loss": 3.785, "step": 15020 }, { "epoch": 4.00131584, "grad_norm": 1.2868890762329102, "learning_rate": 1.9931634315700124e-05, "loss": 3.6365, "step": 15030 }, { "epoch": 4.00134144, "grad_norm": 0.8600379228591919, "learning_rate": 1.9931540281598844e-05, "loss": 3.7968, "step": 15040 }, { "epoch": 4.00136704, "grad_norm": 2.2883002758026123, "learning_rate": 1.9931446183094167e-05, "loss": 3.7784, "step": 15050 }, { "epoch": 4.00139264, "grad_norm": 0.7856008410453796, "learning_rate": 1.99313520201867e-05, "loss": 3.5409, "step": 15060 }, { "epoch": 4.00141824, "grad_norm": 0.9053830504417419, "learning_rate": 1.9931257792877058e-05, "loss": 3.8713, "step": 15070 }, { "epoch": 4.00144384, "grad_norm": 1.0639656782150269, "learning_rate": 1.9931163501165846e-05, "loss": 3.7687, "step": 15080 }, { "epoch": 4.00146944, "grad_norm": 0.9299545288085938, "learning_rate": 1.993106914505368e-05, "loss": 3.963, "step": 15090 }, { "epoch": 4.00149504, "grad_norm": 0.8577538728713989, "learning_rate": 1.9930974724541166e-05, "loss": 3.8139, "step": 15100 }, { "epoch": 4.00152064, "grad_norm": 0.8646170496940613, "learning_rate": 1.9930880239628925e-05, "loss": 3.6982, "step": 15110 }, { "epoch": 4.00154624, "grad_norm": 0.8169735074043274, "learning_rate": 1.9930785690317562e-05, "loss": 3.7978, "step": 15120 }, { "epoch": 4.00157184, "grad_norm": 0.9753700494766235, "learning_rate": 1.9930691076607693e-05, "loss": 4.0781, "step": 15130 }, { "epoch": 4.00159744, "grad_norm": 0.9531400799751282, "learning_rate": 1.9930596398499932e-05, "loss": 3.8458, "step": 15140 }, { "epoch": 4.00162304, "grad_norm": 1.1193654537200928, "learning_rate": 1.9930501655994892e-05, "loss": 3.9156, "step": 15150 }, { "epoch": 4.00164864, "grad_norm": 1.1698077917099, "learning_rate": 1.993040684909319e-05, "loss": 3.7652, "step": 15160 }, { "epoch": 4.00167424, "grad_norm": 0.9437996745109558, "learning_rate": 1.993031197779544e-05, "loss": 3.9663, "step": 15170 }, { "epoch": 4.00169984, "grad_norm": 0.9706278443336487, "learning_rate": 1.993021704210225e-05, "loss": 3.9769, "step": 15180 }, { "epoch": 4.00172544, "grad_norm": 1.056092619895935, "learning_rate": 1.9930122042014245e-05, "loss": 3.9058, "step": 15190 }, { "epoch": 4.00175104, "grad_norm": 0.8224359750747681, "learning_rate": 1.993002697753204e-05, "loss": 3.9118, "step": 15200 }, { "epoch": 4.00177664, "grad_norm": 1.112744927406311, "learning_rate": 1.9929931848656244e-05, "loss": 4.0441, "step": 15210 }, { "epoch": 4.00180224, "grad_norm": 1.4434938430786133, "learning_rate": 1.9929836655387486e-05, "loss": 4.0553, "step": 15220 }, { "epoch": 4.00182784, "grad_norm": 1.023910403251648, "learning_rate": 1.992974139772637e-05, "loss": 3.8447, "step": 15230 }, { "epoch": 4.00185344, "grad_norm": 1.3418813943862915, "learning_rate": 1.9929646075673523e-05, "loss": 3.775, "step": 15240 }, { "epoch": 4.00187904, "grad_norm": 0.9233395457267761, "learning_rate": 1.9929550689229557e-05, "loss": 3.6485, "step": 15250 }, { "epoch": 4.00190464, "grad_norm": 0.9202839136123657, "learning_rate": 1.99294552383951e-05, "loss": 3.7679, "step": 15260 }, { "epoch": 4.00193024, "grad_norm": 0.9084787964820862, "learning_rate": 1.9929359723170757e-05, "loss": 3.5626, "step": 15270 }, { "epoch": 4.00195584, "grad_norm": 1.128015160560608, "learning_rate": 1.992926414355716e-05, "loss": 3.7205, "step": 15280 }, { "epoch": 4.00198144, "grad_norm": 1.3729630708694458, "learning_rate": 1.9929168499554924e-05, "loss": 3.8489, "step": 15290 }, { "epoch": 4.00200704, "grad_norm": 1.003919243812561, "learning_rate": 1.992907279116467e-05, "loss": 3.9443, "step": 15300 }, { "epoch": 4.00203264, "grad_norm": 1.1956148147583008, "learning_rate": 1.9928977018387014e-05, "loss": 3.5346, "step": 15310 }, { "epoch": 4.00205824, "grad_norm": 0.8845064640045166, "learning_rate": 1.992888118122258e-05, "loss": 3.7181, "step": 15320 }, { "epoch": 4.00208384, "grad_norm": 0.9401364326477051, "learning_rate": 1.9928785279671997e-05, "loss": 3.938, "step": 15330 }, { "epoch": 4.00210944, "grad_norm": 0.9421001076698303, "learning_rate": 1.9928689313735875e-05, "loss": 3.8465, "step": 15340 }, { "epoch": 4.00213504, "grad_norm": 1.1207479238510132, "learning_rate": 1.992859328341484e-05, "loss": 3.8076, "step": 15350 }, { "epoch": 4.00216064, "grad_norm": 0.9041663408279419, "learning_rate": 1.992849718870952e-05, "loss": 3.8362, "step": 15360 }, { "epoch": 4.00218624, "grad_norm": 1.1898384094238281, "learning_rate": 1.9928401029620535e-05, "loss": 3.6314, "step": 15370 }, { "epoch": 4.00221184, "grad_norm": 1.0820494890213013, "learning_rate": 1.9928304806148508e-05, "loss": 3.7006, "step": 15380 }, { "epoch": 4.00223744, "grad_norm": 0.9264045357704163, "learning_rate": 1.992820851829406e-05, "loss": 3.7561, "step": 15390 }, { "epoch": 4.00226304, "grad_norm": 0.8914308547973633, "learning_rate": 1.992811216605782e-05, "loss": 3.9673, "step": 15400 }, { "epoch": 4.00228864, "grad_norm": 0.7640278935432434, "learning_rate": 1.992801574944041e-05, "loss": 3.8427, "step": 15410 }, { "epoch": 4.00231424, "grad_norm": 0.8142483234405518, "learning_rate": 1.9927919268442456e-05, "loss": 3.7491, "step": 15420 }, { "epoch": 4.00233984, "grad_norm": 1.3273457288742065, "learning_rate": 1.9927822723064586e-05, "loss": 3.7613, "step": 15430 }, { "epoch": 4.00236544, "grad_norm": 1.2759743928909302, "learning_rate": 1.9927726113307425e-05, "loss": 3.9917, "step": 15440 }, { "epoch": 4.00239104, "grad_norm": 1.0336652994155884, "learning_rate": 1.9927629439171596e-05, "loss": 3.7759, "step": 15450 }, { "epoch": 4.00241664, "grad_norm": 0.8488606810569763, "learning_rate": 1.992753270065773e-05, "loss": 3.7911, "step": 15460 }, { "epoch": 4.00244224, "grad_norm": 0.9356192946434021, "learning_rate": 1.9927435897766454e-05, "loss": 3.7994, "step": 15470 }, { "epoch": 4.00246784, "grad_norm": 0.9129524230957031, "learning_rate": 1.9927339030498395e-05, "loss": 3.7359, "step": 15480 }, { "epoch": 4.00249344, "grad_norm": 0.9089944958686829, "learning_rate": 1.9927242098854176e-05, "loss": 3.7373, "step": 15490 }, { "epoch": 4.00251904, "grad_norm": 0.9447796940803528, "learning_rate": 1.9927145102834432e-05, "loss": 3.9648, "step": 15500 }, { "epoch": 4.00254464, "grad_norm": 0.803959846496582, "learning_rate": 1.9927048042439795e-05, "loss": 3.7159, "step": 15510 }, { "epoch": 4.00257024, "grad_norm": 0.7881433963775635, "learning_rate": 1.9926950917670885e-05, "loss": 3.6338, "step": 15520 }, { "epoch": 4.00259584, "grad_norm": 0.9094733595848083, "learning_rate": 1.9926853728528337e-05, "loss": 3.7183, "step": 15530 }, { "epoch": 4.00262144, "grad_norm": 0.7680348753929138, "learning_rate": 1.9926756475012784e-05, "loss": 3.8557, "step": 15540 }, { "epoch": 4.00264704, "grad_norm": 1.1515918970108032, "learning_rate": 1.9926659157124847e-05, "loss": 3.8321, "step": 15550 }, { "epoch": 4.00267264, "grad_norm": 0.9022713899612427, "learning_rate": 1.992656177486517e-05, "loss": 3.8268, "step": 15560 }, { "epoch": 4.00269824, "grad_norm": 0.8188135623931885, "learning_rate": 1.9926464328234376e-05, "loss": 3.8752, "step": 15570 }, { "epoch": 4.00272384, "grad_norm": 0.9555391669273376, "learning_rate": 1.9926366817233098e-05, "loss": 4.1254, "step": 15580 }, { "epoch": 4.00274944, "grad_norm": 0.8849394917488098, "learning_rate": 1.9926269241861972e-05, "loss": 3.9458, "step": 15590 }, { "epoch": 4.00277504, "grad_norm": 1.046217679977417, "learning_rate": 1.9926171602121624e-05, "loss": 3.6404, "step": 15600 }, { "epoch": 4.00280064, "grad_norm": 0.9726109504699707, "learning_rate": 1.9926073898012694e-05, "loss": 3.8751, "step": 15610 }, { "epoch": 4.00282624, "grad_norm": 0.8573278784751892, "learning_rate": 1.9925976129535813e-05, "loss": 3.6402, "step": 15620 }, { "epoch": 4.00285184, "grad_norm": 1.0052928924560547, "learning_rate": 1.9925878296691614e-05, "loss": 3.544, "step": 15630 }, { "epoch": 4.00287744, "grad_norm": 0.9129832983016968, "learning_rate": 1.9925780399480728e-05, "loss": 3.6766, "step": 15640 }, { "epoch": 4.00290304, "grad_norm": 1.0524177551269531, "learning_rate": 1.99256824379038e-05, "loss": 3.7284, "step": 15650 }, { "epoch": 4.00292864, "grad_norm": 0.821373701095581, "learning_rate": 1.9925584411961458e-05, "loss": 3.9805, "step": 15660 }, { "epoch": 4.00295424, "grad_norm": 0.949928343296051, "learning_rate": 1.992548632165434e-05, "loss": 3.8158, "step": 15670 }, { "epoch": 4.00297984, "grad_norm": 1.1393399238586426, "learning_rate": 1.992538816698308e-05, "loss": 3.7497, "step": 15680 }, { "epoch": 4.00300544, "grad_norm": 0.886012077331543, "learning_rate": 1.9925289947948314e-05, "loss": 3.5577, "step": 15690 }, { "epoch": 4.00303104, "grad_norm": 1.2520060539245605, "learning_rate": 1.9925191664550682e-05, "loss": 3.8173, "step": 15700 }, { "epoch": 4.00305664, "grad_norm": 0.8401501178741455, "learning_rate": 1.992509331679082e-05, "loss": 4.0073, "step": 15710 }, { "epoch": 4.00308224, "grad_norm": 1.0880738496780396, "learning_rate": 1.9924994904669365e-05, "loss": 3.8766, "step": 15720 }, { "epoch": 4.00310784, "grad_norm": 0.9544134140014648, "learning_rate": 1.9924896428186958e-05, "loss": 3.7836, "step": 15730 }, { "epoch": 4.00313344, "grad_norm": 0.87335205078125, "learning_rate": 1.9924797887344232e-05, "loss": 3.8971, "step": 15740 }, { "epoch": 4.00315904, "grad_norm": 0.8628517985343933, "learning_rate": 1.992469928214183e-05, "loss": 3.8541, "step": 15750 }, { "epoch": 4.00318464, "grad_norm": 0.8800371289253235, "learning_rate": 1.9924600612580393e-05, "loss": 3.8252, "step": 15760 }, { "epoch": 4.00321024, "grad_norm": 1.043076992034912, "learning_rate": 1.9924501878660556e-05, "loss": 3.8305, "step": 15770 }, { "epoch": 4.00323584, "grad_norm": 1.635737419128418, "learning_rate": 1.9924403080382963e-05, "loss": 3.7349, "step": 15780 }, { "epoch": 4.00326144, "grad_norm": 1.013009786605835, "learning_rate": 1.9924304217748253e-05, "loss": 3.7833, "step": 15790 }, { "epoch": 4.00328704, "grad_norm": 0.8423873782157898, "learning_rate": 1.992420529075707e-05, "loss": 3.7107, "step": 15800 }, { "epoch": 4.00331264, "grad_norm": 0.8710177540779114, "learning_rate": 1.992410629941005e-05, "loss": 3.8405, "step": 15810 }, { "epoch": 4.00333824, "grad_norm": 1.028880000114441, "learning_rate": 1.9924007243707842e-05, "loss": 3.7184, "step": 15820 }, { "epoch": 4.00336384, "grad_norm": 0.8592199087142944, "learning_rate": 1.9923908123651082e-05, "loss": 3.776, "step": 15830 }, { "epoch": 4.00338944, "grad_norm": 1.1526038646697998, "learning_rate": 1.9923808939240414e-05, "loss": 3.7554, "step": 15840 }, { "epoch": 4.00341504, "grad_norm": 0.9565078616142273, "learning_rate": 1.9923709690476486e-05, "loss": 3.6717, "step": 15850 }, { "epoch": 4.00344064, "grad_norm": 0.7802088856697083, "learning_rate": 1.9923610377359935e-05, "loss": 3.7616, "step": 15860 }, { "epoch": 4.00346624, "grad_norm": 1.1332547664642334, "learning_rate": 1.992351099989141e-05, "loss": 3.8512, "step": 15870 }, { "epoch": 4.00349184, "grad_norm": 0.8939331769943237, "learning_rate": 1.9923411558071553e-05, "loss": 3.8172, "step": 15880 }, { "epoch": 4.00351744, "grad_norm": 0.9109812378883362, "learning_rate": 1.9923312051901008e-05, "loss": 4.0768, "step": 15890 }, { "epoch": 4.00354304, "grad_norm": 0.8555841445922852, "learning_rate": 1.9923212481380424e-05, "loss": 3.6332, "step": 15900 }, { "epoch": 4.00356864, "grad_norm": 0.9279158711433411, "learning_rate": 1.992311284651044e-05, "loss": 3.6585, "step": 15910 }, { "epoch": 4.00359424, "grad_norm": 0.9508206248283386, "learning_rate": 1.9923013147291712e-05, "loss": 3.936, "step": 15920 }, { "epoch": 4.00361984, "grad_norm": 1.020182490348816, "learning_rate": 1.992291338372488e-05, "loss": 3.7168, "step": 15930 }, { "epoch": 4.00364544, "grad_norm": 1.0237654447555542, "learning_rate": 1.992281355581059e-05, "loss": 3.8656, "step": 15940 }, { "epoch": 4.00367104, "grad_norm": 1.8670271635055542, "learning_rate": 1.9922713663549494e-05, "loss": 3.7951, "step": 15950 }, { "epoch": 4.00369664, "grad_norm": 0.9621522426605225, "learning_rate": 1.992261370694223e-05, "loss": 3.8352, "step": 15960 }, { "epoch": 4.00372224, "grad_norm": 0.8763441443443298, "learning_rate": 1.9922513685989463e-05, "loss": 3.6345, "step": 15970 }, { "epoch": 4.00374784, "grad_norm": 0.7783935070037842, "learning_rate": 1.9922413600691825e-05, "loss": 3.9791, "step": 15980 }, { "epoch": 4.00377344, "grad_norm": 0.8560460209846497, "learning_rate": 1.9922313451049974e-05, "loss": 3.9991, "step": 15990 }, { "epoch": 4.00379904, "grad_norm": 0.7861570119857788, "learning_rate": 1.9922213237064555e-05, "loss": 3.6925, "step": 16000 }, { "epoch": 4.00382464, "grad_norm": 2.45591139793396, "learning_rate": 1.9922112958736223e-05, "loss": 3.9196, "step": 16010 }, { "epoch": 4.00385024, "grad_norm": 0.8715717196464539, "learning_rate": 1.9922012616065626e-05, "loss": 3.7371, "step": 16020 }, { "epoch": 4.00387584, "grad_norm": 0.8409456610679626, "learning_rate": 1.9921912209053413e-05, "loss": 3.5664, "step": 16030 }, { "epoch": 4.00390144, "grad_norm": 0.8444534540176392, "learning_rate": 1.9921811737700235e-05, "loss": 3.6641, "step": 16040 }, { "epoch": 4.00392704, "grad_norm": 0.8414536118507385, "learning_rate": 1.9921711202006742e-05, "loss": 3.9248, "step": 16050 }, { "epoch": 4.00395264, "grad_norm": 0.9813508987426758, "learning_rate": 1.9921610601973596e-05, "loss": 3.8486, "step": 16060 }, { "epoch": 4.00397824, "grad_norm": 0.8016359210014343, "learning_rate": 1.9921509937601437e-05, "loss": 3.7976, "step": 16070 }, { "epoch": 4.00400384, "grad_norm": 0.874008297920227, "learning_rate": 1.992140920889093e-05, "loss": 3.7158, "step": 16080 }, { "epoch": 4.00402944, "grad_norm": 9.489341735839844, "learning_rate": 1.992130841584271e-05, "loss": 3.8492, "step": 16090 }, { "epoch": 4.00405504, "grad_norm": 0.7981265783309937, "learning_rate": 1.9921207558457446e-05, "loss": 3.7372, "step": 16100 }, { "epoch": 4.00408064, "grad_norm": 0.9216472506523132, "learning_rate": 1.9921106636735793e-05, "loss": 3.7321, "step": 16110 }, { "epoch": 4.00410624, "grad_norm": 1.1217358112335205, "learning_rate": 1.9921005650678396e-05, "loss": 3.8583, "step": 16120 }, { "epoch": 4.00413184, "grad_norm": 0.9173359870910645, "learning_rate": 1.9920904600285912e-05, "loss": 3.7989, "step": 16130 }, { "epoch": 4.00415744, "grad_norm": 1.1493680477142334, "learning_rate": 1.9920803485559e-05, "loss": 3.7849, "step": 16140 }, { "epoch": 4.00418304, "grad_norm": 1.3486607074737549, "learning_rate": 1.9920702306498316e-05, "loss": 3.6623, "step": 16150 }, { "epoch": 4.00420864, "grad_norm": 0.7878615260124207, "learning_rate": 1.9920601063104513e-05, "loss": 3.6159, "step": 16160 }, { "epoch": 4.00423424, "grad_norm": 0.8858698606491089, "learning_rate": 1.9920499755378247e-05, "loss": 3.7864, "step": 16170 }, { "epoch": 4.00425984, "grad_norm": 1.0241827964782715, "learning_rate": 1.9920398383320177e-05, "loss": 3.614, "step": 16180 }, { "epoch": 4.00428544, "grad_norm": 1.1844708919525146, "learning_rate": 1.9920296946930965e-05, "loss": 3.9316, "step": 16190 }, { "epoch": 4.00431104, "grad_norm": 0.8192345499992371, "learning_rate": 1.9920195446211257e-05, "loss": 3.9251, "step": 16200 }, { "epoch": 4.00433664, "grad_norm": 0.7587572336196899, "learning_rate": 1.992009388116172e-05, "loss": 3.6258, "step": 16210 }, { "epoch": 4.00436224, "grad_norm": 1.0881317853927612, "learning_rate": 1.9919992251783014e-05, "loss": 4.0106, "step": 16220 }, { "epoch": 4.00438784, "grad_norm": 1.3303251266479492, "learning_rate": 1.991989055807579e-05, "loss": 3.8365, "step": 16230 }, { "epoch": 4.00441344, "grad_norm": 0.8414644002914429, "learning_rate": 1.9919788800040717e-05, "loss": 3.6759, "step": 16240 }, { "epoch": 4.00443904, "grad_norm": 1.0473380088806152, "learning_rate": 1.9919686977678445e-05, "loss": 3.4733, "step": 16250 }, { "epoch": 4.00446464, "grad_norm": 0.9445676207542419, "learning_rate": 1.9919585090989645e-05, "loss": 3.6842, "step": 16260 }, { "epoch": 4.00449024, "grad_norm": 0.9594079256057739, "learning_rate": 1.9919483139974964e-05, "loss": 3.8084, "step": 16270 }, { "epoch": 4.00451584, "grad_norm": 0.9513341188430786, "learning_rate": 1.9919381124635076e-05, "loss": 3.7542, "step": 16280 }, { "epoch": 4.00454144, "grad_norm": 0.9413555264472961, "learning_rate": 1.991927904497064e-05, "loss": 3.7152, "step": 16290 }, { "epoch": 4.00456704, "grad_norm": 0.8027446269989014, "learning_rate": 1.991917690098231e-05, "loss": 3.8765, "step": 16300 }, { "epoch": 4.00459264, "grad_norm": 0.8631410002708435, "learning_rate": 1.991907469267076e-05, "loss": 3.8697, "step": 16310 }, { "epoch": 4.00461824, "grad_norm": 1.1349533796310425, "learning_rate": 1.9918972420036642e-05, "loss": 3.6815, "step": 16320 }, { "epoch": 4.00464384, "grad_norm": 1.1921066045761108, "learning_rate": 1.9918870083080625e-05, "loss": 3.7309, "step": 16330 }, { "epoch": 4.00466944, "grad_norm": 0.8445999026298523, "learning_rate": 1.991876768180337e-05, "loss": 3.8619, "step": 16340 }, { "epoch": 4.00469504, "grad_norm": 0.8212587833404541, "learning_rate": 1.991866521620555e-05, "loss": 3.7884, "step": 16350 }, { "epoch": 4.00472064, "grad_norm": 1.0000323057174683, "learning_rate": 1.9918562686287814e-05, "loss": 3.6078, "step": 16360 }, { "epoch": 4.00474624, "grad_norm": 0.9152665734291077, "learning_rate": 1.991846009205084e-05, "loss": 3.8582, "step": 16370 }, { "epoch": 4.00477184, "grad_norm": 0.7815871834754944, "learning_rate": 1.991835743349529e-05, "loss": 3.8695, "step": 16380 }, { "epoch": 4.00479744, "grad_norm": 0.7837058305740356, "learning_rate": 1.991825471062182e-05, "loss": 3.6593, "step": 16390 }, { "epoch": 4.00482304, "grad_norm": 0.8016416430473328, "learning_rate": 1.991815192343111e-05, "loss": 3.9619, "step": 16400 }, { "epoch": 4.00484864, "grad_norm": 0.9937669038772583, "learning_rate": 1.991804907192382e-05, "loss": 3.6223, "step": 16410 }, { "epoch": 4.00487424, "grad_norm": 0.900391161441803, "learning_rate": 1.9917946156100617e-05, "loss": 3.8901, "step": 16420 }, { "epoch": 4.00489984, "grad_norm": 1.039068579673767, "learning_rate": 1.991784317596217e-05, "loss": 3.7641, "step": 16430 }, { "epoch": 4.00492544, "grad_norm": 1.1317979097366333, "learning_rate": 1.9917740131509143e-05, "loss": 3.864, "step": 16440 }, { "epoch": 4.00495104, "grad_norm": 1.2036769390106201, "learning_rate": 1.991763702274221e-05, "loss": 3.9889, "step": 16450 }, { "epoch": 4.00497664, "grad_norm": 0.9288697242736816, "learning_rate": 1.9917533849662036e-05, "loss": 3.858, "step": 16460 }, { "epoch": 4.00500224, "grad_norm": 0.8886618614196777, "learning_rate": 1.991743061226929e-05, "loss": 3.7593, "step": 16470 }, { "epoch": 4.00502784, "grad_norm": 0.9384385943412781, "learning_rate": 1.9917327310564644e-05, "loss": 3.6702, "step": 16480 }, { "epoch": 4.00505344, "grad_norm": 0.8732821941375732, "learning_rate": 1.9917223944548763e-05, "loss": 3.7395, "step": 16490 }, { "epoch": 4.00507904, "grad_norm": 0.7379124760627747, "learning_rate": 1.9917120514222322e-05, "loss": 3.6385, "step": 16500 }, { "epoch": 4.00510464, "grad_norm": 0.9316958785057068, "learning_rate": 1.991701701958599e-05, "loss": 3.6226, "step": 16510 }, { "epoch": 4.00513024, "grad_norm": 0.8325507640838623, "learning_rate": 1.9916913460640438e-05, "loss": 3.7161, "step": 16520 }, { "epoch": 4.00515584, "grad_norm": 1.0520888566970825, "learning_rate": 1.991680983738634e-05, "loss": 3.8878, "step": 16530 }, { "epoch": 4.00518144, "grad_norm": 0.8645333647727966, "learning_rate": 1.991670614982436e-05, "loss": 4.0053, "step": 16540 }, { "epoch": 4.00520704, "grad_norm": 0.9079676270484924, "learning_rate": 1.991660239795518e-05, "loss": 3.7275, "step": 16550 }, { "epoch": 4.00523264, "grad_norm": 0.9437577128410339, "learning_rate": 1.9916498581779473e-05, "loss": 3.8681, "step": 16560 }, { "epoch": 4.00525824, "grad_norm": 0.8243520259857178, "learning_rate": 1.9916394701297905e-05, "loss": 3.7259, "step": 16570 }, { "epoch": 4.00528384, "grad_norm": 0.8860000967979431, "learning_rate": 1.9916290756511153e-05, "loss": 3.7504, "step": 16580 }, { "epoch": 4.00530944, "grad_norm": 0.8871355056762695, "learning_rate": 1.991618674741989e-05, "loss": 3.8927, "step": 16590 }, { "epoch": 4.00533504, "grad_norm": 0.9159587025642395, "learning_rate": 1.991608267402479e-05, "loss": 3.7483, "step": 16600 }, { "epoch": 4.00536064, "grad_norm": 0.839114248752594, "learning_rate": 1.991597853632653e-05, "loss": 3.7495, "step": 16610 }, { "epoch": 4.00538624, "grad_norm": 0.8400610089302063, "learning_rate": 1.9915874334325787e-05, "loss": 3.6883, "step": 16620 }, { "epoch": 4.00541184, "grad_norm": 0.9448635578155518, "learning_rate": 1.9915770068023232e-05, "loss": 3.8338, "step": 16630 }, { "epoch": 4.00543744, "grad_norm": 0.8327813744544983, "learning_rate": 1.9915665737419542e-05, "loss": 3.7784, "step": 16640 }, { "epoch": 4.00546304, "grad_norm": 1.0096491575241089, "learning_rate": 1.99155613425154e-05, "loss": 3.5594, "step": 16650 }, { "epoch": 4.00548864, "grad_norm": 0.9445922374725342, "learning_rate": 1.9915456883311476e-05, "loss": 3.7724, "step": 16660 }, { "epoch": 4.00551424, "grad_norm": 0.7944066524505615, "learning_rate": 1.9915352359808447e-05, "loss": 3.684, "step": 16670 }, { "epoch": 4.00553984, "grad_norm": 1.3146032094955444, "learning_rate": 1.9915247772006994e-05, "loss": 3.7962, "step": 16680 }, { "epoch": 4.00556544, "grad_norm": 0.8294607400894165, "learning_rate": 1.9915143119907796e-05, "loss": 3.9044, "step": 16690 }, { "epoch": 4.00559104, "grad_norm": 1.0345137119293213, "learning_rate": 1.9915038403511526e-05, "loss": 3.8584, "step": 16700 }, { "epoch": 4.00561664, "grad_norm": 0.9372106194496155, "learning_rate": 1.991493362281887e-05, "loss": 3.6509, "step": 16710 }, { "epoch": 4.00564224, "grad_norm": 0.950230598449707, "learning_rate": 1.9914828777830502e-05, "loss": 3.7819, "step": 16720 }, { "epoch": 4.00566784, "grad_norm": 0.9934252500534058, "learning_rate": 1.991472386854711e-05, "loss": 3.829, "step": 16730 }, { "epoch": 4.00569344, "grad_norm": 0.8995223641395569, "learning_rate": 1.9914618894969364e-05, "loss": 3.8, "step": 16740 }, { "epoch": 4.00571904, "grad_norm": 0.862555205821991, "learning_rate": 1.991451385709795e-05, "loss": 3.4963, "step": 16750 }, { "epoch": 4.00574464, "grad_norm": 0.8015438914299011, "learning_rate": 1.9914408754933553e-05, "loss": 3.5177, "step": 16760 }, { "epoch": 4.00577024, "grad_norm": 0.954328179359436, "learning_rate": 1.9914303588476843e-05, "loss": 3.9374, "step": 16770 }, { "epoch": 4.00579584, "grad_norm": 0.9183204770088196, "learning_rate": 1.9914198357728517e-05, "loss": 3.8434, "step": 16780 }, { "epoch": 4.00582144, "grad_norm": 1.012458086013794, "learning_rate": 1.9914093062689244e-05, "loss": 3.9227, "step": 16790 }, { "epoch": 4.00584704, "grad_norm": 1.0907564163208008, "learning_rate": 1.9913987703359714e-05, "loss": 3.8346, "step": 16800 }, { "epoch": 4.00587264, "grad_norm": 0.8876010775566101, "learning_rate": 1.991388227974061e-05, "loss": 3.6504, "step": 16810 }, { "epoch": 4.00589824, "grad_norm": 1.3990055322647095, "learning_rate": 1.991377679183261e-05, "loss": 3.68, "step": 16820 }, { "epoch": 4.00592384, "grad_norm": 2.6982719898223877, "learning_rate": 1.99136712396364e-05, "loss": 4.0382, "step": 16830 }, { "epoch": 4.00594944, "grad_norm": 0.924275279045105, "learning_rate": 1.9913565623152676e-05, "loss": 4.0828, "step": 16840 }, { "epoch": 4.00597504, "grad_norm": 0.7643191814422607, "learning_rate": 1.991345994238211e-05, "loss": 3.8859, "step": 16850 }, { "epoch": 4.00600064, "grad_norm": 1.0789448022842407, "learning_rate": 1.991335419732539e-05, "loss": 3.8524, "step": 16860 }, { "epoch": 4.00602624, "grad_norm": 0.8439668416976929, "learning_rate": 1.9913248387983203e-05, "loss": 3.9521, "step": 16870 }, { "epoch": 4.00605184, "grad_norm": 0.8034178018569946, "learning_rate": 1.9913142514356233e-05, "loss": 3.7215, "step": 16880 }, { "epoch": 4.00607744, "grad_norm": 0.7618510723114014, "learning_rate": 1.991303657644517e-05, "loss": 3.8269, "step": 16890 }, { "epoch": 4.00610304, "grad_norm": 0.9283100962638855, "learning_rate": 1.99129305742507e-05, "loss": 3.7956, "step": 16900 }, { "epoch": 4.00612864, "grad_norm": 0.8700074553489685, "learning_rate": 1.9912824507773506e-05, "loss": 3.8134, "step": 16910 }, { "epoch": 4.00615424, "grad_norm": 0.8905102014541626, "learning_rate": 1.9912718377014286e-05, "loss": 3.8326, "step": 16920 }, { "epoch": 4.00617984, "grad_norm": 0.9278662204742432, "learning_rate": 1.9912612181973715e-05, "loss": 3.7749, "step": 16930 }, { "epoch": 4.00620544, "grad_norm": 1.0053815841674805, "learning_rate": 1.9912505922652494e-05, "loss": 3.7022, "step": 16940 }, { "epoch": 4.00623104, "grad_norm": 0.8496661186218262, "learning_rate": 1.9912399599051306e-05, "loss": 3.7513, "step": 16950 }, { "epoch": 4.00625664, "grad_norm": 0.9305307865142822, "learning_rate": 1.991229321117084e-05, "loss": 3.7446, "step": 16960 }, { "epoch": 4.00628224, "grad_norm": 0.9136264324188232, "learning_rate": 1.9912186759011784e-05, "loss": 3.7284, "step": 16970 }, { "epoch": 4.00630784, "grad_norm": 1.022354006767273, "learning_rate": 1.9912080242574834e-05, "loss": 3.9364, "step": 16980 }, { "epoch": 4.00633344, "grad_norm": 1.2345826625823975, "learning_rate": 1.9911973661860677e-05, "loss": 3.8248, "step": 16990 }, { "epoch": 4.00635904, "grad_norm": 0.8337095379829407, "learning_rate": 1.9911867016870004e-05, "loss": 3.7841, "step": 17000 }, { "epoch": 4.00638464, "grad_norm": 1.1975536346435547, "learning_rate": 1.991176030760351e-05, "loss": 3.709, "step": 17010 }, { "epoch": 4.00641024, "grad_norm": 0.9652045965194702, "learning_rate": 1.9911653534061886e-05, "loss": 3.6959, "step": 17020 }, { "epoch": 4.00643584, "grad_norm": 0.9447763562202454, "learning_rate": 1.991154669624582e-05, "loss": 3.6468, "step": 17030 }, { "epoch": 4.00646144, "grad_norm": 1.0422265529632568, "learning_rate": 1.9911439794156008e-05, "loss": 3.6394, "step": 17040 }, { "epoch": 4.00648704, "grad_norm": 0.9483687281608582, "learning_rate": 1.9911332827793144e-05, "loss": 3.7476, "step": 17050 }, { "epoch": 4.00651264, "grad_norm": 1.0404655933380127, "learning_rate": 1.991122579715792e-05, "loss": 3.68, "step": 17060 }, { "epoch": 4.00653824, "grad_norm": 0.8035969138145447, "learning_rate": 1.9911118702251034e-05, "loss": 3.6588, "step": 17070 }, { "epoch": 4.00656384, "grad_norm": 0.9983558058738708, "learning_rate": 1.9911011543073175e-05, "loss": 3.6595, "step": 17080 }, { "epoch": 4.00658944, "grad_norm": 0.8031166195869446, "learning_rate": 1.991090431962504e-05, "loss": 3.7871, "step": 17090 }, { "epoch": 4.00661504, "grad_norm": 1.0693073272705078, "learning_rate": 1.9910797031907322e-05, "loss": 3.5568, "step": 17100 }, { "epoch": 4.00664064, "grad_norm": 1.0745636224746704, "learning_rate": 1.991068967992072e-05, "loss": 3.9087, "step": 17110 }, { "epoch": 4.00666624, "grad_norm": 1.5281469821929932, "learning_rate": 1.9910582263665933e-05, "loss": 3.7438, "step": 17120 }, { "epoch": 4.00669184, "grad_norm": 1.2707810401916504, "learning_rate": 1.991047478314365e-05, "loss": 3.5539, "step": 17130 }, { "epoch": 4.00671744, "grad_norm": 1.3299673795700073, "learning_rate": 1.9910367238354575e-05, "loss": 3.7229, "step": 17140 }, { "epoch": 4.00674304, "grad_norm": 0.8926682472229004, "learning_rate": 1.99102596292994e-05, "loss": 3.7565, "step": 17150 }, { "epoch": 4.00676864, "grad_norm": 0.9413272142410278, "learning_rate": 1.991015195597883e-05, "loss": 3.6851, "step": 17160 }, { "epoch": 4.00679424, "grad_norm": 1.0012402534484863, "learning_rate": 1.9910044218393555e-05, "loss": 3.9273, "step": 17170 }, { "epoch": 4.00681984, "grad_norm": 0.9268196821212769, "learning_rate": 1.9909936416544273e-05, "loss": 3.7629, "step": 17180 }, { "epoch": 4.00684544, "grad_norm": 0.8710830807685852, "learning_rate": 1.9909828550431693e-05, "loss": 3.702, "step": 17190 }, { "epoch": 4.00687104, "grad_norm": 0.8246394395828247, "learning_rate": 1.9909720620056505e-05, "loss": 3.9721, "step": 17200 }, { "epoch": 4.00689664, "grad_norm": 0.9127031564712524, "learning_rate": 1.9909612625419414e-05, "loss": 3.7402, "step": 17210 }, { "epoch": 4.00692224, "grad_norm": 1.2480958700180054, "learning_rate": 1.9909504566521118e-05, "loss": 3.9628, "step": 17220 }, { "epoch": 4.00694784, "grad_norm": 1.2936424016952515, "learning_rate": 1.990939644336232e-05, "loss": 3.794, "step": 17230 }, { "epoch": 4.00697344, "grad_norm": 0.9307857751846313, "learning_rate": 1.9909288255943717e-05, "loss": 3.8688, "step": 17240 }, { "epoch": 4.00699904, "grad_norm": 1.1273775100708008, "learning_rate": 1.9909180004266015e-05, "loss": 3.8167, "step": 17250 }, { "epoch": 4.00702464, "grad_norm": 1.0711992979049683, "learning_rate": 1.9909071688329916e-05, "loss": 3.8149, "step": 17260 }, { "epoch": 4.00705024, "grad_norm": 1.1008614301681519, "learning_rate": 1.9908963308136116e-05, "loss": 3.7799, "step": 17270 }, { "epoch": 4.00707584, "grad_norm": 1.3581396341323853, "learning_rate": 1.9908854863685326e-05, "loss": 3.8123, "step": 17280 }, { "epoch": 4.00710144, "grad_norm": 1.0664311647415161, "learning_rate": 1.9908746354978244e-05, "loss": 4.0196, "step": 17290 }, { "epoch": 4.00712704, "grad_norm": 0.8992627263069153, "learning_rate": 1.9908637782015575e-05, "loss": 3.6914, "step": 17300 }, { "epoch": 4.00715264, "grad_norm": 0.7564437985420227, "learning_rate": 1.9908529144798026e-05, "loss": 3.8676, "step": 17310 }, { "epoch": 4.00717824, "grad_norm": 0.9088222980499268, "learning_rate": 1.99084204433263e-05, "loss": 3.6466, "step": 17320 }, { "epoch": 4.00720384, "grad_norm": 0.900627076625824, "learning_rate": 1.9908311677601096e-05, "loss": 3.7738, "step": 17330 }, { "epoch": 4.00722944, "grad_norm": 0.797522246837616, "learning_rate": 1.990820284762313e-05, "loss": 3.8132, "step": 17340 }, { "epoch": 4.00725504, "grad_norm": 0.8639200925827026, "learning_rate": 1.9908093953393097e-05, "loss": 3.5966, "step": 17350 }, { "epoch": 4.00728064, "grad_norm": 0.7542192935943604, "learning_rate": 1.9907984994911706e-05, "loss": 3.6377, "step": 17360 }, { "epoch": 4.00730624, "grad_norm": 0.8704935908317566, "learning_rate": 1.9907875972179668e-05, "loss": 3.793, "step": 17370 }, { "epoch": 4.00733184, "grad_norm": 1.2017098665237427, "learning_rate": 1.990776688519769e-05, "loss": 3.8679, "step": 17380 }, { "epoch": 4.00735744, "grad_norm": 1.2128652334213257, "learning_rate": 1.9907657733966475e-05, "loss": 4.1929, "step": 17390 }, { "epoch": 4.00738304, "grad_norm": 1.0004526376724243, "learning_rate": 1.9907548518486733e-05, "loss": 3.8811, "step": 17400 }, { "epoch": 4.00740864, "grad_norm": 0.8911948800086975, "learning_rate": 1.990743923875917e-05, "loss": 3.7626, "step": 17410 }, { "epoch": 4.00743424, "grad_norm": 1.234762191772461, "learning_rate": 1.99073298947845e-05, "loss": 3.7301, "step": 17420 }, { "epoch": 4.00745984, "grad_norm": 0.8870864510536194, "learning_rate": 1.9907220486563426e-05, "loss": 3.7263, "step": 17430 }, { "epoch": 4.00748544, "grad_norm": 0.9854526519775391, "learning_rate": 1.9907111014096662e-05, "loss": 3.6787, "step": 17440 }, { "epoch": 4.00751104, "grad_norm": 0.9183798432350159, "learning_rate": 1.9907001477384917e-05, "loss": 3.7235, "step": 17450 }, { "epoch": 4.00753664, "grad_norm": 0.9103091359138489, "learning_rate": 1.9906891876428897e-05, "loss": 3.7038, "step": 17460 }, { "epoch": 4.00756224, "grad_norm": 1.1097068786621094, "learning_rate": 1.9906782211229317e-05, "loss": 4.0035, "step": 17470 }, { "epoch": 4.00758784, "grad_norm": 0.8416129946708679, "learning_rate": 1.990667248178689e-05, "loss": 3.7418, "step": 17480 }, { "epoch": 4.00761344, "grad_norm": 0.9988736510276794, "learning_rate": 1.990656268810232e-05, "loss": 3.6619, "step": 17490 }, { "epoch": 4.00763904, "grad_norm": 1.111403465270996, "learning_rate": 1.990645283017633e-05, "loss": 3.7645, "step": 17500 }, { "epoch": 4.00766464, "grad_norm": 1.0291551351547241, "learning_rate": 1.990634290800962e-05, "loss": 3.7662, "step": 17510 }, { "epoch": 4.00769024, "grad_norm": 0.8975242972373962, "learning_rate": 1.9906232921602913e-05, "loss": 3.6656, "step": 17520 }, { "epoch": 4.00771584, "grad_norm": 0.9947468638420105, "learning_rate": 1.9906122870956918e-05, "loss": 3.5746, "step": 17530 }, { "epoch": 4.00774144, "grad_norm": 1.317083716392517, "learning_rate": 1.9906012756072347e-05, "loss": 3.7817, "step": 17540 }, { "epoch": 4.00776704, "grad_norm": 1.8855884075164795, "learning_rate": 1.990590257694992e-05, "loss": 4.1316, "step": 17550 }, { "epoch": 4.00779264, "grad_norm": 3.743506908416748, "learning_rate": 1.990579233359034e-05, "loss": 4.2551, "step": 17560 }, { "epoch": 4.00781824, "grad_norm": 1.040006160736084, "learning_rate": 1.9905682025994336e-05, "loss": 3.5743, "step": 17570 }, { "epoch": 4.00784384, "grad_norm": 0.9703771471977234, "learning_rate": 1.9905571654162615e-05, "loss": 3.7737, "step": 17580 }, { "epoch": 4.00786944, "grad_norm": 0.8966724872589111, "learning_rate": 1.9905461218095894e-05, "loss": 3.7576, "step": 17590 }, { "epoch": 4.00789504, "grad_norm": 0.9135972857475281, "learning_rate": 1.990535071779489e-05, "loss": 3.558, "step": 17600 }, { "epoch": 4.00792064, "grad_norm": 0.8905203342437744, "learning_rate": 1.990524015326032e-05, "loss": 3.7644, "step": 17610 }, { "epoch": 4.00794624, "grad_norm": 0.9501385688781738, "learning_rate": 1.9905129524492897e-05, "loss": 3.7819, "step": 17620 }, { "epoch": 4.00797184, "grad_norm": 0.8939642906188965, "learning_rate": 1.9905018831493344e-05, "loss": 3.635, "step": 17630 }, { "epoch": 4.00799744, "grad_norm": 0.9542222023010254, "learning_rate": 1.9904908074262375e-05, "loss": 3.7009, "step": 17640 }, { "epoch": 4.00802304, "grad_norm": 0.9138267040252686, "learning_rate": 1.9904797252800708e-05, "loss": 3.7169, "step": 17650 }, { "epoch": 4.00804864, "grad_norm": 0.9845089316368103, "learning_rate": 1.9904686367109068e-05, "loss": 3.7301, "step": 17660 }, { "epoch": 4.00807424, "grad_norm": 0.7808359265327454, "learning_rate": 1.9904575417188166e-05, "loss": 3.5818, "step": 17670 }, { "epoch": 4.00809984, "grad_norm": 1.0923405885696411, "learning_rate": 1.9904464403038724e-05, "loss": 3.7369, "step": 17680 }, { "epoch": 4.00812544, "grad_norm": 0.9263991713523865, "learning_rate": 1.9904353324661464e-05, "loss": 3.5743, "step": 17690 }, { "epoch": 4.00815104, "grad_norm": 0.8546760678291321, "learning_rate": 1.9904242182057104e-05, "loss": 3.7089, "step": 17700 }, { "epoch": 4.00817664, "grad_norm": 0.9508864283561707, "learning_rate": 1.990413097522637e-05, "loss": 3.8906, "step": 17710 }, { "epoch": 4.00820224, "grad_norm": 0.8857953548431396, "learning_rate": 1.9904019704169975e-05, "loss": 3.8873, "step": 17720 }, { "epoch": 4.00822784, "grad_norm": 1.0176588296890259, "learning_rate": 1.9903908368888642e-05, "loss": 3.7746, "step": 17730 }, { "epoch": 4.00825344, "grad_norm": 1.0434163808822632, "learning_rate": 1.99037969693831e-05, "loss": 3.764, "step": 17740 }, { "epoch": 4.00827904, "grad_norm": 0.8107135891914368, "learning_rate": 1.9903685505654064e-05, "loss": 3.8683, "step": 17750 }, { "epoch": 4.00830464, "grad_norm": 0.8828801512718201, "learning_rate": 1.9903573977702263e-05, "loss": 3.5463, "step": 17760 }, { "epoch": 4.00833024, "grad_norm": 1.048643946647644, "learning_rate": 1.9903462385528414e-05, "loss": 3.7003, "step": 17770 }, { "epoch": 4.00835584, "grad_norm": 0.9362902641296387, "learning_rate": 1.9903350729133242e-05, "loss": 3.5706, "step": 17780 }, { "epoch": 4.00838144, "grad_norm": 1.0586823225021362, "learning_rate": 1.9903239008517475e-05, "loss": 3.7925, "step": 17790 }, { "epoch": 4.00840704, "grad_norm": 0.8690969944000244, "learning_rate": 1.9903127223681835e-05, "loss": 3.7714, "step": 17800 }, { "epoch": 4.00843264, "grad_norm": 0.7225038409233093, "learning_rate": 1.9903015374627043e-05, "loss": 3.4428, "step": 17810 }, { "epoch": 4.00845824, "grad_norm": 0.9129195809364319, "learning_rate": 1.9902903461353832e-05, "loss": 3.5834, "step": 17820 }, { "epoch": 4.00848384, "grad_norm": 1.0625873804092407, "learning_rate": 1.9902791483862923e-05, "loss": 3.7978, "step": 17830 }, { "epoch": 4.00850944, "grad_norm": 0.8704116940498352, "learning_rate": 1.9902679442155045e-05, "loss": 3.6267, "step": 17840 }, { "epoch": 4.00853504, "grad_norm": 0.9160746335983276, "learning_rate": 1.9902567336230922e-05, "loss": 3.6719, "step": 17850 }, { "epoch": 4.00856064, "grad_norm": 0.8166300654411316, "learning_rate": 1.990245516609128e-05, "loss": 3.9399, "step": 17860 }, { "epoch": 4.00858624, "grad_norm": 0.8306975364685059, "learning_rate": 1.9902342931736848e-05, "loss": 3.6686, "step": 17870 }, { "epoch": 4.00861184, "grad_norm": 0.8936006426811218, "learning_rate": 1.9902230633168352e-05, "loss": 3.7989, "step": 17880 }, { "epoch": 4.00863744, "grad_norm": 0.8705113530158997, "learning_rate": 1.9902118270386527e-05, "loss": 3.6385, "step": 17890 }, { "epoch": 4.00866304, "grad_norm": 1.4536430835723877, "learning_rate": 1.9902005843392095e-05, "loss": 3.7739, "step": 17900 }, { "epoch": 4.00868864, "grad_norm": 0.8947390913963318, "learning_rate": 1.9901893352185787e-05, "loss": 3.7896, "step": 17910 }, { "epoch": 4.00871424, "grad_norm": 2.6498491764068604, "learning_rate": 1.990178079676833e-05, "loss": 3.8715, "step": 17920 }, { "epoch": 4.00873984, "grad_norm": 0.9426674246788025, "learning_rate": 1.990166817714046e-05, "loss": 3.7769, "step": 17930 }, { "epoch": 4.00876544, "grad_norm": 0.9545010924339294, "learning_rate": 1.9901555493302898e-05, "loss": 3.6713, "step": 17940 }, { "epoch": 4.00879104, "grad_norm": 0.9486173987388611, "learning_rate": 1.9901442745256383e-05, "loss": 3.8153, "step": 17950 }, { "epoch": 4.00881664, "grad_norm": 1.3151741027832031, "learning_rate": 1.9901329933001643e-05, "loss": 3.813, "step": 17960 }, { "epoch": 4.00884224, "grad_norm": 0.8767807483673096, "learning_rate": 1.990121705653941e-05, "loss": 3.7954, "step": 17970 }, { "epoch": 4.00886784, "grad_norm": 0.9744095802307129, "learning_rate": 1.9901104115870417e-05, "loss": 3.6866, "step": 17980 }, { "epoch": 4.00889344, "grad_norm": 0.9819355607032776, "learning_rate": 1.9900991110995394e-05, "loss": 3.6013, "step": 17990 }, { "epoch": 4.00891904, "grad_norm": 0.9283024668693542, "learning_rate": 1.9900878041915074e-05, "loss": 3.7667, "step": 18000 }, { "epoch": 4.00894464, "grad_norm": 0.9063059091567993, "learning_rate": 1.9900764908630192e-05, "loss": 3.7375, "step": 18010 }, { "epoch": 4.00897024, "grad_norm": 0.9924116134643555, "learning_rate": 1.990065171114148e-05, "loss": 3.7441, "step": 18020 }, { "epoch": 4.00899584, "grad_norm": 1.094007134437561, "learning_rate": 1.9900538449449676e-05, "loss": 3.4635, "step": 18030 }, { "epoch": 4.00902144, "grad_norm": 0.8972987532615662, "learning_rate": 1.990042512355551e-05, "loss": 3.394, "step": 18040 }, { "epoch": 4.00904704, "grad_norm": 0.8841853737831116, "learning_rate": 1.9900311733459717e-05, "loss": 3.6431, "step": 18050 }, { "epoch": 4.00907264, "grad_norm": 0.8544389605522156, "learning_rate": 1.9900198279163036e-05, "loss": 3.7159, "step": 18060 }, { "epoch": 4.00909824, "grad_norm": 0.8336541652679443, "learning_rate": 1.9900084760666197e-05, "loss": 3.7099, "step": 18070 }, { "epoch": 4.00912384, "grad_norm": 0.903679370880127, "learning_rate": 1.989997117796994e-05, "loss": 3.6901, "step": 18080 }, { "epoch": 4.00914944, "grad_norm": 1.5993130207061768, "learning_rate": 1.9899857531075002e-05, "loss": 3.9108, "step": 18090 }, { "epoch": 4.00917504, "grad_norm": 0.8597598671913147, "learning_rate": 1.989974381998212e-05, "loss": 3.8005, "step": 18100 }, { "epoch": 4.00920064, "grad_norm": 4.033047676086426, "learning_rate": 1.989963004469203e-05, "loss": 3.9864, "step": 18110 }, { "epoch": 4.00922624, "grad_norm": 1.0615512132644653, "learning_rate": 1.989951620520547e-05, "loss": 3.7355, "step": 18120 }, { "epoch": 4.00925184, "grad_norm": 1.017261266708374, "learning_rate": 1.9899402301523177e-05, "loss": 3.7353, "step": 18130 }, { "epoch": 4.00927744, "grad_norm": 0.9594956636428833, "learning_rate": 1.9899288333645893e-05, "loss": 3.663, "step": 18140 }, { "epoch": 5.0000128, "grad_norm": 0.9847010374069214, "learning_rate": 1.9899174301574354e-05, "loss": 4.0801, "step": 18150 }, { "epoch": 5.0000384, "grad_norm": 0.9532468914985657, "learning_rate": 1.98990602053093e-05, "loss": 3.6418, "step": 18160 }, { "epoch": 5.000064, "grad_norm": 0.7899117469787598, "learning_rate": 1.989894604485147e-05, "loss": 3.3546, "step": 18170 }, { "epoch": 5.0000896, "grad_norm": 1.0005249977111816, "learning_rate": 1.9898831820201608e-05, "loss": 3.6637, "step": 18180 }, { "epoch": 5.0001152, "grad_norm": 2.1351137161254883, "learning_rate": 1.989871753136045e-05, "loss": 3.6246, "step": 18190 }, { "epoch": 5.0001408, "grad_norm": 1.0167076587677002, "learning_rate": 1.9898603178328744e-05, "loss": 3.5508, "step": 18200 }, { "epoch": 5.0001664, "grad_norm": 1.2388771772384644, "learning_rate": 1.9898488761107228e-05, "loss": 3.7771, "step": 18210 }, { "epoch": 5.000192, "grad_norm": 1.145302414894104, "learning_rate": 1.989837427969664e-05, "loss": 4.099, "step": 18220 }, { "epoch": 5.0002176, "grad_norm": 1.0329171419143677, "learning_rate": 1.9898259734097723e-05, "loss": 3.5226, "step": 18230 }, { "epoch": 5.0002432, "grad_norm": 1.0391077995300293, "learning_rate": 1.9898145124311224e-05, "loss": 3.9832, "step": 18240 }, { "epoch": 5.0002688, "grad_norm": 1.111161708831787, "learning_rate": 1.9898030450337887e-05, "loss": 3.7725, "step": 18250 }, { "epoch": 5.0002944, "grad_norm": 0.9049463868141174, "learning_rate": 1.989791571217845e-05, "loss": 3.475, "step": 18260 }, { "epoch": 5.00032, "grad_norm": 0.8915307521820068, "learning_rate": 1.9897800909833662e-05, "loss": 3.2993, "step": 18270 }, { "epoch": 5.0003456, "grad_norm": 0.8777012228965759, "learning_rate": 1.9897686043304264e-05, "loss": 3.4395, "step": 18280 }, { "epoch": 5.0003712, "grad_norm": 0.9532457590103149, "learning_rate": 1.9897571112591005e-05, "loss": 3.5769, "step": 18290 }, { "epoch": 5.0003968, "grad_norm": 1.0095138549804688, "learning_rate": 1.989745611769463e-05, "loss": 3.6401, "step": 18300 }, { "epoch": 5.0004224, "grad_norm": 1.0028440952301025, "learning_rate": 1.9897341058615875e-05, "loss": 3.5864, "step": 18310 }, { "epoch": 5.000448, "grad_norm": 0.8618952631950378, "learning_rate": 1.9897225935355497e-05, "loss": 3.7445, "step": 18320 }, { "epoch": 5.0004736, "grad_norm": 0.8120489120483398, "learning_rate": 1.989711074791424e-05, "loss": 3.5941, "step": 18330 }, { "epoch": 5.0004992, "grad_norm": 0.9310399293899536, "learning_rate": 1.9896995496292852e-05, "loss": 3.6255, "step": 18340 }, { "epoch": 5.0005248, "grad_norm": 1.232507348060608, "learning_rate": 1.9896880180492077e-05, "loss": 3.4917, "step": 18350 }, { "epoch": 5.0005504, "grad_norm": 0.778928816318512, "learning_rate": 1.989676480051266e-05, "loss": 3.5199, "step": 18360 }, { "epoch": 5.000576, "grad_norm": 0.8279438614845276, "learning_rate": 1.989664935635536e-05, "loss": 3.2797, "step": 18370 }, { "epoch": 5.0006016, "grad_norm": 0.8461528420448303, "learning_rate": 1.9896533848020915e-05, "loss": 3.4157, "step": 18380 }, { "epoch": 5.0006272, "grad_norm": 0.9345951676368713, "learning_rate": 1.989641827551008e-05, "loss": 3.5735, "step": 18390 }, { "epoch": 5.0006528, "grad_norm": 0.8313469290733337, "learning_rate": 1.9896302638823602e-05, "loss": 3.4528, "step": 18400 }, { "epoch": 5.0006784, "grad_norm": 0.7965988516807556, "learning_rate": 1.989618693796223e-05, "loss": 3.6363, "step": 18410 }, { "epoch": 5.000704, "grad_norm": 1.1853500604629517, "learning_rate": 1.989607117292672e-05, "loss": 4.2968, "step": 18420 }, { "epoch": 5.0007296, "grad_norm": 0.9829492568969727, "learning_rate": 1.9895955343717816e-05, "loss": 3.512, "step": 18430 }, { "epoch": 5.0007552, "grad_norm": 1.0464916229248047, "learning_rate": 1.989583945033627e-05, "loss": 3.5812, "step": 18440 }, { "epoch": 5.0007808, "grad_norm": 0.8010080456733704, "learning_rate": 1.9895723492782836e-05, "loss": 3.5283, "step": 18450 }, { "epoch": 5.0008064, "grad_norm": 0.9621171355247498, "learning_rate": 1.9895607471058267e-05, "loss": 3.8169, "step": 18460 }, { "epoch": 5.000832, "grad_norm": 0.7845169305801392, "learning_rate": 1.989549138516331e-05, "loss": 3.484, "step": 18470 }, { "epoch": 5.0008576, "grad_norm": 0.7738804221153259, "learning_rate": 1.9895375235098725e-05, "loss": 3.5414, "step": 18480 }, { "epoch": 5.0008832, "grad_norm": 1.1224826574325562, "learning_rate": 1.989525902086526e-05, "loss": 3.5054, "step": 18490 }, { "epoch": 5.0009088, "grad_norm": 1.2654521465301514, "learning_rate": 1.989514274246367e-05, "loss": 3.6322, "step": 18500 }, { "epoch": 5.0009344, "grad_norm": 0.8920170664787292, "learning_rate": 1.9895026399894713e-05, "loss": 3.5909, "step": 18510 }, { "epoch": 5.00096, "grad_norm": 0.881648063659668, "learning_rate": 1.9894909993159134e-05, "loss": 3.6595, "step": 18520 }, { "epoch": 5.0009856, "grad_norm": 0.9340692758560181, "learning_rate": 1.9894793522257695e-05, "loss": 3.6751, "step": 18530 }, { "epoch": 5.0010112, "grad_norm": 1.073019027709961, "learning_rate": 1.989467698719115e-05, "loss": 3.6172, "step": 18540 }, { "epoch": 5.0010368, "grad_norm": 0.89656001329422, "learning_rate": 1.9894560387960255e-05, "loss": 3.433, "step": 18550 }, { "epoch": 5.0010624, "grad_norm": 0.937275230884552, "learning_rate": 1.9894443724565767e-05, "loss": 3.4319, "step": 18560 }, { "epoch": 5.001088, "grad_norm": 0.8215678334236145, "learning_rate": 1.9894326997008436e-05, "loss": 3.5649, "step": 18570 }, { "epoch": 5.0011136, "grad_norm": 0.8599653840065002, "learning_rate": 1.9894210205289027e-05, "loss": 3.4939, "step": 18580 }, { "epoch": 5.0011392, "grad_norm": 0.8798080682754517, "learning_rate": 1.9894093349408298e-05, "loss": 3.5561, "step": 18590 }, { "epoch": 5.0011648, "grad_norm": 0.9613558053970337, "learning_rate": 1.9893976429367e-05, "loss": 3.3881, "step": 18600 }, { "epoch": 5.0011904, "grad_norm": 1.0617480278015137, "learning_rate": 1.989385944516589e-05, "loss": 3.7329, "step": 18610 }, { "epoch": 5.001216, "grad_norm": 0.9599997401237488, "learning_rate": 1.9893742396805734e-05, "loss": 3.2676, "step": 18620 }, { "epoch": 5.0012416, "grad_norm": 0.9610558152198792, "learning_rate": 1.9893625284287293e-05, "loss": 3.6621, "step": 18630 }, { "epoch": 5.0012672, "grad_norm": 0.7723628282546997, "learning_rate": 1.9893508107611318e-05, "loss": 3.5619, "step": 18640 }, { "epoch": 5.0012928, "grad_norm": 0.8668854832649231, "learning_rate": 1.989339086677857e-05, "loss": 3.6411, "step": 18650 }, { "epoch": 5.0013184, "grad_norm": 1.0473802089691162, "learning_rate": 1.9893273561789816e-05, "loss": 3.4483, "step": 18660 }, { "epoch": 5.001344, "grad_norm": 0.8638676404953003, "learning_rate": 1.9893156192645807e-05, "loss": 3.7107, "step": 18670 }, { "epoch": 5.0013696, "grad_norm": 0.936991810798645, "learning_rate": 1.9893038759347312e-05, "loss": 3.6203, "step": 18680 }, { "epoch": 5.0013952, "grad_norm": 0.9452715516090393, "learning_rate": 1.989292126189509e-05, "loss": 3.4057, "step": 18690 }, { "epoch": 5.0014208, "grad_norm": 1.4044489860534668, "learning_rate": 1.9892803700289902e-05, "loss": 3.7135, "step": 18700 }, { "epoch": 5.0014464, "grad_norm": 0.9918235540390015, "learning_rate": 1.989268607453251e-05, "loss": 3.6522, "step": 18710 }, { "epoch": 5.001472, "grad_norm": 0.9640305638313293, "learning_rate": 1.9892568384623684e-05, "loss": 3.7988, "step": 18720 }, { "epoch": 5.0014976, "grad_norm": 1.1314716339111328, "learning_rate": 1.9892450630564173e-05, "loss": 3.6492, "step": 18730 }, { "epoch": 5.0015232, "grad_norm": 0.9596134424209595, "learning_rate": 1.9892332812354754e-05, "loss": 3.5358, "step": 18740 }, { "epoch": 5.0015488, "grad_norm": 1.019827127456665, "learning_rate": 1.9892214929996186e-05, "loss": 3.7072, "step": 18750 }, { "epoch": 5.0015744, "grad_norm": 1.0085335969924927, "learning_rate": 1.989209698348923e-05, "loss": 3.8663, "step": 18760 }, { "epoch": 5.0016, "grad_norm": 0.898496687412262, "learning_rate": 1.989197897283466e-05, "loss": 3.7016, "step": 18770 }, { "epoch": 5.0016256, "grad_norm": 0.9904465079307556, "learning_rate": 1.9891860898033227e-05, "loss": 3.729, "step": 18780 }, { "epoch": 5.0016512, "grad_norm": 1.0428669452667236, "learning_rate": 1.989174275908571e-05, "loss": 3.6316, "step": 18790 }, { "epoch": 5.0016768, "grad_norm": 0.869860827922821, "learning_rate": 1.989162455599287e-05, "loss": 3.8382, "step": 18800 }, { "epoch": 5.0017024, "grad_norm": 0.8749136924743652, "learning_rate": 1.9891506288755474e-05, "loss": 3.7987, "step": 18810 }, { "epoch": 5.001728, "grad_norm": 1.1731865406036377, "learning_rate": 1.9891387957374288e-05, "loss": 3.7737, "step": 18820 }, { "epoch": 5.0017536, "grad_norm": 0.987602174282074, "learning_rate": 1.989126956185008e-05, "loss": 3.7283, "step": 18830 }, { "epoch": 5.0017792, "grad_norm": 1.0567429065704346, "learning_rate": 1.9891151102183618e-05, "loss": 3.8807, "step": 18840 }, { "epoch": 5.0018048, "grad_norm": 1.4528164863586426, "learning_rate": 1.9891032578375666e-05, "loss": 3.856, "step": 18850 }, { "epoch": 5.0018304, "grad_norm": 0.8829284310340881, "learning_rate": 1.9890913990427e-05, "loss": 3.6748, "step": 18860 }, { "epoch": 5.001856, "grad_norm": 1.0239382982254028, "learning_rate": 1.9890795338338385e-05, "loss": 3.6462, "step": 18870 }, { "epoch": 5.0018816, "grad_norm": 0.9736757874488831, "learning_rate": 1.989067662211059e-05, "loss": 3.5138, "step": 18880 }, { "epoch": 5.0019072, "grad_norm": 0.9118050336837769, "learning_rate": 1.9890557841744387e-05, "loss": 3.5863, "step": 18890 }, { "epoch": 5.0019328, "grad_norm": 0.9673511981964111, "learning_rate": 1.9890438997240545e-05, "loss": 3.4178, "step": 18900 }, { "epoch": 5.0019584, "grad_norm": 0.8585647344589233, "learning_rate": 1.9890320088599836e-05, "loss": 3.5822, "step": 18910 }, { "epoch": 5.001984, "grad_norm": 1.1534724235534668, "learning_rate": 1.9890201115823025e-05, "loss": 3.6892, "step": 18920 }, { "epoch": 5.0020096, "grad_norm": 0.8995459079742432, "learning_rate": 1.989008207891089e-05, "loss": 3.7906, "step": 18930 }, { "epoch": 5.0020352, "grad_norm": 1.0458875894546509, "learning_rate": 1.9889962977864203e-05, "loss": 3.3857, "step": 18940 }, { "epoch": 5.0020608, "grad_norm": 1.3058571815490723, "learning_rate": 1.9889843812683733e-05, "loss": 3.6255, "step": 18950 }, { "epoch": 5.0020864, "grad_norm": 0.9683895707130432, "learning_rate": 1.988972458337026e-05, "loss": 3.7392, "step": 18960 }, { "epoch": 5.002112, "grad_norm": 1.048349142074585, "learning_rate": 1.9889605289924545e-05, "loss": 3.6911, "step": 18970 }, { "epoch": 5.0021376, "grad_norm": 0.9249964356422424, "learning_rate": 1.988948593234737e-05, "loss": 3.6599, "step": 18980 }, { "epoch": 5.0021632, "grad_norm": 1.4511114358901978, "learning_rate": 1.9889366510639504e-05, "loss": 3.6816, "step": 18990 }, { "epoch": 5.0021888, "grad_norm": 1.0243594646453857, "learning_rate": 1.988924702480173e-05, "loss": 3.4662, "step": 19000 }, { "epoch": 5.0022144, "grad_norm": 1.0949140787124634, "learning_rate": 1.988912747483481e-05, "loss": 3.5585, "step": 19010 }, { "epoch": 5.00224, "grad_norm": 0.8565740585327148, "learning_rate": 1.9889007860739534e-05, "loss": 3.6176, "step": 19020 }, { "epoch": 5.0022656, "grad_norm": 1.0474135875701904, "learning_rate": 1.9888888182516666e-05, "loss": 3.8274, "step": 19030 }, { "epoch": 5.0022912, "grad_norm": 1.1421552896499634, "learning_rate": 1.9888768440166987e-05, "loss": 3.6935, "step": 19040 }, { "epoch": 5.0023168, "grad_norm": 0.8182336091995239, "learning_rate": 1.9888648633691273e-05, "loss": 3.6315, "step": 19050 }, { "epoch": 5.0023424, "grad_norm": 1.100650668144226, "learning_rate": 1.9888528763090303e-05, "loss": 3.5838, "step": 19060 }, { "epoch": 5.002368, "grad_norm": 1.0631132125854492, "learning_rate": 1.9888408828364852e-05, "loss": 3.8209, "step": 19070 }, { "epoch": 5.0023936, "grad_norm": 0.9042850732803345, "learning_rate": 1.9888288829515694e-05, "loss": 3.6262, "step": 19080 }, { "epoch": 5.0024192, "grad_norm": 1.165469765663147, "learning_rate": 1.988816876654361e-05, "loss": 3.6632, "step": 19090 }, { "epoch": 5.0024448, "grad_norm": 0.9780548810958862, "learning_rate": 1.9888048639449385e-05, "loss": 3.6317, "step": 19100 }, { "epoch": 5.0024704, "grad_norm": 1.1097922325134277, "learning_rate": 1.988792844823379e-05, "loss": 3.5537, "step": 19110 }, { "epoch": 5.002496, "grad_norm": 0.8263509273529053, "learning_rate": 1.9887808192897606e-05, "loss": 3.6487, "step": 19120 }, { "epoch": 5.0025216, "grad_norm": 0.9239467978477478, "learning_rate": 1.9887687873441616e-05, "loss": 3.7999, "step": 19130 }, { "epoch": 5.0025472, "grad_norm": 0.9514371156692505, "learning_rate": 1.9887567489866596e-05, "loss": 3.6113, "step": 19140 }, { "epoch": 5.0025728, "grad_norm": 0.863140344619751, "learning_rate": 1.9887447042173327e-05, "loss": 3.3996, "step": 19150 }, { "epoch": 5.0025984, "grad_norm": 1.0523637533187866, "learning_rate": 1.9887326530362595e-05, "loss": 3.665, "step": 19160 }, { "epoch": 5.002624, "grad_norm": 1.1918985843658447, "learning_rate": 1.9887205954435175e-05, "loss": 3.7094, "step": 19170 }, { "epoch": 5.0026496, "grad_norm": 1.0705808401107788, "learning_rate": 1.9887085314391857e-05, "loss": 3.6452, "step": 19180 }, { "epoch": 5.0026752, "grad_norm": 1.0018630027770996, "learning_rate": 1.9886964610233414e-05, "loss": 3.7317, "step": 19190 }, { "epoch": 5.0027008, "grad_norm": 1.1957805156707764, "learning_rate": 1.988684384196063e-05, "loss": 3.6978, "step": 19200 }, { "epoch": 5.0027264, "grad_norm": 0.8705518245697021, "learning_rate": 1.98867230095743e-05, "loss": 3.9727, "step": 19210 }, { "epoch": 5.002752, "grad_norm": 1.030706763267517, "learning_rate": 1.9886602113075194e-05, "loss": 3.776, "step": 19220 }, { "epoch": 5.0027776, "grad_norm": 0.9219871759414673, "learning_rate": 1.9886481152464102e-05, "loss": 3.5371, "step": 19230 }, { "epoch": 5.0028032, "grad_norm": 0.9155302047729492, "learning_rate": 1.9886360127741805e-05, "loss": 3.6986, "step": 19240 }, { "epoch": 5.0028288, "grad_norm": 1.0043216943740845, "learning_rate": 1.9886239038909095e-05, "loss": 3.5143, "step": 19250 }, { "epoch": 5.0028544, "grad_norm": 0.8653699159622192, "learning_rate": 1.9886117885966748e-05, "loss": 3.3968, "step": 19260 }, { "epoch": 5.00288, "grad_norm": 0.9029381275177002, "learning_rate": 1.9885996668915554e-05, "loss": 3.5455, "step": 19270 }, { "epoch": 5.0029056, "grad_norm": 1.1833577156066895, "learning_rate": 1.9885875387756298e-05, "loss": 3.5795, "step": 19280 }, { "epoch": 5.0029312, "grad_norm": 0.9649666547775269, "learning_rate": 1.988575404248977e-05, "loss": 3.8848, "step": 19290 }, { "epoch": 5.0029568, "grad_norm": 1.0156702995300293, "learning_rate": 1.9885632633116757e-05, "loss": 3.6256, "step": 19300 }, { "epoch": 5.0029824, "grad_norm": 1.153100848197937, "learning_rate": 1.9885511159638038e-05, "loss": 3.5896, "step": 19310 }, { "epoch": 5.003008, "grad_norm": 0.8816978335380554, "learning_rate": 1.988538962205441e-05, "loss": 3.4206, "step": 19320 }, { "epoch": 5.0030336, "grad_norm": 1.2840759754180908, "learning_rate": 1.9885268020366656e-05, "loss": 3.6991, "step": 19330 }, { "epoch": 5.0030592, "grad_norm": 1.006379246711731, "learning_rate": 1.9885146354575567e-05, "loss": 3.8463, "step": 19340 }, { "epoch": 5.0030848, "grad_norm": 0.9692828059196472, "learning_rate": 1.988502462468193e-05, "loss": 3.6931, "step": 19350 }, { "epoch": 5.0031104, "grad_norm": 0.9580073952674866, "learning_rate": 1.9884902830686538e-05, "loss": 3.627, "step": 19360 }, { "epoch": 5.003136, "grad_norm": 1.0769809484481812, "learning_rate": 1.9884780972590174e-05, "loss": 3.8667, "step": 19370 }, { "epoch": 5.0031616, "grad_norm": 0.8902144432067871, "learning_rate": 1.9884659050393636e-05, "loss": 3.6082, "step": 19380 }, { "epoch": 5.0031872, "grad_norm": 0.8478927612304688, "learning_rate": 1.9884537064097708e-05, "loss": 3.6696, "step": 19390 }, { "epoch": 5.0032128, "grad_norm": 0.8988828063011169, "learning_rate": 1.9884415013703187e-05, "loss": 3.6502, "step": 19400 }, { "epoch": 5.0032384, "grad_norm": 0.91889888048172, "learning_rate": 1.988429289921086e-05, "loss": 3.6531, "step": 19410 }, { "epoch": 5.003264, "grad_norm": 1.0038529634475708, "learning_rate": 1.988417072062152e-05, "loss": 3.5722, "step": 19420 }, { "epoch": 5.0032896, "grad_norm": 0.9205412864685059, "learning_rate": 1.988404847793596e-05, "loss": 3.6343, "step": 19430 }, { "epoch": 5.0033152, "grad_norm": 0.8810191750526428, "learning_rate": 1.9883926171154976e-05, "loss": 3.6822, "step": 19440 }, { "epoch": 5.0033408, "grad_norm": 0.9033856987953186, "learning_rate": 1.9883803800279353e-05, "loss": 3.5273, "step": 19450 }, { "epoch": 5.0033664, "grad_norm": 0.9765467047691345, "learning_rate": 1.9883681365309893e-05, "loss": 3.6781, "step": 19460 }, { "epoch": 5.003392, "grad_norm": 0.9709467887878418, "learning_rate": 1.988355886624738e-05, "loss": 3.5896, "step": 19470 }, { "epoch": 5.0034176, "grad_norm": 0.8510295152664185, "learning_rate": 1.9883436303092622e-05, "loss": 3.5131, "step": 19480 }, { "epoch": 5.0034432, "grad_norm": 1.0297919511795044, "learning_rate": 1.98833136758464e-05, "loss": 3.6617, "step": 19490 }, { "epoch": 5.0034688, "grad_norm": 0.9600341320037842, "learning_rate": 1.988319098450952e-05, "loss": 3.6452, "step": 19500 }, { "epoch": 5.0034944, "grad_norm": 0.9407463669776917, "learning_rate": 1.9883068229082773e-05, "loss": 3.6839, "step": 19510 }, { "epoch": 5.00352, "grad_norm": 1.331692099571228, "learning_rate": 1.988294540956695e-05, "loss": 3.939, "step": 19520 }, { "epoch": 5.0035456, "grad_norm": 0.8375319838523865, "learning_rate": 1.9882822525962858e-05, "loss": 3.4324, "step": 19530 }, { "epoch": 5.0035712, "grad_norm": 0.9793799519538879, "learning_rate": 1.9882699578271286e-05, "loss": 3.5656, "step": 19540 }, { "epoch": 5.0035968, "grad_norm": 0.9121690392494202, "learning_rate": 1.9882576566493036e-05, "loss": 3.8108, "step": 19550 }, { "epoch": 5.0036224, "grad_norm": 0.9782550930976868, "learning_rate": 1.9882453490628903e-05, "loss": 3.543, "step": 19560 }, { "epoch": 5.003648, "grad_norm": 1.1314622163772583, "learning_rate": 1.988233035067968e-05, "loss": 3.7015, "step": 19570 }, { "epoch": 5.0036736, "grad_norm": 1.2158259153366089, "learning_rate": 1.9882207146646177e-05, "loss": 3.6785, "step": 19580 }, { "epoch": 5.0036992, "grad_norm": 1.165172815322876, "learning_rate": 1.9882083878529187e-05, "loss": 3.6455, "step": 19590 }, { "epoch": 5.0037248, "grad_norm": 0.89488685131073, "learning_rate": 1.9881960546329506e-05, "loss": 3.5484, "step": 19600 }, { "epoch": 5.0037504, "grad_norm": 0.76272052526474, "learning_rate": 1.988183715004794e-05, "loss": 3.7932, "step": 19610 }, { "epoch": 5.003776, "grad_norm": 1.0211915969848633, "learning_rate": 1.9881713689685283e-05, "loss": 3.8911, "step": 19620 }, { "epoch": 5.0038016, "grad_norm": 0.9617506861686707, "learning_rate": 1.9881590165242344e-05, "loss": 3.5549, "step": 19630 }, { "epoch": 5.0038272, "grad_norm": 1.0814354419708252, "learning_rate": 1.9881466576719915e-05, "loss": 3.7176, "step": 19640 }, { "epoch": 5.0038528, "grad_norm": 0.82639479637146, "learning_rate": 1.9881342924118804e-05, "loss": 3.5704, "step": 19650 }, { "epoch": 5.0038784, "grad_norm": 0.9392719268798828, "learning_rate": 1.988121920743981e-05, "loss": 3.4701, "step": 19660 }, { "epoch": 5.003904, "grad_norm": 0.7957473993301392, "learning_rate": 1.988109542668373e-05, "loss": 3.5266, "step": 19670 }, { "epoch": 5.0039296, "grad_norm": 0.9304134249687195, "learning_rate": 1.988097158185138e-05, "loss": 3.7726, "step": 19680 }, { "epoch": 5.0039552, "grad_norm": 1.2443279027938843, "learning_rate": 1.988084767294355e-05, "loss": 3.6826, "step": 19690 }, { "epoch": 5.0039808, "grad_norm": 0.8428776264190674, "learning_rate": 1.9880723699961052e-05, "loss": 3.6422, "step": 19700 }, { "epoch": 5.0040064, "grad_norm": 0.8409715890884399, "learning_rate": 1.9880599662904686e-05, "loss": 3.5467, "step": 19710 }, { "epoch": 5.004032, "grad_norm": 5.6016693115234375, "learning_rate": 1.9880475561775258e-05, "loss": 3.7426, "step": 19720 }, { "epoch": 5.0040576, "grad_norm": 0.9461497068405151, "learning_rate": 1.9880351396573568e-05, "loss": 3.5584, "step": 19730 }, { "epoch": 5.0040832, "grad_norm": 0.8588926792144775, "learning_rate": 1.988022716730043e-05, "loss": 3.5966, "step": 19740 }, { "epoch": 5.0041088, "grad_norm": 0.8469528555870056, "learning_rate": 1.9880102873956643e-05, "loss": 3.7376, "step": 19750 }, { "epoch": 5.0041344, "grad_norm": 0.9489337801933289, "learning_rate": 1.9879978516543015e-05, "loss": 3.6354, "step": 19760 }, { "epoch": 5.00416, "grad_norm": 0.9525190591812134, "learning_rate": 1.9879854095060353e-05, "loss": 3.6141, "step": 19770 }, { "epoch": 5.0041856, "grad_norm": 1.4681919813156128, "learning_rate": 1.987972960950946e-05, "loss": 3.4797, "step": 19780 }, { "epoch": 5.0042112, "grad_norm": 0.9928115010261536, "learning_rate": 1.9879605059891144e-05, "loss": 3.4592, "step": 19790 }, { "epoch": 5.0042368, "grad_norm": 0.873275637626648, "learning_rate": 1.987948044620622e-05, "loss": 3.6549, "step": 19800 }, { "epoch": 5.0042624, "grad_norm": 0.8501573801040649, "learning_rate": 1.987935576845549e-05, "loss": 3.4925, "step": 19810 }, { "epoch": 5.004288, "grad_norm": 1.1230922937393188, "learning_rate": 1.987923102663976e-05, "loss": 3.8245, "step": 19820 }, { "epoch": 5.0043136, "grad_norm": 0.8293110132217407, "learning_rate": 1.987910622075985e-05, "loss": 3.7554, "step": 19830 }, { "epoch": 5.0043392, "grad_norm": 1.0497517585754395, "learning_rate": 1.9878981350816556e-05, "loss": 3.528, "step": 19840 }, { "epoch": 5.0043648, "grad_norm": 0.9831824898719788, "learning_rate": 1.9878856416810693e-05, "loss": 3.8307, "step": 19850 }, { "epoch": 5.0043904, "grad_norm": 0.9785191416740417, "learning_rate": 1.9878731418743075e-05, "loss": 3.6049, "step": 19860 }, { "epoch": 5.004416, "grad_norm": 0.9964720010757446, "learning_rate": 1.9878606356614504e-05, "loss": 3.6285, "step": 19870 }, { "epoch": 5.0044416, "grad_norm": 0.8937829732894897, "learning_rate": 1.98784812304258e-05, "loss": 3.3472, "step": 19880 }, { "epoch": 5.0044672, "grad_norm": 0.956779956817627, "learning_rate": 1.987835604017777e-05, "loss": 3.5368, "step": 19890 }, { "epoch": 5.0044928, "grad_norm": 0.933313250541687, "learning_rate": 1.987823078587123e-05, "loss": 3.6588, "step": 19900 }, { "epoch": 5.0045184, "grad_norm": 1.016363263130188, "learning_rate": 1.9878105467506983e-05, "loss": 3.6444, "step": 19910 }, { "epoch": 5.004544, "grad_norm": 0.9328101277351379, "learning_rate": 1.987798008508585e-05, "loss": 3.5509, "step": 19920 }, { "epoch": 5.0045696, "grad_norm": 0.7672699093818665, "learning_rate": 1.9877854638608638e-05, "loss": 3.7319, "step": 19930 }, { "epoch": 5.0045952, "grad_norm": 1.232246994972229, "learning_rate": 1.987772912807617e-05, "loss": 3.7588, "step": 19940 }, { "epoch": 5.0046208, "grad_norm": 0.9818084836006165, "learning_rate": 1.987760355348925e-05, "loss": 3.5355, "step": 19950 }, { "epoch": 5.0046464, "grad_norm": 1.496843934059143, "learning_rate": 1.9877477914848697e-05, "loss": 3.6132, "step": 19960 }, { "epoch": 5.004672, "grad_norm": 0.9659149050712585, "learning_rate": 1.987735221215532e-05, "loss": 3.6308, "step": 19970 }, { "epoch": 5.0046976, "grad_norm": 0.8900889754295349, "learning_rate": 1.987722644540995e-05, "loss": 3.6109, "step": 19980 }, { "epoch": 5.0047232, "grad_norm": 1.076918125152588, "learning_rate": 1.987710061461338e-05, "loss": 3.5189, "step": 19990 }, { "epoch": 5.0047488, "grad_norm": 0.8393824696540833, "learning_rate": 1.9876974719766445e-05, "loss": 3.6667, "step": 20000 }, { "epoch": 5.0047744, "grad_norm": 0.9633308053016663, "learning_rate": 1.9876848760869948e-05, "loss": 3.757, "step": 20010 }, { "epoch": 5.0048, "grad_norm": 0.9853694438934326, "learning_rate": 1.9876722737924718e-05, "loss": 3.5356, "step": 20020 }, { "epoch": 5.0048256, "grad_norm": 0.905399739742279, "learning_rate": 1.987659665093156e-05, "loss": 3.8527, "step": 20030 }, { "epoch": 5.0048512, "grad_norm": 1.1171985864639282, "learning_rate": 1.9876470499891303e-05, "loss": 3.5205, "step": 20040 }, { "epoch": 5.0048768, "grad_norm": 0.8833311796188354, "learning_rate": 1.9876344284804757e-05, "loss": 3.6694, "step": 20050 }, { "epoch": 5.0049024, "grad_norm": 0.9785189032554626, "learning_rate": 1.9876218005672743e-05, "loss": 3.6676, "step": 20060 }, { "epoch": 5.004928, "grad_norm": 0.8838850855827332, "learning_rate": 1.987609166249608e-05, "loss": 3.7421, "step": 20070 }, { "epoch": 5.0049536, "grad_norm": 1.061266541481018, "learning_rate": 1.987596525527559e-05, "loss": 3.7959, "step": 20080 }, { "epoch": 5.0049792, "grad_norm": 0.8770979642868042, "learning_rate": 1.9875838784012088e-05, "loss": 3.717, "step": 20090 }, { "epoch": 5.0050048, "grad_norm": 0.8923527598381042, "learning_rate": 1.9875712248706396e-05, "loss": 3.5914, "step": 20100 }, { "epoch": 5.0050304, "grad_norm": 0.8161956071853638, "learning_rate": 1.9875585649359332e-05, "loss": 3.5409, "step": 20110 }, { "epoch": 5.005056, "grad_norm": 1.2263246774673462, "learning_rate": 1.9875458985971722e-05, "loss": 3.6091, "step": 20120 }, { "epoch": 5.0050816, "grad_norm": 0.9512576460838318, "learning_rate": 1.987533225854439e-05, "loss": 3.4451, "step": 20130 }, { "epoch": 5.0051072, "grad_norm": 0.9134489893913269, "learning_rate": 1.9875205467078148e-05, "loss": 3.5175, "step": 20140 }, { "epoch": 5.0051328, "grad_norm": 0.8733143210411072, "learning_rate": 1.9875078611573823e-05, "loss": 3.5844, "step": 20150 }, { "epoch": 5.0051584, "grad_norm": 0.8987309336662292, "learning_rate": 1.9874951692032235e-05, "loss": 3.7241, "step": 20160 }, { "epoch": 5.005184, "grad_norm": 0.8768064975738525, "learning_rate": 1.9874824708454214e-05, "loss": 3.8417, "step": 20170 }, { "epoch": 5.0052096, "grad_norm": 0.8927314877510071, "learning_rate": 1.9874697660840577e-05, "loss": 3.5784, "step": 20180 }, { "epoch": 5.0052352, "grad_norm": 1.0574785470962524, "learning_rate": 1.987457054919215e-05, "loss": 3.7236, "step": 20190 }, { "epoch": 5.0052608, "grad_norm": 0.8132477402687073, "learning_rate": 1.987444337350976e-05, "loss": 3.5946, "step": 20200 }, { "epoch": 5.0052864, "grad_norm": 0.8781039714813232, "learning_rate": 1.987431613379423e-05, "loss": 3.6308, "step": 20210 }, { "epoch": 5.005312, "grad_norm": 1.0004243850708008, "learning_rate": 1.987418883004638e-05, "loss": 3.7263, "step": 20220 }, { "epoch": 5.0053376, "grad_norm": 0.9243001937866211, "learning_rate": 1.9874061462267042e-05, "loss": 3.6392, "step": 20230 }, { "epoch": 5.0053632, "grad_norm": 0.9342142343521118, "learning_rate": 1.9873934030457037e-05, "loss": 3.5934, "step": 20240 }, { "epoch": 5.0053888, "grad_norm": 0.9326561689376831, "learning_rate": 1.9873806534617197e-05, "loss": 3.6019, "step": 20250 }, { "epoch": 5.0054144, "grad_norm": 1.0080652236938477, "learning_rate": 1.9873678974748344e-05, "loss": 3.655, "step": 20260 }, { "epoch": 5.00544, "grad_norm": 1.1118723154067993, "learning_rate": 1.9873551350851308e-05, "loss": 3.6649, "step": 20270 }, { "epoch": 5.0054656, "grad_norm": 0.9995453953742981, "learning_rate": 1.9873423662926916e-05, "loss": 3.3827, "step": 20280 }, { "epoch": 5.0054912, "grad_norm": 0.8903359174728394, "learning_rate": 1.9873295910975994e-05, "loss": 3.6524, "step": 20290 }, { "epoch": 5.0055168, "grad_norm": 0.9300561547279358, "learning_rate": 1.9873168094999374e-05, "loss": 3.5598, "step": 20300 }, { "epoch": 5.0055424, "grad_norm": 1.0331305265426636, "learning_rate": 1.987304021499788e-05, "loss": 3.6435, "step": 20310 }, { "epoch": 5.005568, "grad_norm": 0.9346664547920227, "learning_rate": 1.9872912270972345e-05, "loss": 3.7839, "step": 20320 }, { "epoch": 5.0055936, "grad_norm": 0.855735719203949, "learning_rate": 1.98727842629236e-05, "loss": 3.6933, "step": 20330 }, { "epoch": 5.0056192, "grad_norm": 1.310826063156128, "learning_rate": 1.987265619085247e-05, "loss": 3.5121, "step": 20340 }, { "epoch": 5.0056448, "grad_norm": 1.0351752042770386, "learning_rate": 1.9872528054759792e-05, "loss": 3.6696, "step": 20350 }, { "epoch": 5.0056704, "grad_norm": 1.136296272277832, "learning_rate": 1.987239985464639e-05, "loss": 3.6624, "step": 20360 }, { "epoch": 5.005696, "grad_norm": 0.8792813420295715, "learning_rate": 1.9872271590513102e-05, "loss": 3.6494, "step": 20370 }, { "epoch": 5.0057216, "grad_norm": 0.9330699443817139, "learning_rate": 1.9872143262360754e-05, "loss": 3.4032, "step": 20380 }, { "epoch": 5.0057472, "grad_norm": 0.9005818963050842, "learning_rate": 1.987201487019018e-05, "loss": 3.3139, "step": 20390 }, { "epoch": 5.0057728, "grad_norm": 1.0027616024017334, "learning_rate": 1.9871886414002217e-05, "loss": 3.8532, "step": 20400 }, { "epoch": 5.0057984, "grad_norm": 0.9960856437683105, "learning_rate": 1.987175789379769e-05, "loss": 3.6948, "step": 20410 }, { "epoch": 5.005824, "grad_norm": 1.0800925493240356, "learning_rate": 1.9871629309577438e-05, "loss": 3.763, "step": 20420 }, { "epoch": 5.0058496, "grad_norm": 0.9356898665428162, "learning_rate": 1.9871500661342296e-05, "loss": 3.6874, "step": 20430 }, { "epoch": 5.0058752, "grad_norm": 0.8422794342041016, "learning_rate": 1.9871371949093094e-05, "loss": 3.4504, "step": 20440 }, { "epoch": 5.0059008, "grad_norm": 1.1570053100585938, "learning_rate": 1.9871243172830667e-05, "loss": 3.6121, "step": 20450 }, { "epoch": 5.0059264, "grad_norm": 1.0461722612380981, "learning_rate": 1.9871114332555854e-05, "loss": 3.8685, "step": 20460 }, { "epoch": 5.005952, "grad_norm": 1.114476203918457, "learning_rate": 1.9870985428269486e-05, "loss": 3.9782, "step": 20470 }, { "epoch": 5.0059776, "grad_norm": 0.8784810900688171, "learning_rate": 1.98708564599724e-05, "loss": 3.6428, "step": 20480 }, { "epoch": 5.0060032, "grad_norm": 1.1624438762664795, "learning_rate": 1.9870727427665434e-05, "loss": 3.6924, "step": 20490 }, { "epoch": 5.0060288, "grad_norm": 0.8483045101165771, "learning_rate": 1.9870598331349423e-05, "loss": 3.8127, "step": 20500 }, { "epoch": 5.0060544, "grad_norm": 0.9134243130683899, "learning_rate": 1.987046917102521e-05, "loss": 3.6138, "step": 20510 }, { "epoch": 5.00608, "grad_norm": 0.812853217124939, "learning_rate": 1.9870339946693623e-05, "loss": 3.6471, "step": 20520 }, { "epoch": 5.0061056, "grad_norm": 0.9015008807182312, "learning_rate": 1.9870210658355503e-05, "loss": 3.6557, "step": 20530 }, { "epoch": 5.0061312, "grad_norm": 0.9019323587417603, "learning_rate": 1.9870081306011695e-05, "loss": 3.6454, "step": 20540 }, { "epoch": 5.0061568, "grad_norm": 1.2034472227096558, "learning_rate": 1.9869951889663032e-05, "loss": 3.728, "step": 20550 }, { "epoch": 5.0061824, "grad_norm": 0.9709050059318542, "learning_rate": 1.986982240931035e-05, "loss": 3.6144, "step": 20560 }, { "epoch": 5.006208, "grad_norm": 1.0843837261199951, "learning_rate": 1.98696928649545e-05, "loss": 3.5621, "step": 20570 }, { "epoch": 5.0062336, "grad_norm": 0.9465495347976685, "learning_rate": 1.9869563256596308e-05, "loss": 3.5924, "step": 20580 }, { "epoch": 5.0062592, "grad_norm": 1.1300389766693115, "learning_rate": 1.9869433584236625e-05, "loss": 3.594, "step": 20590 }, { "epoch": 5.0062848, "grad_norm": 0.9293555617332458, "learning_rate": 1.986930384787629e-05, "loss": 3.6031, "step": 20600 }, { "epoch": 5.0063104, "grad_norm": 0.9585537910461426, "learning_rate": 1.9869174047516138e-05, "loss": 3.7962, "step": 20610 }, { "epoch": 5.006336, "grad_norm": 1.2298730611801147, "learning_rate": 1.9869044183157017e-05, "loss": 3.7008, "step": 20620 }, { "epoch": 5.0063616, "grad_norm": 0.8942738175392151, "learning_rate": 1.9868914254799766e-05, "loss": 3.6007, "step": 20630 }, { "epoch": 5.0063872, "grad_norm": 0.8383017182350159, "learning_rate": 1.986878426244523e-05, "loss": 3.572, "step": 20640 }, { "epoch": 5.0064128, "grad_norm": 0.9431827664375305, "learning_rate": 1.9868654206094252e-05, "loss": 3.5501, "step": 20650 }, { "epoch": 5.0064384, "grad_norm": 0.8096798062324524, "learning_rate": 1.986852408574767e-05, "loss": 3.5146, "step": 20660 }, { "epoch": 5.006464, "grad_norm": 1.073793888092041, "learning_rate": 1.9868393901406335e-05, "loss": 3.5464, "step": 20670 }, { "epoch": 5.0064896, "grad_norm": 0.960399329662323, "learning_rate": 1.986826365307109e-05, "loss": 3.6118, "step": 20680 }, { "epoch": 5.0065152, "grad_norm": 0.9485101103782654, "learning_rate": 1.9868133340742778e-05, "loss": 3.5304, "step": 20690 }, { "epoch": 5.0065408, "grad_norm": 0.8604745268821716, "learning_rate": 1.986800296442224e-05, "loss": 3.5398, "step": 20700 }, { "epoch": 5.0065664, "grad_norm": 0.9334549307823181, "learning_rate": 1.9867872524110327e-05, "loss": 3.5224, "step": 20710 }, { "epoch": 5.006592, "grad_norm": 1.0313233137130737, "learning_rate": 1.9867742019807884e-05, "loss": 3.6598, "step": 20720 }, { "epoch": 5.0066176, "grad_norm": 1.1702232360839844, "learning_rate": 1.9867611451515757e-05, "loss": 3.4213, "step": 20730 }, { "epoch": 5.0066432, "grad_norm": 1.6570030450820923, "learning_rate": 1.986748081923479e-05, "loss": 3.7976, "step": 20740 }, { "epoch": 5.0066688, "grad_norm": 1.3105945587158203, "learning_rate": 1.9867350122965835e-05, "loss": 3.5795, "step": 20750 }, { "epoch": 5.0066944, "grad_norm": 1.1191003322601318, "learning_rate": 1.9867219362709737e-05, "loss": 3.3674, "step": 20760 }, { "epoch": 5.00672, "grad_norm": 1.1383099555969238, "learning_rate": 1.9867088538467344e-05, "loss": 3.5909, "step": 20770 }, { "epoch": 5.0067456, "grad_norm": 0.9377251267433167, "learning_rate": 1.9866957650239503e-05, "loss": 3.6114, "step": 20780 }, { "epoch": 5.0067712, "grad_norm": 1.133545160293579, "learning_rate": 1.9866826698027064e-05, "loss": 3.5766, "step": 20790 }, { "epoch": 5.0067968, "grad_norm": 1.0133014917373657, "learning_rate": 1.9866695681830874e-05, "loss": 3.7426, "step": 20800 }, { "epoch": 5.0068224, "grad_norm": 0.8815727829933167, "learning_rate": 1.9866564601651787e-05, "loss": 3.6408, "step": 20810 }, { "epoch": 5.006848, "grad_norm": 0.9121434092521667, "learning_rate": 1.986643345749065e-05, "loss": 3.5467, "step": 20820 }, { "epoch": 5.0068736, "grad_norm": 0.8365052938461304, "learning_rate": 1.9866302249348318e-05, "loss": 3.8494, "step": 20830 }, { "epoch": 5.0068992, "grad_norm": 0.9072020649909973, "learning_rate": 1.9866170977225632e-05, "loss": 3.6007, "step": 20840 }, { "epoch": 5.0069248, "grad_norm": 0.9765729904174805, "learning_rate": 1.9866039641123452e-05, "loss": 3.8053, "step": 20850 }, { "epoch": 5.0069504, "grad_norm": 0.9245471358299255, "learning_rate": 1.986590824104263e-05, "loss": 3.6301, "step": 20860 }, { "epoch": 5.006976, "grad_norm": 1.9889181852340698, "learning_rate": 1.986577677698401e-05, "loss": 3.7333, "step": 20870 }, { "epoch": 5.0070016, "grad_norm": 1.0142418146133423, "learning_rate": 1.986564524894845e-05, "loss": 3.6897, "step": 20880 }, { "epoch": 5.0070272, "grad_norm": 1.1442248821258545, "learning_rate": 1.986551365693681e-05, "loss": 3.6113, "step": 20890 }, { "epoch": 5.0070528, "grad_norm": 1.0519144535064697, "learning_rate": 1.9865382000949926e-05, "loss": 3.6767, "step": 20900 }, { "epoch": 5.0070784, "grad_norm": 0.9811773896217346, "learning_rate": 1.986525028098867e-05, "loss": 3.7086, "step": 20910 }, { "epoch": 5.007104, "grad_norm": 0.8601074814796448, "learning_rate": 1.9865118497053882e-05, "loss": 3.8413, "step": 20920 }, { "epoch": 5.0071296, "grad_norm": 1.0343043804168701, "learning_rate": 1.9864986649146422e-05, "loss": 3.5573, "step": 20930 }, { "epoch": 5.0071552, "grad_norm": 1.0299640893936157, "learning_rate": 1.9864854737267148e-05, "loss": 3.694, "step": 20940 }, { "epoch": 5.0071808, "grad_norm": 0.8815767765045166, "learning_rate": 1.9864722761416913e-05, "loss": 3.5158, "step": 20950 }, { "epoch": 5.0072064, "grad_norm": 1.1832581758499146, "learning_rate": 1.9864590721596572e-05, "loss": 3.6662, "step": 20960 }, { "epoch": 5.007232, "grad_norm": 0.9321957230567932, "learning_rate": 1.986445861780698e-05, "loss": 3.6334, "step": 20970 }, { "epoch": 5.0072576, "grad_norm": 0.8200331330299377, "learning_rate": 1.9864326450048998e-05, "loss": 3.4621, "step": 20980 }, { "epoch": 5.0072832, "grad_norm": 1.0783493518829346, "learning_rate": 1.986419421832348e-05, "loss": 3.544, "step": 20990 }, { "epoch": 5.0073088, "grad_norm": 0.9095908999443054, "learning_rate": 1.9864061922631283e-05, "loss": 3.6367, "step": 21000 }, { "epoch": 5.0073344, "grad_norm": 1.1062157154083252, "learning_rate": 1.9863929562973266e-05, "loss": 3.7195, "step": 21010 }, { "epoch": 5.00736, "grad_norm": 1.1005058288574219, "learning_rate": 1.986379713935029e-05, "loss": 3.9413, "step": 21020 }, { "epoch": 5.0073856, "grad_norm": 0.9279658794403076, "learning_rate": 1.986366465176321e-05, "loss": 3.7718, "step": 21030 }, { "epoch": 5.0074112, "grad_norm": 0.9362602233886719, "learning_rate": 1.9863532100212883e-05, "loss": 3.5965, "step": 21040 }, { "epoch": 5.0074368, "grad_norm": 1.0553781986236572, "learning_rate": 1.9863399484700175e-05, "loss": 3.553, "step": 21050 }, { "epoch": 5.0074624, "grad_norm": 0.9207058548927307, "learning_rate": 1.9863266805225946e-05, "loss": 3.5778, "step": 21060 }, { "epoch": 5.007488, "grad_norm": 1.1150860786437988, "learning_rate": 1.9863134061791046e-05, "loss": 3.5436, "step": 21070 }, { "epoch": 5.0075136, "grad_norm": 0.9561945199966431, "learning_rate": 1.9863001254396346e-05, "loss": 3.6282, "step": 21080 }, { "epoch": 5.0075392, "grad_norm": 1.233597755432129, "learning_rate": 1.9862868383042703e-05, "loss": 3.5676, "step": 21090 }, { "epoch": 5.0075648, "grad_norm": 1.151341438293457, "learning_rate": 1.9862735447730982e-05, "loss": 3.7936, "step": 21100 }, { "epoch": 5.0075904, "grad_norm": 0.9171431660652161, "learning_rate": 1.9862602448462042e-05, "loss": 3.5859, "step": 21110 }, { "epoch": 5.007616, "grad_norm": 1.0034949779510498, "learning_rate": 1.9862469385236744e-05, "loss": 3.5394, "step": 21120 }, { "epoch": 5.0076416, "grad_norm": 0.8972070813179016, "learning_rate": 1.9862336258055956e-05, "loss": 3.6517, "step": 21130 }, { "epoch": 5.0076672, "grad_norm": 0.86176598072052, "learning_rate": 1.986220306692054e-05, "loss": 3.6178, "step": 21140 }, { "epoch": 5.0076928, "grad_norm": 0.8799459934234619, "learning_rate": 1.9862069811831355e-05, "loss": 3.5129, "step": 21150 }, { "epoch": 5.0077184, "grad_norm": 1.027021050453186, "learning_rate": 1.986193649278927e-05, "loss": 3.4797, "step": 21160 }, { "epoch": 5.007744, "grad_norm": 0.9746131896972656, "learning_rate": 1.9861803109795143e-05, "loss": 3.6373, "step": 21170 }, { "epoch": 5.0077696, "grad_norm": 2.075528621673584, "learning_rate": 1.986166966284985e-05, "loss": 4.0651, "step": 21180 }, { "epoch": 5.0077952, "grad_norm": 1.3076157569885254, "learning_rate": 1.986153615195425e-05, "loss": 3.9246, "step": 21190 }, { "epoch": 5.0078208, "grad_norm": 0.8625836968421936, "learning_rate": 1.9861402577109207e-05, "loss": 3.4516, "step": 21200 }, { "epoch": 5.0078464, "grad_norm": 0.918807864189148, "learning_rate": 1.986126893831559e-05, "loss": 3.6538, "step": 21210 }, { "epoch": 5.007872, "grad_norm": 0.9654533863067627, "learning_rate": 1.986113523557426e-05, "loss": 3.607, "step": 21220 }, { "epoch": 5.0078976, "grad_norm": 0.882384181022644, "learning_rate": 1.986100146888609e-05, "loss": 3.46, "step": 21230 }, { "epoch": 5.0079232, "grad_norm": 1.0767085552215576, "learning_rate": 1.986086763825195e-05, "loss": 3.6192, "step": 21240 }, { "epoch": 5.0079488, "grad_norm": 0.9302254915237427, "learning_rate": 1.9860733743672703e-05, "loss": 3.5982, "step": 21250 }, { "epoch": 5.0079744, "grad_norm": 0.9080090522766113, "learning_rate": 1.9860599785149216e-05, "loss": 3.5261, "step": 21260 }, { "epoch": 5.008, "grad_norm": 0.9624396562576294, "learning_rate": 1.9860465762682365e-05, "loss": 3.5176, "step": 21270 }, { "epoch": 5.0080256, "grad_norm": 0.9145578145980835, "learning_rate": 1.986033167627301e-05, "loss": 3.6019, "step": 21280 }, { "epoch": 5.0080512, "grad_norm": 1.1898853778839111, "learning_rate": 1.9860197525922027e-05, "loss": 3.601, "step": 21290 }, { "epoch": 5.0080768, "grad_norm": 0.9032965302467346, "learning_rate": 1.986006331163028e-05, "loss": 3.4385, "step": 21300 }, { "epoch": 5.0081024, "grad_norm": 1.0417330265045166, "learning_rate": 1.9859929033398647e-05, "loss": 3.5653, "step": 21310 }, { "epoch": 5.008128, "grad_norm": 1.355574369430542, "learning_rate": 1.9859794691227995e-05, "loss": 3.4307, "step": 21320 }, { "epoch": 5.0081536, "grad_norm": 1.1477489471435547, "learning_rate": 1.985966028511919e-05, "loss": 3.5917, "step": 21330 }, { "epoch": 5.0081792, "grad_norm": 1.0541682243347168, "learning_rate": 1.9859525815073114e-05, "loss": 3.7308, "step": 21340 }, { "epoch": 5.0082048, "grad_norm": 0.9677797555923462, "learning_rate": 1.9859391281090628e-05, "loss": 3.7687, "step": 21350 }, { "epoch": 5.0082304, "grad_norm": 1.020050048828125, "learning_rate": 1.9859256683172615e-05, "loss": 3.6715, "step": 21360 }, { "epoch": 5.008256, "grad_norm": 0.8459165692329407, "learning_rate": 1.985912202131994e-05, "loss": 3.5809, "step": 21370 }, { "epoch": 5.0082816, "grad_norm": 0.9039878249168396, "learning_rate": 1.9858987295533477e-05, "loss": 3.71, "step": 21380 }, { "epoch": 5.0083072, "grad_norm": 0.8940899968147278, "learning_rate": 1.9858852505814106e-05, "loss": 3.3842, "step": 21390 }, { "epoch": 5.0083328, "grad_norm": 0.8694629073143005, "learning_rate": 1.9858717652162694e-05, "loss": 3.615, "step": 21400 }, { "epoch": 5.0083584, "grad_norm": 0.8897743225097656, "learning_rate": 1.985858273458012e-05, "loss": 3.4094, "step": 21410 }, { "epoch": 5.008384, "grad_norm": 1.0438646078109741, "learning_rate": 1.9858447753067254e-05, "loss": 3.6472, "step": 21420 }, { "epoch": 5.0084096, "grad_norm": 0.8644671440124512, "learning_rate": 1.9858312707624977e-05, "loss": 3.6105, "step": 21430 }, { "epoch": 5.0084352, "grad_norm": 1.1165801286697388, "learning_rate": 1.985817759825416e-05, "loss": 3.3353, "step": 21440 }, { "epoch": 5.0084608, "grad_norm": 1.2607519626617432, "learning_rate": 1.985804242495568e-05, "loss": 3.4742, "step": 21450 }, { "epoch": 5.0084864, "grad_norm": 0.9147409796714783, "learning_rate": 1.985790718773042e-05, "loss": 3.6492, "step": 21460 }, { "epoch": 5.008512, "grad_norm": 1.0462597608566284, "learning_rate": 1.9857771886579247e-05, "loss": 3.5029, "step": 21470 }, { "epoch": 5.0085376, "grad_norm": 0.9288604855537415, "learning_rate": 1.9857636521503047e-05, "loss": 3.5382, "step": 21480 }, { "epoch": 5.0085632, "grad_norm": 0.9728196263313293, "learning_rate": 1.9857501092502693e-05, "loss": 3.7692, "step": 21490 }, { "epoch": 5.0085888, "grad_norm": 1.3163948059082031, "learning_rate": 1.9857365599579063e-05, "loss": 3.5574, "step": 21500 }, { "epoch": 5.0086144, "grad_norm": 0.9411855340003967, "learning_rate": 1.9857230042733038e-05, "loss": 3.6426, "step": 21510 }, { "epoch": 5.00864, "grad_norm": 0.9372824430465698, "learning_rate": 1.9857094421965496e-05, "loss": 3.5407, "step": 21520 }, { "epoch": 5.0086656, "grad_norm": 1.308225393295288, "learning_rate": 1.9856958737277313e-05, "loss": 3.6209, "step": 21530 }, { "epoch": 5.0086912, "grad_norm": 0.9194871187210083, "learning_rate": 1.9856822988669378e-05, "loss": 3.6493, "step": 21540 }, { "epoch": 5.0087168, "grad_norm": 2.3280229568481445, "learning_rate": 1.985668717614256e-05, "loss": 3.7546, "step": 21550 }, { "epoch": 5.0087424, "grad_norm": 1.0317227840423584, "learning_rate": 1.9856551299697748e-05, "loss": 3.6066, "step": 21560 }, { "epoch": 5.008768, "grad_norm": 1.3491462469100952, "learning_rate": 1.9856415359335818e-05, "loss": 3.522, "step": 21570 }, { "epoch": 5.0087936, "grad_norm": 1.1149942874908447, "learning_rate": 1.985627935505766e-05, "loss": 3.7257, "step": 21580 }, { "epoch": 5.0088192, "grad_norm": 1.0765010118484497, "learning_rate": 1.9856143286864142e-05, "loss": 3.6635, "step": 21590 }, { "epoch": 5.0088448, "grad_norm": 0.8968810439109802, "learning_rate": 1.9856007154756162e-05, "loss": 3.5947, "step": 21600 }, { "epoch": 5.0088704, "grad_norm": 0.9388608336448669, "learning_rate": 1.985587095873459e-05, "loss": 3.5524, "step": 21610 }, { "epoch": 5.008896, "grad_norm": 0.8871246576309204, "learning_rate": 1.9855734698800315e-05, "loss": 3.4569, "step": 21620 }, { "epoch": 5.0089216, "grad_norm": 1.1434757709503174, "learning_rate": 1.985559837495422e-05, "loss": 3.6056, "step": 21630 }, { "epoch": 5.0089472, "grad_norm": 1.0732030868530273, "learning_rate": 1.9855461987197188e-05, "loss": 3.6314, "step": 21640 }, { "epoch": 5.0089728, "grad_norm": 0.847326934337616, "learning_rate": 1.9855325535530104e-05, "loss": 3.5962, "step": 21650 }, { "epoch": 5.0089984, "grad_norm": 0.8850942850112915, "learning_rate": 1.9855189019953856e-05, "loss": 3.295, "step": 21660 }, { "epoch": 5.009024, "grad_norm": 1.0516775846481323, "learning_rate": 1.9855052440469324e-05, "loss": 3.234, "step": 21670 }, { "epoch": 5.0090496, "grad_norm": 1.0484367609024048, "learning_rate": 1.98549157970774e-05, "loss": 3.572, "step": 21680 }, { "epoch": 5.0090752, "grad_norm": 1.0034716129302979, "learning_rate": 1.985477908977896e-05, "loss": 3.5176, "step": 21690 }, { "epoch": 5.0091008, "grad_norm": 0.8641106486320496, "learning_rate": 1.98546423185749e-05, "loss": 3.6147, "step": 21700 }, { "epoch": 5.0091264, "grad_norm": 0.9613854289054871, "learning_rate": 1.9854505483466106e-05, "loss": 3.5651, "step": 21710 }, { "epoch": 5.009152, "grad_norm": 1.7063300609588623, "learning_rate": 1.985436858445346e-05, "loss": 3.8494, "step": 21720 }, { "epoch": 5.0091776, "grad_norm": 0.84525465965271, "learning_rate": 1.985423162153785e-05, "loss": 3.5509, "step": 21730 }, { "epoch": 5.0092032, "grad_norm": 0.9652246236801147, "learning_rate": 1.9854094594720173e-05, "loss": 3.8514, "step": 21740 }, { "epoch": 5.0092288, "grad_norm": 1.4731849431991577, "learning_rate": 1.9853957504001306e-05, "loss": 3.5413, "step": 21750 }, { "epoch": 5.0092544, "grad_norm": 1.1367361545562744, "learning_rate": 1.9853820349382145e-05, "loss": 3.6265, "step": 21760 }, { "epoch": 5.00928, "grad_norm": 1.1041395664215088, "learning_rate": 1.985368313086358e-05, "loss": 3.5269, "step": 21770 }, { "epoch": 6.00001536, "grad_norm": 0.9315940141677856, "learning_rate": 1.9853545848446498e-05, "loss": 3.9418, "step": 21780 }, { "epoch": 6.00004096, "grad_norm": 0.9272583723068237, "learning_rate": 1.985340850213179e-05, "loss": 3.4595, "step": 21790 }, { "epoch": 6.00006656, "grad_norm": 0.9280377626419067, "learning_rate": 1.9853271091920347e-05, "loss": 3.279, "step": 21800 }, { "epoch": 6.00009216, "grad_norm": 0.9172782897949219, "learning_rate": 1.9853133617813057e-05, "loss": 3.5231, "step": 21810 }, { "epoch": 6.00011776, "grad_norm": 1.7153499126434326, "learning_rate": 1.9852996079810816e-05, "loss": 3.493, "step": 21820 }, { "epoch": 6.00014336, "grad_norm": 1.1186741590499878, "learning_rate": 1.985285847791451e-05, "loss": 3.3836, "step": 21830 }, { "epoch": 6.00016896, "grad_norm": 0.9395714402198792, "learning_rate": 1.985272081212504e-05, "loss": 3.6687, "step": 21840 }, { "epoch": 6.00019456, "grad_norm": 1.1350083351135254, "learning_rate": 1.9852583082443296e-05, "loss": 3.9485, "step": 21850 }, { "epoch": 6.00022016, "grad_norm": 1.2404857873916626, "learning_rate": 1.9852445288870164e-05, "loss": 3.3959, "step": 21860 }, { "epoch": 6.00024576, "grad_norm": 1.0128862857818604, "learning_rate": 1.9852307431406547e-05, "loss": 3.8558, "step": 21870 }, { "epoch": 6.00027136, "grad_norm": 0.9909709095954895, "learning_rate": 1.9852169510053332e-05, "loss": 3.626, "step": 21880 }, { "epoch": 6.00029696, "grad_norm": 1.166375994682312, "learning_rate": 1.9852031524811416e-05, "loss": 3.3365, "step": 21890 }, { "epoch": 6.00032256, "grad_norm": 0.8911279439926147, "learning_rate": 1.9851893475681695e-05, "loss": 3.2197, "step": 21900 }, { "epoch": 6.00034816, "grad_norm": 1.0814683437347412, "learning_rate": 1.9851755362665063e-05, "loss": 3.3457, "step": 21910 }, { "epoch": 6.00037376, "grad_norm": 0.9757185578346252, "learning_rate": 1.9851617185762414e-05, "loss": 3.4361, "step": 21920 }, { "epoch": 6.00039936, "grad_norm": 0.8420057892799377, "learning_rate": 1.985147894497465e-05, "loss": 3.5342, "step": 21930 }, { "epoch": 6.00042496, "grad_norm": 0.8393572568893433, "learning_rate": 1.985134064030266e-05, "loss": 3.4514, "step": 21940 }, { "epoch": 6.00045056, "grad_norm": 1.2733182907104492, "learning_rate": 1.9851202271747346e-05, "loss": 3.6576, "step": 21950 }, { "epoch": 6.00047616, "grad_norm": 0.8036996722221375, "learning_rate": 1.9851063839309602e-05, "loss": 3.4396, "step": 21960 }, { "epoch": 6.00050176, "grad_norm": 0.9370121955871582, "learning_rate": 1.9850925342990327e-05, "loss": 3.5087, "step": 21970 }, { "epoch": 6.00052736, "grad_norm": 1.4224975109100342, "learning_rate": 1.9850786782790423e-05, "loss": 3.333, "step": 21980 }, { "epoch": 6.00055296, "grad_norm": 0.7999767661094666, "learning_rate": 1.985064815871078e-05, "loss": 3.4412, "step": 21990 }, { "epoch": 6.00057856, "grad_norm": 1.1425944566726685, "learning_rate": 1.9850509470752303e-05, "loss": 3.179, "step": 22000 }, { "epoch": 6.00060416, "grad_norm": 0.7911369800567627, "learning_rate": 1.985037071891589e-05, "loss": 3.2514, "step": 22010 }, { "epoch": 6.00062976, "grad_norm": 0.9561915993690491, "learning_rate": 1.9850231903202443e-05, "loss": 3.4926, "step": 22020 }, { "epoch": 6.00065536, "grad_norm": 1.0161044597625732, "learning_rate": 1.9850093023612855e-05, "loss": 3.3977, "step": 22030 }, { "epoch": 6.00068096, "grad_norm": 0.8929912447929382, "learning_rate": 1.9849954080148035e-05, "loss": 3.5045, "step": 22040 }, { "epoch": 6.00070656, "grad_norm": 1.2449575662612915, "learning_rate": 1.984981507280888e-05, "loss": 4.1757, "step": 22050 }, { "epoch": 6.00073216, "grad_norm": 0.9008142352104187, "learning_rate": 1.9849676001596293e-05, "loss": 3.3678, "step": 22060 }, { "epoch": 6.00075776, "grad_norm": 0.9929481148719788, "learning_rate": 1.9849536866511173e-05, "loss": 3.4532, "step": 22070 }, { "epoch": 6.00078336, "grad_norm": 0.9942212700843811, "learning_rate": 1.9849397667554423e-05, "loss": 3.4405, "step": 22080 }, { "epoch": 6.00080896, "grad_norm": 0.9344143867492676, "learning_rate": 1.9849258404726953e-05, "loss": 3.7059, "step": 22090 }, { "epoch": 6.00083456, "grad_norm": 1.2954798936843872, "learning_rate": 1.9849119078029652e-05, "loss": 3.3412, "step": 22100 }, { "epoch": 6.00086016, "grad_norm": 0.819534957408905, "learning_rate": 1.9848979687463438e-05, "loss": 3.4033, "step": 22110 }, { "epoch": 6.00088576, "grad_norm": 0.989454984664917, "learning_rate": 1.9848840233029204e-05, "loss": 3.4079, "step": 22120 }, { "epoch": 6.00091136, "grad_norm": 1.588668704032898, "learning_rate": 1.984870071472786e-05, "loss": 3.5694, "step": 22130 }, { "epoch": 6.00093696, "grad_norm": 0.9040716290473938, "learning_rate": 1.984856113256031e-05, "loss": 3.4724, "step": 22140 }, { "epoch": 6.00096256, "grad_norm": 1.4885178804397583, "learning_rate": 1.984842148652746e-05, "loss": 3.5404, "step": 22150 }, { "epoch": 6.00098816, "grad_norm": 0.8786513805389404, "learning_rate": 1.9848281776630207e-05, "loss": 3.5423, "step": 22160 }, { "epoch": 6.00101376, "grad_norm": 1.0796295404434204, "learning_rate": 1.9848142002869472e-05, "loss": 3.5432, "step": 22170 }, { "epoch": 6.00103936, "grad_norm": 1.0602816343307495, "learning_rate": 1.9848002165246146e-05, "loss": 3.3253, "step": 22180 }, { "epoch": 6.00106496, "grad_norm": 0.8346995115280151, "learning_rate": 1.984786226376115e-05, "loss": 3.3027, "step": 22190 }, { "epoch": 6.00109056, "grad_norm": 1.0485035181045532, "learning_rate": 1.9847722298415382e-05, "loss": 3.4899, "step": 22200 }, { "epoch": 6.00111616, "grad_norm": 1.040126919746399, "learning_rate": 1.9847582269209754e-05, "loss": 3.3498, "step": 22210 }, { "epoch": 6.00114176, "grad_norm": 0.9467681646347046, "learning_rate": 1.9847442176145167e-05, "loss": 3.4167, "step": 22220 }, { "epoch": 6.00116736, "grad_norm": 0.9514057040214539, "learning_rate": 1.984730201922254e-05, "loss": 3.2838, "step": 22230 }, { "epoch": 6.00119296, "grad_norm": 0.8728824257850647, "learning_rate": 1.9847161798442772e-05, "loss": 3.6455, "step": 22240 }, { "epoch": 6.00121856, "grad_norm": 0.9738779067993164, "learning_rate": 1.984702151380678e-05, "loss": 3.1744, "step": 22250 }, { "epoch": 6.00124416, "grad_norm": 0.9389808773994446, "learning_rate": 1.984688116531547e-05, "loss": 3.5124, "step": 22260 }, { "epoch": 6.00126976, "grad_norm": 0.837030827999115, "learning_rate": 1.9846740752969752e-05, "loss": 3.4619, "step": 22270 }, { "epoch": 6.00129536, "grad_norm": 0.8739522099494934, "learning_rate": 1.984660027677054e-05, "loss": 3.518, "step": 22280 }, { "epoch": 6.00132096, "grad_norm": 0.8689491748809814, "learning_rate": 1.984645973671874e-05, "loss": 3.3368, "step": 22290 }, { "epoch": 6.00134656, "grad_norm": 0.9490770101547241, "learning_rate": 1.9846319132815263e-05, "loss": 3.6019, "step": 22300 }, { "epoch": 6.00137216, "grad_norm": 0.8908547163009644, "learning_rate": 1.9846178465061027e-05, "loss": 3.4591, "step": 22310 }, { "epoch": 6.00139776, "grad_norm": 0.9270046949386597, "learning_rate": 1.9846037733456936e-05, "loss": 3.3075, "step": 22320 }, { "epoch": 6.00142336, "grad_norm": 1.0630298852920532, "learning_rate": 1.984589693800391e-05, "loss": 3.6001, "step": 22330 }, { "epoch": 6.00144896, "grad_norm": 1.4413440227508545, "learning_rate": 1.984575607870286e-05, "loss": 3.5559, "step": 22340 }, { "epoch": 6.00147456, "grad_norm": 1.9647103548049927, "learning_rate": 1.9845615155554697e-05, "loss": 3.6812, "step": 22350 }, { "epoch": 6.00150016, "grad_norm": 1.0331560373306274, "learning_rate": 1.984547416856034e-05, "loss": 3.4756, "step": 22360 }, { "epoch": 6.00152576, "grad_norm": 0.9159643650054932, "learning_rate": 1.9845333117720695e-05, "loss": 3.397, "step": 22370 }, { "epoch": 6.00155136, "grad_norm": 1.3350623846054077, "learning_rate": 1.9845192003036683e-05, "loss": 3.564, "step": 22380 }, { "epoch": 6.00157696, "grad_norm": 0.9780877828598022, "learning_rate": 1.9845050824509217e-05, "loss": 3.7514, "step": 22390 }, { "epoch": 6.00160256, "grad_norm": 2.6164305210113525, "learning_rate": 1.9844909582139214e-05, "loss": 3.6262, "step": 22400 }, { "epoch": 6.00162816, "grad_norm": 0.9168088436126709, "learning_rate": 1.9844768275927585e-05, "loss": 3.5269, "step": 22410 }, { "epoch": 6.00165376, "grad_norm": 0.9505119323730469, "learning_rate": 1.9844626905875253e-05, "loss": 3.5535, "step": 22420 }, { "epoch": 6.00167936, "grad_norm": 1.1092889308929443, "learning_rate": 1.9844485471983135e-05, "loss": 3.7035, "step": 22430 }, { "epoch": 6.00170496, "grad_norm": 1.8206716775894165, "learning_rate": 1.984434397425214e-05, "loss": 3.634, "step": 22440 }, { "epoch": 6.00173056, "grad_norm": 0.9548474550247192, "learning_rate": 1.9844202412683194e-05, "loss": 3.6464, "step": 22450 }, { "epoch": 6.00175616, "grad_norm": 1.1021654605865479, "learning_rate": 1.9844060787277208e-05, "loss": 3.6052, "step": 22460 }, { "epoch": 6.00178176, "grad_norm": 0.9561390280723572, "learning_rate": 1.984391909803511e-05, "loss": 3.7278, "step": 22470 }, { "epoch": 6.00180736, "grad_norm": 1.445909023284912, "learning_rate": 1.9843777344957805e-05, "loss": 3.7603, "step": 22480 }, { "epoch": 6.00183296, "grad_norm": 0.8927164077758789, "learning_rate": 1.9843635528046224e-05, "loss": 3.4733, "step": 22490 }, { "epoch": 6.00185856, "grad_norm": 1.1853920221328735, "learning_rate": 1.9843493647301286e-05, "loss": 3.5553, "step": 22500 }, { "epoch": 6.00188416, "grad_norm": 1.3130899667739868, "learning_rate": 1.9843351702723903e-05, "loss": 3.4026, "step": 22510 }, { "epoch": 6.00190976, "grad_norm": 0.9194611310958862, "learning_rate": 1.9843209694315005e-05, "loss": 3.3854, "step": 22520 }, { "epoch": 6.00193536, "grad_norm": 0.9190330505371094, "learning_rate": 1.9843067622075504e-05, "loss": 3.3195, "step": 22530 }, { "epoch": 6.00196096, "grad_norm": 0.8606675863265991, "learning_rate": 1.9842925486006327e-05, "loss": 3.4754, "step": 22540 }, { "epoch": 6.00198656, "grad_norm": 1.2891103029251099, "learning_rate": 1.9842783286108393e-05, "loss": 3.5699, "step": 22550 }, { "epoch": 6.00201216, "grad_norm": 0.9286776781082153, "learning_rate": 1.9842641022382626e-05, "loss": 3.6497, "step": 22560 }, { "epoch": 6.00203776, "grad_norm": 1.0489343404769897, "learning_rate": 1.984249869482995e-05, "loss": 3.2469, "step": 22570 }, { "epoch": 6.00206336, "grad_norm": 1.088576316833496, "learning_rate": 1.9842356303451283e-05, "loss": 3.5157, "step": 22580 }, { "epoch": 6.00208896, "grad_norm": 0.9554169178009033, "learning_rate": 1.9842213848247552e-05, "loss": 3.5872, "step": 22590 }, { "epoch": 6.00211456, "grad_norm": 0.949150025844574, "learning_rate": 1.984207132921968e-05, "loss": 3.5805, "step": 22600 }, { "epoch": 6.00214016, "grad_norm": 1.0500894784927368, "learning_rate": 1.9841928746368588e-05, "loss": 3.497, "step": 22610 }, { "epoch": 6.00216576, "grad_norm": 1.1601513624191284, "learning_rate": 1.9841786099695207e-05, "loss": 3.6185, "step": 22620 }, { "epoch": 6.00219136, "grad_norm": 0.9790710210800171, "learning_rate": 1.984164338920046e-05, "loss": 3.3057, "step": 22630 }, { "epoch": 6.00221696, "grad_norm": 0.9805000424385071, "learning_rate": 1.984150061488527e-05, "loss": 3.4368, "step": 22640 }, { "epoch": 6.00224256, "grad_norm": 1.17438542842865, "learning_rate": 1.9841357776750563e-05, "loss": 3.5415, "step": 22650 }, { "epoch": 6.00226816, "grad_norm": 1.0667064189910889, "learning_rate": 1.9841214874797265e-05, "loss": 3.6757, "step": 22660 }, { "epoch": 6.00229376, "grad_norm": 1.0849783420562744, "learning_rate": 1.9841071909026306e-05, "loss": 3.5434, "step": 22670 }, { "epoch": 6.00231936, "grad_norm": 1.0280050039291382, "learning_rate": 1.9840928879438613e-05, "loss": 3.5468, "step": 22680 }, { "epoch": 6.00234496, "grad_norm": 1.7654035091400146, "learning_rate": 1.9840785786035107e-05, "loss": 3.4688, "step": 22690 }, { "epoch": 6.00237056, "grad_norm": 1.2136527299880981, "learning_rate": 1.9840642628816727e-05, "loss": 3.6552, "step": 22700 }, { "epoch": 6.00239616, "grad_norm": 1.1108427047729492, "learning_rate": 1.984049940778439e-05, "loss": 3.4848, "step": 22710 }, { "epoch": 6.00242176, "grad_norm": 1.0160720348358154, "learning_rate": 1.984035612293903e-05, "loss": 3.5449, "step": 22720 }, { "epoch": 6.00244736, "grad_norm": 1.0160369873046875, "learning_rate": 1.984021277428158e-05, "loss": 3.4817, "step": 22730 }, { "epoch": 6.00247296, "grad_norm": 0.8867342472076416, "learning_rate": 1.984006936181296e-05, "loss": 3.3969, "step": 22740 }, { "epoch": 6.00249856, "grad_norm": 1.0155985355377197, "learning_rate": 1.983992588553411e-05, "loss": 3.5892, "step": 22750 }, { "epoch": 6.00252416, "grad_norm": 0.8771054744720459, "learning_rate": 1.9839782345445952e-05, "loss": 3.6674, "step": 22760 }, { "epoch": 6.00254976, "grad_norm": 1.1221802234649658, "learning_rate": 1.983963874154942e-05, "loss": 3.5038, "step": 22770 }, { "epoch": 6.00257536, "grad_norm": 2.1012930870056152, "learning_rate": 1.983949507384545e-05, "loss": 3.261, "step": 22780 }, { "epoch": 6.00260096, "grad_norm": 0.9525576233863831, "learning_rate": 1.9839351342334963e-05, "loss": 3.5655, "step": 22790 }, { "epoch": 6.00262656, "grad_norm": 1.0402941703796387, "learning_rate": 1.9839207547018902e-05, "loss": 3.548, "step": 22800 }, { "epoch": 6.00265216, "grad_norm": 0.9922022819519043, "learning_rate": 1.9839063687898198e-05, "loss": 3.5283, "step": 22810 }, { "epoch": 6.00267776, "grad_norm": 1.0026695728302002, "learning_rate": 1.9838919764973777e-05, "loss": 3.6501, "step": 22820 }, { "epoch": 6.00270336, "grad_norm": 1.0699176788330078, "learning_rate": 1.9838775778246576e-05, "loss": 3.5248, "step": 22830 }, { "epoch": 6.00272896, "grad_norm": 1.0394203662872314, "learning_rate": 1.983863172771753e-05, "loss": 3.8857, "step": 22840 }, { "epoch": 6.00275456, "grad_norm": 1.0349936485290527, "learning_rate": 1.983848761338757e-05, "loss": 3.6136, "step": 22850 }, { "epoch": 6.00278016, "grad_norm": 0.9806698560714722, "learning_rate": 1.9838343435257635e-05, "loss": 3.4301, "step": 22860 }, { "epoch": 6.00280576, "grad_norm": 0.9272599220275879, "learning_rate": 1.9838199193328656e-05, "loss": 3.5503, "step": 22870 }, { "epoch": 6.00283136, "grad_norm": 0.9866198897361755, "learning_rate": 1.9838054887601573e-05, "loss": 3.4095, "step": 22880 }, { "epoch": 6.00285696, "grad_norm": 0.852547824382782, "learning_rate": 1.9837910518077313e-05, "loss": 3.2561, "step": 22890 }, { "epoch": 6.00288256, "grad_norm": 1.1096726655960083, "learning_rate": 1.9837766084756822e-05, "loss": 3.4568, "step": 22900 }, { "epoch": 6.00290816, "grad_norm": 0.9746274948120117, "learning_rate": 1.9837621587641035e-05, "loss": 3.4675, "step": 22910 }, { "epoch": 6.00293376, "grad_norm": 0.9981305599212646, "learning_rate": 1.983747702673088e-05, "loss": 3.7873, "step": 22920 }, { "epoch": 6.00295936, "grad_norm": 0.9358292818069458, "learning_rate": 1.9837332402027307e-05, "loss": 3.4569, "step": 22930 }, { "epoch": 6.00298496, "grad_norm": 0.9496201276779175, "learning_rate": 1.9837187713531242e-05, "loss": 3.4255, "step": 22940 }, { "epoch": 6.00301056, "grad_norm": 1.060788631439209, "learning_rate": 1.9837042961243632e-05, "loss": 3.3622, "step": 22950 }, { "epoch": 6.00303616, "grad_norm": 1.0433067083358765, "learning_rate": 1.9836898145165415e-05, "loss": 3.5343, "step": 22960 }, { "epoch": 6.00306176, "grad_norm": 1.020155906677246, "learning_rate": 1.9836753265297527e-05, "loss": 3.6799, "step": 22970 }, { "epoch": 6.00308736, "grad_norm": 0.9349377751350403, "learning_rate": 1.9836608321640908e-05, "loss": 3.5513, "step": 22980 }, { "epoch": 6.00311296, "grad_norm": 0.9619227647781372, "learning_rate": 1.9836463314196495e-05, "loss": 3.5533, "step": 22990 }, { "epoch": 6.00313856, "grad_norm": 1.0537526607513428, "learning_rate": 1.9836318242965236e-05, "loss": 3.7513, "step": 23000 }, { "epoch": 6.00316416, "grad_norm": 0.8952500820159912, "learning_rate": 1.9836173107948065e-05, "loss": 3.4665, "step": 23010 }, { "epoch": 6.00318976, "grad_norm": 0.8386407494544983, "learning_rate": 1.9836027909145926e-05, "loss": 3.5444, "step": 23020 }, { "epoch": 6.00321536, "grad_norm": 0.8942707777023315, "learning_rate": 1.9835882646559763e-05, "loss": 3.4818, "step": 23030 }, { "epoch": 6.00324096, "grad_norm": 0.8961383700370789, "learning_rate": 1.9835737320190507e-05, "loss": 3.5876, "step": 23040 }, { "epoch": 6.00326656, "grad_norm": 1.0092530250549316, "learning_rate": 1.9835591930039116e-05, "loss": 3.4236, "step": 23050 }, { "epoch": 6.00329216, "grad_norm": 0.9609701633453369, "learning_rate": 1.9835446476106522e-05, "loss": 3.527, "step": 23060 }, { "epoch": 6.00331776, "grad_norm": 0.9238085746765137, "learning_rate": 1.9835300958393673e-05, "loss": 3.5467, "step": 23070 }, { "epoch": 6.00334336, "grad_norm": 0.9872964024543762, "learning_rate": 1.983515537690151e-05, "loss": 3.3795, "step": 23080 }, { "epoch": 6.00336896, "grad_norm": 0.916348934173584, "learning_rate": 1.983500973163098e-05, "loss": 3.601, "step": 23090 }, { "epoch": 6.00339456, "grad_norm": 1.0374776124954224, "learning_rate": 1.9834864022583024e-05, "loss": 3.4702, "step": 23100 }, { "epoch": 6.00342016, "grad_norm": 0.8861818313598633, "learning_rate": 1.9834718249758592e-05, "loss": 3.3471, "step": 23110 }, { "epoch": 6.00344576, "grad_norm": 0.9821520447731018, "learning_rate": 1.983457241315862e-05, "loss": 3.5599, "step": 23120 }, { "epoch": 6.00347136, "grad_norm": 1.307286262512207, "learning_rate": 1.9834426512784065e-05, "loss": 3.4857, "step": 23130 }, { "epoch": 6.00349696, "grad_norm": 1.0179846286773682, "learning_rate": 1.9834280548635864e-05, "loss": 3.5713, "step": 23140 }, { "epoch": 6.00352256, "grad_norm": 1.008167028427124, "learning_rate": 1.9834134520714968e-05, "loss": 3.7842, "step": 23150 }, { "epoch": 6.00354816, "grad_norm": 0.8682231307029724, "learning_rate": 1.9833988429022327e-05, "loss": 3.3185, "step": 23160 }, { "epoch": 6.00357376, "grad_norm": 0.9627460241317749, "learning_rate": 1.983384227355888e-05, "loss": 3.4714, "step": 23170 }, { "epoch": 6.00359936, "grad_norm": 1.1030712127685547, "learning_rate": 1.983369605432558e-05, "loss": 3.6711, "step": 23180 }, { "epoch": 6.00362496, "grad_norm": 1.038503885269165, "learning_rate": 1.9833549771323377e-05, "loss": 3.4509, "step": 23190 }, { "epoch": 6.00365056, "grad_norm": 1.0481945276260376, "learning_rate": 1.983340342455322e-05, "loss": 3.6, "step": 23200 }, { "epoch": 6.00367616, "grad_norm": 1.385785460472107, "learning_rate": 1.983325701401605e-05, "loss": 3.5435, "step": 23210 }, { "epoch": 6.00370176, "grad_norm": 1.0501435995101929, "learning_rate": 1.983311053971282e-05, "loss": 3.4778, "step": 23220 }, { "epoch": 6.00372736, "grad_norm": 1.3748376369476318, "learning_rate": 1.9832964001644487e-05, "loss": 3.4729, "step": 23230 }, { "epoch": 6.00375296, "grad_norm": 0.8755690455436707, "learning_rate": 1.983281739981199e-05, "loss": 3.6534, "step": 23240 }, { "epoch": 6.00377856, "grad_norm": 1.0603936910629272, "learning_rate": 1.9832670734216292e-05, "loss": 3.7889, "step": 23250 }, { "epoch": 6.00380416, "grad_norm": 1.1505272388458252, "learning_rate": 1.983252400485833e-05, "loss": 3.4027, "step": 23260 }, { "epoch": 6.00382976, "grad_norm": 1.007007360458374, "learning_rate": 1.9832377211739067e-05, "loss": 3.5818, "step": 23270 }, { "epoch": 6.00385536, "grad_norm": 0.8197838068008423, "learning_rate": 1.983223035485945e-05, "loss": 3.443, "step": 23280 }, { "epoch": 6.00388096, "grad_norm": 0.9117670655250549, "learning_rate": 1.9832083434220435e-05, "loss": 3.3909, "step": 23290 }, { "epoch": 6.00390656, "grad_norm": 1.337895393371582, "learning_rate": 1.983193644982297e-05, "loss": 3.472, "step": 23300 }, { "epoch": 6.00393216, "grad_norm": 1.0130181312561035, "learning_rate": 1.983178940166801e-05, "loss": 3.5742, "step": 23310 }, { "epoch": 6.00395776, "grad_norm": 0.9220149517059326, "learning_rate": 1.983164228975651e-05, "loss": 3.5173, "step": 23320 }, { "epoch": 6.00398336, "grad_norm": 0.8606331944465637, "learning_rate": 1.983149511408942e-05, "loss": 3.5489, "step": 23330 }, { "epoch": 6.00400896, "grad_norm": 0.8912082314491272, "learning_rate": 1.9831347874667698e-05, "loss": 3.3864, "step": 23340 }, { "epoch": 6.00403456, "grad_norm": 1.1599122285842896, "learning_rate": 1.98312005714923e-05, "loss": 3.6392, "step": 23350 }, { "epoch": 6.00406016, "grad_norm": 0.9329705238342285, "learning_rate": 1.9831053204564178e-05, "loss": 3.4371, "step": 23360 }, { "epoch": 6.00408576, "grad_norm": 0.9292133450508118, "learning_rate": 1.9830905773884287e-05, "loss": 3.4574, "step": 23370 }, { "epoch": 6.00411136, "grad_norm": 0.8610599040985107, "learning_rate": 1.9830758279453586e-05, "loss": 3.6385, "step": 23380 }, { "epoch": 6.00413696, "grad_norm": 0.9829214811325073, "learning_rate": 1.983061072127303e-05, "loss": 3.4806, "step": 23390 }, { "epoch": 6.00416256, "grad_norm": 1.0598055124282837, "learning_rate": 1.9830463099343576e-05, "loss": 3.5012, "step": 23400 }, { "epoch": 6.00418816, "grad_norm": 1.0863460302352905, "learning_rate": 1.9830315413666185e-05, "loss": 3.3471, "step": 23410 }, { "epoch": 6.00421376, "grad_norm": 1.12995183467865, "learning_rate": 1.9830167664241806e-05, "loss": 3.3336, "step": 23420 }, { "epoch": 6.00423936, "grad_norm": 0.8941740393638611, "learning_rate": 1.9830019851071404e-05, "loss": 3.4883, "step": 23430 }, { "epoch": 6.00426496, "grad_norm": 0.9981285929679871, "learning_rate": 1.982987197415594e-05, "loss": 3.3753, "step": 23440 }, { "epoch": 6.00429056, "grad_norm": 0.9647899866104126, "learning_rate": 1.982972403349636e-05, "loss": 3.7113, "step": 23450 }, { "epoch": 6.00431616, "grad_norm": 0.8545606732368469, "learning_rate": 1.982957602909364e-05, "loss": 3.6558, "step": 23460 }, { "epoch": 6.00434176, "grad_norm": 1.0307762622833252, "learning_rate": 1.982942796094873e-05, "loss": 3.4375, "step": 23470 }, { "epoch": 6.00436736, "grad_norm": 0.912539005279541, "learning_rate": 1.982927982906259e-05, "loss": 3.6622, "step": 23480 }, { "epoch": 6.00439296, "grad_norm": 1.017470121383667, "learning_rate": 1.9829131633436182e-05, "loss": 3.4569, "step": 23490 }, { "epoch": 6.00441856, "grad_norm": 1.0100840330123901, "learning_rate": 1.982898337407047e-05, "loss": 3.5119, "step": 23500 }, { "epoch": 6.00444416, "grad_norm": 1.1770118474960327, "learning_rate": 1.982883505096641e-05, "loss": 3.2743, "step": 23510 }, { "epoch": 6.00446976, "grad_norm": 0.8671426773071289, "learning_rate": 1.9828686664124968e-05, "loss": 3.4079, "step": 23520 }, { "epoch": 6.00449536, "grad_norm": 0.9697526097297668, "learning_rate": 1.982853821354711e-05, "loss": 3.5196, "step": 23530 }, { "epoch": 6.00452096, "grad_norm": 1.1925065517425537, "learning_rate": 1.9828389699233787e-05, "loss": 3.5608, "step": 23540 }, { "epoch": 6.00454656, "grad_norm": 0.9183837175369263, "learning_rate": 1.9828241121185973e-05, "loss": 3.4197, "step": 23550 }, { "epoch": 6.00457216, "grad_norm": 1.0895119905471802, "learning_rate": 1.9828092479404625e-05, "loss": 3.5982, "step": 23560 }, { "epoch": 6.00459776, "grad_norm": 0.9300475716590881, "learning_rate": 1.9827943773890708e-05, "loss": 3.6446, "step": 23570 }, { "epoch": 6.00462336, "grad_norm": 1.2014268636703491, "learning_rate": 1.9827795004645193e-05, "loss": 3.4057, "step": 23580 }, { "epoch": 6.00464896, "grad_norm": 1.0856108665466309, "learning_rate": 1.9827646171669034e-05, "loss": 3.4661, "step": 23590 }, { "epoch": 6.00467456, "grad_norm": 1.050417184829712, "learning_rate": 1.9827497274963206e-05, "loss": 3.4953, "step": 23600 }, { "epoch": 6.00470016, "grad_norm": 0.9325752258300781, "learning_rate": 1.982734831452867e-05, "loss": 3.4623, "step": 23610 }, { "epoch": 6.00472576, "grad_norm": 0.9984118938446045, "learning_rate": 1.9827199290366387e-05, "loss": 3.4608, "step": 23620 }, { "epoch": 6.00475136, "grad_norm": 0.8927995562553406, "learning_rate": 1.9827050202477333e-05, "loss": 3.4991, "step": 23630 }, { "epoch": 6.00477696, "grad_norm": 1.2951949834823608, "learning_rate": 1.982690105086247e-05, "loss": 3.6314, "step": 23640 }, { "epoch": 6.00480256, "grad_norm": 1.1956866979599, "learning_rate": 1.9826751835522763e-05, "loss": 3.4992, "step": 23650 }, { "epoch": 6.00482816, "grad_norm": 0.9949907064437866, "learning_rate": 1.982660255645918e-05, "loss": 3.6847, "step": 23660 }, { "epoch": 6.00485376, "grad_norm": 1.1564884185791016, "learning_rate": 1.9826453213672693e-05, "loss": 3.4478, "step": 23670 }, { "epoch": 6.00487936, "grad_norm": 1.0236625671386719, "learning_rate": 1.982630380716427e-05, "loss": 3.4876, "step": 23680 }, { "epoch": 6.00490496, "grad_norm": 0.9994959235191345, "learning_rate": 1.9826154336934877e-05, "loss": 3.5645, "step": 23690 }, { "epoch": 6.00493056, "grad_norm": 1.5321308374404907, "learning_rate": 1.9826004802985486e-05, "loss": 3.6391, "step": 23700 }, { "epoch": 6.00495616, "grad_norm": 0.994819164276123, "learning_rate": 1.9825855205317062e-05, "loss": 3.6163, "step": 23710 }, { "epoch": 6.00498176, "grad_norm": 0.906197190284729, "learning_rate": 1.9825705543930582e-05, "loss": 3.6117, "step": 23720 }, { "epoch": 6.00500736, "grad_norm": 0.9072557687759399, "learning_rate": 1.982555581882701e-05, "loss": 3.4439, "step": 23730 }, { "epoch": 6.00503296, "grad_norm": 1.0816984176635742, "learning_rate": 1.9825406030007323e-05, "loss": 3.4508, "step": 23740 }, { "epoch": 6.00505856, "grad_norm": 0.9978890419006348, "learning_rate": 1.982525617747249e-05, "loss": 3.4751, "step": 23750 }, { "epoch": 6.00508416, "grad_norm": 0.9868170619010925, "learning_rate": 1.9825106261223478e-05, "loss": 3.2934, "step": 23760 }, { "epoch": 6.00510976, "grad_norm": 0.9681140184402466, "learning_rate": 1.9824956281261265e-05, "loss": 3.4192, "step": 23770 }, { "epoch": 6.00513536, "grad_norm": 1.0060933828353882, "learning_rate": 1.982480623758682e-05, "loss": 3.4793, "step": 23780 }, { "epoch": 6.00516096, "grad_norm": 1.0104504823684692, "learning_rate": 1.982465613020112e-05, "loss": 3.6299, "step": 23790 }, { "epoch": 6.00518656, "grad_norm": 1.0590468645095825, "learning_rate": 1.9824505959105137e-05, "loss": 3.6396, "step": 23800 }, { "epoch": 6.00521216, "grad_norm": 1.03486967086792, "learning_rate": 1.9824355724299838e-05, "loss": 3.5129, "step": 23810 }, { "epoch": 6.00523776, "grad_norm": 1.0196166038513184, "learning_rate": 1.9824205425786206e-05, "loss": 3.5717, "step": 23820 }, { "epoch": 6.00526336, "grad_norm": 1.206041693687439, "learning_rate": 1.9824055063565217e-05, "loss": 3.4718, "step": 23830 }, { "epoch": 6.00528896, "grad_norm": 0.907505989074707, "learning_rate": 1.9823904637637837e-05, "loss": 3.5376, "step": 23840 }, { "epoch": 6.00531456, "grad_norm": 1.0613900423049927, "learning_rate": 1.982375414800505e-05, "loss": 3.573, "step": 23850 }, { "epoch": 6.00534016, "grad_norm": 0.9013208746910095, "learning_rate": 1.9823603594667825e-05, "loss": 3.5396, "step": 23860 }, { "epoch": 6.00536576, "grad_norm": 1.0533733367919922, "learning_rate": 1.982345297762714e-05, "loss": 3.445, "step": 23870 }, { "epoch": 6.00539136, "grad_norm": 1.029103398323059, "learning_rate": 1.982330229688398e-05, "loss": 3.5442, "step": 23880 }, { "epoch": 6.00541696, "grad_norm": 1.1262567043304443, "learning_rate": 1.982315155243931e-05, "loss": 3.496, "step": 23890 }, { "epoch": 6.00544256, "grad_norm": 0.964357852935791, "learning_rate": 1.9823000744294115e-05, "loss": 3.539, "step": 23900 }, { "epoch": 6.00546816, "grad_norm": 0.8830447793006897, "learning_rate": 1.982284987244937e-05, "loss": 3.2737, "step": 23910 }, { "epoch": 6.00549376, "grad_norm": 0.8904331922531128, "learning_rate": 1.9822698936906054e-05, "loss": 3.5459, "step": 23920 }, { "epoch": 6.00551936, "grad_norm": 1.115660309791565, "learning_rate": 1.9822547937665148e-05, "loss": 3.4546, "step": 23930 }, { "epoch": 6.00554496, "grad_norm": 0.9992704391479492, "learning_rate": 1.9822396874727622e-05, "loss": 3.5013, "step": 23940 }, { "epoch": 6.00557056, "grad_norm": 1.1372772455215454, "learning_rate": 1.982224574809447e-05, "loss": 3.6928, "step": 23950 }, { "epoch": 6.00559616, "grad_norm": 0.9494563341140747, "learning_rate": 1.9822094557766664e-05, "loss": 3.5416, "step": 23960 }, { "epoch": 6.00562176, "grad_norm": 1.214876413345337, "learning_rate": 1.9821943303745183e-05, "loss": 3.4222, "step": 23970 }, { "epoch": 6.00564736, "grad_norm": 1.19051194190979, "learning_rate": 1.9821791986031012e-05, "loss": 3.5285, "step": 23980 }, { "epoch": 6.00567296, "grad_norm": 1.0091373920440674, "learning_rate": 1.9821640604625128e-05, "loss": 3.5462, "step": 23990 }, { "epoch": 6.00569856, "grad_norm": 0.9582409858703613, "learning_rate": 1.9821489159528515e-05, "loss": 3.5206, "step": 24000 }, { "epoch": 6.00572416, "grad_norm": 0.8960213661193848, "learning_rate": 1.982133765074216e-05, "loss": 3.2851, "step": 24010 }, { "epoch": 6.00574976, "grad_norm": 0.9400068521499634, "learning_rate": 1.9821186078267035e-05, "loss": 3.1747, "step": 24020 }, { "epoch": 6.00577536, "grad_norm": 0.9860766530036926, "learning_rate": 1.982103444210413e-05, "loss": 3.7714, "step": 24030 }, { "epoch": 6.00580096, "grad_norm": 1.2473639249801636, "learning_rate": 1.9820882742254426e-05, "loss": 3.5869, "step": 24040 }, { "epoch": 6.00582656, "grad_norm": 1.0060572624206543, "learning_rate": 1.982073097871891e-05, "loss": 3.5832, "step": 24050 }, { "epoch": 6.00585216, "grad_norm": 0.9854660034179688, "learning_rate": 1.9820579151498562e-05, "loss": 3.5719, "step": 24060 }, { "epoch": 6.00587776, "grad_norm": 0.8460894227027893, "learning_rate": 1.9820427260594367e-05, "loss": 3.2863, "step": 24070 }, { "epoch": 6.00590336, "grad_norm": 1.1373733282089233, "learning_rate": 1.982027530600731e-05, "loss": 3.5348, "step": 24080 }, { "epoch": 6.00592896, "grad_norm": 1.0302973985671997, "learning_rate": 1.9820123287738383e-05, "loss": 3.7544, "step": 24090 }, { "epoch": 6.00595456, "grad_norm": 0.9275426864624023, "learning_rate": 1.9819971205788563e-05, "loss": 3.8079, "step": 24100 }, { "epoch": 6.00598016, "grad_norm": 1.0467538833618164, "learning_rate": 1.981981906015884e-05, "loss": 3.5394, "step": 24110 }, { "epoch": 6.00600576, "grad_norm": 0.9515910148620605, "learning_rate": 1.98196668508502e-05, "loss": 3.5267, "step": 24120 }, { "epoch": 6.00603136, "grad_norm": 1.2527981996536255, "learning_rate": 1.981951457786363e-05, "loss": 3.7154, "step": 24130 }, { "epoch": 6.0060569600000004, "grad_norm": 0.9306503534317017, "learning_rate": 1.9819362241200118e-05, "loss": 3.503, "step": 24140 }, { "epoch": 6.00608256, "grad_norm": 0.9614654779434204, "learning_rate": 1.981920984086065e-05, "loss": 3.4916, "step": 24150 }, { "epoch": 6.00610816, "grad_norm": 0.9298986792564392, "learning_rate": 1.981905737684622e-05, "loss": 3.5252, "step": 24160 }, { "epoch": 6.00613376, "grad_norm": 1.087937355041504, "learning_rate": 1.981890484915781e-05, "loss": 3.5253, "step": 24170 }, { "epoch": 6.00615936, "grad_norm": 1.0597693920135498, "learning_rate": 1.9818752257796408e-05, "loss": 3.6056, "step": 24180 }, { "epoch": 6.00618496, "grad_norm": 0.9341207146644592, "learning_rate": 1.9818599602763012e-05, "loss": 3.5025, "step": 24190 }, { "epoch": 6.00621056, "grad_norm": 1.2063651084899902, "learning_rate": 1.9818446884058606e-05, "loss": 3.4435, "step": 24200 }, { "epoch": 6.00623616, "grad_norm": 1.0824795961380005, "learning_rate": 1.981829410168418e-05, "loss": 3.4425, "step": 24210 }, { "epoch": 6.00626176, "grad_norm": 1.5067508220672607, "learning_rate": 1.9818141255640728e-05, "loss": 3.462, "step": 24220 }, { "epoch": 6.00628736, "grad_norm": 0.9491271376609802, "learning_rate": 1.9817988345929238e-05, "loss": 3.5159, "step": 24230 }, { "epoch": 6.00631296, "grad_norm": 0.9698442220687866, "learning_rate": 1.9817835372550705e-05, "loss": 3.6525, "step": 24240 }, { "epoch": 6.00633856, "grad_norm": 1.170267939567566, "learning_rate": 1.9817682335506117e-05, "loss": 3.5565, "step": 24250 }, { "epoch": 6.00636416, "grad_norm": 0.9067039489746094, "learning_rate": 1.981752923479647e-05, "loss": 3.4864, "step": 24260 }, { "epoch": 6.00638976, "grad_norm": 0.8730793595314026, "learning_rate": 1.981737607042275e-05, "loss": 3.4445, "step": 24270 }, { "epoch": 6.00641536, "grad_norm": 0.9130116701126099, "learning_rate": 1.981722284238596e-05, "loss": 3.4283, "step": 24280 }, { "epoch": 6.00644096, "grad_norm": 0.935050904750824, "learning_rate": 1.9817069550687085e-05, "loss": 3.4028, "step": 24290 }, { "epoch": 6.00646656, "grad_norm": 0.9685462117195129, "learning_rate": 1.981691619532713e-05, "loss": 3.4829, "step": 24300 }, { "epoch": 6.00649216, "grad_norm": 0.9612316489219666, "learning_rate": 1.9816762776307078e-05, "loss": 3.4271, "step": 24310 }, { "epoch": 6.00651776, "grad_norm": 0.965351939201355, "learning_rate": 1.981660929362793e-05, "loss": 3.4389, "step": 24320 }, { "epoch": 6.00654336, "grad_norm": 0.8829487562179565, "learning_rate": 1.9816455747290677e-05, "loss": 3.4303, "step": 24330 }, { "epoch": 6.00656896, "grad_norm": 0.938879132270813, "learning_rate": 1.981630213729632e-05, "loss": 3.4139, "step": 24340 }, { "epoch": 6.00659456, "grad_norm": 1.002519130706787, "learning_rate": 1.981614846364585e-05, "loss": 3.5352, "step": 24350 }, { "epoch": 6.00662016, "grad_norm": 0.9031131863594055, "learning_rate": 1.9815994726340268e-05, "loss": 3.3286, "step": 24360 }, { "epoch": 6.00664576, "grad_norm": 1.3633449077606201, "learning_rate": 1.9815840925380568e-05, "loss": 3.6912, "step": 24370 }, { "epoch": 6.00667136, "grad_norm": 0.9892861843109131, "learning_rate": 1.9815687060767748e-05, "loss": 3.4566, "step": 24380 }, { "epoch": 6.00669696, "grad_norm": 1.0641204118728638, "learning_rate": 1.9815533132502808e-05, "loss": 3.2074, "step": 24390 }, { "epoch": 6.00672256, "grad_norm": 1.02101469039917, "learning_rate": 1.981537914058674e-05, "loss": 3.4499, "step": 24400 }, { "epoch": 2.56e-05, "grad_norm": 1.2738677263259888, "learning_rate": 1.9815225085020554e-05, "loss": 3.5784, "step": 24410 }, { "epoch": 5.12e-05, "grad_norm": 1.8850935697555542, "learning_rate": 1.9815070965805236e-05, "loss": 3.6025, "step": 24420 }, { "epoch": 7.68e-05, "grad_norm": 1.1868728399276733, "learning_rate": 1.9814916782941792e-05, "loss": 3.4122, "step": 24430 }, { "epoch": 0.0001024, "grad_norm": 1.159222960472107, "learning_rate": 1.9814762536431225e-05, "loss": 3.6018, "step": 24440 }, { "epoch": 0.000128, "grad_norm": 1.342866063117981, "learning_rate": 1.981460822627453e-05, "loss": 3.9024, "step": 24450 }, { "epoch": 0.0001536, "grad_norm": 1.0900126695632935, "learning_rate": 1.9814453852472708e-05, "loss": 3.6743, "step": 24460 }, { "epoch": 0.0001792, "grad_norm": 1.2144883871078491, "learning_rate": 1.981429941502676e-05, "loss": 3.5754, "step": 24470 }, { "epoch": 0.0002048, "grad_norm": 0.9003142714500427, "learning_rate": 1.981414491393769e-05, "loss": 3.7723, "step": 24480 }, { "epoch": 0.0002304, "grad_norm": 1.004772663116455, "learning_rate": 1.9813990349206496e-05, "loss": 3.5702, "step": 24490 }, { "epoch": 0.000256, "grad_norm": 1.245143175125122, "learning_rate": 1.9813835720834185e-05, "loss": 3.6693, "step": 24500 }, { "epoch": 0.0002816, "grad_norm": 1.0097745656967163, "learning_rate": 1.9813681028821757e-05, "loss": 3.9352, "step": 24510 }, { "epoch": 0.0003072, "grad_norm": 1.158562421798706, "learning_rate": 1.981352627317022e-05, "loss": 3.704, "step": 24520 }, { "epoch": 0.0003328, "grad_norm": 0.9780899286270142, "learning_rate": 1.9813371453880568e-05, "loss": 3.4638, "step": 24530 }, { "epoch": 0.0003584, "grad_norm": 1.1465950012207031, "learning_rate": 1.981321657095381e-05, "loss": 3.5118, "step": 24540 }, { "epoch": 0.000384, "grad_norm": 0.8826435804367065, "learning_rate": 1.9813061624390952e-05, "loss": 3.4416, "step": 24550 }, { "epoch": 0.0004096, "grad_norm": 1.2140052318572998, "learning_rate": 1.9812906614192997e-05, "loss": 3.853, "step": 24560 }, { "epoch": 0.0004352, "grad_norm": 0.9177186489105225, "learning_rate": 1.9812751540360952e-05, "loss": 3.5534, "step": 24570 }, { "epoch": 0.0004608, "grad_norm": 0.9738857746124268, "learning_rate": 1.9812596402895822e-05, "loss": 3.6998, "step": 24580 }, { "epoch": 0.0004864, "grad_norm": 1.2239844799041748, "learning_rate": 1.981244120179861e-05, "loss": 3.4876, "step": 24590 }, { "epoch": 0.000512, "grad_norm": 1.0885549783706665, "learning_rate": 1.9812285937070323e-05, "loss": 3.4188, "step": 24600 }, { "epoch": 0.0005376, "grad_norm": 0.8454264402389526, "learning_rate": 1.9812130608711972e-05, "loss": 3.5678, "step": 24610 }, { "epoch": 0.0005632, "grad_norm": 2.211449146270752, "learning_rate": 1.981197521672456e-05, "loss": 3.6416, "step": 24620 }, { "epoch": 0.0005888, "grad_norm": 2.06425404548645, "learning_rate": 1.98118197611091e-05, "loss": 4.0021, "step": 24630 }, { "epoch": 0.0006144, "grad_norm": 0.9263545274734497, "learning_rate": 1.9811664241866592e-05, "loss": 3.6637, "step": 24640 }, { "epoch": 0.00064, "grad_norm": 0.9325092434883118, "learning_rate": 1.9811508658998054e-05, "loss": 3.5531, "step": 24650 }, { "epoch": 0.0006656, "grad_norm": 0.9248952269554138, "learning_rate": 1.981135301250449e-05, "loss": 3.7056, "step": 24660 }, { "epoch": 0.0006912, "grad_norm": 1.0091015100479126, "learning_rate": 1.9811197302386903e-05, "loss": 3.6303, "step": 24670 }, { "epoch": 0.0007168, "grad_norm": 1.4298962354660034, "learning_rate": 1.9811041528646313e-05, "loss": 3.4666, "step": 24680 }, { "epoch": 0.0007424, "grad_norm": 1.000925898551941, "learning_rate": 1.9810885691283727e-05, "loss": 3.5413, "step": 24690 }, { "epoch": 0.000768, "grad_norm": 1.0015076398849487, "learning_rate": 1.981072979030015e-05, "loss": 3.3224, "step": 24700 }, { "epoch": 0.0007936, "grad_norm": 0.8938690423965454, "learning_rate": 1.9810573825696603e-05, "loss": 3.6006, "step": 24710 }, { "epoch": 0.0008192, "grad_norm": 1.96420156955719, "learning_rate": 1.9810417797474092e-05, "loss": 3.5694, "step": 24720 }, { "epoch": 0.0008448, "grad_norm": 0.9304109811782837, "learning_rate": 1.9810261705633627e-05, "loss": 3.414, "step": 24730 }, { "epoch": 0.0008704, "grad_norm": 1.264208197593689, "learning_rate": 1.9810105550176223e-05, "loss": 3.3435, "step": 24740 }, { "epoch": 0.000896, "grad_norm": 1.0087867975234985, "learning_rate": 1.9809949331102888e-05, "loss": 3.5156, "step": 24750 }, { "epoch": 0.0009216, "grad_norm": 1.1607409715652466, "learning_rate": 1.9809793048414644e-05, "loss": 3.7517, "step": 24760 }, { "epoch": 0.0009472, "grad_norm": 3.6315505504608154, "learning_rate": 1.9809636702112494e-05, "loss": 3.6711, "step": 24770 }, { "epoch": 0.0009728, "grad_norm": 0.8663115501403809, "learning_rate": 1.9809480292197462e-05, "loss": 4.1901, "step": 24780 }, { "epoch": 0.0009984, "grad_norm": 1.3552931547164917, "learning_rate": 1.9809323818670554e-05, "loss": 3.9026, "step": 24790 }, { "epoch": 0.001024, "grad_norm": 1.0767821073532104, "learning_rate": 1.980916728153279e-05, "loss": 3.4156, "step": 24800 }, { "epoch": 0.0010496, "grad_norm": 1.4963289499282837, "learning_rate": 1.9809010680785184e-05, "loss": 3.5589, "step": 24810 }, { "epoch": 0.0010752, "grad_norm": 1.0184794664382935, "learning_rate": 1.9808854016428746e-05, "loss": 3.6753, "step": 24820 }, { "epoch": 0.0011008, "grad_norm": 1.375101089477539, "learning_rate": 1.98086972884645e-05, "loss": 3.7265, "step": 24830 }, { "epoch": 0.0011264, "grad_norm": 0.9188999533653259, "learning_rate": 1.9808540496893462e-05, "loss": 3.6089, "step": 24840 }, { "epoch": 0.001152, "grad_norm": 0.8654535412788391, "learning_rate": 1.980838364171664e-05, "loss": 3.3276, "step": 24850 }, { "epoch": 0.0011776, "grad_norm": 1.8707001209259033, "learning_rate": 1.9808226722935057e-05, "loss": 3.6422, "step": 24860 }, { "epoch": 0.0012032, "grad_norm": 1.1683796644210815, "learning_rate": 1.9808069740549735e-05, "loss": 3.6525, "step": 24870 }, { "epoch": 0.0012288, "grad_norm": 0.9464643597602844, "learning_rate": 1.9807912694561687e-05, "loss": 3.8395, "step": 24880 }, { "epoch": 0.0012544, "grad_norm": 1.1922777891159058, "learning_rate": 1.9807755584971925e-05, "loss": 3.4067, "step": 24890 }, { "epoch": 0.00128, "grad_norm": 0.9105675220489502, "learning_rate": 1.980759841178148e-05, "loss": 4.1032, "step": 24900 }, { "epoch": 0.0013056, "grad_norm": 0.796916127204895, "learning_rate": 1.9807441174991367e-05, "loss": 3.6043, "step": 24910 }, { "epoch": 0.0013312, "grad_norm": 1.3942128419876099, "learning_rate": 1.9807283874602602e-05, "loss": 3.5728, "step": 24920 }, { "epoch": 0.0013568, "grad_norm": 1.1318241357803345, "learning_rate": 1.9807126510616207e-05, "loss": 3.3875, "step": 24930 }, { "epoch": 0.0013824, "grad_norm": 1.223592758178711, "learning_rate": 1.9806969083033207e-05, "loss": 3.5885, "step": 24940 }, { "epoch": 0.001408, "grad_norm": 0.9569821357727051, "learning_rate": 1.9806811591854614e-05, "loss": 3.7073, "step": 24950 }, { "epoch": 0.0014336, "grad_norm": 0.8020431399345398, "learning_rate": 1.9806654037081458e-05, "loss": 3.6754, "step": 24960 }, { "epoch": 0.0014592, "grad_norm": 1.1240068674087524, "learning_rate": 1.9806496418714752e-05, "loss": 3.7985, "step": 24970 }, { "epoch": 0.0014848, "grad_norm": 1.7544887065887451, "learning_rate": 1.980633873675553e-05, "loss": 3.92, "step": 24980 }, { "epoch": 0.0015104, "grad_norm": 1.3376867771148682, "learning_rate": 1.98061809912048e-05, "loss": 3.7557, "step": 24990 }, { "epoch": 0.001536, "grad_norm": 2.4072680473327637, "learning_rate": 1.98060231820636e-05, "loss": 3.8271, "step": 25000 }, { "epoch": 0.0015616, "grad_norm": 0.9041154384613037, "learning_rate": 1.980586530933294e-05, "loss": 3.5308, "step": 25010 }, { "epoch": 0.0015872, "grad_norm": 1.2175354957580566, "learning_rate": 1.980570737301385e-05, "loss": 3.6725, "step": 25020 }, { "epoch": 0.0016128, "grad_norm": 1.013716459274292, "learning_rate": 1.9805549373107357e-05, "loss": 3.8107, "step": 25030 }, { "epoch": 0.0016384, "grad_norm": 1.0328656435012817, "learning_rate": 1.980539130961448e-05, "loss": 3.9331, "step": 25040 }, { "epoch": 0.001664, "grad_norm": 0.8555101156234741, "learning_rate": 1.980523318253625e-05, "loss": 3.7602, "step": 25050 }, { "epoch": 0.0016896, "grad_norm": 0.79630047082901, "learning_rate": 1.9805074991873684e-05, "loss": 3.4999, "step": 25060 }, { "epoch": 0.0017152, "grad_norm": 1.5769304037094116, "learning_rate": 1.9804916737627815e-05, "loss": 3.569, "step": 25070 }, { "epoch": 0.0017408, "grad_norm": 0.7940043807029724, "learning_rate": 1.9804758419799665e-05, "loss": 3.6854, "step": 25080 }, { "epoch": 0.0017664, "grad_norm": 0.9728817939758301, "learning_rate": 1.9804600038390264e-05, "loss": 3.7268, "step": 25090 }, { "epoch": 0.001792, "grad_norm": 0.9858490228652954, "learning_rate": 1.9804441593400636e-05, "loss": 3.7682, "step": 25100 }, { "epoch": 0.0018176, "grad_norm": 1.006318211555481, "learning_rate": 1.980428308483181e-05, "loss": 3.7478, "step": 25110 }, { "epoch": 0.0018432, "grad_norm": 0.9697992205619812, "learning_rate": 1.9804124512684816e-05, "loss": 3.8609, "step": 25120 }, { "epoch": 0.0018688, "grad_norm": 0.8455150723457336, "learning_rate": 1.9803965876960678e-05, "loss": 3.6338, "step": 25130 }, { "epoch": 0.0018944, "grad_norm": 1.384097933769226, "learning_rate": 1.980380717766043e-05, "loss": 3.5928, "step": 25140 }, { "epoch": 0.00192, "grad_norm": 0.9349937438964844, "learning_rate": 1.9803648414785095e-05, "loss": 3.9065, "step": 25150 }, { "epoch": 0.0019456, "grad_norm": 0.9823383688926697, "learning_rate": 1.9803489588335706e-05, "loss": 3.8488, "step": 25160 }, { "epoch": 0.0019712, "grad_norm": 1.7377262115478516, "learning_rate": 1.9803330698313294e-05, "loss": 3.6394, "step": 25170 }, { "epoch": 0.0019968, "grad_norm": 0.8433765172958374, "learning_rate": 1.9803171744718887e-05, "loss": 3.7036, "step": 25180 }, { "epoch": 0.0020224, "grad_norm": 0.8731498122215271, "learning_rate": 1.980301272755351e-05, "loss": 3.7032, "step": 25190 }, { "epoch": 0.002048, "grad_norm": 1.369222640991211, "learning_rate": 1.980285364681821e-05, "loss": 3.9635, "step": 25200 }, { "epoch": 0.0020736, "grad_norm": 1.1231915950775146, "learning_rate": 1.9802694502514005e-05, "loss": 3.7701, "step": 25210 }, { "epoch": 0.0020992, "grad_norm": 1.1349964141845703, "learning_rate": 1.9802535294641933e-05, "loss": 3.7863, "step": 25220 }, { "epoch": 0.0021248, "grad_norm": 1.0243291854858398, "learning_rate": 1.9802376023203024e-05, "loss": 3.8791, "step": 25230 }, { "epoch": 0.0021504, "grad_norm": 1.0406817197799683, "learning_rate": 1.9802216688198314e-05, "loss": 3.702, "step": 25240 }, { "epoch": 0.002176, "grad_norm": 0.9510494470596313, "learning_rate": 1.9802057289628832e-05, "loss": 3.82, "step": 25250 }, { "epoch": 0.0022016, "grad_norm": 0.96031653881073, "learning_rate": 1.9801897827495613e-05, "loss": 3.7431, "step": 25260 }, { "epoch": 0.0022272, "grad_norm": 0.9077900052070618, "learning_rate": 1.980173830179969e-05, "loss": 3.9651, "step": 25270 }, { "epoch": 0.0022528, "grad_norm": 1.3593766689300537, "learning_rate": 1.9801578712542103e-05, "loss": 3.5968, "step": 25280 }, { "epoch": 0.0022784, "grad_norm": 1.4116233587265015, "learning_rate": 1.980141905972388e-05, "loss": 3.7138, "step": 25290 }, { "epoch": 0.002304, "grad_norm": 0.9679179787635803, "learning_rate": 1.9801259343346063e-05, "loss": 3.7475, "step": 25300 }, { "epoch": 0.0023296, "grad_norm": 0.721107006072998, "learning_rate": 1.980109956340968e-05, "loss": 3.5418, "step": 25310 }, { "epoch": 0.0023552, "grad_norm": 0.9098236560821533, "learning_rate": 1.9800939719915774e-05, "loss": 3.5953, "step": 25320 }, { "epoch": 0.0023808, "grad_norm": 0.8582794666290283, "learning_rate": 1.9800779812865377e-05, "loss": 3.6546, "step": 25330 }, { "epoch": 0.0024064, "grad_norm": 2.1311559677124023, "learning_rate": 1.980061984225953e-05, "loss": 3.9173, "step": 25340 }, { "epoch": 0.002432, "grad_norm": 0.9317250847816467, "learning_rate": 1.9800459808099267e-05, "loss": 3.7164, "step": 25350 }, { "epoch": 0.0024576, "grad_norm": 0.9286590218544006, "learning_rate": 1.9800299710385623e-05, "loss": 3.6407, "step": 25360 }, { "epoch": 0.0024832, "grad_norm": 0.9031126499176025, "learning_rate": 1.9800139549119642e-05, "loss": 3.6527, "step": 25370 }, { "epoch": 0.0025088, "grad_norm": 1.5863643884658813, "learning_rate": 1.9799979324302362e-05, "loss": 3.7402, "step": 25380 }, { "epoch": 0.0025344, "grad_norm": 0.8928690552711487, "learning_rate": 1.9799819035934822e-05, "loss": 3.8229, "step": 25390 }, { "epoch": 0.00256, "grad_norm": 1.1865850687026978, "learning_rate": 1.9799658684018056e-05, "loss": 3.6005, "step": 25400 }, { "epoch": 0.0025856, "grad_norm": 1.0609142780303955, "learning_rate": 1.979949826855311e-05, "loss": 3.7018, "step": 25410 }, { "epoch": 0.0026112, "grad_norm": 1.0902798175811768, "learning_rate": 1.9799337789541023e-05, "loss": 3.7346, "step": 25420 }, { "epoch": 0.0026368, "grad_norm": 0.8561603426933289, "learning_rate": 1.9799177246982836e-05, "loss": 3.9069, "step": 25430 }, { "epoch": 0.0026624, "grad_norm": 1.2329574823379517, "learning_rate": 1.9799016640879583e-05, "loss": 3.6535, "step": 25440 }, { "epoch": 0.002688, "grad_norm": 1.1639752388000488, "learning_rate": 1.979885597123232e-05, "loss": 3.6655, "step": 25450 }, { "epoch": 0.0027136, "grad_norm": 0.8986809253692627, "learning_rate": 1.9798695238042073e-05, "loss": 3.5546, "step": 25460 }, { "epoch": 0.0027392, "grad_norm": 1.3986883163452148, "learning_rate": 1.9798534441309895e-05, "loss": 3.6141, "step": 25470 }, { "epoch": 0.0027648, "grad_norm": 1.0209676027297974, "learning_rate": 1.9798373581036825e-05, "loss": 3.5706, "step": 25480 }, { "epoch": 0.0027904, "grad_norm": 1.019755482673645, "learning_rate": 1.9798212657223905e-05, "loss": 3.5359, "step": 25490 }, { "epoch": 0.002816, "grad_norm": 1.0034728050231934, "learning_rate": 1.979805166987218e-05, "loss": 3.6774, "step": 25500 }, { "epoch": 0.0028416, "grad_norm": 0.8753938674926758, "learning_rate": 1.9797890618982697e-05, "loss": 3.6605, "step": 25510 }, { "epoch": 0.0028672, "grad_norm": 1.0857714414596558, "learning_rate": 1.9797729504556494e-05, "loss": 3.5902, "step": 25520 }, { "epoch": 0.0028928, "grad_norm": 1.1613565683364868, "learning_rate": 1.979756832659462e-05, "loss": 3.6009, "step": 25530 }, { "epoch": 0.0029184, "grad_norm": 1.5539449453353882, "learning_rate": 1.979740708509812e-05, "loss": 3.7765, "step": 25540 }, { "epoch": 0.002944, "grad_norm": 2.715522050857544, "learning_rate": 1.9797245780068034e-05, "loss": 3.6775, "step": 25550 }, { "epoch": 0.0029696, "grad_norm": 1.193668007850647, "learning_rate": 1.979708441150542e-05, "loss": 3.6465, "step": 25560 }, { "epoch": 0.0029952, "grad_norm": 2.5698981285095215, "learning_rate": 1.9796922979411316e-05, "loss": 3.8029, "step": 25570 }, { "epoch": 0.0030208, "grad_norm": 0.8829193115234375, "learning_rate": 1.9796761483786767e-05, "loss": 3.9057, "step": 25580 }, { "epoch": 0.0030464, "grad_norm": 0.7988272905349731, "learning_rate": 1.9796599924632824e-05, "loss": 3.6548, "step": 25590 }, { "epoch": 0.003072, "grad_norm": 0.7866848707199097, "learning_rate": 1.9796438301950535e-05, "loss": 3.7232, "step": 25600 }, { "epoch": 0.0030976, "grad_norm": 0.7985216379165649, "learning_rate": 1.9796276615740948e-05, "loss": 3.7854, "step": 25610 }, { "epoch": 0.0031232, "grad_norm": 1.584104061126709, "learning_rate": 1.979611486600511e-05, "loss": 3.5582, "step": 25620 }, { "epoch": 0.0031488, "grad_norm": 0.8780615925788879, "learning_rate": 1.979595305274407e-05, "loss": 3.9612, "step": 25630 }, { "epoch": 0.0031744, "grad_norm": 0.7765239477157593, "learning_rate": 1.979579117595888e-05, "loss": 3.7414, "step": 25640 }, { "epoch": 0.0032, "grad_norm": 1.0670642852783203, "learning_rate": 1.9795629235650587e-05, "loss": 3.7945, "step": 25650 }, { "epoch": 0.0032256, "grad_norm": 1.1348719596862793, "learning_rate": 1.979546723182024e-05, "loss": 3.5782, "step": 25660 }, { "epoch": 0.0032512, "grad_norm": 1.129472255706787, "learning_rate": 1.9795305164468893e-05, "loss": 3.7591, "step": 25670 }, { "epoch": 0.0032768, "grad_norm": 0.9725901484489441, "learning_rate": 1.9795143033597593e-05, "loss": 3.7781, "step": 25680 }, { "epoch": 0.0033024, "grad_norm": 1.1546356678009033, "learning_rate": 1.9794980839207398e-05, "loss": 3.7119, "step": 25690 }, { "epoch": 0.003328, "grad_norm": 0.8546295762062073, "learning_rate": 1.979481858129935e-05, "loss": 3.6076, "step": 25700 }, { "epoch": 0.0033536, "grad_norm": 1.0240569114685059, "learning_rate": 1.979465625987451e-05, "loss": 3.7958, "step": 25710 }, { "epoch": 0.0033792, "grad_norm": 1.258156657218933, "learning_rate": 1.9794493874933927e-05, "loss": 3.4693, "step": 25720 }, { "epoch": 0.0034048, "grad_norm": 1.1522893905639648, "learning_rate": 1.9794331426478656e-05, "loss": 3.867, "step": 25730 }, { "epoch": 0.0034304, "grad_norm": 0.9446315169334412, "learning_rate": 1.9794168914509743e-05, "loss": 3.9761, "step": 25740 }, { "epoch": 0.003456, "grad_norm": 0.9893064498901367, "learning_rate": 1.979400633902825e-05, "loss": 3.3986, "step": 25750 }, { "epoch": 0.0034816, "grad_norm": 1.2002201080322266, "learning_rate": 1.9793843700035232e-05, "loss": 3.6581, "step": 25760 }, { "epoch": 0.0035072, "grad_norm": 1.349570631980896, "learning_rate": 1.979368099753174e-05, "loss": 3.5474, "step": 25770 }, { "epoch": 0.0035328, "grad_norm": 0.7609353065490723, "learning_rate": 1.9793518231518826e-05, "loss": 3.5682, "step": 25780 }, { "epoch": 0.0035584, "grad_norm": 1.1654949188232422, "learning_rate": 1.979335540199755e-05, "loss": 3.5098, "step": 25790 }, { "epoch": 0.003584, "grad_norm": 1.1993207931518555, "learning_rate": 1.979319250896897e-05, "loss": 3.7542, "step": 25800 }, { "epoch": 0.0036096, "grad_norm": 0.8157981634140015, "learning_rate": 1.9793029552434137e-05, "loss": 3.6405, "step": 25810 }, { "epoch": 0.0036352, "grad_norm": 0.9842674732208252, "learning_rate": 1.979286653239411e-05, "loss": 3.697, "step": 25820 }, { "epoch": 0.0036608, "grad_norm": 1.7082393169403076, "learning_rate": 1.9792703448849944e-05, "loss": 3.7175, "step": 25830 }, { "epoch": 0.0036864, "grad_norm": 1.2766109704971313, "learning_rate": 1.9792540301802705e-05, "loss": 3.7986, "step": 25840 }, { "epoch": 0.003712, "grad_norm": 1.988588809967041, "learning_rate": 1.9792377091253442e-05, "loss": 3.8419, "step": 25850 }, { "epoch": 0.0037376, "grad_norm": 0.9006958603858948, "learning_rate": 1.9792213817203214e-05, "loss": 3.5036, "step": 25860 }, { "epoch": 0.0037632, "grad_norm": 1.1726326942443848, "learning_rate": 1.979205047965308e-05, "loss": 3.6974, "step": 25870 }, { "epoch": 0.0037888, "grad_norm": 1.3375605344772339, "learning_rate": 1.9791887078604106e-05, "loss": 3.6819, "step": 25880 }, { "epoch": 0.0038144, "grad_norm": 1.5899752378463745, "learning_rate": 1.9791723614057347e-05, "loss": 3.7593, "step": 25890 }, { "epoch": 0.00384, "grad_norm": 0.8649042248725891, "learning_rate": 1.979156008601386e-05, "loss": 3.6062, "step": 25900 }, { "epoch": 0.0038656, "grad_norm": 0.8285888433456421, "learning_rate": 1.9791396494474705e-05, "loss": 3.717, "step": 25910 }, { "epoch": 0.0038912, "grad_norm": 1.077317237854004, "learning_rate": 1.979123283944095e-05, "loss": 3.4766, "step": 25920 }, { "epoch": 0.0039168, "grad_norm": 1.0425300598144531, "learning_rate": 1.9791069120913655e-05, "loss": 3.7341, "step": 25930 }, { "epoch": 0.0039424, "grad_norm": 1.272279143333435, "learning_rate": 1.979090533889387e-05, "loss": 3.5814, "step": 25940 }, { "epoch": 0.003968, "grad_norm": 0.9731799960136414, "learning_rate": 1.9790741493382674e-05, "loss": 3.7637, "step": 25950 }, { "epoch": 0.0039936, "grad_norm": 1.4509689807891846, "learning_rate": 1.9790577584381122e-05, "loss": 3.6307, "step": 25960 }, { "epoch": 0.0040192, "grad_norm": 0.8525829911231995, "learning_rate": 1.9790413611890274e-05, "loss": 3.6731, "step": 25970 }, { "epoch": 0.0040448, "grad_norm": 1.1361207962036133, "learning_rate": 1.9790249575911192e-05, "loss": 3.6634, "step": 25980 }, { "epoch": 0.0040704, "grad_norm": 1.0060869455337524, "learning_rate": 1.9790085476444948e-05, "loss": 3.7064, "step": 25990 }, { "epoch": 0.004096, "grad_norm": 1.1032525300979614, "learning_rate": 1.97899213134926e-05, "loss": 3.6441, "step": 26000 }, { "epoch": 0.0041216, "grad_norm": 1.4533594846725464, "learning_rate": 1.9789757087055214e-05, "loss": 3.694, "step": 26010 }, { "epoch": 0.0041472, "grad_norm": 0.9619110822677612, "learning_rate": 1.9789592797133856e-05, "loss": 3.7883, "step": 26020 }, { "epoch": 0.0041728, "grad_norm": 0.803139865398407, "learning_rate": 1.978942844372959e-05, "loss": 3.7812, "step": 26030 }, { "epoch": 0.0041984, "grad_norm": 0.8162810802459717, "learning_rate": 1.9789264026843482e-05, "loss": 3.7197, "step": 26040 }, { "epoch": 0.004224, "grad_norm": 0.8795602321624756, "learning_rate": 1.97890995464766e-05, "loss": 3.9145, "step": 26050 }, { "epoch": 0.0042496, "grad_norm": 0.7452648282051086, "learning_rate": 1.978893500263001e-05, "loss": 3.7914, "step": 26060 }, { "epoch": 0.0042752, "grad_norm": 0.8578802943229675, "learning_rate": 1.978877039530477e-05, "loss": 3.8694, "step": 26070 }, { "epoch": 0.0043008, "grad_norm": 0.9651628136634827, "learning_rate": 1.9788605724501963e-05, "loss": 3.4954, "step": 26080 }, { "epoch": 0.0043264, "grad_norm": 1.7823125123977661, "learning_rate": 1.978844099022265e-05, "loss": 3.6024, "step": 26090 }, { "epoch": 0.004352, "grad_norm": 0.9781813621520996, "learning_rate": 1.9788276192467894e-05, "loss": 3.5954, "step": 26100 }, { "epoch": 0.0043776, "grad_norm": 0.9655224680900574, "learning_rate": 1.9788111331238774e-05, "loss": 3.5433, "step": 26110 }, { "epoch": 0.0044032, "grad_norm": 0.9248391389846802, "learning_rate": 1.9787946406536346e-05, "loss": 3.7019, "step": 26120 }, { "epoch": 0.0044288, "grad_norm": 1.0506577491760254, "learning_rate": 1.9787781418361694e-05, "loss": 3.9033, "step": 26130 }, { "epoch": 0.0044544, "grad_norm": 1.3195048570632935, "learning_rate": 1.9787616366715874e-05, "loss": 3.9343, "step": 26140 }, { "epoch": 0.00448, "grad_norm": 0.891253650188446, "learning_rate": 1.978745125159997e-05, "loss": 3.7084, "step": 26150 }, { "epoch": 0.0045056, "grad_norm": 0.9116626381874084, "learning_rate": 1.978728607301504e-05, "loss": 3.7852, "step": 26160 }, { "epoch": 0.0045312, "grad_norm": 1.2692384719848633, "learning_rate": 1.9787120830962164e-05, "loss": 3.7287, "step": 26170 }, { "epoch": 0.0045568, "grad_norm": 2.0702948570251465, "learning_rate": 1.978695552544241e-05, "loss": 3.8377, "step": 26180 }, { "epoch": 0.0045824, "grad_norm": 0.8058087825775146, "learning_rate": 1.978679015645685e-05, "loss": 3.6619, "step": 26190 }, { "epoch": 0.004608, "grad_norm": 1.300865888595581, "learning_rate": 1.978662472400656e-05, "loss": 3.5977, "step": 26200 }, { "epoch": 0.0046336, "grad_norm": 0.9231269955635071, "learning_rate": 1.9786459228092605e-05, "loss": 3.7864, "step": 26210 }, { "epoch": 0.0046592, "grad_norm": 1.1034339666366577, "learning_rate": 1.9786293668716064e-05, "loss": 3.8332, "step": 26220 }, { "epoch": 0.0046848, "grad_norm": 1.5580600500106812, "learning_rate": 1.978612804587801e-05, "loss": 3.8845, "step": 26230 }, { "epoch": 0.0047104, "grad_norm": 0.8672987818717957, "learning_rate": 1.978596235957952e-05, "loss": 3.7475, "step": 26240 }, { "epoch": 0.004736, "grad_norm": 1.207991123199463, "learning_rate": 1.978579660982166e-05, "loss": 3.7808, "step": 26250 }, { "epoch": 0.0047616, "grad_norm": 1.0992119312286377, "learning_rate": 1.978563079660551e-05, "loss": 3.7042, "step": 26260 }, { "epoch": 0.0047872, "grad_norm": 0.8619397878646851, "learning_rate": 1.978546491993215e-05, "loss": 3.8804, "step": 26270 }, { "epoch": 0.0048128, "grad_norm": 0.7159697413444519, "learning_rate": 1.9785298979802644e-05, "loss": 3.553, "step": 26280 }, { "epoch": 0.0048384, "grad_norm": 1.3232108354568481, "learning_rate": 1.9785132976218078e-05, "loss": 3.7245, "step": 26290 }, { "epoch": 0.004864, "grad_norm": 0.9580478668212891, "learning_rate": 1.9784966909179526e-05, "loss": 3.8758, "step": 26300 }, { "epoch": 0.0048896, "grad_norm": 1.34110689163208, "learning_rate": 1.9784800778688065e-05, "loss": 3.8929, "step": 26310 }, { "epoch": 0.0049152, "grad_norm": 1.3980754613876343, "learning_rate": 1.978463458474477e-05, "loss": 3.6157, "step": 26320 }, { "epoch": 0.0049408, "grad_norm": 0.874900221824646, "learning_rate": 1.9784468327350715e-05, "loss": 4.3142, "step": 26330 }, { "epoch": 0.0049664, "grad_norm": 0.9306855201721191, "learning_rate": 1.9784302006506994e-05, "loss": 3.6611, "step": 26340 }, { "epoch": 0.004992, "grad_norm": 1.2446106672286987, "learning_rate": 1.9784135622214667e-05, "loss": 3.6985, "step": 26350 }, { "epoch": 0.0050176, "grad_norm": 2.607722520828247, "learning_rate": 1.9783969174474822e-05, "loss": 4.0993, "step": 26360 }, { "epoch": 0.0050432, "grad_norm": 1.0903834104537964, "learning_rate": 1.9783802663288537e-05, "loss": 3.7854, "step": 26370 }, { "epoch": 0.0050688, "grad_norm": 0.903714656829834, "learning_rate": 1.9783636088656895e-05, "loss": 3.5671, "step": 26380 }, { "epoch": 0.0050944, "grad_norm": 1.4652470350265503, "learning_rate": 1.978346945058097e-05, "loss": 3.626, "step": 26390 }, { "epoch": 0.00512, "grad_norm": 0.9851975440979004, "learning_rate": 1.9783302749061847e-05, "loss": 3.9561, "step": 26400 }, { "epoch": 0.0051456, "grad_norm": 0.9192158579826355, "learning_rate": 1.9783135984100604e-05, "loss": 3.6779, "step": 26410 }, { "epoch": 0.0051712, "grad_norm": 1.2056617736816406, "learning_rate": 1.9782969155698325e-05, "loss": 3.8174, "step": 26420 }, { "epoch": 0.0051968, "grad_norm": 1.0082956552505493, "learning_rate": 1.9782802263856094e-05, "loss": 3.6209, "step": 26430 }, { "epoch": 0.0052224, "grad_norm": 1.022995948791504, "learning_rate": 1.9782635308574985e-05, "loss": 3.6988, "step": 26440 }, { "epoch": 0.005248, "grad_norm": 0.9703688621520996, "learning_rate": 1.9782468289856092e-05, "loss": 3.6058, "step": 26450 }, { "epoch": 0.0052736, "grad_norm": 1.2067567110061646, "learning_rate": 1.9782301207700486e-05, "loss": 3.5957, "step": 26460 }, { "epoch": 0.0052992, "grad_norm": 0.8838315010070801, "learning_rate": 1.978213406210926e-05, "loss": 3.5142, "step": 26470 }, { "epoch": 0.0053248, "grad_norm": 0.8528091907501221, "learning_rate": 1.9781966853083495e-05, "loss": 3.7023, "step": 26480 }, { "epoch": 0.0053504, "grad_norm": 0.8781428337097168, "learning_rate": 1.978179958062427e-05, "loss": 3.7684, "step": 26490 }, { "epoch": 0.005376, "grad_norm": 0.866707444190979, "learning_rate": 1.9781632244732682e-05, "loss": 3.8344, "step": 26500 }, { "epoch": 0.0054016, "grad_norm": 0.7657451033592224, "learning_rate": 1.9781464845409804e-05, "loss": 3.7028, "step": 26510 }, { "epoch": 0.0054272, "grad_norm": 1.1062520742416382, "learning_rate": 1.9781297382656724e-05, "loss": 3.7462, "step": 26520 }, { "epoch": 0.0054528, "grad_norm": 0.8343626856803894, "learning_rate": 1.9781129856474534e-05, "loss": 3.6896, "step": 26530 }, { "epoch": 0.0054784, "grad_norm": 0.9227779507637024, "learning_rate": 1.9780962266864315e-05, "loss": 3.6984, "step": 26540 }, { "epoch": 0.005504, "grad_norm": 1.7023229598999023, "learning_rate": 1.9780794613827154e-05, "loss": 3.8787, "step": 26550 }, { "epoch": 0.0055296, "grad_norm": 1.183564305305481, "learning_rate": 1.9780626897364142e-05, "loss": 3.831, "step": 26560 }, { "epoch": 0.0055552, "grad_norm": 1.0612499713897705, "learning_rate": 1.9780459117476364e-05, "loss": 3.6685, "step": 26570 }, { "epoch": 0.0055808, "grad_norm": 1.095914602279663, "learning_rate": 1.9780291274164903e-05, "loss": 3.5558, "step": 26580 }, { "epoch": 0.0056064, "grad_norm": 1.0924491882324219, "learning_rate": 1.9780123367430858e-05, "loss": 3.7558, "step": 26590 }, { "epoch": 0.005632, "grad_norm": 1.0873265266418457, "learning_rate": 1.977995539727531e-05, "loss": 3.8455, "step": 26600 }, { "epoch": 0.0056576, "grad_norm": 1.1586145162582397, "learning_rate": 1.977978736369935e-05, "loss": 3.5715, "step": 26610 }, { "epoch": 0.0056832, "grad_norm": 0.97591233253479, "learning_rate": 1.977961926670407e-05, "loss": 4.4021, "step": 26620 }, { "epoch": 0.0057088, "grad_norm": 1.0540977716445923, "learning_rate": 1.9779451106290555e-05, "loss": 3.7263, "step": 26630 }, { "epoch": 0.0057344, "grad_norm": 1.1405786275863647, "learning_rate": 1.97792828824599e-05, "loss": 3.7181, "step": 26640 }, { "epoch": 0.00576, "grad_norm": 1.0155171155929565, "learning_rate": 1.9779114595213196e-05, "loss": 3.7977, "step": 26650 }, { "epoch": 0.0057856, "grad_norm": 0.7162613868713379, "learning_rate": 1.977894624455153e-05, "loss": 3.6643, "step": 26660 }, { "epoch": 0.0058112, "grad_norm": 1.000745177268982, "learning_rate": 1.9778777830476e-05, "loss": 3.3499, "step": 26670 }, { "epoch": 0.0058368, "grad_norm": 0.9935820698738098, "learning_rate": 1.977860935298769e-05, "loss": 3.7459, "step": 26680 }, { "epoch": 0.0058624, "grad_norm": 0.8161382079124451, "learning_rate": 1.97784408120877e-05, "loss": 3.7021, "step": 26690 }, { "epoch": 0.005888, "grad_norm": 1.0300573110580444, "learning_rate": 1.977827220777712e-05, "loss": 3.8408, "step": 26700 }, { "epoch": 0.0059136, "grad_norm": 0.9428972601890564, "learning_rate": 1.9778103540057048e-05, "loss": 3.9508, "step": 26710 }, { "epoch": 0.0059392, "grad_norm": 1.0745337009429932, "learning_rate": 1.9777934808928566e-05, "loss": 3.8306, "step": 26720 }, { "epoch": 0.0059648, "grad_norm": 0.8630613088607788, "learning_rate": 1.977776601439278e-05, "loss": 3.7844, "step": 26730 }, { "epoch": 0.0059904, "grad_norm": 0.9464666247367859, "learning_rate": 1.9777597156450777e-05, "loss": 3.7752, "step": 26740 }, { "epoch": 0.006016, "grad_norm": 0.8472693562507629, "learning_rate": 1.977742823510366e-05, "loss": 3.6709, "step": 26750 }, { "epoch": 0.0060416, "grad_norm": 0.8914229273796082, "learning_rate": 1.9777259250352513e-05, "loss": 3.7739, "step": 26760 }, { "epoch": 0.0060672, "grad_norm": 0.9148170948028564, "learning_rate": 1.977709020219844e-05, "loss": 3.5944, "step": 26770 }, { "epoch": 0.0060928, "grad_norm": 0.8214684724807739, "learning_rate": 1.977692109064254e-05, "loss": 3.6279, "step": 26780 }, { "epoch": 0.0061184, "grad_norm": 1.38887357711792, "learning_rate": 1.97767519156859e-05, "loss": 3.7493, "step": 26790 }, { "epoch": 0.006144, "grad_norm": 1.1011775732040405, "learning_rate": 1.9776582677329622e-05, "loss": 3.7177, "step": 26800 }, { "epoch": 0.0061696, "grad_norm": 0.7700881361961365, "learning_rate": 1.9776413375574807e-05, "loss": 3.6693, "step": 26810 }, { "epoch": 0.0061952, "grad_norm": 0.8361510038375854, "learning_rate": 1.9776244010422546e-05, "loss": 3.6346, "step": 26820 }, { "epoch": 0.0062208, "grad_norm": 0.8049379587173462, "learning_rate": 1.9776074581873943e-05, "loss": 3.5577, "step": 26830 }, { "epoch": 0.0062464, "grad_norm": 1.3250408172607422, "learning_rate": 1.9775905089930093e-05, "loss": 3.9531, "step": 26840 }, { "epoch": 0.006272, "grad_norm": 0.9594144821166992, "learning_rate": 1.9775735534592102e-05, "loss": 4.3125, "step": 26850 }, { "epoch": 0.0062976, "grad_norm": 0.9316559433937073, "learning_rate": 1.9775565915861057e-05, "loss": 3.6648, "step": 26860 }, { "epoch": 0.0063232, "grad_norm": 0.8279734253883362, "learning_rate": 1.977539623373807e-05, "loss": 3.7973, "step": 26870 }, { "epoch": 0.0063488, "grad_norm": 0.9537367224693298, "learning_rate": 1.9775226488224235e-05, "loss": 3.7061, "step": 26880 }, { "epoch": 0.0063744, "grad_norm": 0.8869485855102539, "learning_rate": 1.9775056679320653e-05, "loss": 3.8976, "step": 26890 }, { "epoch": 0.0064, "grad_norm": 0.8653908967971802, "learning_rate": 1.977488680702843e-05, "loss": 3.6756, "step": 26900 }, { "epoch": 0.0064256, "grad_norm": 1.3749995231628418, "learning_rate": 1.9774716871348657e-05, "loss": 3.9114, "step": 26910 }, { "epoch": 0.0064512, "grad_norm": 1.6109416484832764, "learning_rate": 1.9774546872282448e-05, "loss": 3.6411, "step": 26920 }, { "epoch": 0.0064768, "grad_norm": 0.9398743510246277, "learning_rate": 1.9774376809830897e-05, "loss": 3.7493, "step": 26930 }, { "epoch": 0.0065024, "grad_norm": 0.9599744081497192, "learning_rate": 1.9774206683995115e-05, "loss": 3.7009, "step": 26940 }, { "epoch": 0.006528, "grad_norm": 0.8756610155105591, "learning_rate": 1.9774036494776196e-05, "loss": 3.527, "step": 26950 }, { "epoch": 0.0065536, "grad_norm": 0.8595810532569885, "learning_rate": 1.9773866242175252e-05, "loss": 3.4382, "step": 26960 }, { "epoch": 0.0065792, "grad_norm": 2.479483127593994, "learning_rate": 1.977369592619338e-05, "loss": 3.9057, "step": 26970 }, { "epoch": 0.0066048, "grad_norm": 1.9807227849960327, "learning_rate": 1.9773525546831685e-05, "loss": 3.8541, "step": 26980 }, { "epoch": 0.0066304, "grad_norm": 0.8632097840309143, "learning_rate": 1.977335510409128e-05, "loss": 3.637, "step": 26990 }, { "epoch": 0.006656, "grad_norm": 1.0657916069030762, "learning_rate": 1.9773184597973262e-05, "loss": 3.664, "step": 27000 }, { "epoch": 0.0066816, "grad_norm": 1.134243130683899, "learning_rate": 1.977301402847874e-05, "loss": 3.82, "step": 27010 }, { "epoch": 0.0067072, "grad_norm": 3.174077033996582, "learning_rate": 1.9772843395608817e-05, "loss": 3.7769, "step": 27020 }, { "epoch": 0.0067328, "grad_norm": 0.8370055556297302, "learning_rate": 1.9772672699364603e-05, "loss": 3.71, "step": 27030 }, { "epoch": 0.0067584, "grad_norm": 0.8513873219490051, "learning_rate": 1.9772501939747205e-05, "loss": 3.8935, "step": 27040 }, { "epoch": 0.006784, "grad_norm": 0.9226246476173401, "learning_rate": 1.9772331116757727e-05, "loss": 3.7158, "step": 27050 }, { "epoch": 0.0068096, "grad_norm": 1.102384328842163, "learning_rate": 1.9772160230397282e-05, "loss": 3.9267, "step": 27060 }, { "epoch": 0.0068352, "grad_norm": 0.9446311593055725, "learning_rate": 1.9771989280666972e-05, "loss": 3.7857, "step": 27070 }, { "epoch": 0.0068608, "grad_norm": 1.0611298084259033, "learning_rate": 1.977181826756791e-05, "loss": 3.5546, "step": 27080 }, { "epoch": 0.0068864, "grad_norm": 1.0201520919799805, "learning_rate": 1.9771647191101205e-05, "loss": 3.7712, "step": 27090 }, { "epoch": 0.006912, "grad_norm": 0.8802912831306458, "learning_rate": 1.9771476051267963e-05, "loss": 3.7176, "step": 27100 }, { "epoch": 0.0069376, "grad_norm": 1.1344144344329834, "learning_rate": 1.9771304848069298e-05, "loss": 4.0053, "step": 27110 }, { "epoch": 0.0069632, "grad_norm": 0.8532383441925049, "learning_rate": 1.9771133581506314e-05, "loss": 3.7219, "step": 27120 }, { "epoch": 0.0069888, "grad_norm": 1.315616250038147, "learning_rate": 1.977096225158013e-05, "loss": 3.9514, "step": 27130 }, { "epoch": 0.0070144, "grad_norm": 0.9166871309280396, "learning_rate": 1.9770790858291848e-05, "loss": 3.5913, "step": 27140 }, { "epoch": 0.00704, "grad_norm": 0.7842753529548645, "learning_rate": 1.9770619401642585e-05, "loss": 3.8237, "step": 27150 }, { "epoch": 0.0070656, "grad_norm": 1.0440181493759155, "learning_rate": 1.9770447881633455e-05, "loss": 3.7258, "step": 27160 }, { "epoch": 0.0070912, "grad_norm": 1.2890771627426147, "learning_rate": 1.9770276298265566e-05, "loss": 3.7387, "step": 27170 }, { "epoch": 0.0071168, "grad_norm": 1.1421282291412354, "learning_rate": 1.977010465154003e-05, "loss": 3.6105, "step": 27180 }, { "epoch": 0.0071424, "grad_norm": 1.1611143350601196, "learning_rate": 1.9769932941457964e-05, "loss": 3.8729, "step": 27190 }, { "epoch": 0.007168, "grad_norm": 0.8938225507736206, "learning_rate": 1.976976116802048e-05, "loss": 3.9862, "step": 27200 }, { "epoch": 0.0071936, "grad_norm": 0.7961449027061462, "learning_rate": 1.9769589331228688e-05, "loss": 3.56, "step": 27210 }, { "epoch": 0.0072192, "grad_norm": 0.7501410841941833, "learning_rate": 1.9769417431083707e-05, "loss": 3.9052, "step": 27220 }, { "epoch": 0.0072448, "grad_norm": 0.8139942288398743, "learning_rate": 1.976924546758665e-05, "loss": 3.9952, "step": 27230 }, { "epoch": 0.0072704, "grad_norm": 1.6573948860168457, "learning_rate": 1.9769073440738635e-05, "loss": 3.9239, "step": 27240 }, { "epoch": 0.007296, "grad_norm": 0.7937262058258057, "learning_rate": 1.9768901350540776e-05, "loss": 3.7435, "step": 27250 }, { "epoch": 0.0073216, "grad_norm": 0.9715018272399902, "learning_rate": 1.9768729196994183e-05, "loss": 3.6093, "step": 27260 }, { "epoch": 0.0073472, "grad_norm": 0.9111701846122742, "learning_rate": 1.976855698009998e-05, "loss": 3.8475, "step": 27270 }, { "epoch": 0.0073728, "grad_norm": 0.9184595346450806, "learning_rate": 1.976838469985928e-05, "loss": 3.8779, "step": 27280 }, { "epoch": 0.0073984, "grad_norm": 1.1216949224472046, "learning_rate": 1.9768212356273202e-05, "loss": 3.8497, "step": 27290 }, { "epoch": 0.007424, "grad_norm": 0.8435598611831665, "learning_rate": 1.9768039949342866e-05, "loss": 3.5623, "step": 27300 }, { "epoch": 0.0074496, "grad_norm": 1.0188097953796387, "learning_rate": 1.9767867479069384e-05, "loss": 4.2253, "step": 27310 }, { "epoch": 0.0074752, "grad_norm": 0.8270360231399536, "learning_rate": 1.9767694945453877e-05, "loss": 3.7213, "step": 27320 }, { "epoch": 0.0075008, "grad_norm": 1.179268717765808, "learning_rate": 1.9767522348497463e-05, "loss": 3.6564, "step": 27330 }, { "epoch": 0.0075264, "grad_norm": 1.5051449537277222, "learning_rate": 1.9767349688201266e-05, "loss": 4.0853, "step": 27340 }, { "epoch": 0.007552, "grad_norm": 1.3315389156341553, "learning_rate": 1.97671769645664e-05, "loss": 4.2349, "step": 27350 }, { "epoch": 0.0075776, "grad_norm": 0.8882696032524109, "learning_rate": 1.976700417759399e-05, "loss": 3.7509, "step": 27360 }, { "epoch": 0.0076032, "grad_norm": 0.9721698760986328, "learning_rate": 1.976683132728515e-05, "loss": 3.6683, "step": 27370 }, { "epoch": 0.0076288, "grad_norm": 0.8882979154586792, "learning_rate": 1.9766658413641005e-05, "loss": 3.6238, "step": 27380 }, { "epoch": 0.0076544, "grad_norm": 0.969214141368866, "learning_rate": 1.9766485436662676e-05, "loss": 3.7628, "step": 27390 }, { "epoch": 0.00768, "grad_norm": 0.8228253126144409, "learning_rate": 1.9766312396351287e-05, "loss": 3.6147, "step": 27400 }, { "epoch": 2.56e-05, "grad_norm": 1.0117019414901733, "learning_rate": 1.9766139292707956e-05, "loss": 3.5515, "step": 27410 }, { "epoch": 5.12e-05, "grad_norm": 0.8879646062850952, "learning_rate": 1.9765966125733805e-05, "loss": 3.3196, "step": 27420 }, { "epoch": 7.68e-05, "grad_norm": 2.2280561923980713, "learning_rate": 1.976579289542996e-05, "loss": 3.7108, "step": 27430 }, { "epoch": 0.0001024, "grad_norm": 1.1298894882202148, "learning_rate": 1.9765619601797547e-05, "loss": 3.3574, "step": 27440 }, { "epoch": 0.000128, "grad_norm": 1.2068027257919312, "learning_rate": 1.9765446244837684e-05, "loss": 3.4323, "step": 27450 }, { "epoch": 0.0001536, "grad_norm": 1.266340970993042, "learning_rate": 1.97652728245515e-05, "loss": 3.8332, "step": 27460 }, { "epoch": 0.0001792, "grad_norm": 1.2631169557571411, "learning_rate": 1.976509934094011e-05, "loss": 3.4918, "step": 27470 }, { "epoch": 0.0002048, "grad_norm": 0.8306694626808167, "learning_rate": 1.9764925794004656e-05, "loss": 3.4862, "step": 27480 }, { "epoch": 0.0002304, "grad_norm": 0.9339821934700012, "learning_rate": 1.9764752183746244e-05, "loss": 3.5314, "step": 27490 }, { "epoch": 0.000256, "grad_norm": 1.9653249979019165, "learning_rate": 1.9764578510166017e-05, "loss": 3.5856, "step": 27500 }, { "epoch": 0.0002816, "grad_norm": 2.2943508625030518, "learning_rate": 1.976440477326509e-05, "loss": 3.7036, "step": 27510 }, { "epoch": 0.0003072, "grad_norm": 0.8901084065437317, "learning_rate": 1.976423097304459e-05, "loss": 3.4158, "step": 27520 }, { "epoch": 0.0003328, "grad_norm": 0.8363661766052246, "learning_rate": 1.976405710950565e-05, "loss": 3.4083, "step": 27530 }, { "epoch": 0.0003584, "grad_norm": 1.02206289768219, "learning_rate": 1.9763883182649396e-05, "loss": 3.3898, "step": 27540 }, { "epoch": 0.000384, "grad_norm": 1.1584395170211792, "learning_rate": 1.976370919247695e-05, "loss": 3.3736, "step": 27550 }, { "epoch": 0.0004096, "grad_norm": 0.7786848545074463, "learning_rate": 1.976353513898945e-05, "loss": 3.4271, "step": 27560 }, { "epoch": 0.0004352, "grad_norm": 0.9938954710960388, "learning_rate": 1.976336102218802e-05, "loss": 3.5168, "step": 27570 }, { "epoch": 0.0004608, "grad_norm": 0.846645712852478, "learning_rate": 1.9763186842073785e-05, "loss": 3.5372, "step": 27580 }, { "epoch": 0.0004864, "grad_norm": 1.0372785329818726, "learning_rate": 1.976301259864788e-05, "loss": 3.5101, "step": 27590 }, { "epoch": 0.000512, "grad_norm": 1.00088369846344, "learning_rate": 1.976283829191143e-05, "loss": 3.6091, "step": 27600 }, { "epoch": 0.0005376, "grad_norm": 1.133019208908081, "learning_rate": 1.976266392186557e-05, "loss": 3.534, "step": 27610 }, { "epoch": 0.0005632, "grad_norm": 0.9245799779891968, "learning_rate": 1.9762489488511432e-05, "loss": 3.6284, "step": 27620 }, { "epoch": 0.0005888, "grad_norm": 0.7179731130599976, "learning_rate": 1.976231499185014e-05, "loss": 3.3998, "step": 27630 }, { "epoch": 0.0006144, "grad_norm": 1.0163795948028564, "learning_rate": 1.976214043188283e-05, "loss": 3.6954, "step": 27640 }, { "epoch": 0.00064, "grad_norm": 1.2261146306991577, "learning_rate": 1.9761965808610637e-05, "loss": 3.6842, "step": 27650 }, { "epoch": 0.0006656, "grad_norm": 1.0051946640014648, "learning_rate": 1.976179112203469e-05, "loss": 3.4793, "step": 27660 }, { "epoch": 0.0006912, "grad_norm": 0.8425965905189514, "learning_rate": 1.9761616372156117e-05, "loss": 3.6925, "step": 27670 }, { "epoch": 0.0007168, "grad_norm": 0.7887319922447205, "learning_rate": 1.9761441558976062e-05, "loss": 3.7669, "step": 27680 }, { "epoch": 0.0007424, "grad_norm": 1.07819402217865, "learning_rate": 1.976126668249565e-05, "loss": 3.65, "step": 27690 }, { "epoch": 0.000768, "grad_norm": 1.0133533477783203, "learning_rate": 1.976109174271602e-05, "loss": 3.4739, "step": 27700 }, { "epoch": 0.0007936, "grad_norm": 1.090541124343872, "learning_rate": 1.97609167396383e-05, "loss": 3.4157, "step": 27710 }, { "epoch": 0.0008192, "grad_norm": 0.97044438123703, "learning_rate": 1.9760741673263634e-05, "loss": 3.4651, "step": 27720 }, { "epoch": 0.0008448, "grad_norm": 0.8614478707313538, "learning_rate": 1.9760566543593147e-05, "loss": 3.349, "step": 27730 }, { "epoch": 0.0008704, "grad_norm": 0.9268606305122375, "learning_rate": 1.9760391350627984e-05, "loss": 3.4601, "step": 27740 }, { "epoch": 0.000896, "grad_norm": 1.150446891784668, "learning_rate": 1.9760216094369277e-05, "loss": 3.5278, "step": 27750 }, { "epoch": 0.0009216, "grad_norm": 0.911454439163208, "learning_rate": 1.9760040774818157e-05, "loss": 3.4913, "step": 27760 }, { "epoch": 0.0009472, "grad_norm": 0.875829815864563, "learning_rate": 1.9759865391975773e-05, "loss": 3.6589, "step": 27770 }, { "epoch": 0.0009728, "grad_norm": 0.9124084711074829, "learning_rate": 1.9759689945843254e-05, "loss": 3.5764, "step": 27780 }, { "epoch": 0.0009984, "grad_norm": 0.8872256875038147, "learning_rate": 1.975951443642174e-05, "loss": 3.7273, "step": 27790 }, { "epoch": 0.001024, "grad_norm": 1.4032951593399048, "learning_rate": 1.9759338863712364e-05, "loss": 3.6028, "step": 27800 }, { "epoch": 0.0010496, "grad_norm": 0.8222949504852295, "learning_rate": 1.9759163227716278e-05, "loss": 3.4601, "step": 27810 }, { "epoch": 0.0010752, "grad_norm": 0.7114614844322205, "learning_rate": 1.9758987528434606e-05, "loss": 3.4138, "step": 27820 }, { "epoch": 0.0011008, "grad_norm": 1.1176161766052246, "learning_rate": 1.9758811765868493e-05, "loss": 3.6457, "step": 27830 }, { "epoch": 0.0011264, "grad_norm": 1.1524637937545776, "learning_rate": 1.9758635940019082e-05, "loss": 3.7652, "step": 27840 }, { "epoch": 0.001152, "grad_norm": 0.9368055462837219, "learning_rate": 1.9758460050887508e-05, "loss": 3.1772, "step": 27850 }, { "epoch": 0.0011776, "grad_norm": 0.8674799203872681, "learning_rate": 1.9758284098474915e-05, "loss": 3.5678, "step": 27860 }, { "epoch": 0.0012032, "grad_norm": 0.8745134472846985, "learning_rate": 1.9758108082782446e-05, "loss": 3.4906, "step": 27870 }, { "epoch": 0.0012288, "grad_norm": 1.594111680984497, "learning_rate": 1.9757932003811235e-05, "loss": 3.2949, "step": 27880 }, { "epoch": 0.0012544, "grad_norm": 0.9877114295959473, "learning_rate": 1.975775586156243e-05, "loss": 3.6435, "step": 27890 }, { "epoch": 0.00128, "grad_norm": 0.9376205205917358, "learning_rate": 1.9757579656037173e-05, "loss": 3.7942, "step": 27900 }, { "epoch": 0.0013056, "grad_norm": 0.8528583645820618, "learning_rate": 1.9757403387236606e-05, "loss": 3.529, "step": 27910 }, { "epoch": 0.0013312, "grad_norm": 1.093148946762085, "learning_rate": 1.9757227055161867e-05, "loss": 3.4036, "step": 27920 }, { "epoch": 0.0013568, "grad_norm": 0.8589590191841125, "learning_rate": 1.9757050659814106e-05, "loss": 3.9788, "step": 27930 }, { "epoch": 0.0013824, "grad_norm": 0.8193768858909607, "learning_rate": 1.9756874201194465e-05, "loss": 3.5031, "step": 27940 }, { "epoch": 0.001408, "grad_norm": 2.1704509258270264, "learning_rate": 1.9756697679304088e-05, "loss": 3.7002, "step": 27950 }, { "epoch": 0.0014336, "grad_norm": 0.8448336124420166, "learning_rate": 1.9756521094144118e-05, "loss": 3.6689, "step": 27960 }, { "epoch": 0.0014592, "grad_norm": 1.1172915697097778, "learning_rate": 1.9756344445715702e-05, "loss": 3.6444, "step": 27970 }, { "epoch": 0.0014848, "grad_norm": 1.143086314201355, "learning_rate": 1.9756167734019987e-05, "loss": 3.6907, "step": 27980 }, { "epoch": 0.0015104, "grad_norm": 2.0642881393432617, "learning_rate": 1.9755990959058118e-05, "loss": 3.6358, "step": 27990 }, { "epoch": 0.001536, "grad_norm": 1.1466975212097168, "learning_rate": 1.975581412083124e-05, "loss": 3.6827, "step": 28000 }, { "epoch": 0.0015616, "grad_norm": 1.0992164611816406, "learning_rate": 1.9755637219340495e-05, "loss": 3.6653, "step": 28010 }, { "epoch": 0.0015872, "grad_norm": 0.9618198871612549, "learning_rate": 1.975546025458704e-05, "loss": 3.7226, "step": 28020 }, { "epoch": 0.0016128, "grad_norm": 2.1130101680755615, "learning_rate": 1.9755283226572017e-05, "loss": 3.832, "step": 28030 }, { "epoch": 0.0016384, "grad_norm": 1.1790531873703003, "learning_rate": 1.9755106135296572e-05, "loss": 3.6065, "step": 28040 }, { "epoch": 0.001664, "grad_norm": 1.100949764251709, "learning_rate": 1.975492898076186e-05, "loss": 3.7101, "step": 28050 }, { "epoch": 0.0016896, "grad_norm": 0.9766319990158081, "learning_rate": 1.9754751762969025e-05, "loss": 3.5962, "step": 28060 }, { "epoch": 0.0017152, "grad_norm": 0.9524109959602356, "learning_rate": 1.975457448191922e-05, "loss": 3.4227, "step": 28070 }, { "epoch": 0.0017408, "grad_norm": 0.9745418429374695, "learning_rate": 1.9754397137613588e-05, "loss": 3.5862, "step": 28080 }, { "epoch": 0.0017664, "grad_norm": 0.9311702251434326, "learning_rate": 1.9754219730053286e-05, "loss": 3.7546, "step": 28090 }, { "epoch": 0.001792, "grad_norm": 0.9118032455444336, "learning_rate": 1.975404225923946e-05, "loss": 3.7526, "step": 28100 }, { "epoch": 0.0018176, "grad_norm": 1.0286507606506348, "learning_rate": 1.975386472517326e-05, "loss": 3.4869, "step": 28110 }, { "epoch": 0.0018432, "grad_norm": 1.1474844217300415, "learning_rate": 1.9753687127855844e-05, "loss": 3.8321, "step": 28120 }, { "epoch": 0.0018688, "grad_norm": 0.7971705198287964, "learning_rate": 1.9753509467288356e-05, "loss": 3.7386, "step": 28130 }, { "epoch": 0.0018944, "grad_norm": 1.2166430950164795, "learning_rate": 1.9753331743471952e-05, "loss": 3.8379, "step": 28140 }, { "epoch": 0.00192, "grad_norm": 0.8434909582138062, "learning_rate": 1.9753153956407788e-05, "loss": 3.8194, "step": 28150 }, { "epoch": 0.0019456, "grad_norm": 2.04183030128479, "learning_rate": 1.9752976106097006e-05, "loss": 3.7999, "step": 28160 }, { "epoch": 0.0019712, "grad_norm": 0.8984522223472595, "learning_rate": 1.9752798192540766e-05, "loss": 3.7547, "step": 28170 }, { "epoch": 0.0019968, "grad_norm": 0.7232092022895813, "learning_rate": 1.9752620215740226e-05, "loss": 3.6939, "step": 28180 }, { "epoch": 0.0020224, "grad_norm": 0.958294689655304, "learning_rate": 1.9752442175696537e-05, "loss": 3.5654, "step": 28190 }, { "epoch": 0.002048, "grad_norm": 0.9175833463668823, "learning_rate": 1.9752264072410846e-05, "loss": 3.5271, "step": 28200 }, { "epoch": 0.0020736, "grad_norm": 0.7590486407279968, "learning_rate": 1.9752085905884318e-05, "loss": 3.7393, "step": 28210 }, { "epoch": 0.0020992, "grad_norm": 1.0468664169311523, "learning_rate": 1.9751907676118106e-05, "loss": 3.6458, "step": 28220 }, { "epoch": 0.0021248, "grad_norm": 0.9122461676597595, "learning_rate": 1.9751729383113367e-05, "loss": 4.0837, "step": 28230 }, { "epoch": 0.0021504, "grad_norm": 0.8593602180480957, "learning_rate": 1.9751551026871248e-05, "loss": 4.0538, "step": 28240 }, { "epoch": 0.002176, "grad_norm": 0.8618978261947632, "learning_rate": 1.9751372607392912e-05, "loss": 3.6065, "step": 28250 }, { "epoch": 0.0022016, "grad_norm": 0.8529167771339417, "learning_rate": 1.975119412467952e-05, "loss": 3.8343, "step": 28260 }, { "epoch": 0.0022272, "grad_norm": 0.9704717993736267, "learning_rate": 1.9751015578732228e-05, "loss": 3.7361, "step": 28270 }, { "epoch": 0.0022528, "grad_norm": 0.7666947245597839, "learning_rate": 1.9750836969552186e-05, "loss": 3.5032, "step": 28280 }, { "epoch": 0.0022784, "grad_norm": 1.0805498361587524, "learning_rate": 1.9750658297140563e-05, "loss": 3.6247, "step": 28290 }, { "epoch": 0.002304, "grad_norm": 1.0598112344741821, "learning_rate": 1.9750479561498506e-05, "loss": 3.7926, "step": 28300 }, { "epoch": 0.0023296, "grad_norm": 0.9535551071166992, "learning_rate": 1.9750300762627184e-05, "loss": 3.5498, "step": 28310 }, { "epoch": 0.0023552, "grad_norm": 1.6106666326522827, "learning_rate": 1.9750121900527753e-05, "loss": 3.8084, "step": 28320 }, { "epoch": 0.0023808, "grad_norm": 1.0141377449035645, "learning_rate": 1.9749942975201372e-05, "loss": 3.8218, "step": 28330 }, { "epoch": 0.0024064, "grad_norm": 1.449714183807373, "learning_rate": 1.9749763986649202e-05, "loss": 3.8259, "step": 28340 }, { "epoch": 0.002432, "grad_norm": 0.9339233040809631, "learning_rate": 1.9749584934872406e-05, "loss": 3.7622, "step": 28350 }, { "epoch": 0.0024576, "grad_norm": 1.4429783821105957, "learning_rate": 1.974940581987214e-05, "loss": 3.7441, "step": 28360 }, { "epoch": 0.0024832, "grad_norm": 1.0001418590545654, "learning_rate": 1.974922664164957e-05, "loss": 3.7664, "step": 28370 }, { "epoch": 0.0025088, "grad_norm": 0.8686537742614746, "learning_rate": 1.9749047400205858e-05, "loss": 3.843, "step": 28380 }, { "epoch": 0.0025344, "grad_norm": 0.7112681269645691, "learning_rate": 1.974886809554216e-05, "loss": 3.4861, "step": 28390 }, { "epoch": 0.00256, "grad_norm": 1.2466380596160889, "learning_rate": 1.9748688727659645e-05, "loss": 3.6994, "step": 28400 }, { "epoch": 0.0025856, "grad_norm": 0.7876825928688049, "learning_rate": 1.9748509296559476e-05, "loss": 3.7343, "step": 28410 }, { "epoch": 0.0026112, "grad_norm": 0.8533666133880615, "learning_rate": 1.9748329802242815e-05, "loss": 3.6181, "step": 28420 }, { "epoch": 0.0026368, "grad_norm": 0.9850829243659973, "learning_rate": 1.9748150244710825e-05, "loss": 3.6598, "step": 28430 }, { "epoch": 0.0026624, "grad_norm": 1.523909091949463, "learning_rate": 1.974797062396467e-05, "loss": 3.6155, "step": 28440 }, { "epoch": 0.002688, "grad_norm": 1.7499377727508545, "learning_rate": 1.9747790940005517e-05, "loss": 3.7164, "step": 28450 }, { "epoch": 0.0027136, "grad_norm": 0.878546953201294, "learning_rate": 1.9747611192834532e-05, "loss": 3.6239, "step": 28460 }, { "epoch": 0.0027392, "grad_norm": 1.1615023612976074, "learning_rate": 1.974743138245288e-05, "loss": 3.6126, "step": 28470 }, { "epoch": 0.0027648, "grad_norm": 1.0945698022842407, "learning_rate": 1.9747251508861724e-05, "loss": 3.5516, "step": 28480 }, { "epoch": 0.0027904, "grad_norm": 0.8823060393333435, "learning_rate": 1.9747071572062234e-05, "loss": 3.588, "step": 28490 }, { "epoch": 0.002816, "grad_norm": 1.9522554874420166, "learning_rate": 1.9746891572055574e-05, "loss": 4.0319, "step": 28500 }, { "epoch": 0.0028416, "grad_norm": 0.7604995965957642, "learning_rate": 1.9746711508842912e-05, "loss": 3.7122, "step": 28510 }, { "epoch": 0.0028672, "grad_norm": 1.6035096645355225, "learning_rate": 1.9746531382425414e-05, "loss": 3.7261, "step": 28520 }, { "epoch": 0.0028928, "grad_norm": 1.0400598049163818, "learning_rate": 1.9746351192804256e-05, "loss": 3.7046, "step": 28530 }, { "epoch": 0.0029184, "grad_norm": 1.3627322912216187, "learning_rate": 1.9746170939980595e-05, "loss": 3.7004, "step": 28540 }, { "epoch": 0.002944, "grad_norm": 0.784325897693634, "learning_rate": 1.974599062395561e-05, "loss": 3.5549, "step": 28550 }, { "epoch": 0.0029696, "grad_norm": 1.0840983390808105, "learning_rate": 1.9745810244730465e-05, "loss": 3.3469, "step": 28560 }, { "epoch": 0.0029952, "grad_norm": 0.8778899312019348, "learning_rate": 1.974562980230633e-05, "loss": 3.7477, "step": 28570 }, { "epoch": 0.0030208, "grad_norm": 0.8332865834236145, "learning_rate": 1.9745449296684375e-05, "loss": 3.4628, "step": 28580 }, { "epoch": 0.0030464, "grad_norm": 0.7544131875038147, "learning_rate": 1.9745268727865774e-05, "loss": 3.6437, "step": 28590 }, { "epoch": 0.003072, "grad_norm": 1.2494159936904907, "learning_rate": 1.9745088095851694e-05, "loss": 3.7821, "step": 28600 }, { "epoch": 0.0030976, "grad_norm": 1.893868088722229, "learning_rate": 1.9744907400643308e-05, "loss": 3.9547, "step": 28610 }, { "epoch": 0.0031232, "grad_norm": 0.8365073800086975, "learning_rate": 1.9744726642241785e-05, "loss": 3.8806, "step": 28620 }, { "epoch": 0.0031488, "grad_norm": 1.3617337942123413, "learning_rate": 1.97445458206483e-05, "loss": 3.7611, "step": 28630 }, { "epoch": 0.0031744, "grad_norm": 0.8698869347572327, "learning_rate": 1.974436493586403e-05, "loss": 3.8887, "step": 28640 }, { "epoch": 0.0032, "grad_norm": 0.9536399841308594, "learning_rate": 1.974418398789014e-05, "loss": 3.7555, "step": 28650 }, { "epoch": 0.0032256, "grad_norm": 0.9353407025337219, "learning_rate": 1.9744002976727805e-05, "loss": 3.5438, "step": 28660 }, { "epoch": 0.0032512, "grad_norm": 0.8457173705101013, "learning_rate": 1.97438219023782e-05, "loss": 3.7186, "step": 28670 }, { "epoch": 0.0032768, "grad_norm": 0.8477301597595215, "learning_rate": 1.9743640764842502e-05, "loss": 3.4343, "step": 28680 }, { "epoch": 0.0033024, "grad_norm": 0.7500773072242737, "learning_rate": 1.9743459564121882e-05, "loss": 3.8049, "step": 28690 }, { "epoch": 0.003328, "grad_norm": 0.7715641260147095, "learning_rate": 1.9743278300217517e-05, "loss": 3.5238, "step": 28700 }, { "epoch": 0.0033536, "grad_norm": 0.9367055892944336, "learning_rate": 1.974309697313058e-05, "loss": 3.4371, "step": 28710 }, { "epoch": 0.0033792, "grad_norm": 0.9851276874542236, "learning_rate": 1.974291558286225e-05, "loss": 3.5993, "step": 28720 }, { "epoch": 0.0034048, "grad_norm": 0.8639968633651733, "learning_rate": 1.9742734129413698e-05, "loss": 3.5966, "step": 28730 }, { "epoch": 0.0034304, "grad_norm": 1.0381178855895996, "learning_rate": 1.974255261278611e-05, "loss": 3.3797, "step": 28740 }, { "epoch": 0.003456, "grad_norm": 0.957491397857666, "learning_rate": 1.9742371032980657e-05, "loss": 3.5816, "step": 28750 }, { "epoch": 0.0034816, "grad_norm": 0.895439863204956, "learning_rate": 1.9742189389998515e-05, "loss": 3.7645, "step": 28760 }, { "epoch": 0.0035072, "grad_norm": 0.9546801447868347, "learning_rate": 1.9742007683840863e-05, "loss": 3.7506, "step": 28770 }, { "epoch": 0.0035328, "grad_norm": 0.8426961898803711, "learning_rate": 1.9741825914508882e-05, "loss": 3.5669, "step": 28780 }, { "epoch": 0.0035584, "grad_norm": 0.8752999901771545, "learning_rate": 1.974164408200375e-05, "loss": 3.5299, "step": 28790 }, { "epoch": 0.003584, "grad_norm": 0.9128421545028687, "learning_rate": 1.9741462186326642e-05, "loss": 3.4067, "step": 28800 }, { "epoch": 0.0036096, "grad_norm": 0.8887921571731567, "learning_rate": 1.9741280227478744e-05, "loss": 3.7076, "step": 28810 }, { "epoch": 0.0036352, "grad_norm": 1.0036323070526123, "learning_rate": 1.9741098205461232e-05, "loss": 3.5997, "step": 28820 }, { "epoch": 0.0036608, "grad_norm": 1.3386856317520142, "learning_rate": 1.9740916120275283e-05, "loss": 3.5252, "step": 28830 }, { "epoch": 0.0036864, "grad_norm": 1.0386580228805542, "learning_rate": 1.974073397192209e-05, "loss": 3.8318, "step": 28840 }, { "epoch": 0.003712, "grad_norm": 0.8712669610977173, "learning_rate": 1.974055176040282e-05, "loss": 3.8241, "step": 28850 }, { "epoch": 0.0037376, "grad_norm": 0.7610027194023132, "learning_rate": 1.974036948571866e-05, "loss": 3.6421, "step": 28860 }, { "epoch": 0.0037632, "grad_norm": 0.9945865273475647, "learning_rate": 1.9740187147870796e-05, "loss": 3.5539, "step": 28870 }, { "epoch": 0.0037888, "grad_norm": 1.2216367721557617, "learning_rate": 1.9740004746860403e-05, "loss": 3.7262, "step": 28880 }, { "epoch": 0.0038144, "grad_norm": 0.8061459064483643, "learning_rate": 1.973982228268867e-05, "loss": 3.4967, "step": 28890 }, { "epoch": 0.00384, "grad_norm": 0.8977207541465759, "learning_rate": 1.973963975535678e-05, "loss": 3.6139, "step": 28900 }, { "epoch": 2.56e-05, "grad_norm": 1.0819969177246094, "learning_rate": 1.9739457164865912e-05, "loss": 3.7175, "step": 28910 }, { "epoch": 5.12e-05, "grad_norm": 2.0522427558898926, "learning_rate": 1.9739274511217253e-05, "loss": 3.6593, "step": 28920 }, { "epoch": 7.68e-05, "grad_norm": 0.7739502787590027, "learning_rate": 1.973909179441199e-05, "loss": 3.7488, "step": 28930 }, { "epoch": 0.0001024, "grad_norm": 1.043535590171814, "learning_rate": 1.97389090144513e-05, "loss": 3.5972, "step": 28940 }, { "epoch": 0.000128, "grad_norm": 0.8904197812080383, "learning_rate": 1.9738726171336377e-05, "loss": 3.3991, "step": 28950 }, { "epoch": 0.0001536, "grad_norm": 1.1344929933547974, "learning_rate": 1.9738543265068402e-05, "loss": 3.6513, "step": 28960 }, { "epoch": 0.0001792, "grad_norm": 0.9359645843505859, "learning_rate": 1.973836029564856e-05, "loss": 3.6457, "step": 28970 }, { "epoch": 0.0002048, "grad_norm": 0.8810282349586487, "learning_rate": 1.9738177263078043e-05, "loss": 3.4816, "step": 28980 }, { "epoch": 0.0002304, "grad_norm": 1.0869038105010986, "learning_rate": 1.9737994167358033e-05, "loss": 3.4572, "step": 28990 }, { "epoch": 0.000256, "grad_norm": 1.0631994009017944, "learning_rate": 1.973781100848972e-05, "loss": 3.9003, "step": 29000 }, { "epoch": 0.0002816, "grad_norm": 0.8372664451599121, "learning_rate": 1.9737627786474285e-05, "loss": 3.8696, "step": 29010 }, { "epoch": 0.0003072, "grad_norm": 1.6184099912643433, "learning_rate": 1.9737444501312932e-05, "loss": 3.5255, "step": 29020 }, { "epoch": 0.0003328, "grad_norm": 2.6844019889831543, "learning_rate": 1.973726115300683e-05, "loss": 3.6039, "step": 29030 }, { "epoch": 0.0003584, "grad_norm": 1.1548436880111694, "learning_rate": 1.9737077741557184e-05, "loss": 3.5819, "step": 29040 }, { "epoch": 0.000384, "grad_norm": 1.9604755640029907, "learning_rate": 1.9736894266965172e-05, "loss": 3.8557, "step": 29050 }, { "epoch": 0.0004096, "grad_norm": 0.9383719563484192, "learning_rate": 1.973671072923199e-05, "loss": 3.3836, "step": 29060 }, { "epoch": 0.0004352, "grad_norm": 1.046452283859253, "learning_rate": 1.9736527128358828e-05, "loss": 3.6442, "step": 29070 }, { "epoch": 0.0004608, "grad_norm": 1.0312952995300293, "learning_rate": 1.9736343464346874e-05, "loss": 3.4272, "step": 29080 }, { "epoch": 0.0004864, "grad_norm": 0.844529390335083, "learning_rate": 1.973615973719732e-05, "loss": 3.5227, "step": 29090 }, { "epoch": 0.000512, "grad_norm": 0.7849445343017578, "learning_rate": 1.973597594691136e-05, "loss": 3.713, "step": 29100 }, { "epoch": 0.0005376, "grad_norm": 0.8832810521125793, "learning_rate": 1.9735792093490177e-05, "loss": 3.6275, "step": 29110 }, { "epoch": 0.0005632, "grad_norm": 0.9508944153785706, "learning_rate": 1.9735608176934974e-05, "loss": 3.4479, "step": 29120 }, { "epoch": 0.0005888, "grad_norm": 1.3570733070373535, "learning_rate": 1.9735424197246943e-05, "loss": 3.8961, "step": 29130 }, { "epoch": 0.0006144, "grad_norm": 1.051883339881897, "learning_rate": 1.973524015442727e-05, "loss": 3.5363, "step": 29140 }, { "epoch": 0.00064, "grad_norm": 0.8706480264663696, "learning_rate": 1.973505604847715e-05, "loss": 3.575, "step": 29150 }, { "epoch": 0.0006656, "grad_norm": 0.7950775027275085, "learning_rate": 1.9734871879397782e-05, "loss": 3.5598, "step": 29160 }, { "epoch": 0.0006912, "grad_norm": 0.9359655380249023, "learning_rate": 1.9734687647190357e-05, "loss": 3.6347, "step": 29170 }, { "epoch": 0.0007168, "grad_norm": 0.8140844702720642, "learning_rate": 1.9734503351856067e-05, "loss": 3.4157, "step": 29180 }, { "epoch": 0.0007424, "grad_norm": 0.8476618528366089, "learning_rate": 1.973431899339611e-05, "loss": 3.431, "step": 29190 }, { "epoch": 0.000768, "grad_norm": 0.6975016593933105, "learning_rate": 1.9734134571811686e-05, "loss": 3.5846, "step": 29200 }, { "epoch": 0.0007936, "grad_norm": 0.9270142912864685, "learning_rate": 1.973395008710398e-05, "loss": 3.5491, "step": 29210 }, { "epoch": 0.0008192, "grad_norm": 0.9243552088737488, "learning_rate": 1.97337655392742e-05, "loss": 3.6729, "step": 29220 }, { "epoch": 0.0008448, "grad_norm": 0.8183483481407166, "learning_rate": 1.9733580928323535e-05, "loss": 3.6298, "step": 29230 }, { "epoch": 0.0008704, "grad_norm": 0.8490518927574158, "learning_rate": 1.9733396254253184e-05, "loss": 3.4786, "step": 29240 }, { "epoch": 0.000896, "grad_norm": 0.9460092782974243, "learning_rate": 1.9733211517064347e-05, "loss": 3.4914, "step": 29250 }, { "epoch": 0.0009216, "grad_norm": 0.9898030161857605, "learning_rate": 1.973302671675822e-05, "loss": 3.5716, "step": 29260 }, { "epoch": 0.0009472, "grad_norm": 1.0099496841430664, "learning_rate": 1.9732841853335998e-05, "loss": 3.5815, "step": 29270 }, { "epoch": 0.0009728, "grad_norm": 0.8587186336517334, "learning_rate": 1.9732656926798888e-05, "loss": 3.6189, "step": 29280 }, { "epoch": 0.0009984, "grad_norm": 1.1568843126296997, "learning_rate": 1.973247193714808e-05, "loss": 3.5695, "step": 29290 }, { "epoch": 0.001024, "grad_norm": 1.1769750118255615, "learning_rate": 1.9732286884384782e-05, "loss": 3.5993, "step": 29300 }, { "epoch": 0.0010496, "grad_norm": 0.811223030090332, "learning_rate": 1.9732101768510187e-05, "loss": 3.6526, "step": 29310 }, { "epoch": 0.0010752, "grad_norm": 0.8966854810714722, "learning_rate": 1.97319165895255e-05, "loss": 3.7586, "step": 29320 }, { "epoch": 0.0011008, "grad_norm": 0.7537301778793335, "learning_rate": 1.973173134743192e-05, "loss": 3.544, "step": 29330 }, { "epoch": 0.0011264, "grad_norm": 0.8115180730819702, "learning_rate": 1.973154604223065e-05, "loss": 3.643, "step": 29340 }, { "epoch": 0.001152, "grad_norm": 0.7016899585723877, "learning_rate": 1.973136067392289e-05, "loss": 3.4257, "step": 29350 }, { "epoch": 0.0011776, "grad_norm": 0.7840659022331238, "learning_rate": 1.973117524250984e-05, "loss": 3.3556, "step": 29360 }, { "epoch": 0.0012032, "grad_norm": 0.8927159309387207, "learning_rate": 1.9730989747992706e-05, "loss": 3.3518, "step": 29370 }, { "epoch": 0.0012288, "grad_norm": 0.8908610939979553, "learning_rate": 1.973080419037269e-05, "loss": 3.5148, "step": 29380 }, { "epoch": 0.0012544, "grad_norm": 0.870636522769928, "learning_rate": 1.9730618569650995e-05, "loss": 3.5332, "step": 29390 }, { "epoch": 0.00128, "grad_norm": 0.8961625099182129, "learning_rate": 1.9730432885828822e-05, "loss": 3.4836, "step": 29400 }, { "epoch": 0.0013056, "grad_norm": 0.9051622748374939, "learning_rate": 1.9730247138907384e-05, "loss": 3.457, "step": 29410 }, { "epoch": 0.0013312, "grad_norm": 0.8038799166679382, "learning_rate": 1.9730061328887877e-05, "loss": 3.2532, "step": 29420 }, { "epoch": 0.0013568, "grad_norm": 2.565659284591675, "learning_rate": 1.9729875455771506e-05, "loss": 3.6124, "step": 29430 }, { "epoch": 0.0013824, "grad_norm": 1.0581661462783813, "learning_rate": 1.9729689519559478e-05, "loss": 3.3969, "step": 29440 }, { "epoch": 0.001408, "grad_norm": 0.9998385310173035, "learning_rate": 1.9729503520253e-05, "loss": 3.9739, "step": 29450 }, { "epoch": 0.0014336, "grad_norm": 1.4023901224136353, "learning_rate": 1.972931745785328e-05, "loss": 3.552, "step": 29460 }, { "epoch": 0.0014592, "grad_norm": 0.753542423248291, "learning_rate": 1.9729131332361517e-05, "loss": 3.2758, "step": 29470 }, { "epoch": 0.0014848, "grad_norm": 1.2017488479614258, "learning_rate": 1.9728945143778926e-05, "loss": 3.5635, "step": 29480 }, { "epoch": 0.0015104, "grad_norm": 0.945216715335846, "learning_rate": 1.9728758892106713e-05, "loss": 3.5756, "step": 29490 }, { "epoch": 0.001536, "grad_norm": 0.7437297105789185, "learning_rate": 1.9728572577346083e-05, "loss": 3.252, "step": 29500 }, { "epoch": 0.0015616, "grad_norm": 0.7875266671180725, "learning_rate": 1.9728386199498245e-05, "loss": 3.3881, "step": 29510 }, { "epoch": 0.0015872, "grad_norm": 0.9541364312171936, "learning_rate": 1.9728199758564407e-05, "loss": 3.7296, "step": 29520 }, { "epoch": 0.0016128, "grad_norm": 0.9645223021507263, "learning_rate": 1.972801325454578e-05, "loss": 3.5924, "step": 29530 }, { "epoch": 0.0016384, "grad_norm": 0.7998558282852173, "learning_rate": 1.9727826687443574e-05, "loss": 3.5983, "step": 29540 }, { "epoch": 0.001664, "grad_norm": 1.487729787826538, "learning_rate": 1.9727640057258997e-05, "loss": 3.6515, "step": 29550 }, { "epoch": 0.0016896, "grad_norm": 0.9798567295074463, "learning_rate": 1.9727453363993258e-05, "loss": 3.8461, "step": 29560 }, { "epoch": 0.0017152, "grad_norm": 0.8353801965713501, "learning_rate": 1.972726660764757e-05, "loss": 3.5034, "step": 29570 }, { "epoch": 0.0017408, "grad_norm": 0.979978621006012, "learning_rate": 1.972707978822314e-05, "loss": 3.6896, "step": 29580 }, { "epoch": 0.0017664, "grad_norm": 0.9337724447250366, "learning_rate": 1.9726892905721188e-05, "loss": 3.4807, "step": 29590 }, { "epoch": 0.001792, "grad_norm": 0.7650423049926758, "learning_rate": 1.9726705960142917e-05, "loss": 3.3938, "step": 29600 }, { "epoch": 0.0018176, "grad_norm": 0.7877510786056519, "learning_rate": 1.9726518951489542e-05, "loss": 3.3123, "step": 29610 }, { "epoch": 0.0018432, "grad_norm": 0.887415885925293, "learning_rate": 1.972633187976228e-05, "loss": 3.5225, "step": 29620 }, { "epoch": 0.0018688, "grad_norm": 0.7799769639968872, "learning_rate": 1.9726144744962337e-05, "loss": 3.5309, "step": 29630 }, { "epoch": 0.0018944, "grad_norm": 1.1272855997085571, "learning_rate": 1.9725957547090933e-05, "loss": 3.6076, "step": 29640 }, { "epoch": 0.00192, "grad_norm": 0.9207348823547363, "learning_rate": 1.9725770286149283e-05, "loss": 3.5001, "step": 29650 }, { "epoch": 0.0019456, "grad_norm": 0.7262068390846252, "learning_rate": 1.9725582962138593e-05, "loss": 3.4137, "step": 29660 }, { "epoch": 0.0019712, "grad_norm": 1.7737741470336914, "learning_rate": 1.972539557506008e-05, "loss": 3.9011, "step": 29670 }, { "epoch": 0.0019968, "grad_norm": 1.1442204713821411, "learning_rate": 1.972520812491496e-05, "loss": 3.5959, "step": 29680 }, { "epoch": 0.0020224, "grad_norm": 0.8727182149887085, "learning_rate": 1.9725020611704456e-05, "loss": 3.835, "step": 29690 }, { "epoch": 0.002048, "grad_norm": 0.7925219535827637, "learning_rate": 1.9724833035429777e-05, "loss": 3.6158, "step": 29700 }, { "epoch": 0.0020736, "grad_norm": 0.7493284940719604, "learning_rate": 1.972464539609214e-05, "loss": 3.6567, "step": 29710 }, { "epoch": 0.0020992, "grad_norm": 1.374817967414856, "learning_rate": 1.972445769369276e-05, "loss": 3.4025, "step": 29720 }, { "epoch": 0.0021248, "grad_norm": 0.7434877157211304, "learning_rate": 1.9724269928232858e-05, "loss": 3.6983, "step": 29730 }, { "epoch": 0.0021504, "grad_norm": 1.0330034494400024, "learning_rate": 1.972408209971365e-05, "loss": 3.8704, "step": 29740 }, { "epoch": 0.002176, "grad_norm": 1.0901135206222534, "learning_rate": 1.9723894208136354e-05, "loss": 3.6368, "step": 29750 }, { "epoch": 0.0022016, "grad_norm": 0.8841471672058105, "learning_rate": 1.9723706253502187e-05, "loss": 3.6011, "step": 29760 }, { "epoch": 0.0022272, "grad_norm": 0.837200403213501, "learning_rate": 1.9723518235812368e-05, "loss": 3.7386, "step": 29770 }, { "epoch": 0.0022528, "grad_norm": 0.805018424987793, "learning_rate": 1.9723330155068117e-05, "loss": 3.8234, "step": 29780 }, { "epoch": 0.0022784, "grad_norm": 0.980557918548584, "learning_rate": 1.972314201127066e-05, "loss": 3.6127, "step": 29790 }, { "epoch": 0.002304, "grad_norm": 0.8434127569198608, "learning_rate": 1.9722953804421205e-05, "loss": 3.5759, "step": 29800 }, { "epoch": 0.0023296, "grad_norm": 0.9834818243980408, "learning_rate": 1.9722765534520983e-05, "loss": 3.8314, "step": 29810 }, { "epoch": 0.0023552, "grad_norm": 1.0158599615097046, "learning_rate": 1.972257720157121e-05, "loss": 3.9172, "step": 29820 }, { "epoch": 0.0023808, "grad_norm": 0.7836599946022034, "learning_rate": 1.9722388805573104e-05, "loss": 3.7175, "step": 29830 }, { "epoch": 0.0024064, "grad_norm": 1.0060549974441528, "learning_rate": 1.9722200346527893e-05, "loss": 3.4449, "step": 29840 }, { "epoch": 0.002432, "grad_norm": 1.040461778640747, "learning_rate": 1.9722011824436797e-05, "loss": 3.5866, "step": 29850 }, { "epoch": 0.0024576, "grad_norm": 0.8963359594345093, "learning_rate": 1.972182323930104e-05, "loss": 3.5991, "step": 29860 }, { "epoch": 0.0024832, "grad_norm": 1.0157196521759033, "learning_rate": 1.9721634591121838e-05, "loss": 3.4822, "step": 29870 }, { "epoch": 0.0025088, "grad_norm": 0.9207935333251953, "learning_rate": 1.972144587990042e-05, "loss": 3.7576, "step": 29880 }, { "epoch": 0.0025344, "grad_norm": 0.851226806640625, "learning_rate": 1.9721257105638015e-05, "loss": 3.9024, "step": 29890 }, { "epoch": 0.00256, "grad_norm": 1.034816861152649, "learning_rate": 1.9721068268335835e-05, "loss": 3.6995, "step": 29900 }, { "epoch": 0.0025856, "grad_norm": 0.8085922598838806, "learning_rate": 1.9720879367995116e-05, "loss": 3.5185, "step": 29910 }, { "epoch": 0.0026112, "grad_norm": 1.0203967094421387, "learning_rate": 1.9720690404617073e-05, "loss": 3.7142, "step": 29920 }, { "epoch": 0.0026368, "grad_norm": 1.3052736520767212, "learning_rate": 1.972050137820294e-05, "loss": 3.4353, "step": 29930 }, { "epoch": 0.0026624, "grad_norm": 0.8363834023475647, "learning_rate": 1.972031228875394e-05, "loss": 3.5122, "step": 29940 }, { "epoch": 0.002688, "grad_norm": 0.810799241065979, "learning_rate": 1.972012313627129e-05, "loss": 3.6143, "step": 29950 }, { "epoch": 0.0027136, "grad_norm": 0.8849419951438904, "learning_rate": 1.9719933920756233e-05, "loss": 3.5142, "step": 29960 }, { "epoch": 0.0027392, "grad_norm": 0.9568013548851013, "learning_rate": 1.9719744642209986e-05, "loss": 3.7198, "step": 29970 }, { "epoch": 0.0027648, "grad_norm": 0.7661420106887817, "learning_rate": 1.9719555300633776e-05, "loss": 3.4302, "step": 29980 }, { "epoch": 0.0027904, "grad_norm": 1.2140839099884033, "learning_rate": 1.9719365896028833e-05, "loss": 3.5232, "step": 29990 }, { "epoch": 0.002816, "grad_norm": 0.9050735831260681, "learning_rate": 1.9719176428396386e-05, "loss": 3.6408, "step": 30000 }, { "epoch": 0.0028416, "grad_norm": 1.0931711196899414, "learning_rate": 1.9718986897737665e-05, "loss": 3.5953, "step": 30010 }, { "epoch": 0.0028672, "grad_norm": 1.2248711585998535, "learning_rate": 1.9718797304053895e-05, "loss": 3.6435, "step": 30020 }, { "epoch": 0.0028928, "grad_norm": 1.6157785654067993, "learning_rate": 1.971860764734631e-05, "loss": 3.7908, "step": 30030 }, { "epoch": 0.0029184, "grad_norm": 0.9990795850753784, "learning_rate": 1.9718417927616132e-05, "loss": 3.7628, "step": 30040 }, { "epoch": 0.002944, "grad_norm": 0.8969384431838989, "learning_rate": 1.9718228144864603e-05, "loss": 3.7144, "step": 30050 }, { "epoch": 0.0029696, "grad_norm": 0.9512036442756653, "learning_rate": 1.971803829909294e-05, "loss": 3.4977, "step": 30060 }, { "epoch": 0.0029952, "grad_norm": 1.0451701879501343, "learning_rate": 1.9717848390302388e-05, "loss": 3.6272, "step": 30070 }, { "epoch": 0.0030208, "grad_norm": 0.7959398031234741, "learning_rate": 1.971765841849417e-05, "loss": 3.6104, "step": 30080 }, { "epoch": 0.0030464, "grad_norm": 0.9229720830917358, "learning_rate": 1.9717468383669514e-05, "loss": 3.667, "step": 30090 }, { "epoch": 0.003072, "grad_norm": 0.8100448846817017, "learning_rate": 1.9717278285829667e-05, "loss": 3.6479, "step": 30100 }, { "epoch": 0.0030976, "grad_norm": 1.3358234167099, "learning_rate": 1.971708812497585e-05, "loss": 3.72, "step": 30110 }, { "epoch": 0.0031232, "grad_norm": 0.8572596907615662, "learning_rate": 1.97168979011093e-05, "loss": 3.8259, "step": 30120 }, { "epoch": 0.0031488, "grad_norm": 0.8366508483886719, "learning_rate": 1.9716707614231244e-05, "loss": 3.5639, "step": 30130 }, { "epoch": 0.0031744, "grad_norm": 2.849576473236084, "learning_rate": 1.9716517264342926e-05, "loss": 4.069, "step": 30140 }, { "epoch": 0.0032, "grad_norm": 0.9116349816322327, "learning_rate": 1.9716326851445573e-05, "loss": 4.1499, "step": 30150 }, { "epoch": 0.0032256, "grad_norm": 0.7907918691635132, "learning_rate": 1.9716136375540423e-05, "loss": 3.8103, "step": 30160 }, { "epoch": 0.0032512, "grad_norm": 0.7408362030982971, "learning_rate": 1.9715945836628715e-05, "loss": 3.7473, "step": 30170 }, { "epoch": 0.0032768, "grad_norm": 0.9934759736061096, "learning_rate": 1.9715755234711676e-05, "loss": 3.5829, "step": 30180 }, { "epoch": 0.0033024, "grad_norm": 0.8161402940750122, "learning_rate": 1.971556456979055e-05, "loss": 3.7433, "step": 30190 }, { "epoch": 0.003328, "grad_norm": 1.047668695449829, "learning_rate": 1.971537384186657e-05, "loss": 3.6289, "step": 30200 }, { "epoch": 0.0033536, "grad_norm": 0.9014951586723328, "learning_rate": 1.971518305094097e-05, "loss": 3.6389, "step": 30210 }, { "epoch": 0.0033792, "grad_norm": 1.2322276830673218, "learning_rate": 1.971499219701499e-05, "loss": 3.6602, "step": 30220 }, { "epoch": 0.0034048, "grad_norm": 1.0003584623336792, "learning_rate": 1.9714801280089868e-05, "loss": 3.8656, "step": 30230 }, { "epoch": 0.0034304, "grad_norm": 1.0179678201675415, "learning_rate": 1.971461030016684e-05, "loss": 3.83, "step": 30240 }, { "epoch": 0.003456, "grad_norm": 1.4509203433990479, "learning_rate": 1.971441925724715e-05, "loss": 3.8639, "step": 30250 }, { "epoch": 0.0034816, "grad_norm": 0.7357967495918274, "learning_rate": 1.971422815133203e-05, "loss": 3.7397, "step": 30260 }, { "epoch": 0.0035072, "grad_norm": 0.79340660572052, "learning_rate": 1.971403698242272e-05, "loss": 3.5765, "step": 30270 }, { "epoch": 0.0035328, "grad_norm": 0.7425922155380249, "learning_rate": 1.9713845750520466e-05, "loss": 3.6786, "step": 30280 }, { "epoch": 0.0035584, "grad_norm": 2.4944257736206055, "learning_rate": 1.9713654455626505e-05, "loss": 3.6947, "step": 30290 }, { "epoch": 0.003584, "grad_norm": 0.8844621777534485, "learning_rate": 1.9713463097742075e-05, "loss": 4.0543, "step": 30300 }, { "epoch": 0.0036096, "grad_norm": 0.9110368490219116, "learning_rate": 1.9713271676868416e-05, "loss": 3.6331, "step": 30310 }, { "epoch": 0.0036352, "grad_norm": 0.8885475993156433, "learning_rate": 1.9713080193006772e-05, "loss": 3.5827, "step": 30320 }, { "epoch": 0.0036608, "grad_norm": 0.9458448886871338, "learning_rate": 1.9712888646158385e-05, "loss": 3.6583, "step": 30330 }, { "epoch": 0.0036864, "grad_norm": 0.7318917512893677, "learning_rate": 1.9712697036324497e-05, "loss": 3.7091, "step": 30340 }, { "epoch": 0.003712, "grad_norm": 0.8376695513725281, "learning_rate": 1.9712505363506353e-05, "loss": 3.5545, "step": 30350 }, { "epoch": 0.0037376, "grad_norm": 0.7324709892272949, "learning_rate": 1.971231362770519e-05, "loss": 4.1019, "step": 30360 }, { "epoch": 0.0037632, "grad_norm": 0.9814066886901855, "learning_rate": 1.9712121828922256e-05, "loss": 3.4452, "step": 30370 }, { "epoch": 0.0037888, "grad_norm": 0.8543195128440857, "learning_rate": 1.971192996715879e-05, "loss": 3.9708, "step": 30380 }, { "epoch": 0.0038144, "grad_norm": 0.8500795960426331, "learning_rate": 1.971173804241604e-05, "loss": 3.481, "step": 30390 }, { "epoch": 0.00384, "grad_norm": 1.0972888469696045, "learning_rate": 1.9711546054695254e-05, "loss": 3.3821, "step": 30400 }, { "epoch": 0.0038656, "grad_norm": 0.8807580471038818, "learning_rate": 1.9711354003997667e-05, "loss": 3.5539, "step": 30410 }, { "epoch": 0.0038912, "grad_norm": 0.9438031315803528, "learning_rate": 1.9711161890324532e-05, "loss": 3.503, "step": 30420 }, { "epoch": 0.0039168, "grad_norm": 1.1865733861923218, "learning_rate": 1.9710969713677093e-05, "loss": 3.6811, "step": 30430 }, { "epoch": 0.0039424, "grad_norm": 13.535170555114746, "learning_rate": 1.97107774740566e-05, "loss": 3.6467, "step": 30440 }, { "epoch": 0.003968, "grad_norm": 1.3967015743255615, "learning_rate": 1.9710585171464293e-05, "loss": 4.0642, "step": 30450 }, { "epoch": 0.0039936, "grad_norm": 0.8263975381851196, "learning_rate": 1.9710392805901422e-05, "loss": 3.5979, "step": 30460 }, { "epoch": 0.0040192, "grad_norm": 0.8945654034614563, "learning_rate": 1.9710200377369234e-05, "loss": 3.6866, "step": 30470 }, { "epoch": 0.0040448, "grad_norm": 1.0943915843963623, "learning_rate": 1.9710007885868976e-05, "loss": 3.4099, "step": 30480 }, { "epoch": 0.0040704, "grad_norm": 0.8231286406517029, "learning_rate": 1.97098153314019e-05, "loss": 3.5862, "step": 30490 }, { "epoch": 0.004096, "grad_norm": 0.7610021233558655, "learning_rate": 1.970962271396925e-05, "loss": 3.704, "step": 30500 }, { "epoch": 0.0041216, "grad_norm": 1.5207033157348633, "learning_rate": 1.970943003357228e-05, "loss": 3.6102, "step": 30510 }, { "epoch": 0.0041472, "grad_norm": 1.3027619123458862, "learning_rate": 1.9709237290212233e-05, "loss": 3.6072, "step": 30520 }, { "epoch": 0.0041728, "grad_norm": 0.8976724743843079, "learning_rate": 1.9709044483890364e-05, "loss": 4.0736, "step": 30530 }, { "epoch": 0.0041984, "grad_norm": 0.8159211874008179, "learning_rate": 1.9708851614607926e-05, "loss": 3.6912, "step": 30540 }, { "epoch": 0.004224, "grad_norm": 0.7295176982879639, "learning_rate": 1.970865868236616e-05, "loss": 3.5223, "step": 30550 }, { "epoch": 0.0042496, "grad_norm": 0.9746666550636292, "learning_rate": 1.9708465687166324e-05, "loss": 3.7291, "step": 30560 }, { "epoch": 0.0042752, "grad_norm": 1.1087285280227661, "learning_rate": 1.9708272629009668e-05, "loss": 3.4992, "step": 30570 }, { "epoch": 0.0043008, "grad_norm": 1.0083417892456055, "learning_rate": 1.9708079507897446e-05, "loss": 3.6448, "step": 30580 }, { "epoch": 0.0043264, "grad_norm": 0.9560156464576721, "learning_rate": 1.9707886323830907e-05, "loss": 3.5479, "step": 30590 }, { "epoch": 0.004352, "grad_norm": 1.390923023223877, "learning_rate": 1.9707693076811308e-05, "loss": 3.7231, "step": 30600 }, { "epoch": 0.0043776, "grad_norm": 0.8419608473777771, "learning_rate": 1.9707499766839894e-05, "loss": 3.5328, "step": 30610 }, { "epoch": 0.0044032, "grad_norm": 0.916790783405304, "learning_rate": 1.970730639391793e-05, "loss": 3.612, "step": 30620 }, { "epoch": 0.0044288, "grad_norm": 0.9404973387718201, "learning_rate": 1.970711295804666e-05, "loss": 3.6973, "step": 30630 }, { "epoch": 0.0044544, "grad_norm": 0.8470897674560547, "learning_rate": 1.9706919459227343e-05, "loss": 3.6462, "step": 30640 }, { "epoch": 0.00448, "grad_norm": 0.8268787264823914, "learning_rate": 1.9706725897461237e-05, "loss": 3.4581, "step": 30650 }, { "epoch": 0.0045056, "grad_norm": 0.8927762508392334, "learning_rate": 1.970653227274959e-05, "loss": 3.6962, "step": 30660 }, { "epoch": 0.0045312, "grad_norm": 0.9168525338172913, "learning_rate": 1.970633858509366e-05, "loss": 3.5299, "step": 30670 }, { "epoch": 0.0045568, "grad_norm": 1.0355300903320312, "learning_rate": 1.9706144834494703e-05, "loss": 3.7738, "step": 30680 }, { "epoch": 0.0045824, "grad_norm": 0.8575823307037354, "learning_rate": 1.970595102095398e-05, "loss": 3.6513, "step": 30690 }, { "epoch": 0.004608, "grad_norm": 0.8114187717437744, "learning_rate": 1.970575714447274e-05, "loss": 3.7033, "step": 30700 }, { "epoch": 0.0046336, "grad_norm": 1.173700213432312, "learning_rate": 1.9705563205052245e-05, "loss": 3.5994, "step": 30710 }, { "epoch": 0.0046592, "grad_norm": 1.045298457145691, "learning_rate": 1.9705369202693756e-05, "loss": 3.4626, "step": 30720 }, { "epoch": 0.0046848, "grad_norm": 1.5597785711288452, "learning_rate": 1.9705175137398523e-05, "loss": 3.3878, "step": 30730 }, { "epoch": 0.0047104, "grad_norm": 0.8237354755401611, "learning_rate": 1.970498100916781e-05, "loss": 3.64, "step": 30740 }, { "epoch": 0.004736, "grad_norm": 1.0006662607192993, "learning_rate": 1.9704786818002875e-05, "loss": 3.5775, "step": 30750 }, { "epoch": 0.0047616, "grad_norm": 0.9305443167686462, "learning_rate": 1.9704592563904977e-05, "loss": 3.4759, "step": 30760 }, { "epoch": 0.0047872, "grad_norm": 0.8256446719169617, "learning_rate": 1.9704398246875374e-05, "loss": 3.3445, "step": 30770 }, { "epoch": 0.0048128, "grad_norm": 2.09421968460083, "learning_rate": 1.970420386691533e-05, "loss": 3.891, "step": 30780 }, { "epoch": 0.0048384, "grad_norm": 1.1371363401412964, "learning_rate": 1.97040094240261e-05, "loss": 3.5154, "step": 30790 }, { "epoch": 0.004864, "grad_norm": 2.1658802032470703, "learning_rate": 1.970381491820895e-05, "loss": 3.5033, "step": 30800 }, { "epoch": 0.0048896, "grad_norm": 0.8112155199050903, "learning_rate": 1.9703620349465137e-05, "loss": 3.7122, "step": 30810 }, { "epoch": 0.0049152, "grad_norm": 0.9048958420753479, "learning_rate": 1.9703425717795926e-05, "loss": 3.6864, "step": 30820 }, { "epoch": 0.0049408, "grad_norm": 0.7237329483032227, "learning_rate": 1.970323102320258e-05, "loss": 3.836, "step": 30830 }, { "epoch": 0.0049664, "grad_norm": 0.9984995722770691, "learning_rate": 1.9703036265686357e-05, "loss": 3.6487, "step": 30840 }, { "epoch": 0.004992, "grad_norm": 3.700568437576294, "learning_rate": 1.9702841445248526e-05, "loss": 4.9457, "step": 30850 }, { "epoch": 0.0050176, "grad_norm": 1.0347309112548828, "learning_rate": 1.9702646561890343e-05, "loss": 3.4064, "step": 30860 }, { "epoch": 0.0050432, "grad_norm": 0.9269953370094299, "learning_rate": 1.970245161561308e-05, "loss": 3.6693, "step": 30870 }, { "epoch": 0.0050688, "grad_norm": 0.8981368541717529, "learning_rate": 1.970225660641799e-05, "loss": 3.8798, "step": 30880 }, { "epoch": 0.0050944, "grad_norm": 0.7780205011367798, "learning_rate": 1.9702061534306353e-05, "loss": 3.6695, "step": 30890 }, { "epoch": 0.00512, "grad_norm": 0.8890566229820251, "learning_rate": 1.9701866399279418e-05, "loss": 3.6135, "step": 30900 }, { "epoch": 0.0051456, "grad_norm": 0.7338903546333313, "learning_rate": 1.970167120133846e-05, "loss": 3.7319, "step": 30910 }, { "epoch": 0.0051712, "grad_norm": 0.805595338344574, "learning_rate": 1.9701475940484743e-05, "loss": 3.3937, "step": 30920 }, { "epoch": 0.0051968, "grad_norm": 1.2691140174865723, "learning_rate": 1.9701280616719535e-05, "loss": 3.605, "step": 30930 }, { "epoch": 0.0052224, "grad_norm": 0.7891610860824585, "learning_rate": 1.97010852300441e-05, "loss": 3.4665, "step": 30940 }, { "epoch": 0.005248, "grad_norm": 0.7796364426612854, "learning_rate": 1.97008897804597e-05, "loss": 3.5863, "step": 30950 }, { "epoch": 0.0052736, "grad_norm": 1.357865333557129, "learning_rate": 1.9700694267967617e-05, "loss": 3.7705, "step": 30960 }, { "epoch": 0.0052992, "grad_norm": 1.2066031694412231, "learning_rate": 1.9700498692569107e-05, "loss": 3.5673, "step": 30970 }, { "epoch": 0.0053248, "grad_norm": 0.8494134545326233, "learning_rate": 1.9700303054265437e-05, "loss": 3.6721, "step": 30980 }, { "epoch": 0.0053504, "grad_norm": 1.1676945686340332, "learning_rate": 1.970010735305788e-05, "loss": 3.7312, "step": 30990 }, { "epoch": 0.005376, "grad_norm": 0.9177849888801575, "learning_rate": 1.9699911588947708e-05, "loss": 3.6175, "step": 31000 }, { "epoch": 0.0054016, "grad_norm": 0.8014964461326599, "learning_rate": 1.9699715761936184e-05, "loss": 3.5033, "step": 31010 }, { "epoch": 0.0054272, "grad_norm": 1.7166807651519775, "learning_rate": 1.9699519872024585e-05, "loss": 3.9834, "step": 31020 }, { "epoch": 0.0054528, "grad_norm": 0.7953618168830872, "learning_rate": 1.9699323919214177e-05, "loss": 3.6564, "step": 31030 }, { "epoch": 0.0054784, "grad_norm": 1.1500803232192993, "learning_rate": 1.969912790350623e-05, "loss": 4.1343, "step": 31040 }, { "epoch": 0.005504, "grad_norm": 0.9052059650421143, "learning_rate": 1.9698931824902013e-05, "loss": 3.946, "step": 31050 }, { "epoch": 0.0055296, "grad_norm": 1.1805975437164307, "learning_rate": 1.9698735683402805e-05, "loss": 3.4722, "step": 31060 }, { "epoch": 0.0055552, "grad_norm": 1.372495412826538, "learning_rate": 1.9698539479009872e-05, "loss": 3.7025, "step": 31070 }, { "epoch": 0.0055808, "grad_norm": 0.7626193761825562, "learning_rate": 1.969834321172449e-05, "loss": 3.761, "step": 31080 }, { "epoch": 0.0056064, "grad_norm": 1.1581231355667114, "learning_rate": 1.9698146881547927e-05, "loss": 3.983, "step": 31090 }, { "epoch": 0.005632, "grad_norm": 0.7420107126235962, "learning_rate": 1.969795048848146e-05, "loss": 3.6018, "step": 31100 }, { "epoch": 0.0056576, "grad_norm": 0.7431421875953674, "learning_rate": 1.9697754032526364e-05, "loss": 3.5343, "step": 31110 }, { "epoch": 0.0056832, "grad_norm": 1.02370023727417, "learning_rate": 1.969755751368391e-05, "loss": 3.7658, "step": 31120 }, { "epoch": 0.0057088, "grad_norm": 0.9967777729034424, "learning_rate": 1.969736093195537e-05, "loss": 3.601, "step": 31130 }, { "epoch": 0.0057344, "grad_norm": 1.1386473178863525, "learning_rate": 1.9697164287342022e-05, "loss": 3.5698, "step": 31140 }, { "epoch": 0.00576, "grad_norm": 0.7715117931365967, "learning_rate": 1.969696757984514e-05, "loss": 3.774, "step": 31150 }, { "epoch": 0.0057856, "grad_norm": 0.8684498071670532, "learning_rate": 1.9696770809466e-05, "loss": 3.6358, "step": 31160 }, { "epoch": 0.0058112, "grad_norm": 0.9714090824127197, "learning_rate": 1.9696573976205885e-05, "loss": 3.4796, "step": 31170 }, { "epoch": 0.0058368, "grad_norm": 0.7954586148262024, "learning_rate": 1.9696377080066057e-05, "loss": 3.8542, "step": 31180 }, { "epoch": 0.0058624, "grad_norm": 0.895664632320404, "learning_rate": 1.9696180121047806e-05, "loss": 3.7932, "step": 31190 }, { "epoch": 0.005888, "grad_norm": 0.8530347943305969, "learning_rate": 1.96959830991524e-05, "loss": 3.4049, "step": 31200 }, { "epoch": 0.0059136, "grad_norm": 0.7500506639480591, "learning_rate": 1.9695786014381122e-05, "loss": 3.6341, "step": 31210 }, { "epoch": 0.0059392, "grad_norm": 0.8608022928237915, "learning_rate": 1.9695588866735247e-05, "loss": 3.6162, "step": 31220 }, { "epoch": 0.0059648, "grad_norm": 1.262508749961853, "learning_rate": 1.969539165621606e-05, "loss": 3.6038, "step": 31230 }, { "epoch": 0.0059904, "grad_norm": 1.0747876167297363, "learning_rate": 1.969519438282483e-05, "loss": 3.7434, "step": 31240 }, { "epoch": 0.006016, "grad_norm": 0.893563449382782, "learning_rate": 1.9694997046562843e-05, "loss": 3.459, "step": 31250 }, { "epoch": 0.0060416, "grad_norm": 1.190782070159912, "learning_rate": 1.969479964743138e-05, "loss": 3.7388, "step": 31260 }, { "epoch": 0.0060672, "grad_norm": 1.0019135475158691, "learning_rate": 1.969460218543171e-05, "loss": 3.9169, "step": 31270 }, { "epoch": 0.0060928, "grad_norm": 1.1004853248596191, "learning_rate": 1.9694404660565125e-05, "loss": 3.7372, "step": 31280 }, { "epoch": 0.0061184, "grad_norm": 0.7360408306121826, "learning_rate": 1.96942070728329e-05, "loss": 3.8103, "step": 31290 }, { "epoch": 0.006144, "grad_norm": 0.6722855567932129, "learning_rate": 1.9694009422236325e-05, "loss": 3.6745, "step": 31300 }, { "epoch": 0.0061696, "grad_norm": 0.8602474927902222, "learning_rate": 1.969381170877667e-05, "loss": 3.5626, "step": 31310 }, { "epoch": 0.0061952, "grad_norm": 1.1627458333969116, "learning_rate": 1.9693613932455223e-05, "loss": 3.8448, "step": 31320 }, { "epoch": 0.0062208, "grad_norm": 0.9760617613792419, "learning_rate": 1.969341609327327e-05, "loss": 4.0313, "step": 31330 }, { "epoch": 0.0062464, "grad_norm": 0.8323933482170105, "learning_rate": 1.9693218191232083e-05, "loss": 3.6409, "step": 31340 }, { "epoch": 0.006272, "grad_norm": 0.7560579776763916, "learning_rate": 1.9693020226332957e-05, "loss": 3.4641, "step": 31350 }, { "epoch": 0.0062976, "grad_norm": 1.098282814025879, "learning_rate": 1.969282219857717e-05, "loss": 3.8339, "step": 31360 }, { "epoch": 0.0063232, "grad_norm": 0.7869873642921448, "learning_rate": 1.969262410796601e-05, "loss": 3.7347, "step": 31370 }, { "epoch": 0.0063488, "grad_norm": 0.9253185391426086, "learning_rate": 1.969242595450075e-05, "loss": 3.8381, "step": 31380 }, { "epoch": 0.0063744, "grad_norm": 0.8417391180992126, "learning_rate": 1.9692227738182693e-05, "loss": 3.7984, "step": 31390 }, { "epoch": 0.0064, "grad_norm": 0.9681099057197571, "learning_rate": 1.969202945901311e-05, "loss": 3.4542, "step": 31400 }, { "epoch": 0.0064256, "grad_norm": 0.780180811882019, "learning_rate": 1.9691831116993293e-05, "loss": 3.664, "step": 31410 }, { "epoch": 0.0064512, "grad_norm": 0.7596474885940552, "learning_rate": 1.9691632712124526e-05, "loss": 3.6182, "step": 31420 }, { "epoch": 0.0064768, "grad_norm": 0.8337158560752869, "learning_rate": 1.96914342444081e-05, "loss": 3.7257, "step": 31430 }, { "epoch": 0.0065024, "grad_norm": 0.9659503698348999, "learning_rate": 1.9691235713845295e-05, "loss": 3.6677, "step": 31440 }, { "epoch": 0.006528, "grad_norm": 1.817419171333313, "learning_rate": 1.9691037120437402e-05, "loss": 3.8282, "step": 31450 }, { "epoch": 0.0065536, "grad_norm": 1.1092267036437988, "learning_rate": 1.9690838464185712e-05, "loss": 3.7668, "step": 31460 }, { "epoch": 0.0065792, "grad_norm": 0.8917662501335144, "learning_rate": 1.9690639745091506e-05, "loss": 3.6601, "step": 31470 }, { "epoch": 0.0066048, "grad_norm": 0.8195468783378601, "learning_rate": 1.969044096315608e-05, "loss": 3.9525, "step": 31480 }, { "epoch": 0.0066304, "grad_norm": 1.2093379497528076, "learning_rate": 1.9690242118380716e-05, "loss": 3.6308, "step": 31490 }, { "epoch": 0.006656, "grad_norm": 0.9162184596061707, "learning_rate": 1.969004321076671e-05, "loss": 3.57, "step": 31500 }, { "epoch": 0.0066816, "grad_norm": 1.7837611436843872, "learning_rate": 1.968984424031535e-05, "loss": 3.6374, "step": 31510 }, { "epoch": 0.0067072, "grad_norm": 1.1884442567825317, "learning_rate": 1.9689645207027922e-05, "loss": 3.6432, "step": 31520 }, { "epoch": 0.0067328, "grad_norm": 0.8470492362976074, "learning_rate": 1.9689446110905724e-05, "loss": 3.6679, "step": 31530 }, { "epoch": 0.0067584, "grad_norm": 3.241162061691284, "learning_rate": 1.9689246951950043e-05, "loss": 3.6802, "step": 31540 }, { "epoch": 0.006784, "grad_norm": 0.733995258808136, "learning_rate": 1.9689047730162166e-05, "loss": 3.7003, "step": 31550 }, { "epoch": 0.0068096, "grad_norm": 0.8016207814216614, "learning_rate": 1.9688848445543394e-05, "loss": 3.6681, "step": 31560 }, { "epoch": 0.0068352, "grad_norm": 0.9235904812812805, "learning_rate": 1.968864909809501e-05, "loss": 3.6301, "step": 31570 }, { "epoch": 0.0068608, "grad_norm": 0.8874274492263794, "learning_rate": 1.9688449687818316e-05, "loss": 3.6009, "step": 31580 }, { "epoch": 0.0068864, "grad_norm": 0.9275990128517151, "learning_rate": 1.96882502147146e-05, "loss": 3.5962, "step": 31590 }, { "epoch": 0.006912, "grad_norm": 0.8575594425201416, "learning_rate": 1.9688050678785155e-05, "loss": 3.6009, "step": 31600 }, { "epoch": 0.0069376, "grad_norm": 0.7074971199035645, "learning_rate": 1.968785108003128e-05, "loss": 3.4936, "step": 31610 }, { "epoch": 0.0069632, "grad_norm": 1.0793919563293457, "learning_rate": 1.968765141845426e-05, "loss": 3.7459, "step": 31620 }, { "epoch": 0.0069888, "grad_norm": 0.8202930092811584, "learning_rate": 1.96874516940554e-05, "loss": 3.6587, "step": 31630 }, { "epoch": 0.0070144, "grad_norm": 0.7755850553512573, "learning_rate": 1.9687251906835984e-05, "loss": 3.8985, "step": 31640 }, { "epoch": 0.00704, "grad_norm": 1.2053821086883545, "learning_rate": 1.968705205679732e-05, "loss": 3.5122, "step": 31650 }, { "epoch": 0.0070656, "grad_norm": 1.1179043054580688, "learning_rate": 1.9686852143940695e-05, "loss": 3.4965, "step": 31660 }, { "epoch": 0.0070912, "grad_norm": 0.8844748139381409, "learning_rate": 1.968665216826741e-05, "loss": 3.5637, "step": 31670 }, { "epoch": 0.0071168, "grad_norm": 0.9429481625556946, "learning_rate": 1.968645212977876e-05, "loss": 3.6086, "step": 31680 }, { "epoch": 0.0071424, "grad_norm": 0.913507878780365, "learning_rate": 1.9686252028476042e-05, "loss": 3.738, "step": 31690 }, { "epoch": 0.007168, "grad_norm": 1.4472781419754028, "learning_rate": 1.968605186436055e-05, "loss": 3.5857, "step": 31700 }, { "epoch": 0.0071936, "grad_norm": 1.0717699527740479, "learning_rate": 1.968585163743359e-05, "loss": 3.4525, "step": 31710 }, { "epoch": 0.0072192, "grad_norm": 1.1310391426086426, "learning_rate": 1.9685651347696453e-05, "loss": 3.7288, "step": 31720 }, { "epoch": 0.0072448, "grad_norm": 0.6760324239730835, "learning_rate": 1.9685450995150444e-05, "loss": 3.7378, "step": 31730 }, { "epoch": 0.0072704, "grad_norm": 0.778531551361084, "learning_rate": 1.968525057979686e-05, "loss": 3.5353, "step": 31740 }, { "epoch": 0.007296, "grad_norm": 0.8432754874229431, "learning_rate": 1.9685050101637e-05, "loss": 3.4975, "step": 31750 }, { "epoch": 0.0073216, "grad_norm": 0.7669180035591125, "learning_rate": 1.9684849560672164e-05, "loss": 3.5758, "step": 31760 }, { "epoch": 0.0073472, "grad_norm": 1.129860281944275, "learning_rate": 1.968464895690365e-05, "loss": 3.4008, "step": 31770 }, { "epoch": 0.0073728, "grad_norm": 1.204668402671814, "learning_rate": 1.9684448290332762e-05, "loss": 3.5701, "step": 31780 }, { "epoch": 0.0073984, "grad_norm": 1.030198574066162, "learning_rate": 1.9684247560960802e-05, "loss": 3.7183, "step": 31790 }, { "epoch": 0.007424, "grad_norm": 0.8821504712104797, "learning_rate": 1.968404676878907e-05, "loss": 3.6337, "step": 31800 }, { "epoch": 0.0074496, "grad_norm": 0.9117991328239441, "learning_rate": 1.9683845913818872e-05, "loss": 3.5959, "step": 31810 }, { "epoch": 0.0074752, "grad_norm": 0.8527608513832092, "learning_rate": 1.96836449960515e-05, "loss": 3.5854, "step": 31820 }, { "epoch": 0.0075008, "grad_norm": 0.6885613203048706, "learning_rate": 1.9683444015488267e-05, "loss": 3.7159, "step": 31830 }, { "epoch": 0.0075264, "grad_norm": 1.0054042339324951, "learning_rate": 1.9683242972130474e-05, "loss": 3.5045, "step": 31840 }, { "epoch": 0.007552, "grad_norm": 1.1227303743362427, "learning_rate": 1.9683041865979424e-05, "loss": 3.4157, "step": 31850 }, { "epoch": 0.0075776, "grad_norm": 0.8263418078422546, "learning_rate": 1.968284069703642e-05, "loss": 3.3922, "step": 31860 }, { "epoch": 0.0076032, "grad_norm": 1.1869601011276245, "learning_rate": 1.9682639465302767e-05, "loss": 3.5918, "step": 31870 }, { "epoch": 0.0076288, "grad_norm": 0.7133355140686035, "learning_rate": 1.9682438170779773e-05, "loss": 3.6764, "step": 31880 }, { "epoch": 0.0076544, "grad_norm": 0.8676489591598511, "learning_rate": 1.9682236813468736e-05, "loss": 3.8966, "step": 31890 }, { "epoch": 0.00768, "grad_norm": 0.8610178232192993, "learning_rate": 1.968203539337097e-05, "loss": 3.6708, "step": 31900 }, { "epoch": 0.0077056, "grad_norm": 1.0046731233596802, "learning_rate": 1.9681833910487776e-05, "loss": 3.5689, "step": 31910 }, { "epoch": 0.0077312, "grad_norm": 1.1485934257507324, "learning_rate": 1.968163236482046e-05, "loss": 3.7585, "step": 31920 }, { "epoch": 0.0077568, "grad_norm": 0.7886174917221069, "learning_rate": 1.9681430756370335e-05, "loss": 3.6399, "step": 31930 }, { "epoch": 0.0077824, "grad_norm": 0.993740439414978, "learning_rate": 1.9681229085138707e-05, "loss": 3.7661, "step": 31940 }, { "epoch": 0.007808, "grad_norm": 1.19083571434021, "learning_rate": 1.9681027351126877e-05, "loss": 3.8687, "step": 31950 }, { "epoch": 0.0078336, "grad_norm": 0.8032439351081848, "learning_rate": 1.9680825554336157e-05, "loss": 3.5394, "step": 31960 }, { "epoch": 0.0078592, "grad_norm": 1.1274811029434204, "learning_rate": 1.968062369476786e-05, "loss": 3.6129, "step": 31970 }, { "epoch": 0.0078848, "grad_norm": 0.9838938117027283, "learning_rate": 1.968042177242329e-05, "loss": 3.6594, "step": 31980 }, { "epoch": 0.0079104, "grad_norm": 0.8378875851631165, "learning_rate": 1.9680219787303755e-05, "loss": 3.4268, "step": 31990 }, { "epoch": 0.007936, "grad_norm": 0.7893176078796387, "learning_rate": 1.968001773941057e-05, "loss": 3.6616, "step": 32000 }, { "epoch": 0.0079616, "grad_norm": 0.8490405678749084, "learning_rate": 1.967981562874504e-05, "loss": 3.7901, "step": 32010 }, { "epoch": 0.0079872, "grad_norm": 2.9370245933532715, "learning_rate": 1.967961345530848e-05, "loss": 3.6859, "step": 32020 }, { "epoch": 0.0080128, "grad_norm": 0.7923609614372253, "learning_rate": 1.96794112191022e-05, "loss": 3.5188, "step": 32030 }, { "epoch": 0.0080384, "grad_norm": 0.7823167443275452, "learning_rate": 1.9679208920127508e-05, "loss": 3.5023, "step": 32040 }, { "epoch": 0.008064, "grad_norm": 0.7891188859939575, "learning_rate": 1.9679006558385717e-05, "loss": 3.4887, "step": 32050 }, { "epoch": 0.0080896, "grad_norm": 0.7746818661689758, "learning_rate": 1.9678804133878148e-05, "loss": 3.5226, "step": 32060 }, { "epoch": 0.0081152, "grad_norm": 1.0272256135940552, "learning_rate": 1.96786016466061e-05, "loss": 3.5753, "step": 32070 }, { "epoch": 0.0081408, "grad_norm": 1.0908838510513306, "learning_rate": 1.9678399096570896e-05, "loss": 3.7297, "step": 32080 }, { "epoch": 0.0081664, "grad_norm": 0.8882974982261658, "learning_rate": 1.9678196483773845e-05, "loss": 3.5632, "step": 32090 }, { "epoch": 0.008192, "grad_norm": 1.0824118852615356, "learning_rate": 1.967799380821626e-05, "loss": 3.7346, "step": 32100 }, { "epoch": 0.0082176, "grad_norm": 1.0895320177078247, "learning_rate": 1.967779106989946e-05, "loss": 3.6322, "step": 32110 }, { "epoch": 0.0082432, "grad_norm": 0.8757520318031311, "learning_rate": 1.9677588268824757e-05, "loss": 3.8073, "step": 32120 }, { "epoch": 0.0082688, "grad_norm": 1.0351595878601074, "learning_rate": 1.9677385404993467e-05, "loss": 3.8533, "step": 32130 }, { "epoch": 0.0082944, "grad_norm": 2.6525330543518066, "learning_rate": 1.9677182478406905e-05, "loss": 3.7856, "step": 32140 }, { "epoch": 0.00832, "grad_norm": 2.9656524658203125, "learning_rate": 1.9676979489066382e-05, "loss": 3.6369, "step": 32150 }, { "epoch": 0.0083456, "grad_norm": 1.520920753479004, "learning_rate": 1.967677643697322e-05, "loss": 3.4752, "step": 32160 }, { "epoch": 0.0083712, "grad_norm": 0.9989417195320129, "learning_rate": 1.9676573322128737e-05, "loss": 3.5961, "step": 32170 }, { "epoch": 0.0083968, "grad_norm": 1.2838554382324219, "learning_rate": 1.9676370144534248e-05, "loss": 3.8403, "step": 32180 }, { "epoch": 0.0084224, "grad_norm": 1.7205960750579834, "learning_rate": 1.967616690419107e-05, "loss": 3.6197, "step": 32190 }, { "epoch": 0.008448, "grad_norm": 0.8264669179916382, "learning_rate": 1.967596360110052e-05, "loss": 3.701, "step": 32200 }, { "epoch": 0.0084736, "grad_norm": 0.8827752470970154, "learning_rate": 1.9675760235263916e-05, "loss": 3.4299, "step": 32210 }, { "epoch": 0.0084992, "grad_norm": 1.004047155380249, "learning_rate": 1.967555680668258e-05, "loss": 3.6768, "step": 32220 }, { "epoch": 0.0085248, "grad_norm": 1.4823219776153564, "learning_rate": 1.9675353315357833e-05, "loss": 3.752, "step": 32230 }, { "epoch": 0.0085504, "grad_norm": 0.873120129108429, "learning_rate": 1.9675149761290985e-05, "loss": 3.4077, "step": 32240 }, { "epoch": 0.008576, "grad_norm": 0.8717175126075745, "learning_rate": 1.9674946144483366e-05, "loss": 3.5257, "step": 32250 }, { "epoch": 0.0086016, "grad_norm": 1.243452548980713, "learning_rate": 1.967474246493629e-05, "loss": 3.5146, "step": 32260 }, { "epoch": 0.0086272, "grad_norm": 1.1123137474060059, "learning_rate": 1.9674538722651082e-05, "loss": 3.5573, "step": 32270 }, { "epoch": 0.0086528, "grad_norm": 0.8600099682807922, "learning_rate": 1.9674334917629063e-05, "loss": 3.4949, "step": 32280 }, { "epoch": 0.0086784, "grad_norm": 0.8926434516906738, "learning_rate": 1.967413104987155e-05, "loss": 3.728, "step": 32290 }, { "epoch": 0.008704, "grad_norm": 0.7720945477485657, "learning_rate": 1.9673927119379874e-05, "loss": 3.7079, "step": 32300 }, { "epoch": 0.0087296, "grad_norm": 1.29965078830719, "learning_rate": 1.9673723126155346e-05, "loss": 3.7803, "step": 32310 }, { "epoch": 0.0087552, "grad_norm": 0.8471423983573914, "learning_rate": 1.9673519070199295e-05, "loss": 3.6338, "step": 32320 }, { "epoch": 0.0087808, "grad_norm": 0.9163434505462646, "learning_rate": 1.9673314951513044e-05, "loss": 3.4238, "step": 32330 }, { "epoch": 0.0088064, "grad_norm": 0.9780977964401245, "learning_rate": 1.9673110770097917e-05, "loss": 3.5932, "step": 32340 }, { "epoch": 0.008832, "grad_norm": 0.8980678915977478, "learning_rate": 1.967290652595524e-05, "loss": 3.6557, "step": 32350 }, { "epoch": 0.0088576, "grad_norm": 0.8867409229278564, "learning_rate": 1.967270221908633e-05, "loss": 3.4568, "step": 32360 }, { "epoch": 0.0088832, "grad_norm": 0.9275981783866882, "learning_rate": 1.967249784949252e-05, "loss": 3.6341, "step": 32370 }, { "epoch": 0.0089088, "grad_norm": 1.0816458463668823, "learning_rate": 1.9672293417175133e-05, "loss": 3.7325, "step": 32380 }, { "epoch": 0.0089344, "grad_norm": 0.7571274638175964, "learning_rate": 1.9672088922135494e-05, "loss": 3.7515, "step": 32390 }, { "epoch": 0.00896, "grad_norm": 1.1262835264205933, "learning_rate": 1.9671884364374926e-05, "loss": 3.9204, "step": 32400 }, { "epoch": 0.0089856, "grad_norm": 0.7211765050888062, "learning_rate": 1.967167974389476e-05, "loss": 3.5668, "step": 32410 }, { "epoch": 0.0090112, "grad_norm": 1.101572036743164, "learning_rate": 1.967147506069632e-05, "loss": 3.566, "step": 32420 }, { "epoch": 0.0090368, "grad_norm": 1.0772581100463867, "learning_rate": 1.9671270314780937e-05, "loss": 4.098, "step": 32430 }, { "epoch": 0.0090624, "grad_norm": 0.8489611148834229, "learning_rate": 1.9671065506149935e-05, "loss": 3.8024, "step": 32440 }, { "epoch": 0.009088, "grad_norm": 0.9283691644668579, "learning_rate": 1.9670860634804643e-05, "loss": 3.4068, "step": 32450 }, { "epoch": 0.0091136, "grad_norm": 0.870420515537262, "learning_rate": 1.9670655700746388e-05, "loss": 3.8083, "step": 32460 }, { "epoch": 0.0091392, "grad_norm": 0.8768624067306519, "learning_rate": 1.9670450703976503e-05, "loss": 3.6242, "step": 32470 }, { "epoch": 0.0091648, "grad_norm": 1.5579551458358765, "learning_rate": 1.9670245644496314e-05, "loss": 3.4431, "step": 32480 }, { "epoch": 0.0091904, "grad_norm": 0.8054280877113342, "learning_rate": 1.9670040522307155e-05, "loss": 3.6268, "step": 32490 }, { "epoch": 0.009216, "grad_norm": 0.9953359961509705, "learning_rate": 1.9669835337410352e-05, "loss": 3.7642, "step": 32500 }, { "epoch": 0.0092416, "grad_norm": 0.7877715229988098, "learning_rate": 1.966963008980724e-05, "loss": 3.6337, "step": 32510 }, { "epoch": 0.0092672, "grad_norm": 0.7734318971633911, "learning_rate": 1.966942477949914e-05, "loss": 3.6794, "step": 32520 }, { "epoch": 0.0092928, "grad_norm": 0.9780212044715881, "learning_rate": 1.966921940648739e-05, "loss": 3.5915, "step": 32530 }, { "epoch": 0.0093184, "grad_norm": 0.7168958187103271, "learning_rate": 1.966901397077333e-05, "loss": 3.4955, "step": 32540 }, { "epoch": 0.009344, "grad_norm": 0.9491490125656128, "learning_rate": 1.966880847235828e-05, "loss": 3.6907, "step": 32550 }, { "epoch": 0.0093696, "grad_norm": 0.8746066093444824, "learning_rate": 1.9668602911243573e-05, "loss": 3.7158, "step": 32560 }, { "epoch": 0.0093952, "grad_norm": 1.053406834602356, "learning_rate": 1.966839728743055e-05, "loss": 3.6454, "step": 32570 }, { "epoch": 0.0094208, "grad_norm": 0.8406070470809937, "learning_rate": 1.966819160092054e-05, "loss": 3.6553, "step": 32580 }, { "epoch": 0.0094464, "grad_norm": 0.9226546287536621, "learning_rate": 1.9667985851714875e-05, "loss": 3.7509, "step": 32590 }, { "epoch": 0.009472, "grad_norm": 0.7868872284889221, "learning_rate": 1.9667780039814892e-05, "loss": 3.5017, "step": 32600 }, { "epoch": 0.0094976, "grad_norm": 1.2729601860046387, "learning_rate": 1.966757416522192e-05, "loss": 3.7073, "step": 32610 }, { "epoch": 0.0095232, "grad_norm": 1.1075797080993652, "learning_rate": 1.9667368227937308e-05, "loss": 3.5003, "step": 32620 }, { "epoch": 0.0095488, "grad_norm": 0.886673092842102, "learning_rate": 1.966716222796238e-05, "loss": 3.6654, "step": 32630 }, { "epoch": 0.0095744, "grad_norm": 0.8767901659011841, "learning_rate": 1.966695616529847e-05, "loss": 3.8093, "step": 32640 }, { "epoch": 0.0096, "grad_norm": 0.9224374890327454, "learning_rate": 1.9666750039946924e-05, "loss": 3.5078, "step": 32650 }, { "epoch": 0.0096256, "grad_norm": 0.989217221736908, "learning_rate": 1.966654385190907e-05, "loss": 3.4369, "step": 32660 }, { "epoch": 0.0096512, "grad_norm": 1.2191537618637085, "learning_rate": 1.9666337601186245e-05, "loss": 3.6927, "step": 32670 }, { "epoch": 0.0096768, "grad_norm": 1.0337666273117065, "learning_rate": 1.9666131287779792e-05, "loss": 3.7581, "step": 32680 }, { "epoch": 0.0097024, "grad_norm": 1.1801862716674805, "learning_rate": 1.966592491169105e-05, "loss": 3.8988, "step": 32690 }, { "epoch": 0.009728, "grad_norm": 0.6284043788909912, "learning_rate": 1.966571847292135e-05, "loss": 3.7214, "step": 32700 }, { "epoch": 0.0097536, "grad_norm": 0.9368493556976318, "learning_rate": 1.9665511971472035e-05, "loss": 3.6155, "step": 32710 }, { "epoch": 0.0097792, "grad_norm": 3.556227207183838, "learning_rate": 1.9665305407344446e-05, "loss": 3.6159, "step": 32720 }, { "epoch": 0.0098048, "grad_norm": 0.7931320071220398, "learning_rate": 1.9665098780539916e-05, "loss": 3.5428, "step": 32730 }, { "epoch": 0.0098304, "grad_norm": 0.7800467014312744, "learning_rate": 1.9664892091059793e-05, "loss": 3.709, "step": 32740 }, { "epoch": 0.009856, "grad_norm": 0.8633655309677124, "learning_rate": 1.966468533890541e-05, "loss": 3.4709, "step": 32750 }, { "epoch": 0.0098816, "grad_norm": 1.625071406364441, "learning_rate": 1.9664478524078113e-05, "loss": 3.7724, "step": 32760 }, { "epoch": 0.0099072, "grad_norm": 1.739474892616272, "learning_rate": 1.9664271646579244e-05, "loss": 3.7658, "step": 32770 }, { "epoch": 0.0099328, "grad_norm": 2.8578834533691406, "learning_rate": 1.966406470641014e-05, "loss": 3.8949, "step": 32780 }, { "epoch": 0.0099584, "grad_norm": 1.4442286491394043, "learning_rate": 1.966385770357214e-05, "loss": 3.6731, "step": 32790 }, { "epoch": 0.009984, "grad_norm": 0.7787548899650574, "learning_rate": 1.9663650638066598e-05, "loss": 3.691, "step": 32800 }, { "epoch": 0.0100096, "grad_norm": 0.9004014730453491, "learning_rate": 1.9663443509894847e-05, "loss": 3.5481, "step": 32810 }, { "epoch": 0.0100352, "grad_norm": 0.8760932087898254, "learning_rate": 1.9663236319058233e-05, "loss": 3.6046, "step": 32820 }, { "epoch": 0.0100608, "grad_norm": 0.8360670208930969, "learning_rate": 1.9663029065558098e-05, "loss": 3.6868, "step": 32830 }, { "epoch": 0.0100864, "grad_norm": 1.2847661972045898, "learning_rate": 1.966282174939579e-05, "loss": 3.5567, "step": 32840 }, { "epoch": 0.010112, "grad_norm": 0.7965390682220459, "learning_rate": 1.966261437057265e-05, "loss": 3.3212, "step": 32850 }, { "epoch": 0.0101376, "grad_norm": 0.8953307867050171, "learning_rate": 1.9662406929090023e-05, "loss": 3.6921, "step": 32860 }, { "epoch": 0.0101632, "grad_norm": 1.0307230949401855, "learning_rate": 1.9662199424949257e-05, "loss": 3.6374, "step": 32870 }, { "epoch": 0.0101888, "grad_norm": 7.024886131286621, "learning_rate": 1.9661991858151696e-05, "loss": 3.7767, "step": 32880 }, { "epoch": 0.0102144, "grad_norm": 0.8957598805427551, "learning_rate": 1.9661784228698684e-05, "loss": 3.7892, "step": 32890 }, { "epoch": 0.01024, "grad_norm": 0.9481748938560486, "learning_rate": 1.966157653659157e-05, "loss": 3.7494, "step": 32900 }, { "epoch": 0.0102656, "grad_norm": 0.9067980051040649, "learning_rate": 1.96613687818317e-05, "loss": 3.6241, "step": 32910 }, { "epoch": 0.0102912, "grad_norm": 3.8509039878845215, "learning_rate": 1.966116096442042e-05, "loss": 3.732, "step": 32920 }, { "epoch": 0.0103168, "grad_norm": 0.9139289855957031, "learning_rate": 1.966095308435908e-05, "loss": 3.5669, "step": 32930 }, { "epoch": 0.0103424, "grad_norm": 0.7847647070884705, "learning_rate": 1.9660745141649028e-05, "loss": 3.7767, "step": 32940 }, { "epoch": 0.010368, "grad_norm": 0.8888584971427917, "learning_rate": 1.966053713629161e-05, "loss": 3.7187, "step": 32950 }, { "epoch": 0.0103936, "grad_norm": 1.3024060726165771, "learning_rate": 1.9660329068288174e-05, "loss": 3.6568, "step": 32960 }, { "epoch": 0.0104192, "grad_norm": 0.902301013469696, "learning_rate": 1.966012093764007e-05, "loss": 3.7816, "step": 32970 }, { "epoch": 0.0104448, "grad_norm": 0.807094156742096, "learning_rate": 1.9659912744348655e-05, "loss": 3.5584, "step": 32980 }, { "epoch": 0.0104704, "grad_norm": 0.699296772480011, "learning_rate": 1.965970448841527e-05, "loss": 3.6982, "step": 32990 }, { "epoch": 0.010496, "grad_norm": 0.9706164002418518, "learning_rate": 1.965949616984127e-05, "loss": 3.5559, "step": 33000 }, { "epoch": 0.0105216, "grad_norm": 1.0761550664901733, "learning_rate": 1.9659287788628e-05, "loss": 3.4988, "step": 33010 }, { "epoch": 0.0105472, "grad_norm": 1.3833917379379272, "learning_rate": 1.965907934477682e-05, "loss": 3.5263, "step": 33020 }, { "epoch": 0.0105728, "grad_norm": 0.7693754434585571, "learning_rate": 1.965887083828908e-05, "loss": 3.6202, "step": 33030 }, { "epoch": 0.0105984, "grad_norm": 0.8598027229309082, "learning_rate": 1.9658662269166123e-05, "loss": 3.6378, "step": 33040 }, { "epoch": 0.010624, "grad_norm": 0.7657833695411682, "learning_rate": 1.965845363740931e-05, "loss": 3.7842, "step": 33050 }, { "epoch": 0.0106496, "grad_norm": 0.7541384696960449, "learning_rate": 1.9658244943019996e-05, "loss": 3.6428, "step": 33060 }, { "epoch": 0.0106752, "grad_norm": 0.9769710302352905, "learning_rate": 1.965803618599953e-05, "loss": 3.4073, "step": 33070 }, { "epoch": 0.0107008, "grad_norm": 0.982292652130127, "learning_rate": 1.9657827366349262e-05, "loss": 3.5297, "step": 33080 }, { "epoch": 0.0107264, "grad_norm": 0.8037064671516418, "learning_rate": 1.9657618484070553e-05, "loss": 3.5416, "step": 33090 }, { "epoch": 0.010752, "grad_norm": 0.9626526236534119, "learning_rate": 1.9657409539164752e-05, "loss": 3.8734, "step": 33100 }, { "epoch": 0.0107776, "grad_norm": 0.7597658634185791, "learning_rate": 1.965720053163322e-05, "loss": 3.6006, "step": 33110 }, { "epoch": 0.0108032, "grad_norm": 0.8787328004837036, "learning_rate": 1.965699146147731e-05, "loss": 3.4669, "step": 33120 }, { "epoch": 0.0108288, "grad_norm": 1.3956265449523926, "learning_rate": 1.9656782328698374e-05, "loss": 3.4972, "step": 33130 }, { "epoch": 0.0108544, "grad_norm": 0.7563385367393494, "learning_rate": 1.9656573133297772e-05, "loss": 3.5991, "step": 33140 }, { "epoch": 0.01088, "grad_norm": 0.9153509736061096, "learning_rate": 1.965636387527686e-05, "loss": 3.5909, "step": 33150 }, { "epoch": 0.0109056, "grad_norm": 0.7553024888038635, "learning_rate": 1.9656154554636995e-05, "loss": 3.5457, "step": 33160 }, { "epoch": 0.0109312, "grad_norm": 1.4208472967147827, "learning_rate": 1.9655945171379535e-05, "loss": 3.5257, "step": 33170 }, { "epoch": 0.0109568, "grad_norm": 0.9322847723960876, "learning_rate": 1.965573572550583e-05, "loss": 3.9196, "step": 33180 }, { "epoch": 0.0109824, "grad_norm": 0.7867373824119568, "learning_rate": 1.9655526217017254e-05, "loss": 3.5244, "step": 33190 }, { "epoch": 0.011008, "grad_norm": 1.0471867322921753, "learning_rate": 1.9655316645915154e-05, "loss": 3.7115, "step": 33200 }, { "epoch": 0.0110336, "grad_norm": 0.7403574585914612, "learning_rate": 1.965510701220089e-05, "loss": 3.574, "step": 33210 }, { "epoch": 0.0110592, "grad_norm": 0.8772501349449158, "learning_rate": 1.965489731587582e-05, "loss": 3.5609, "step": 33220 }, { "epoch": 0.0110848, "grad_norm": 1.392651915550232, "learning_rate": 1.965468755694131e-05, "loss": 3.7814, "step": 33230 }, { "epoch": 0.0111104, "grad_norm": 0.672082781791687, "learning_rate": 1.965447773539872e-05, "loss": 3.6484, "step": 33240 }, { "epoch": 0.011136, "grad_norm": 0.9519385099411011, "learning_rate": 1.9654267851249405e-05, "loss": 3.6372, "step": 33250 }, { "epoch": 0.0111616, "grad_norm": 0.7314754724502563, "learning_rate": 1.965405790449473e-05, "loss": 3.6462, "step": 33260 }, { "epoch": 0.0111872, "grad_norm": 1.341450572013855, "learning_rate": 1.965384789513605e-05, "loss": 3.6609, "step": 33270 }, { "epoch": 0.0112128, "grad_norm": 0.718078076839447, "learning_rate": 1.9653637823174742e-05, "loss": 3.6407, "step": 33280 }, { "epoch": 0.0112384, "grad_norm": 0.8422191143035889, "learning_rate": 1.965342768861215e-05, "loss": 3.4402, "step": 33290 }, { "epoch": 0.011264, "grad_norm": 1.0213465690612793, "learning_rate": 1.9653217491449648e-05, "loss": 3.6142, "step": 33300 }, { "epoch": 0.0112896, "grad_norm": 0.9416978359222412, "learning_rate": 1.9653007231688596e-05, "loss": 3.5932, "step": 33310 }, { "epoch": 0.0113152, "grad_norm": 1.1391353607177734, "learning_rate": 1.9652796909330355e-05, "loss": 3.699, "step": 33320 }, { "epoch": 0.0113408, "grad_norm": 0.7708802223205566, "learning_rate": 1.9652586524376292e-05, "loss": 3.5662, "step": 33330 }, { "epoch": 0.0113664, "grad_norm": 0.8511199355125427, "learning_rate": 1.9652376076827775e-05, "loss": 3.6414, "step": 33340 }, { "epoch": 0.011392, "grad_norm": 0.8314061760902405, "learning_rate": 1.9652165566686162e-05, "loss": 3.4809, "step": 33350 }, { "epoch": 0.0114176, "grad_norm": 1.1804291009902954, "learning_rate": 1.9651954993952823e-05, "loss": 3.4139, "step": 33360 }, { "epoch": 0.0114432, "grad_norm": 0.7547909617424011, "learning_rate": 1.9651744358629116e-05, "loss": 3.5245, "step": 33370 }, { "epoch": 0.0114688, "grad_norm": 0.8757628202438354, "learning_rate": 1.9651533660716416e-05, "loss": 3.7487, "step": 33380 }, { "epoch": 0.0114944, "grad_norm": 0.8913978338241577, "learning_rate": 1.9651322900216082e-05, "loss": 3.6018, "step": 33390 }, { "epoch": 0.01152, "grad_norm": 0.8036907315254211, "learning_rate": 1.9651112077129487e-05, "loss": 3.6087, "step": 33400 }, { "epoch": 0.0115456, "grad_norm": 0.7879661917686462, "learning_rate": 1.9650901191457992e-05, "loss": 3.565, "step": 33410 }, { "epoch": 0.0115712, "grad_norm": 0.8009902834892273, "learning_rate": 1.9650690243202974e-05, "loss": 3.598, "step": 33420 }, { "epoch": 0.0115968, "grad_norm": 3.3401083946228027, "learning_rate": 1.9650479232365786e-05, "loss": 3.8991, "step": 33430 }, { "epoch": 0.0116224, "grad_norm": 0.8771002888679504, "learning_rate": 1.965026815894781e-05, "loss": 3.5707, "step": 33440 }, { "epoch": 0.011648, "grad_norm": 0.8216012716293335, "learning_rate": 1.9650057022950408e-05, "loss": 3.2589, "step": 33450 }, { "epoch": 0.0116736, "grad_norm": 0.8883204460144043, "learning_rate": 1.964984582437495e-05, "loss": 3.5986, "step": 33460 }, { "epoch": 0.0116992, "grad_norm": 1.2808887958526611, "learning_rate": 1.964963456322281e-05, "loss": 3.8421, "step": 33470 }, { "epoch": 0.0117248, "grad_norm": 0.7468708753585815, "learning_rate": 1.9649423239495352e-05, "loss": 3.7003, "step": 33480 }, { "epoch": 0.0117504, "grad_norm": 0.7606188654899597, "learning_rate": 1.964921185319395e-05, "loss": 3.555, "step": 33490 }, { "epoch": 0.011776, "grad_norm": 3.353644847869873, "learning_rate": 1.9649000404319974e-05, "loss": 3.8225, "step": 33500 }, { "epoch": 0.0118016, "grad_norm": 0.8447765111923218, "learning_rate": 1.9648788892874795e-05, "loss": 3.5623, "step": 33510 }, { "epoch": 0.0118272, "grad_norm": 1.1042269468307495, "learning_rate": 1.964857731885978e-05, "loss": 3.7408, "step": 33520 }, { "epoch": 0.0118528, "grad_norm": 0.9168924689292908, "learning_rate": 1.964836568227631e-05, "loss": 3.6356, "step": 33530 }, { "epoch": 0.0118784, "grad_norm": 0.724132776260376, "learning_rate": 1.9648153983125752e-05, "loss": 3.5227, "step": 33540 }, { "epoch": 0.011904, "grad_norm": 0.7364735007286072, "learning_rate": 1.964794222140948e-05, "loss": 3.451, "step": 33550 }, { "epoch": 0.0119296, "grad_norm": 1.1828749179840088, "learning_rate": 1.9647730397128863e-05, "loss": 3.5662, "step": 33560 }, { "epoch": 0.0119552, "grad_norm": 0.895529568195343, "learning_rate": 1.9647518510285282e-05, "loss": 3.5948, "step": 33570 }, { "epoch": 0.0119808, "grad_norm": 0.8252629041671753, "learning_rate": 1.9647306560880105e-05, "loss": 3.6703, "step": 33580 }, { "epoch": 0.0120064, "grad_norm": 1.0138291120529175, "learning_rate": 1.9647094548914715e-05, "loss": 3.5741, "step": 33590 }, { "epoch": 0.012032, "grad_norm": 1.2840327024459839, "learning_rate": 1.9646882474390472e-05, "loss": 3.597, "step": 33600 }, { "epoch": 0.0120576, "grad_norm": 3.540834426879883, "learning_rate": 1.9646670337308764e-05, "loss": 3.4852, "step": 33610 }, { "epoch": 0.0120832, "grad_norm": 0.9570579528808594, "learning_rate": 1.9646458137670966e-05, "loss": 3.4958, "step": 33620 }, { "epoch": 0.0121088, "grad_norm": 1.0988235473632812, "learning_rate": 1.9646245875478446e-05, "loss": 3.613, "step": 33630 }, { "epoch": 0.0121344, "grad_norm": 1.1006380319595337, "learning_rate": 1.9646033550732587e-05, "loss": 3.5303, "step": 33640 }, { "epoch": 0.01216, "grad_norm": 1.4839986562728882, "learning_rate": 1.9645821163434765e-05, "loss": 3.6766, "step": 33650 }, { "epoch": 0.0121856, "grad_norm": 0.7840291857719421, "learning_rate": 1.964560871358635e-05, "loss": 3.5384, "step": 33660 }, { "epoch": 0.0122112, "grad_norm": 0.9601384997367859, "learning_rate": 1.9645396201188735e-05, "loss": 3.352, "step": 33670 }, { "epoch": 0.0122368, "grad_norm": 1.0380107164382935, "learning_rate": 1.9645183626243286e-05, "loss": 3.5912, "step": 33680 }, { "epoch": 0.0122624, "grad_norm": 0.7543262243270874, "learning_rate": 1.9644970988751382e-05, "loss": 3.517, "step": 33690 }, { "epoch": 0.012288, "grad_norm": 0.9423026442527771, "learning_rate": 1.9644758288714406e-05, "loss": 3.6558, "step": 33700 }, { "epoch": 0.0123136, "grad_norm": 0.9623668193817139, "learning_rate": 1.9644545526133737e-05, "loss": 3.3981, "step": 33710 }, { "epoch": 0.0123392, "grad_norm": 0.8645933270454407, "learning_rate": 1.964433270101075e-05, "loss": 3.7577, "step": 33720 }, { "epoch": 0.0123648, "grad_norm": 0.878866970539093, "learning_rate": 1.9644119813346832e-05, "loss": 3.4756, "step": 33730 }, { "epoch": 0.0123904, "grad_norm": 1.0759053230285645, "learning_rate": 1.964390686314336e-05, "loss": 3.4842, "step": 33740 }, { "epoch": 0.012416, "grad_norm": 0.9189815521240234, "learning_rate": 1.9643693850401716e-05, "loss": 3.4125, "step": 33750 }, { "epoch": 0.0124416, "grad_norm": 1.1648715734481812, "learning_rate": 1.9643480775123277e-05, "loss": 3.5694, "step": 33760 }, { "epoch": 0.0124672, "grad_norm": 0.844158411026001, "learning_rate": 1.9643267637309428e-05, "loss": 3.4489, "step": 33770 }, { "epoch": 0.0124928, "grad_norm": 0.71067214012146, "learning_rate": 1.9643054436961556e-05, "loss": 3.6517, "step": 33780 }, { "epoch": 0.0125184, "grad_norm": 0.9466609358787537, "learning_rate": 1.9642841174081035e-05, "loss": 3.7985, "step": 33790 }, { "epoch": 0.012544, "grad_norm": 0.9326655268669128, "learning_rate": 1.964262784866925e-05, "loss": 3.4928, "step": 33800 }, { "epoch": 0.0125696, "grad_norm": 0.844619631767273, "learning_rate": 1.964241446072759e-05, "loss": 3.464, "step": 33810 }, { "epoch": 0.0125952, "grad_norm": 1.2851388454437256, "learning_rate": 1.9642201010257433e-05, "loss": 3.4947, "step": 33820 }, { "epoch": 0.0126208, "grad_norm": 1.5530813932418823, "learning_rate": 1.9641987497260165e-05, "loss": 3.7803, "step": 33830 }, { "epoch": 0.0126464, "grad_norm": 0.8390510678291321, "learning_rate": 1.9641773921737172e-05, "loss": 3.5489, "step": 33840 }, { "epoch": 0.012672, "grad_norm": 1.0454497337341309, "learning_rate": 1.964156028368984e-05, "loss": 3.4081, "step": 33850 }, { "epoch": 0.0126976, "grad_norm": 1.0115954875946045, "learning_rate": 1.9641346583119545e-05, "loss": 3.8039, "step": 33860 }, { "epoch": 0.0127232, "grad_norm": 1.1521954536437988, "learning_rate": 1.9641132820027682e-05, "loss": 3.4784, "step": 33870 }, { "epoch": 0.0127488, "grad_norm": 1.133725881576538, "learning_rate": 1.964091899441564e-05, "loss": 3.4324, "step": 33880 }, { "epoch": 0.0127744, "grad_norm": 0.7598399519920349, "learning_rate": 1.9640705106284793e-05, "loss": 3.8892, "step": 33890 }, { "epoch": 0.0128, "grad_norm": 1.033149242401123, "learning_rate": 1.9640491155636542e-05, "loss": 3.8373, "step": 33900 }, { "epoch": 0.0128256, "grad_norm": 1.0730929374694824, "learning_rate": 1.9640277142472266e-05, "loss": 3.8583, "step": 33910 }, { "epoch": 0.0128512, "grad_norm": 0.8915746212005615, "learning_rate": 1.9640063066793353e-05, "loss": 3.6871, "step": 33920 }, { "epoch": 0.0128768, "grad_norm": 0.9758279323577881, "learning_rate": 1.9639848928601193e-05, "loss": 3.6858, "step": 33930 }, { "epoch": 0.0129024, "grad_norm": 0.7804235816001892, "learning_rate": 1.9639634727897178e-05, "loss": 3.5188, "step": 33940 }, { "epoch": 0.012928, "grad_norm": 0.7196527719497681, "learning_rate": 1.963942046468269e-05, "loss": 3.3854, "step": 33950 }, { "epoch": 0.0129536, "grad_norm": 1.5514785051345825, "learning_rate": 1.9639206138959122e-05, "loss": 3.6292, "step": 33960 }, { "epoch": 0.0129792, "grad_norm": 0.8512642979621887, "learning_rate": 1.9638991750727868e-05, "loss": 3.5597, "step": 33970 }, { "epoch": 0.0130048, "grad_norm": 0.7546426057815552, "learning_rate": 1.963877729999031e-05, "loss": 3.71, "step": 33980 }, { "epoch": 0.0130304, "grad_norm": 0.9357023239135742, "learning_rate": 1.9638562786747845e-05, "loss": 3.6207, "step": 33990 }, { "epoch": 0.013056, "grad_norm": 0.8391178250312805, "learning_rate": 1.9638348211001866e-05, "loss": 3.7578, "step": 34000 }, { "epoch": 0.0130816, "grad_norm": 0.7116932272911072, "learning_rate": 1.9638133572753756e-05, "loss": 3.6685, "step": 34010 }, { "epoch": 0.0131072, "grad_norm": 0.7978698015213013, "learning_rate": 1.963791887200491e-05, "loss": 3.5434, "step": 34020 }, { "epoch": 0.0131328, "grad_norm": 0.7676711678504944, "learning_rate": 1.9637704108756724e-05, "loss": 3.4356, "step": 34030 }, { "epoch": 0.0131584, "grad_norm": 1.0229848623275757, "learning_rate": 1.9637489283010587e-05, "loss": 3.5768, "step": 34040 }, { "epoch": 0.013184, "grad_norm": 0.9543655514717102, "learning_rate": 1.9637274394767892e-05, "loss": 3.6063, "step": 34050 }, { "epoch": 0.0132096, "grad_norm": 0.8649645447731018, "learning_rate": 1.9637059444030036e-05, "loss": 3.4641, "step": 34060 }, { "epoch": 0.0132352, "grad_norm": 0.9126257300376892, "learning_rate": 1.963684443079841e-05, "loss": 3.8249, "step": 34070 }, { "epoch": 0.0132608, "grad_norm": 0.9268119931221008, "learning_rate": 1.963662935507441e-05, "loss": 3.5813, "step": 34080 }, { "epoch": 0.0132864, "grad_norm": 0.9418911337852478, "learning_rate": 1.963641421685943e-05, "loss": 3.6309, "step": 34090 }, { "epoch": 0.013312, "grad_norm": 0.7928049564361572, "learning_rate": 1.9636199016154864e-05, "loss": 3.7006, "step": 34100 }, { "epoch": 0.0133376, "grad_norm": 0.8145092725753784, "learning_rate": 1.963598375296211e-05, "loss": 3.8651, "step": 34110 }, { "epoch": 0.0133632, "grad_norm": 1.0291504859924316, "learning_rate": 1.963576842728256e-05, "loss": 3.7745, "step": 34120 }, { "epoch": 0.0133888, "grad_norm": 1.0895979404449463, "learning_rate": 1.963555303911761e-05, "loss": 3.792, "step": 34130 }, { "epoch": 0.0134144, "grad_norm": 1.2164735794067383, "learning_rate": 1.9635337588468665e-05, "loss": 3.5666, "step": 34140 }, { "epoch": 0.01344, "grad_norm": 1.210940957069397, "learning_rate": 1.9635122075337114e-05, "loss": 3.7608, "step": 34150 }, { "epoch": 0.0134656, "grad_norm": 1.1315391063690186, "learning_rate": 1.9634906499724357e-05, "loss": 3.6416, "step": 34160 }, { "epoch": 0.0134912, "grad_norm": 2.476398229598999, "learning_rate": 1.9634690861631793e-05, "loss": 3.8679, "step": 34170 }, { "epoch": 0.0135168, "grad_norm": 1.0207159519195557, "learning_rate": 1.9634475161060816e-05, "loss": 3.6293, "step": 34180 }, { "epoch": 0.0135424, "grad_norm": 1.1839556694030762, "learning_rate": 1.963425939801283e-05, "loss": 3.8588, "step": 34190 }, { "epoch": 0.013568, "grad_norm": 0.7035020589828491, "learning_rate": 1.9634043572489237e-05, "loss": 3.8005, "step": 34200 }, { "epoch": 0.0135936, "grad_norm": 0.9899457693099976, "learning_rate": 1.9633827684491423e-05, "loss": 3.697, "step": 34210 }, { "epoch": 0.0136192, "grad_norm": 1.1913491487503052, "learning_rate": 1.9633611734020803e-05, "loss": 3.7719, "step": 34220 }, { "epoch": 0.0136448, "grad_norm": 1.2456656694412231, "learning_rate": 1.9633395721078772e-05, "loss": 3.7277, "step": 34230 }, { "epoch": 0.0136704, "grad_norm": 0.7735298275947571, "learning_rate": 1.9633179645666727e-05, "loss": 3.6758, "step": 34240 }, { "epoch": 0.013696, "grad_norm": 0.860828161239624, "learning_rate": 1.9632963507786073e-05, "loss": 3.6438, "step": 34250 }, { "epoch": 0.0137216, "grad_norm": 0.9757567048072815, "learning_rate": 1.963274730743821e-05, "loss": 3.7478, "step": 34260 }, { "epoch": 0.0137472, "grad_norm": 0.7476038932800293, "learning_rate": 1.9632531044624542e-05, "loss": 3.7891, "step": 34270 }, { "epoch": 0.0137728, "grad_norm": 1.253288984298706, "learning_rate": 1.963231471934647e-05, "loss": 3.6402, "step": 34280 }, { "epoch": 0.0137984, "grad_norm": 0.7387740015983582, "learning_rate": 1.96320983316054e-05, "loss": 3.6953, "step": 34290 }, { "epoch": 0.013824, "grad_norm": 0.9911642670631409, "learning_rate": 1.9631881881402724e-05, "loss": 3.8658, "step": 34300 }, { "epoch": 0.0138496, "grad_norm": 0.7428621649742126, "learning_rate": 1.963166536873986e-05, "loss": 3.7397, "step": 34310 }, { "epoch": 0.0138752, "grad_norm": 1.634724497795105, "learning_rate": 1.9631448793618207e-05, "loss": 3.7421, "step": 34320 }, { "epoch": 0.0139008, "grad_norm": 0.7808132767677307, "learning_rate": 1.9631232156039163e-05, "loss": 3.7479, "step": 34330 }, { "epoch": 0.0139264, "grad_norm": 0.7634571194648743, "learning_rate": 1.9631015456004144e-05, "loss": 3.634, "step": 34340 }, { "epoch": 0.013952, "grad_norm": 0.8126035332679749, "learning_rate": 1.9630798693514544e-05, "loss": 3.6885, "step": 34350 }, { "epoch": 0.0139776, "grad_norm": 1.1040159463882446, "learning_rate": 1.9630581868571775e-05, "loss": 3.7171, "step": 34360 }, { "epoch": 0.0140032, "grad_norm": 0.9563593864440918, "learning_rate": 1.9630364981177246e-05, "loss": 3.6723, "step": 34370 }, { "epoch": 0.0140288, "grad_norm": 0.9876836538314819, "learning_rate": 1.963014803133236e-05, "loss": 3.4394, "step": 34380 }, { "epoch": 0.0140544, "grad_norm": 0.9490966796875, "learning_rate": 1.9629931019038517e-05, "loss": 3.6214, "step": 34390 }, { "epoch": 0.01408, "grad_norm": 0.6981646418571472, "learning_rate": 1.9629713944297136e-05, "loss": 3.5886, "step": 34400 }, { "epoch": 0.0141056, "grad_norm": 0.7747970819473267, "learning_rate": 1.9629496807109616e-05, "loss": 3.6639, "step": 34410 }, { "epoch": 0.0141312, "grad_norm": 0.8276783227920532, "learning_rate": 1.962927960747737e-05, "loss": 3.647, "step": 34420 }, { "epoch": 0.0141568, "grad_norm": 0.9366831183433533, "learning_rate": 1.9629062345401805e-05, "loss": 3.6155, "step": 34430 }, { "epoch": 0.0141824, "grad_norm": 1.007202386856079, "learning_rate": 1.962884502088433e-05, "loss": 3.6517, "step": 34440 }, { "epoch": 0.014208, "grad_norm": 1.851232647895813, "learning_rate": 1.9628627633926354e-05, "loss": 3.4761, "step": 34450 }, { "epoch": 0.0142336, "grad_norm": 0.7331991195678711, "learning_rate": 1.9628410184529286e-05, "loss": 3.8269, "step": 34460 }, { "epoch": 0.0142592, "grad_norm": 0.835442066192627, "learning_rate": 1.962819267269454e-05, "loss": 4.3683, "step": 34470 }, { "epoch": 0.0142848, "grad_norm": 0.7545817494392395, "learning_rate": 1.9627975098423518e-05, "loss": 3.6837, "step": 34480 }, { "epoch": 0.0143104, "grad_norm": 0.7959833145141602, "learning_rate": 1.962775746171764e-05, "loss": 3.596, "step": 34490 }, { "epoch": 0.014336, "grad_norm": 1.7026352882385254, "learning_rate": 1.9627539762578312e-05, "loss": 3.6995, "step": 34500 }, { "epoch": 0.0143616, "grad_norm": 0.958898663520813, "learning_rate": 1.9627322001006947e-05, "loss": 3.7341, "step": 34510 }, { "epoch": 0.0143872, "grad_norm": 1.0440590381622314, "learning_rate": 1.962710417700496e-05, "loss": 3.5807, "step": 34520 }, { "epoch": 0.0144128, "grad_norm": 0.8277835249900818, "learning_rate": 1.962688629057376e-05, "loss": 3.5023, "step": 34530 }, { "epoch": 0.0144384, "grad_norm": 0.8511258959770203, "learning_rate": 1.962666834171476e-05, "loss": 3.6477, "step": 34540 }, { "epoch": 0.014464, "grad_norm": 0.9271231293678284, "learning_rate": 1.9626450330429375e-05, "loss": 3.5629, "step": 34550 }, { "epoch": 0.0144896, "grad_norm": 1.0396870374679565, "learning_rate": 1.9626232256719016e-05, "loss": 3.6277, "step": 34560 }, { "epoch": 0.0145152, "grad_norm": 0.8435205817222595, "learning_rate": 1.9626014120585104e-05, "loss": 3.6717, "step": 34570 }, { "epoch": 0.0145408, "grad_norm": 0.867088258266449, "learning_rate": 1.9625795922029045e-05, "loss": 3.6874, "step": 34580 }, { "epoch": 0.0145664, "grad_norm": 1.000057578086853, "learning_rate": 1.962557766105226e-05, "loss": 3.6075, "step": 34590 }, { "epoch": 0.014592, "grad_norm": 1.022192358970642, "learning_rate": 1.9625359337656157e-05, "loss": 3.7077, "step": 34600 }, { "epoch": 0.0146176, "grad_norm": 2.9202842712402344, "learning_rate": 1.9625140951842164e-05, "loss": 4.0028, "step": 34610 }, { "epoch": 0.0146432, "grad_norm": 0.9899886846542358, "learning_rate": 1.9624922503611685e-05, "loss": 3.5454, "step": 34620 }, { "epoch": 0.0146688, "grad_norm": 0.9600430727005005, "learning_rate": 1.962470399296614e-05, "loss": 3.6591, "step": 34630 }, { "epoch": 0.0146944, "grad_norm": 0.9531959891319275, "learning_rate": 1.962448541990695e-05, "loss": 3.5259, "step": 34640 }, { "epoch": 0.01472, "grad_norm": 1.6740893125534058, "learning_rate": 1.9624266784435533e-05, "loss": 3.6192, "step": 34650 }, { "epoch": 0.0147456, "grad_norm": 1.2703746557235718, "learning_rate": 1.96240480865533e-05, "loss": 3.4914, "step": 34660 }, { "epoch": 0.0147712, "grad_norm": 0.9146012663841248, "learning_rate": 1.9623829326261675e-05, "loss": 3.5393, "step": 34670 }, { "epoch": 0.0147968, "grad_norm": 0.9853202700614929, "learning_rate": 1.962361050356207e-05, "loss": 3.7084, "step": 34680 }, { "epoch": 0.0148224, "grad_norm": 0.9158510565757751, "learning_rate": 1.9623391618455914e-05, "loss": 3.5684, "step": 34690 }, { "epoch": 0.014848, "grad_norm": 0.8688972592353821, "learning_rate": 1.962317267094462e-05, "loss": 3.6329, "step": 34700 }, { "epoch": 0.0148736, "grad_norm": 0.8632927536964417, "learning_rate": 1.9622953661029608e-05, "loss": 3.6745, "step": 34710 }, { "epoch": 0.0148992, "grad_norm": 0.9723210334777832, "learning_rate": 1.9622734588712297e-05, "loss": 3.5008, "step": 34720 }, { "epoch": 0.0149248, "grad_norm": 1.2882771492004395, "learning_rate": 1.962251545399411e-05, "loss": 3.6785, "step": 34730 }, { "epoch": 0.0149504, "grad_norm": 0.9129915833473206, "learning_rate": 1.962229625687647e-05, "loss": 3.7187, "step": 34740 }, { "epoch": 0.014976, "grad_norm": 0.7999345064163208, "learning_rate": 1.9622076997360795e-05, "loss": 3.6026, "step": 34750 }, { "epoch": 0.0150016, "grad_norm": 0.8213152885437012, "learning_rate": 1.962185767544851e-05, "loss": 3.956, "step": 34760 }, { "epoch": 0.0150272, "grad_norm": 0.812612771987915, "learning_rate": 1.9621638291141032e-05, "loss": 3.5232, "step": 34770 }, { "epoch": 0.0150528, "grad_norm": 0.8147515654563904, "learning_rate": 1.9621418844439787e-05, "loss": 3.7079, "step": 34780 }, { "epoch": 0.0150784, "grad_norm": 1.4639391899108887, "learning_rate": 1.96211993353462e-05, "loss": 3.6649, "step": 34790 }, { "epoch": 0.015104, "grad_norm": 0.7488465905189514, "learning_rate": 1.962097976386169e-05, "loss": 3.7194, "step": 34800 }, { "epoch": 0.0151296, "grad_norm": 0.8138414621353149, "learning_rate": 1.9620760129987683e-05, "loss": 3.8823, "step": 34810 }, { "epoch": 0.0151552, "grad_norm": 1.4020501375198364, "learning_rate": 1.9620540433725605e-05, "loss": 3.5843, "step": 34820 }, { "epoch": 0.0151808, "grad_norm": 0.9627105593681335, "learning_rate": 1.9620320675076877e-05, "loss": 3.9547, "step": 34830 }, { "epoch": 0.0152064, "grad_norm": 0.8114227056503296, "learning_rate": 1.962010085404293e-05, "loss": 3.6023, "step": 34840 }, { "epoch": 0.015232, "grad_norm": 0.7989040613174438, "learning_rate": 1.961988097062518e-05, "loss": 3.4803, "step": 34850 }, { "epoch": 0.0152576, "grad_norm": 0.9249349236488342, "learning_rate": 1.9619661024825066e-05, "loss": 3.4599, "step": 34860 }, { "epoch": 0.0152832, "grad_norm": 0.9518409371376038, "learning_rate": 1.9619441016644e-05, "loss": 3.4843, "step": 34870 }, { "epoch": 0.0153088, "grad_norm": 0.7739441394805908, "learning_rate": 1.9619220946083423e-05, "loss": 3.5034, "step": 34880 }, { "epoch": 0.0153344, "grad_norm": 0.8960883617401123, "learning_rate": 1.9619000813144748e-05, "loss": 3.681, "step": 34890 }, { "epoch": 0.01536, "grad_norm": 0.83669114112854, "learning_rate": 1.961878061782941e-05, "loss": 3.762, "step": 34900 }, { "epoch": 0.0153856, "grad_norm": 0.8097332119941711, "learning_rate": 1.961856036013884e-05, "loss": 3.5214, "step": 34910 }, { "epoch": 0.0154112, "grad_norm": 0.9204548001289368, "learning_rate": 1.9618340040074458e-05, "loss": 3.4377, "step": 34920 }, { "epoch": 1.00002048, "grad_norm": 0.9606540203094482, "learning_rate": 1.96181196576377e-05, "loss": 3.6734, "step": 34930 }, { "epoch": 1.00004608, "grad_norm": 0.8379903435707092, "learning_rate": 1.961789921282999e-05, "loss": 3.4506, "step": 34940 }, { "epoch": 1.00007168, "grad_norm": 0.9098504781723022, "learning_rate": 1.961767870565276e-05, "loss": 3.6912, "step": 34950 }, { "epoch": 1.00009728, "grad_norm": 0.8106098175048828, "learning_rate": 1.9617458136107442e-05, "loss": 3.3347, "step": 34960 }, { "epoch": 1.00012288, "grad_norm": 1.094664454460144, "learning_rate": 1.961723750419546e-05, "loss": 3.2944, "step": 34970 }, { "epoch": 1.00014848, "grad_norm": 1.1400786638259888, "learning_rate": 1.9617016809918253e-05, "loss": 3.4389, "step": 34980 }, { "epoch": 1.00017408, "grad_norm": 0.7769654989242554, "learning_rate": 1.9616796053277248e-05, "loss": 3.5361, "step": 34990 }, { "epoch": 1.00019968, "grad_norm": 1.2008336782455444, "learning_rate": 1.9616575234273873e-05, "loss": 3.3375, "step": 35000 }, { "epoch": 1.00022528, "grad_norm": 0.9843141436576843, "learning_rate": 1.9616354352909567e-05, "loss": 3.2798, "step": 35010 }, { "epoch": 1.00025088, "grad_norm": 1.1131449937820435, "learning_rate": 1.9616133409185754e-05, "loss": 3.6077, "step": 35020 }, { "epoch": 1.00027648, "grad_norm": 0.884164571762085, "learning_rate": 1.9615912403103873e-05, "loss": 3.7651, "step": 35030 }, { "epoch": 1.00030208, "grad_norm": 0.7754958271980286, "learning_rate": 1.9615691334665358e-05, "loss": 3.3821, "step": 35040 }, { "epoch": 1.00032768, "grad_norm": 0.9211655855178833, "learning_rate": 1.9615470203871642e-05, "loss": 3.4103, "step": 35050 }, { "epoch": 1.00035328, "grad_norm": 0.7390272617340088, "learning_rate": 1.961524901072415e-05, "loss": 3.4184, "step": 35060 }, { "epoch": 1.00037888, "grad_norm": 1.1540591716766357, "learning_rate": 1.961502775522433e-05, "loss": 3.5312, "step": 35070 }, { "epoch": 1.00040448, "grad_norm": 0.8228347301483154, "learning_rate": 1.9614806437373608e-05, "loss": 3.373, "step": 35080 }, { "epoch": 1.00043008, "grad_norm": 0.7965388894081116, "learning_rate": 1.9614585057173422e-05, "loss": 3.4322, "step": 35090 }, { "epoch": 1.00045568, "grad_norm": 0.877102792263031, "learning_rate": 1.9614363614625204e-05, "loss": 3.3596, "step": 35100 }, { "epoch": 1.00048128, "grad_norm": 0.9442691802978516, "learning_rate": 1.96141421097304e-05, "loss": 3.2218, "step": 35110 }, { "epoch": 1.00050688, "grad_norm": 0.816055953502655, "learning_rate": 1.9613920542490433e-05, "loss": 3.6173, "step": 35120 }, { "epoch": 1.00053248, "grad_norm": 0.9011704325675964, "learning_rate": 1.961369891290675e-05, "loss": 3.3825, "step": 35130 }, { "epoch": 1.00055808, "grad_norm": 0.6922595500946045, "learning_rate": 1.9613477220980784e-05, "loss": 3.3791, "step": 35140 }, { "epoch": 1.00058368, "grad_norm": 0.8362528681755066, "learning_rate": 1.9613255466713972e-05, "loss": 3.6457, "step": 35150 }, { "epoch": 1.00060928, "grad_norm": 0.7272346019744873, "learning_rate": 1.9613033650107755e-05, "loss": 3.4853, "step": 35160 }, { "epoch": 1.00063488, "grad_norm": 0.7619903087615967, "learning_rate": 1.961281177116357e-05, "loss": 3.4142, "step": 35170 }, { "epoch": 1.00066048, "grad_norm": 0.9437701106071472, "learning_rate": 1.961258982988285e-05, "loss": 3.4001, "step": 35180 }, { "epoch": 1.00068608, "grad_norm": 1.1311631202697754, "learning_rate": 1.9612367826267045e-05, "loss": 3.441, "step": 35190 }, { "epoch": 1.00071168, "grad_norm": 0.8052496910095215, "learning_rate": 1.9612145760317588e-05, "loss": 3.2268, "step": 35200 }, { "epoch": 1.00073728, "grad_norm": 0.713004469871521, "learning_rate": 1.961192363203592e-05, "loss": 3.3339, "step": 35210 }, { "epoch": 1.00076288, "grad_norm": 0.723074734210968, "learning_rate": 1.9611701441423482e-05, "loss": 3.3811, "step": 35220 }, { "epoch": 1.00078848, "grad_norm": 0.7885905504226685, "learning_rate": 1.9611479188481713e-05, "loss": 3.3582, "step": 35230 }, { "epoch": 1.00081408, "grad_norm": 1.0247942209243774, "learning_rate": 1.9611256873212062e-05, "loss": 3.504, "step": 35240 }, { "epoch": 1.00083968, "grad_norm": 0.8322592973709106, "learning_rate": 1.961103449561596e-05, "loss": 3.4776, "step": 35250 }, { "epoch": 1.00086528, "grad_norm": 0.719934344291687, "learning_rate": 1.9610812055694854e-05, "loss": 3.3654, "step": 35260 }, { "epoch": 1.00089088, "grad_norm": 0.802848219871521, "learning_rate": 1.9610589553450184e-05, "loss": 3.3101, "step": 35270 }, { "epoch": 1.00091648, "grad_norm": 0.7889386415481567, "learning_rate": 1.9610366988883398e-05, "loss": 3.3988, "step": 35280 }, { "epoch": 1.00094208, "grad_norm": 0.7773065567016602, "learning_rate": 1.9610144361995936e-05, "loss": 3.4513, "step": 35290 }, { "epoch": 1.00096768, "grad_norm": 0.8886108994483948, "learning_rate": 1.9609921672789238e-05, "loss": 3.4471, "step": 35300 }, { "epoch": 1.00099328, "grad_norm": 0.7170579433441162, "learning_rate": 1.9609698921264754e-05, "loss": 3.3747, "step": 35310 }, { "epoch": 1.00101888, "grad_norm": 0.8819681406021118, "learning_rate": 1.9609476107423927e-05, "loss": 3.4837, "step": 35320 }, { "epoch": 1.00104448, "grad_norm": 0.9172781705856323, "learning_rate": 1.9609253231268204e-05, "loss": 3.5489, "step": 35330 }, { "epoch": 1.00107008, "grad_norm": 0.8692842125892639, "learning_rate": 1.9609030292799026e-05, "loss": 3.5147, "step": 35340 }, { "epoch": 1.00109568, "grad_norm": 0.7441876530647278, "learning_rate": 1.960880729201784e-05, "loss": 3.4452, "step": 35350 }, { "epoch": 1.00112128, "grad_norm": 0.7160735726356506, "learning_rate": 1.960858422892609e-05, "loss": 3.3975, "step": 35360 }, { "epoch": 1.00114688, "grad_norm": 0.7579998970031738, "learning_rate": 1.9608361103525228e-05, "loss": 3.3244, "step": 35370 }, { "epoch": 1.00117248, "grad_norm": 0.7607085108757019, "learning_rate": 1.9608137915816696e-05, "loss": 3.2466, "step": 35380 }, { "epoch": 1.00119808, "grad_norm": 0.8198503851890564, "learning_rate": 1.9607914665801945e-05, "loss": 3.1978, "step": 35390 }, { "epoch": 1.00122368, "grad_norm": 0.837131917476654, "learning_rate": 1.960769135348242e-05, "loss": 3.3807, "step": 35400 }, { "epoch": 1.00124928, "grad_norm": 0.6872273683547974, "learning_rate": 1.960746797885957e-05, "loss": 3.3069, "step": 35410 }, { "epoch": 1.00127488, "grad_norm": 0.8140255212783813, "learning_rate": 1.9607244541934842e-05, "loss": 3.3967, "step": 35420 }, { "epoch": 1.00130048, "grad_norm": 0.8145366311073303, "learning_rate": 1.960702104270969e-05, "loss": 3.3144, "step": 35430 }, { "epoch": 1.00132608, "grad_norm": 0.8238507509231567, "learning_rate": 1.9606797481185555e-05, "loss": 3.1204, "step": 35440 }, { "epoch": 1.00135168, "grad_norm": 0.8958831429481506, "learning_rate": 1.9606573857363895e-05, "loss": 3.3885, "step": 35450 }, { "epoch": 1.00137728, "grad_norm": 0.9295539855957031, "learning_rate": 1.9606350171246154e-05, "loss": 3.2265, "step": 35460 }, { "epoch": 1.00140288, "grad_norm": 1.0069403648376465, "learning_rate": 1.960612642283379e-05, "loss": 3.7817, "step": 35470 }, { "epoch": 1.00142848, "grad_norm": 0.9438199400901794, "learning_rate": 1.9605902612128245e-05, "loss": 3.3762, "step": 35480 }, { "epoch": 1.00145408, "grad_norm": 0.6611096858978271, "learning_rate": 1.9605678739130976e-05, "loss": 3.1867, "step": 35490 }, { "epoch": 1.00147968, "grad_norm": 0.7726284861564636, "learning_rate": 1.9605454803843433e-05, "loss": 3.3364, "step": 35500 }, { "epoch": 1.00150528, "grad_norm": 0.7366737127304077, "learning_rate": 1.9605230806267066e-05, "loss": 3.431, "step": 35510 }, { "epoch": 1.00153088, "grad_norm": 0.6469897031784058, "learning_rate": 1.9605006746403332e-05, "loss": 3.1853, "step": 35520 }, { "epoch": 1.00155648, "grad_norm": 0.7983824610710144, "learning_rate": 1.9604782624253683e-05, "loss": 3.2734, "step": 35530 }, { "epoch": 1.00158208, "grad_norm": 1.1870425939559937, "learning_rate": 1.960455843981957e-05, "loss": 3.341, "step": 35540 }, { "epoch": 1.00160768, "grad_norm": 0.8195818066596985, "learning_rate": 1.960433419310245e-05, "loss": 3.4708, "step": 35550 }, { "epoch": 1.00163328, "grad_norm": 0.7783684730529785, "learning_rate": 1.9604109884103773e-05, "loss": 3.5218, "step": 35560 }, { "epoch": 1.00165888, "grad_norm": 1.3138251304626465, "learning_rate": 1.9603885512824998e-05, "loss": 3.3856, "step": 35570 }, { "epoch": 1.00168448, "grad_norm": 1.3767602443695068, "learning_rate": 1.9603661079267576e-05, "loss": 3.642, "step": 35580 }, { "epoch": 1.00171008, "grad_norm": 1.3366105556488037, "learning_rate": 1.9603436583432965e-05, "loss": 3.264, "step": 35590 }, { "epoch": 1.00173568, "grad_norm": 0.908986508846283, "learning_rate": 1.960321202532262e-05, "loss": 3.545, "step": 35600 }, { "epoch": 1.00176128, "grad_norm": 0.7765007019042969, "learning_rate": 1.9602987404937995e-05, "loss": 3.4954, "step": 35610 }, { "epoch": 1.00178688, "grad_norm": 0.7934430837631226, "learning_rate": 1.9602762722280553e-05, "loss": 3.1355, "step": 35620 }, { "epoch": 1.00181248, "grad_norm": 0.8075481653213501, "learning_rate": 1.9602537977351746e-05, "loss": 3.1262, "step": 35630 }, { "epoch": 1.00183808, "grad_norm": 0.8290247321128845, "learning_rate": 1.9602313170153028e-05, "loss": 3.3886, "step": 35640 }, { "epoch": 1.00186368, "grad_norm": 0.7880347371101379, "learning_rate": 1.9602088300685865e-05, "loss": 3.3107, "step": 35650 }, { "epoch": 1.00188928, "grad_norm": 0.7585127353668213, "learning_rate": 1.9601863368951714e-05, "loss": 3.5366, "step": 35660 }, { "epoch": 1.00191488, "grad_norm": 0.7698643803596497, "learning_rate": 1.9601638374952025e-05, "loss": 3.3737, "step": 35670 }, { "epoch": 1.00194048, "grad_norm": 0.7547111511230469, "learning_rate": 1.9601413318688266e-05, "loss": 3.2567, "step": 35680 }, { "epoch": 1.00196608, "grad_norm": 1.2465136051177979, "learning_rate": 1.9601188200161893e-05, "loss": 3.6478, "step": 35690 }, { "epoch": 1.00199168, "grad_norm": 0.8100875616073608, "learning_rate": 1.9600963019374367e-05, "loss": 3.4659, "step": 35700 }, { "epoch": 1.00201728, "grad_norm": 0.7529834508895874, "learning_rate": 1.9600737776327144e-05, "loss": 3.5629, "step": 35710 }, { "epoch": 1.00204288, "grad_norm": 0.6876733303070068, "learning_rate": 1.9600512471021692e-05, "loss": 3.4999, "step": 35720 }, { "epoch": 1.00206848, "grad_norm": 0.75758957862854, "learning_rate": 1.9600287103459464e-05, "loss": 3.537, "step": 35730 }, { "epoch": 1.00209408, "grad_norm": 0.7009768486022949, "learning_rate": 1.9600061673641934e-05, "loss": 3.0744, "step": 35740 }, { "epoch": 1.00211968, "grad_norm": 0.8394862413406372, "learning_rate": 1.9599836181570545e-05, "loss": 3.6033, "step": 35750 }, { "epoch": 1.00214528, "grad_norm": 0.8861285448074341, "learning_rate": 1.9599610627246775e-05, "loss": 3.6034, "step": 35760 }, { "epoch": 1.00217088, "grad_norm": 0.7471907734870911, "learning_rate": 1.959938501067208e-05, "loss": 3.6594, "step": 35770 }, { "epoch": 1.00219648, "grad_norm": 0.768373966217041, "learning_rate": 1.9599159331847927e-05, "loss": 3.2859, "step": 35780 }, { "epoch": 1.00222208, "grad_norm": 0.7698176503181458, "learning_rate": 1.9598933590775774e-05, "loss": 3.561, "step": 35790 }, { "epoch": 1.00224768, "grad_norm": 0.8838414549827576, "learning_rate": 1.959870778745709e-05, "loss": 3.6379, "step": 35800 }, { "epoch": 1.00227328, "grad_norm": 0.800270676612854, "learning_rate": 1.9598481921893336e-05, "loss": 3.5067, "step": 35810 }, { "epoch": 1.00229888, "grad_norm": 0.8029117584228516, "learning_rate": 1.9598255994085977e-05, "loss": 3.4223, "step": 35820 }, { "epoch": 1.00232448, "grad_norm": 0.729805588722229, "learning_rate": 1.9598030004036478e-05, "loss": 3.6119, "step": 35830 }, { "epoch": 1.00235008, "grad_norm": 1.0788723230361938, "learning_rate": 1.9597803951746307e-05, "loss": 3.6361, "step": 35840 }, { "epoch": 1.00237568, "grad_norm": 1.1738969087600708, "learning_rate": 1.9597577837216927e-05, "loss": 3.619, "step": 35850 }, { "epoch": 1.00240128, "grad_norm": 0.7420452833175659, "learning_rate": 1.959735166044981e-05, "loss": 3.2902, "step": 35860 }, { "epoch": 1.00242688, "grad_norm": 0.7898780107498169, "learning_rate": 1.9597125421446414e-05, "loss": 3.4898, "step": 35870 }, { "epoch": 1.00245248, "grad_norm": 0.8986024260520935, "learning_rate": 1.9596899120208208e-05, "loss": 3.3778, "step": 35880 }, { "epoch": 1.00247808, "grad_norm": 0.8223957419395447, "learning_rate": 1.9596672756736665e-05, "loss": 3.3284, "step": 35890 }, { "epoch": 1.00250368, "grad_norm": 0.9020829796791077, "learning_rate": 1.959644633103325e-05, "loss": 3.6122, "step": 35900 }, { "epoch": 1.00252928, "grad_norm": 0.7535799145698547, "learning_rate": 1.959621984309943e-05, "loss": 3.6801, "step": 35910 }, { "epoch": 1.00255488, "grad_norm": 1.0273510217666626, "learning_rate": 1.9595993292936676e-05, "loss": 3.5049, "step": 35920 }, { "epoch": 1.00258048, "grad_norm": 2.094574451446533, "learning_rate": 1.9595766680546453e-05, "loss": 3.402, "step": 35930 }, { "epoch": 1.00260608, "grad_norm": 0.8199256062507629, "learning_rate": 1.9595540005930235e-05, "loss": 3.5521, "step": 35940 }, { "epoch": 1.00263168, "grad_norm": 0.9019724726676941, "learning_rate": 1.959531326908949e-05, "loss": 3.1642, "step": 35950 }, { "epoch": 1.00265728, "grad_norm": 0.7315996289253235, "learning_rate": 1.9595086470025687e-05, "loss": 3.4397, "step": 35960 }, { "epoch": 1.00268288, "grad_norm": 0.7923860549926758, "learning_rate": 1.95948596087403e-05, "loss": 3.4633, "step": 35970 }, { "epoch": 1.00270848, "grad_norm": 0.8383647799491882, "learning_rate": 1.95946326852348e-05, "loss": 3.2479, "step": 35980 }, { "epoch": 1.00273408, "grad_norm": 0.8029789924621582, "learning_rate": 1.9594405699510654e-05, "loss": 3.661, "step": 35990 }, { "epoch": 1.00275968, "grad_norm": 0.8765255808830261, "learning_rate": 1.959417865156934e-05, "loss": 3.3435, "step": 36000 }, { "epoch": 1.00278528, "grad_norm": 0.7797889709472656, "learning_rate": 1.9593951541412325e-05, "loss": 3.1589, "step": 36010 }, { "epoch": 1.00281088, "grad_norm": 0.9062844514846802, "learning_rate": 1.9593724369041082e-05, "loss": 3.5214, "step": 36020 }, { "epoch": 1.00283648, "grad_norm": 0.7711690068244934, "learning_rate": 1.959349713445709e-05, "loss": 3.4113, "step": 36030 }, { "epoch": 1.00286208, "grad_norm": 0.8105394244194031, "learning_rate": 1.9593269837661818e-05, "loss": 3.4799, "step": 36040 }, { "epoch": 1.00288768, "grad_norm": 0.8641887307167053, "learning_rate": 1.959304247865674e-05, "loss": 3.4678, "step": 36050 }, { "epoch": 1.00291328, "grad_norm": 0.8537810444831848, "learning_rate": 1.959281505744333e-05, "loss": 3.6184, "step": 36060 }, { "epoch": 1.00293888, "grad_norm": 1.2121882438659668, "learning_rate": 1.9592587574023065e-05, "loss": 3.5484, "step": 36070 }, { "epoch": 1.00296448, "grad_norm": 0.8965107798576355, "learning_rate": 1.9592360028397423e-05, "loss": 3.4747, "step": 36080 }, { "epoch": 1.00299008, "grad_norm": 0.7705751061439514, "learning_rate": 1.959213242056787e-05, "loss": 3.2792, "step": 36090 }, { "epoch": 1.00301568, "grad_norm": 0.8706433176994324, "learning_rate": 1.959190475053589e-05, "loss": 3.5372, "step": 36100 }, { "epoch": 1.00304128, "grad_norm": 0.738684892654419, "learning_rate": 1.9591677018302953e-05, "loss": 3.5716, "step": 36110 }, { "epoch": 1.00306688, "grad_norm": 0.8105684518814087, "learning_rate": 1.9591449223870545e-05, "loss": 3.3633, "step": 36120 }, { "epoch": 1.00309248, "grad_norm": 1.0106794834136963, "learning_rate": 1.9591221367240134e-05, "loss": 3.5326, "step": 36130 }, { "epoch": 1.00311808, "grad_norm": 1.0810610055923462, "learning_rate": 1.9590993448413205e-05, "loss": 3.6042, "step": 36140 }, { "epoch": 1.00314368, "grad_norm": 0.7437257766723633, "learning_rate": 1.9590765467391232e-05, "loss": 3.4878, "step": 36150 }, { "epoch": 1.00316928, "grad_norm": 0.9970358610153198, "learning_rate": 1.959053742417569e-05, "loss": 3.6127, "step": 36160 }, { "epoch": 1.00319488, "grad_norm": 1.0468796491622925, "learning_rate": 1.9590309318768065e-05, "loss": 3.9844, "step": 36170 }, { "epoch": 1.00322048, "grad_norm": 1.1133619546890259, "learning_rate": 1.959008115116983e-05, "loss": 3.681, "step": 36180 }, { "epoch": 1.00324608, "grad_norm": 0.6956177949905396, "learning_rate": 1.958985292138247e-05, "loss": 3.585, "step": 36190 }, { "epoch": 1.00327168, "grad_norm": 0.7493845224380493, "learning_rate": 1.9589624629407464e-05, "loss": 3.39, "step": 36200 }, { "epoch": 1.00329728, "grad_norm": 0.8320032358169556, "learning_rate": 1.958939627524629e-05, "loss": 3.6013, "step": 36210 }, { "epoch": 1.00332288, "grad_norm": 0.9438285231590271, "learning_rate": 1.9589167858900426e-05, "loss": 3.4858, "step": 36220 }, { "epoch": 1.00334848, "grad_norm": 0.9758398532867432, "learning_rate": 1.9588939380371358e-05, "loss": 3.5178, "step": 36230 }, { "epoch": 1.00337408, "grad_norm": 3.5022923946380615, "learning_rate": 1.958871083966057e-05, "loss": 3.4382, "step": 36240 }, { "epoch": 1.00339968, "grad_norm": 1.1437456607818604, "learning_rate": 1.9588482236769535e-05, "loss": 3.5735, "step": 36250 }, { "epoch": 1.00342528, "grad_norm": 0.7234240770339966, "learning_rate": 1.9588253571699746e-05, "loss": 3.6965, "step": 36260 }, { "epoch": 1.00345088, "grad_norm": 1.2503994703292847, "learning_rate": 1.958802484445268e-05, "loss": 3.6706, "step": 36270 }, { "epoch": 1.00347648, "grad_norm": 0.7415453791618347, "learning_rate": 1.958779605502982e-05, "loss": 3.6292, "step": 36280 }, { "epoch": 1.00350208, "grad_norm": 0.6554509401321411, "learning_rate": 1.9587567203432653e-05, "loss": 3.3964, "step": 36290 }, { "epoch": 1.00352768, "grad_norm": 0.7508052587509155, "learning_rate": 1.9587338289662658e-05, "loss": 3.5709, "step": 36300 }, { "epoch": 1.00355328, "grad_norm": 0.8020958304405212, "learning_rate": 1.9587109313721324e-05, "loss": 3.4871, "step": 36310 }, { "epoch": 1.00357888, "grad_norm": 1.021205186843872, "learning_rate": 1.9586880275610134e-05, "loss": 3.7482, "step": 36320 }, { "epoch": 1.00360448, "grad_norm": 0.7347571849822998, "learning_rate": 1.958665117533057e-05, "loss": 3.4543, "step": 36330 }, { "epoch": 1.00363008, "grad_norm": 0.8837571144104004, "learning_rate": 1.9586422012884126e-05, "loss": 3.502, "step": 36340 }, { "epoch": 1.00365568, "grad_norm": 0.7923222184181213, "learning_rate": 1.958619278827228e-05, "loss": 3.4403, "step": 36350 }, { "epoch": 1.00368128, "grad_norm": 0.7712441086769104, "learning_rate": 1.958596350149652e-05, "loss": 3.5308, "step": 36360 }, { "epoch": 1.00370688, "grad_norm": 0.7756124138832092, "learning_rate": 1.958573415255834e-05, "loss": 3.4644, "step": 36370 }, { "epoch": 1.00373248, "grad_norm": 0.8402660489082336, "learning_rate": 1.9585504741459216e-05, "loss": 3.7784, "step": 36380 }, { "epoch": 1.00375808, "grad_norm": 0.6688399314880371, "learning_rate": 1.9585275268200642e-05, "loss": 3.3648, "step": 36390 }, { "epoch": 1.00378368, "grad_norm": 2.1768972873687744, "learning_rate": 1.9585045732784106e-05, "loss": 3.7013, "step": 36400 }, { "epoch": 1.00380928, "grad_norm": 0.7815853953361511, "learning_rate": 1.95848161352111e-05, "loss": 3.4683, "step": 36410 }, { "epoch": 1.00383488, "grad_norm": 0.7890573740005493, "learning_rate": 1.9584586475483104e-05, "loss": 3.2102, "step": 36420 }, { "epoch": 1.00386048, "grad_norm": 0.8473378419876099, "learning_rate": 1.9584356753601613e-05, "loss": 3.3522, "step": 36430 }, { "epoch": 1.00388608, "grad_norm": 0.8533082008361816, "learning_rate": 1.9584126969568117e-05, "loss": 3.3254, "step": 36440 }, { "epoch": 1.00391168, "grad_norm": 0.7424465417861938, "learning_rate": 1.95838971233841e-05, "loss": 3.5598, "step": 36450 }, { "epoch": 1.00393728, "grad_norm": 0.8488658666610718, "learning_rate": 1.9583667215051064e-05, "loss": 3.3906, "step": 36460 }, { "epoch": 1.00396288, "grad_norm": 0.8709390759468079, "learning_rate": 1.9583437244570488e-05, "loss": 3.7704, "step": 36470 }, { "epoch": 1.00398848, "grad_norm": 0.8249363303184509, "learning_rate": 1.958320721194387e-05, "loss": 3.465, "step": 36480 }, { "epoch": 1.00401408, "grad_norm": 0.699187159538269, "learning_rate": 1.9582977117172703e-05, "loss": 3.4559, "step": 36490 }, { "epoch": 1.00403968, "grad_norm": 0.8718985319137573, "learning_rate": 1.958274696025847e-05, "loss": 3.3002, "step": 36500 }, { "epoch": 1.00406528, "grad_norm": 0.7696202397346497, "learning_rate": 1.9582516741202674e-05, "loss": 3.378, "step": 36510 }, { "epoch": 1.00409088, "grad_norm": 0.7639551162719727, "learning_rate": 1.9582286460006804e-05, "loss": 3.5469, "step": 36520 }, { "epoch": 1.00411648, "grad_norm": 0.7012820243835449, "learning_rate": 1.958205611667235e-05, "loss": 3.3758, "step": 36530 }, { "epoch": 1.00414208, "grad_norm": 0.7624715566635132, "learning_rate": 1.9581825711200814e-05, "loss": 3.4286, "step": 36540 }, { "epoch": 1.00416768, "grad_norm": 0.8000349402427673, "learning_rate": 1.958159524359368e-05, "loss": 3.8098, "step": 36550 }, { "epoch": 1.00419328, "grad_norm": 0.7472072243690491, "learning_rate": 1.958136471385245e-05, "loss": 3.5159, "step": 36560 }, { "epoch": 1.00421888, "grad_norm": 0.7336896061897278, "learning_rate": 1.9581134121978617e-05, "loss": 3.4862, "step": 36570 }, { "epoch": 1.00424448, "grad_norm": 0.8226104974746704, "learning_rate": 1.9580903467973675e-05, "loss": 3.5101, "step": 36580 }, { "epoch": 1.00427008, "grad_norm": 0.7767869830131531, "learning_rate": 1.9580672751839116e-05, "loss": 3.2553, "step": 36590 }, { "epoch": 1.00429568, "grad_norm": 0.7694026231765747, "learning_rate": 1.9580441973576443e-05, "loss": 3.5221, "step": 36600 }, { "epoch": 1.00432128, "grad_norm": 0.7052871584892273, "learning_rate": 1.9580211133187152e-05, "loss": 3.3898, "step": 36610 }, { "epoch": 1.00434688, "grad_norm": 0.897311270236969, "learning_rate": 1.9579980230672734e-05, "loss": 3.5659, "step": 36620 }, { "epoch": 1.00437248, "grad_norm": 0.9811800122261047, "learning_rate": 1.9579749266034694e-05, "loss": 3.4149, "step": 36630 }, { "epoch": 1.00439808, "grad_norm": 0.812889814376831, "learning_rate": 1.9579518239274523e-05, "loss": 3.423, "step": 36640 }, { "epoch": 1.00442368, "grad_norm": 0.7265908122062683, "learning_rate": 1.9579287150393727e-05, "loss": 3.5411, "step": 36650 }, { "epoch": 1.00444928, "grad_norm": 0.7402195334434509, "learning_rate": 1.9579055999393795e-05, "loss": 3.431, "step": 36660 }, { "epoch": 1.00447488, "grad_norm": 0.8000377416610718, "learning_rate": 1.957882478627623e-05, "loss": 3.4056, "step": 36670 }, { "epoch": 1.00450048, "grad_norm": 0.8880795240402222, "learning_rate": 1.9578593511042536e-05, "loss": 3.4392, "step": 36680 }, { "epoch": 1.00452608, "grad_norm": 0.6998052000999451, "learning_rate": 1.9578362173694207e-05, "loss": 3.4405, "step": 36690 }, { "epoch": 1.00455168, "grad_norm": 0.7037399411201477, "learning_rate": 1.9578130774232746e-05, "loss": 3.5581, "step": 36700 }, { "epoch": 1.00457728, "grad_norm": 0.9875528812408447, "learning_rate": 1.957789931265965e-05, "loss": 3.4541, "step": 36710 }, { "epoch": 1.00460288, "grad_norm": 0.9637550115585327, "learning_rate": 1.9577667788976425e-05, "loss": 3.4579, "step": 36720 }, { "epoch": 1.00462848, "grad_norm": 0.7454763650894165, "learning_rate": 1.9577436203184568e-05, "loss": 3.4636, "step": 36730 }, { "epoch": 1.00465408, "grad_norm": 0.6686562299728394, "learning_rate": 1.957720455528558e-05, "loss": 3.3738, "step": 36740 }, { "epoch": 1.00467968, "grad_norm": 0.8009961247444153, "learning_rate": 1.9576972845280974e-05, "loss": 3.2722, "step": 36750 }, { "epoch": 1.00470528, "grad_norm": 0.780605137348175, "learning_rate": 1.9576741073172236e-05, "loss": 3.3408, "step": 36760 }, { "epoch": 1.00473088, "grad_norm": 0.8768825531005859, "learning_rate": 1.9576509238960882e-05, "loss": 3.4715, "step": 36770 }, { "epoch": 1.00475648, "grad_norm": 0.7349922060966492, "learning_rate": 1.957627734264841e-05, "loss": 3.3465, "step": 36780 }, { "epoch": 1.00478208, "grad_norm": 0.9299008846282959, "learning_rate": 1.9576045384236324e-05, "loss": 3.1213, "step": 36790 }, { "epoch": 1.00480768, "grad_norm": 1.0427372455596924, "learning_rate": 1.9575813363726128e-05, "loss": 3.4966, "step": 36800 }, { "epoch": 1.00483328, "grad_norm": 0.8159831166267395, "learning_rate": 1.957558128111933e-05, "loss": 3.5826, "step": 36810 }, { "epoch": 1.00485888, "grad_norm": 0.7803835272789001, "learning_rate": 1.9575349136417427e-05, "loss": 3.1963, "step": 36820 }, { "epoch": 1.00488448, "grad_norm": 0.8020631074905396, "learning_rate": 1.9575116929621935e-05, "loss": 3.6389, "step": 36830 }, { "epoch": 1.00491008, "grad_norm": 0.8054378032684326, "learning_rate": 1.9574884660734353e-05, "loss": 3.5104, "step": 36840 }, { "epoch": 1.00493568, "grad_norm": 0.982232391834259, "learning_rate": 1.9574652329756184e-05, "loss": 3.5586, "step": 36850 }, { "epoch": 1.00496128, "grad_norm": 0.7345327138900757, "learning_rate": 1.9574419936688947e-05, "loss": 3.489, "step": 36860 }, { "epoch": 1.00498688, "grad_norm": 3.0115771293640137, "learning_rate": 1.957418748153414e-05, "loss": 4.1849, "step": 36870 }, { "epoch": 1.00501248, "grad_norm": 0.8478348255157471, "learning_rate": 1.9573954964293265e-05, "loss": 3.6662, "step": 36880 }, { "epoch": 1.00503808, "grad_norm": 0.7998169660568237, "learning_rate": 1.957372238496784e-05, "loss": 3.3741, "step": 36890 }, { "epoch": 1.00506368, "grad_norm": 1.089414119720459, "learning_rate": 1.957348974355937e-05, "loss": 3.7156, "step": 36900 }, { "epoch": 1.00508928, "grad_norm": 0.7949447631835938, "learning_rate": 1.9573257040069365e-05, "loss": 3.5483, "step": 36910 }, { "epoch": 1.00511488, "grad_norm": 0.9031728506088257, "learning_rate": 1.9573024274499332e-05, "loss": 3.4417, "step": 36920 }, { "epoch": 1.00514048, "grad_norm": 0.7657273411750793, "learning_rate": 1.957279144685078e-05, "loss": 3.6039, "step": 36930 }, { "epoch": 1.00516608, "grad_norm": 0.9200237989425659, "learning_rate": 1.957255855712522e-05, "loss": 3.1968, "step": 36940 }, { "epoch": 1.00519168, "grad_norm": 0.7976690530776978, "learning_rate": 1.957232560532416e-05, "loss": 3.4082, "step": 36950 }, { "epoch": 1.00521728, "grad_norm": 0.779694676399231, "learning_rate": 1.9572092591449114e-05, "loss": 3.3091, "step": 36960 }, { "epoch": 1.00524288, "grad_norm": 0.8937925100326538, "learning_rate": 1.957185951550159e-05, "loss": 3.4471, "step": 36970 }, { "epoch": 1.00526848, "grad_norm": 0.9778732657432556, "learning_rate": 1.9571626377483105e-05, "loss": 3.5191, "step": 36980 }, { "epoch": 1.00529408, "grad_norm": 0.7947215437889099, "learning_rate": 1.9571393177395165e-05, "loss": 3.4136, "step": 36990 }, { "epoch": 1.00531968, "grad_norm": 0.868006706237793, "learning_rate": 1.957115991523928e-05, "loss": 3.4763, "step": 37000 }, { "epoch": 1.00534528, "grad_norm": 0.8605491518974304, "learning_rate": 1.957092659101697e-05, "loss": 3.5442, "step": 37010 }, { "epoch": 1.00537088, "grad_norm": 0.7468289136886597, "learning_rate": 1.9570693204729743e-05, "loss": 3.5031, "step": 37020 }, { "epoch": 1.00539648, "grad_norm": 0.9073086380958557, "learning_rate": 1.957045975637912e-05, "loss": 3.324, "step": 37030 }, { "epoch": 1.00542208, "grad_norm": 1.1382519006729126, "learning_rate": 1.9570226245966603e-05, "loss": 3.6607, "step": 37040 }, { "epoch": 1.00544768, "grad_norm": 0.8523823618888855, "learning_rate": 1.9569992673493713e-05, "loss": 3.5372, "step": 37050 }, { "epoch": 1.00547328, "grad_norm": 0.8041700124740601, "learning_rate": 1.9569759038961965e-05, "loss": 3.8422, "step": 37060 }, { "epoch": 1.00549888, "grad_norm": 1.185334324836731, "learning_rate": 1.9569525342372873e-05, "loss": 3.7881, "step": 37070 }, { "epoch": 1.00552448, "grad_norm": 0.8231343030929565, "learning_rate": 1.9569291583727952e-05, "loss": 3.3995, "step": 37080 }, { "epoch": 1.00555008, "grad_norm": 0.8616853952407837, "learning_rate": 1.9569057763028717e-05, "loss": 3.4384, "step": 37090 }, { "epoch": 1.00557568, "grad_norm": 0.8467494249343872, "learning_rate": 1.9568823880276688e-05, "loss": 3.6779, "step": 37100 }, { "epoch": 1.00560128, "grad_norm": 1.4732980728149414, "learning_rate": 1.956858993547338e-05, "loss": 3.6428, "step": 37110 }, { "epoch": 1.00562688, "grad_norm": 0.7664240598678589, "learning_rate": 1.9568355928620307e-05, "loss": 3.4703, "step": 37120 }, { "epoch": 1.00565248, "grad_norm": 0.8794015049934387, "learning_rate": 1.956812185971899e-05, "loss": 3.3789, "step": 37130 }, { "epoch": 1.00567808, "grad_norm": 0.757449209690094, "learning_rate": 1.9567887728770943e-05, "loss": 3.5201, "step": 37140 }, { "epoch": 1.00570368, "grad_norm": 1.1339365243911743, "learning_rate": 1.9567653535777688e-05, "loss": 3.4379, "step": 37150 }, { "epoch": 1.00572928, "grad_norm": 0.7894696593284607, "learning_rate": 1.9567419280740746e-05, "loss": 3.334, "step": 37160 }, { "epoch": 1.00575488, "grad_norm": 0.7136163711547852, "learning_rate": 1.956718496366163e-05, "loss": 3.5841, "step": 37170 }, { "epoch": 1.00578048, "grad_norm": 0.8015162348747253, "learning_rate": 1.9566950584541863e-05, "loss": 3.4688, "step": 37180 }, { "epoch": 1.00580608, "grad_norm": 0.7482042908668518, "learning_rate": 1.9566716143382963e-05, "loss": 3.321, "step": 37190 }, { "epoch": 1.00583168, "grad_norm": 0.8837457895278931, "learning_rate": 1.956648164018645e-05, "loss": 3.6297, "step": 37200 }, { "epoch": 1.00585728, "grad_norm": 0.7388659119606018, "learning_rate": 1.9566247074953848e-05, "loss": 3.584, "step": 37210 }, { "epoch": 1.00588288, "grad_norm": 0.7275218963623047, "learning_rate": 1.956601244768668e-05, "loss": 3.3025, "step": 37220 }, { "epoch": 1.00590848, "grad_norm": 0.7745441794395447, "learning_rate": 1.956577775838646e-05, "loss": 3.4384, "step": 37230 }, { "epoch": 1.00593408, "grad_norm": 0.8121082186698914, "learning_rate": 1.956554300705471e-05, "loss": 3.5214, "step": 37240 }, { "epoch": 1.00595968, "grad_norm": 0.8126327395439148, "learning_rate": 1.9565308193692958e-05, "loss": 3.4364, "step": 37250 }, { "epoch": 1.00598528, "grad_norm": 0.7797713279724121, "learning_rate": 1.9565073318302726e-05, "loss": 3.5664, "step": 37260 }, { "epoch": 1.00601088, "grad_norm": 0.7772783637046814, "learning_rate": 1.9564838380885532e-05, "loss": 3.3315, "step": 37270 }, { "epoch": 1.00603648, "grad_norm": 0.8342816233634949, "learning_rate": 1.9564603381442908e-05, "loss": 3.5387, "step": 37280 }, { "epoch": 1.00606208, "grad_norm": 0.9273583889007568, "learning_rate": 1.9564368319976373e-05, "loss": 3.6237, "step": 37290 }, { "epoch": 1.00608768, "grad_norm": 0.8936719298362732, "learning_rate": 1.956413319648745e-05, "loss": 3.518, "step": 37300 }, { "epoch": 1.00611328, "grad_norm": 0.8165277242660522, "learning_rate": 1.956389801097766e-05, "loss": 3.6234, "step": 37310 }, { "epoch": 1.00613888, "grad_norm": 0.6914087533950806, "learning_rate": 1.956366276344854e-05, "loss": 3.5603, "step": 37320 }, { "epoch": 1.00616448, "grad_norm": 0.9154125452041626, "learning_rate": 1.9563427453901606e-05, "loss": 3.4317, "step": 37330 }, { "epoch": 1.00619008, "grad_norm": 0.9250445365905762, "learning_rate": 1.956319208233839e-05, "loss": 3.6395, "step": 37340 }, { "epoch": 1.00621568, "grad_norm": 1.070394515991211, "learning_rate": 1.956295664876041e-05, "loss": 3.8156, "step": 37350 }, { "epoch": 1.00624128, "grad_norm": 0.8368149399757385, "learning_rate": 1.95627211531692e-05, "loss": 3.4566, "step": 37360 }, { "epoch": 1.00626688, "grad_norm": 0.8527957201004028, "learning_rate": 1.9562485595566284e-05, "loss": 3.3611, "step": 37370 }, { "epoch": 1.00629248, "grad_norm": 1.19145929813385, "learning_rate": 1.9562249975953193e-05, "loss": 3.5788, "step": 37380 }, { "epoch": 1.00631808, "grad_norm": 0.6970329880714417, "learning_rate": 1.956201429433145e-05, "loss": 3.614, "step": 37390 }, { "epoch": 1.00634368, "grad_norm": 0.9251118302345276, "learning_rate": 1.9561778550702583e-05, "loss": 3.6184, "step": 37400 }, { "epoch": 1.00636928, "grad_norm": 0.8025612831115723, "learning_rate": 1.9561542745068128e-05, "loss": 3.6995, "step": 37410 }, { "epoch": 1.00639488, "grad_norm": 0.8247721195220947, "learning_rate": 1.9561306877429604e-05, "loss": 3.3029, "step": 37420 }, { "epoch": 1.00642048, "grad_norm": 0.9255365133285522, "learning_rate": 1.956107094778855e-05, "loss": 3.4007, "step": 37430 }, { "epoch": 1.00644608, "grad_norm": 0.749237596988678, "learning_rate": 1.9560834956146492e-05, "loss": 3.456, "step": 37440 }, { "epoch": 1.00647168, "grad_norm": 0.8891429305076599, "learning_rate": 1.956059890250496e-05, "loss": 3.6085, "step": 37450 }, { "epoch": 1.00649728, "grad_norm": 0.7182863354682922, "learning_rate": 1.9560362786865482e-05, "loss": 3.34, "step": 37460 }, { "epoch": 1.00652288, "grad_norm": 0.7835971713066101, "learning_rate": 1.9560126609229596e-05, "loss": 3.6207, "step": 37470 }, { "epoch": 1.00654848, "grad_norm": 0.9386635422706604, "learning_rate": 1.955989036959883e-05, "loss": 3.5913, "step": 37480 }, { "epoch": 1.00657408, "grad_norm": 0.675503134727478, "learning_rate": 1.9559654067974715e-05, "loss": 3.5254, "step": 37490 }, { "epoch": 1.00659968, "grad_norm": 0.9340787529945374, "learning_rate": 1.955941770435878e-05, "loss": 3.6992, "step": 37500 }, { "epoch": 1.00662528, "grad_norm": 1.3815706968307495, "learning_rate": 1.9559181278752565e-05, "loss": 3.5986, "step": 37510 }, { "epoch": 1.00665088, "grad_norm": 0.7895622849464417, "learning_rate": 1.9558944791157603e-05, "loss": 3.3154, "step": 37520 }, { "epoch": 1.00667648, "grad_norm": 0.7398686408996582, "learning_rate": 1.9558708241575418e-05, "loss": 3.4734, "step": 37530 }, { "epoch": 1.00670208, "grad_norm": 0.9179908037185669, "learning_rate": 1.9558471630007554e-05, "loss": 3.4635, "step": 37540 }, { "epoch": 1.00672768, "grad_norm": 0.9834953546524048, "learning_rate": 1.955823495645554e-05, "loss": 3.4999, "step": 37550 }, { "epoch": 1.00675328, "grad_norm": 0.8713352680206299, "learning_rate": 1.9557998220920917e-05, "loss": 3.3656, "step": 37560 }, { "epoch": 1.00677888, "grad_norm": 0.7954431772232056, "learning_rate": 1.9557761423405212e-05, "loss": 3.6449, "step": 37570 }, { "epoch": 1.00680448, "grad_norm": 0.7694615125656128, "learning_rate": 1.9557524563909968e-05, "loss": 3.5282, "step": 37580 }, { "epoch": 1.00683008, "grad_norm": 0.8899155855178833, "learning_rate": 1.9557287642436714e-05, "loss": 3.5009, "step": 37590 }, { "epoch": 1.00685568, "grad_norm": 0.8127608299255371, "learning_rate": 1.955705065898699e-05, "loss": 3.4169, "step": 37600 }, { "epoch": 1.00688128, "grad_norm": 0.9192294478416443, "learning_rate": 1.9556813613562333e-05, "loss": 3.4722, "step": 37610 }, { "epoch": 1.00690688, "grad_norm": 1.0550103187561035, "learning_rate": 1.9556576506164276e-05, "loss": 3.3995, "step": 37620 }, { "epoch": 1.00693248, "grad_norm": 0.7823398113250732, "learning_rate": 1.9556339336794365e-05, "loss": 3.299, "step": 37630 }, { "epoch": 1.00695808, "grad_norm": 1.0499087572097778, "learning_rate": 1.9556102105454134e-05, "loss": 3.5177, "step": 37640 }, { "epoch": 1.00698368, "grad_norm": 0.850179135799408, "learning_rate": 1.9555864812145115e-05, "loss": 3.5003, "step": 37650 }, { "epoch": 1.00700928, "grad_norm": 1.0187066793441772, "learning_rate": 1.9555627456868857e-05, "loss": 3.6893, "step": 37660 }, { "epoch": 1.00703488, "grad_norm": 0.7980759739875793, "learning_rate": 1.9555390039626893e-05, "loss": 3.3477, "step": 37670 }, { "epoch": 1.00706048, "grad_norm": 0.880274772644043, "learning_rate": 1.9555152560420763e-05, "loss": 3.3375, "step": 37680 }, { "epoch": 1.00708608, "grad_norm": 0.7671177983283997, "learning_rate": 1.955491501925201e-05, "loss": 3.3501, "step": 37690 }, { "epoch": 1.00711168, "grad_norm": 0.7384423017501831, "learning_rate": 1.9554677416122172e-05, "loss": 3.4298, "step": 37700 }, { "epoch": 1.00713728, "grad_norm": 0.752685010433197, "learning_rate": 1.955443975103279e-05, "loss": 3.598, "step": 37710 }, { "epoch": 1.00716288, "grad_norm": 0.7510886788368225, "learning_rate": 1.9554202023985404e-05, "loss": 3.4977, "step": 37720 }, { "epoch": 1.00718848, "grad_norm": 0.8547483682632446, "learning_rate": 1.955396423498156e-05, "loss": 3.3501, "step": 37730 }, { "epoch": 1.00721408, "grad_norm": 0.7446702718734741, "learning_rate": 1.9553726384022797e-05, "loss": 3.4791, "step": 37740 }, { "epoch": 1.00723968, "grad_norm": 0.7382928729057312, "learning_rate": 1.9553488471110654e-05, "loss": 3.641, "step": 37750 }, { "epoch": 1.00726528, "grad_norm": 0.7044509053230286, "learning_rate": 1.9553250496246678e-05, "loss": 3.3812, "step": 37760 }, { "epoch": 1.00729088, "grad_norm": 0.8012375235557556, "learning_rate": 1.9553012459432416e-05, "loss": 3.3641, "step": 37770 }, { "epoch": 1.00731648, "grad_norm": 0.8028831481933594, "learning_rate": 1.9552774360669403e-05, "loss": 3.3767, "step": 37780 }, { "epoch": 1.00734208, "grad_norm": 1.087914228439331, "learning_rate": 1.9552536199959188e-05, "loss": 3.2318, "step": 37790 }, { "epoch": 1.00736768, "grad_norm": 0.9997387528419495, "learning_rate": 1.9552297977303315e-05, "loss": 3.3805, "step": 37800 }, { "epoch": 1.00739328, "grad_norm": 0.7862571477890015, "learning_rate": 1.9552059692703327e-05, "loss": 3.5603, "step": 37810 }, { "epoch": 1.00741888, "grad_norm": 0.8985859155654907, "learning_rate": 1.9551821346160772e-05, "loss": 3.4406, "step": 37820 }, { "epoch": 1.00744448, "grad_norm": 0.8496285676956177, "learning_rate": 1.9551582937677194e-05, "loss": 3.4519, "step": 37830 }, { "epoch": 1.00747008, "grad_norm": 0.7637202739715576, "learning_rate": 1.955134446725414e-05, "loss": 3.4183, "step": 37840 }, { "epoch": 1.00749568, "grad_norm": 0.8242153525352478, "learning_rate": 1.955110593489315e-05, "loss": 3.5576, "step": 37850 }, { "epoch": 1.00752128, "grad_norm": 0.7852720022201538, "learning_rate": 1.9550867340595783e-05, "loss": 3.4116, "step": 37860 }, { "epoch": 1.00754688, "grad_norm": 0.7352519631385803, "learning_rate": 1.955062868436358e-05, "loss": 3.1786, "step": 37870 }, { "epoch": 1.00757248, "grad_norm": 0.7574599385261536, "learning_rate": 1.955038996619808e-05, "loss": 3.2822, "step": 37880 }, { "epoch": 1.00759808, "grad_norm": 0.9491204619407654, "learning_rate": 1.9550151186100847e-05, "loss": 3.4006, "step": 37890 }, { "epoch": 1.00762368, "grad_norm": 0.79114830493927, "learning_rate": 1.954991234407342e-05, "loss": 3.5243, "step": 37900 }, { "epoch": 1.00764928, "grad_norm": 0.8672481775283813, "learning_rate": 1.954967344011735e-05, "loss": 3.6805, "step": 37910 }, { "epoch": 1.00767488, "grad_norm": 0.7339709997177124, "learning_rate": 1.9549434474234183e-05, "loss": 3.5516, "step": 37920 }, { "epoch": 1.00770048, "grad_norm": 0.7972648739814758, "learning_rate": 1.9549195446425475e-05, "loss": 3.4061, "step": 37930 }, { "epoch": 1.00772608, "grad_norm": 0.840208888053894, "learning_rate": 1.954895635669277e-05, "loss": 3.589, "step": 37940 }, { "epoch": 1.00775168, "grad_norm": 0.8563634157180786, "learning_rate": 1.954871720503762e-05, "loss": 3.5024, "step": 37950 }, { "epoch": 1.00777728, "grad_norm": 0.8246386647224426, "learning_rate": 1.954847799146158e-05, "loss": 3.5345, "step": 37960 }, { "epoch": 1.00780288, "grad_norm": 1.8322172164916992, "learning_rate": 1.9548238715966194e-05, "loss": 3.5362, "step": 37970 }, { "epoch": 1.00782848, "grad_norm": 0.7237630486488342, "learning_rate": 1.954799937855302e-05, "loss": 3.3273, "step": 37980 }, { "epoch": 1.00785408, "grad_norm": 0.8094584345817566, "learning_rate": 1.9547759979223607e-05, "loss": 3.534, "step": 37990 }, { "epoch": 1.00787968, "grad_norm": 0.8554574847221375, "learning_rate": 1.954752051797951e-05, "loss": 3.4156, "step": 38000 }, { "epoch": 1.00790528, "grad_norm": 0.7931087613105774, "learning_rate": 1.9547280994822277e-05, "loss": 3.3504, "step": 38010 }, { "epoch": 1.00793088, "grad_norm": 0.8883897662162781, "learning_rate": 1.9547041409753465e-05, "loss": 3.4928, "step": 38020 }, { "epoch": 1.00795648, "grad_norm": 0.846890926361084, "learning_rate": 1.9546801762774626e-05, "loss": 3.5233, "step": 38030 }, { "epoch": 1.00798208, "grad_norm": 0.8166642785072327, "learning_rate": 1.9546562053887314e-05, "loss": 3.4197, "step": 38040 }, { "epoch": 1.00800768, "grad_norm": 0.7133126258850098, "learning_rate": 1.9546322283093088e-05, "loss": 3.4743, "step": 38050 }, { "epoch": 1.00803328, "grad_norm": 0.7769073843955994, "learning_rate": 1.9546082450393496e-05, "loss": 3.3062, "step": 38060 }, { "epoch": 1.00805888, "grad_norm": 0.8805080652236938, "learning_rate": 1.9545842555790098e-05, "loss": 3.3003, "step": 38070 }, { "epoch": 1.00808448, "grad_norm": 0.8310803771018982, "learning_rate": 1.9545602599284447e-05, "loss": 3.3909, "step": 38080 }, { "epoch": 1.00811008, "grad_norm": 0.7961034774780273, "learning_rate": 1.95453625808781e-05, "loss": 3.3721, "step": 38090 }, { "epoch": 1.00813568, "grad_norm": 0.9111803770065308, "learning_rate": 1.9545122500572616e-05, "loss": 3.5396, "step": 38100 }, { "epoch": 1.00816128, "grad_norm": 0.7046401500701904, "learning_rate": 1.9544882358369547e-05, "loss": 3.478, "step": 38110 }, { "epoch": 1.00818688, "grad_norm": 0.7360559105873108, "learning_rate": 1.954464215427045e-05, "loss": 3.479, "step": 38120 }, { "epoch": 1.00821248, "grad_norm": 0.7943379878997803, "learning_rate": 1.954440188827689e-05, "loss": 3.4476, "step": 38130 }, { "epoch": 1.00823808, "grad_norm": 0.8776357173919678, "learning_rate": 1.9544161560390416e-05, "loss": 3.5233, "step": 38140 }, { "epoch": 1.00826368, "grad_norm": 0.7685678601264954, "learning_rate": 1.9543921170612593e-05, "loss": 3.6604, "step": 38150 }, { "epoch": 1.00828928, "grad_norm": 0.8674782514572144, "learning_rate": 1.9543680718944983e-05, "loss": 3.6567, "step": 38160 }, { "epoch": 1.00831488, "grad_norm": 0.7285463213920593, "learning_rate": 1.954344020538913e-05, "loss": 3.5123, "step": 38170 }, { "epoch": 1.00834048, "grad_norm": 0.9006087183952332, "learning_rate": 1.954319962994661e-05, "loss": 3.3028, "step": 38180 }, { "epoch": 1.00836608, "grad_norm": 1.2645950317382812, "learning_rate": 1.9542958992618975e-05, "loss": 3.3613, "step": 38190 }, { "epoch": 1.00839168, "grad_norm": 1.1479216814041138, "learning_rate": 1.954271829340779e-05, "loss": 3.6117, "step": 38200 }, { "epoch": 1.00841728, "grad_norm": 0.9072645902633667, "learning_rate": 1.9542477532314608e-05, "loss": 3.4349, "step": 38210 }, { "epoch": 1.00844288, "grad_norm": 0.960934579372406, "learning_rate": 1.9542236709341e-05, "loss": 3.5373, "step": 38220 }, { "epoch": 1.00846848, "grad_norm": 0.7747402787208557, "learning_rate": 1.954199582448852e-05, "loss": 3.2715, "step": 38230 }, { "epoch": 1.00849408, "grad_norm": 0.9734393358230591, "learning_rate": 1.9541754877758733e-05, "loss": 3.4763, "step": 38240 }, { "epoch": 1.00851968, "grad_norm": 0.9835291504859924, "learning_rate": 1.9541513869153204e-05, "loss": 3.5733, "step": 38250 }, { "epoch": 1.00854528, "grad_norm": 0.7621899247169495, "learning_rate": 1.954127279867349e-05, "loss": 3.354, "step": 38260 }, { "epoch": 1.00857088, "grad_norm": 0.843529224395752, "learning_rate": 1.954103166632116e-05, "loss": 3.3129, "step": 38270 }, { "epoch": 1.00859648, "grad_norm": 0.9613127112388611, "learning_rate": 1.9540790472097774e-05, "loss": 3.386, "step": 38280 }, { "epoch": 1.00862208, "grad_norm": 0.8488774299621582, "learning_rate": 1.95405492160049e-05, "loss": 3.3677, "step": 38290 }, { "epoch": 1.00864768, "grad_norm": 0.8541886806488037, "learning_rate": 1.9540307898044096e-05, "loss": 3.3565, "step": 38300 }, { "epoch": 1.00867328, "grad_norm": 0.8689907789230347, "learning_rate": 1.9540066518216932e-05, "loss": 3.4145, "step": 38310 }, { "epoch": 1.00869888, "grad_norm": 0.7206857800483704, "learning_rate": 1.9539825076524975e-05, "loss": 3.5915, "step": 38320 }, { "epoch": 1.00872448, "grad_norm": 1.0066014528274536, "learning_rate": 1.9539583572969784e-05, "loss": 3.4195, "step": 38330 }, { "epoch": 1.00875008, "grad_norm": 0.8019399642944336, "learning_rate": 1.953934200755293e-05, "loss": 3.5962, "step": 38340 }, { "epoch": 1.00877568, "grad_norm": 0.8779288530349731, "learning_rate": 1.9539100380275978e-05, "loss": 3.2682, "step": 38350 }, { "epoch": 1.00880128, "grad_norm": 0.7966319918632507, "learning_rate": 1.9538858691140496e-05, "loss": 3.403, "step": 38360 }, { "epoch": 1.00882688, "grad_norm": 0.7994953393936157, "learning_rate": 1.953861694014805e-05, "loss": 3.4146, "step": 38370 }, { "epoch": 1.00885248, "grad_norm": 0.7058055400848389, "learning_rate": 1.953837512730021e-05, "loss": 3.3517, "step": 38380 }, { "epoch": 1.00887808, "grad_norm": 1.7169275283813477, "learning_rate": 1.9538133252598542e-05, "loss": 3.4683, "step": 38390 }, { "epoch": 1.00890368, "grad_norm": 1.0589447021484375, "learning_rate": 1.953789131604461e-05, "loss": 3.4708, "step": 38400 }, { "epoch": 1.00892928, "grad_norm": 0.7613821029663086, "learning_rate": 1.9537649317639996e-05, "loss": 3.6564, "step": 38410 }, { "epoch": 1.00895488, "grad_norm": 0.8835191130638123, "learning_rate": 1.9537407257386255e-05, "loss": 3.7181, "step": 38420 }, { "epoch": 1.00898048, "grad_norm": 0.844754159450531, "learning_rate": 1.9537165135284962e-05, "loss": 3.5068, "step": 38430 }, { "epoch": 1.00900608, "grad_norm": 0.7207252383232117, "learning_rate": 1.9536922951337695e-05, "loss": 3.3487, "step": 38440 }, { "epoch": 1.00903168, "grad_norm": 0.9031351208686829, "learning_rate": 1.9536680705546008e-05, "loss": 3.8347, "step": 38450 }, { "epoch": 1.00905728, "grad_norm": 0.8062012195587158, "learning_rate": 1.9536438397911486e-05, "loss": 3.5945, "step": 38460 }, { "epoch": 1.00908288, "grad_norm": 0.8931095600128174, "learning_rate": 1.9536196028435696e-05, "loss": 3.3331, "step": 38470 }, { "epoch": 1.00910848, "grad_norm": 0.9386063814163208, "learning_rate": 1.953595359712021e-05, "loss": 3.5199, "step": 38480 }, { "epoch": 1.00913408, "grad_norm": 0.8118954300880432, "learning_rate": 1.9535711103966596e-05, "loss": 3.5249, "step": 38490 }, { "epoch": 1.00915968, "grad_norm": 1.556565523147583, "learning_rate": 1.9535468548976428e-05, "loss": 3.3449, "step": 38500 }, { "epoch": 1.00918528, "grad_norm": 0.8110575675964355, "learning_rate": 1.953522593215129e-05, "loss": 3.2929, "step": 38510 }, { "epoch": 1.00921088, "grad_norm": 0.9146286845207214, "learning_rate": 1.9534983253492738e-05, "loss": 3.5673, "step": 38520 }, { "epoch": 1.00923648, "grad_norm": 0.8003819584846497, "learning_rate": 1.9534740513002354e-05, "loss": 3.4286, "step": 38530 }, { "epoch": 1.00926208, "grad_norm": 0.8477046489715576, "learning_rate": 1.9534497710681713e-05, "loss": 3.5553, "step": 38540 }, { "epoch": 1.00928768, "grad_norm": 0.7855861783027649, "learning_rate": 1.953425484653239e-05, "loss": 3.4109, "step": 38550 }, { "epoch": 1.00931328, "grad_norm": 0.7172048687934875, "learning_rate": 1.9534011920555956e-05, "loss": 3.377, "step": 38560 }, { "epoch": 1.00933888, "grad_norm": 0.8352746367454529, "learning_rate": 1.9533768932753988e-05, "loss": 3.5262, "step": 38570 }, { "epoch": 1.00936448, "grad_norm": 0.8011744618415833, "learning_rate": 1.9533525883128064e-05, "loss": 3.5267, "step": 38580 }, { "epoch": 1.00939008, "grad_norm": 0.7180629968643188, "learning_rate": 1.9533282771679757e-05, "loss": 3.5742, "step": 38590 }, { "epoch": 1.00941568, "grad_norm": 0.825060248374939, "learning_rate": 1.9533039598410645e-05, "loss": 3.4625, "step": 38600 }, { "epoch": 1.00944128, "grad_norm": 0.7291908860206604, "learning_rate": 1.9532796363322307e-05, "loss": 3.4954, "step": 38610 }, { "epoch": 1.00946688, "grad_norm": 0.8341073393821716, "learning_rate": 1.9532553066416316e-05, "loss": 3.4553, "step": 38620 }, { "epoch": 1.00949248, "grad_norm": 0.9267721772193909, "learning_rate": 1.9532309707694253e-05, "loss": 3.5169, "step": 38630 }, { "epoch": 1.00951808, "grad_norm": 0.8356528878211975, "learning_rate": 1.9532066287157694e-05, "loss": 3.3214, "step": 38640 }, { "epoch": 1.00954368, "grad_norm": 0.8886595964431763, "learning_rate": 1.9531822804808218e-05, "loss": 3.4321, "step": 38650 }, { "epoch": 1.00956928, "grad_norm": 0.7872552871704102, "learning_rate": 1.9531579260647407e-05, "loss": 3.6286, "step": 38660 }, { "epoch": 1.00959488, "grad_norm": 0.8159332871437073, "learning_rate": 1.9531335654676833e-05, "loss": 3.3674, "step": 38670 }, { "epoch": 1.00962048, "grad_norm": 0.849978506565094, "learning_rate": 1.9531091986898086e-05, "loss": 3.2683, "step": 38680 }, { "epoch": 1.00964608, "grad_norm": 0.9366746544837952, "learning_rate": 1.9530848257312737e-05, "loss": 3.5006, "step": 38690 }, { "epoch": 1.00967168, "grad_norm": 0.9085193276405334, "learning_rate": 1.953060446592237e-05, "loss": 3.6252, "step": 38700 }, { "epoch": 1.00969728, "grad_norm": 0.8360149264335632, "learning_rate": 1.9530360612728566e-05, "loss": 3.5711, "step": 38710 }, { "epoch": 1.00972288, "grad_norm": 0.6750808358192444, "learning_rate": 1.953011669773291e-05, "loss": 3.7378, "step": 38720 }, { "epoch": 1.00974848, "grad_norm": 1.0023386478424072, "learning_rate": 1.9529872720936976e-05, "loss": 3.4243, "step": 38730 }, { "epoch": 1.00977408, "grad_norm": 0.7031369805335999, "learning_rate": 1.9529628682342353e-05, "loss": 3.4078, "step": 38740 }, { "epoch": 1.00979968, "grad_norm": 0.9227405190467834, "learning_rate": 1.9529384581950618e-05, "loss": 3.4034, "step": 38750 }, { "epoch": 1.00982528, "grad_norm": 0.7856470346450806, "learning_rate": 1.9529140419763357e-05, "loss": 3.5356, "step": 38760 }, { "epoch": 1.00985088, "grad_norm": 0.7844724059104919, "learning_rate": 1.9528896195782153e-05, "loss": 3.3853, "step": 38770 }, { "epoch": 1.00987648, "grad_norm": 0.9667842984199524, "learning_rate": 1.952865191000859e-05, "loss": 3.3023, "step": 38780 }, { "epoch": 1.00990208, "grad_norm": 0.8436921834945679, "learning_rate": 1.9528407562444253e-05, "loss": 3.489, "step": 38790 }, { "epoch": 1.00992768, "grad_norm": 0.9511328935623169, "learning_rate": 1.952816315309072e-05, "loss": 3.6639, "step": 38800 }, { "epoch": 1.00995328, "grad_norm": 0.7982999086380005, "learning_rate": 1.9527918681949586e-05, "loss": 3.3936, "step": 38810 }, { "epoch": 1.00997888, "grad_norm": 0.7528390884399414, "learning_rate": 1.952767414902243e-05, "loss": 3.6094, "step": 38820 }, { "epoch": 1.01000448, "grad_norm": 0.8658792972564697, "learning_rate": 1.9527429554310844e-05, "loss": 3.4467, "step": 38830 }, { "epoch": 1.01003008, "grad_norm": 1.383049488067627, "learning_rate": 1.9527184897816406e-05, "loss": 3.3122, "step": 38840 }, { "epoch": 1.01005568, "grad_norm": 0.9568002820014954, "learning_rate": 1.9526940179540706e-05, "loss": 3.5575, "step": 38850 }, { "epoch": 1.01008128, "grad_norm": 0.7525402307510376, "learning_rate": 1.952669539948533e-05, "loss": 3.5061, "step": 38860 }, { "epoch": 1.01010688, "grad_norm": 0.7720332741737366, "learning_rate": 1.9526450557651866e-05, "loss": 3.1292, "step": 38870 }, { "epoch": 1.01013248, "grad_norm": 0.8591662645339966, "learning_rate": 1.9526205654041904e-05, "loss": 3.4737, "step": 38880 }, { "epoch": 1.01015808, "grad_norm": 0.7440454363822937, "learning_rate": 1.9525960688657027e-05, "loss": 3.3855, "step": 38890 }, { "epoch": 1.01018368, "grad_norm": 0.8125277161598206, "learning_rate": 1.952571566149883e-05, "loss": 3.5122, "step": 38900 }, { "epoch": 1.01020928, "grad_norm": 1.0546009540557861, "learning_rate": 1.9525470572568898e-05, "loss": 3.6902, "step": 38910 }, { "epoch": 1.01023488, "grad_norm": 0.8082741498947144, "learning_rate": 1.9525225421868817e-05, "loss": 3.6132, "step": 38920 }, { "epoch": 1.01026048, "grad_norm": 0.7209550142288208, "learning_rate": 1.9524980209400187e-05, "loss": 3.4089, "step": 38930 }, { "epoch": 1.01028608, "grad_norm": 1.0934141874313354, "learning_rate": 1.9524734935164587e-05, "loss": 3.6098, "step": 38940 }, { "epoch": 1.01031168, "grad_norm": 0.7422130107879639, "learning_rate": 1.952448959916361e-05, "loss": 3.3474, "step": 38950 }, { "epoch": 1.01033728, "grad_norm": 0.8252828121185303, "learning_rate": 1.9524244201398853e-05, "loss": 3.5899, "step": 38960 }, { "epoch": 1.01036288, "grad_norm": 1.0754042863845825, "learning_rate": 1.9523998741871905e-05, "loss": 3.5942, "step": 38970 }, { "epoch": 1.01038848, "grad_norm": 1.1180356740951538, "learning_rate": 1.9523753220584356e-05, "loss": 3.3688, "step": 38980 }, { "epoch": 1.01041408, "grad_norm": 2.1932077407836914, "learning_rate": 1.95235076375378e-05, "loss": 3.6085, "step": 38990 }, { "epoch": 1.01043968, "grad_norm": 0.8476832509040833, "learning_rate": 1.9523261992733822e-05, "loss": 3.4093, "step": 39000 }, { "epoch": 1.01046528, "grad_norm": 0.920208215713501, "learning_rate": 1.9523016286174025e-05, "loss": 3.5317, "step": 39010 }, { "epoch": 1.01049088, "grad_norm": 0.9812808036804199, "learning_rate": 1.9522770517859996e-05, "loss": 3.3774, "step": 39020 }, { "epoch": 1.01051648, "grad_norm": 0.6876496076583862, "learning_rate": 1.9522524687793335e-05, "loss": 3.4234, "step": 39030 }, { "epoch": 1.01054208, "grad_norm": 0.8516364693641663, "learning_rate": 1.952227879597563e-05, "loss": 3.2487, "step": 39040 }, { "epoch": 1.01056768, "grad_norm": 0.7403070330619812, "learning_rate": 1.9522032842408477e-05, "loss": 3.5172, "step": 39050 }, { "epoch": 1.01059328, "grad_norm": 0.8754279613494873, "learning_rate": 1.952178682709347e-05, "loss": 3.4677, "step": 39060 }, { "epoch": 1.01061888, "grad_norm": 0.8395757675170898, "learning_rate": 1.952154075003221e-05, "loss": 3.5415, "step": 39070 }, { "epoch": 1.01064448, "grad_norm": 0.8232977390289307, "learning_rate": 1.9521294611226286e-05, "loss": 3.5209, "step": 39080 }, { "epoch": 1.01067008, "grad_norm": 0.698445737361908, "learning_rate": 1.95210484106773e-05, "loss": 3.2607, "step": 39090 }, { "epoch": 1.01069568, "grad_norm": 0.7356345653533936, "learning_rate": 1.9520802148386843e-05, "loss": 3.4346, "step": 39100 }, { "epoch": 1.01072128, "grad_norm": 0.7836998701095581, "learning_rate": 1.9520555824356514e-05, "loss": 3.2869, "step": 39110 }, { "epoch": 1.01074688, "grad_norm": 0.8266628384590149, "learning_rate": 1.9520309438587912e-05, "loss": 3.6933, "step": 39120 }, { "epoch": 1.01077248, "grad_norm": 0.8339905142784119, "learning_rate": 1.9520062991082632e-05, "loss": 3.4904, "step": 39130 }, { "epoch": 1.01079808, "grad_norm": 0.729978084564209, "learning_rate": 1.9519816481842274e-05, "loss": 3.3599, "step": 39140 }, { "epoch": 1.01082368, "grad_norm": 1.0501505136489868, "learning_rate": 1.9519569910868437e-05, "loss": 3.2662, "step": 39150 }, { "epoch": 1.01084928, "grad_norm": 0.9116112589836121, "learning_rate": 1.951932327816272e-05, "loss": 3.4285, "step": 39160 }, { "epoch": 1.01087488, "grad_norm": 0.8308584690093994, "learning_rate": 1.951907658372672e-05, "loss": 3.4529, "step": 39170 }, { "epoch": 1.01090048, "grad_norm": 0.916697084903717, "learning_rate": 1.951882982756204e-05, "loss": 3.3643, "step": 39180 }, { "epoch": 1.01092608, "grad_norm": 1.0334596633911133, "learning_rate": 1.9518583009670277e-05, "loss": 3.2175, "step": 39190 }, { "epoch": 1.01095168, "grad_norm": 1.5719285011291504, "learning_rate": 1.9518336130053033e-05, "loss": 3.701, "step": 39200 }, { "epoch": 1.01097728, "grad_norm": 0.8755955696105957, "learning_rate": 1.9518089188711905e-05, "loss": 3.3446, "step": 39210 }, { "epoch": 1.01100288, "grad_norm": 5.199332237243652, "learning_rate": 1.9517842185648503e-05, "loss": 3.5072, "step": 39220 }, { "epoch": 1.01102848, "grad_norm": 0.7679300308227539, "learning_rate": 1.9517595120864423e-05, "loss": 3.5031, "step": 39230 }, { "epoch": 1.01105408, "grad_norm": 0.8894584774971008, "learning_rate": 1.951734799436127e-05, "loss": 3.3724, "step": 39240 }, { "epoch": 1.01107968, "grad_norm": 1.3404107093811035, "learning_rate": 1.9517100806140643e-05, "loss": 3.4646, "step": 39250 }, { "epoch": 1.01110528, "grad_norm": 0.7550526857376099, "learning_rate": 1.9516853556204145e-05, "loss": 3.5629, "step": 39260 }, { "epoch": 1.01113088, "grad_norm": 0.7608568072319031, "learning_rate": 1.9516606244553383e-05, "loss": 3.4587, "step": 39270 }, { "epoch": 1.01115648, "grad_norm": 0.8971434235572815, "learning_rate": 1.9516358871189958e-05, "loss": 3.4749, "step": 39280 }, { "epoch": 1.01118208, "grad_norm": 0.8355923295021057, "learning_rate": 1.9516111436115473e-05, "loss": 3.4678, "step": 39290 }, { "epoch": 1.01120768, "grad_norm": 0.7032548189163208, "learning_rate": 1.951586393933154e-05, "loss": 3.5675, "step": 39300 }, { "epoch": 1.01123328, "grad_norm": 0.9056819081306458, "learning_rate": 1.9515616380839752e-05, "loss": 3.2411, "step": 39310 }, { "epoch": 1.01125888, "grad_norm": 0.8285930156707764, "learning_rate": 1.9515368760641725e-05, "loss": 3.3421, "step": 39320 }, { "epoch": 1.01128448, "grad_norm": 0.7532290816307068, "learning_rate": 1.9515121078739058e-05, "loss": 3.4575, "step": 39330 }, { "epoch": 1.01131008, "grad_norm": 0.7469289302825928, "learning_rate": 1.951487333513336e-05, "loss": 3.5904, "step": 39340 }, { "epoch": 1.01133568, "grad_norm": 0.8173041939735413, "learning_rate": 1.9514625529826237e-05, "loss": 3.3575, "step": 39350 }, { "epoch": 1.01136128, "grad_norm": 0.7655494213104248, "learning_rate": 1.9514377662819298e-05, "loss": 3.5176, "step": 39360 }, { "epoch": 1.01138688, "grad_norm": 0.8506954312324524, "learning_rate": 1.9514129734114148e-05, "loss": 3.3727, "step": 39370 }, { "epoch": 1.01141248, "grad_norm": 0.7313495874404907, "learning_rate": 1.9513881743712395e-05, "loss": 3.2349, "step": 39380 }, { "epoch": 1.01143808, "grad_norm": 0.7618494629859924, "learning_rate": 1.9513633691615645e-05, "loss": 3.3557, "step": 39390 }, { "epoch": 1.01146368, "grad_norm": 0.8081052899360657, "learning_rate": 1.951338557782551e-05, "loss": 3.5395, "step": 39400 }, { "epoch": 1.01148928, "grad_norm": 0.7702832221984863, "learning_rate": 1.95131374023436e-05, "loss": 3.5363, "step": 39410 }, { "epoch": 1.01151488, "grad_norm": 0.918881356716156, "learning_rate": 1.951288916517152e-05, "loss": 3.4104, "step": 39420 }, { "epoch": 1.01154048, "grad_norm": 1.0064730644226074, "learning_rate": 1.9512640866310885e-05, "loss": 3.4557, "step": 39430 }, { "epoch": 1.01156608, "grad_norm": 1.3546873331069946, "learning_rate": 1.95123925057633e-05, "loss": 3.4134, "step": 39440 }, { "epoch": 1.01159168, "grad_norm": 2.9499285221099854, "learning_rate": 1.9512144083530377e-05, "loss": 3.4348, "step": 39450 }, { "epoch": 1.01161728, "grad_norm": 0.8604133129119873, "learning_rate": 1.9511895599613727e-05, "loss": 3.452, "step": 39460 }, { "epoch": 1.01164288, "grad_norm": 0.799577534198761, "learning_rate": 1.9511647054014962e-05, "loss": 3.1415, "step": 39470 }, { "epoch": 1.01166848, "grad_norm": 0.8678033947944641, "learning_rate": 1.95113984467357e-05, "loss": 3.4014, "step": 39480 }, { "epoch": 1.01169408, "grad_norm": 0.7555897831916809, "learning_rate": 1.951114977777754e-05, "loss": 3.6197, "step": 39490 }, { "epoch": 1.01171968, "grad_norm": 0.9290664196014404, "learning_rate": 1.9510901047142102e-05, "loss": 3.5485, "step": 39500 }, { "epoch": 1.01174528, "grad_norm": 0.7311877012252808, "learning_rate": 1.9510652254831002e-05, "loss": 3.4736, "step": 39510 }, { "epoch": 1.01177088, "grad_norm": 0.7958731055259705, "learning_rate": 1.9510403400845844e-05, "loss": 3.4471, "step": 39520 }, { "epoch": 1.01179648, "grad_norm": 0.8214541077613831, "learning_rate": 1.9510154485188254e-05, "loss": 3.4596, "step": 39530 }, { "epoch": 1.01182208, "grad_norm": 0.8331996202468872, "learning_rate": 1.9509905507859835e-05, "loss": 3.6114, "step": 39540 }, { "epoch": 1.01184768, "grad_norm": 0.7757537961006165, "learning_rate": 1.9509656468862206e-05, "loss": 3.5387, "step": 39550 }, { "epoch": 1.01187328, "grad_norm": 0.8265349864959717, "learning_rate": 1.9509407368196988e-05, "loss": 3.4317, "step": 39560 }, { "epoch": 1.01189888, "grad_norm": 0.7648820281028748, "learning_rate": 1.9509158205865786e-05, "loss": 3.1676, "step": 39570 }, { "epoch": 1.01192448, "grad_norm": 0.9096166491508484, "learning_rate": 1.950890898187022e-05, "loss": 3.4072, "step": 39580 }, { "epoch": 1.01195008, "grad_norm": 0.8342686295509338, "learning_rate": 1.950865969621191e-05, "loss": 3.4889, "step": 39590 }, { "epoch": 1.01197568, "grad_norm": 0.7495956420898438, "learning_rate": 1.9508410348892463e-05, "loss": 3.5074, "step": 39600 }, { "epoch": 1.01200128, "grad_norm": 0.7632516026496887, "learning_rate": 1.9508160939913506e-05, "loss": 3.3594, "step": 39610 }, { "epoch": 1.01202688, "grad_norm": 0.7901861667633057, "learning_rate": 1.9507911469276653e-05, "loss": 3.4371, "step": 39620 }, { "epoch": 1.01205248, "grad_norm": 2.9978554248809814, "learning_rate": 1.950766193698352e-05, "loss": 3.3387, "step": 39630 }, { "epoch": 1.01207808, "grad_norm": 0.7883576154708862, "learning_rate": 1.9507412343035725e-05, "loss": 3.1854, "step": 39640 }, { "epoch": 1.01210368, "grad_norm": 0.7984568476676941, "learning_rate": 1.950716268743489e-05, "loss": 3.3963, "step": 39650 }, { "epoch": 1.01212928, "grad_norm": 0.7617549896240234, "learning_rate": 1.9506912970182628e-05, "loss": 3.4668, "step": 39660 }, { "epoch": 1.01215488, "grad_norm": 1.3763511180877686, "learning_rate": 1.9506663191280564e-05, "loss": 3.5106, "step": 39670 }, { "epoch": 1.01218048, "grad_norm": 0.7450870871543884, "learning_rate": 1.9506413350730318e-05, "loss": 3.3513, "step": 39680 }, { "epoch": 1.01220608, "grad_norm": 1.0812838077545166, "learning_rate": 1.9506163448533505e-05, "loss": 3.3121, "step": 39690 }, { "epoch": 1.01223168, "grad_norm": 0.8200197815895081, "learning_rate": 1.950591348469175e-05, "loss": 3.3133, "step": 39700 }, { "epoch": 1.01225728, "grad_norm": 0.8914932012557983, "learning_rate": 1.950566345920667e-05, "loss": 3.369, "step": 39710 }, { "epoch": 1.01228288, "grad_norm": 0.7688124775886536, "learning_rate": 1.950541337207989e-05, "loss": 3.5066, "step": 39720 }, { "epoch": 1.01230848, "grad_norm": 0.9174800515174866, "learning_rate": 1.950516322331303e-05, "loss": 3.3019, "step": 39730 }, { "epoch": 1.01233408, "grad_norm": 0.8174201250076294, "learning_rate": 1.9504913012907717e-05, "loss": 3.5334, "step": 39740 }, { "epoch": 1.01235968, "grad_norm": 0.7814203500747681, "learning_rate": 1.9504662740865562e-05, "loss": 3.3604, "step": 39750 }, { "epoch": 1.01238528, "grad_norm": 0.8223546743392944, "learning_rate": 1.95044124071882e-05, "loss": 3.325, "step": 39760 }, { "epoch": 1.01241088, "grad_norm": 1.4733035564422607, "learning_rate": 1.950416201187725e-05, "loss": 3.2161, "step": 39770 }, { "epoch": 1.01243648, "grad_norm": 0.9584980607032776, "learning_rate": 1.950391155493433e-05, "loss": 3.3855, "step": 39780 }, { "epoch": 1.01246208, "grad_norm": 0.7776952981948853, "learning_rate": 1.9503661036361076e-05, "loss": 3.1974, "step": 39790 }, { "epoch": 1.01248768, "grad_norm": 0.7087141871452332, "learning_rate": 1.9503410456159102e-05, "loss": 3.5195, "step": 39800 }, { "epoch": 1.01251328, "grad_norm": 0.7953963875770569, "learning_rate": 1.950315981433004e-05, "loss": 3.6231, "step": 39810 }, { "epoch": 1.01253888, "grad_norm": 0.9348750710487366, "learning_rate": 1.950290911087551e-05, "loss": 3.3253, "step": 39820 }, { "epoch": 1.01256448, "grad_norm": 0.7724438905715942, "learning_rate": 1.9502658345797142e-05, "loss": 3.263, "step": 39830 }, { "epoch": 1.01259008, "grad_norm": 0.7718709707260132, "learning_rate": 1.9502407519096556e-05, "loss": 3.3667, "step": 39840 }, { "epoch": 1.01261568, "grad_norm": 0.8816287517547607, "learning_rate": 1.9502156630775387e-05, "loss": 3.4925, "step": 39850 }, { "epoch": 1.01264128, "grad_norm": 0.8714044690132141, "learning_rate": 1.9501905680835255e-05, "loss": 3.3749, "step": 39860 }, { "epoch": 1.01266688, "grad_norm": 0.9301815032958984, "learning_rate": 1.950165466927779e-05, "loss": 3.2988, "step": 39870 }, { "epoch": 1.01269248, "grad_norm": 0.865889847278595, "learning_rate": 1.950140359610462e-05, "loss": 3.5396, "step": 39880 }, { "epoch": 1.01271808, "grad_norm": 0.750606119632721, "learning_rate": 1.9501152461317373e-05, "loss": 3.4121, "step": 39890 }, { "epoch": 1.01274368, "grad_norm": 1.1903704404830933, "learning_rate": 1.9500901264917677e-05, "loss": 3.2182, "step": 39900 }, { "epoch": 1.01276928, "grad_norm": 0.7271758913993835, "learning_rate": 1.9500650006907165e-05, "loss": 3.5892, "step": 39910 }, { "epoch": 1.01279488, "grad_norm": 0.87350994348526, "learning_rate": 1.9500398687287458e-05, "loss": 3.7219, "step": 39920 }, { "epoch": 1.01282048, "grad_norm": 1.0011698007583618, "learning_rate": 1.9500147306060192e-05, "loss": 3.666, "step": 39930 }, { "epoch": 1.01284608, "grad_norm": 1.0859397649765015, "learning_rate": 1.9499895863226993e-05, "loss": 3.5416, "step": 39940 }, { "epoch": 1.01287168, "grad_norm": 0.7691282033920288, "learning_rate": 1.94996443587895e-05, "loss": 3.5899, "step": 39950 }, { "epoch": 1.01289728, "grad_norm": 0.7541866898536682, "learning_rate": 1.9499392792749333e-05, "loss": 3.3065, "step": 39960 }, { "epoch": 1.01292288, "grad_norm": 0.7634449601173401, "learning_rate": 1.949914116510813e-05, "loss": 3.2723, "step": 39970 }, { "epoch": 1.01294848, "grad_norm": 1.016884207725525, "learning_rate": 1.9498889475867525e-05, "loss": 3.4045, "step": 39980 }, { "epoch": 1.01297408, "grad_norm": 1.1288161277770996, "learning_rate": 1.9498637725029138e-05, "loss": 3.3742, "step": 39990 }, { "epoch": 1.01299968, "grad_norm": 0.8059718608856201, "learning_rate": 1.9498385912594617e-05, "loss": 3.5155, "step": 40000 }, { "epoch": 1.01302528, "grad_norm": 0.7704803347587585, "learning_rate": 1.9498134038565585e-05, "loss": 3.4994, "step": 40010 }, { "epoch": 1.01305088, "grad_norm": 0.7460649013519287, "learning_rate": 1.9497882102943677e-05, "loss": 3.635, "step": 40020 }, { "epoch": 1.01307648, "grad_norm": 0.811040461063385, "learning_rate": 1.949763010573053e-05, "loss": 3.52, "step": 40030 }, { "epoch": 1.01310208, "grad_norm": 0.8638127446174622, "learning_rate": 1.9497378046927774e-05, "loss": 3.4002, "step": 40040 }, { "epoch": 1.01312768, "grad_norm": 0.8873514533042908, "learning_rate": 1.9497125926537047e-05, "loss": 3.2955, "step": 40050 }, { "epoch": 1.01315328, "grad_norm": 0.8063274025917053, "learning_rate": 1.949687374455998e-05, "loss": 3.3706, "step": 40060 }, { "epoch": 1.01317888, "grad_norm": 0.7947402000427246, "learning_rate": 1.9496621500998213e-05, "loss": 3.3907, "step": 40070 }, { "epoch": 1.01320448, "grad_norm": 0.9408482313156128, "learning_rate": 1.949636919585338e-05, "loss": 3.3855, "step": 40080 }, { "epoch": 1.01323008, "grad_norm": 2.0490169525146484, "learning_rate": 1.9496116829127113e-05, "loss": 3.477, "step": 40090 }, { "epoch": 1.01325568, "grad_norm": 0.7563177347183228, "learning_rate": 1.9495864400821055e-05, "loss": 3.5469, "step": 40100 }, { "epoch": 1.01328128, "grad_norm": 0.789638340473175, "learning_rate": 1.949561191093684e-05, "loss": 3.3911, "step": 40110 }, { "epoch": 1.01330688, "grad_norm": 0.9257838726043701, "learning_rate": 1.9495359359476108e-05, "loss": 3.5888, "step": 40120 }, { "epoch": 1.01333248, "grad_norm": 0.7704899311065674, "learning_rate": 1.949510674644049e-05, "loss": 3.6261, "step": 40130 }, { "epoch": 1.01335808, "grad_norm": 0.7492403984069824, "learning_rate": 1.949485407183163e-05, "loss": 3.7075, "step": 40140 }, { "epoch": 1.01338368, "grad_norm": 0.8378415107727051, "learning_rate": 1.949460133565116e-05, "loss": 3.5856, "step": 40150 }, { "epoch": 1.01340928, "grad_norm": 0.8199307322502136, "learning_rate": 1.949434853790073e-05, "loss": 3.4802, "step": 40160 }, { "epoch": 1.01343488, "grad_norm": 1.2678848505020142, "learning_rate": 1.9494095678581974e-05, "loss": 3.4269, "step": 40170 }, { "epoch": 1.01346048, "grad_norm": 0.8343449831008911, "learning_rate": 1.9493842757696522e-05, "loss": 3.6796, "step": 40180 }, { "epoch": 1.01348608, "grad_norm": 1.4116069078445435, "learning_rate": 1.949358977524603e-05, "loss": 3.4141, "step": 40190 }, { "epoch": 1.01351168, "grad_norm": 0.9381754398345947, "learning_rate": 1.9493336731232132e-05, "loss": 3.6565, "step": 40200 }, { "epoch": 1.01353728, "grad_norm": 0.9388043284416199, "learning_rate": 1.9493083625656467e-05, "loss": 3.5709, "step": 40210 }, { "epoch": 1.01356288, "grad_norm": 0.8831541538238525, "learning_rate": 1.9492830458520676e-05, "loss": 3.7327, "step": 40220 }, { "epoch": 1.01358848, "grad_norm": 2.050117254257202, "learning_rate": 1.9492577229826402e-05, "loss": 3.4017, "step": 40230 }, { "epoch": 1.01361408, "grad_norm": 1.3823217153549194, "learning_rate": 1.9492323939575287e-05, "loss": 3.5053, "step": 40240 }, { "epoch": 1.01363968, "grad_norm": 1.033773422241211, "learning_rate": 1.949207058776898e-05, "loss": 3.5422, "step": 40250 }, { "epoch": 1.01366528, "grad_norm": 0.772869884967804, "learning_rate": 1.949181717440911e-05, "loss": 3.623, "step": 40260 }, { "epoch": 1.01369088, "grad_norm": 0.9318835139274597, "learning_rate": 1.9491563699497335e-05, "loss": 3.3834, "step": 40270 }, { "epoch": 1.01371648, "grad_norm": 0.904830813407898, "learning_rate": 1.9491310163035287e-05, "loss": 3.5833, "step": 40280 }, { "epoch": 1.01374208, "grad_norm": 0.8163405060768127, "learning_rate": 1.9491056565024617e-05, "loss": 3.6549, "step": 40290 }, { "epoch": 1.01376768, "grad_norm": 1.2476170063018799, "learning_rate": 1.9490802905466967e-05, "loss": 3.4649, "step": 40300 }, { "epoch": 1.01379328, "grad_norm": 0.7893875241279602, "learning_rate": 1.9490549184363985e-05, "loss": 3.5381, "step": 40310 }, { "epoch": 1.01381888, "grad_norm": 0.765066921710968, "learning_rate": 1.9490295401717313e-05, "loss": 3.5788, "step": 40320 }, { "epoch": 1.01384448, "grad_norm": 0.8552426695823669, "learning_rate": 1.9490041557528596e-05, "loss": 3.5702, "step": 40330 }, { "epoch": 1.01387008, "grad_norm": 1.1302037239074707, "learning_rate": 1.9489787651799482e-05, "loss": 3.5412, "step": 40340 }, { "epoch": 1.01389568, "grad_norm": 0.7856495380401611, "learning_rate": 1.9489533684531615e-05, "loss": 3.5474, "step": 40350 }, { "epoch": 1.01392128, "grad_norm": 0.771257221698761, "learning_rate": 1.948927965572665e-05, "loss": 3.5857, "step": 40360 }, { "epoch": 1.01394688, "grad_norm": 0.9111387729644775, "learning_rate": 1.9489025565386223e-05, "loss": 3.4881, "step": 40370 }, { "epoch": 1.01397248, "grad_norm": 0.9356427788734436, "learning_rate": 1.9488771413511987e-05, "loss": 3.55, "step": 40380 }, { "epoch": 1.01399808, "grad_norm": 0.9175236821174622, "learning_rate": 1.9488517200105592e-05, "loss": 3.5058, "step": 40390 }, { "epoch": 1.01402368, "grad_norm": 0.7824336290359497, "learning_rate": 1.9488262925168688e-05, "loss": 3.2864, "step": 40400 }, { "epoch": 1.01404928, "grad_norm": 1.0044898986816406, "learning_rate": 1.9488008588702914e-05, "loss": 3.4494, "step": 40410 }, { "epoch": 1.01407488, "grad_norm": 0.7720074653625488, "learning_rate": 1.948775419070993e-05, "loss": 3.3342, "step": 40420 }, { "epoch": 1.01410048, "grad_norm": 0.977275550365448, "learning_rate": 1.9487499731191383e-05, "loss": 3.6214, "step": 40430 }, { "epoch": 1.01412608, "grad_norm": 0.732754647731781, "learning_rate": 1.9487245210148917e-05, "loss": 3.4666, "step": 40440 }, { "epoch": 1.0141516799999999, "grad_norm": 0.8354089260101318, "learning_rate": 1.948699062758419e-05, "loss": 3.4309, "step": 40450 }, { "epoch": 1.01417728, "grad_norm": 0.8090373277664185, "learning_rate": 1.948673598349885e-05, "loss": 3.5236, "step": 40460 }, { "epoch": 1.01420288, "grad_norm": 0.783000648021698, "learning_rate": 1.9486481277894548e-05, "loss": 3.3809, "step": 40470 }, { "epoch": 1.01422848, "grad_norm": 0.7169411778450012, "learning_rate": 1.948622651077294e-05, "loss": 3.4851, "step": 40480 }, { "epoch": 1.01425408, "grad_norm": 1.7506847381591797, "learning_rate": 1.9485971682135667e-05, "loss": 3.6707, "step": 40490 }, { "epoch": 1.01427968, "grad_norm": 0.8018583059310913, "learning_rate": 1.9485716791984394e-05, "loss": 3.5195, "step": 40500 }, { "epoch": 1.01430528, "grad_norm": 0.8385321497917175, "learning_rate": 1.948546184032077e-05, "loss": 3.5264, "step": 40510 }, { "epoch": 1.01433088, "grad_norm": 0.7556445598602295, "learning_rate": 1.9485206827146437e-05, "loss": 3.4383, "step": 40520 }, { "epoch": 1.01435648, "grad_norm": 0.8104725480079651, "learning_rate": 1.9484951752463067e-05, "loss": 3.578, "step": 40530 }, { "epoch": 1.01438208, "grad_norm": 0.8221552968025208, "learning_rate": 1.9484696616272305e-05, "loss": 3.3699, "step": 40540 }, { "epoch": 1.01440768, "grad_norm": 1.0550929307937622, "learning_rate": 1.9484441418575806e-05, "loss": 3.4462, "step": 40550 }, { "epoch": 1.01443328, "grad_norm": 0.724113404750824, "learning_rate": 1.9484186159375226e-05, "loss": 3.4316, "step": 40560 }, { "epoch": 1.01445888, "grad_norm": 0.792799174785614, "learning_rate": 1.9483930838672217e-05, "loss": 3.4067, "step": 40570 }, { "epoch": 1.01448448, "grad_norm": 1.3816900253295898, "learning_rate": 1.9483675456468437e-05, "loss": 3.4207, "step": 40580 }, { "epoch": 1.01451008, "grad_norm": 0.8579064607620239, "learning_rate": 1.9483420012765544e-05, "loss": 3.5059, "step": 40590 }, { "epoch": 1.01453568, "grad_norm": 0.9228204488754272, "learning_rate": 1.948316450756519e-05, "loss": 3.4856, "step": 40600 }, { "epoch": 1.01456128, "grad_norm": 0.7607482671737671, "learning_rate": 1.9482908940869037e-05, "loss": 3.4703, "step": 40610 }, { "epoch": 1.01458688, "grad_norm": 0.9261395335197449, "learning_rate": 1.948265331267874e-05, "loss": 3.5606, "step": 40620 }, { "epoch": 1.01461248, "grad_norm": 2.3345448970794678, "learning_rate": 1.948239762299595e-05, "loss": 3.7176, "step": 40630 }, { "epoch": 1.01463808, "grad_norm": 0.8671140670776367, "learning_rate": 1.948214187182234e-05, "loss": 3.164, "step": 40640 }, { "epoch": 1.01466368, "grad_norm": 0.7943164110183716, "learning_rate": 1.9481886059159558e-05, "loss": 3.4967, "step": 40650 }, { "epoch": 1.01468928, "grad_norm": 0.94368577003479, "learning_rate": 1.9481630185009264e-05, "loss": 3.4482, "step": 40660 }, { "epoch": 1.01471488, "grad_norm": 0.8197217583656311, "learning_rate": 1.948137424937312e-05, "loss": 3.4365, "step": 40670 }, { "epoch": 1.01474048, "grad_norm": 0.9035544395446777, "learning_rate": 1.9481118252252776e-05, "loss": 3.2925, "step": 40680 }, { "epoch": 1.01476608, "grad_norm": 0.917404055595398, "learning_rate": 1.9480862193649907e-05, "loss": 3.3721, "step": 40690 }, { "epoch": 1.01479168, "grad_norm": 0.9252608418464661, "learning_rate": 1.9480606073566168e-05, "loss": 3.5052, "step": 40700 }, { "epoch": 1.01481728, "grad_norm": 0.7468551993370056, "learning_rate": 1.9480349892003212e-05, "loss": 3.5086, "step": 40710 }, { "epoch": 1.01484288, "grad_norm": 0.804385244846344, "learning_rate": 1.948009364896271e-05, "loss": 3.4455, "step": 40720 }, { "epoch": 1.01486848, "grad_norm": 0.928210437297821, "learning_rate": 1.947983734444632e-05, "loss": 3.4812, "step": 40730 }, { "epoch": 1.01489408, "grad_norm": 0.8612575531005859, "learning_rate": 1.9479580978455704e-05, "loss": 3.4503, "step": 40740 }, { "epoch": 1.01491968, "grad_norm": 0.8112159371376038, "learning_rate": 1.9479324550992524e-05, "loss": 3.4962, "step": 40750 }, { "epoch": 1.01494528, "grad_norm": 0.9062279462814331, "learning_rate": 1.9479068062058445e-05, "loss": 3.4645, "step": 40760 }, { "epoch": 1.01497088, "grad_norm": 1.4741557836532593, "learning_rate": 1.9478811511655128e-05, "loss": 3.5264, "step": 40770 }, { "epoch": 1.01499648, "grad_norm": 3.551309585571289, "learning_rate": 1.947855489978424e-05, "loss": 3.762, "step": 40780 }, { "epoch": 1.01502208, "grad_norm": 0.8012361526489258, "learning_rate": 1.9478298226447442e-05, "loss": 3.3851, "step": 40790 }, { "epoch": 1.01504768, "grad_norm": 0.9211165904998779, "learning_rate": 1.9478041491646394e-05, "loss": 3.5527, "step": 40800 }, { "epoch": 1.01507328, "grad_norm": 1.0110102891921997, "learning_rate": 1.947778469538277e-05, "loss": 3.4229, "step": 40810 }, { "epoch": 1.01509888, "grad_norm": 0.9730212092399597, "learning_rate": 1.947752783765823e-05, "loss": 3.5941, "step": 40820 }, { "epoch": 1.01512448, "grad_norm": 0.9794178009033203, "learning_rate": 1.9477270918474442e-05, "loss": 3.7043, "step": 40830 }, { "epoch": 1.01515008, "grad_norm": 1.1735478639602661, "learning_rate": 1.947701393783307e-05, "loss": 3.4304, "step": 40840 }, { "epoch": 1.01517568, "grad_norm": 0.8594624996185303, "learning_rate": 1.9476756895735782e-05, "loss": 3.7308, "step": 40850 }, { "epoch": 1.01520128, "grad_norm": 0.8255285620689392, "learning_rate": 1.9476499792184243e-05, "loss": 3.5577, "step": 40860 }, { "epoch": 1.01522688, "grad_norm": 0.9129996299743652, "learning_rate": 1.947624262718012e-05, "loss": 3.315, "step": 40870 }, { "epoch": 1.01525248, "grad_norm": 0.8135256767272949, "learning_rate": 1.9475985400725087e-05, "loss": 3.3273, "step": 40880 }, { "epoch": 1.01527808, "grad_norm": 0.8819308280944824, "learning_rate": 1.9475728112820805e-05, "loss": 3.3002, "step": 40890 }, { "epoch": 1.01530368, "grad_norm": 0.7360608577728271, "learning_rate": 1.9475470763468944e-05, "loss": 3.3178, "step": 40900 }, { "epoch": 1.01532928, "grad_norm": 0.7949578166007996, "learning_rate": 1.947521335267117e-05, "loss": 3.5066, "step": 40910 }, { "epoch": 1.01535488, "grad_norm": 1.0850927829742432, "learning_rate": 1.9474955880429162e-05, "loss": 3.5816, "step": 40920 }, { "epoch": 1.01538048, "grad_norm": 0.787568986415863, "learning_rate": 1.9474698346744578e-05, "loss": 3.4212, "step": 40930 }, { "epoch": 1.01540608, "grad_norm": 0.7281793355941772, "learning_rate": 1.9474440751619094e-05, "loss": 3.2212, "step": 40940 }, { "epoch": 2.00001536, "grad_norm": 0.8648423552513123, "learning_rate": 1.947418309505438e-05, "loss": 3.6211, "step": 40950 }, { "epoch": 2.00004096, "grad_norm": 0.7784690260887146, "learning_rate": 1.947392537705211e-05, "loss": 3.3548, "step": 40960 }, { "epoch": 2.00006656, "grad_norm": 2.4770829677581787, "learning_rate": 1.947366759761395e-05, "loss": 3.5866, "step": 40970 }, { "epoch": 2.00009216, "grad_norm": 0.7442464828491211, "learning_rate": 1.9473409756741573e-05, "loss": 3.1827, "step": 40980 }, { "epoch": 2.00011776, "grad_norm": 0.7437440156936646, "learning_rate": 1.947315185443665e-05, "loss": 3.2054, "step": 40990 }, { "epoch": 2.00014336, "grad_norm": 0.7740707397460938, "learning_rate": 1.9472893890700855e-05, "loss": 3.2605, "step": 41000 }, { "epoch": 2.00016896, "grad_norm": 0.9012874364852905, "learning_rate": 1.9472635865535863e-05, "loss": 3.4843, "step": 41010 }, { "epoch": 2.00019456, "grad_norm": 0.9461014270782471, "learning_rate": 1.9472377778943344e-05, "loss": 3.2144, "step": 41020 }, { "epoch": 2.00022016, "grad_norm": 1.1425701379776, "learning_rate": 1.9472119630924974e-05, "loss": 3.123, "step": 41030 }, { "epoch": 2.00024576, "grad_norm": 1.2034679651260376, "learning_rate": 1.947186142148242e-05, "loss": 3.359, "step": 41040 }, { "epoch": 2.00027136, "grad_norm": 0.9737411141395569, "learning_rate": 1.9471603150617366e-05, "loss": 3.7104, "step": 41050 }, { "epoch": 2.00029696, "grad_norm": 0.791409432888031, "learning_rate": 1.9471344818331485e-05, "loss": 3.3945, "step": 41060 }, { "epoch": 2.00032256, "grad_norm": 0.8414984941482544, "learning_rate": 1.9471086424626445e-05, "loss": 3.2116, "step": 41070 }, { "epoch": 2.00034816, "grad_norm": 0.8392379283905029, "learning_rate": 1.947082796950393e-05, "loss": 3.3067, "step": 41080 }, { "epoch": 2.00037376, "grad_norm": 1.1956470012664795, "learning_rate": 1.9470569452965612e-05, "loss": 3.3059, "step": 41090 }, { "epoch": 2.00039936, "grad_norm": 0.8013750314712524, "learning_rate": 1.947031087501317e-05, "loss": 3.3533, "step": 41100 }, { "epoch": 2.00042496, "grad_norm": 0.7957696914672852, "learning_rate": 1.9470052235648276e-05, "loss": 3.2242, "step": 41110 }, { "epoch": 2.00045056, "grad_norm": 0.9354164004325867, "learning_rate": 1.946979353487261e-05, "loss": 3.3345, "step": 41120 }, { "epoch": 2.00047616, "grad_norm": 0.8468334674835205, "learning_rate": 1.9469534772687853e-05, "loss": 3.1306, "step": 41130 }, { "epoch": 2.00050176, "grad_norm": 0.7972702980041504, "learning_rate": 1.9469275949095676e-05, "loss": 3.4667, "step": 41140 }, { "epoch": 2.00052736, "grad_norm": 0.8703827857971191, "learning_rate": 1.9469017064097764e-05, "loss": 3.2108, "step": 41150 }, { "epoch": 2.00055296, "grad_norm": 0.7656765580177307, "learning_rate": 1.946875811769579e-05, "loss": 3.3614, "step": 41160 }, { "epoch": 2.00057856, "grad_norm": 0.9516866207122803, "learning_rate": 1.9468499109891438e-05, "loss": 3.3026, "step": 41170 }, { "epoch": 2.00060416, "grad_norm": 0.720917820930481, "learning_rate": 1.9468240040686384e-05, "loss": 3.5371, "step": 41180 }, { "epoch": 2.00062976, "grad_norm": 0.9357711672782898, "learning_rate": 1.946798091008231e-05, "loss": 3.2834, "step": 41190 }, { "epoch": 2.00065536, "grad_norm": 0.7122368216514587, "learning_rate": 1.94677217180809e-05, "loss": 3.342, "step": 41200 }, { "epoch": 2.00068096, "grad_norm": 0.9021980166435242, "learning_rate": 1.946746246468383e-05, "loss": 3.2597, "step": 41210 }, { "epoch": 2.00070656, "grad_norm": 0.9155310392379761, "learning_rate": 1.946720314989278e-05, "loss": 3.1112, "step": 41220 }, { "epoch": 2.00073216, "grad_norm": 0.7080984711647034, "learning_rate": 1.946694377370943e-05, "loss": 3.2759, "step": 41230 }, { "epoch": 2.00075776, "grad_norm": 0.7439106702804565, "learning_rate": 1.946668433613547e-05, "loss": 3.1791, "step": 41240 }, { "epoch": 2.00078336, "grad_norm": 0.8304629921913147, "learning_rate": 1.946642483717258e-05, "loss": 3.2563, "step": 41250 }, { "epoch": 2.00080896, "grad_norm": 0.9732460379600525, "learning_rate": 1.9466165276822437e-05, "loss": 3.3722, "step": 41260 }, { "epoch": 2.00083456, "grad_norm": 1.0085422992706299, "learning_rate": 1.946590565508673e-05, "loss": 3.3872, "step": 41270 }, { "epoch": 2.00086016, "grad_norm": 0.7980553507804871, "learning_rate": 1.946564597196714e-05, "loss": 3.2441, "step": 41280 }, { "epoch": 2.00088576, "grad_norm": 0.8257339596748352, "learning_rate": 1.9465386227465348e-05, "loss": 3.2008, "step": 41290 }, { "epoch": 2.00091136, "grad_norm": 1.4083353281021118, "learning_rate": 1.946512642158305e-05, "loss": 3.2712, "step": 41300 }, { "epoch": 2.00093696, "grad_norm": 0.9230853319168091, "learning_rate": 1.9464866554321916e-05, "loss": 3.387, "step": 41310 }, { "epoch": 2.00096256, "grad_norm": 0.7581651210784912, "learning_rate": 1.946460662568364e-05, "loss": 3.312, "step": 41320 }, { "epoch": 2.00098816, "grad_norm": 0.7344048619270325, "learning_rate": 1.9464346635669904e-05, "loss": 3.3166, "step": 41330 }, { "epoch": 2.00101376, "grad_norm": 1.1038278341293335, "learning_rate": 1.9464086584282396e-05, "loss": 3.3404, "step": 41340 }, { "epoch": 2.00103936, "grad_norm": 3.4707045555114746, "learning_rate": 1.94638264715228e-05, "loss": 3.4603, "step": 41350 }, { "epoch": 2.00106496, "grad_norm": 0.8204007744789124, "learning_rate": 1.946356629739281e-05, "loss": 3.2638, "step": 41360 }, { "epoch": 2.00109056, "grad_norm": 0.7749536633491516, "learning_rate": 1.9463306061894104e-05, "loss": 3.4467, "step": 41370 }, { "epoch": 2.00111616, "grad_norm": 0.7366728186607361, "learning_rate": 1.9463045765028374e-05, "loss": 3.2798, "step": 41380 }, { "epoch": 2.00114176, "grad_norm": 0.7280868887901306, "learning_rate": 1.9462785406797305e-05, "loss": 3.1984, "step": 41390 }, { "epoch": 2.00116736, "grad_norm": 0.7528578639030457, "learning_rate": 1.946252498720259e-05, "loss": 3.1763, "step": 41400 }, { "epoch": 2.00119296, "grad_norm": 0.8281322121620178, "learning_rate": 1.9462264506245914e-05, "loss": 3.1532, "step": 41410 }, { "epoch": 2.00121856, "grad_norm": 0.7992222309112549, "learning_rate": 1.9462003963928972e-05, "loss": 3.1871, "step": 41420 }, { "epoch": 2.00124416, "grad_norm": 0.7155008912086487, "learning_rate": 1.946174336025344e-05, "loss": 3.1822, "step": 41430 }, { "epoch": 2.00126976, "grad_norm": 0.7951751947402954, "learning_rate": 1.9461482695221025e-05, "loss": 3.2914, "step": 41440 }, { "epoch": 2.00129536, "grad_norm": 0.8736405372619629, "learning_rate": 1.9461221968833405e-05, "loss": 3.2396, "step": 41450 }, { "epoch": 2.00132096, "grad_norm": 0.6903493404388428, "learning_rate": 1.946096118109228e-05, "loss": 2.9883, "step": 41460 }, { "epoch": 2.00134656, "grad_norm": 0.7785794138908386, "learning_rate": 1.946070033199933e-05, "loss": 3.2574, "step": 41470 }, { "epoch": 2.00137216, "grad_norm": 0.7033239603042603, "learning_rate": 1.9460439421556257e-05, "loss": 3.1328, "step": 41480 }, { "epoch": 2.00139776, "grad_norm": 0.9447778463363647, "learning_rate": 1.9460178449764748e-05, "loss": 3.5512, "step": 41490 }, { "epoch": 2.00142336, "grad_norm": 0.7046921849250793, "learning_rate": 1.945991741662649e-05, "loss": 3.2672, "step": 41500 }, { "epoch": 2.00144896, "grad_norm": 0.6857909560203552, "learning_rate": 1.9459656322143187e-05, "loss": 3.2072, "step": 41510 }, { "epoch": 2.00147456, "grad_norm": 0.8030847907066345, "learning_rate": 1.9459395166316527e-05, "loss": 3.1758, "step": 41520 }, { "epoch": 2.00150016, "grad_norm": 0.9666990041732788, "learning_rate": 1.9459133949148202e-05, "loss": 3.2459, "step": 41530 }, { "epoch": 2.00152576, "grad_norm": 0.8098236918449402, "learning_rate": 1.945887267063991e-05, "loss": 3.1243, "step": 41540 }, { "epoch": 2.00155136, "grad_norm": 0.8207499980926514, "learning_rate": 1.945861133079334e-05, "loss": 3.2117, "step": 41550 }, { "epoch": 2.00157696, "grad_norm": 0.7343152761459351, "learning_rate": 1.945834992961019e-05, "loss": 3.0015, "step": 41560 }, { "epoch": 2.00160256, "grad_norm": 0.8237646222114563, "learning_rate": 1.9458088467092155e-05, "loss": 3.5352, "step": 41570 }, { "epoch": 2.00162816, "grad_norm": 0.7469272613525391, "learning_rate": 1.9457826943240928e-05, "loss": 3.3171, "step": 41580 }, { "epoch": 2.00165376, "grad_norm": 0.7699095010757446, "learning_rate": 1.945756535805821e-05, "loss": 3.1946, "step": 41590 }, { "epoch": 2.00167936, "grad_norm": 0.9547687768936157, "learning_rate": 1.9457303711545693e-05, "loss": 3.5711, "step": 41600 }, { "epoch": 2.00170496, "grad_norm": 0.7984176874160767, "learning_rate": 1.945704200370507e-05, "loss": 3.1427, "step": 41610 }, { "epoch": 2.00173056, "grad_norm": 0.9222204089164734, "learning_rate": 1.9456780234538048e-05, "loss": 3.3349, "step": 41620 }, { "epoch": 2.00175616, "grad_norm": 0.7615705132484436, "learning_rate": 1.9456518404046322e-05, "loss": 3.433, "step": 41630 }, { "epoch": 2.00178176, "grad_norm": 0.7023246884346008, "learning_rate": 1.9456256512231582e-05, "loss": 3.064, "step": 41640 }, { "epoch": 2.00180736, "grad_norm": 0.7352251410484314, "learning_rate": 1.9455994559095533e-05, "loss": 2.9622, "step": 41650 }, { "epoch": 2.00183296, "grad_norm": 0.8827303647994995, "learning_rate": 1.9455732544639873e-05, "loss": 3.2579, "step": 41660 }, { "epoch": 2.00185856, "grad_norm": 0.8761966824531555, "learning_rate": 1.94554704688663e-05, "loss": 3.2191, "step": 41670 }, { "epoch": 2.00188416, "grad_norm": 0.728275716304779, "learning_rate": 1.9455208331776515e-05, "loss": 3.374, "step": 41680 }, { "epoch": 2.00190976, "grad_norm": 0.8720196485519409, "learning_rate": 1.9454946133372217e-05, "loss": 3.2822, "step": 41690 }, { "epoch": 2.00193536, "grad_norm": 0.8942943215370178, "learning_rate": 1.9454683873655107e-05, "loss": 3.2146, "step": 41700 }, { "epoch": 2.00196096, "grad_norm": 0.8785948157310486, "learning_rate": 1.9454421552626882e-05, "loss": 3.4015, "step": 41710 }, { "epoch": 2.00198656, "grad_norm": 0.9365178942680359, "learning_rate": 1.9454159170289247e-05, "loss": 3.3786, "step": 41720 }, { "epoch": 2.00201216, "grad_norm": 0.8896686434745789, "learning_rate": 1.9453896726643903e-05, "loss": 3.3312, "step": 41730 }, { "epoch": 2.00203776, "grad_norm": 0.7717710137367249, "learning_rate": 1.9453634221692554e-05, "loss": 3.4479, "step": 41740 }, { "epoch": 2.00206336, "grad_norm": 1.1964367628097534, "learning_rate": 1.9453371655436893e-05, "loss": 3.444, "step": 41750 }, { "epoch": 2.00208896, "grad_norm": 0.7446661591529846, "learning_rate": 1.9453109027878632e-05, "loss": 3.0088, "step": 41760 }, { "epoch": 2.00211456, "grad_norm": 1.0454846620559692, "learning_rate": 1.9452846339019476e-05, "loss": 3.3579, "step": 41770 }, { "epoch": 2.00214016, "grad_norm": 0.8673444986343384, "learning_rate": 1.945258358886112e-05, "loss": 3.4045, "step": 41780 }, { "epoch": 2.00216576, "grad_norm": 2.5839149951934814, "learning_rate": 1.945232077740527e-05, "loss": 3.6345, "step": 41790 }, { "epoch": 2.00219136, "grad_norm": 1.0108568668365479, "learning_rate": 1.9452057904653633e-05, "loss": 3.1218, "step": 41800 }, { "epoch": 2.00221696, "grad_norm": 0.9563668966293335, "learning_rate": 1.9451794970607915e-05, "loss": 3.4042, "step": 41810 }, { "epoch": 2.00224256, "grad_norm": 0.9976035356521606, "learning_rate": 1.9451531975269815e-05, "loss": 3.54, "step": 41820 }, { "epoch": 2.00226816, "grad_norm": 0.7743564248085022, "learning_rate": 1.9451268918641042e-05, "loss": 3.4011, "step": 41830 }, { "epoch": 2.00229376, "grad_norm": 0.8732303977012634, "learning_rate": 1.9451005800723302e-05, "loss": 3.3285, "step": 41840 }, { "epoch": 2.00231936, "grad_norm": 0.9241203665733337, "learning_rate": 1.9450742621518307e-05, "loss": 3.4461, "step": 41850 }, { "epoch": 2.00234496, "grad_norm": 1.0077213048934937, "learning_rate": 1.9450479381027755e-05, "loss": 3.389, "step": 41860 }, { "epoch": 2.00237056, "grad_norm": 0.766306459903717, "learning_rate": 1.9450216079253352e-05, "loss": 3.5614, "step": 41870 }, { "epoch": 2.00239616, "grad_norm": 0.7011285424232483, "learning_rate": 1.944995271619681e-05, "loss": 3.2129, "step": 41880 }, { "epoch": 2.00242176, "grad_norm": 0.8721188306808472, "learning_rate": 1.944968929185984e-05, "loss": 3.3786, "step": 41890 }, { "epoch": 2.00244736, "grad_norm": 1.3622061014175415, "learning_rate": 1.9449425806244142e-05, "loss": 3.2336, "step": 41900 }, { "epoch": 2.00247296, "grad_norm": 0.7643356919288635, "learning_rate": 1.9449162259351432e-05, "loss": 3.1654, "step": 41910 }, { "epoch": 2.00249856, "grad_norm": 0.8530015349388123, "learning_rate": 1.9448898651183413e-05, "loss": 3.4848, "step": 41920 }, { "epoch": 2.00252416, "grad_norm": 0.7693105340003967, "learning_rate": 1.94486349817418e-05, "loss": 3.5245, "step": 41930 }, { "epoch": 2.00254976, "grad_norm": 0.757128894329071, "learning_rate": 1.9448371251028303e-05, "loss": 3.4589, "step": 41940 }, { "epoch": 2.00257536, "grad_norm": 0.8323243856430054, "learning_rate": 1.9448107459044624e-05, "loss": 3.3825, "step": 41950 }, { "epoch": 2.00260096, "grad_norm": 0.8420889973640442, "learning_rate": 1.9447843605792483e-05, "loss": 3.2757, "step": 41960 }, { "epoch": 2.00262656, "grad_norm": 0.8405770659446716, "learning_rate": 1.9447579691273586e-05, "loss": 3.0822, "step": 41970 }, { "epoch": 2.00265216, "grad_norm": 0.79940265417099, "learning_rate": 1.9447315715489642e-05, "loss": 3.2982, "step": 41980 }, { "epoch": 2.00267776, "grad_norm": 0.8558135628700256, "learning_rate": 1.9447051678442374e-05, "loss": 3.3682, "step": 41990 }, { "epoch": 2.00270336, "grad_norm": 0.7391214966773987, "learning_rate": 1.9446787580133478e-05, "loss": 3.0873, "step": 42000 }, { "epoch": 2.00272896, "grad_norm": 0.8447456955909729, "learning_rate": 1.9446523420564683e-05, "loss": 3.5377, "step": 42010 }, { "epoch": 2.00275456, "grad_norm": 0.8212832808494568, "learning_rate": 1.944625919973769e-05, "loss": 3.3074, "step": 42020 }, { "epoch": 2.00278016, "grad_norm": 0.7414726614952087, "learning_rate": 1.9445994917654218e-05, "loss": 3.0384, "step": 42030 }, { "epoch": 2.00280576, "grad_norm": 0.751285970211029, "learning_rate": 1.9445730574315976e-05, "loss": 3.3943, "step": 42040 }, { "epoch": 2.00283136, "grad_norm": 0.9553493857383728, "learning_rate": 1.9445466169724683e-05, "loss": 3.2061, "step": 42050 }, { "epoch": 2.00285696, "grad_norm": 1.2302289009094238, "learning_rate": 1.9445201703882056e-05, "loss": 3.4304, "step": 42060 }, { "epoch": 2.00288256, "grad_norm": 0.8694784641265869, "learning_rate": 1.9444937176789805e-05, "loss": 3.2546, "step": 42070 }, { "epoch": 2.00290816, "grad_norm": 0.8866481184959412, "learning_rate": 1.9444672588449642e-05, "loss": 3.4289, "step": 42080 }, { "epoch": 2.00293376, "grad_norm": 0.8902485966682434, "learning_rate": 1.944440793886329e-05, "loss": 3.5032, "step": 42090 }, { "epoch": 2.00295936, "grad_norm": 0.8002330660820007, "learning_rate": 1.9444143228032467e-05, "loss": 3.4022, "step": 42100 }, { "epoch": 2.00298496, "grad_norm": 0.8670795559883118, "learning_rate": 1.944387845595888e-05, "loss": 3.0887, "step": 42110 }, { "epoch": 2.00301056, "grad_norm": 1.0630995035171509, "learning_rate": 1.944361362264425e-05, "loss": 3.4206, "step": 42120 }, { "epoch": 2.00303616, "grad_norm": 0.8461490273475647, "learning_rate": 1.94433487280903e-05, "loss": 3.4831, "step": 42130 }, { "epoch": 2.00306176, "grad_norm": 0.9201670289039612, "learning_rate": 1.944308377229874e-05, "loss": 3.2626, "step": 42140 }, { "epoch": 2.00308736, "grad_norm": 1.0701876878738403, "learning_rate": 1.9442818755271292e-05, "loss": 3.3309, "step": 42150 }, { "epoch": 2.00311296, "grad_norm": 1.2193632125854492, "learning_rate": 1.9442553677009674e-05, "loss": 3.46, "step": 42160 }, { "epoch": 2.00313856, "grad_norm": 0.7987017035484314, "learning_rate": 1.9442288537515605e-05, "loss": 3.4223, "step": 42170 }, { "epoch": 2.00316416, "grad_norm": 0.8642638325691223, "learning_rate": 1.9442023336790807e-05, "loss": 3.4153, "step": 42180 }, { "epoch": 2.00318976, "grad_norm": 1.2820405960083008, "learning_rate": 1.9441758074836993e-05, "loss": 3.716, "step": 42190 }, { "epoch": 2.00321536, "grad_norm": 0.7560237646102905, "learning_rate": 1.944149275165589e-05, "loss": 3.5739, "step": 42200 }, { "epoch": 2.00324096, "grad_norm": 1.0044692754745483, "learning_rate": 1.9441227367249213e-05, "loss": 3.4628, "step": 42210 }, { "epoch": 2.00326656, "grad_norm": 0.8528819680213928, "learning_rate": 1.944096192161869e-05, "loss": 3.2813, "step": 42220 }, { "epoch": 2.00329216, "grad_norm": 0.7982478737831116, "learning_rate": 1.9440696414766035e-05, "loss": 3.4551, "step": 42230 }, { "epoch": 2.00331776, "grad_norm": 0.8531129956245422, "learning_rate": 1.9440430846692972e-05, "loss": 3.4145, "step": 42240 }, { "epoch": 2.00334336, "grad_norm": 0.7930809855461121, "learning_rate": 1.9440165217401223e-05, "loss": 3.4298, "step": 42250 }, { "epoch": 2.00336896, "grad_norm": 0.8083459138870239, "learning_rate": 1.9439899526892515e-05, "loss": 3.2215, "step": 42260 }, { "epoch": 2.00339456, "grad_norm": 0.9345690608024597, "learning_rate": 1.9439633775168568e-05, "loss": 3.3716, "step": 42270 }, { "epoch": 2.00342016, "grad_norm": 0.7334737777709961, "learning_rate": 1.94393679622311e-05, "loss": 3.6301, "step": 42280 }, { "epoch": 2.00344576, "grad_norm": 0.8810000419616699, "learning_rate": 1.9439102088081842e-05, "loss": 3.5419, "step": 42290 }, { "epoch": 2.00347136, "grad_norm": 0.9264945387840271, "learning_rate": 1.9438836152722517e-05, "loss": 3.5615, "step": 42300 }, { "epoch": 2.00349696, "grad_norm": 0.7354495525360107, "learning_rate": 1.9438570156154846e-05, "loss": 3.325, "step": 42310 }, { "epoch": 2.00352256, "grad_norm": 0.7954890727996826, "learning_rate": 1.9438304098380557e-05, "loss": 3.3993, "step": 42320 }, { "epoch": 2.00354816, "grad_norm": 0.7852109670639038, "learning_rate": 1.9438037979401375e-05, "loss": 3.3476, "step": 42330 }, { "epoch": 2.00357376, "grad_norm": 1.25584077835083, "learning_rate": 1.9437771799219024e-05, "loss": 3.5378, "step": 42340 }, { "epoch": 2.00359936, "grad_norm": 1.053558349609375, "learning_rate": 1.943750555783523e-05, "loss": 3.4161, "step": 42350 }, { "epoch": 2.00362496, "grad_norm": 1.1064519882202148, "learning_rate": 1.9437239255251723e-05, "loss": 3.4069, "step": 42360 }, { "epoch": 2.00365056, "grad_norm": 0.8860671520233154, "learning_rate": 1.9436972891470227e-05, "loss": 3.27, "step": 42370 }, { "epoch": 2.00367616, "grad_norm": 1.1098068952560425, "learning_rate": 1.943670646649247e-05, "loss": 3.3656, "step": 42380 }, { "epoch": 2.00370176, "grad_norm": 0.761218249797821, "learning_rate": 1.9436439980320177e-05, "loss": 3.4061, "step": 42390 }, { "epoch": 2.00372736, "grad_norm": 0.8152667880058289, "learning_rate": 1.9436173432955082e-05, "loss": 3.5649, "step": 42400 }, { "epoch": 2.00375296, "grad_norm": 0.753808319568634, "learning_rate": 1.9435906824398907e-05, "loss": 3.3733, "step": 42410 }, { "epoch": 2.00377856, "grad_norm": 0.7616114020347595, "learning_rate": 1.9435640154653386e-05, "loss": 3.3748, "step": 42420 }, { "epoch": 2.00380416, "grad_norm": 0.7541891932487488, "learning_rate": 1.9435373423720246e-05, "loss": 3.4807, "step": 42430 }, { "epoch": 2.00382976, "grad_norm": 0.7626081109046936, "learning_rate": 1.9435106631601216e-05, "loss": 3.1157, "step": 42440 }, { "epoch": 2.00385536, "grad_norm": 0.756325900554657, "learning_rate": 1.943483977829803e-05, "loss": 3.1932, "step": 42450 }, { "epoch": 2.00388096, "grad_norm": 0.8073113560676575, "learning_rate": 1.943457286381241e-05, "loss": 3.2698, "step": 42460 }, { "epoch": 2.00390656, "grad_norm": 6.041598796844482, "learning_rate": 1.94343058881461e-05, "loss": 3.3595, "step": 42470 }, { "epoch": 2.00393216, "grad_norm": 0.7474682331085205, "learning_rate": 1.943403885130082e-05, "loss": 3.297, "step": 42480 }, { "epoch": 2.00395776, "grad_norm": 0.8909639120101929, "learning_rate": 1.94337717532783e-05, "loss": 3.544, "step": 42490 }, { "epoch": 2.00398336, "grad_norm": 0.8275143504142761, "learning_rate": 1.9433504594080283e-05, "loss": 3.325, "step": 42500 }, { "epoch": 2.00400896, "grad_norm": 0.93780916929245, "learning_rate": 1.9433237373708497e-05, "loss": 3.3041, "step": 42510 }, { "epoch": 2.00403456, "grad_norm": 1.0178409814834595, "learning_rate": 1.9432970092164672e-05, "loss": 3.1806, "step": 42520 }, { "epoch": 2.00406016, "grad_norm": 0.8069435954093933, "learning_rate": 1.943270274945054e-05, "loss": 3.2676, "step": 42530 }, { "epoch": 2.00408576, "grad_norm": 0.7594735622406006, "learning_rate": 1.9432435345567843e-05, "loss": 3.3751, "step": 42540 }, { "epoch": 2.00411136, "grad_norm": 1.1220500469207764, "learning_rate": 1.9432167880518308e-05, "loss": 3.3479, "step": 42550 }, { "epoch": 2.00413696, "grad_norm": 0.8570191860198975, "learning_rate": 1.943190035430367e-05, "loss": 3.2025, "step": 42560 }, { "epoch": 2.00416256, "grad_norm": 0.7805286645889282, "learning_rate": 1.943163276692566e-05, "loss": 3.7334, "step": 42570 }, { "epoch": 2.00418816, "grad_norm": 0.844593346118927, "learning_rate": 1.9431365118386027e-05, "loss": 3.3517, "step": 42580 }, { "epoch": 2.00421376, "grad_norm": 0.8023747205734253, "learning_rate": 1.943109740868649e-05, "loss": 3.399, "step": 42590 }, { "epoch": 2.00423936, "grad_norm": 1.331939697265625, "learning_rate": 1.94308296378288e-05, "loss": 3.3496, "step": 42600 }, { "epoch": 2.00426496, "grad_norm": 0.7331408858299255, "learning_rate": 1.943056180581468e-05, "loss": 3.2028, "step": 42610 }, { "epoch": 2.00429056, "grad_norm": 0.7737034559249878, "learning_rate": 1.9430293912645878e-05, "loss": 3.3251, "step": 42620 }, { "epoch": 2.00431616, "grad_norm": 0.7464633584022522, "learning_rate": 1.9430025958324122e-05, "loss": 3.3174, "step": 42630 }, { "epoch": 2.00434176, "grad_norm": 0.9290969371795654, "learning_rate": 1.9429757942851158e-05, "loss": 3.4112, "step": 42640 }, { "epoch": 2.00436736, "grad_norm": 0.8831943869590759, "learning_rate": 1.9429489866228714e-05, "loss": 3.3338, "step": 42650 }, { "epoch": 2.00439296, "grad_norm": 0.9311950206756592, "learning_rate": 1.942922172845854e-05, "loss": 3.2221, "step": 42660 }, { "epoch": 2.00441856, "grad_norm": 0.9885244369506836, "learning_rate": 1.9428953529542367e-05, "loss": 3.4068, "step": 42670 }, { "epoch": 2.00444416, "grad_norm": 0.7935484051704407, "learning_rate": 1.9428685269481934e-05, "loss": 3.2944, "step": 42680 }, { "epoch": 2.00446976, "grad_norm": 0.8320603966712952, "learning_rate": 1.9428416948278988e-05, "loss": 3.3903, "step": 42690 }, { "epoch": 2.00449536, "grad_norm": 0.6789170503616333, "learning_rate": 1.942814856593526e-05, "loss": 3.2715, "step": 42700 }, { "epoch": 2.00452096, "grad_norm": 0.7785191535949707, "learning_rate": 1.9427880122452496e-05, "loss": 3.3682, "step": 42710 }, { "epoch": 2.00454656, "grad_norm": 0.8190088868141174, "learning_rate": 1.9427611617832432e-05, "loss": 3.3989, "step": 42720 }, { "epoch": 2.00457216, "grad_norm": 0.9458824992179871, "learning_rate": 1.9427343052076816e-05, "loss": 3.3729, "step": 42730 }, { "epoch": 2.00459776, "grad_norm": 0.8479299545288086, "learning_rate": 1.9427074425187386e-05, "loss": 3.2623, "step": 42740 }, { "epoch": 2.00462336, "grad_norm": 0.7550108432769775, "learning_rate": 1.942680573716588e-05, "loss": 3.4176, "step": 42750 }, { "epoch": 2.00464896, "grad_norm": 0.7127549648284912, "learning_rate": 1.942653698801405e-05, "loss": 3.2365, "step": 42760 }, { "epoch": 2.00467456, "grad_norm": 0.8944438099861145, "learning_rate": 1.9426268177733628e-05, "loss": 3.178, "step": 42770 }, { "epoch": 2.00470016, "grad_norm": 0.9739983081817627, "learning_rate": 1.9425999306326362e-05, "loss": 3.1062, "step": 42780 }, { "epoch": 2.00472576, "grad_norm": 0.8316158056259155, "learning_rate": 1.9425730373793998e-05, "loss": 3.3504, "step": 42790 }, { "epoch": 2.00475136, "grad_norm": 0.7777910232543945, "learning_rate": 1.9425461380138275e-05, "loss": 3.253, "step": 42800 }, { "epoch": 2.00477696, "grad_norm": 0.7728469371795654, "learning_rate": 1.9425192325360944e-05, "loss": 3.0021, "step": 42810 }, { "epoch": 2.00480256, "grad_norm": 0.7941750288009644, "learning_rate": 1.942492320946374e-05, "loss": 3.3054, "step": 42820 }, { "epoch": 2.00482816, "grad_norm": 0.7803716063499451, "learning_rate": 1.9424654032448417e-05, "loss": 3.5271, "step": 42830 }, { "epoch": 2.00485376, "grad_norm": 0.9340884685516357, "learning_rate": 1.9424384794316716e-05, "loss": 3.1237, "step": 42840 }, { "epoch": 2.00487936, "grad_norm": 0.817162811756134, "learning_rate": 1.942411549507039e-05, "loss": 3.4034, "step": 42850 }, { "epoch": 2.00490496, "grad_norm": 0.7323970794677734, "learning_rate": 1.9423846134711172e-05, "loss": 3.3765, "step": 42860 }, { "epoch": 2.00493056, "grad_norm": 0.8832039833068848, "learning_rate": 1.9423576713240823e-05, "loss": 3.3543, "step": 42870 }, { "epoch": 2.00495616, "grad_norm": 0.7306975722312927, "learning_rate": 1.9423307230661078e-05, "loss": 3.4177, "step": 42880 }, { "epoch": 2.00498176, "grad_norm": 3.4984259605407715, "learning_rate": 1.9423037686973693e-05, "loss": 3.6147, "step": 42890 }, { "epoch": 2.00500736, "grad_norm": 0.7161373496055603, "learning_rate": 1.9422768082180413e-05, "loss": 3.9815, "step": 42900 }, { "epoch": 2.00503296, "grad_norm": 0.8318964242935181, "learning_rate": 1.9422498416282982e-05, "loss": 3.1778, "step": 42910 }, { "epoch": 2.00505856, "grad_norm": 1.1448005437850952, "learning_rate": 1.942222868928316e-05, "loss": 3.52, "step": 42920 }, { "epoch": 2.00508416, "grad_norm": 0.8880332708358765, "learning_rate": 1.9421958901182683e-05, "loss": 3.4445, "step": 42930 }, { "epoch": 2.00510976, "grad_norm": 0.7576709389686584, "learning_rate": 1.9421689051983307e-05, "loss": 3.2784, "step": 42940 }, { "epoch": 2.00513536, "grad_norm": 0.858751118183136, "learning_rate": 1.9421419141686784e-05, "loss": 3.5375, "step": 42950 }, { "epoch": 2.00516096, "grad_norm": 1.053041696548462, "learning_rate": 1.9421149170294863e-05, "loss": 3.1235, "step": 42960 }, { "epoch": 2.00518656, "grad_norm": 0.8357564210891724, "learning_rate": 1.942087913780929e-05, "loss": 3.2068, "step": 42970 }, { "epoch": 2.00521216, "grad_norm": 0.8960314393043518, "learning_rate": 1.9420609044231823e-05, "loss": 3.2281, "step": 42980 }, { "epoch": 2.00523776, "grad_norm": 0.8287016153335571, "learning_rate": 1.9420338889564208e-05, "loss": 3.2917, "step": 42990 }, { "epoch": 2.00526336, "grad_norm": 0.8474080562591553, "learning_rate": 1.9420068673808197e-05, "loss": 3.3406, "step": 43000 }, { "epoch": 2.00528896, "grad_norm": 0.8462146520614624, "learning_rate": 1.941979839696555e-05, "loss": 3.3521, "step": 43010 }, { "epoch": 2.00531456, "grad_norm": 0.8648440837860107, "learning_rate": 1.941952805903801e-05, "loss": 3.2961, "step": 43020 }, { "epoch": 2.00534016, "grad_norm": 0.8927026391029358, "learning_rate": 1.9419257660027332e-05, "loss": 3.4416, "step": 43030 }, { "epoch": 2.00536576, "grad_norm": 0.8026115298271179, "learning_rate": 1.9418987199935272e-05, "loss": 3.3889, "step": 43040 }, { "epoch": 2.00539136, "grad_norm": 0.7161507606506348, "learning_rate": 1.9418716678763586e-05, "loss": 3.1786, "step": 43050 }, { "epoch": 2.00541696, "grad_norm": 0.9401573538780212, "learning_rate": 1.9418446096514027e-05, "loss": 3.5098, "step": 43060 }, { "epoch": 2.00544256, "grad_norm": 0.8352519869804382, "learning_rate": 1.9418175453188348e-05, "loss": 3.4152, "step": 43070 }, { "epoch": 2.00546816, "grad_norm": 0.8250135779380798, "learning_rate": 1.94179047487883e-05, "loss": 3.665, "step": 43080 }, { "epoch": 2.00549376, "grad_norm": 0.8948635458946228, "learning_rate": 1.941763398331564e-05, "loss": 3.6343, "step": 43090 }, { "epoch": 2.00551936, "grad_norm": 0.785982608795166, "learning_rate": 1.9417363156772134e-05, "loss": 3.3789, "step": 43100 }, { "epoch": 2.00554496, "grad_norm": 0.7773634791374207, "learning_rate": 1.9417092269159527e-05, "loss": 3.2577, "step": 43110 }, { "epoch": 2.00557056, "grad_norm": 0.7452014684677124, "learning_rate": 1.9416821320479582e-05, "loss": 3.5556, "step": 43120 }, { "epoch": 2.00559616, "grad_norm": 0.9315606951713562, "learning_rate": 1.9416550310734053e-05, "loss": 3.4523, "step": 43130 }, { "epoch": 2.00562176, "grad_norm": 0.8685118556022644, "learning_rate": 1.9416279239924693e-05, "loss": 3.3346, "step": 43140 }, { "epoch": 2.00564736, "grad_norm": 0.717784583568573, "learning_rate": 1.941600810805327e-05, "loss": 3.3255, "step": 43150 }, { "epoch": 2.00567296, "grad_norm": 0.8300848007202148, "learning_rate": 1.9415736915121534e-05, "loss": 3.3432, "step": 43160 }, { "epoch": 2.00569856, "grad_norm": 0.894929826259613, "learning_rate": 1.9415465661131247e-05, "loss": 3.3229, "step": 43170 }, { "epoch": 2.00572416, "grad_norm": 0.834225058555603, "learning_rate": 1.941519434608417e-05, "loss": 3.2139, "step": 43180 }, { "epoch": 2.00574976, "grad_norm": 0.8167722225189209, "learning_rate": 1.941492296998206e-05, "loss": 3.3779, "step": 43190 }, { "epoch": 2.00577536, "grad_norm": 0.8008967041969299, "learning_rate": 1.941465153282667e-05, "loss": 3.3595, "step": 43200 }, { "epoch": 2.00580096, "grad_norm": 0.6970574259757996, "learning_rate": 1.941438003461977e-05, "loss": 3.2647, "step": 43210 }, { "epoch": 2.00582656, "grad_norm": 0.79864901304245, "learning_rate": 1.9414108475363123e-05, "loss": 3.4624, "step": 43220 }, { "epoch": 2.00585216, "grad_norm": 1.6633002758026123, "learning_rate": 1.941383685505848e-05, "loss": 3.4282, "step": 43230 }, { "epoch": 2.00587776, "grad_norm": 0.9510062336921692, "learning_rate": 1.941356517370761e-05, "loss": 3.2536, "step": 43240 }, { "epoch": 2.00590336, "grad_norm": 0.8215360641479492, "learning_rate": 1.9413293431312267e-05, "loss": 3.2634, "step": 43250 }, { "epoch": 2.00592896, "grad_norm": 0.7544612288475037, "learning_rate": 1.9413021627874224e-05, "loss": 3.4355, "step": 43260 }, { "epoch": 2.00595456, "grad_norm": 0.8463592529296875, "learning_rate": 1.9412749763395232e-05, "loss": 3.3707, "step": 43270 }, { "epoch": 2.00598016, "grad_norm": 0.755536675453186, "learning_rate": 1.941247783787706e-05, "loss": 3.3794, "step": 43280 }, { "epoch": 2.00600576, "grad_norm": 0.7719517946243286, "learning_rate": 1.9412205851321472e-05, "loss": 3.2151, "step": 43290 }, { "epoch": 2.00603136, "grad_norm": 0.8030624389648438, "learning_rate": 1.941193380373023e-05, "loss": 3.4238, "step": 43300 }, { "epoch": 2.00605696, "grad_norm": 1.1497125625610352, "learning_rate": 1.94116616951051e-05, "loss": 3.4154, "step": 43310 }, { "epoch": 2.00608256, "grad_norm": 0.9513013362884521, "learning_rate": 1.9411389525447844e-05, "loss": 3.3428, "step": 43320 }, { "epoch": 2.00610816, "grad_norm": 0.7261576056480408, "learning_rate": 1.9411117294760226e-05, "loss": 3.5202, "step": 43330 }, { "epoch": 2.00613376, "grad_norm": 0.7615669965744019, "learning_rate": 1.9410845003044017e-05, "loss": 3.4403, "step": 43340 }, { "epoch": 2.00615936, "grad_norm": 0.8112645149230957, "learning_rate": 1.9410572650300977e-05, "loss": 3.3181, "step": 43350 }, { "epoch": 2.00618496, "grad_norm": 0.8362141251564026, "learning_rate": 1.9410300236532876e-05, "loss": 3.5058, "step": 43360 }, { "epoch": 2.00621056, "grad_norm": 1.272189736366272, "learning_rate": 1.941002776174148e-05, "loss": 3.4961, "step": 43370 }, { "epoch": 2.00623616, "grad_norm": 0.903727650642395, "learning_rate": 1.940975522592855e-05, "loss": 3.5599, "step": 43380 }, { "epoch": 2.00626176, "grad_norm": 0.7754265069961548, "learning_rate": 1.9409482629095864e-05, "loss": 3.3098, "step": 43390 }, { "epoch": 2.00628736, "grad_norm": 0.8850070238113403, "learning_rate": 1.940920997124518e-05, "loss": 3.3183, "step": 43400 }, { "epoch": 2.00631296, "grad_norm": 0.7632863521575928, "learning_rate": 1.9408937252378272e-05, "loss": 3.4994, "step": 43410 }, { "epoch": 2.00633856, "grad_norm": 1.1384581327438354, "learning_rate": 1.9408664472496907e-05, "loss": 3.4802, "step": 43420 }, { "epoch": 2.00636416, "grad_norm": 0.8072495460510254, "learning_rate": 1.9408391631602854e-05, "loss": 3.5166, "step": 43430 }, { "epoch": 2.00638976, "grad_norm": 0.7452082633972168, "learning_rate": 1.9408118729697878e-05, "loss": 3.2942, "step": 43440 }, { "epoch": 2.00641536, "grad_norm": 0.8815551400184631, "learning_rate": 1.9407845766783755e-05, "loss": 3.282, "step": 43450 }, { "epoch": 2.00644096, "grad_norm": 0.8525589108467102, "learning_rate": 1.940757274286225e-05, "loss": 3.3441, "step": 43460 }, { "epoch": 2.00646656, "grad_norm": 1.1192713975906372, "learning_rate": 1.940729965793514e-05, "loss": 3.4923, "step": 43470 }, { "epoch": 2.00649216, "grad_norm": 0.947225034236908, "learning_rate": 1.940702651200419e-05, "loss": 3.2131, "step": 43480 }, { "epoch": 2.00651776, "grad_norm": 0.7493842840194702, "learning_rate": 1.9406753305071174e-05, "loss": 3.4016, "step": 43490 }, { "epoch": 2.00654336, "grad_norm": 0.7503270506858826, "learning_rate": 1.940648003713786e-05, "loss": 3.4672, "step": 43500 }, { "epoch": 2.00656896, "grad_norm": 0.7335778474807739, "learning_rate": 1.9406206708206026e-05, "loss": 3.4024, "step": 43510 }, { "epoch": 2.00659456, "grad_norm": 0.8418139815330505, "learning_rate": 1.9405933318277438e-05, "loss": 3.4464, "step": 43520 }, { "epoch": 2.00662016, "grad_norm": 0.9329205751419067, "learning_rate": 1.9405659867353875e-05, "loss": 3.6259, "step": 43530 }, { "epoch": 2.00664576, "grad_norm": 0.870391845703125, "learning_rate": 1.940538635543711e-05, "loss": 3.2338, "step": 43540 }, { "epoch": 2.00667136, "grad_norm": 0.8236629366874695, "learning_rate": 1.940511278252891e-05, "loss": 3.2655, "step": 43550 }, { "epoch": 2.00669696, "grad_norm": 0.9945080280303955, "learning_rate": 1.9404839148631054e-05, "loss": 3.3826, "step": 43560 }, { "epoch": 2.00672256, "grad_norm": 0.873941957950592, "learning_rate": 1.9404565453745316e-05, "loss": 3.2417, "step": 43570 }, { "epoch": 2.00674816, "grad_norm": 0.6874954700469971, "learning_rate": 1.940429169787347e-05, "loss": 3.3463, "step": 43580 }, { "epoch": 2.00677376, "grad_norm": 0.8484203219413757, "learning_rate": 1.940401788101729e-05, "loss": 3.4752, "step": 43590 }, { "epoch": 2.00679936, "grad_norm": 0.8349608182907104, "learning_rate": 1.9403744003178558e-05, "loss": 3.4216, "step": 43600 }, { "epoch": 2.00682496, "grad_norm": 0.8044393062591553, "learning_rate": 1.940347006435904e-05, "loss": 3.4123, "step": 43610 }, { "epoch": 2.00685056, "grad_norm": 0.9599044919013977, "learning_rate": 1.9403196064560522e-05, "loss": 3.29, "step": 43620 }, { "epoch": 2.00687616, "grad_norm": 0.9689335823059082, "learning_rate": 1.9402922003784774e-05, "loss": 3.339, "step": 43630 }, { "epoch": 2.00690176, "grad_norm": 0.8972829580307007, "learning_rate": 1.9402647882033577e-05, "loss": 3.3184, "step": 43640 }, { "epoch": 2.00692736, "grad_norm": 0.8599194884300232, "learning_rate": 1.940237369930871e-05, "loss": 3.1341, "step": 43650 }, { "epoch": 2.00695296, "grad_norm": 0.8312616944313049, "learning_rate": 1.9402099455611944e-05, "loss": 3.3419, "step": 43660 }, { "epoch": 2.00697856, "grad_norm": 0.8778447508811951, "learning_rate": 1.9401825150945062e-05, "loss": 3.3984, "step": 43670 }, { "epoch": 2.00700416, "grad_norm": 1.507351040840149, "learning_rate": 1.9401550785309843e-05, "loss": 3.4934, "step": 43680 }, { "epoch": 2.00702976, "grad_norm": 0.7597252130508423, "learning_rate": 1.9401276358708066e-05, "loss": 3.2677, "step": 43690 }, { "epoch": 2.00705536, "grad_norm": 0.9331944584846497, "learning_rate": 1.9401001871141513e-05, "loss": 3.2003, "step": 43700 }, { "epoch": 2.00708096, "grad_norm": 0.8288670778274536, "learning_rate": 1.940072732261196e-05, "loss": 3.214, "step": 43710 }, { "epoch": 2.00710656, "grad_norm": 0.788266122341156, "learning_rate": 1.9400452713121186e-05, "loss": 3.3575, "step": 43720 }, { "epoch": 2.00713216, "grad_norm": 0.8606510758399963, "learning_rate": 1.940017804267098e-05, "loss": 3.4489, "step": 43730 }, { "epoch": 2.00715776, "grad_norm": 0.8713988661766052, "learning_rate": 1.9399903311263112e-05, "loss": 3.4007, "step": 43740 }, { "epoch": 2.00718336, "grad_norm": 0.7563909888267517, "learning_rate": 1.9399628518899373e-05, "loss": 3.2525, "step": 43750 }, { "epoch": 2.00720896, "grad_norm": 0.8134707808494568, "learning_rate": 1.939935366558154e-05, "loss": 3.312, "step": 43760 }, { "epoch": 2.00723456, "grad_norm": 0.8346105217933655, "learning_rate": 1.9399078751311396e-05, "loss": 3.5911, "step": 43770 }, { "epoch": 2.00726016, "grad_norm": 0.7846826910972595, "learning_rate": 1.9398803776090722e-05, "loss": 3.256, "step": 43780 }, { "epoch": 2.00728576, "grad_norm": 0.9000205397605896, "learning_rate": 1.939852873992131e-05, "loss": 3.2793, "step": 43790 }, { "epoch": 2.00731136, "grad_norm": 0.7674069404602051, "learning_rate": 1.939825364280493e-05, "loss": 3.2158, "step": 43800 }, { "epoch": 2.00733696, "grad_norm": 0.8092535138130188, "learning_rate": 1.9397978484743376e-05, "loss": 3.1953, "step": 43810 }, { "epoch": 2.00736256, "grad_norm": 0.866233766078949, "learning_rate": 1.939770326573843e-05, "loss": 3.1939, "step": 43820 }, { "epoch": 2.00738816, "grad_norm": 0.9026059508323669, "learning_rate": 1.9397427985791874e-05, "loss": 3.3483, "step": 43830 }, { "epoch": 2.00741376, "grad_norm": 0.9076895713806152, "learning_rate": 1.9397152644905495e-05, "loss": 3.4198, "step": 43840 }, { "epoch": 2.00743936, "grad_norm": 0.7962132096290588, "learning_rate": 1.939687724308108e-05, "loss": 3.3291, "step": 43850 }, { "epoch": 2.00746496, "grad_norm": 0.830407977104187, "learning_rate": 1.9396601780320415e-05, "loss": 3.3372, "step": 43860 }, { "epoch": 2.00749056, "grad_norm": 0.7995101809501648, "learning_rate": 1.939632625662528e-05, "loss": 3.3811, "step": 43870 }, { "epoch": 2.00751616, "grad_norm": 1.0583992004394531, "learning_rate": 1.939605067199747e-05, "loss": 3.3266, "step": 43880 }, { "epoch": 2.00754176, "grad_norm": 0.7632225751876831, "learning_rate": 1.9395775026438768e-05, "loss": 3.0902, "step": 43890 }, { "epoch": 2.00756736, "grad_norm": 0.7473136186599731, "learning_rate": 1.9395499319950963e-05, "loss": 3.1298, "step": 43900 }, { "epoch": 2.00759296, "grad_norm": 0.8088992238044739, "learning_rate": 1.939522355253584e-05, "loss": 3.2619, "step": 43910 }, { "epoch": 2.00761856, "grad_norm": 0.7867265343666077, "learning_rate": 1.939494772419519e-05, "loss": 3.391, "step": 43920 }, { "epoch": 2.00764416, "grad_norm": 0.9312130808830261, "learning_rate": 1.93946718349308e-05, "loss": 3.5381, "step": 43930 }, { "epoch": 2.00766976, "grad_norm": 0.7362246513366699, "learning_rate": 1.9394395884744465e-05, "loss": 3.4709, "step": 43940 }, { "epoch": 2.00769536, "grad_norm": 0.8123497366905212, "learning_rate": 1.9394119873637964e-05, "loss": 3.2726, "step": 43950 }, { "epoch": 2.00772096, "grad_norm": 0.8654736280441284, "learning_rate": 1.9393843801613094e-05, "loss": 3.4589, "step": 43960 }, { "epoch": 2.00774656, "grad_norm": 0.851921796798706, "learning_rate": 1.9393567668671644e-05, "loss": 3.4266, "step": 43970 }, { "epoch": 2.00777216, "grad_norm": 0.9386417865753174, "learning_rate": 1.9393291474815403e-05, "loss": 3.3181, "step": 43980 }, { "epoch": 2.00779776, "grad_norm": 1.3885009288787842, "learning_rate": 1.939301522004616e-05, "loss": 3.436, "step": 43990 }, { "epoch": 2.00782336, "grad_norm": 0.9801545143127441, "learning_rate": 1.9392738904365716e-05, "loss": 3.1729, "step": 44000 }, { "epoch": 2.00784896, "grad_norm": 0.8122705221176147, "learning_rate": 1.9392462527775852e-05, "loss": 3.4112, "step": 44010 }, { "epoch": 2.00787456, "grad_norm": 0.8596982359886169, "learning_rate": 1.9392186090278366e-05, "loss": 3.2743, "step": 44020 }, { "epoch": 2.00790016, "grad_norm": 0.8298258781433105, "learning_rate": 1.939190959187505e-05, "loss": 3.2286, "step": 44030 }, { "epoch": 2.00792576, "grad_norm": 0.8616294264793396, "learning_rate": 1.9391633032567698e-05, "loss": 3.3762, "step": 44040 }, { "epoch": 2.00795136, "grad_norm": 2.2627110481262207, "learning_rate": 1.9391356412358097e-05, "loss": 3.3744, "step": 44050 }, { "epoch": 2.00797696, "grad_norm": 0.7949172258377075, "learning_rate": 1.939107973124805e-05, "loss": 3.2555, "step": 44060 }, { "epoch": 2.00800256, "grad_norm": 0.8509587049484253, "learning_rate": 1.9390802989239347e-05, "loss": 3.4083, "step": 44070 }, { "epoch": 2.00802816, "grad_norm": 0.7729277610778809, "learning_rate": 1.9390526186333782e-05, "loss": 3.1189, "step": 44080 }, { "epoch": 2.00805376, "grad_norm": 0.8243733048439026, "learning_rate": 1.9390249322533145e-05, "loss": 3.3162, "step": 44090 }, { "epoch": 2.00807936, "grad_norm": 0.8040969371795654, "learning_rate": 1.938997239783924e-05, "loss": 3.1664, "step": 44100 }, { "epoch": 2.00810496, "grad_norm": 0.793065071105957, "learning_rate": 1.938969541225386e-05, "loss": 3.3027, "step": 44110 }, { "epoch": 2.00813056, "grad_norm": 0.8641877174377441, "learning_rate": 1.9389418365778802e-05, "loss": 3.3134, "step": 44120 }, { "epoch": 2.00815616, "grad_norm": 0.7851496934890747, "learning_rate": 1.9389141258415863e-05, "loss": 3.396, "step": 44130 }, { "epoch": 2.00818176, "grad_norm": 0.794818103313446, "learning_rate": 1.9388864090166835e-05, "loss": 3.3068, "step": 44140 }, { "epoch": 2.00820736, "grad_norm": 0.7803791761398315, "learning_rate": 1.938858686103352e-05, "loss": 3.3455, "step": 44150 }, { "epoch": 2.00823296, "grad_norm": 1.0016427040100098, "learning_rate": 1.9388309571017714e-05, "loss": 3.3445, "step": 44160 }, { "epoch": 2.00825856, "grad_norm": 0.8108857274055481, "learning_rate": 1.9388032220121212e-05, "loss": 3.4943, "step": 44170 }, { "epoch": 2.00828416, "grad_norm": 1.1307684183120728, "learning_rate": 1.9387754808345824e-05, "loss": 3.5578, "step": 44180 }, { "epoch": 2.00830976, "grad_norm": 0.8283551931381226, "learning_rate": 1.9387477335693334e-05, "loss": 3.3866, "step": 44190 }, { "epoch": 2.00833536, "grad_norm": 0.8738216161727905, "learning_rate": 1.938719980216555e-05, "loss": 3.1742, "step": 44200 }, { "epoch": 2.00836096, "grad_norm": 0.831752598285675, "learning_rate": 1.9386922207764272e-05, "loss": 3.2338, "step": 44210 }, { "epoch": 2.00838656, "grad_norm": 0.8775911927223206, "learning_rate": 1.93866445524913e-05, "loss": 3.4298, "step": 44220 }, { "epoch": 2.00841216, "grad_norm": 1.084606647491455, "learning_rate": 1.9386366836348428e-05, "loss": 3.3362, "step": 44230 }, { "epoch": 2.00843776, "grad_norm": 0.8707879185676575, "learning_rate": 1.9386089059337464e-05, "loss": 3.4552, "step": 44240 }, { "epoch": 2.00846336, "grad_norm": 0.7232532501220703, "learning_rate": 1.9385811221460206e-05, "loss": 3.1681, "step": 44250 }, { "epoch": 2.00848896, "grad_norm": 0.9294540286064148, "learning_rate": 1.9385533322718455e-05, "loss": 3.2937, "step": 44260 }, { "epoch": 2.00851456, "grad_norm": 0.8998742699623108, "learning_rate": 1.9385255363114018e-05, "loss": 3.4858, "step": 44270 }, { "epoch": 2.00854016, "grad_norm": 0.910643994808197, "learning_rate": 1.9384977342648695e-05, "loss": 3.2693, "step": 44280 }, { "epoch": 2.00856576, "grad_norm": 0.7371492981910706, "learning_rate": 1.9384699261324286e-05, "loss": 3.1715, "step": 44290 }, { "epoch": 2.00859136, "grad_norm": 0.767402708530426, "learning_rate": 1.9384421119142595e-05, "loss": 3.2641, "step": 44300 }, { "epoch": 2.00861696, "grad_norm": 0.9245347380638123, "learning_rate": 1.938414291610543e-05, "loss": 3.2077, "step": 44310 }, { "epoch": 2.00864256, "grad_norm": 0.8987410068511963, "learning_rate": 1.938386465221459e-05, "loss": 3.2877, "step": 44320 }, { "epoch": 2.00866816, "grad_norm": 0.9183356165885925, "learning_rate": 1.9383586327471882e-05, "loss": 3.227, "step": 44330 }, { "epoch": 2.00869376, "grad_norm": 0.7601636648178101, "learning_rate": 1.9383307941879113e-05, "loss": 3.4463, "step": 44340 }, { "epoch": 2.00871936, "grad_norm": 0.9007793664932251, "learning_rate": 1.9383029495438085e-05, "loss": 3.2784, "step": 44350 }, { "epoch": 2.00874496, "grad_norm": 0.797709584236145, "learning_rate": 1.93827509881506e-05, "loss": 3.4766, "step": 44360 }, { "epoch": 2.00877056, "grad_norm": 1.063438057899475, "learning_rate": 1.9382472420018478e-05, "loss": 3.204, "step": 44370 }, { "epoch": 2.00879616, "grad_norm": 1.0012916326522827, "learning_rate": 1.9382193791043507e-05, "loss": 3.1942, "step": 44380 }, { "epoch": 2.00882176, "grad_norm": 0.7918983101844788, "learning_rate": 1.9381915101227507e-05, "loss": 3.2824, "step": 44390 }, { "epoch": 2.00884736, "grad_norm": 0.8158343434333801, "learning_rate": 1.9381636350572276e-05, "loss": 3.2947, "step": 44400 }, { "epoch": 2.00887296, "grad_norm": 0.8324217200279236, "learning_rate": 1.938135753907963e-05, "loss": 3.2938, "step": 44410 }, { "epoch": 2.00889856, "grad_norm": 0.8405930399894714, "learning_rate": 1.9381078666751376e-05, "loss": 3.3034, "step": 44420 }, { "epoch": 2.00892416, "grad_norm": 0.8459407687187195, "learning_rate": 1.938079973358932e-05, "loss": 3.5859, "step": 44430 }, { "epoch": 2.00894976, "grad_norm": 1.223720669746399, "learning_rate": 1.9380520739595268e-05, "loss": 3.5048, "step": 44440 }, { "epoch": 2.00897536, "grad_norm": 0.8949490785598755, "learning_rate": 1.938024168477103e-05, "loss": 3.4639, "step": 44450 }, { "epoch": 2.00900096, "grad_norm": 0.8244220614433289, "learning_rate": 1.9379962569118422e-05, "loss": 3.2456, "step": 44460 }, { "epoch": 2.00902656, "grad_norm": 0.9646621346473694, "learning_rate": 1.9379683392639246e-05, "loss": 3.6405, "step": 44470 }, { "epoch": 2.00905216, "grad_norm": 0.8465326428413391, "learning_rate": 1.937940415533532e-05, "loss": 3.4215, "step": 44480 }, { "epoch": 2.00907776, "grad_norm": 0.8225970268249512, "learning_rate": 1.937912485720845e-05, "loss": 3.2667, "step": 44490 }, { "epoch": 2.00910336, "grad_norm": 1.0218722820281982, "learning_rate": 1.9378845498260448e-05, "loss": 3.299, "step": 44500 }, { "epoch": 2.00912896, "grad_norm": 0.9395655989646912, "learning_rate": 1.9378566078493125e-05, "loss": 3.4595, "step": 44510 }, { "epoch": 2.00915456, "grad_norm": 1.0614802837371826, "learning_rate": 1.937828659790829e-05, "loss": 3.2224, "step": 44520 }, { "epoch": 2.00918016, "grad_norm": 0.8481937646865845, "learning_rate": 1.9378007056507762e-05, "loss": 3.1451, "step": 44530 }, { "epoch": 2.00920576, "grad_norm": 1.0100189447402954, "learning_rate": 1.937772745429335e-05, "loss": 3.4108, "step": 44540 }, { "epoch": 2.00923136, "grad_norm": 0.7798805236816406, "learning_rate": 1.937744779126687e-05, "loss": 3.257, "step": 44550 }, { "epoch": 2.00925696, "grad_norm": 1.0305794477462769, "learning_rate": 1.937716806743013e-05, "loss": 3.5061, "step": 44560 }, { "epoch": 2.00928256, "grad_norm": 0.8462920784950256, "learning_rate": 1.937688828278495e-05, "loss": 3.2684, "step": 44570 }, { "epoch": 2.00930816, "grad_norm": 0.8766233325004578, "learning_rate": 1.937660843733314e-05, "loss": 3.2838, "step": 44580 }, { "epoch": 2.00933376, "grad_norm": 0.85310298204422, "learning_rate": 1.9376328531076515e-05, "loss": 3.337, "step": 44590 }, { "epoch": 2.00935936, "grad_norm": 0.7807635068893433, "learning_rate": 1.9376048564016895e-05, "loss": 3.4276, "step": 44600 }, { "epoch": 2.00938496, "grad_norm": 0.9179003238677979, "learning_rate": 1.937576853615609e-05, "loss": 3.513, "step": 44610 }, { "epoch": 2.00941056, "grad_norm": 0.8737676739692688, "learning_rate": 1.9375488447495917e-05, "loss": 3.35, "step": 44620 }, { "epoch": 2.00943616, "grad_norm": 0.9086151123046875, "learning_rate": 1.9375208298038194e-05, "loss": 3.3389, "step": 44630 }, { "epoch": 2.00946176, "grad_norm": 0.7694413065910339, "learning_rate": 1.937492808778474e-05, "loss": 3.4855, "step": 44640 }, { "epoch": 2.00948736, "grad_norm": 0.8234263062477112, "learning_rate": 1.9374647816737365e-05, "loss": 3.2487, "step": 44650 }, { "epoch": 2.00951296, "grad_norm": 0.9367861747741699, "learning_rate": 1.937436748489789e-05, "loss": 3.2857, "step": 44660 }, { "epoch": 2.00953856, "grad_norm": 0.9883652329444885, "learning_rate": 1.9374087092268135e-05, "loss": 3.218, "step": 44670 }, { "epoch": 2.00956416, "grad_norm": 0.9277909994125366, "learning_rate": 1.9373806638849915e-05, "loss": 3.5035, "step": 44680 }, { "epoch": 2.00958976, "grad_norm": 0.8616549968719482, "learning_rate": 1.937352612464505e-05, "loss": 3.2352, "step": 44690 }, { "epoch": 2.00961536, "grad_norm": 0.7359296679496765, "learning_rate": 1.9373245549655362e-05, "loss": 3.2402, "step": 44700 }, { "epoch": 2.00964096, "grad_norm": 0.8442458510398865, "learning_rate": 1.9372964913882665e-05, "loss": 3.3396, "step": 44710 }, { "epoch": 2.00966656, "grad_norm": 0.9040518403053284, "learning_rate": 1.937268421732878e-05, "loss": 3.4269, "step": 44720 }, { "epoch": 2.00969216, "grad_norm": 1.009523630142212, "learning_rate": 1.9372403459995532e-05, "loss": 3.3864, "step": 44730 }, { "epoch": 2.00971776, "grad_norm": 0.7776818871498108, "learning_rate": 1.937212264188474e-05, "loss": 3.6786, "step": 44740 }, { "epoch": 2.00974336, "grad_norm": 1.174385666847229, "learning_rate": 1.937184176299822e-05, "loss": 3.2831, "step": 44750 }, { "epoch": 2.00976896, "grad_norm": 0.7915676236152649, "learning_rate": 1.9371560823337797e-05, "loss": 3.2918, "step": 44760 }, { "epoch": 2.00979456, "grad_norm": 1.0528897047042847, "learning_rate": 1.9371279822905296e-05, "loss": 3.2978, "step": 44770 }, { "epoch": 2.00982016, "grad_norm": 0.8654963970184326, "learning_rate": 1.9370998761702532e-05, "loss": 3.3738, "step": 44780 }, { "epoch": 2.00984576, "grad_norm": 0.785496711730957, "learning_rate": 1.9370717639731334e-05, "loss": 3.3742, "step": 44790 }, { "epoch": 2.00987136, "grad_norm": 1.7914525270462036, "learning_rate": 1.937043645699352e-05, "loss": 2.9859, "step": 44800 }, { "epoch": 2.00989696, "grad_norm": 0.9218713641166687, "learning_rate": 1.937015521349092e-05, "loss": 3.3562, "step": 44810 }, { "epoch": 2.00992256, "grad_norm": 0.8970616459846497, "learning_rate": 1.936987390922535e-05, "loss": 3.4395, "step": 44820 }, { "epoch": 2.00994816, "grad_norm": 0.8212094306945801, "learning_rate": 1.9369592544198645e-05, "loss": 3.2178, "step": 44830 }, { "epoch": 2.00997376, "grad_norm": 0.7757486701011658, "learning_rate": 1.9369311118412612e-05, "loss": 3.4804, "step": 44840 }, { "epoch": 2.00999936, "grad_norm": 1.4643969535827637, "learning_rate": 1.9369029631869096e-05, "loss": 3.3413, "step": 44850 }, { "epoch": 2.01002496, "grad_norm": 0.925460159778595, "learning_rate": 1.9368748084569905e-05, "loss": 3.1581, "step": 44860 }, { "epoch": 2.01005056, "grad_norm": 1.056118130683899, "learning_rate": 1.9368466476516877e-05, "loss": 3.4681, "step": 44870 }, { "epoch": 2.01007616, "grad_norm": 0.7514379024505615, "learning_rate": 1.9368184807711833e-05, "loss": 3.4036, "step": 44880 }, { "epoch": 2.01010176, "grad_norm": 0.9432419538497925, "learning_rate": 1.93679030781566e-05, "loss": 3.0618, "step": 44890 }, { "epoch": 2.01012736, "grad_norm": 0.7651854753494263, "learning_rate": 1.9367621287853005e-05, "loss": 3.288, "step": 44900 }, { "epoch": 2.01015296, "grad_norm": 0.8428829312324524, "learning_rate": 1.9367339436802877e-05, "loss": 3.2149, "step": 44910 }, { "epoch": 2.01017856, "grad_norm": 0.7878979444503784, "learning_rate": 1.9367057525008038e-05, "loss": 3.491, "step": 44920 }, { "epoch": 2.01020416, "grad_norm": 0.8385924100875854, "learning_rate": 1.9366775552470325e-05, "loss": 3.5031, "step": 44930 }, { "epoch": 2.01022976, "grad_norm": 0.7518267035484314, "learning_rate": 1.9366493519191558e-05, "loss": 3.5179, "step": 44940 }, { "epoch": 2.01025536, "grad_norm": 0.9563095569610596, "learning_rate": 1.9366211425173573e-05, "loss": 3.2389, "step": 44950 }, { "epoch": 2.01028096, "grad_norm": 0.8478637337684631, "learning_rate": 1.9365929270418195e-05, "loss": 3.5045, "step": 44960 }, { "epoch": 2.01030656, "grad_norm": 0.8754043579101562, "learning_rate": 1.9365647054927256e-05, "loss": 3.2391, "step": 44970 }, { "epoch": 2.01033216, "grad_norm": 1.0109573602676392, "learning_rate": 1.9365364778702583e-05, "loss": 3.4282, "step": 44980 }, { "epoch": 2.01035776, "grad_norm": 0.9181865453720093, "learning_rate": 1.936508244174601e-05, "loss": 3.4513, "step": 44990 }, { "epoch": 2.01038336, "grad_norm": 0.9398366808891296, "learning_rate": 1.9364800044059365e-05, "loss": 3.1853, "step": 45000 }, { "epoch": 2.01040896, "grad_norm": 1.8391650915145874, "learning_rate": 1.9364517585644482e-05, "loss": 3.5014, "step": 45010 }, { "epoch": 2.01043456, "grad_norm": 0.8324936628341675, "learning_rate": 1.9364235066503192e-05, "loss": 3.321, "step": 45020 }, { "epoch": 2.01046016, "grad_norm": 0.7785422801971436, "learning_rate": 1.9363952486637326e-05, "loss": 3.3583, "step": 45030 }, { "epoch": 2.01048576, "grad_norm": 0.8873288631439209, "learning_rate": 1.9363669846048716e-05, "loss": 3.3184, "step": 45040 }, { "epoch": 2.01051136, "grad_norm": 0.7288200259208679, "learning_rate": 1.9363387144739194e-05, "loss": 3.3337, "step": 45050 }, { "epoch": 2.01053696, "grad_norm": 0.9578812718391418, "learning_rate": 1.93631043827106e-05, "loss": 3.0906, "step": 45060 }, { "epoch": 2.01056256, "grad_norm": 0.8314415216445923, "learning_rate": 1.9362821559964755e-05, "loss": 3.4307, "step": 45070 }, { "epoch": 2.01058816, "grad_norm": 0.9613412022590637, "learning_rate": 1.9362538676503508e-05, "loss": 3.3292, "step": 45080 }, { "epoch": 2.01061376, "grad_norm": 0.948477029800415, "learning_rate": 1.936225573232868e-05, "loss": 3.3443, "step": 45090 }, { "epoch": 2.01063936, "grad_norm": 0.8293636441230774, "learning_rate": 1.9361972727442112e-05, "loss": 3.47, "step": 45100 }, { "epoch": 2.01066496, "grad_norm": 0.801087498664856, "learning_rate": 1.936168966184564e-05, "loss": 3.1973, "step": 45110 }, { "epoch": 2.01069056, "grad_norm": 0.8767096400260925, "learning_rate": 1.93614065355411e-05, "loss": 3.2838, "step": 45120 }, { "epoch": 2.01071616, "grad_norm": 0.8419689536094666, "learning_rate": 1.9361123348530327e-05, "loss": 3.1344, "step": 45130 }, { "epoch": 2.01074176, "grad_norm": 0.9556267261505127, "learning_rate": 1.9360840100815158e-05, "loss": 3.5128, "step": 45140 }, { "epoch": 2.01076736, "grad_norm": 0.9048598408699036, "learning_rate": 1.9360556792397423e-05, "loss": 3.4445, "step": 45150 }, { "epoch": 2.01079296, "grad_norm": 0.7284145355224609, "learning_rate": 1.936027342327897e-05, "loss": 3.2524, "step": 45160 }, { "epoch": 2.01081856, "grad_norm": 0.9997521638870239, "learning_rate": 1.935998999346163e-05, "loss": 3.0993, "step": 45170 }, { "epoch": 2.01084416, "grad_norm": 1.093866229057312, "learning_rate": 1.9359706502947246e-05, "loss": 3.307, "step": 45180 }, { "epoch": 2.01086976, "grad_norm": 0.8423595428466797, "learning_rate": 1.9359422951737648e-05, "loss": 3.3615, "step": 45190 }, { "epoch": 2.01089536, "grad_norm": 0.8813533186912537, "learning_rate": 1.9359139339834683e-05, "loss": 3.2383, "step": 45200 }, { "epoch": 2.01092096, "grad_norm": 1.0094197988510132, "learning_rate": 1.9358855667240184e-05, "loss": 3.0997, "step": 45210 }, { "epoch": 2.01094656, "grad_norm": 0.8422948718070984, "learning_rate": 1.9358571933955992e-05, "loss": 3.5003, "step": 45220 }, { "epoch": 2.01097216, "grad_norm": 0.7321246862411499, "learning_rate": 1.9358288139983953e-05, "loss": 3.2053, "step": 45230 }, { "epoch": 2.01099776, "grad_norm": 1.0771522521972656, "learning_rate": 1.93580042853259e-05, "loss": 3.4317, "step": 45240 }, { "epoch": 2.01102336, "grad_norm": 0.9207165241241455, "learning_rate": 1.9357720369983674e-05, "loss": 3.4252, "step": 45250 }, { "epoch": 2.01104896, "grad_norm": 0.9176579117774963, "learning_rate": 1.935743639395912e-05, "loss": 3.2174, "step": 45260 }, { "epoch": 2.01107456, "grad_norm": 0.9243714213371277, "learning_rate": 1.935715235725408e-05, "loss": 3.3559, "step": 45270 }, { "epoch": 2.01110016, "grad_norm": 0.9237977266311646, "learning_rate": 1.9356868259870393e-05, "loss": 3.4195, "step": 45280 }, { "epoch": 2.01112576, "grad_norm": 0.8006853461265564, "learning_rate": 1.93565841018099e-05, "loss": 3.3651, "step": 45290 }, { "epoch": 2.01115136, "grad_norm": 0.8023568391799927, "learning_rate": 1.935629988307445e-05, "loss": 3.3112, "step": 45300 }, { "epoch": 2.01117696, "grad_norm": 0.8417333364486694, "learning_rate": 1.9356015603665877e-05, "loss": 3.4124, "step": 45310 }, { "epoch": 2.01120256, "grad_norm": 0.896838366985321, "learning_rate": 1.935573126358603e-05, "loss": 3.4132, "step": 45320 }, { "epoch": 2.01122816, "grad_norm": 0.8614992499351501, "learning_rate": 1.9355446862836756e-05, "loss": 3.1635, "step": 45330 }, { "epoch": 2.01125376, "grad_norm": 0.8367227911949158, "learning_rate": 1.9355162401419895e-05, "loss": 3.1747, "step": 45340 }, { "epoch": 2.01127936, "grad_norm": 0.8872437477111816, "learning_rate": 1.9354877879337287e-05, "loss": 3.3807, "step": 45350 }, { "epoch": 2.01130496, "grad_norm": 0.8647363185882568, "learning_rate": 1.9354593296590786e-05, "loss": 3.3686, "step": 45360 }, { "epoch": 2.01133056, "grad_norm": 0.8310175538063049, "learning_rate": 1.935430865318223e-05, "loss": 3.2959, "step": 45370 }, { "epoch": 2.01135616, "grad_norm": 0.8492024540901184, "learning_rate": 1.9354023949113476e-05, "loss": 3.3569, "step": 45380 }, { "epoch": 2.01138176, "grad_norm": 0.9865275025367737, "learning_rate": 1.9353739184386358e-05, "loss": 3.3588, "step": 45390 }, { "epoch": 2.01140736, "grad_norm": 0.7825721502304077, "learning_rate": 1.935345435900273e-05, "loss": 3.0759, "step": 45400 }, { "epoch": 2.01143296, "grad_norm": 0.9518454670906067, "learning_rate": 1.935316947296443e-05, "loss": 3.2236, "step": 45410 }, { "epoch": 2.01145856, "grad_norm": 0.7509359121322632, "learning_rate": 1.9352884526273315e-05, "loss": 3.3819, "step": 45420 }, { "epoch": 2.01148416, "grad_norm": 0.7784274220466614, "learning_rate": 1.935259951893123e-05, "loss": 3.4589, "step": 45430 }, { "epoch": 2.01150976, "grad_norm": 0.8526710271835327, "learning_rate": 1.9352314450940025e-05, "loss": 3.2975, "step": 45440 }, { "epoch": 2.01153536, "grad_norm": 0.7996909022331238, "learning_rate": 1.9352029322301545e-05, "loss": 3.2904, "step": 45450 }, { "epoch": 2.01156096, "grad_norm": 0.7847739458084106, "learning_rate": 1.9351744133017637e-05, "loss": 3.2962, "step": 45460 }, { "epoch": 2.01158656, "grad_norm": 0.7883087992668152, "learning_rate": 1.935145888309016e-05, "loss": 3.305, "step": 45470 }, { "epoch": 2.01161216, "grad_norm": 0.796135663986206, "learning_rate": 1.935117357252095e-05, "loss": 3.3003, "step": 45480 }, { "epoch": 2.01163776, "grad_norm": 0.9042933583259583, "learning_rate": 1.935088820131187e-05, "loss": 3.0864, "step": 45490 }, { "epoch": 2.01166336, "grad_norm": 0.9341117143630981, "learning_rate": 1.9350602769464763e-05, "loss": 3.194, "step": 45500 }, { "epoch": 2.01168896, "grad_norm": 0.7772819995880127, "learning_rate": 1.935031727698148e-05, "loss": 3.4819, "step": 45510 }, { "epoch": 2.01171456, "grad_norm": 0.8194620609283447, "learning_rate": 1.935003172386388e-05, "loss": 3.5188, "step": 45520 }, { "epoch": 2.01174016, "grad_norm": 0.7996038794517517, "learning_rate": 1.9349746110113807e-05, "loss": 3.3254, "step": 45530 }, { "epoch": 2.01176576, "grad_norm": 0.9237564206123352, "learning_rate": 1.9349460435733114e-05, "loss": 3.2327, "step": 45540 }, { "epoch": 2.01179136, "grad_norm": 0.8896080851554871, "learning_rate": 1.9349174700723655e-05, "loss": 3.3747, "step": 45550 }, { "epoch": 2.01181696, "grad_norm": 0.9095994234085083, "learning_rate": 1.9348888905087282e-05, "loss": 3.4694, "step": 45560 }, { "epoch": 2.01184256, "grad_norm": 0.8475574254989624, "learning_rate": 1.934860304882585e-05, "loss": 3.466, "step": 45570 }, { "epoch": 2.01186816, "grad_norm": 0.8016791343688965, "learning_rate": 1.9348317131941213e-05, "loss": 3.2837, "step": 45580 }, { "epoch": 2.01189376, "grad_norm": 0.8058612942695618, "learning_rate": 1.9348031154435224e-05, "loss": 3.0648, "step": 45590 }, { "epoch": 2.01191936, "grad_norm": 0.7177265882492065, "learning_rate": 1.9347745116309737e-05, "loss": 3.2941, "step": 45600 }, { "epoch": 2.01194496, "grad_norm": 0.8005762100219727, "learning_rate": 1.9347459017566605e-05, "loss": 3.4088, "step": 45610 }, { "epoch": 2.01197056, "grad_norm": 0.8839901089668274, "learning_rate": 1.9347172858207687e-05, "loss": 3.3897, "step": 45620 }, { "epoch": 2.01199616, "grad_norm": 0.9292455315589905, "learning_rate": 1.9346886638234838e-05, "loss": 3.1824, "step": 45630 }, { "epoch": 2.01202176, "grad_norm": 0.8403429388999939, "learning_rate": 1.9346600357649913e-05, "loss": 3.3338, "step": 45640 }, { "epoch": 2.01204736, "grad_norm": 0.9038650393486023, "learning_rate": 1.9346314016454767e-05, "loss": 3.2296, "step": 45650 }, { "epoch": 2.01207296, "grad_norm": 1.0657334327697754, "learning_rate": 1.9346027614651262e-05, "loss": 3.0498, "step": 45660 }, { "epoch": 2.01209856, "grad_norm": 0.9745791554450989, "learning_rate": 1.934574115224125e-05, "loss": 3.2377, "step": 45670 }, { "epoch": 2.01212416, "grad_norm": 0.911446213722229, "learning_rate": 1.934545462922659e-05, "loss": 3.4455, "step": 45680 }, { "epoch": 2.01214976, "grad_norm": 0.9747688174247742, "learning_rate": 1.934516804560914e-05, "loss": 3.333, "step": 45690 }, { "epoch": 2.01217536, "grad_norm": 0.838780403137207, "learning_rate": 1.934488140139076e-05, "loss": 3.2644, "step": 45700 }, { "epoch": 2.01220096, "grad_norm": 0.8067469000816345, "learning_rate": 1.9344594696573307e-05, "loss": 3.1873, "step": 45710 }, { "epoch": 2.01222656, "grad_norm": 0.8382267355918884, "learning_rate": 1.934430793115864e-05, "loss": 3.1463, "step": 45720 }, { "epoch": 2.01225216, "grad_norm": 0.8640348315238953, "learning_rate": 1.9344021105148624e-05, "loss": 3.2487, "step": 45730 }, { "epoch": 2.01227776, "grad_norm": 0.7594525218009949, "learning_rate": 1.934373421854511e-05, "loss": 3.3813, "step": 45740 }, { "epoch": 2.01230336, "grad_norm": 0.9002884030342102, "learning_rate": 1.9343447271349963e-05, "loss": 3.2354, "step": 45750 }, { "epoch": 2.01232896, "grad_norm": 0.9722437262535095, "learning_rate": 1.9343160263565044e-05, "loss": 3.3476, "step": 45760 }, { "epoch": 2.01235456, "grad_norm": 0.8380035161972046, "learning_rate": 1.9342873195192214e-05, "loss": 3.2851, "step": 45770 }, { "epoch": 2.01238016, "grad_norm": 0.7849381566047668, "learning_rate": 1.9342586066233333e-05, "loss": 3.2377, "step": 45780 }, { "epoch": 2.01240576, "grad_norm": 1.2555127143859863, "learning_rate": 1.9342298876690265e-05, "loss": 3.2873, "step": 45790 }, { "epoch": 2.01243136, "grad_norm": 0.7119148373603821, "learning_rate": 1.9342011626564873e-05, "loss": 3.0637, "step": 45800 }, { "epoch": 2.01245696, "grad_norm": 1.514107584953308, "learning_rate": 1.934172431585902e-05, "loss": 3.0562, "step": 45810 }, { "epoch": 2.01248256, "grad_norm": 0.8009641766548157, "learning_rate": 1.9341436944574564e-05, "loss": 3.3446, "step": 45820 }, { "epoch": 2.01250816, "grad_norm": 0.9188195466995239, "learning_rate": 1.9341149512713374e-05, "loss": 3.439, "step": 45830 }, { "epoch": 2.01253376, "grad_norm": 0.9853242635726929, "learning_rate": 1.9340862020277314e-05, "loss": 3.2941, "step": 45840 }, { "epoch": 2.01255936, "grad_norm": 0.847341001033783, "learning_rate": 1.9340574467268244e-05, "loss": 3.1536, "step": 45850 }, { "epoch": 2.01258496, "grad_norm": 0.8440207839012146, "learning_rate": 1.934028685368803e-05, "loss": 3.2766, "step": 45860 }, { "epoch": 2.01261056, "grad_norm": 1.1593058109283447, "learning_rate": 1.933999917953854e-05, "loss": 3.1488, "step": 45870 }, { "epoch": 2.01263616, "grad_norm": 0.7893470525741577, "learning_rate": 1.9339711444821636e-05, "loss": 3.3341, "step": 45880 }, { "epoch": 2.01266176, "grad_norm": 0.8600842356681824, "learning_rate": 1.9339423649539185e-05, "loss": 3.2018, "step": 45890 }, { "epoch": 2.01268736, "grad_norm": 0.8961156606674194, "learning_rate": 1.9339135793693055e-05, "loss": 3.354, "step": 45900 }, { "epoch": 2.01271296, "grad_norm": 0.8042927980422974, "learning_rate": 1.9338847877285116e-05, "loss": 3.3822, "step": 45910 }, { "epoch": 2.01273856, "grad_norm": 0.99775230884552, "learning_rate": 1.933855990031722e-05, "loss": 3.0575, "step": 45920 }, { "epoch": 2.01276416, "grad_norm": 0.8795724511146545, "learning_rate": 1.9338271862791255e-05, "loss": 3.4434, "step": 45930 }, { "epoch": 2.01278976, "grad_norm": 0.9266924858093262, "learning_rate": 1.9337983764709073e-05, "loss": 3.5626, "step": 45940 }, { "epoch": 2.01281536, "grad_norm": 1.0039705038070679, "learning_rate": 1.9337695606072547e-05, "loss": 3.5532, "step": 45950 }, { "epoch": 2.01284096, "grad_norm": 0.8540468811988831, "learning_rate": 1.933740738688355e-05, "loss": 3.4306, "step": 45960 }, { "epoch": 2.01286656, "grad_norm": 1.0673397779464722, "learning_rate": 1.933711910714395e-05, "loss": 3.4283, "step": 45970 }, { "epoch": 2.01289216, "grad_norm": 1.0232058763504028, "learning_rate": 1.933683076685561e-05, "loss": 3.2231, "step": 45980 }, { "epoch": 2.01291776, "grad_norm": 0.8653547167778015, "learning_rate": 1.9336542366020404e-05, "loss": 3.1931, "step": 45990 }, { "epoch": 2.01294336, "grad_norm": 0.8669877648353577, "learning_rate": 1.9336253904640205e-05, "loss": 3.2351, "step": 46000 }, { "epoch": 2.01296896, "grad_norm": 0.8280447125434875, "learning_rate": 1.9335965382716875e-05, "loss": 3.2464, "step": 46010 }, { "epoch": 2.01299456, "grad_norm": 1.1902505159378052, "learning_rate": 1.9335676800252296e-05, "loss": 3.3538, "step": 46020 }, { "epoch": 2.01302016, "grad_norm": 0.8045118451118469, "learning_rate": 1.9335388157248335e-05, "loss": 3.4244, "step": 46030 }, { "epoch": 2.01304576, "grad_norm": 0.8739528656005859, "learning_rate": 1.9335099453706857e-05, "loss": 3.4176, "step": 46040 }, { "epoch": 2.01307136, "grad_norm": 1.2521947622299194, "learning_rate": 1.9334810689629747e-05, "loss": 3.4756, "step": 46050 }, { "epoch": 2.01309696, "grad_norm": 0.9397523999214172, "learning_rate": 1.9334521865018865e-05, "loss": 3.318, "step": 46060 }, { "epoch": 2.01312256, "grad_norm": 1.337814211845398, "learning_rate": 1.9334232979876092e-05, "loss": 3.198, "step": 46070 }, { "epoch": 2.01314816, "grad_norm": 0.8908774256706238, "learning_rate": 1.9333944034203297e-05, "loss": 3.3343, "step": 46080 }, { "epoch": 2.01317376, "grad_norm": 0.8713054656982422, "learning_rate": 1.9333655028002355e-05, "loss": 3.1465, "step": 46090 }, { "epoch": 2.01319936, "grad_norm": 0.9514274597167969, "learning_rate": 1.9333365961275145e-05, "loss": 3.2231, "step": 46100 }, { "epoch": 2.01322496, "grad_norm": 0.8743636608123779, "learning_rate": 1.9333076834023535e-05, "loss": 3.2815, "step": 46110 }, { "epoch": 2.01325056, "grad_norm": 0.8461045622825623, "learning_rate": 1.93327876462494e-05, "loss": 3.5511, "step": 46120 }, { "epoch": 2.01327616, "grad_norm": 0.8024890422821045, "learning_rate": 1.933249839795462e-05, "loss": 3.2523, "step": 46130 }, { "epoch": 2.01330176, "grad_norm": 0.9169163107872009, "learning_rate": 1.933220908914107e-05, "loss": 3.4873, "step": 46140 }, { "epoch": 2.01332736, "grad_norm": 1.0672749280929565, "learning_rate": 1.933191971981062e-05, "loss": 3.4565, "step": 46150 }, { "epoch": 2.01335296, "grad_norm": 0.9649171829223633, "learning_rate": 1.9331630289965157e-05, "loss": 3.6273, "step": 46160 }, { "epoch": 2.01337856, "grad_norm": 2.288767099380493, "learning_rate": 1.9331340799606548e-05, "loss": 3.4488, "step": 46170 }, { "epoch": 2.01340416, "grad_norm": 0.8575504422187805, "learning_rate": 1.933105124873667e-05, "loss": 3.3425, "step": 46180 }, { "epoch": 2.01342976, "grad_norm": 0.8762047290802002, "learning_rate": 1.933076163735741e-05, "loss": 3.2429, "step": 46190 }, { "epoch": 2.01345536, "grad_norm": 0.8375945687294006, "learning_rate": 1.9330471965470635e-05, "loss": 3.6101, "step": 46200 }, { "epoch": 2.01348096, "grad_norm": 0.8733552098274231, "learning_rate": 1.9330182233078233e-05, "loss": 3.2496, "step": 46210 }, { "epoch": 2.01350656, "grad_norm": 0.8339818120002747, "learning_rate": 1.9329892440182078e-05, "loss": 3.5559, "step": 46220 }, { "epoch": 2.01353216, "grad_norm": 0.7763755321502686, "learning_rate": 1.932960258678405e-05, "loss": 3.482, "step": 46230 }, { "epoch": 2.01355776, "grad_norm": 0.9808928966522217, "learning_rate": 1.932931267288603e-05, "loss": 3.5686, "step": 46240 }, { "epoch": 2.01358336, "grad_norm": 0.8733530640602112, "learning_rate": 1.9329022698489895e-05, "loss": 3.3334, "step": 46250 }, { "epoch": 2.01360896, "grad_norm": 0.795516312122345, "learning_rate": 1.9328732663597528e-05, "loss": 3.25, "step": 46260 }, { "epoch": 2.01363456, "grad_norm": 0.8542541265487671, "learning_rate": 1.9328442568210814e-05, "loss": 3.4567, "step": 46270 }, { "epoch": 2.01366016, "grad_norm": 0.9593092203140259, "learning_rate": 1.932815241233162e-05, "loss": 3.4978, "step": 46280 }, { "epoch": 2.01368576, "grad_norm": 0.8675176501274109, "learning_rate": 1.9327862195961843e-05, "loss": 3.2045, "step": 46290 }, { "epoch": 2.01371136, "grad_norm": 1.0188523530960083, "learning_rate": 1.9327571919103357e-05, "loss": 3.4969, "step": 46300 }, { "epoch": 2.01373696, "grad_norm": 1.288403034210205, "learning_rate": 1.9327281581758046e-05, "loss": 3.4686, "step": 46310 }, { "epoch": 2.01376256, "grad_norm": 0.963250458240509, "learning_rate": 1.9326991183927792e-05, "loss": 3.4255, "step": 46320 }, { "epoch": 2.01378816, "grad_norm": 0.8852587342262268, "learning_rate": 1.932670072561448e-05, "loss": 3.3467, "step": 46330 }, { "epoch": 2.01381376, "grad_norm": 0.9345433115959167, "learning_rate": 1.9326410206819993e-05, "loss": 3.4399, "step": 46340 }, { "epoch": 2.01383936, "grad_norm": 0.8604869246482849, "learning_rate": 1.9326119627546212e-05, "loss": 3.4408, "step": 46350 }, { "epoch": 2.01386496, "grad_norm": 0.8421736359596252, "learning_rate": 1.9325828987795024e-05, "loss": 3.4374, "step": 46360 }, { "epoch": 2.01389056, "grad_norm": 0.8294129371643066, "learning_rate": 1.9325538287568316e-05, "loss": 3.3977, "step": 46370 }, { "epoch": 2.01391616, "grad_norm": 0.902603030204773, "learning_rate": 1.932524752686797e-05, "loss": 3.4581, "step": 46380 }, { "epoch": 2.01394176, "grad_norm": 0.8594677448272705, "learning_rate": 1.932495670569587e-05, "loss": 3.4028, "step": 46390 }, { "epoch": 2.01396736, "grad_norm": 0.8309995532035828, "learning_rate": 1.9324665824053905e-05, "loss": 3.4444, "step": 46400 }, { "epoch": 2.01399296, "grad_norm": 0.7946376204490662, "learning_rate": 1.9324374881943962e-05, "loss": 3.3444, "step": 46410 }, { "epoch": 2.01401856, "grad_norm": 0.8205945491790771, "learning_rate": 1.9324083879367926e-05, "loss": 3.1659, "step": 46420 }, { "epoch": 2.01404416, "grad_norm": 0.9402245283126831, "learning_rate": 1.9323792816327682e-05, "loss": 3.3531, "step": 46430 }, { "epoch": 2.01406976, "grad_norm": 0.9070357084274292, "learning_rate": 1.9323501692825122e-05, "loss": 3.1628, "step": 46440 }, { "epoch": 2.01409536, "grad_norm": 0.8845308423042297, "learning_rate": 1.932321050886213e-05, "loss": 3.543, "step": 46450 }, { "epoch": 2.01412096, "grad_norm": 0.8598587512969971, "learning_rate": 1.9322919264440596e-05, "loss": 3.3376, "step": 46460 }, { "epoch": 2.01414656, "grad_norm": 1.017756700515747, "learning_rate": 1.9322627959562407e-05, "loss": 3.332, "step": 46470 }, { "epoch": 2.01417216, "grad_norm": 0.821219265460968, "learning_rate": 1.9322336594229454e-05, "loss": 3.4246, "step": 46480 }, { "epoch": 2.01419776, "grad_norm": 0.8330651521682739, "learning_rate": 1.9322045168443627e-05, "loss": 3.2303, "step": 46490 }, { "epoch": 2.01422336, "grad_norm": 0.7780098915100098, "learning_rate": 1.932175368220681e-05, "loss": 3.3606, "step": 46500 }, { "epoch": 2.01424896, "grad_norm": 1.7992048263549805, "learning_rate": 1.9321462135520905e-05, "loss": 3.3783, "step": 46510 }, { "epoch": 2.01427456, "grad_norm": 1.811747670173645, "learning_rate": 1.932117052838779e-05, "loss": 3.3454, "step": 46520 }, { "epoch": 2.01430016, "grad_norm": 0.9820504784584045, "learning_rate": 1.9320878860809365e-05, "loss": 3.378, "step": 46530 }, { "epoch": 2.01432576, "grad_norm": 0.9206458926200867, "learning_rate": 1.9320587132787516e-05, "loss": 3.379, "step": 46540 }, { "epoch": 2.01435136, "grad_norm": 0.8983114957809448, "learning_rate": 1.9320295344324136e-05, "loss": 3.3788, "step": 46550 }, { "epoch": 2.01437696, "grad_norm": 0.8122814893722534, "learning_rate": 1.9320003495421117e-05, "loss": 3.3121, "step": 46560 }, { "epoch": 2.01440256, "grad_norm": 0.7558205127716064, "learning_rate": 1.9319711586080356e-05, "loss": 3.3445, "step": 46570 }, { "epoch": 2.01442816, "grad_norm": 0.9365245699882507, "learning_rate": 1.931941961630374e-05, "loss": 3.3193, "step": 46580 }, { "epoch": 2.01445376, "grad_norm": 0.7885687947273254, "learning_rate": 1.9319127586093165e-05, "loss": 3.2619, "step": 46590 }, { "epoch": 2.01447936, "grad_norm": 0.8287737369537354, "learning_rate": 1.9318835495450524e-05, "loss": 3.2647, "step": 46600 }, { "epoch": 2.01450496, "grad_norm": 0.8164097666740417, "learning_rate": 1.931854334437771e-05, "loss": 3.3952, "step": 46610 }, { "epoch": 2.01453056, "grad_norm": 0.9771021008491516, "learning_rate": 1.931825113287662e-05, "loss": 3.3611, "step": 46620 }, { "epoch": 2.01455616, "grad_norm": 0.7818776965141296, "learning_rate": 1.9317958860949152e-05, "loss": 3.3882, "step": 46630 }, { "epoch": 2.01458176, "grad_norm": 0.8122216463088989, "learning_rate": 1.931766652859719e-05, "loss": 3.4219, "step": 46640 }, { "epoch": 2.01460736, "grad_norm": 1.0326308012008667, "learning_rate": 1.9317374135822642e-05, "loss": 3.5849, "step": 46650 }, { "epoch": 2.01463296, "grad_norm": 0.9421903491020203, "learning_rate": 1.9317081682627396e-05, "loss": 2.9237, "step": 46660 }, { "epoch": 2.01465856, "grad_norm": 0.8246031403541565, "learning_rate": 1.9316789169013354e-05, "loss": 3.3565, "step": 46670 }, { "epoch": 2.01468416, "grad_norm": 0.8027917146682739, "learning_rate": 1.9316496594982406e-05, "loss": 3.3532, "step": 46680 }, { "epoch": 2.01470976, "grad_norm": 0.8755142688751221, "learning_rate": 1.9316203960536458e-05, "loss": 3.3034, "step": 46690 }, { "epoch": 2.01473536, "grad_norm": 0.9783415198326111, "learning_rate": 1.9315911265677402e-05, "loss": 3.1993, "step": 46700 }, { "epoch": 2.01476096, "grad_norm": 0.8043222427368164, "learning_rate": 1.931561851040714e-05, "loss": 3.2404, "step": 46710 }, { "epoch": 2.01478656, "grad_norm": 0.9814298152923584, "learning_rate": 1.9315325694727566e-05, "loss": 3.3657, "step": 46720 }, { "epoch": 2.01481216, "grad_norm": 0.8817463517189026, "learning_rate": 1.9315032818640577e-05, "loss": 3.3746, "step": 46730 }, { "epoch": 2.01483776, "grad_norm": 1.284407377243042, "learning_rate": 1.931473988214808e-05, "loss": 3.3138, "step": 46740 }, { "epoch": 2.01486336, "grad_norm": 0.8207771182060242, "learning_rate": 1.9314446885251967e-05, "loss": 3.3907, "step": 46750 }, { "epoch": 2.01488896, "grad_norm": 0.8067053556442261, "learning_rate": 1.9314153827954144e-05, "loss": 3.4203, "step": 46760 }, { "epoch": 2.01491456, "grad_norm": 1.189260721206665, "learning_rate": 1.9313860710256505e-05, "loss": 3.2978, "step": 46770 }, { "epoch": 2.01494016, "grad_norm": 0.9282261729240417, "learning_rate": 1.931356753216096e-05, "loss": 3.295, "step": 46780 }, { "epoch": 2.01496576, "grad_norm": 1.1627376079559326, "learning_rate": 1.9313274293669405e-05, "loss": 3.4476, "step": 46790 }, { "epoch": 2.01499136, "grad_norm": 0.8461252450942993, "learning_rate": 1.9312980994783735e-05, "loss": 3.523, "step": 46800 }, { "epoch": 2.01501696, "grad_norm": 0.9249751567840576, "learning_rate": 1.9312687635505866e-05, "loss": 3.4686, "step": 46810 }, { "epoch": 2.01504256, "grad_norm": 0.8077806234359741, "learning_rate": 1.9312394215837685e-05, "loss": 3.3694, "step": 46820 }, { "epoch": 2.01506816, "grad_norm": 0.8075308799743652, "learning_rate": 1.931210073578111e-05, "loss": 3.2913, "step": 46830 }, { "epoch": 2.01509376, "grad_norm": 0.8311822414398193, "learning_rate": 1.931180719533803e-05, "loss": 3.4466, "step": 46840 }, { "epoch": 2.01511936, "grad_norm": 0.8818550705909729, "learning_rate": 1.931151359451036e-05, "loss": 3.5527, "step": 46850 }, { "epoch": 2.01514496, "grad_norm": 0.8214224576950073, "learning_rate": 1.9311219933299998e-05, "loss": 3.3183, "step": 46860 }, { "epoch": 2.01517056, "grad_norm": 0.8832558989524841, "learning_rate": 1.9310926211708848e-05, "loss": 3.6399, "step": 46870 }, { "epoch": 2.01519616, "grad_norm": 0.9241962432861328, "learning_rate": 1.9310632429738817e-05, "loss": 3.4033, "step": 46880 }, { "epoch": 2.01522176, "grad_norm": 0.80161052942276, "learning_rate": 1.9310338587391807e-05, "loss": 3.2524, "step": 46890 }, { "epoch": 2.01524736, "grad_norm": 0.8685395121574402, "learning_rate": 1.9310044684669728e-05, "loss": 3.1646, "step": 46900 }, { "epoch": 2.01527296, "grad_norm": 0.8970438838005066, "learning_rate": 1.9309750721574482e-05, "loss": 3.2301, "step": 46910 }, { "epoch": 2.01529856, "grad_norm": 1.0542004108428955, "learning_rate": 1.930945669810798e-05, "loss": 3.2151, "step": 46920 }, { "epoch": 2.01532416, "grad_norm": 1.2343945503234863, "learning_rate": 1.9309162614272123e-05, "loss": 3.2993, "step": 46930 }, { "epoch": 2.01534976, "grad_norm": 0.8467494249343872, "learning_rate": 1.930886847006882e-05, "loss": 3.5006, "step": 46940 }, { "epoch": 2.01537536, "grad_norm": 0.8175849318504333, "learning_rate": 1.930857426549998e-05, "loss": 3.3384, "step": 46950 }, { "epoch": 2.01540096, "grad_norm": 0.8355841040611267, "learning_rate": 1.930828000056751e-05, "loss": 3.1298, "step": 46960 }, { "epoch": 3.00001024, "grad_norm": 0.9134494662284851, "learning_rate": 1.930798567527332e-05, "loss": 3.4347, "step": 46970 }, { "epoch": 3.00003584, "grad_norm": 0.8662451505661011, "learning_rate": 1.930769128961931e-05, "loss": 3.2902, "step": 46980 }, { "epoch": 3.00006144, "grad_norm": 0.8769835829734802, "learning_rate": 1.9307396843607404e-05, "loss": 3.5014, "step": 46990 }, { "epoch": 3.00008704, "grad_norm": 0.739088237285614, "learning_rate": 1.9307102337239496e-05, "loss": 3.1177, "step": 47000 }, { "epoch": 3.00011264, "grad_norm": 0.755979597568512, "learning_rate": 1.9306807770517505e-05, "loss": 3.107, "step": 47010 }, { "epoch": 3.00013824, "grad_norm": 0.7216302156448364, "learning_rate": 1.930651314344334e-05, "loss": 3.1394, "step": 47020 }, { "epoch": 3.00016384, "grad_norm": 0.8089730143547058, "learning_rate": 1.9306218456018914e-05, "loss": 3.3417, "step": 47030 }, { "epoch": 3.00018944, "grad_norm": 0.9689599871635437, "learning_rate": 1.930592370824613e-05, "loss": 3.1462, "step": 47040 }, { "epoch": 3.00021504, "grad_norm": 0.8560208678245544, "learning_rate": 1.9305628900126904e-05, "loss": 3.028, "step": 47050 }, { "epoch": 3.00024064, "grad_norm": 0.8823887705802917, "learning_rate": 1.9305334031663148e-05, "loss": 3.0684, "step": 47060 }, { "epoch": 3.00026624, "grad_norm": 1.2212971448898315, "learning_rate": 1.9305039102856775e-05, "loss": 3.7274, "step": 47070 }, { "epoch": 3.00029184, "grad_norm": 0.8350810408592224, "learning_rate": 1.9304744113709696e-05, "loss": 3.3574, "step": 47080 }, { "epoch": 3.00031744, "grad_norm": 0.9538081884384155, "learning_rate": 1.9304449064223825e-05, "loss": 3.0844, "step": 47090 }, { "epoch": 3.00034304, "grad_norm": 0.7796695232391357, "learning_rate": 1.9304153954401074e-05, "loss": 3.1889, "step": 47100 }, { "epoch": 3.00036864, "grad_norm": 0.8187363147735596, "learning_rate": 1.930385878424336e-05, "loss": 3.0979, "step": 47110 }, { "epoch": 3.00039424, "grad_norm": 0.7312328815460205, "learning_rate": 1.930356355375259e-05, "loss": 3.3463, "step": 47120 }, { "epoch": 3.00041984, "grad_norm": 0.8351225852966309, "learning_rate": 1.9303268262930683e-05, "loss": 3.1109, "step": 47130 }, { "epoch": 3.00044544, "grad_norm": 0.7864370346069336, "learning_rate": 1.9302972911779558e-05, "loss": 3.229, "step": 47140 }, { "epoch": 3.00047104, "grad_norm": 1.508212924003601, "learning_rate": 1.9302677500301123e-05, "loss": 3.0238, "step": 47150 }, { "epoch": 3.00049664, "grad_norm": 0.7586274743080139, "learning_rate": 1.9302382028497296e-05, "loss": 3.3949, "step": 47160 }, { "epoch": 3.00052224, "grad_norm": 0.8348051309585571, "learning_rate": 1.930208649637e-05, "loss": 3.157, "step": 47170 }, { "epoch": 3.00054784, "grad_norm": 0.8261901140213013, "learning_rate": 1.930179090392114e-05, "loss": 3.2325, "step": 47180 }, { "epoch": 3.00057344, "grad_norm": 0.7139047384262085, "learning_rate": 1.9301495251152636e-05, "loss": 3.1579, "step": 47190 }, { "epoch": 3.00059904, "grad_norm": 0.9000856876373291, "learning_rate": 1.9301199538066412e-05, "loss": 3.4822, "step": 47200 }, { "epoch": 3.00062464, "grad_norm": 0.9047387838363647, "learning_rate": 1.930090376466438e-05, "loss": 3.105, "step": 47210 }, { "epoch": 3.00065024, "grad_norm": 0.7663514614105225, "learning_rate": 1.9300607930948456e-05, "loss": 3.3243, "step": 47220 }, { "epoch": 3.00067584, "grad_norm": 1.010168433189392, "learning_rate": 1.9300312036920562e-05, "loss": 3.1226, "step": 47230 }, { "epoch": 3.00070144, "grad_norm": 0.7621601819992065, "learning_rate": 1.930001608258262e-05, "loss": 3.0876, "step": 47240 }, { "epoch": 3.00072704, "grad_norm": 0.8633519411087036, "learning_rate": 1.929972006793654e-05, "loss": 3.1226, "step": 47250 }, { "epoch": 3.00075264, "grad_norm": 0.7064617276191711, "learning_rate": 1.9299423992984252e-05, "loss": 3.1304, "step": 47260 }, { "epoch": 3.00077824, "grad_norm": 0.7666050791740417, "learning_rate": 1.929912785772767e-05, "loss": 3.1136, "step": 47270 }, { "epoch": 3.00080384, "grad_norm": 0.935614824295044, "learning_rate": 1.929883166216871e-05, "loss": 3.237, "step": 47280 }, { "epoch": 3.00082944, "grad_norm": 0.7870259881019592, "learning_rate": 1.9298535406309305e-05, "loss": 3.2983, "step": 47290 }, { "epoch": 3.00085504, "grad_norm": 0.8237197399139404, "learning_rate": 1.9298239090151364e-05, "loss": 3.2059, "step": 47300 }, { "epoch": 3.00088064, "grad_norm": 0.8332134485244751, "learning_rate": 1.9297942713696817e-05, "loss": 3.0905, "step": 47310 }, { "epoch": 3.00090624, "grad_norm": 0.7579834461212158, "learning_rate": 1.9297646276947582e-05, "loss": 3.1485, "step": 47320 }, { "epoch": 3.00093184, "grad_norm": 0.8795813918113708, "learning_rate": 1.9297349779905583e-05, "loss": 3.3199, "step": 47330 }, { "epoch": 3.00095744, "grad_norm": 0.8283055424690247, "learning_rate": 1.9297053222572737e-05, "loss": 3.2177, "step": 47340 }, { "epoch": 3.00098304, "grad_norm": 0.7829639911651611, "learning_rate": 1.9296756604950976e-05, "loss": 3.2869, "step": 47350 }, { "epoch": 3.00100864, "grad_norm": 0.7452629208564758, "learning_rate": 1.9296459927042218e-05, "loss": 3.2339, "step": 47360 }, { "epoch": 3.00103424, "grad_norm": 0.8824563026428223, "learning_rate": 1.929616318884839e-05, "loss": 3.3051, "step": 47370 }, { "epoch": 3.00105984, "grad_norm": 1.4687252044677734, "learning_rate": 1.929586639037141e-05, "loss": 3.1145, "step": 47380 }, { "epoch": 3.00108544, "grad_norm": 0.7616282105445862, "learning_rate": 1.929556953161321e-05, "loss": 3.4008, "step": 47390 }, { "epoch": 3.00111104, "grad_norm": 0.7999421954154968, "learning_rate": 1.929527261257571e-05, "loss": 3.1901, "step": 47400 }, { "epoch": 3.00113664, "grad_norm": 0.7424261569976807, "learning_rate": 1.929497563326084e-05, "loss": 3.2307, "step": 47410 }, { "epoch": 3.00116224, "grad_norm": 0.811145544052124, "learning_rate": 1.9294678593670522e-05, "loss": 3.0311, "step": 47420 }, { "epoch": 3.00118784, "grad_norm": 1.0431123971939087, "learning_rate": 1.9294381493806684e-05, "loss": 3.0066, "step": 47430 }, { "epoch": 3.00121344, "grad_norm": 0.8788858652114868, "learning_rate": 1.929408433367125e-05, "loss": 3.0903, "step": 47440 }, { "epoch": 3.00123904, "grad_norm": 0.8297306299209595, "learning_rate": 1.9293787113266153e-05, "loss": 3.1086, "step": 47450 }, { "epoch": 3.00126464, "grad_norm": 0.7843814492225647, "learning_rate": 1.9293489832593316e-05, "loss": 3.1621, "step": 47460 }, { "epoch": 3.00129024, "grad_norm": 0.8444182872772217, "learning_rate": 1.9293192491654665e-05, "loss": 3.1774, "step": 47470 }, { "epoch": 3.00131584, "grad_norm": 0.7991797924041748, "learning_rate": 1.929289509045213e-05, "loss": 2.9869, "step": 47480 }, { "epoch": 3.00134144, "grad_norm": 3.1001763343811035, "learning_rate": 1.9292597628987643e-05, "loss": 3.0539, "step": 47490 }, { "epoch": 3.00136704, "grad_norm": 0.7232283353805542, "learning_rate": 1.9292300107263128e-05, "loss": 3.1392, "step": 47500 }, { "epoch": 3.00139264, "grad_norm": 0.8263140320777893, "learning_rate": 1.9292002525280516e-05, "loss": 3.3092, "step": 47510 }, { "epoch": 3.00141824, "grad_norm": 0.7313686013221741, "learning_rate": 1.929170488304174e-05, "loss": 3.2118, "step": 47520 }, { "epoch": 3.00144384, "grad_norm": 1.2495604753494263, "learning_rate": 1.9291407180548724e-05, "loss": 3.1596, "step": 47530 }, { "epoch": 3.00146944, "grad_norm": 0.9430737495422363, "learning_rate": 1.9291109417803403e-05, "loss": 3.0068, "step": 47540 }, { "epoch": 3.00149504, "grad_norm": 0.8663125038146973, "learning_rate": 1.9290811594807706e-05, "loss": 3.1139, "step": 47550 }, { "epoch": 3.00152064, "grad_norm": 0.741861879825592, "learning_rate": 1.9290513711563566e-05, "loss": 3.0977, "step": 47560 }, { "epoch": 3.00154624, "grad_norm": 0.8164311051368713, "learning_rate": 1.9290215768072917e-05, "loss": 2.995, "step": 47570 }, { "epoch": 3.00157184, "grad_norm": 0.8284039497375488, "learning_rate": 1.9289917764337683e-05, "loss": 3.0573, "step": 47580 }, { "epoch": 3.00159744, "grad_norm": 1.1393606662750244, "learning_rate": 1.92896197003598e-05, "loss": 3.4416, "step": 47590 }, { "epoch": 3.00162304, "grad_norm": 0.8369698524475098, "learning_rate": 1.9289321576141205e-05, "loss": 3.0782, "step": 47600 }, { "epoch": 3.00164864, "grad_norm": 0.7968539595603943, "learning_rate": 1.928902339168383e-05, "loss": 3.1383, "step": 47610 }, { "epoch": 3.00167424, "grad_norm": 0.802893340587616, "learning_rate": 1.9288725146989607e-05, "loss": 3.4356, "step": 47620 }, { "epoch": 3.00169984, "grad_norm": 0.7826440930366516, "learning_rate": 1.9288426842060465e-05, "loss": 3.1419, "step": 47630 }, { "epoch": 3.00172544, "grad_norm": 0.8394054770469666, "learning_rate": 1.928812847689835e-05, "loss": 3.1399, "step": 47640 }, { "epoch": 3.00175104, "grad_norm": 0.7954345941543579, "learning_rate": 1.9287830051505184e-05, "loss": 3.3604, "step": 47650 }, { "epoch": 3.00177664, "grad_norm": 0.7139648199081421, "learning_rate": 1.9287531565882915e-05, "loss": 2.992, "step": 47660 }, { "epoch": 3.00180224, "grad_norm": 0.9439398646354675, "learning_rate": 1.9287233020033465e-05, "loss": 2.9605, "step": 47670 }, { "epoch": 3.00182784, "grad_norm": 0.8541329503059387, "learning_rate": 1.9286934413958784e-05, "loss": 3.0282, "step": 47680 }, { "epoch": 3.00185344, "grad_norm": 0.9688230752944946, "learning_rate": 1.92866357476608e-05, "loss": 3.1735, "step": 47690 }, { "epoch": 3.00187904, "grad_norm": 0.8951865434646606, "learning_rate": 1.9286337021141448e-05, "loss": 3.2045, "step": 47700 }, { "epoch": 3.00190464, "grad_norm": 0.9027773141860962, "learning_rate": 1.9286038234402672e-05, "loss": 3.1715, "step": 47710 }, { "epoch": 3.00193024, "grad_norm": 0.8172804117202759, "learning_rate": 1.92857393874464e-05, "loss": 3.1511, "step": 47720 }, { "epoch": 3.00195584, "grad_norm": 1.6804311275482178, "learning_rate": 1.9285440480274582e-05, "loss": 3.2771, "step": 47730 }, { "epoch": 3.00198144, "grad_norm": 0.7624708414077759, "learning_rate": 1.9285141512889147e-05, "loss": 3.2434, "step": 47740 }, { "epoch": 3.00200704, "grad_norm": 1.6067569255828857, "learning_rate": 1.928484248529204e-05, "loss": 3.1632, "step": 47750 }, { "epoch": 3.00203264, "grad_norm": 0.8683189153671265, "learning_rate": 1.9284543397485195e-05, "loss": 3.4102, "step": 47760 }, { "epoch": 3.00205824, "grad_norm": 1.195521354675293, "learning_rate": 1.9284244249470554e-05, "loss": 3.3473, "step": 47770 }, { "epoch": 3.00208384, "grad_norm": 0.872073233127594, "learning_rate": 1.9283945041250057e-05, "loss": 2.9977, "step": 47780 }, { "epoch": 3.00210944, "grad_norm": 0.727834165096283, "learning_rate": 1.9283645772825644e-05, "loss": 3.1155, "step": 47790 }, { "epoch": 3.00213504, "grad_norm": 0.9452624917030334, "learning_rate": 1.9283346444199254e-05, "loss": 3.3163, "step": 47800 }, { "epoch": 3.00216064, "grad_norm": 0.8312985897064209, "learning_rate": 1.9283047055372827e-05, "loss": 3.4842, "step": 47810 }, { "epoch": 3.00218624, "grad_norm": 0.8107131123542786, "learning_rate": 1.9282747606348315e-05, "loss": 3.0292, "step": 47820 }, { "epoch": 3.00221184, "grad_norm": 0.8025489449501038, "learning_rate": 1.9282448097127643e-05, "loss": 3.2866, "step": 47830 }, { "epoch": 3.00223744, "grad_norm": 0.7685579657554626, "learning_rate": 1.928214852771277e-05, "loss": 3.4337, "step": 47840 }, { "epoch": 3.00226304, "grad_norm": 0.7837840914726257, "learning_rate": 1.9281848898105622e-05, "loss": 3.3112, "step": 47850 }, { "epoch": 3.00228864, "grad_norm": 0.7280016541481018, "learning_rate": 1.928154920830816e-05, "loss": 3.2808, "step": 47860 }, { "epoch": 3.00231424, "grad_norm": 0.7777348756790161, "learning_rate": 1.9281249458322313e-05, "loss": 3.2215, "step": 47870 }, { "epoch": 3.00233984, "grad_norm": 0.8721612691879272, "learning_rate": 1.928094964815003e-05, "loss": 3.3431, "step": 47880 }, { "epoch": 3.00236544, "grad_norm": 0.7799971699714661, "learning_rate": 1.9280649777793256e-05, "loss": 3.4511, "step": 47890 }, { "epoch": 3.00239104, "grad_norm": 0.785844087600708, "learning_rate": 1.928034984725393e-05, "loss": 3.2206, "step": 47900 }, { "epoch": 3.00241664, "grad_norm": 0.8247954845428467, "learning_rate": 1.928004985653401e-05, "loss": 3.1919, "step": 47910 }, { "epoch": 3.00244224, "grad_norm": 2.4665415287017822, "learning_rate": 1.9279749805635427e-05, "loss": 3.1582, "step": 47920 }, { "epoch": 3.00246784, "grad_norm": 0.857519805431366, "learning_rate": 1.9279449694560137e-05, "loss": 3.0572, "step": 47930 }, { "epoch": 3.00249344, "grad_norm": 1.0224924087524414, "learning_rate": 1.9279149523310078e-05, "loss": 3.2704, "step": 47940 }, { "epoch": 3.00251904, "grad_norm": 0.8429641723632812, "learning_rate": 1.92788492918872e-05, "loss": 3.4095, "step": 47950 }, { "epoch": 3.00254464, "grad_norm": 0.8550659418106079, "learning_rate": 1.9278549000293455e-05, "loss": 3.4548, "step": 47960 }, { "epoch": 3.00257024, "grad_norm": 0.8658035397529602, "learning_rate": 1.9278248648530785e-05, "loss": 3.3079, "step": 47970 }, { "epoch": 3.00259584, "grad_norm": 0.9257025718688965, "learning_rate": 1.9277948236601136e-05, "loss": 3.1115, "step": 47980 }, { "epoch": 3.00262144, "grad_norm": 0.7859559655189514, "learning_rate": 1.927764776450646e-05, "loss": 3.0601, "step": 47990 }, { "epoch": 3.00264704, "grad_norm": 0.9934312701225281, "learning_rate": 1.9277347232248704e-05, "loss": 3.1672, "step": 48000 }, { "epoch": 3.00267264, "grad_norm": 1.0619676113128662, "learning_rate": 1.9277046639829814e-05, "loss": 3.2291, "step": 48010 }, { "epoch": 3.00269824, "grad_norm": 0.7217944860458374, "learning_rate": 1.9276745987251745e-05, "loss": 3.0785, "step": 48020 }, { "epoch": 3.00272384, "grad_norm": 0.8353272080421448, "learning_rate": 1.9276445274516443e-05, "loss": 3.3317, "step": 48030 }, { "epoch": 3.00274944, "grad_norm": 0.7339105010032654, "learning_rate": 1.9276144501625858e-05, "loss": 3.2781, "step": 48040 }, { "epoch": 3.00277504, "grad_norm": 0.7816954851150513, "learning_rate": 1.927584366858194e-05, "loss": 2.994, "step": 48050 }, { "epoch": 3.00280064, "grad_norm": 0.950210690498352, "learning_rate": 1.9275542775386645e-05, "loss": 3.206, "step": 48060 }, { "epoch": 3.00282624, "grad_norm": 0.8468984365463257, "learning_rate": 1.9275241822041915e-05, "loss": 3.1176, "step": 48070 }, { "epoch": 3.00285184, "grad_norm": 0.8308567404747009, "learning_rate": 1.9274940808549713e-05, "loss": 3.1871, "step": 48080 }, { "epoch": 3.00287744, "grad_norm": 0.9380219578742981, "learning_rate": 1.927463973491198e-05, "loss": 3.2569, "step": 48090 }, { "epoch": 3.00290304, "grad_norm": 0.7814929485321045, "learning_rate": 1.9274338601130673e-05, "loss": 3.316, "step": 48100 }, { "epoch": 3.00292864, "grad_norm": 0.8547413945198059, "learning_rate": 1.927403740720775e-05, "loss": 3.3847, "step": 48110 }, { "epoch": 3.00295424, "grad_norm": 1.0395383834838867, "learning_rate": 1.9273736153145153e-05, "loss": 3.3191, "step": 48120 }, { "epoch": 3.00297984, "grad_norm": 0.9232662916183472, "learning_rate": 1.9273434838944847e-05, "loss": 2.9803, "step": 48130 }, { "epoch": 3.00300544, "grad_norm": 0.8590767979621887, "learning_rate": 1.9273133464608777e-05, "loss": 3.3047, "step": 48140 }, { "epoch": 3.00303104, "grad_norm": 1.1211609840393066, "learning_rate": 1.9272832030138903e-05, "loss": 3.3595, "step": 48150 }, { "epoch": 3.00305664, "grad_norm": 0.7844284772872925, "learning_rate": 1.9272530535537175e-05, "loss": 3.2326, "step": 48160 }, { "epoch": 3.00308224, "grad_norm": 0.9550862312316895, "learning_rate": 1.9272228980805553e-05, "loss": 3.2756, "step": 48170 }, { "epoch": 3.00310784, "grad_norm": 0.7752067446708679, "learning_rate": 1.927192736594599e-05, "loss": 3.2083, "step": 48180 }, { "epoch": 3.00313344, "grad_norm": 0.8387207984924316, "learning_rate": 1.9271625690960444e-05, "loss": 3.3672, "step": 48190 }, { "epoch": 3.00315904, "grad_norm": 0.8121353387832642, "learning_rate": 1.927132395585087e-05, "loss": 3.3028, "step": 48200 }, { "epoch": 3.00318464, "grad_norm": 0.971295177936554, "learning_rate": 1.9271022160619218e-05, "loss": 3.4937, "step": 48210 }, { "epoch": 3.00321024, "grad_norm": 0.8835462331771851, "learning_rate": 1.9270720305267455e-05, "loss": 3.484, "step": 48220 }, { "epoch": 3.00323584, "grad_norm": 0.9001634120941162, "learning_rate": 1.927041838979754e-05, "loss": 3.3432, "step": 48230 }, { "epoch": 3.00326144, "grad_norm": 0.8683317303657532, "learning_rate": 1.927011641421142e-05, "loss": 3.2214, "step": 48240 }, { "epoch": 3.00328704, "grad_norm": 0.8301811814308167, "learning_rate": 1.9269814378511056e-05, "loss": 3.3354, "step": 48250 }, { "epoch": 3.00331264, "grad_norm": 0.83517986536026, "learning_rate": 1.9269512282698414e-05, "loss": 3.3272, "step": 48260 }, { "epoch": 3.00333824, "grad_norm": 0.8143470883369446, "learning_rate": 1.9269210126775447e-05, "loss": 3.3125, "step": 48270 }, { "epoch": 3.00336384, "grad_norm": 0.892159104347229, "learning_rate": 1.9268907910744116e-05, "loss": 3.1256, "step": 48280 }, { "epoch": 3.00338944, "grad_norm": 0.8304699659347534, "learning_rate": 1.926860563460638e-05, "loss": 3.2608, "step": 48290 }, { "epoch": 3.00341504, "grad_norm": 0.9010924100875854, "learning_rate": 1.92683032983642e-05, "loss": 3.5777, "step": 48300 }, { "epoch": 3.00344064, "grad_norm": 0.7990552186965942, "learning_rate": 1.9268000902019537e-05, "loss": 3.4291, "step": 48310 }, { "epoch": 3.00346624, "grad_norm": 0.8198301792144775, "learning_rate": 1.926769844557435e-05, "loss": 3.3993, "step": 48320 }, { "epoch": 3.00349184, "grad_norm": 0.8408704996109009, "learning_rate": 1.9267395929030603e-05, "loss": 3.3073, "step": 48330 }, { "epoch": 3.00351744, "grad_norm": 0.866562008857727, "learning_rate": 1.9267093352390255e-05, "loss": 3.226, "step": 48340 }, { "epoch": 3.00354304, "grad_norm": 0.8100970983505249, "learning_rate": 1.926679071565527e-05, "loss": 3.2647, "step": 48350 }, { "epoch": 3.00356864, "grad_norm": 0.8180371522903442, "learning_rate": 1.926648801882761e-05, "loss": 3.3437, "step": 48360 }, { "epoch": 3.00359424, "grad_norm": 0.8206444978713989, "learning_rate": 1.926618526190924e-05, "loss": 3.3638, "step": 48370 }, { "epoch": 3.00361984, "grad_norm": 0.9729020595550537, "learning_rate": 1.926588244490212e-05, "loss": 3.3369, "step": 48380 }, { "epoch": 3.00364544, "grad_norm": 0.909837543964386, "learning_rate": 1.926557956780821e-05, "loss": 3.1677, "step": 48390 }, { "epoch": 3.00367104, "grad_norm": 0.8776983022689819, "learning_rate": 1.9265276630629488e-05, "loss": 3.2874, "step": 48400 }, { "epoch": 3.00369664, "grad_norm": 0.7802743315696716, "learning_rate": 1.9264973633367906e-05, "loss": 3.2583, "step": 48410 }, { "epoch": 3.00372224, "grad_norm": 1.0963715314865112, "learning_rate": 1.926467057602543e-05, "loss": 3.3562, "step": 48420 }, { "epoch": 3.00374784, "grad_norm": 0.7774801254272461, "learning_rate": 1.926436745860403e-05, "loss": 3.395, "step": 48430 }, { "epoch": 3.00377344, "grad_norm": 0.9428799152374268, "learning_rate": 1.9264064281105667e-05, "loss": 3.2286, "step": 48440 }, { "epoch": 3.00379904, "grad_norm": 0.7720880508422852, "learning_rate": 1.926376104353231e-05, "loss": 3.3631, "step": 48450 }, { "epoch": 3.00382464, "grad_norm": 0.79073166847229, "learning_rate": 1.9263457745885926e-05, "loss": 3.0666, "step": 48460 }, { "epoch": 3.00385024, "grad_norm": 0.85267174243927, "learning_rate": 1.9263154388168477e-05, "loss": 3.0649, "step": 48470 }, { "epoch": 3.00387584, "grad_norm": 0.9262882471084595, "learning_rate": 1.9262850970381935e-05, "loss": 3.1692, "step": 48480 }, { "epoch": 3.00390144, "grad_norm": 0.8184542655944824, "learning_rate": 1.926254749252827e-05, "loss": 3.2568, "step": 48490 }, { "epoch": 3.00392704, "grad_norm": 0.8465251922607422, "learning_rate": 1.926224395460944e-05, "loss": 3.1962, "step": 48500 }, { "epoch": 3.00395264, "grad_norm": 1.4065887928009033, "learning_rate": 1.9261940356627426e-05, "loss": 3.3565, "step": 48510 }, { "epoch": 3.00397824, "grad_norm": 0.7874218821525574, "learning_rate": 1.9261636698584186e-05, "loss": 3.232, "step": 48520 }, { "epoch": 3.00400384, "grad_norm": 1.032374382019043, "learning_rate": 1.9261332980481696e-05, "loss": 3.1447, "step": 48530 }, { "epoch": 3.00402944, "grad_norm": 0.726047158241272, "learning_rate": 1.926102920232192e-05, "loss": 3.1842, "step": 48540 }, { "epoch": 3.00405504, "grad_norm": 0.8891833424568176, "learning_rate": 1.9260725364106835e-05, "loss": 3.1462, "step": 48550 }, { "epoch": 3.00408064, "grad_norm": 0.8194966912269592, "learning_rate": 1.9260421465838404e-05, "loss": 3.2349, "step": 48560 }, { "epoch": 3.00410624, "grad_norm": 0.7573374509811401, "learning_rate": 1.92601175075186e-05, "loss": 3.2566, "step": 48570 }, { "epoch": 3.00413184, "grad_norm": 0.7624163627624512, "learning_rate": 1.9259813489149398e-05, "loss": 3.1475, "step": 48580 }, { "epoch": 3.00415744, "grad_norm": 2.1995978355407715, "learning_rate": 1.9259509410732767e-05, "loss": 3.3266, "step": 48590 }, { "epoch": 3.00418304, "grad_norm": 0.8197693824768066, "learning_rate": 1.9259205272270674e-05, "loss": 3.4241, "step": 48600 }, { "epoch": 3.00420864, "grad_norm": 0.8932395577430725, "learning_rate": 1.92589010737651e-05, "loss": 3.3125, "step": 48610 }, { "epoch": 3.00423424, "grad_norm": 0.762293815612793, "learning_rate": 1.9258596815218008e-05, "loss": 3.167, "step": 48620 }, { "epoch": 3.00425984, "grad_norm": 0.7689289450645447, "learning_rate": 1.925829249663138e-05, "loss": 3.2191, "step": 48630 }, { "epoch": 3.00428544, "grad_norm": 0.8281218409538269, "learning_rate": 1.925798811800719e-05, "loss": 3.1679, "step": 48640 }, { "epoch": 3.00431104, "grad_norm": 0.8317775726318359, "learning_rate": 1.92576836793474e-05, "loss": 3.2565, "step": 48650 }, { "epoch": 3.00433664, "grad_norm": 0.8436621427536011, "learning_rate": 1.9257379180653997e-05, "loss": 3.2005, "step": 48660 }, { "epoch": 3.00436224, "grad_norm": 0.8322432637214661, "learning_rate": 1.925707462192895e-05, "loss": 3.3618, "step": 48670 }, { "epoch": 3.00438784, "grad_norm": 0.8670104742050171, "learning_rate": 1.925677000317423e-05, "loss": 3.077, "step": 48680 }, { "epoch": 3.00441344, "grad_norm": 0.8334774374961853, "learning_rate": 1.9256465324391818e-05, "loss": 3.2551, "step": 48690 }, { "epoch": 3.00443904, "grad_norm": 0.7717075347900391, "learning_rate": 1.925616058558369e-05, "loss": 3.2271, "step": 48700 }, { "epoch": 3.00446464, "grad_norm": 0.7647733688354492, "learning_rate": 1.925585578675182e-05, "loss": 3.371, "step": 48710 }, { "epoch": 3.00449024, "grad_norm": 0.7595325708389282, "learning_rate": 1.9255550927898186e-05, "loss": 3.1259, "step": 48720 }, { "epoch": 3.00451584, "grad_norm": 0.8278961777687073, "learning_rate": 1.925524600902476e-05, "loss": 3.2859, "step": 48730 }, { "epoch": 3.00454144, "grad_norm": 1.6033520698547363, "learning_rate": 1.925494103013353e-05, "loss": 3.2507, "step": 48740 }, { "epoch": 3.00456704, "grad_norm": 0.9810839295387268, "learning_rate": 1.9254635991226465e-05, "loss": 3.2882, "step": 48750 }, { "epoch": 3.00459264, "grad_norm": 0.8063149452209473, "learning_rate": 1.9254330892305542e-05, "loss": 3.1551, "step": 48760 }, { "epoch": 3.00461824, "grad_norm": 0.8418417572975159, "learning_rate": 1.9254025733372744e-05, "loss": 3.2923, "step": 48770 }, { "epoch": 3.00464384, "grad_norm": 0.7860402464866638, "learning_rate": 1.925372051443005e-05, "loss": 3.178, "step": 48780 }, { "epoch": 3.00466944, "grad_norm": 0.8203387260437012, "learning_rate": 1.9253415235479437e-05, "loss": 3.0558, "step": 48790 }, { "epoch": 3.00469504, "grad_norm": 0.9737443327903748, "learning_rate": 1.9253109896522885e-05, "loss": 2.9749, "step": 48800 }, { "epoch": 3.00472064, "grad_norm": 0.9628830552101135, "learning_rate": 1.9252804497562377e-05, "loss": 3.2614, "step": 48810 }, { "epoch": 3.00474624, "grad_norm": 0.8412245512008667, "learning_rate": 1.9252499038599887e-05, "loss": 3.204, "step": 48820 }, { "epoch": 3.00477184, "grad_norm": 0.8053393363952637, "learning_rate": 1.9252193519637402e-05, "loss": 2.8718, "step": 48830 }, { "epoch": 3.00479744, "grad_norm": 0.8150424361228943, "learning_rate": 1.92518879406769e-05, "loss": 3.1629, "step": 48840 }, { "epoch": 3.00482304, "grad_norm": 0.759723424911499, "learning_rate": 1.9251582301720364e-05, "loss": 3.3944, "step": 48850 }, { "epoch": 3.00484864, "grad_norm": 0.7775519490242004, "learning_rate": 1.9251276602769778e-05, "loss": 3.1119, "step": 48860 }, { "epoch": 3.00487424, "grad_norm": 0.9921696186065674, "learning_rate": 1.9250970843827118e-05, "loss": 3.1573, "step": 48870 }, { "epoch": 3.00489984, "grad_norm": 0.8442431092262268, "learning_rate": 1.9250665024894372e-05, "loss": 3.2803, "step": 48880 }, { "epoch": 3.00492544, "grad_norm": 1.036828637123108, "learning_rate": 1.9250359145973524e-05, "loss": 3.2232, "step": 48890 }, { "epoch": 3.00495104, "grad_norm": 0.8073385953903198, "learning_rate": 1.9250053207066554e-05, "loss": 3.3856, "step": 48900 }, { "epoch": 3.00497664, "grad_norm": 0.8636882901191711, "learning_rate": 1.924974720817545e-05, "loss": 3.3378, "step": 48910 }, { "epoch": 3.00500224, "grad_norm": 0.8294476270675659, "learning_rate": 1.9249441149302186e-05, "loss": 4.0489, "step": 48920 }, { "epoch": 3.00502784, "grad_norm": 0.8968760967254639, "learning_rate": 1.924913503044876e-05, "loss": 3.062, "step": 48930 }, { "epoch": 3.00505344, "grad_norm": 0.879733681678772, "learning_rate": 1.9248828851617148e-05, "loss": 3.3613, "step": 48940 }, { "epoch": 3.00507904, "grad_norm": 0.950592577457428, "learning_rate": 1.9248522612809344e-05, "loss": 3.3914, "step": 48950 }, { "epoch": 3.00510464, "grad_norm": 0.7701599597930908, "learning_rate": 1.9248216314027325e-05, "loss": 3.1704, "step": 48960 }, { "epoch": 3.00513024, "grad_norm": 0.8802741765975952, "learning_rate": 1.9247909955273082e-05, "loss": 3.3098, "step": 48970 }, { "epoch": 3.00515584, "grad_norm": 1.0334745645523071, "learning_rate": 1.9247603536548602e-05, "loss": 3.1798, "step": 48980 }, { "epoch": 3.00518144, "grad_norm": 0.9397733807563782, "learning_rate": 1.9247297057855867e-05, "loss": 3.0171, "step": 48990 }, { "epoch": 3.00520704, "grad_norm": 0.8106107115745544, "learning_rate": 1.9246990519196875e-05, "loss": 3.1718, "step": 49000 }, { "epoch": 3.00523264, "grad_norm": 0.8879449963569641, "learning_rate": 1.92466839205736e-05, "loss": 3.1888, "step": 49010 }, { "epoch": 3.00525824, "grad_norm": 0.8595956563949585, "learning_rate": 1.924637726198804e-05, "loss": 3.204, "step": 49020 }, { "epoch": 3.00528384, "grad_norm": 1.049877405166626, "learning_rate": 1.924607054344218e-05, "loss": 3.2923, "step": 49030 }, { "epoch": 3.00530944, "grad_norm": 0.9947653412818909, "learning_rate": 1.9245763764938013e-05, "loss": 3.1321, "step": 49040 }, { "epoch": 3.00533504, "grad_norm": 0.9306682348251343, "learning_rate": 1.9245456926477523e-05, "loss": 3.3087, "step": 49050 }, { "epoch": 3.00536064, "grad_norm": 0.916431188583374, "learning_rate": 1.9245150028062703e-05, "loss": 3.3089, "step": 49060 }, { "epoch": 3.00538624, "grad_norm": 0.7578913569450378, "learning_rate": 1.924484306969554e-05, "loss": 3.1146, "step": 49070 }, { "epoch": 3.00541184, "grad_norm": 1.3073902130126953, "learning_rate": 1.9244536051378027e-05, "loss": 3.3746, "step": 49080 }, { "epoch": 3.00543744, "grad_norm": 0.9163573384284973, "learning_rate": 1.9244228973112156e-05, "loss": 3.2729, "step": 49090 }, { "epoch": 3.00546304, "grad_norm": 0.9073221683502197, "learning_rate": 1.9243921834899916e-05, "loss": 3.4636, "step": 49100 }, { "epoch": 3.00548864, "grad_norm": 0.8648418188095093, "learning_rate": 1.92436146367433e-05, "loss": 3.5554, "step": 49110 }, { "epoch": 3.00551424, "grad_norm": 0.8451858758926392, "learning_rate": 1.9243307378644302e-05, "loss": 3.3136, "step": 49120 }, { "epoch": 3.00553984, "grad_norm": 0.8680051565170288, "learning_rate": 1.924300006060491e-05, "loss": 3.1864, "step": 49130 }, { "epoch": 3.00556544, "grad_norm": 0.957120418548584, "learning_rate": 1.924269268262712e-05, "loss": 3.3572, "step": 49140 }, { "epoch": 3.00559104, "grad_norm": 0.9151686429977417, "learning_rate": 1.924238524471292e-05, "loss": 3.355, "step": 49150 }, { "epoch": 3.00561664, "grad_norm": 0.8120043873786926, "learning_rate": 1.9242077746864315e-05, "loss": 3.2742, "step": 49160 }, { "epoch": 3.00564224, "grad_norm": 0.8027195930480957, "learning_rate": 1.9241770189083287e-05, "loss": 3.2267, "step": 49170 }, { "epoch": 3.00566784, "grad_norm": 1.4851405620574951, "learning_rate": 1.9241462571371836e-05, "loss": 3.2052, "step": 49180 }, { "epoch": 3.00569344, "grad_norm": 0.8559970855712891, "learning_rate": 1.924115489373196e-05, "loss": 3.167, "step": 49190 }, { "epoch": 3.00571904, "grad_norm": 1.680759072303772, "learning_rate": 1.9240847156165644e-05, "loss": 3.2176, "step": 49200 }, { "epoch": 3.00574464, "grad_norm": 0.8476759195327759, "learning_rate": 1.9240539358674895e-05, "loss": 3.1797, "step": 49210 }, { "epoch": 3.00577024, "grad_norm": 0.7732351422309875, "learning_rate": 1.9240231501261705e-05, "loss": 3.273, "step": 49220 }, { "epoch": 3.00579584, "grad_norm": 0.8231801986694336, "learning_rate": 1.923992358392807e-05, "loss": 3.2314, "step": 49230 }, { "epoch": 3.00582144, "grad_norm": 0.9668828248977661, "learning_rate": 1.923961560667598e-05, "loss": 3.2568, "step": 49240 }, { "epoch": 3.00584704, "grad_norm": 0.898717999458313, "learning_rate": 1.9239307569507443e-05, "loss": 3.434, "step": 49250 }, { "epoch": 3.00587264, "grad_norm": 0.7834280729293823, "learning_rate": 1.923899947242445e-05, "loss": 3.1409, "step": 49260 }, { "epoch": 3.00589824, "grad_norm": 0.832675576210022, "learning_rate": 1.9238691315429003e-05, "loss": 3.1118, "step": 49270 }, { "epoch": 3.00592384, "grad_norm": 0.8992850184440613, "learning_rate": 1.9238383098523097e-05, "loss": 3.3691, "step": 49280 }, { "epoch": 3.00594944, "grad_norm": 0.8209416270256042, "learning_rate": 1.9238074821708732e-05, "loss": 3.2533, "step": 49290 }, { "epoch": 3.00597504, "grad_norm": 0.8723741173744202, "learning_rate": 1.923776648498791e-05, "loss": 3.2369, "step": 49300 }, { "epoch": 3.00600064, "grad_norm": 0.7939255833625793, "learning_rate": 1.9237458088362625e-05, "loss": 3.2023, "step": 49310 }, { "epoch": 3.00602624, "grad_norm": 0.9012367725372314, "learning_rate": 1.9237149631834876e-05, "loss": 3.2346, "step": 49320 }, { "epoch": 3.00605184, "grad_norm": 1.6953270435333252, "learning_rate": 1.923684111540667e-05, "loss": 3.3345, "step": 49330 }, { "epoch": 3.00607744, "grad_norm": 0.9147545695304871, "learning_rate": 1.9236532539080003e-05, "loss": 3.1784, "step": 49340 }, { "epoch": 3.00610304, "grad_norm": 0.8467957377433777, "learning_rate": 1.9236223902856877e-05, "loss": 3.3936, "step": 49350 }, { "epoch": 3.00612864, "grad_norm": 0.884475588798523, "learning_rate": 1.9235915206739297e-05, "loss": 3.3725, "step": 49360 }, { "epoch": 3.00615424, "grad_norm": 0.8631933927536011, "learning_rate": 1.923560645072926e-05, "loss": 3.2943, "step": 49370 }, { "epoch": 3.00617984, "grad_norm": 0.860070526599884, "learning_rate": 1.923529763482877e-05, "loss": 3.3273, "step": 49380 }, { "epoch": 3.00620544, "grad_norm": 0.789783775806427, "learning_rate": 1.9234988759039822e-05, "loss": 3.3329, "step": 49390 }, { "epoch": 3.00623104, "grad_norm": 0.9097357392311096, "learning_rate": 1.9234679823364434e-05, "loss": 3.5214, "step": 49400 }, { "epoch": 3.00625664, "grad_norm": 0.8758026361465454, "learning_rate": 1.9234370827804603e-05, "loss": 3.2039, "step": 49410 }, { "epoch": 3.00628224, "grad_norm": 0.8431061506271362, "learning_rate": 1.9234061772362327e-05, "loss": 3.1872, "step": 49420 }, { "epoch": 3.00630784, "grad_norm": 0.8265750408172607, "learning_rate": 1.9233752657039616e-05, "loss": 3.3952, "step": 49430 }, { "epoch": 3.00633344, "grad_norm": 1.0834892988204956, "learning_rate": 1.9233443481838472e-05, "loss": 3.3424, "step": 49440 }, { "epoch": 3.00635904, "grad_norm": 0.8977164030075073, "learning_rate": 1.92331342467609e-05, "loss": 3.4902, "step": 49450 }, { "epoch": 3.00638464, "grad_norm": 0.8461244702339172, "learning_rate": 1.923282495180891e-05, "loss": 3.2286, "step": 49460 }, { "epoch": 3.00641024, "grad_norm": 0.833489179611206, "learning_rate": 1.92325155969845e-05, "loss": 3.1812, "step": 49470 }, { "epoch": 3.00643584, "grad_norm": 0.8214621543884277, "learning_rate": 1.9232206182289684e-05, "loss": 3.2246, "step": 49480 }, { "epoch": 3.00646144, "grad_norm": 0.7799249291419983, "learning_rate": 1.9231896707726463e-05, "loss": 3.3278, "step": 49490 }, { "epoch": 3.00648704, "grad_norm": 1.734113097190857, "learning_rate": 1.9231587173296848e-05, "loss": 3.2205, "step": 49500 }, { "epoch": 3.00651264, "grad_norm": 0.815165102481842, "learning_rate": 1.923127757900284e-05, "loss": 3.1895, "step": 49510 }, { "epoch": 3.00653824, "grad_norm": 0.820873498916626, "learning_rate": 1.9230967924846448e-05, "loss": 3.356, "step": 49520 }, { "epoch": 3.00656384, "grad_norm": 0.7985717058181763, "learning_rate": 1.9230658210829686e-05, "loss": 3.2653, "step": 49530 }, { "epoch": 3.00658944, "grad_norm": 0.9047940969467163, "learning_rate": 1.923034843695456e-05, "loss": 3.393, "step": 49540 }, { "epoch": 3.00661504, "grad_norm": 0.9568749070167542, "learning_rate": 1.9230038603223073e-05, "loss": 3.4605, "step": 49550 }, { "epoch": 3.00664064, "grad_norm": 0.8666346073150635, "learning_rate": 1.922972870963724e-05, "loss": 3.1608, "step": 49560 }, { "epoch": 3.00666624, "grad_norm": 0.8921893835067749, "learning_rate": 1.922941875619907e-05, "loss": 3.1798, "step": 49570 }, { "epoch": 3.00669184, "grad_norm": 0.8124176263809204, "learning_rate": 1.9229108742910576e-05, "loss": 3.2605, "step": 49580 }, { "epoch": 3.00671744, "grad_norm": 1.0627657175064087, "learning_rate": 1.9228798669773762e-05, "loss": 3.0793, "step": 49590 }, { "epoch": 3.00674304, "grad_norm": 0.7714450359344482, "learning_rate": 1.922848853679064e-05, "loss": 3.3323, "step": 49600 }, { "epoch": 3.00676864, "grad_norm": 0.8971641659736633, "learning_rate": 1.922817834396322e-05, "loss": 3.3198, "step": 49610 }, { "epoch": 3.00679424, "grad_norm": 0.9023619294166565, "learning_rate": 1.9227868091293525e-05, "loss": 3.329, "step": 49620 }, { "epoch": 3.00681984, "grad_norm": 0.9269682168960571, "learning_rate": 1.922755777878355e-05, "loss": 3.3171, "step": 49630 }, { "epoch": 3.00684544, "grad_norm": 1.107803225517273, "learning_rate": 1.9227247406435315e-05, "loss": 3.1628, "step": 49640 }, { "epoch": 3.00687104, "grad_norm": 1.0294688940048218, "learning_rate": 1.9226936974250837e-05, "loss": 3.1825, "step": 49650 }, { "epoch": 3.00689664, "grad_norm": 0.9509176015853882, "learning_rate": 1.9226626482232124e-05, "loss": 3.2987, "step": 49660 }, { "epoch": 3.00692224, "grad_norm": 0.8989757895469666, "learning_rate": 1.922631593038119e-05, "loss": 3.0679, "step": 49670 }, { "epoch": 3.00694784, "grad_norm": 0.8751895427703857, "learning_rate": 1.922600531870005e-05, "loss": 3.1609, "step": 49680 }, { "epoch": 3.00697344, "grad_norm": 0.8610190153121948, "learning_rate": 1.9225694647190717e-05, "loss": 3.2736, "step": 49690 }, { "epoch": 3.00699904, "grad_norm": 0.9496526718139648, "learning_rate": 1.9225383915855205e-05, "loss": 3.3478, "step": 49700 }, { "epoch": 3.00702464, "grad_norm": 0.804206907749176, "learning_rate": 1.922507312469553e-05, "loss": 3.1843, "step": 49710 }, { "epoch": 3.00705024, "grad_norm": 0.805916428565979, "learning_rate": 1.9224762273713705e-05, "loss": 3.0734, "step": 49720 }, { "epoch": 3.00707584, "grad_norm": 0.8896265029907227, "learning_rate": 1.9224451362911752e-05, "loss": 3.1405, "step": 49730 }, { "epoch": 3.00710144, "grad_norm": 1.1071974039077759, "learning_rate": 1.9224140392291682e-05, "loss": 3.2682, "step": 49740 }, { "epoch": 3.00712704, "grad_norm": 0.8788272142410278, "learning_rate": 1.9223829361855517e-05, "loss": 3.3546, "step": 49750 }, { "epoch": 3.00715264, "grad_norm": 0.7520207166671753, "learning_rate": 1.9223518271605267e-05, "loss": 3.3025, "step": 49760 }, { "epoch": 3.00717824, "grad_norm": 0.8398561477661133, "learning_rate": 1.922320712154295e-05, "loss": 3.204, "step": 49770 }, { "epoch": 3.00720384, "grad_norm": 0.939149022102356, "learning_rate": 1.922289591167059e-05, "loss": 3.163, "step": 49780 }, { "epoch": 3.00722944, "grad_norm": 0.8895670175552368, "learning_rate": 1.92225846419902e-05, "loss": 3.4515, "step": 49790 }, { "epoch": 3.00725504, "grad_norm": 0.8869264125823975, "learning_rate": 1.9222273312503797e-05, "loss": 3.2104, "step": 49800 }, { "epoch": 3.00728064, "grad_norm": 0.8488876223564148, "learning_rate": 1.9221961923213407e-05, "loss": 3.2118, "step": 49810 }, { "epoch": 3.00730624, "grad_norm": 0.8529369831085205, "learning_rate": 1.922165047412104e-05, "loss": 3.1052, "step": 49820 }, { "epoch": 3.00733184, "grad_norm": 0.8567706346511841, "learning_rate": 1.9221338965228723e-05, "loss": 3.146, "step": 49830 }, { "epoch": 3.00735744, "grad_norm": 0.9024309515953064, "learning_rate": 1.9221027396538473e-05, "loss": 3.0863, "step": 49840 }, { "epoch": 3.00738304, "grad_norm": 0.8228935599327087, "learning_rate": 1.922071576805231e-05, "loss": 3.1226, "step": 49850 }, { "epoch": 3.00740864, "grad_norm": 0.9254204034805298, "learning_rate": 1.922040407977226e-05, "loss": 3.4214, "step": 49860 }, { "epoch": 3.00743424, "grad_norm": 0.8306892514228821, "learning_rate": 1.922009233170034e-05, "loss": 3.1995, "step": 49870 }, { "epoch": 3.00745984, "grad_norm": 2.92641282081604, "learning_rate": 1.921978052383857e-05, "loss": 3.2344, "step": 49880 }, { "epoch": 3.00748544, "grad_norm": 0.8166998028755188, "learning_rate": 1.9219468656188967e-05, "loss": 3.274, "step": 49890 }, { "epoch": 3.00751104, "grad_norm": 0.8709633350372314, "learning_rate": 1.9219156728753568e-05, "loss": 3.2639, "step": 49900 }, { "epoch": 2.56e-05, "grad_norm": 1.0473015308380127, "learning_rate": 1.9218844741534384e-05, "loss": 3.388, "step": 49910 }, { "epoch": 5.12e-05, "grad_norm": 0.821375846862793, "learning_rate": 1.9218532694533444e-05, "loss": 3.2843, "step": 49920 }, { "epoch": 7.68e-05, "grad_norm": 0.8334003686904907, "learning_rate": 1.921822058775277e-05, "loss": 3.2326, "step": 49930 }, { "epoch": 0.0001024, "grad_norm": 1.0485197305679321, "learning_rate": 1.9217908421194383e-05, "loss": 3.2328, "step": 49940 }, { "epoch": 0.000128, "grad_norm": 0.8448278307914734, "learning_rate": 1.9217596194860312e-05, "loss": 3.3863, "step": 49950 }, { "epoch": 0.0001536, "grad_norm": 1.0759507417678833, "learning_rate": 1.921728390875258e-05, "loss": 3.244, "step": 49960 }, { "epoch": 0.0001792, "grad_norm": 0.8067512512207031, "learning_rate": 1.9216971562873207e-05, "loss": 3.3298, "step": 49970 }, { "epoch": 0.0002048, "grad_norm": 1.0333224534988403, "learning_rate": 1.9216659157224226e-05, "loss": 3.4416, "step": 49980 }, { "epoch": 0.0002304, "grad_norm": 0.9939389824867249, "learning_rate": 1.921634669180766e-05, "loss": 3.3286, "step": 49990 }, { "epoch": 0.000256, "grad_norm": 1.4069640636444092, "learning_rate": 1.9216034166625533e-05, "loss": 3.3052, "step": 50000 }, { "epoch": 0.0002816, "grad_norm": 0.9489545226097107, "learning_rate": 1.9215721581679877e-05, "loss": 3.5908, "step": 50010 }, { "epoch": 0.0003072, "grad_norm": 0.8339141011238098, "learning_rate": 1.9215408936972716e-05, "loss": 3.4703, "step": 50020 }, { "epoch": 0.0003328, "grad_norm": 0.9692524671554565, "learning_rate": 1.9215096232506072e-05, "loss": 3.3431, "step": 50030 }, { "epoch": 0.0003584, "grad_norm": 0.9102902412414551, "learning_rate": 1.9214783468281983e-05, "loss": 3.9787, "step": 50040 }, { "epoch": 0.000384, "grad_norm": 0.8984925746917725, "learning_rate": 1.921447064430247e-05, "loss": 3.3152, "step": 50050 }, { "epoch": 0.0004096, "grad_norm": 0.7049500346183777, "learning_rate": 1.9214157760569568e-05, "loss": 3.2751, "step": 50060 }, { "epoch": 0.0004352, "grad_norm": 1.1793454885482788, "learning_rate": 1.9213844817085297e-05, "loss": 3.2738, "step": 50070 }, { "epoch": 0.0004608, "grad_norm": 0.9149006605148315, "learning_rate": 1.921353181385169e-05, "loss": 3.3787, "step": 50080 }, { "epoch": 0.0004864, "grad_norm": 0.8349671363830566, "learning_rate": 1.9213218750870784e-05, "loss": 3.1914, "step": 50090 }, { "epoch": 0.000512, "grad_norm": 1.175945520401001, "learning_rate": 1.9212905628144598e-05, "loss": 3.4819, "step": 50100 }, { "epoch": 0.0005376, "grad_norm": 1.020561933517456, "learning_rate": 1.921259244567517e-05, "loss": 3.6777, "step": 50110 }, { "epoch": 0.0005632, "grad_norm": 0.8415781855583191, "learning_rate": 1.9212279203464525e-05, "loss": 3.3519, "step": 50120 }, { "epoch": 0.0005888, "grad_norm": 2.1361162662506104, "learning_rate": 1.92119659015147e-05, "loss": 3.076, "step": 50130 }, { "epoch": 0.0006144, "grad_norm": 0.8190310597419739, "learning_rate": 1.9211652539827724e-05, "loss": 3.2875, "step": 50140 }, { "epoch": 0.00064, "grad_norm": 1.1587870121002197, "learning_rate": 1.921133911840563e-05, "loss": 3.2163, "step": 50150 }, { "epoch": 0.0006656, "grad_norm": 0.8150973320007324, "learning_rate": 1.921102563725045e-05, "loss": 3.351, "step": 50160 }, { "epoch": 0.0006912, "grad_norm": 0.8328425884246826, "learning_rate": 1.9210712096364216e-05, "loss": 3.1602, "step": 50170 }, { "epoch": 0.0007168, "grad_norm": 1.0065261125564575, "learning_rate": 1.9210398495748963e-05, "loss": 3.2335, "step": 50180 }, { "epoch": 0.0007424, "grad_norm": 0.8038985133171082, "learning_rate": 1.9210084835406723e-05, "loss": 3.4701, "step": 50190 }, { "epoch": 0.000768, "grad_norm": 0.7008118033409119, "learning_rate": 1.9209771115339532e-05, "loss": 3.2626, "step": 50200 }, { "epoch": 0.0007936, "grad_norm": 0.9528414607048035, "learning_rate": 1.920945733554942e-05, "loss": 3.204, "step": 50210 }, { "epoch": 0.0008192, "grad_norm": 1.0372893810272217, "learning_rate": 1.9209143496038425e-05, "loss": 3.3857, "step": 50220 }, { "epoch": 0.0008448, "grad_norm": 0.731360137462616, "learning_rate": 1.9208829596808584e-05, "loss": 3.2859, "step": 50230 }, { "epoch": 0.0008704, "grad_norm": 0.856288492679596, "learning_rate": 1.9208515637861927e-05, "loss": 3.5487, "step": 50240 }, { "epoch": 0.000896, "grad_norm": 0.7921684384346008, "learning_rate": 1.9208201619200498e-05, "loss": 3.3387, "step": 50250 }, { "epoch": 0.0009216, "grad_norm": 0.7689756155014038, "learning_rate": 1.9207887540826326e-05, "loss": 2.9857, "step": 50260 }, { "epoch": 0.0009472, "grad_norm": 1.3350849151611328, "learning_rate": 1.920757340274145e-05, "loss": 3.5821, "step": 50270 }, { "epoch": 0.0009728, "grad_norm": 0.8343431353569031, "learning_rate": 1.9207259204947912e-05, "loss": 3.2139, "step": 50280 }, { "epoch": 0.0009984, "grad_norm": 0.7965306043624878, "learning_rate": 1.9206944947447742e-05, "loss": 3.4155, "step": 50290 }, { "epoch": 0.001024, "grad_norm": 0.7844344973564148, "learning_rate": 1.920663063024298e-05, "loss": 3.3935, "step": 50300 }, { "epoch": 0.0010496, "grad_norm": 0.783789873123169, "learning_rate": 1.9206316253335665e-05, "loss": 3.3852, "step": 50310 }, { "epoch": 0.0010752, "grad_norm": 1.8613735437393188, "learning_rate": 1.9206001816727838e-05, "loss": 3.1834, "step": 50320 }, { "epoch": 0.0011008, "grad_norm": 0.9454888701438904, "learning_rate": 1.9205687320421535e-05, "loss": 3.6251, "step": 50330 }, { "epoch": 0.0011264, "grad_norm": 1.5669504404067993, "learning_rate": 1.9205372764418796e-05, "loss": 3.4545, "step": 50340 }, { "epoch": 0.001152, "grad_norm": 0.8011676669120789, "learning_rate": 1.920505814872166e-05, "loss": 3.3206, "step": 50350 }, { "epoch": 0.0011776, "grad_norm": 0.8102993965148926, "learning_rate": 1.920474347333217e-05, "loss": 3.2186, "step": 50360 }, { "epoch": 0.0012032, "grad_norm": 0.9118268489837646, "learning_rate": 1.9204428738252363e-05, "loss": 3.1791, "step": 50370 }, { "epoch": 0.0012288, "grad_norm": 0.8180035948753357, "learning_rate": 1.9204113943484284e-05, "loss": 3.2132, "step": 50380 }, { "epoch": 0.0012544, "grad_norm": 0.9537824392318726, "learning_rate": 1.920379908902997e-05, "loss": 3.2324, "step": 50390 }, { "epoch": 0.00128, "grad_norm": 0.8662207126617432, "learning_rate": 1.9203484174891466e-05, "loss": 3.4703, "step": 50400 }, { "epoch": 0.0013056, "grad_norm": 1.0788236856460571, "learning_rate": 1.9203169201070813e-05, "loss": 3.3386, "step": 50410 }, { "epoch": 0.0013312, "grad_norm": 1.7924749851226807, "learning_rate": 1.920285416757006e-05, "loss": 3.2746, "step": 50420 }, { "epoch": 0.0013568, "grad_norm": 1.9440467357635498, "learning_rate": 1.9202539074391232e-05, "loss": 3.3033, "step": 50430 }, { "epoch": 0.0013824, "grad_norm": 0.9742114543914795, "learning_rate": 1.920222392153639e-05, "loss": 3.3892, "step": 50440 }, { "epoch": 0.001408, "grad_norm": 0.8638966083526611, "learning_rate": 1.920190870900757e-05, "loss": 3.534, "step": 50450 }, { "epoch": 0.0014336, "grad_norm": 0.8894153833389282, "learning_rate": 1.9201593436806824e-05, "loss": 3.4136, "step": 50460 }, { "epoch": 0.0014592, "grad_norm": 0.9052250385284424, "learning_rate": 1.920127810493618e-05, "loss": 3.1908, "step": 50470 }, { "epoch": 0.0014848, "grad_norm": 0.7026944160461426, "learning_rate": 1.9200962713397698e-05, "loss": 3.1191, "step": 50480 }, { "epoch": 0.0015104, "grad_norm": 0.7672899961471558, "learning_rate": 1.920064726219342e-05, "loss": 3.1516, "step": 50490 }, { "epoch": 0.001536, "grad_norm": 0.7445535659790039, "learning_rate": 1.9200331751325388e-05, "loss": 3.143, "step": 50500 }, { "epoch": 0.0015616, "grad_norm": 0.8533522486686707, "learning_rate": 1.920001618079565e-05, "loss": 3.4909, "step": 50510 }, { "epoch": 0.0015872, "grad_norm": 1.3326066732406616, "learning_rate": 1.9199700550606248e-05, "loss": 3.1633, "step": 50520 }, { "epoch": 0.0016128, "grad_norm": 0.8023968935012817, "learning_rate": 1.919938486075924e-05, "loss": 3.5156, "step": 50530 }, { "epoch": 0.0016384, "grad_norm": 0.781933069229126, "learning_rate": 1.9199069111256662e-05, "loss": 3.3229, "step": 50540 }, { "epoch": 0.001664, "grad_norm": 0.8116629719734192, "learning_rate": 1.9198753302100567e-05, "loss": 3.2019, "step": 50550 }, { "epoch": 0.0016896, "grad_norm": 1.1200625896453857, "learning_rate": 1.9198437433293e-05, "loss": 3.2428, "step": 50560 }, { "epoch": 0.0017152, "grad_norm": 0.8959218263626099, "learning_rate": 1.919812150483601e-05, "loss": 3.498, "step": 50570 }, { "epoch": 0.0017408, "grad_norm": 0.8357654213905334, "learning_rate": 1.9197805516731648e-05, "loss": 3.3692, "step": 50580 }, { "epoch": 0.0017664, "grad_norm": 1.0095006227493286, "learning_rate": 1.9197489468981964e-05, "loss": 3.4644, "step": 50590 }, { "epoch": 0.001792, "grad_norm": 1.0447067022323608, "learning_rate": 1.9197173361589e-05, "loss": 3.0926, "step": 50600 }, { "epoch": 0.0018176, "grad_norm": 0.8192207217216492, "learning_rate": 1.9196857194554813e-05, "loss": 3.1582, "step": 50610 }, { "epoch": 0.0018432, "grad_norm": 1.3237494230270386, "learning_rate": 1.9196540967881453e-05, "loss": 3.3957, "step": 50620 }, { "epoch": 0.0018688, "grad_norm": 0.8668148517608643, "learning_rate": 1.919622468157097e-05, "loss": 3.3997, "step": 50630 }, { "epoch": 0.0018944, "grad_norm": 0.7766328454017639, "learning_rate": 1.9195908335625413e-05, "loss": 3.3949, "step": 50640 }, { "epoch": 0.00192, "grad_norm": 0.8279550671577454, "learning_rate": 1.9195591930046833e-05, "loss": 3.4655, "step": 50650 }, { "epoch": 0.0019456, "grad_norm": 0.9241006374359131, "learning_rate": 1.9195275464837283e-05, "loss": 3.3231, "step": 50660 }, { "epoch": 0.0019712, "grad_norm": 0.8062259554862976, "learning_rate": 1.9194958939998818e-05, "loss": 3.5151, "step": 50670 }, { "epoch": 0.0019968, "grad_norm": 0.8898740410804749, "learning_rate": 1.9194642355533486e-05, "loss": 3.4509, "step": 50680 }, { "epoch": 0.0020224, "grad_norm": 0.7560621500015259, "learning_rate": 1.9194325711443344e-05, "loss": 3.4243, "step": 50690 }, { "epoch": 0.002048, "grad_norm": 0.8112359046936035, "learning_rate": 1.919400900773044e-05, "loss": 3.4738, "step": 50700 }, { "epoch": 0.0020736, "grad_norm": 0.7506688833236694, "learning_rate": 1.9193692244396832e-05, "loss": 3.4713, "step": 50710 }, { "epoch": 0.0020992, "grad_norm": 0.9445849657058716, "learning_rate": 1.9193375421444577e-05, "loss": 3.3475, "step": 50720 }, { "epoch": 0.0021248, "grad_norm": 0.8335155248641968, "learning_rate": 1.919305853887572e-05, "loss": 3.5439, "step": 50730 }, { "epoch": 0.0021504, "grad_norm": 0.6986977458000183, "learning_rate": 1.9192741596692325e-05, "loss": 3.5147, "step": 50740 }, { "epoch": 0.002176, "grad_norm": 0.8053309321403503, "learning_rate": 1.9192424594896445e-05, "loss": 3.3186, "step": 50750 }, { "epoch": 0.0022016, "grad_norm": 0.8746131062507629, "learning_rate": 1.919210753349013e-05, "loss": 3.4481, "step": 50760 }, { "epoch": 0.0022272, "grad_norm": 0.9002577066421509, "learning_rate": 1.9191790412475444e-05, "loss": 3.6448, "step": 50770 }, { "epoch": 0.0022528, "grad_norm": 0.912501335144043, "learning_rate": 1.919147323185444e-05, "loss": 3.2392, "step": 50780 }, { "epoch": 0.0022784, "grad_norm": 0.8904776573181152, "learning_rate": 1.9191155991629173e-05, "loss": 3.3692, "step": 50790 }, { "epoch": 0.002304, "grad_norm": 0.824496865272522, "learning_rate": 1.91908386918017e-05, "loss": 3.5144, "step": 50800 }, { "epoch": 0.0023296, "grad_norm": 0.797749936580658, "learning_rate": 1.9190521332374088e-05, "loss": 3.4353, "step": 50810 }, { "epoch": 0.0023552, "grad_norm": 1.093535304069519, "learning_rate": 1.9190203913348383e-05, "loss": 3.3795, "step": 50820 }, { "epoch": 0.0023808, "grad_norm": 2.4866676330566406, "learning_rate": 1.918988643472665e-05, "loss": 3.1143, "step": 50830 }, { "epoch": 0.0024064, "grad_norm": 0.8918775320053101, "learning_rate": 1.9189568896510943e-05, "loss": 3.2877, "step": 50840 }, { "epoch": 0.002432, "grad_norm": 1.6419771909713745, "learning_rate": 1.9189251298703325e-05, "loss": 3.4525, "step": 50850 }, { "epoch": 0.0024576, "grad_norm": 0.9773121476173401, "learning_rate": 1.9188933641305858e-05, "loss": 3.2796, "step": 50860 }, { "epoch": 0.0024832, "grad_norm": 1.0026919841766357, "learning_rate": 1.9188615924320596e-05, "loss": 3.5814, "step": 50870 }, { "epoch": 0.0025088, "grad_norm": 1.1792265176773071, "learning_rate": 1.9188298147749602e-05, "loss": 3.5781, "step": 50880 }, { "epoch": 0.0025344, "grad_norm": 0.9230345487594604, "learning_rate": 1.9187980311594936e-05, "loss": 3.1972, "step": 50890 }, { "epoch": 0.00256, "grad_norm": 0.7395026087760925, "learning_rate": 1.9187662415858657e-05, "loss": 3.3026, "step": 50900 }, { "epoch": 0.0025856, "grad_norm": 0.7756673097610474, "learning_rate": 1.918734446054283e-05, "loss": 3.4978, "step": 50910 }, { "epoch": 0.0026112, "grad_norm": 0.8194543719291687, "learning_rate": 1.9187026445649518e-05, "loss": 3.5058, "step": 50920 }, { "epoch": 0.0026368, "grad_norm": 0.8672339916229248, "learning_rate": 1.9186708371180784e-05, "loss": 3.3239, "step": 50930 }, { "epoch": 0.0026624, "grad_norm": 1.070157766342163, "learning_rate": 1.9186390237138684e-05, "loss": 3.2897, "step": 50940 }, { "epoch": 0.002688, "grad_norm": 0.7587520480155945, "learning_rate": 1.9186072043525285e-05, "loss": 3.3784, "step": 50950 }, { "epoch": 0.0027136, "grad_norm": 0.9310948848724365, "learning_rate": 1.9185753790342648e-05, "loss": 3.5102, "step": 50960 }, { "epoch": 0.0027392, "grad_norm": 1.3065446615219116, "learning_rate": 1.9185435477592843e-05, "loss": 3.4303, "step": 50970 }, { "epoch": 0.0027648, "grad_norm": 0.9145968556404114, "learning_rate": 1.918511710527793e-05, "loss": 3.4345, "step": 50980 }, { "epoch": 0.0027904, "grad_norm": 0.8288535475730896, "learning_rate": 1.918479867339997e-05, "loss": 3.4035, "step": 50990 }, { "epoch": 0.002816, "grad_norm": 0.8290491104125977, "learning_rate": 1.9184480181961035e-05, "loss": 3.382, "step": 51000 }, { "epoch": 0.0028416, "grad_norm": 0.9490634202957153, "learning_rate": 1.9184161630963186e-05, "loss": 3.3709, "step": 51010 }, { "epoch": 0.0028672, "grad_norm": 0.8090299963951111, "learning_rate": 1.9183843020408488e-05, "loss": 3.3633, "step": 51020 }, { "epoch": 0.0028928, "grad_norm": 0.7164315581321716, "learning_rate": 1.918352435029901e-05, "loss": 3.5544, "step": 51030 }, { "epoch": 0.0029184, "grad_norm": 0.9522548317909241, "learning_rate": 1.9183205620636816e-05, "loss": 3.4507, "step": 51040 }, { "epoch": 0.002944, "grad_norm": 1.025025725364685, "learning_rate": 1.9182886831423977e-05, "loss": 3.3413, "step": 51050 }, { "epoch": 0.0029696, "grad_norm": 0.8448939919471741, "learning_rate": 1.9182567982662554e-05, "loss": 3.2682, "step": 51060 }, { "epoch": 0.0029952, "grad_norm": 0.7693493962287903, "learning_rate": 1.918224907435462e-05, "loss": 3.3501, "step": 51070 }, { "epoch": 0.0030208, "grad_norm": 0.7498416900634766, "learning_rate": 1.9181930106502242e-05, "loss": 3.3355, "step": 51080 }, { "epoch": 0.0030464, "grad_norm": 0.9833826422691345, "learning_rate": 1.9181611079107486e-05, "loss": 3.4008, "step": 51090 }, { "epoch": 0.003072, "grad_norm": 0.9509050846099854, "learning_rate": 1.9181291992172423e-05, "loss": 3.223, "step": 51100 }, { "epoch": 0.0030976, "grad_norm": 0.8279082179069519, "learning_rate": 1.9180972845699123e-05, "loss": 3.6416, "step": 51110 }, { "epoch": 0.0031232, "grad_norm": 1.0010310411453247, "learning_rate": 1.9180653639689652e-05, "loss": 3.2688, "step": 51120 }, { "epoch": 0.0031488, "grad_norm": 0.7758146524429321, "learning_rate": 1.918033437414608e-05, "loss": 3.4534, "step": 51130 }, { "epoch": 0.0031744, "grad_norm": 0.9352713823318481, "learning_rate": 1.9180015049070486e-05, "loss": 3.8548, "step": 51140 }, { "epoch": 0.0032, "grad_norm": 0.9245810508728027, "learning_rate": 1.9179695664464928e-05, "loss": 3.4689, "step": 51150 }, { "epoch": 0.0032256, "grad_norm": 0.7978968620300293, "learning_rate": 1.9179376220331483e-05, "loss": 3.3354, "step": 51160 }, { "epoch": 0.0032512, "grad_norm": 1.2991931438446045, "learning_rate": 1.917905671667223e-05, "loss": 3.4805, "step": 51170 }, { "epoch": 0.0032768, "grad_norm": 0.8042384386062622, "learning_rate": 1.9178737153489225e-05, "loss": 3.4632, "step": 51180 }, { "epoch": 0.0033024, "grad_norm": 0.7402207255363464, "learning_rate": 1.9178417530784556e-05, "loss": 3.3689, "step": 51190 }, { "epoch": 0.003328, "grad_norm": 0.8266066312789917, "learning_rate": 1.9178097848560284e-05, "loss": 3.193, "step": 51200 }, { "epoch": 0.0033536, "grad_norm": 1.1437128782272339, "learning_rate": 1.917777810681849e-05, "loss": 3.3308, "step": 51210 }, { "epoch": 0.0033792, "grad_norm": 0.7940859198570251, "learning_rate": 1.917745830556124e-05, "loss": 3.4351, "step": 51220 }, { "epoch": 0.0034048, "grad_norm": 0.7635122537612915, "learning_rate": 1.9177138444790618e-05, "loss": 3.3838, "step": 51230 }, { "epoch": 0.0034304, "grad_norm": 0.7702778577804565, "learning_rate": 1.9176818524508688e-05, "loss": 3.4357, "step": 51240 }, { "epoch": 0.003456, "grad_norm": 0.7873014211654663, "learning_rate": 1.917649854471753e-05, "loss": 3.1646, "step": 51250 }, { "epoch": 0.0034816, "grad_norm": 0.9576598405838013, "learning_rate": 1.9176178505419214e-05, "loss": 3.5434, "step": 51260 }, { "epoch": 0.0035072, "grad_norm": 0.8330161571502686, "learning_rate": 1.9175858406615824e-05, "loss": 3.3475, "step": 51270 }, { "epoch": 0.0035328, "grad_norm": 0.7344838976860046, "learning_rate": 1.917553824830943e-05, "loss": 3.4143, "step": 51280 }, { "epoch": 0.0035584, "grad_norm": 1.014691710472107, "learning_rate": 1.917521803050211e-05, "loss": 3.3648, "step": 51290 }, { "epoch": 0.003584, "grad_norm": 0.8946030735969543, "learning_rate": 1.9174897753195934e-05, "loss": 3.32, "step": 51300 }, { "epoch": 0.0036096, "grad_norm": 0.8278323411941528, "learning_rate": 1.917457741639299e-05, "loss": 3.4602, "step": 51310 }, { "epoch": 0.0036352, "grad_norm": 0.9388791918754578, "learning_rate": 1.9174257020095348e-05, "loss": 3.3944, "step": 51320 }, { "epoch": 0.0036608, "grad_norm": 0.8323115706443787, "learning_rate": 1.9173936564305087e-05, "loss": 3.3048, "step": 51330 }, { "epoch": 0.0036864, "grad_norm": 0.8638271689414978, "learning_rate": 1.9173616049024284e-05, "loss": 3.4085, "step": 51340 }, { "epoch": 0.003712, "grad_norm": 0.9158612489700317, "learning_rate": 1.9173295474255022e-05, "loss": 3.2601, "step": 51350 }, { "epoch": 0.0037376, "grad_norm": 0.8267661929130554, "learning_rate": 1.9172974839999375e-05, "loss": 3.6078, "step": 51360 }, { "epoch": 0.0037632, "grad_norm": 1.3643466234207153, "learning_rate": 1.9172654146259426e-05, "loss": 3.4176, "step": 51370 }, { "epoch": 0.0037888, "grad_norm": 0.7306091785430908, "learning_rate": 1.9172333393037253e-05, "loss": 3.564, "step": 51380 }, { "epoch": 0.0038144, "grad_norm": 1.544960618019104, "learning_rate": 1.9172012580334932e-05, "loss": 3.7516, "step": 51390 }, { "epoch": 0.00384, "grad_norm": 0.7302253246307373, "learning_rate": 1.917169170815455e-05, "loss": 3.2973, "step": 51400 }, { "epoch": 0.0038656, "grad_norm": 0.8453110456466675, "learning_rate": 1.9171370776498185e-05, "loss": 3.3703, "step": 51410 }, { "epoch": 0.0038912, "grad_norm": 1.3163468837738037, "learning_rate": 1.9171049785367916e-05, "loss": 3.5216, "step": 51420 }, { "epoch": 0.0039168, "grad_norm": 0.7946518063545227, "learning_rate": 1.9170728734765827e-05, "loss": 3.2878, "step": 51430 }, { "epoch": 0.0039424, "grad_norm": 1.0763065814971924, "learning_rate": 1.9170407624694e-05, "loss": 3.7, "step": 51440 }, { "epoch": 0.003968, "grad_norm": 0.6551638245582581, "learning_rate": 1.9170086455154516e-05, "loss": 3.1257, "step": 51450 }, { "epoch": 0.0039936, "grad_norm": 0.8738053441047668, "learning_rate": 1.916976522614946e-05, "loss": 3.4121, "step": 51460 }, { "epoch": 0.0040192, "grad_norm": 0.8520444631576538, "learning_rate": 1.9169443937680913e-05, "loss": 3.4549, "step": 51470 }, { "epoch": 0.0040448, "grad_norm": 0.7552017569541931, "learning_rate": 1.9169122589750958e-05, "loss": 3.3275, "step": 51480 }, { "epoch": 0.0040704, "grad_norm": 1.1988773345947266, "learning_rate": 1.9168801182361682e-05, "loss": 3.2802, "step": 51490 }, { "epoch": 0.004096, "grad_norm": 0.8837918043136597, "learning_rate": 1.9168479715515165e-05, "loss": 3.2519, "step": 51500 }, { "epoch": 0.0041216, "grad_norm": 0.8248169422149658, "learning_rate": 1.916815818921349e-05, "loss": 3.6555, "step": 51510 }, { "epoch": 0.0041472, "grad_norm": 1.7678892612457275, "learning_rate": 1.9167836603458755e-05, "loss": 3.4438, "step": 51520 }, { "epoch": 0.0041728, "grad_norm": 0.9365816116333008, "learning_rate": 1.916751495825303e-05, "loss": 3.4049, "step": 51530 }, { "epoch": 0.0041984, "grad_norm": 1.2264634370803833, "learning_rate": 1.9167193253598408e-05, "loss": 3.5683, "step": 51540 }, { "epoch": 0.004224, "grad_norm": 0.7942685484886169, "learning_rate": 1.9166871489496974e-05, "loss": 3.3909, "step": 51550 }, { "epoch": 0.0042496, "grad_norm": 0.8757058382034302, "learning_rate": 1.9166549665950813e-05, "loss": 3.6016, "step": 51560 }, { "epoch": 0.0042752, "grad_norm": 0.9371997117996216, "learning_rate": 1.9166227782962017e-05, "loss": 3.3779, "step": 51570 }, { "epoch": 0.0043008, "grad_norm": 0.9541649222373962, "learning_rate": 1.916590584053267e-05, "loss": 3.4003, "step": 51580 }, { "epoch": 0.0043264, "grad_norm": 0.803624153137207, "learning_rate": 1.9165583838664856e-05, "loss": 3.3628, "step": 51590 }, { "epoch": 0.004352, "grad_norm": 0.765300989151001, "learning_rate": 1.916526177736067e-05, "loss": 3.4634, "step": 51600 }, { "epoch": 0.0043776, "grad_norm": 1.6770600080490112, "learning_rate": 1.9164939656622197e-05, "loss": 3.3573, "step": 51610 }, { "epoch": 0.0044032, "grad_norm": 0.673304557800293, "learning_rate": 1.9164617476451522e-05, "loss": 3.3563, "step": 51620 }, { "epoch": 0.0044288, "grad_norm": 0.7612051963806152, "learning_rate": 1.9164295236850743e-05, "loss": 3.1515, "step": 51630 }, { "epoch": 0.0044544, "grad_norm": 0.8102282285690308, "learning_rate": 1.916397293782194e-05, "loss": 3.2747, "step": 51640 }, { "epoch": 0.00448, "grad_norm": 1.3035560846328735, "learning_rate": 1.9163650579367213e-05, "loss": 3.3487, "step": 51650 }, { "epoch": 0.0045056, "grad_norm": 0.8461115956306458, "learning_rate": 1.9163328161488647e-05, "loss": 3.3575, "step": 51660 }, { "epoch": 0.0045312, "grad_norm": 0.9175460934638977, "learning_rate": 1.9163005684188332e-05, "loss": 3.4386, "step": 51670 }, { "epoch": 0.0045568, "grad_norm": 0.8188594579696655, "learning_rate": 1.916268314746836e-05, "loss": 3.3809, "step": 51680 }, { "epoch": 0.0045824, "grad_norm": 0.766705334186554, "learning_rate": 1.9162360551330823e-05, "loss": 3.0792, "step": 51690 }, { "epoch": 0.004608, "grad_norm": 0.9345227479934692, "learning_rate": 1.9162037895777815e-05, "loss": 3.3877, "step": 51700 }, { "epoch": 0.0046336, "grad_norm": 0.9078870415687561, "learning_rate": 1.9161715180811425e-05, "loss": 3.7519, "step": 51710 }, { "epoch": 0.0046592, "grad_norm": 0.8896827101707458, "learning_rate": 1.9161392406433744e-05, "loss": 3.3714, "step": 51720 }, { "epoch": 0.0046848, "grad_norm": 0.8582196235656738, "learning_rate": 1.916106957264687e-05, "loss": 3.3218, "step": 51730 }, { "epoch": 0.0047104, "grad_norm": 0.9462164044380188, "learning_rate": 1.9160746679452896e-05, "loss": 3.384, "step": 51740 }, { "epoch": 0.004736, "grad_norm": 0.8590997457504272, "learning_rate": 1.9160423726853914e-05, "loss": 3.3162, "step": 51750 }, { "epoch": 0.0047616, "grad_norm": 1.3515887260437012, "learning_rate": 1.916010071485202e-05, "loss": 3.6756, "step": 51760 }, { "epoch": 0.0047872, "grad_norm": 0.8368569612503052, "learning_rate": 1.9159777643449303e-05, "loss": 3.651, "step": 51770 }, { "epoch": 0.0048128, "grad_norm": 0.9340158700942993, "learning_rate": 1.9159454512647863e-05, "loss": 3.2525, "step": 51780 }, { "epoch": 0.0048384, "grad_norm": 1.1315683126449585, "learning_rate": 1.91591313224498e-05, "loss": 3.4626, "step": 51790 }, { "epoch": 0.004864, "grad_norm": 0.7166658043861389, "learning_rate": 1.91588080728572e-05, "loss": 3.3579, "step": 51800 }, { "epoch": 0.0048896, "grad_norm": 0.7641305327415466, "learning_rate": 1.9158484763872166e-05, "loss": 3.2488, "step": 51810 }, { "epoch": 0.0049152, "grad_norm": 0.8666667342185974, "learning_rate": 1.9158161395496792e-05, "loss": 3.7554, "step": 51820 }, { "epoch": 0.0049408, "grad_norm": 0.790448784828186, "learning_rate": 1.915783796773317e-05, "loss": 3.1279, "step": 51830 }, { "epoch": 0.0049664, "grad_norm": 0.7680529952049255, "learning_rate": 1.915751448058341e-05, "loss": 3.4238, "step": 51840 }, { "epoch": 0.004992, "grad_norm": 0.6800299882888794, "learning_rate": 1.91571909340496e-05, "loss": 3.2293, "step": 51850 }, { "epoch": 0.0050176, "grad_norm": 0.7633655667304993, "learning_rate": 1.9156867328133838e-05, "loss": 3.438, "step": 51860 }, { "epoch": 0.0050432, "grad_norm": 0.984475314617157, "learning_rate": 1.9156543662838226e-05, "loss": 3.3389, "step": 51870 }, { "epoch": 0.0050688, "grad_norm": 0.7803070545196533, "learning_rate": 1.9156219938164863e-05, "loss": 3.5882, "step": 51880 }, { "epoch": 0.0050944, "grad_norm": 1.083443522453308, "learning_rate": 1.915589615411585e-05, "loss": 3.4089, "step": 51890 }, { "epoch": 0.00512, "grad_norm": 0.7384083271026611, "learning_rate": 1.915557231069328e-05, "loss": 3.4653, "step": 51900 }, { "epoch": 0.0051456, "grad_norm": 1.397839069366455, "learning_rate": 1.9155248407899257e-05, "loss": 3.2806, "step": 51910 }, { "epoch": 0.0051712, "grad_norm": 1.1677265167236328, "learning_rate": 1.915492444573588e-05, "loss": 3.3241, "step": 51920 }, { "epoch": 0.0051968, "grad_norm": 0.7636027336120605, "learning_rate": 1.9154600424205255e-05, "loss": 3.4014, "step": 51930 }, { "epoch": 0.0052224, "grad_norm": 1.0550776720046997, "learning_rate": 1.9154276343309475e-05, "loss": 3.53, "step": 51940 }, { "epoch": 0.005248, "grad_norm": 0.9606035947799683, "learning_rate": 1.915395220305065e-05, "loss": 3.1865, "step": 51950 }, { "epoch": 0.0052736, "grad_norm": 0.849803626537323, "learning_rate": 1.9153628003430875e-05, "loss": 3.3193, "step": 51960 }, { "epoch": 0.0052992, "grad_norm": 0.9491750001907349, "learning_rate": 1.9153303744452256e-05, "loss": 3.5757, "step": 51970 }, { "epoch": 0.0053248, "grad_norm": 0.7306290864944458, "learning_rate": 1.9152979426116896e-05, "loss": 3.4149, "step": 51980 }, { "epoch": 0.0053504, "grad_norm": 0.7534753084182739, "learning_rate": 1.9152655048426894e-05, "loss": 3.4352, "step": 51990 }, { "epoch": 0.005376, "grad_norm": 0.808108389377594, "learning_rate": 1.9152330611384358e-05, "loss": 3.3572, "step": 52000 }, { "epoch": 0.0054016, "grad_norm": 0.9823683500289917, "learning_rate": 1.915200611499139e-05, "loss": 3.3739, "step": 52010 }, { "epoch": 0.0054272, "grad_norm": 0.8485146760940552, "learning_rate": 1.9151681559250095e-05, "loss": 3.3592, "step": 52020 }, { "epoch": 0.0054528, "grad_norm": 0.8011845350265503, "learning_rate": 1.915135694416258e-05, "loss": 3.4727, "step": 52030 }, { "epoch": 0.0054784, "grad_norm": 1.1874979734420776, "learning_rate": 1.9151032269730945e-05, "loss": 3.5223, "step": 52040 }, { "epoch": 0.005504, "grad_norm": 1.22538423538208, "learning_rate": 1.9150707535957297e-05, "loss": 3.4715, "step": 52050 }, { "epoch": 0.0055296, "grad_norm": 0.7986845970153809, "learning_rate": 1.9150382742843744e-05, "loss": 3.5993, "step": 52060 }, { "epoch": 0.0055552, "grad_norm": 0.7604173421859741, "learning_rate": 1.9150057890392388e-05, "loss": 3.255, "step": 52070 }, { "epoch": 0.0055808, "grad_norm": 0.7493487000465393, "learning_rate": 1.914973297860534e-05, "loss": 3.2482, "step": 52080 }, { "epoch": 0.0056064, "grad_norm": 1.0149930715560913, "learning_rate": 1.9149408007484707e-05, "loss": 3.9003, "step": 52090 }, { "epoch": 0.005632, "grad_norm": 0.7716232538223267, "learning_rate": 1.9149082977032593e-05, "loss": 3.6232, "step": 52100 }, { "epoch": 0.0056576, "grad_norm": 0.7222391963005066, "learning_rate": 1.914875788725111e-05, "loss": 3.4476, "step": 52110 }, { "epoch": 0.0056832, "grad_norm": 0.7517922520637512, "learning_rate": 1.914843273814236e-05, "loss": 3.7419, "step": 52120 }, { "epoch": 0.0057088, "grad_norm": 1.1946861743927002, "learning_rate": 1.9148107529708455e-05, "loss": 3.1543, "step": 52130 }, { "epoch": 0.0057344, "grad_norm": 0.7569536566734314, "learning_rate": 1.9147782261951506e-05, "loss": 3.2318, "step": 52140 }, { "epoch": 0.00576, "grad_norm": 0.9194373488426208, "learning_rate": 1.914745693487362e-05, "loss": 3.1593, "step": 52150 }, { "epoch": 0.0057856, "grad_norm": 0.9019550085067749, "learning_rate": 1.9147131548476906e-05, "loss": 3.2676, "step": 52160 }, { "epoch": 0.0058112, "grad_norm": 0.7519251704216003, "learning_rate": 1.9146806102763475e-05, "loss": 3.5025, "step": 52170 }, { "epoch": 0.0058368, "grad_norm": 0.8144854307174683, "learning_rate": 1.914648059773544e-05, "loss": 3.347, "step": 52180 }, { "epoch": 0.0058624, "grad_norm": 0.9873960614204407, "learning_rate": 1.9146155033394908e-05, "loss": 3.4433, "step": 52190 }, { "epoch": 0.005888, "grad_norm": 0.9137505888938904, "learning_rate": 1.9145829409743987e-05, "loss": 3.6946, "step": 52200 }, { "epoch": 0.0059136, "grad_norm": 0.7972686886787415, "learning_rate": 1.91455037267848e-05, "loss": 3.4583, "step": 52210 }, { "epoch": 0.0059392, "grad_norm": 0.9466798305511475, "learning_rate": 1.9145177984519447e-05, "loss": 3.4975, "step": 52220 }, { "epoch": 0.0059648, "grad_norm": 0.7256863713264465, "learning_rate": 1.9144852182950046e-05, "loss": 3.4433, "step": 52230 }, { "epoch": 0.0059904, "grad_norm": 0.8645552396774292, "learning_rate": 1.914452632207871e-05, "loss": 3.3256, "step": 52240 }, { "epoch": 0.006016, "grad_norm": 0.7597209811210632, "learning_rate": 1.914420040190755e-05, "loss": 3.4533, "step": 52250 }, { "epoch": 0.0060416, "grad_norm": 0.7258925437927246, "learning_rate": 1.9143874422438682e-05, "loss": 3.4561, "step": 52260 }, { "epoch": 0.0060672, "grad_norm": 0.7088329792022705, "learning_rate": 1.9143548383674218e-05, "loss": 3.4414, "step": 52270 }, { "epoch": 0.0060928, "grad_norm": 0.914735734462738, "learning_rate": 1.914322228561627e-05, "loss": 3.4532, "step": 52280 }, { "epoch": 0.0061184, "grad_norm": 0.879095196723938, "learning_rate": 1.914289612826696e-05, "loss": 3.8253, "step": 52290 }, { "epoch": 0.006144, "grad_norm": 0.813201904296875, "learning_rate": 1.91425699116284e-05, "loss": 3.4751, "step": 52300 }, { "epoch": 0.0061696, "grad_norm": 0.8471327424049377, "learning_rate": 1.9142243635702698e-05, "loss": 3.2385, "step": 52310 }, { "epoch": 0.0061952, "grad_norm": 0.9475533366203308, "learning_rate": 1.914191730049198e-05, "loss": 3.4116, "step": 52320 }, { "epoch": 0.0062208, "grad_norm": 0.7040161490440369, "learning_rate": 1.9141590905998356e-05, "loss": 3.4898, "step": 52330 }, { "epoch": 0.0062464, "grad_norm": 0.8275058269500732, "learning_rate": 1.9141264452223942e-05, "loss": 3.4688, "step": 52340 }, { "epoch": 0.006272, "grad_norm": 0.7614169716835022, "learning_rate": 1.914093793917086e-05, "loss": 3.4121, "step": 52350 }, { "epoch": 0.0062976, "grad_norm": 0.9195089936256409, "learning_rate": 1.9140611366841226e-05, "loss": 3.2683, "step": 52360 }, { "epoch": 0.0063232, "grad_norm": 0.8387550115585327, "learning_rate": 1.9140284735237155e-05, "loss": 3.3496, "step": 52370 }, { "epoch": 0.0063488, "grad_norm": 0.7184533476829529, "learning_rate": 1.9139958044360767e-05, "loss": 3.1903, "step": 52380 }, { "epoch": 0.0063744, "grad_norm": 0.8317385315895081, "learning_rate": 1.913963129421418e-05, "loss": 3.2905, "step": 52390 }, { "epoch": 0.0064, "grad_norm": 0.7864481806755066, "learning_rate": 1.9139304484799513e-05, "loss": 3.5521, "step": 52400 }, { "epoch": 0.0064256, "grad_norm": 0.8797986507415771, "learning_rate": 1.913897761611889e-05, "loss": 3.4695, "step": 52410 }, { "epoch": 0.0064512, "grad_norm": 0.8071987628936768, "learning_rate": 1.9138650688174422e-05, "loss": 3.2925, "step": 52420 }, { "epoch": 0.0064768, "grad_norm": 0.8802893161773682, "learning_rate": 1.913832370096823e-05, "loss": 3.3537, "step": 52430 }, { "epoch": 0.0065024, "grad_norm": 0.8061981797218323, "learning_rate": 1.913799665450244e-05, "loss": 3.3349, "step": 52440 }, { "epoch": 0.006528, "grad_norm": 0.7870141267776489, "learning_rate": 1.9137669548779172e-05, "loss": 3.3506, "step": 52450 }, { "epoch": 0.0065536, "grad_norm": 0.7314761877059937, "learning_rate": 1.9137342383800544e-05, "loss": 3.4338, "step": 52460 }, { "epoch": 0.0065792, "grad_norm": 1.0003278255462646, "learning_rate": 1.9137015159568675e-05, "loss": 3.4295, "step": 52470 }, { "epoch": 0.0066048, "grad_norm": 3.7966554164886475, "learning_rate": 1.91366878760857e-05, "loss": 3.4308, "step": 52480 }, { "epoch": 0.0066304, "grad_norm": 1.07282292842865, "learning_rate": 1.9136360533353725e-05, "loss": 3.4486, "step": 52490 }, { "epoch": 0.006656, "grad_norm": 0.7419129014015198, "learning_rate": 1.9136033131374886e-05, "loss": 3.3884, "step": 52500 }, { "epoch": 0.0066816, "grad_norm": 0.8414716124534607, "learning_rate": 1.9135705670151293e-05, "loss": 3.4675, "step": 52510 }, { "epoch": 0.0067072, "grad_norm": 1.593376636505127, "learning_rate": 1.9135378149685084e-05, "loss": 3.4511, "step": 52520 }, { "epoch": 0.0067328, "grad_norm": 0.8006657361984253, "learning_rate": 1.913505056997837e-05, "loss": 3.4273, "step": 52530 }, { "epoch": 0.0067584, "grad_norm": 0.8606917858123779, "learning_rate": 1.9134722931033284e-05, "loss": 3.5183, "step": 52540 }, { "epoch": 0.006784, "grad_norm": 0.8345127105712891, "learning_rate": 1.9134395232851946e-05, "loss": 3.4424, "step": 52550 }, { "epoch": 0.0068096, "grad_norm": 0.8137942552566528, "learning_rate": 1.9134067475436483e-05, "loss": 3.4263, "step": 52560 }, { "epoch": 0.0068352, "grad_norm": 1.174970269203186, "learning_rate": 1.9133739658789022e-05, "loss": 3.2705, "step": 52570 }, { "epoch": 0.0068608, "grad_norm": 0.7373625040054321, "learning_rate": 1.9133411782911685e-05, "loss": 3.407, "step": 52580 }, { "epoch": 0.0068864, "grad_norm": 0.9200279116630554, "learning_rate": 1.91330838478066e-05, "loss": 3.2884, "step": 52590 }, { "epoch": 0.006912, "grad_norm": 0.8306771516799927, "learning_rate": 1.9132755853475894e-05, "loss": 3.3494, "step": 52600 }, { "epoch": 0.0069376, "grad_norm": 1.0493897199630737, "learning_rate": 1.913242779992169e-05, "loss": 3.5801, "step": 52610 }, { "epoch": 0.0069632, "grad_norm": 1.3508747816085815, "learning_rate": 1.9132099687146124e-05, "loss": 3.2882, "step": 52620 }, { "epoch": 0.0069888, "grad_norm": 0.9450843334197998, "learning_rate": 1.9131771515151313e-05, "loss": 3.4308, "step": 52630 }, { "epoch": 0.0070144, "grad_norm": 0.8079192638397217, "learning_rate": 1.9131443283939395e-05, "loss": 3.3173, "step": 52640 }, { "epoch": 0.00704, "grad_norm": 1.0667847394943237, "learning_rate": 1.913111499351249e-05, "loss": 3.2698, "step": 52650 }, { "epoch": 0.0070656, "grad_norm": 0.9146875739097595, "learning_rate": 1.913078664387274e-05, "loss": 3.4121, "step": 52660 }, { "epoch": 0.0070912, "grad_norm": 0.8816933035850525, "learning_rate": 1.9130458235022255e-05, "loss": 3.4527, "step": 52670 }, { "epoch": 0.0071168, "grad_norm": 0.773676335811615, "learning_rate": 1.913012976696318e-05, "loss": 3.2217, "step": 52680 }, { "epoch": 0.0071424, "grad_norm": 0.7720034718513489, "learning_rate": 1.912980123969764e-05, "loss": 3.4456, "step": 52690 }, { "epoch": 0.007168, "grad_norm": 0.8126999735832214, "learning_rate": 1.9129472653227764e-05, "loss": 3.5359, "step": 52700 }, { "epoch": 0.0071936, "grad_norm": 0.7909855246543884, "learning_rate": 1.9129144007555686e-05, "loss": 3.3413, "step": 52710 }, { "epoch": 0.0072192, "grad_norm": 0.9696994423866272, "learning_rate": 1.9128815302683534e-05, "loss": 3.3968, "step": 52720 }, { "epoch": 0.0072448, "grad_norm": 0.6997357606887817, "learning_rate": 1.912848653861344e-05, "loss": 3.4144, "step": 52730 }, { "epoch": 0.0072704, "grad_norm": 0.8125617504119873, "learning_rate": 1.9128157715347537e-05, "loss": 3.1236, "step": 52740 }, { "epoch": 0.007296, "grad_norm": 0.7238814830780029, "learning_rate": 1.912782883288796e-05, "loss": 3.2821, "step": 52750 }, { "epoch": 0.0073216, "grad_norm": 0.7679125666618347, "learning_rate": 1.912749989123684e-05, "loss": 3.3153, "step": 52760 }, { "epoch": 0.0073472, "grad_norm": 0.7929610013961792, "learning_rate": 1.9127170890396304e-05, "loss": 3.4554, "step": 52770 }, { "epoch": 0.0073728, "grad_norm": 0.8494355082511902, "learning_rate": 1.912684183036849e-05, "loss": 3.5096, "step": 52780 }, { "epoch": 0.0073984, "grad_norm": 0.9094053506851196, "learning_rate": 1.9126512711155537e-05, "loss": 3.2188, "step": 52790 }, { "epoch": 0.007424, "grad_norm": 0.9180529713630676, "learning_rate": 1.9126183532759573e-05, "loss": 3.3626, "step": 52800 }, { "epoch": 0.0074496, "grad_norm": 0.879849374294281, "learning_rate": 1.9125854295182732e-05, "loss": 3.5144, "step": 52810 }, { "epoch": 0.0074752, "grad_norm": 0.7918165326118469, "learning_rate": 1.912552499842715e-05, "loss": 3.2852, "step": 52820 }, { "epoch": 0.0075008, "grad_norm": 0.6958999037742615, "learning_rate": 1.9125195642494973e-05, "loss": 3.3136, "step": 52830 }, { "epoch": 0.0075264, "grad_norm": 0.7788407206535339, "learning_rate": 1.9124866227388316e-05, "loss": 3.3548, "step": 52840 }, { "epoch": 0.007552, "grad_norm": 0.9823695421218872, "learning_rate": 1.9124536753109334e-05, "loss": 3.4543, "step": 52850 }, { "epoch": 0.0075776, "grad_norm": 0.827656090259552, "learning_rate": 1.9124207219660154e-05, "loss": 3.3554, "step": 52860 }, { "epoch": 0.0076032, "grad_norm": 0.7717979550361633, "learning_rate": 1.9123877627042914e-05, "loss": 3.4032, "step": 52870 }, { "epoch": 0.0076288, "grad_norm": 1.0166841745376587, "learning_rate": 1.9123547975259752e-05, "loss": 3.329, "step": 52880 }, { "epoch": 0.0076544, "grad_norm": 1.1382694244384766, "learning_rate": 1.912321826431281e-05, "loss": 3.4825, "step": 52890 }, { "epoch": 0.00768, "grad_norm": 1.0350862741470337, "learning_rate": 1.912288849420422e-05, "loss": 3.4162, "step": 52900 }, { "epoch": 0.0077056, "grad_norm": 0.8794443607330322, "learning_rate": 1.912255866493612e-05, "loss": 3.374, "step": 52910 }, { "epoch": 0.0077312, "grad_norm": 0.7813350558280945, "learning_rate": 1.9122228776510656e-05, "loss": 3.261, "step": 52920 }, { "epoch": 0.0077568, "grad_norm": 0.7927353978157043, "learning_rate": 1.9121898828929958e-05, "loss": 3.3374, "step": 52930 }, { "epoch": 0.0077824, "grad_norm": 0.7837910056114197, "learning_rate": 1.9121568822196175e-05, "loss": 3.4648, "step": 52940 }, { "epoch": 0.007808, "grad_norm": 0.8122113347053528, "learning_rate": 1.912123875631144e-05, "loss": 3.1753, "step": 52950 }, { "epoch": 0.0078336, "grad_norm": 0.7463278770446777, "learning_rate": 1.9120908631277898e-05, "loss": 3.3682, "step": 52960 }, { "epoch": 0.0078592, "grad_norm": 1.0048950910568237, "learning_rate": 1.9120578447097686e-05, "loss": 3.2857, "step": 52970 }, { "epoch": 0.0078848, "grad_norm": 0.8104274272918701, "learning_rate": 1.9120248203772944e-05, "loss": 3.2362, "step": 52980 }, { "epoch": 0.0079104, "grad_norm": 0.7781168222427368, "learning_rate": 1.911991790130582e-05, "loss": 3.3541, "step": 52990 }, { "epoch": 0.007936, "grad_norm": 1.3675205707550049, "learning_rate": 1.9119587539698454e-05, "loss": 3.3858, "step": 53000 }, { "epoch": 0.0079616, "grad_norm": 0.8082374334335327, "learning_rate": 1.911925711895298e-05, "loss": 3.3503, "step": 53010 }, { "epoch": 0.0079872, "grad_norm": 0.8310667276382446, "learning_rate": 1.911892663907155e-05, "loss": 3.4457, "step": 53020 }, { "epoch": 0.0080128, "grad_norm": 0.7653807997703552, "learning_rate": 1.9118596100056308e-05, "loss": 3.2702, "step": 53030 }, { "epoch": 0.0080384, "grad_norm": 0.7050586938858032, "learning_rate": 1.911826550190939e-05, "loss": 3.4678, "step": 53040 }, { "epoch": 0.008064, "grad_norm": 0.8066926002502441, "learning_rate": 1.9117934844632942e-05, "loss": 3.5376, "step": 53050 }, { "epoch": 0.0080896, "grad_norm": 0.6960774660110474, "learning_rate": 1.911760412822911e-05, "loss": 3.3405, "step": 53060 }, { "epoch": 0.0081152, "grad_norm": 1.7113546133041382, "learning_rate": 1.9117273352700044e-05, "loss": 3.3657, "step": 53070 }, { "epoch": 0.0081408, "grad_norm": 0.6604812741279602, "learning_rate": 1.911694251804788e-05, "loss": 3.3274, "step": 53080 }, { "epoch": 0.0081664, "grad_norm": 0.7159879803657532, "learning_rate": 1.9116611624274767e-05, "loss": 3.3678, "step": 53090 }, { "epoch": 0.008192, "grad_norm": 0.7096481919288635, "learning_rate": 1.911628067138285e-05, "loss": 3.4058, "step": 53100 }, { "epoch": 0.0082176, "grad_norm": 0.7893007397651672, "learning_rate": 1.9115949659374277e-05, "loss": 3.3434, "step": 53110 }, { "epoch": 0.0082432, "grad_norm": 1.016305923461914, "learning_rate": 1.911561858825119e-05, "loss": 3.3234, "step": 53120 }, { "epoch": 0.0082688, "grad_norm": 0.8789947628974915, "learning_rate": 1.911528745801574e-05, "loss": 3.5914, "step": 53130 }, { "epoch": 0.0082944, "grad_norm": 0.7483550310134888, "learning_rate": 1.9114956268670078e-05, "loss": 3.4401, "step": 53140 }, { "epoch": 0.00832, "grad_norm": 0.9029511213302612, "learning_rate": 1.9114625020216345e-05, "loss": 3.5614, "step": 53150 }, { "epoch": 0.0083456, "grad_norm": 1.2494350671768188, "learning_rate": 1.9114293712656688e-05, "loss": 3.4032, "step": 53160 }, { "epoch": 0.0083712, "grad_norm": 0.7425829768180847, "learning_rate": 1.911396234599326e-05, "loss": 3.4414, "step": 53170 }, { "epoch": 0.0083968, "grad_norm": 0.7879291772842407, "learning_rate": 1.9113630920228207e-05, "loss": 3.5148, "step": 53180 }, { "epoch": 0.0084224, "grad_norm": 0.980535626411438, "learning_rate": 1.9113299435363682e-05, "loss": 3.3326, "step": 53190 }, { "epoch": 0.008448, "grad_norm": 0.9858416318893433, "learning_rate": 1.9112967891401832e-05, "loss": 3.4776, "step": 53200 }, { "epoch": 0.0084736, "grad_norm": 0.9376601576805115, "learning_rate": 1.9112636288344807e-05, "loss": 3.2277, "step": 53210 }, { "epoch": 0.0084992, "grad_norm": 1.0083107948303223, "learning_rate": 1.9112304626194755e-05, "loss": 3.2171, "step": 53220 }, { "epoch": 0.0085248, "grad_norm": 0.8343210220336914, "learning_rate": 1.9111972904953832e-05, "loss": 3.4647, "step": 53230 }, { "epoch": 0.0085504, "grad_norm": 1.0264462232589722, "learning_rate": 1.911164112462419e-05, "loss": 3.6256, "step": 53240 }, { "epoch": 0.008576, "grad_norm": 0.6351854205131531, "learning_rate": 1.911130928520797e-05, "loss": 3.1444, "step": 53250 }, { "epoch": 0.0086016, "grad_norm": 0.8559780716896057, "learning_rate": 1.9110977386707333e-05, "loss": 3.3895, "step": 53260 }, { "epoch": 0.0086272, "grad_norm": 0.7874879240989685, "learning_rate": 1.9110645429124427e-05, "loss": 3.2705, "step": 53270 }, { "epoch": 0.0086528, "grad_norm": 0.8564940094947815, "learning_rate": 1.911031341246141e-05, "loss": 3.3721, "step": 53280 }, { "epoch": 0.0086784, "grad_norm": 0.7127236127853394, "learning_rate": 1.910998133672043e-05, "loss": 3.2214, "step": 53290 }, { "epoch": 0.008704, "grad_norm": 0.8747738599777222, "learning_rate": 1.9109649201903644e-05, "loss": 3.4221, "step": 53300 }, { "epoch": 0.0087296, "grad_norm": 0.7608972787857056, "learning_rate": 1.9109317008013203e-05, "loss": 3.3297, "step": 53310 }, { "epoch": 0.0087552, "grad_norm": 0.9670854210853577, "learning_rate": 1.9108984755051267e-05, "loss": 3.366, "step": 53320 }, { "epoch": 0.0087808, "grad_norm": 0.7961580157279968, "learning_rate": 1.9108652443019977e-05, "loss": 3.4149, "step": 53330 }, { "epoch": 0.0088064, "grad_norm": 0.94575035572052, "learning_rate": 1.9108320071921503e-05, "loss": 3.2709, "step": 53340 }, { "epoch": 0.008832, "grad_norm": 1.0736572742462158, "learning_rate": 1.910798764175799e-05, "loss": 3.5107, "step": 53350 }, { "epoch": 0.0088576, "grad_norm": 0.793067455291748, "learning_rate": 1.91076551525316e-05, "loss": 3.3523, "step": 53360 }, { "epoch": 0.0088832, "grad_norm": 0.7811720967292786, "learning_rate": 1.9107322604244488e-05, "loss": 3.2985, "step": 53370 }, { "epoch": 0.0089088, "grad_norm": 0.8525583744049072, "learning_rate": 1.9106989996898807e-05, "loss": 3.4926, "step": 53380 }, { "epoch": 0.0089344, "grad_norm": 0.8399509191513062, "learning_rate": 1.9106657330496715e-05, "loss": 3.4783, "step": 53390 }, { "epoch": 0.00896, "grad_norm": 0.9689513444900513, "learning_rate": 1.9106324605040374e-05, "loss": 3.4049, "step": 53400 }, { "epoch": 0.0089856, "grad_norm": 0.8125225305557251, "learning_rate": 1.9105991820531934e-05, "loss": 3.2811, "step": 53410 }, { "epoch": 0.0090112, "grad_norm": 0.8052246570587158, "learning_rate": 1.9105658976973562e-05, "loss": 3.3648, "step": 53420 }, { "epoch": 0.0090368, "grad_norm": 1.1056640148162842, "learning_rate": 1.910532607436741e-05, "loss": 3.5202, "step": 53430 }, { "epoch": 0.0090624, "grad_norm": 1.0505605936050415, "learning_rate": 1.9104993112715635e-05, "loss": 3.4518, "step": 53440 }, { "epoch": 0.009088, "grad_norm": 0.7999097108840942, "learning_rate": 1.9104660092020403e-05, "loss": 3.2974, "step": 53450 }, { "epoch": 0.0091136, "grad_norm": 0.7907309532165527, "learning_rate": 1.9104327012283868e-05, "loss": 3.1515, "step": 53460 }, { "epoch": 0.0091392, "grad_norm": 0.8092707991600037, "learning_rate": 1.9103993873508195e-05, "loss": 3.5392, "step": 53470 }, { "epoch": 0.0091648, "grad_norm": 1.3098664283752441, "learning_rate": 1.9103660675695538e-05, "loss": 3.3367, "step": 53480 }, { "epoch": 0.0091904, "grad_norm": 0.7080894112586975, "learning_rate": 1.910332741884806e-05, "loss": 3.451, "step": 53490 }, { "epoch": 0.009216, "grad_norm": 1.0270874500274658, "learning_rate": 1.9102994102967927e-05, "loss": 3.3522, "step": 53500 }, { "epoch": 0.0092416, "grad_norm": 0.7996050119400024, "learning_rate": 1.9102660728057297e-05, "loss": 3.8609, "step": 53510 }, { "epoch": 0.0092672, "grad_norm": 0.8222805261611938, "learning_rate": 1.910232729411833e-05, "loss": 3.629, "step": 53520 }, { "epoch": 0.0092928, "grad_norm": 0.925222635269165, "learning_rate": 1.9101993801153186e-05, "loss": 3.5136, "step": 53530 }, { "epoch": 0.0093184, "grad_norm": 0.8881663084030151, "learning_rate": 1.9101660249164036e-05, "loss": 3.5392, "step": 53540 }, { "epoch": 0.009344, "grad_norm": 0.8884826898574829, "learning_rate": 1.9101326638153034e-05, "loss": 3.1878, "step": 53550 }, { "epoch": 0.0093696, "grad_norm": 0.8637227416038513, "learning_rate": 1.9100992968122352e-05, "loss": 3.2683, "step": 53560 }, { "epoch": 0.0093952, "grad_norm": 0.8458925485610962, "learning_rate": 1.9100659239074148e-05, "loss": 3.3846, "step": 53570 }, { "epoch": 0.0094208, "grad_norm": 0.782639741897583, "learning_rate": 1.9100325451010583e-05, "loss": 3.1587, "step": 53580 }, { "epoch": 0.0094464, "grad_norm": 8.16277027130127, "learning_rate": 1.9099991603933833e-05, "loss": 3.3203, "step": 53590 }, { "epoch": 0.009472, "grad_norm": 0.9050723910331726, "learning_rate": 1.9099657697846053e-05, "loss": 3.6601, "step": 53600 }, { "epoch": 0.0094976, "grad_norm": 0.8130155801773071, "learning_rate": 1.909932373274941e-05, "loss": 3.4672, "step": 53610 }, { "epoch": 0.0095232, "grad_norm": 0.8366908431053162, "learning_rate": 1.909898970864607e-05, "loss": 3.3236, "step": 53620 }, { "epoch": 0.0095488, "grad_norm": 0.8704665303230286, "learning_rate": 1.9098655625538203e-05, "loss": 3.5224, "step": 53630 }, { "epoch": 0.0095744, "grad_norm": 1.1319912672042847, "learning_rate": 1.9098321483427972e-05, "loss": 3.2789, "step": 53640 }, { "epoch": 0.0096, "grad_norm": 0.6950610280036926, "learning_rate": 1.9097987282317542e-05, "loss": 3.1257, "step": 53650 }, { "epoch": 0.0096256, "grad_norm": 1.6883004903793335, "learning_rate": 1.9097653022209085e-05, "loss": 3.3841, "step": 53660 }, { "epoch": 0.0096512, "grad_norm": 0.9219053983688354, "learning_rate": 1.9097318703104763e-05, "loss": 3.301, "step": 53670 }, { "epoch": 0.0096768, "grad_norm": 0.7388838529586792, "learning_rate": 1.9096984325006752e-05, "loss": 3.2486, "step": 53680 }, { "epoch": 0.0097024, "grad_norm": 0.7838048338890076, "learning_rate": 1.909664988791721e-05, "loss": 3.6581, "step": 53690 }, { "epoch": 0.009728, "grad_norm": 0.856458306312561, "learning_rate": 1.9096315391838312e-05, "loss": 3.3249, "step": 53700 }, { "epoch": 0.0097536, "grad_norm": 0.9568027257919312, "learning_rate": 1.9095980836772227e-05, "loss": 3.5328, "step": 53710 }, { "epoch": 0.0097792, "grad_norm": 0.8122425675392151, "learning_rate": 1.9095646222721123e-05, "loss": 3.3577, "step": 53720 }, { "epoch": 0.0098048, "grad_norm": 0.8832615613937378, "learning_rate": 1.9095311549687174e-05, "loss": 3.4717, "step": 53730 }, { "epoch": 0.0098304, "grad_norm": 0.7488026022911072, "learning_rate": 1.9094976817672543e-05, "loss": 3.3901, "step": 53740 }, { "epoch": 0.009856, "grad_norm": 0.9207197427749634, "learning_rate": 1.9094642026679407e-05, "loss": 3.3306, "step": 53750 }, { "epoch": 0.0098816, "grad_norm": 0.7580482959747314, "learning_rate": 1.9094307176709935e-05, "loss": 3.424, "step": 53760 }, { "epoch": 0.0099072, "grad_norm": 1.1358562707901, "learning_rate": 1.9093972267766295e-05, "loss": 3.3629, "step": 53770 }, { "epoch": 0.0099328, "grad_norm": 0.7302141785621643, "learning_rate": 1.9093637299850664e-05, "loss": 3.5649, "step": 53780 }, { "epoch": 0.0099584, "grad_norm": 0.8109645843505859, "learning_rate": 1.909330227296521e-05, "loss": 3.4157, "step": 53790 }, { "epoch": 0.009984, "grad_norm": 0.8131275773048401, "learning_rate": 1.909296718711211e-05, "loss": 3.315, "step": 53800 }, { "epoch": 0.0100096, "grad_norm": 0.916786253452301, "learning_rate": 1.9092632042293532e-05, "loss": 3.2914, "step": 53810 }, { "epoch": 0.0100352, "grad_norm": 1.2530817985534668, "learning_rate": 1.9092296838511656e-05, "loss": 3.4715, "step": 53820 }, { "epoch": 0.0100608, "grad_norm": 0.7819297313690186, "learning_rate": 1.9091961575768646e-05, "loss": 3.2589, "step": 53830 }, { "epoch": 0.0100864, "grad_norm": 1.055288553237915, "learning_rate": 1.9091626254066682e-05, "loss": 3.4142, "step": 53840 }, { "epoch": 0.010112, "grad_norm": 0.8233078718185425, "learning_rate": 1.9091290873407938e-05, "loss": 3.4701, "step": 53850 }, { "epoch": 0.0101376, "grad_norm": 0.7632480263710022, "learning_rate": 1.9090955433794593e-05, "loss": 3.5472, "step": 53860 }, { "epoch": 0.0101632, "grad_norm": 0.8015027642250061, "learning_rate": 1.9090619935228815e-05, "loss": 3.2843, "step": 53870 }, { "epoch": 0.0101888, "grad_norm": 0.8616126775741577, "learning_rate": 1.9090284377712783e-05, "loss": 3.5675, "step": 53880 }, { "epoch": 0.0102144, "grad_norm": 0.862774670124054, "learning_rate": 1.9089948761248672e-05, "loss": 3.479, "step": 53890 }, { "epoch": 0.01024, "grad_norm": 1.3958830833435059, "learning_rate": 1.9089613085838664e-05, "loss": 3.6679, "step": 53900 }, { "epoch": 0.0102656, "grad_norm": 0.8955956101417542, "learning_rate": 1.9089277351484928e-05, "loss": 3.3219, "step": 53910 }, { "epoch": 0.0102912, "grad_norm": 0.9258504509925842, "learning_rate": 1.9088941558189642e-05, "loss": 3.3218, "step": 53920 }, { "epoch": 0.0103168, "grad_norm": 0.8918859958648682, "learning_rate": 1.9088605705954984e-05, "loss": 3.6258, "step": 53930 }, { "epoch": 0.0103424, "grad_norm": 0.955093502998352, "learning_rate": 1.9088269794783137e-05, "loss": 3.771, "step": 53940 }, { "epoch": 0.010368, "grad_norm": 1.0672637224197388, "learning_rate": 1.9087933824676276e-05, "loss": 3.667, "step": 53950 }, { "epoch": 0.0103936, "grad_norm": 0.9440933465957642, "learning_rate": 1.9087597795636575e-05, "loss": 3.3846, "step": 53960 }, { "epoch": 0.0104192, "grad_norm": 1.333201289176941, "learning_rate": 1.9087261707666225e-05, "loss": 3.3384, "step": 53970 }, { "epoch": 0.0104448, "grad_norm": 0.7518957853317261, "learning_rate": 1.9086925560767392e-05, "loss": 3.4565, "step": 53980 }, { "epoch": 0.0104704, "grad_norm": 0.7655506730079651, "learning_rate": 1.9086589354942263e-05, "loss": 3.3087, "step": 53990 }, { "epoch": 0.010496, "grad_norm": 1.4062615633010864, "learning_rate": 1.908625309019302e-05, "loss": 3.5247, "step": 54000 }, { "epoch": 0.0105216, "grad_norm": 0.9637539982795715, "learning_rate": 1.9085916766521836e-05, "loss": 3.4047, "step": 54010 }, { "epoch": 0.0105472, "grad_norm": 0.722126305103302, "learning_rate": 1.9085580383930898e-05, "loss": 3.4371, "step": 54020 }, { "epoch": 0.0105728, "grad_norm": 0.7420392632484436, "learning_rate": 1.908524394242239e-05, "loss": 3.4383, "step": 54030 }, { "epoch": 0.0105984, "grad_norm": 0.7792378067970276, "learning_rate": 1.9084907441998482e-05, "loss": 3.3185, "step": 54040 }, { "epoch": 0.010624, "grad_norm": 0.8454945087432861, "learning_rate": 1.9084570882661367e-05, "loss": 3.3181, "step": 54050 }, { "epoch": 0.0106496, "grad_norm": 0.8021668791770935, "learning_rate": 1.908423426441323e-05, "loss": 4.3785, "step": 54060 }, { "epoch": 0.0106752, "grad_norm": 0.9713153839111328, "learning_rate": 1.908389758725624e-05, "loss": 3.414, "step": 54070 }, { "epoch": 0.0107008, "grad_norm": 0.861659049987793, "learning_rate": 1.908356085119259e-05, "loss": 3.5796, "step": 54080 }, { "epoch": 0.0107264, "grad_norm": 0.8972513675689697, "learning_rate": 1.9083224056224464e-05, "loss": 3.4712, "step": 54090 }, { "epoch": 0.010752, "grad_norm": 0.9111940860748291, "learning_rate": 1.9082887202354044e-05, "loss": 3.9931, "step": 54100 }, { "epoch": 0.0107776, "grad_norm": 0.8326712846755981, "learning_rate": 1.9082550289583515e-05, "loss": 3.2017, "step": 54110 }, { "epoch": 0.0108032, "grad_norm": 1.0353102684020996, "learning_rate": 1.9082213317915057e-05, "loss": 3.3834, "step": 54120 }, { "epoch": 0.0108288, "grad_norm": 0.8236315250396729, "learning_rate": 1.9081876287350864e-05, "loss": 3.8792, "step": 54130 }, { "epoch": 0.0108544, "grad_norm": 0.7834768295288086, "learning_rate": 1.9081539197893114e-05, "loss": 3.3662, "step": 54140 }, { "epoch": 0.01088, "grad_norm": 0.7366164326667786, "learning_rate": 1.9081202049544e-05, "loss": 3.3955, "step": 54150 }, { "epoch": 0.0109056, "grad_norm": 2.038027048110962, "learning_rate": 1.90808648423057e-05, "loss": 3.8281, "step": 54160 }, { "epoch": 0.0109312, "grad_norm": 1.0969524383544922, "learning_rate": 1.9080527576180405e-05, "loss": 3.6388, "step": 54170 }, { "epoch": 0.0109568, "grad_norm": 0.791443407535553, "learning_rate": 1.9080190251170304e-05, "loss": 3.4296, "step": 54180 }, { "epoch": 0.0109824, "grad_norm": 0.7538648247718811, "learning_rate": 1.9079852867277582e-05, "loss": 3.3522, "step": 54190 }, { "epoch": 0.011008, "grad_norm": 0.9581408500671387, "learning_rate": 1.9079515424504426e-05, "loss": 3.3785, "step": 54200 }, { "epoch": 0.0110336, "grad_norm": 0.7649551630020142, "learning_rate": 1.9079177922853027e-05, "loss": 3.3655, "step": 54210 }, { "epoch": 0.0110592, "grad_norm": 0.8011370301246643, "learning_rate": 1.9078840362325574e-05, "loss": 3.347, "step": 54220 }, { "epoch": 0.0110848, "grad_norm": 0.8102001547813416, "learning_rate": 1.9078502742924252e-05, "loss": 3.6338, "step": 54230 }, { "epoch": 0.0111104, "grad_norm": 1.0020924806594849, "learning_rate": 1.907816506465125e-05, "loss": 3.367, "step": 54240 }, { "epoch": 0.011136, "grad_norm": 0.8279439806938171, "learning_rate": 1.9077827327508765e-05, "loss": 3.2292, "step": 54250 }, { "epoch": 0.0111616, "grad_norm": 1.0665043592453003, "learning_rate": 1.907748953149898e-05, "loss": 3.3093, "step": 54260 }, { "epoch": 0.0111872, "grad_norm": 0.8400219678878784, "learning_rate": 1.9077151676624088e-05, "loss": 3.4782, "step": 54270 }, { "epoch": 0.0112128, "grad_norm": 0.7959520816802979, "learning_rate": 1.907681376288628e-05, "loss": 3.3381, "step": 54280 }, { "epoch": 0.0112384, "grad_norm": 0.8977729678153992, "learning_rate": 1.9076475790287745e-05, "loss": 3.3514, "step": 54290 }, { "epoch": 0.011264, "grad_norm": 1.1113812923431396, "learning_rate": 1.9076137758830678e-05, "loss": 3.4134, "step": 54300 }, { "epoch": 0.0112896, "grad_norm": 0.7841269969940186, "learning_rate": 1.907579966851727e-05, "loss": 3.2584, "step": 54310 }, { "epoch": 0.0113152, "grad_norm": 1.6503944396972656, "learning_rate": 1.9075461519349714e-05, "loss": 3.3545, "step": 54320 }, { "epoch": 0.0113408, "grad_norm": 0.8291256427764893, "learning_rate": 1.9075123311330205e-05, "loss": 3.1028, "step": 54330 }, { "epoch": 0.0113664, "grad_norm": 0.9694011211395264, "learning_rate": 1.9074785044460927e-05, "loss": 3.4041, "step": 54340 }, { "epoch": 0.011392, "grad_norm": 0.8217309713363647, "learning_rate": 1.9074446718744082e-05, "loss": 3.3983, "step": 54350 }, { "epoch": 0.0114176, "grad_norm": 1.4872312545776367, "learning_rate": 1.9074108334181863e-05, "loss": 3.3983, "step": 54360 }, { "epoch": 0.0114432, "grad_norm": 1.121881365776062, "learning_rate": 1.9073769890776465e-05, "loss": 3.6359, "step": 54370 }, { "epoch": 0.0114688, "grad_norm": 0.8479176759719849, "learning_rate": 1.9073431388530073e-05, "loss": 3.4649, "step": 54380 }, { "epoch": 0.0114944, "grad_norm": 0.8019997477531433, "learning_rate": 1.9073092827444897e-05, "loss": 3.499, "step": 54390 }, { "epoch": 0.01152, "grad_norm": 1.5478769540786743, "learning_rate": 1.9072754207523122e-05, "loss": 3.4888, "step": 54400 }, { "epoch": 0.0115456, "grad_norm": 0.926859974861145, "learning_rate": 1.907241552876695e-05, "loss": 3.4762, "step": 54410 }, { "epoch": 0.0115712, "grad_norm": 0.8402296900749207, "learning_rate": 1.9072076791178572e-05, "loss": 3.459, "step": 54420 }, { "epoch": 0.0115968, "grad_norm": 0.7318375706672668, "learning_rate": 1.907173799476019e-05, "loss": 3.3394, "step": 54430 }, { "epoch": 0.0116224, "grad_norm": 0.8874414563179016, "learning_rate": 1.9071399139513996e-05, "loss": 3.3281, "step": 54440 }, { "epoch": 0.011648, "grad_norm": 0.752440333366394, "learning_rate": 1.9071060225442188e-05, "loss": 3.6261, "step": 54450 }, { "epoch": 0.0116736, "grad_norm": 1.0279080867767334, "learning_rate": 1.9070721252546967e-05, "loss": 3.3376, "step": 54460 }, { "epoch": 0.0116992, "grad_norm": 0.8748027086257935, "learning_rate": 1.907038222083053e-05, "loss": 3.503, "step": 54470 }, { "epoch": 0.0117248, "grad_norm": 1.1636806726455688, "learning_rate": 1.9070043130295073e-05, "loss": 3.4396, "step": 54480 }, { "epoch": 0.0117504, "grad_norm": 0.8723304271697998, "learning_rate": 1.9069703980942798e-05, "loss": 3.6898, "step": 54490 }, { "epoch": 0.011776, "grad_norm": 0.7291685342788696, "learning_rate": 1.9069364772775905e-05, "loss": 3.3217, "step": 54500 }, { "epoch": 0.0118016, "grad_norm": 0.9598138928413391, "learning_rate": 1.9069025505796588e-05, "loss": 3.4751, "step": 54510 }, { "epoch": 0.0118272, "grad_norm": 0.7899320721626282, "learning_rate": 1.9068686180007052e-05, "loss": 3.4146, "step": 54520 }, { "epoch": 0.0118528, "grad_norm": 0.7623290419578552, "learning_rate": 1.90683467954095e-05, "loss": 3.3561, "step": 54530 }, { "epoch": 0.0118784, "grad_norm": 0.7564775943756104, "learning_rate": 1.9068007352006124e-05, "loss": 3.2912, "step": 54540 }, { "epoch": 0.011904, "grad_norm": 0.6434191465377808, "learning_rate": 1.9067667849799133e-05, "loss": 3.1192, "step": 54550 }, { "epoch": 0.0119296, "grad_norm": 0.8709043860435486, "learning_rate": 1.9067328288790726e-05, "loss": 3.2308, "step": 54560 }, { "epoch": 0.0119552, "grad_norm": 0.8401127457618713, "learning_rate": 1.90669886689831e-05, "loss": 3.3867, "step": 54570 }, { "epoch": 0.0119808, "grad_norm": 0.8549070358276367, "learning_rate": 1.906664899037847e-05, "loss": 3.3733, "step": 54580 }, { "epoch": 0.0120064, "grad_norm": 0.7109686732292175, "learning_rate": 1.9066309252979026e-05, "loss": 3.1533, "step": 54590 }, { "epoch": 0.012032, "grad_norm": 0.7708207964897156, "learning_rate": 1.9065969456786978e-05, "loss": 3.2881, "step": 54600 }, { "epoch": 0.0120576, "grad_norm": 0.8005481362342834, "learning_rate": 1.9065629601804527e-05, "loss": 3.3717, "step": 54610 }, { "epoch": 0.0120832, "grad_norm": 0.7673234343528748, "learning_rate": 1.9065289688033877e-05, "loss": 3.2876, "step": 54620 }, { "epoch": 0.0121088, "grad_norm": 0.7191876769065857, "learning_rate": 1.9064949715477233e-05, "loss": 3.2755, "step": 54630 }, { "epoch": 0.0121344, "grad_norm": 1.2705157995224, "learning_rate": 1.9064609684136797e-05, "loss": 3.4395, "step": 54640 }, { "epoch": 0.01216, "grad_norm": 1.1761566400527954, "learning_rate": 1.906426959401478e-05, "loss": 3.3635, "step": 54650 }, { "epoch": 0.0121856, "grad_norm": 1.1987155675888062, "learning_rate": 1.9063929445113383e-05, "loss": 3.3086, "step": 54660 }, { "epoch": 0.0122112, "grad_norm": 4.854874134063721, "learning_rate": 1.906358923743481e-05, "loss": 3.4586, "step": 54670 }, { "epoch": 0.0122368, "grad_norm": 0.7528548240661621, "learning_rate": 1.9063248970981273e-05, "loss": 3.3581, "step": 54680 }, { "epoch": 0.0122624, "grad_norm": 1.1959303617477417, "learning_rate": 1.9062908645754973e-05, "loss": 3.3484, "step": 54690 }, { "epoch": 0.012288, "grad_norm": 0.9284861087799072, "learning_rate": 1.906256826175812e-05, "loss": 3.1012, "step": 54700 }, { "epoch": 0.0123136, "grad_norm": 1.7011364698410034, "learning_rate": 1.9062227818992917e-05, "loss": 3.2823, "step": 54710 }, { "epoch": 0.0123392, "grad_norm": 1.2214287519454956, "learning_rate": 1.9061887317461578e-05, "loss": 2.9765, "step": 54720 }, { "epoch": 0.0123648, "grad_norm": 0.9077929258346558, "learning_rate": 1.906154675716631e-05, "loss": 3.3919, "step": 54730 }, { "epoch": 0.0123904, "grad_norm": 0.8024473190307617, "learning_rate": 1.9061206138109314e-05, "loss": 3.3872, "step": 54740 }, { "epoch": 0.012416, "grad_norm": 0.7477264404296875, "learning_rate": 1.906086546029281e-05, "loss": 3.2515, "step": 54750 }, { "epoch": 0.0124416, "grad_norm": 0.8074702024459839, "learning_rate": 1.9060524723718998e-05, "loss": 3.3474, "step": 54760 }, { "epoch": 0.0124672, "grad_norm": 1.0080413818359375, "learning_rate": 1.9060183928390094e-05, "loss": 3.4414, "step": 54770 }, { "epoch": 0.0124928, "grad_norm": 1.4084750413894653, "learning_rate": 1.9059843074308304e-05, "loss": 3.4159, "step": 54780 }, { "epoch": 0.0125184, "grad_norm": 0.9967642426490784, "learning_rate": 1.905950216147584e-05, "loss": 3.4796, "step": 54790 }, { "epoch": 0.012544, "grad_norm": 0.8647030591964722, "learning_rate": 1.905916118989491e-05, "loss": 3.2798, "step": 54800 }, { "epoch": 0.0125696, "grad_norm": 0.9445406198501587, "learning_rate": 1.9058820159567732e-05, "loss": 3.4181, "step": 54810 }, { "epoch": 0.0125952, "grad_norm": 2.350957155227661, "learning_rate": 1.905847907049651e-05, "loss": 3.3864, "step": 54820 }, { "epoch": 0.0126208, "grad_norm": 0.6744645237922668, "learning_rate": 1.9058137922683456e-05, "loss": 3.4021, "step": 54830 }, { "epoch": 0.0126464, "grad_norm": 0.8049789071083069, "learning_rate": 1.9057796716130786e-05, "loss": 3.5641, "step": 54840 }, { "epoch": 0.012672, "grad_norm": 1.0750455856323242, "learning_rate": 1.9057455450840715e-05, "loss": 3.5083, "step": 54850 }, { "epoch": 0.0126976, "grad_norm": 1.0497865676879883, "learning_rate": 1.9057114126815448e-05, "loss": 3.4067, "step": 54860 }, { "epoch": 0.0127232, "grad_norm": 1.835624098777771, "learning_rate": 1.9056772744057204e-05, "loss": 3.3813, "step": 54870 }, { "epoch": 0.0127488, "grad_norm": 0.6905769109725952, "learning_rate": 1.9056431302568196e-05, "loss": 3.3876, "step": 54880 }, { "epoch": 0.0127744, "grad_norm": 0.755840539932251, "learning_rate": 1.905608980235064e-05, "loss": 3.3881, "step": 54890 }, { "epoch": 0.0128, "grad_norm": 0.8636132478713989, "learning_rate": 1.9055748243406748e-05, "loss": 3.3034, "step": 54900 }, { "epoch": 0.0128256, "grad_norm": 0.8680897951126099, "learning_rate": 1.9055406625738736e-05, "loss": 3.5441, "step": 54910 }, { "epoch": 0.0128512, "grad_norm": 0.9762598276138306, "learning_rate": 1.9055064949348816e-05, "loss": 3.3537, "step": 54920 }, { "epoch": 0.0128768, "grad_norm": 0.8838036060333252, "learning_rate": 1.905472321423921e-05, "loss": 3.8874, "step": 54930 }, { "epoch": 0.0129024, "grad_norm": 0.7623813152313232, "learning_rate": 1.9054381420412125e-05, "loss": 3.452, "step": 54940 }, { "epoch": 0.012928, "grad_norm": 0.9361720681190491, "learning_rate": 1.9054039567869788e-05, "loss": 3.152, "step": 54950 }, { "epoch": 0.0129536, "grad_norm": 0.8735730051994324, "learning_rate": 1.905369765661441e-05, "loss": 3.2469, "step": 54960 }, { "epoch": 0.0129792, "grad_norm": 0.9769587516784668, "learning_rate": 1.9053355686648204e-05, "loss": 3.2891, "step": 54970 }, { "epoch": 0.0130048, "grad_norm": 0.8584842681884766, "learning_rate": 1.90530136579734e-05, "loss": 3.3752, "step": 54980 }, { "epoch": 0.0130304, "grad_norm": 0.7822132706642151, "learning_rate": 1.9052671570592204e-05, "loss": 3.8528, "step": 54990 }, { "epoch": 0.013056, "grad_norm": 0.7577342987060547, "learning_rate": 1.9052329424506838e-05, "loss": 3.2503, "step": 55000 }, { "epoch": 0.0130816, "grad_norm": 0.8397850394248962, "learning_rate": 1.905198721971952e-05, "loss": 3.6828, "step": 55010 }, { "epoch": 0.0131072, "grad_norm": 0.7123457193374634, "learning_rate": 1.9051644956232475e-05, "loss": 3.4026, "step": 55020 }, { "epoch": 0.0131328, "grad_norm": 0.8440022468566895, "learning_rate": 1.9051302634047916e-05, "loss": 3.3756, "step": 55030 }, { "epoch": 0.0131584, "grad_norm": 0.8039204478263855, "learning_rate": 1.9050960253168067e-05, "loss": 3.3199, "step": 55040 }, { "epoch": 0.013184, "grad_norm": 0.6943132877349854, "learning_rate": 1.9050617813595146e-05, "loss": 3.2637, "step": 55050 }, { "epoch": 0.0132096, "grad_norm": 1.0431795120239258, "learning_rate": 1.905027531533137e-05, "loss": 3.5083, "step": 55060 }, { "epoch": 0.0132352, "grad_norm": 0.8296928405761719, "learning_rate": 1.904993275837897e-05, "loss": 3.3735, "step": 55070 }, { "epoch": 0.0132608, "grad_norm": 1.1807233095169067, "learning_rate": 1.9049590142740153e-05, "loss": 3.6149, "step": 55080 }, { "epoch": 0.0132864, "grad_norm": 0.9971669912338257, "learning_rate": 1.9049247468417157e-05, "loss": 3.4965, "step": 55090 }, { "epoch": 0.013312, "grad_norm": 0.8669178485870361, "learning_rate": 1.9048904735412193e-05, "loss": 3.3155, "step": 55100 }, { "epoch": 0.0133376, "grad_norm": 1.072997808456421, "learning_rate": 1.9048561943727483e-05, "loss": 3.4066, "step": 55110 }, { "epoch": 0.0133632, "grad_norm": 1.0108113288879395, "learning_rate": 1.9048219093365257e-05, "loss": 3.3313, "step": 55120 }, { "epoch": 0.0133888, "grad_norm": 0.7814788222312927, "learning_rate": 1.9047876184327738e-05, "loss": 3.4393, "step": 55130 }, { "epoch": 0.0134144, "grad_norm": 0.732915997505188, "learning_rate": 1.9047533216617144e-05, "loss": 3.4039, "step": 55140 }, { "epoch": 0.01344, "grad_norm": 0.7347687482833862, "learning_rate": 1.90471901902357e-05, "loss": 3.3887, "step": 55150 }, { "epoch": 0.0134656, "grad_norm": 1.2019654512405396, "learning_rate": 1.9046847105185633e-05, "loss": 3.4557, "step": 55160 }, { "epoch": 0.0134912, "grad_norm": 0.7423869371414185, "learning_rate": 1.9046503961469168e-05, "loss": 3.4355, "step": 55170 }, { "epoch": 0.0135168, "grad_norm": 0.8280105590820312, "learning_rate": 1.9046160759088526e-05, "loss": 3.4325, "step": 55180 }, { "epoch": 0.0135424, "grad_norm": 0.7619309425354004, "learning_rate": 1.904581749804594e-05, "loss": 3.4616, "step": 55190 }, { "epoch": 0.013568, "grad_norm": 0.7534810304641724, "learning_rate": 1.904547417834363e-05, "loss": 3.2519, "step": 55200 }, { "epoch": 0.0135936, "grad_norm": 0.8209118843078613, "learning_rate": 1.904513079998382e-05, "loss": 4.5421, "step": 55210 }, { "epoch": 0.0136192, "grad_norm": 0.9519653916358948, "learning_rate": 1.9044787362968745e-05, "loss": 3.3774, "step": 55220 }, { "epoch": 0.0136448, "grad_norm": 0.7978696823120117, "learning_rate": 1.9044443867300627e-05, "loss": 3.2603, "step": 55230 }, { "epoch": 0.0136704, "grad_norm": 0.8712645173072815, "learning_rate": 1.9044100312981696e-05, "loss": 3.5274, "step": 55240 }, { "epoch": 0.013696, "grad_norm": 0.734870970249176, "learning_rate": 1.9043756700014176e-05, "loss": 3.2729, "step": 55250 }, { "epoch": 0.0137216, "grad_norm": 0.9004398584365845, "learning_rate": 1.9043413028400297e-05, "loss": 3.313, "step": 55260 }, { "epoch": 0.0137472, "grad_norm": 0.8160423040390015, "learning_rate": 1.9043069298142288e-05, "loss": 3.3979, "step": 55270 }, { "epoch": 0.0137728, "grad_norm": 0.894887387752533, "learning_rate": 1.9042725509242383e-05, "loss": 3.2478, "step": 55280 }, { "epoch": 0.0137984, "grad_norm": 9.912923812866211, "learning_rate": 1.9042381661702796e-05, "loss": 3.5072, "step": 55290 }, { "epoch": 0.013824, "grad_norm": 0.8006153702735901, "learning_rate": 1.9042037755525775e-05, "loss": 3.3825, "step": 55300 }, { "epoch": 0.0138496, "grad_norm": 0.8558513522148132, "learning_rate": 1.904169379071354e-05, "loss": 3.4271, "step": 55310 }, { "epoch": 0.0138752, "grad_norm": 0.824190616607666, "learning_rate": 1.9041349767268323e-05, "loss": 3.2634, "step": 55320 }, { "epoch": 0.0139008, "grad_norm": 0.7903153896331787, "learning_rate": 1.904100568519236e-05, "loss": 3.3865, "step": 55330 }, { "epoch": 0.0139264, "grad_norm": 0.9826022982597351, "learning_rate": 1.904066154448787e-05, "loss": 3.5229, "step": 55340 }, { "epoch": 0.013952, "grad_norm": 0.7229871153831482, "learning_rate": 1.90403173451571e-05, "loss": 3.4627, "step": 55350 }, { "epoch": 0.0139776, "grad_norm": 0.8322436213493347, "learning_rate": 1.9039973087202272e-05, "loss": 3.4103, "step": 55360 }, { "epoch": 0.0140032, "grad_norm": 0.8977500796318054, "learning_rate": 1.903962877062562e-05, "loss": 3.4618, "step": 55370 }, { "epoch": 0.0140288, "grad_norm": 0.9477660059928894, "learning_rate": 1.9039284395429378e-05, "loss": 3.3126, "step": 55380 }, { "epoch": 0.0140544, "grad_norm": 0.8992390632629395, "learning_rate": 1.903893996161578e-05, "loss": 3.6871, "step": 55390 }, { "epoch": 0.01408, "grad_norm": 0.7378250360488892, "learning_rate": 1.903859546918706e-05, "loss": 3.3824, "step": 55400 }, { "epoch": 0.0141056, "grad_norm": 0.7297703623771667, "learning_rate": 1.903825091814545e-05, "loss": 3.5769, "step": 55410 }, { "epoch": 0.0141312, "grad_norm": 0.8510831594467163, "learning_rate": 1.9037906308493183e-05, "loss": 3.4716, "step": 55420 }, { "epoch": 0.0141568, "grad_norm": 1.0621898174285889, "learning_rate": 1.9037561640232495e-05, "loss": 3.465, "step": 55430 }, { "epoch": 0.0141824, "grad_norm": 0.7558580636978149, "learning_rate": 1.9037216913365624e-05, "loss": 3.4801, "step": 55440 }, { "epoch": 0.014208, "grad_norm": 0.7317044138908386, "learning_rate": 1.90368721278948e-05, "loss": 3.2379, "step": 55450 }, { "epoch": 0.0142336, "grad_norm": 0.866299569606781, "learning_rate": 1.903652728382227e-05, "loss": 3.5678, "step": 55460 }, { "epoch": 0.0142592, "grad_norm": 0.8472599387168884, "learning_rate": 1.9036182381150254e-05, "loss": 3.237, "step": 55470 }, { "epoch": 0.0142848, "grad_norm": 0.7792297005653381, "learning_rate": 1.9035837419880998e-05, "loss": 3.455, "step": 55480 }, { "epoch": 0.0143104, "grad_norm": 0.7448993921279907, "learning_rate": 1.903549240001674e-05, "loss": 3.4177, "step": 55490 }, { "epoch": 0.014336, "grad_norm": 0.7245193719863892, "learning_rate": 1.9035147321559716e-05, "loss": 3.4611, "step": 55500 }, { "epoch": 0.0143616, "grad_norm": 1.9775407314300537, "learning_rate": 1.9034802184512157e-05, "loss": 3.5664, "step": 55510 }, { "epoch": 0.0143872, "grad_norm": 0.9745055437088013, "learning_rate": 1.903445698887631e-05, "loss": 3.3517, "step": 55520 }, { "epoch": 0.0144128, "grad_norm": 0.8690014481544495, "learning_rate": 1.9034111734654414e-05, "loss": 3.4088, "step": 55530 }, { "epoch": 0.0144384, "grad_norm": 0.8967826962471008, "learning_rate": 1.90337664218487e-05, "loss": 3.3761, "step": 55540 }, { "epoch": 0.014464, "grad_norm": 0.801646888256073, "learning_rate": 1.9033421050461416e-05, "loss": 3.3225, "step": 55550 }, { "epoch": 0.0144896, "grad_norm": 0.8182793855667114, "learning_rate": 1.9033075620494793e-05, "loss": 3.3824, "step": 55560 }, { "epoch": 0.0145152, "grad_norm": 1.3486568927764893, "learning_rate": 1.9032730131951077e-05, "loss": 3.154, "step": 55570 }, { "epoch": 0.0145408, "grad_norm": 0.7456823587417603, "learning_rate": 1.9032384584832506e-05, "loss": 3.3143, "step": 55580 }, { "epoch": 0.0145664, "grad_norm": 0.764039158821106, "learning_rate": 1.9032038979141326e-05, "loss": 3.4029, "step": 55590 }, { "epoch": 0.014592, "grad_norm": 0.9720666408538818, "learning_rate": 1.903169331487977e-05, "loss": 3.5451, "step": 55600 }, { "epoch": 0.0146176, "grad_norm": 0.8496038913726807, "learning_rate": 1.903134759205008e-05, "loss": 3.2752, "step": 55610 }, { "epoch": 0.0146432, "grad_norm": 0.9090603590011597, "learning_rate": 1.9031001810654506e-05, "loss": 3.4105, "step": 55620 }, { "epoch": 0.0146688, "grad_norm": 0.9076352119445801, "learning_rate": 1.9030655970695283e-05, "loss": 3.5562, "step": 55630 }, { "epoch": 0.0146944, "grad_norm": 0.6880495548248291, "learning_rate": 1.9030310072174655e-05, "loss": 3.4758, "step": 55640 }, { "epoch": 0.01472, "grad_norm": 0.8159661293029785, "learning_rate": 1.902996411509487e-05, "loss": 3.3377, "step": 55650 }, { "epoch": 0.0147456, "grad_norm": 1.4558244943618774, "learning_rate": 1.9029618099458166e-05, "loss": 3.3228, "step": 55660 }, { "epoch": 0.0147712, "grad_norm": 0.9515659809112549, "learning_rate": 1.9029272025266787e-05, "loss": 3.3685, "step": 55670 }, { "epoch": 0.0147968, "grad_norm": 0.9300277829170227, "learning_rate": 1.902892589252298e-05, "loss": 3.2792, "step": 55680 }, { "epoch": 0.0148224, "grad_norm": 0.7546036243438721, "learning_rate": 1.9028579701228987e-05, "loss": 3.3992, "step": 55690 }, { "epoch": 0.014848, "grad_norm": 0.7481986284255981, "learning_rate": 1.9028233451387055e-05, "loss": 3.3901, "step": 55700 }, { "epoch": 0.0148736, "grad_norm": 0.8611340522766113, "learning_rate": 1.9027887142999425e-05, "loss": 3.3567, "step": 55710 }, { "epoch": 0.0148992, "grad_norm": 0.7615721821784973, "learning_rate": 1.902754077606835e-05, "loss": 3.4107, "step": 55720 }, { "epoch": 0.0149248, "grad_norm": 0.739303708076477, "learning_rate": 1.9027194350596075e-05, "loss": 3.3691, "step": 55730 }, { "epoch": 0.0149504, "grad_norm": 0.8009814023971558, "learning_rate": 1.902684786658484e-05, "loss": 3.3363, "step": 55740 }, { "epoch": 0.014976, "grad_norm": 0.820163369178772, "learning_rate": 1.9026501324036893e-05, "loss": 3.3681, "step": 55750 }, { "epoch": 0.0150016, "grad_norm": 1.145176887512207, "learning_rate": 1.902615472295449e-05, "loss": 3.3073, "step": 55760 }, { "epoch": 0.0150272, "grad_norm": 0.8468707799911499, "learning_rate": 1.902580806333987e-05, "loss": 3.361, "step": 55770 }, { "epoch": 0.0150528, "grad_norm": 1.6040095090866089, "learning_rate": 1.9025461345195284e-05, "loss": 3.558, "step": 55780 }, { "epoch": 0.0150784, "grad_norm": 0.7954235672950745, "learning_rate": 1.9025114568522975e-05, "loss": 3.4822, "step": 55790 }, { "epoch": 0.015104, "grad_norm": 0.8522366881370544, "learning_rate": 1.9024767733325202e-05, "loss": 3.2327, "step": 55800 }, { "epoch": 0.0151296, "grad_norm": 1.3795713186264038, "learning_rate": 1.9024420839604206e-05, "loss": 3.4789, "step": 55810 }, { "epoch": 0.0151552, "grad_norm": 0.8984681367874146, "learning_rate": 1.902407388736224e-05, "loss": 3.441, "step": 55820 }, { "epoch": 0.0151808, "grad_norm": 0.7262293696403503, "learning_rate": 1.9023726876601556e-05, "loss": 3.6963, "step": 55830 }, { "epoch": 0.0152064, "grad_norm": 0.8946739435195923, "learning_rate": 1.90233798073244e-05, "loss": 3.4118, "step": 55840 }, { "epoch": 0.015232, "grad_norm": 0.9721665382385254, "learning_rate": 1.9023032679533024e-05, "loss": 3.6231, "step": 55850 }, { "epoch": 0.0152576, "grad_norm": 0.7577748894691467, "learning_rate": 1.9022685493229676e-05, "loss": 3.4045, "step": 55860 }, { "epoch": 0.0152832, "grad_norm": 0.8363291621208191, "learning_rate": 1.9022338248416613e-05, "loss": 3.0775, "step": 55870 }, { "epoch": 0.0153088, "grad_norm": 0.8087781071662903, "learning_rate": 1.9021990945096084e-05, "loss": 3.297, "step": 55880 }, { "epoch": 0.0153344, "grad_norm": 0.8398188948631287, "learning_rate": 1.9021643583270342e-05, "loss": 3.4341, "step": 55890 }, { "epoch": 0.01536, "grad_norm": 0.9840174317359924, "learning_rate": 1.9021296162941637e-05, "loss": 3.5613, "step": 55900 }, { "epoch": 0.0153856, "grad_norm": 0.8625457286834717, "learning_rate": 1.9020948684112225e-05, "loss": 3.2355, "step": 55910 }, { "epoch": 0.0154112, "grad_norm": 0.8007645010948181, "learning_rate": 1.902060114678436e-05, "loss": 3.463, "step": 55920 }, { "epoch": 0.0154368, "grad_norm": 0.7539651393890381, "learning_rate": 1.902025355096029e-05, "loss": 3.36, "step": 55930 }, { "epoch": 0.0154624, "grad_norm": 0.8124077320098877, "learning_rate": 1.901990589664228e-05, "loss": 3.2958, "step": 55940 }, { "epoch": 1.00000512, "grad_norm": 0.8641634583473206, "learning_rate": 1.9019558183832575e-05, "loss": 3.9523, "step": 55950 }, { "epoch": 1.00003072, "grad_norm": 0.8353500962257385, "learning_rate": 1.9019210412533427e-05, "loss": 3.2255, "step": 55960 }, { "epoch": 1.00005632, "grad_norm": 0.8613099455833435, "learning_rate": 1.90188625827471e-05, "loss": 3.188, "step": 55970 }, { "epoch": 1.00008192, "grad_norm": 0.6954614520072937, "learning_rate": 1.9018514694475848e-05, "loss": 3.0096, "step": 55980 }, { "epoch": 1.00010752, "grad_norm": 0.7205066680908203, "learning_rate": 1.9018166747721923e-05, "loss": 3.1807, "step": 55990 }, { "epoch": 1.00013312, "grad_norm": 0.7605234384536743, "learning_rate": 1.9017818742487583e-05, "loss": 3.2864, "step": 56000 }, { "epoch": 1.00015872, "grad_norm": 0.8357560634613037, "learning_rate": 1.9017470678775084e-05, "loss": 3.0166, "step": 56010 }, { "epoch": 1.00018432, "grad_norm": 0.8006376624107361, "learning_rate": 1.901712255658669e-05, "loss": 3.3121, "step": 56020 }, { "epoch": 1.00020992, "grad_norm": 0.8235640525817871, "learning_rate": 1.9016774375924646e-05, "loss": 3.2391, "step": 56030 }, { "epoch": 1.00023552, "grad_norm": 0.8319931626319885, "learning_rate": 1.9016426136791217e-05, "loss": 3.2114, "step": 56040 }, { "epoch": 1.00026112, "grad_norm": 0.7671266198158264, "learning_rate": 1.9016077839188666e-05, "loss": 3.2798, "step": 56050 }, { "epoch": 1.00028672, "grad_norm": 0.7483651041984558, "learning_rate": 1.9015729483119242e-05, "loss": 3.4287, "step": 56060 }, { "epoch": 1.00031232, "grad_norm": 0.6839696168899536, "learning_rate": 1.901538106858521e-05, "loss": 3.2928, "step": 56070 }, { "epoch": 1.00033792, "grad_norm": 0.8824326992034912, "learning_rate": 1.9015032595588826e-05, "loss": 3.2758, "step": 56080 }, { "epoch": 1.00036352, "grad_norm": 0.7540054321289062, "learning_rate": 1.9014684064132353e-05, "loss": 3.7292, "step": 56090 }, { "epoch": 1.00038912, "grad_norm": 0.7446434497833252, "learning_rate": 1.901433547421805e-05, "loss": 3.2862, "step": 56100 }, { "epoch": 1.00041472, "grad_norm": 0.7121774554252625, "learning_rate": 1.9013986825848175e-05, "loss": 3.0381, "step": 56110 }, { "epoch": 1.00044032, "grad_norm": 0.7845984101295471, "learning_rate": 1.9013638119024996e-05, "loss": 3.2988, "step": 56120 }, { "epoch": 1.00046592, "grad_norm": 0.6420475840568542, "learning_rate": 1.9013289353750767e-05, "loss": 3.1211, "step": 56130 }, { "epoch": 1.00049152, "grad_norm": 0.8075686097145081, "learning_rate": 1.9012940530027748e-05, "loss": 3.1765, "step": 56140 }, { "epoch": 1.00051712, "grad_norm": 0.9021642804145813, "learning_rate": 1.901259164785821e-05, "loss": 3.5035, "step": 56150 }, { "epoch": 1.00054272, "grad_norm": 0.808372437953949, "learning_rate": 1.901224270724441e-05, "loss": 3.5497, "step": 56160 }, { "epoch": 1.00056832, "grad_norm": 0.7882552146911621, "learning_rate": 1.901189370818861e-05, "loss": 3.0711, "step": 56170 }, { "epoch": 1.00059392, "grad_norm": 0.8714888095855713, "learning_rate": 1.901154465069307e-05, "loss": 2.9061, "step": 56180 }, { "epoch": 1.00061952, "grad_norm": 0.7306024432182312, "learning_rate": 1.9011195534760067e-05, "loss": 3.2268, "step": 56190 }, { "epoch": 1.00064512, "grad_norm": 0.7814797163009644, "learning_rate": 1.901084636039185e-05, "loss": 3.1669, "step": 56200 }, { "epoch": 1.00067072, "grad_norm": 0.7357686161994934, "learning_rate": 1.901049712759069e-05, "loss": 3.185, "step": 56210 }, { "epoch": 1.00069632, "grad_norm": 0.7075182199478149, "learning_rate": 1.901014783635885e-05, "loss": 3.0003, "step": 56220 }, { "epoch": 1.00072192, "grad_norm": 0.7548544406890869, "learning_rate": 1.9009798486698596e-05, "loss": 3.1941, "step": 56230 }, { "epoch": 1.00074752, "grad_norm": 0.8383930325508118, "learning_rate": 1.9009449078612195e-05, "loss": 3.3609, "step": 56240 }, { "epoch": 1.00077312, "grad_norm": 0.8636829853057861, "learning_rate": 1.900909961210191e-05, "loss": 3.1926, "step": 56250 }, { "epoch": 1.00079872, "grad_norm": 0.9321315884590149, "learning_rate": 1.9008750087170007e-05, "loss": 3.0125, "step": 56260 }, { "epoch": 1.00082432, "grad_norm": 0.7790606617927551, "learning_rate": 1.9008400503818757e-05, "loss": 3.2484, "step": 56270 }, { "epoch": 1.00084992, "grad_norm": 0.9229978322982788, "learning_rate": 1.9008050862050423e-05, "loss": 3.2504, "step": 56280 }, { "epoch": 1.00087552, "grad_norm": 0.8524863719940186, "learning_rate": 1.900770116186727e-05, "loss": 3.3912, "step": 56290 }, { "epoch": 1.00090112, "grad_norm": 0.8119648694992065, "learning_rate": 1.9007351403271573e-05, "loss": 3.0694, "step": 56300 }, { "epoch": 1.00092672, "grad_norm": 0.7303422093391418, "learning_rate": 1.9007001586265596e-05, "loss": 2.9674, "step": 56310 }, { "epoch": 1.00095232, "grad_norm": 0.756351888179779, "learning_rate": 1.9006651710851606e-05, "loss": 3.4742, "step": 56320 }, { "epoch": 1.00097792, "grad_norm": 0.8559258580207825, "learning_rate": 1.9006301777031872e-05, "loss": 3.0755, "step": 56330 }, { "epoch": 1.00100352, "grad_norm": 0.6980085372924805, "learning_rate": 1.9005951784808667e-05, "loss": 3.3393, "step": 56340 }, { "epoch": 1.00102912, "grad_norm": 0.7725419998168945, "learning_rate": 1.9005601734184255e-05, "loss": 3.2277, "step": 56350 }, { "epoch": 1.00105472, "grad_norm": 0.776841402053833, "learning_rate": 1.9005251625160915e-05, "loss": 3.2667, "step": 56360 }, { "epoch": 1.00108032, "grad_norm": 0.9828737378120422, "learning_rate": 1.900490145774091e-05, "loss": 3.1881, "step": 56370 }, { "epoch": 1.00110592, "grad_norm": 0.7249999642372131, "learning_rate": 1.9004551231926505e-05, "loss": 3.3443, "step": 56380 }, { "epoch": 1.00113152, "grad_norm": 0.7825957536697388, "learning_rate": 1.9004200947719984e-05, "loss": 3.3088, "step": 56390 }, { "epoch": 1.00115712, "grad_norm": 1.1208544969558716, "learning_rate": 1.9003850605123613e-05, "loss": 3.0979, "step": 56400 }, { "epoch": 1.00118272, "grad_norm": 0.7485799789428711, "learning_rate": 1.9003500204139662e-05, "loss": 3.1425, "step": 56410 }, { "epoch": 1.00120832, "grad_norm": 0.7531086802482605, "learning_rate": 1.9003149744770407e-05, "loss": 2.968, "step": 56420 }, { "epoch": 1.00123392, "grad_norm": 0.7183863520622253, "learning_rate": 1.900279922701812e-05, "loss": 3.0885, "step": 56430 }, { "epoch": 1.00125952, "grad_norm": 0.8110902905464172, "learning_rate": 1.900244865088507e-05, "loss": 3.1383, "step": 56440 }, { "epoch": 1.00128512, "grad_norm": 0.8306276798248291, "learning_rate": 1.9002098016373533e-05, "loss": 3.3248, "step": 56450 }, { "epoch": 1.00131072, "grad_norm": 0.750758945941925, "learning_rate": 1.9001747323485782e-05, "loss": 3.2037, "step": 56460 }, { "epoch": 1.00133632, "grad_norm": 0.7533720135688782, "learning_rate": 1.9001396572224095e-05, "loss": 3.0824, "step": 56470 }, { "epoch": 1.00136192, "grad_norm": 0.7880015969276428, "learning_rate": 1.900104576259074e-05, "loss": 3.3133, "step": 56480 }, { "epoch": 1.00138752, "grad_norm": 0.7073848843574524, "learning_rate": 1.9000694894588e-05, "loss": 3.2348, "step": 56490 }, { "epoch": 1.00141312, "grad_norm": 0.7446432113647461, "learning_rate": 1.900034396821814e-05, "loss": 3.4526, "step": 56500 }, { "epoch": 1.00143872, "grad_norm": 0.709836483001709, "learning_rate": 1.8999992983483446e-05, "loss": 3.1819, "step": 56510 }, { "epoch": 1.00146432, "grad_norm": 0.9371073842048645, "learning_rate": 1.899964194038619e-05, "loss": 3.0673, "step": 56520 }, { "epoch": 1.00148992, "grad_norm": 0.7646338939666748, "learning_rate": 1.8999290838928643e-05, "loss": 2.9912, "step": 56530 }, { "epoch": 1.00151552, "grad_norm": 0.6532027125358582, "learning_rate": 1.8998939679113092e-05, "loss": 3.0095, "step": 56540 }, { "epoch": 1.00154112, "grad_norm": 0.7601944208145142, "learning_rate": 1.8998588460941803e-05, "loss": 3.2209, "step": 56550 }, { "epoch": 1.00156672, "grad_norm": 0.6731627583503723, "learning_rate": 1.8998237184417062e-05, "loss": 3.3376, "step": 56560 }, { "epoch": 1.00159232, "grad_norm": 0.8883897066116333, "learning_rate": 1.8997885849541146e-05, "loss": 3.0999, "step": 56570 }, { "epoch": 1.00161792, "grad_norm": 0.7976786494255066, "learning_rate": 1.8997534456316332e-05, "loss": 3.2643, "step": 56580 }, { "epoch": 1.00164352, "grad_norm": 0.8477092981338501, "learning_rate": 1.8997183004744894e-05, "loss": 3.2366, "step": 56590 }, { "epoch": 1.00166912, "grad_norm": 0.9788588285446167, "learning_rate": 1.8996831494829118e-05, "loss": 3.0552, "step": 56600 }, { "epoch": 1.00169472, "grad_norm": 1.0021188259124756, "learning_rate": 1.8996479926571282e-05, "loss": 3.0739, "step": 56610 }, { "epoch": 1.00172032, "grad_norm": 0.8154973983764648, "learning_rate": 1.8996128299973665e-05, "loss": 3.3578, "step": 56620 }, { "epoch": 1.00174592, "grad_norm": 0.9032917022705078, "learning_rate": 1.8995776615038544e-05, "loss": 3.2449, "step": 56630 }, { "epoch": 1.00177152, "grad_norm": 0.7511865496635437, "learning_rate": 1.8995424871768203e-05, "loss": 3.2868, "step": 56640 }, { "epoch": 1.00179712, "grad_norm": 0.8593527674674988, "learning_rate": 1.8995073070164923e-05, "loss": 2.8817, "step": 56650 }, { "epoch": 1.00182272, "grad_norm": 4.127388000488281, "learning_rate": 1.8994721210230986e-05, "loss": 3.3671, "step": 56660 }, { "epoch": 1.00184832, "grad_norm": 0.7762357592582703, "learning_rate": 1.899436929196867e-05, "loss": 3.0986, "step": 56670 }, { "epoch": 1.00187392, "grad_norm": 0.742041826248169, "learning_rate": 1.899401731538026e-05, "loss": 3.225, "step": 56680 }, { "epoch": 1.00189952, "grad_norm": 0.7415924668312073, "learning_rate": 1.899366528046804e-05, "loss": 3.3128, "step": 56690 }, { "epoch": 1.00192512, "grad_norm": 0.842047393321991, "learning_rate": 1.8993313187234294e-05, "loss": 3.3743, "step": 56700 }, { "epoch": 1.00195072, "grad_norm": 0.7043729424476624, "learning_rate": 1.8992961035681296e-05, "loss": 3.1405, "step": 56710 }, { "epoch": 1.00197632, "grad_norm": 0.7017388939857483, "learning_rate": 1.8992608825811337e-05, "loss": 3.4006, "step": 56720 }, { "epoch": 1.00200192, "grad_norm": 0.7739050388336182, "learning_rate": 1.89922565576267e-05, "loss": 3.3196, "step": 56730 }, { "epoch": 1.00202752, "grad_norm": 0.858113169670105, "learning_rate": 1.899190423112967e-05, "loss": 3.3117, "step": 56740 }, { "epoch": 1.00205312, "grad_norm": 1.43377685546875, "learning_rate": 1.8991551846322533e-05, "loss": 3.3181, "step": 56750 }, { "epoch": 1.00207872, "grad_norm": 0.7271255254745483, "learning_rate": 1.8991199403207567e-05, "loss": 3.189, "step": 56760 }, { "epoch": 1.00210432, "grad_norm": 0.8209180235862732, "learning_rate": 1.8990846901787065e-05, "loss": 3.2554, "step": 56770 }, { "epoch": 1.00212992, "grad_norm": 0.7565674781799316, "learning_rate": 1.899049434206331e-05, "loss": 3.4496, "step": 56780 }, { "epoch": 1.00215552, "grad_norm": 0.6755463480949402, "learning_rate": 1.8990141724038588e-05, "loss": 3.3383, "step": 56790 }, { "epoch": 1.00218112, "grad_norm": 0.8079738616943359, "learning_rate": 1.8989789047715188e-05, "loss": 3.2393, "step": 56800 }, { "epoch": 1.00220672, "grad_norm": 1.3621597290039062, "learning_rate": 1.8989436313095392e-05, "loss": 3.3734, "step": 56810 }, { "epoch": 1.00223232, "grad_norm": 0.7700051665306091, "learning_rate": 1.8989083520181494e-05, "loss": 3.3838, "step": 56820 }, { "epoch": 1.00225792, "grad_norm": 0.781829833984375, "learning_rate": 1.8988730668975775e-05, "loss": 3.2127, "step": 56830 }, { "epoch": 1.00228352, "grad_norm": 0.8631604909896851, "learning_rate": 1.8988377759480528e-05, "loss": 3.2983, "step": 56840 }, { "epoch": 1.00230912, "grad_norm": 0.8332275748252869, "learning_rate": 1.8988024791698044e-05, "loss": 3.3619, "step": 56850 }, { "epoch": 1.00233472, "grad_norm": 0.7478851675987244, "learning_rate": 1.8987671765630603e-05, "loss": 3.2558, "step": 56860 }, { "epoch": 1.00236032, "grad_norm": 0.7027563452720642, "learning_rate": 1.89873186812805e-05, "loss": 3.2223, "step": 56870 }, { "epoch": 1.00238592, "grad_norm": 0.7234609723091125, "learning_rate": 1.8986965538650027e-05, "loss": 2.975, "step": 56880 }, { "epoch": 1.00241152, "grad_norm": 0.8766506910324097, "learning_rate": 1.8986612337741467e-05, "loss": 3.2211, "step": 56890 }, { "epoch": 1.00243712, "grad_norm": 0.743363082408905, "learning_rate": 1.8986259078557118e-05, "loss": 3.3704, "step": 56900 }, { "epoch": 1.00246272, "grad_norm": 1.2066177129745483, "learning_rate": 1.8985905761099267e-05, "loss": 3.0867, "step": 56910 }, { "epoch": 1.00248832, "grad_norm": 0.9958369135856628, "learning_rate": 1.89855523853702e-05, "loss": 3.4045, "step": 56920 }, { "epoch": 1.00251392, "grad_norm": 0.8919192552566528, "learning_rate": 1.8985198951372218e-05, "loss": 3.3765, "step": 56930 }, { "epoch": 1.00253952, "grad_norm": 0.7479714751243591, "learning_rate": 1.8984845459107607e-05, "loss": 3.1476, "step": 56940 }, { "epoch": 1.00256512, "grad_norm": 0.7850143313407898, "learning_rate": 1.8984491908578662e-05, "loss": 3.228, "step": 56950 }, { "epoch": 1.00259072, "grad_norm": 0.8112268447875977, "learning_rate": 1.8984138299787678e-05, "loss": 3.4195, "step": 56960 }, { "epoch": 1.00261632, "grad_norm": 0.7907868027687073, "learning_rate": 1.8983784632736938e-05, "loss": 3.3595, "step": 56970 }, { "epoch": 1.00264192, "grad_norm": 0.8447638154029846, "learning_rate": 1.898343090742875e-05, "loss": 3.1366, "step": 56980 }, { "epoch": 1.00266752, "grad_norm": 1.156601905822754, "learning_rate": 1.8983077123865394e-05, "loss": 3.1625, "step": 56990 }, { "epoch": 1.00269312, "grad_norm": 1.0324808359146118, "learning_rate": 1.898272328204917e-05, "loss": 3.3076, "step": 57000 }, { "epoch": 1.00271872, "grad_norm": 0.999040961265564, "learning_rate": 1.8982369381982377e-05, "loss": 3.3943, "step": 57010 }, { "epoch": 1.00274432, "grad_norm": 0.7667579650878906, "learning_rate": 1.8982015423667306e-05, "loss": 3.1712, "step": 57020 }, { "epoch": 1.00276992, "grad_norm": 0.845429003238678, "learning_rate": 1.898166140710625e-05, "loss": 3.4059, "step": 57030 }, { "epoch": 1.00279552, "grad_norm": 0.882304847240448, "learning_rate": 1.8981307332301508e-05, "loss": 3.2674, "step": 57040 }, { "epoch": 1.00282112, "grad_norm": 0.7530461549758911, "learning_rate": 1.898095319925537e-05, "loss": 3.2237, "step": 57050 }, { "epoch": 1.00284672, "grad_norm": 0.7535930275917053, "learning_rate": 1.8980599007970143e-05, "loss": 3.2696, "step": 57060 }, { "epoch": 1.00287232, "grad_norm": 0.9050033688545227, "learning_rate": 1.8980244758448116e-05, "loss": 3.2559, "step": 57070 }, { "epoch": 1.00289792, "grad_norm": 0.6778621673583984, "learning_rate": 1.8979890450691588e-05, "loss": 3.4183, "step": 57080 }, { "epoch": 1.00292352, "grad_norm": 0.7168656587600708, "learning_rate": 1.897953608470286e-05, "loss": 3.2473, "step": 57090 }, { "epoch": 1.00294912, "grad_norm": 0.8226510286331177, "learning_rate": 1.8979181660484224e-05, "loss": 3.198, "step": 57100 }, { "epoch": 1.00297472, "grad_norm": 0.9791408777236938, "learning_rate": 1.8978827178037983e-05, "loss": 3.2734, "step": 57110 }, { "epoch": 1.00300032, "grad_norm": 0.802794873714447, "learning_rate": 1.8978472637366433e-05, "loss": 3.2025, "step": 57120 }, { "epoch": 1.00302592, "grad_norm": 0.7413483262062073, "learning_rate": 1.8978118038471874e-05, "loss": 3.2113, "step": 57130 }, { "epoch": 1.00305152, "grad_norm": 0.8746786117553711, "learning_rate": 1.8977763381356603e-05, "loss": 3.1614, "step": 57140 }, { "epoch": 1.00307712, "grad_norm": 0.7535056471824646, "learning_rate": 1.897740866602293e-05, "loss": 3.147, "step": 57150 }, { "epoch": 1.00310272, "grad_norm": 0.7078986167907715, "learning_rate": 1.897705389247314e-05, "loss": 3.443, "step": 57160 }, { "epoch": 1.00312832, "grad_norm": 0.8061680793762207, "learning_rate": 1.8976699060709546e-05, "loss": 3.2271, "step": 57170 }, { "epoch": 1.00315392, "grad_norm": 0.7608914971351624, "learning_rate": 1.8976344170734443e-05, "loss": 3.2569, "step": 57180 }, { "epoch": 1.00317952, "grad_norm": 0.8224732279777527, "learning_rate": 1.8975989222550134e-05, "loss": 3.7595, "step": 57190 }, { "epoch": 1.00320512, "grad_norm": 0.7195225954055786, "learning_rate": 1.8975634216158917e-05, "loss": 3.3395, "step": 57200 }, { "epoch": 1.00323072, "grad_norm": 0.8760059475898743, "learning_rate": 1.8975279151563102e-05, "loss": 3.2365, "step": 57210 }, { "epoch": 1.00325632, "grad_norm": 1.338944911956787, "learning_rate": 1.8974924028764984e-05, "loss": 3.3159, "step": 57220 }, { "epoch": 1.00328192, "grad_norm": 1.0196378231048584, "learning_rate": 1.8974568847766873e-05, "loss": 3.3557, "step": 57230 }, { "epoch": 1.00330752, "grad_norm": 0.7118000388145447, "learning_rate": 1.8974213608571063e-05, "loss": 3.1995, "step": 57240 }, { "epoch": 1.00333312, "grad_norm": 0.862682580947876, "learning_rate": 1.8973858311179864e-05, "loss": 3.1348, "step": 57250 }, { "epoch": 1.00335872, "grad_norm": 1.0252286195755005, "learning_rate": 1.897350295559558e-05, "loss": 3.2041, "step": 57260 }, { "epoch": 1.00338432, "grad_norm": 0.7132575511932373, "learning_rate": 1.8973147541820517e-05, "loss": 3.2833, "step": 57270 }, { "epoch": 1.00340992, "grad_norm": 0.8318221569061279, "learning_rate": 1.8972792069856973e-05, "loss": 3.2566, "step": 57280 }, { "epoch": 1.00343552, "grad_norm": 0.7451911568641663, "learning_rate": 1.897243653970726e-05, "loss": 3.2313, "step": 57290 }, { "epoch": 1.00346112, "grad_norm": 0.8196182250976562, "learning_rate": 1.897208095137368e-05, "loss": 3.141, "step": 57300 }, { "epoch": 1.00348672, "grad_norm": 0.7774221897125244, "learning_rate": 1.8971725304858537e-05, "loss": 3.3341, "step": 57310 }, { "epoch": 1.00351232, "grad_norm": 1.078660011291504, "learning_rate": 1.8971369600164142e-05, "loss": 3.2655, "step": 57320 }, { "epoch": 1.00353792, "grad_norm": 0.8841590285301208, "learning_rate": 1.8971013837292798e-05, "loss": 3.2666, "step": 57330 }, { "epoch": 1.00356352, "grad_norm": 0.8265101909637451, "learning_rate": 1.8970658016246813e-05, "loss": 3.1888, "step": 57340 }, { "epoch": 1.00358912, "grad_norm": 0.8168798685073853, "learning_rate": 1.8970302137028497e-05, "loss": 3.2764, "step": 57350 }, { "epoch": 1.00361472, "grad_norm": 0.7259615659713745, "learning_rate": 1.8969946199640155e-05, "loss": 3.3064, "step": 57360 }, { "epoch": 1.00364032, "grad_norm": 0.7952498197555542, "learning_rate": 1.8969590204084096e-05, "loss": 3.2248, "step": 57370 }, { "epoch": 1.00366592, "grad_norm": 0.7278545498847961, "learning_rate": 1.8969234150362627e-05, "loss": 3.1577, "step": 57380 }, { "epoch": 1.00369152, "grad_norm": 0.7397390007972717, "learning_rate": 1.896887803847806e-05, "loss": 3.3138, "step": 57390 }, { "epoch": 1.00371712, "grad_norm": 0.8716837167739868, "learning_rate": 1.8968521868432703e-05, "loss": 3.171, "step": 57400 }, { "epoch": 1.00374272, "grad_norm": 0.8199489712715149, "learning_rate": 1.896816564022886e-05, "loss": 3.3693, "step": 57410 }, { "epoch": 1.00376832, "grad_norm": 0.8588352203369141, "learning_rate": 1.8967809353868854e-05, "loss": 3.3614, "step": 57420 }, { "epoch": 1.00379392, "grad_norm": 0.8084310293197632, "learning_rate": 1.896745300935498e-05, "loss": 3.4176, "step": 57430 }, { "epoch": 1.00381952, "grad_norm": 1.7912770509719849, "learning_rate": 1.8967096606689563e-05, "loss": 3.5959, "step": 57440 }, { "epoch": 1.00384512, "grad_norm": 0.9663181304931641, "learning_rate": 1.8966740145874905e-05, "loss": 3.1455, "step": 57450 }, { "epoch": 1.00387072, "grad_norm": 0.7216455340385437, "learning_rate": 1.896638362691332e-05, "loss": 3.1974, "step": 57460 }, { "epoch": 1.00389632, "grad_norm": 0.8046810030937195, "learning_rate": 1.8966027049807125e-05, "loss": 3.3513, "step": 57470 }, { "epoch": 1.00392192, "grad_norm": 0.7219101786613464, "learning_rate": 1.896567041455862e-05, "loss": 3.2099, "step": 57480 }, { "epoch": 1.00394752, "grad_norm": 0.9289111495018005, "learning_rate": 1.8965313721170128e-05, "loss": 3.4672, "step": 57490 }, { "epoch": 1.00397312, "grad_norm": 0.8264131546020508, "learning_rate": 1.896495696964396e-05, "loss": 3.0484, "step": 57500 }, { "epoch": 1.00399872, "grad_norm": 0.874557614326477, "learning_rate": 1.8964600159982428e-05, "loss": 3.2797, "step": 57510 }, { "epoch": 1.00402432, "grad_norm": 0.7472132444381714, "learning_rate": 1.8964243292187845e-05, "loss": 3.3142, "step": 57520 }, { "epoch": 1.00404992, "grad_norm": 0.7612395882606506, "learning_rate": 1.896388636626253e-05, "loss": 3.236, "step": 57530 }, { "epoch": 1.00407552, "grad_norm": 0.7665838003158569, "learning_rate": 1.8963529382208793e-05, "loss": 3.0467, "step": 57540 }, { "epoch": 1.00410112, "grad_norm": 0.8491336703300476, "learning_rate": 1.8963172340028947e-05, "loss": 3.2614, "step": 57550 }, { "epoch": 1.00412672, "grad_norm": 0.7944394946098328, "learning_rate": 1.8962815239725315e-05, "loss": 3.4481, "step": 57560 }, { "epoch": 1.00415232, "grad_norm": 1.1130355596542358, "learning_rate": 1.8962458081300206e-05, "loss": 3.3433, "step": 57570 }, { "epoch": 1.00417792, "grad_norm": 1.2765320539474487, "learning_rate": 1.896210086475594e-05, "loss": 3.3343, "step": 57580 }, { "epoch": 1.00420352, "grad_norm": 0.7478703856468201, "learning_rate": 1.896174359009483e-05, "loss": 3.4038, "step": 57590 }, { "epoch": 1.00422912, "grad_norm": 1.158678412437439, "learning_rate": 1.8961386257319193e-05, "loss": 3.291, "step": 57600 }, { "epoch": 1.00425472, "grad_norm": 0.7821869254112244, "learning_rate": 1.896102886643135e-05, "loss": 3.4493, "step": 57610 }, { "epoch": 1.00428032, "grad_norm": 0.9418575167655945, "learning_rate": 1.8960671417433615e-05, "loss": 3.2386, "step": 57620 }, { "epoch": 1.00430592, "grad_norm": 0.7135117650032043, "learning_rate": 1.896031391032831e-05, "loss": 3.2682, "step": 57630 }, { "epoch": 1.00433152, "grad_norm": 0.8001186847686768, "learning_rate": 1.8959956345117748e-05, "loss": 3.255, "step": 57640 }, { "epoch": 1.00435712, "grad_norm": 0.6722073554992676, "learning_rate": 1.8959598721804254e-05, "loss": 3.3747, "step": 57650 }, { "epoch": 1.00438272, "grad_norm": 1.1750407218933105, "learning_rate": 1.895924104039014e-05, "loss": 3.1524, "step": 57660 }, { "epoch": 1.00440832, "grad_norm": 0.7735525965690613, "learning_rate": 1.895888330087773e-05, "loss": 3.2373, "step": 57670 }, { "epoch": 1.00443392, "grad_norm": 1.1643660068511963, "learning_rate": 1.8958525503269342e-05, "loss": 3.0777, "step": 57680 }, { "epoch": 1.00445952, "grad_norm": 0.7751561999320984, "learning_rate": 1.8958167647567297e-05, "loss": 3.1848, "step": 57690 }, { "epoch": 1.00448512, "grad_norm": 0.7995643019676208, "learning_rate": 1.8957809733773917e-05, "loss": 3.2137, "step": 57700 }, { "epoch": 1.00451072, "grad_norm": 0.7577954530715942, "learning_rate": 1.895745176189152e-05, "loss": 3.3278, "step": 57710 }, { "epoch": 1.00453632, "grad_norm": 0.684726893901825, "learning_rate": 1.895709373192243e-05, "loss": 3.2173, "step": 57720 }, { "epoch": 1.00456192, "grad_norm": 0.7886570692062378, "learning_rate": 1.8956735643868968e-05, "loss": 3.2548, "step": 57730 }, { "epoch": 1.00458752, "grad_norm": 0.696211040019989, "learning_rate": 1.8956377497733454e-05, "loss": 3.0369, "step": 57740 }, { "epoch": 1.00461312, "grad_norm": 1.0184825658798218, "learning_rate": 1.8956019293518212e-05, "loss": 3.3853, "step": 57750 }, { "epoch": 1.00463872, "grad_norm": 0.8849095702171326, "learning_rate": 1.8955661031225567e-05, "loss": 3.5048, "step": 57760 }, { "epoch": 1.00466432, "grad_norm": 0.839787483215332, "learning_rate": 1.8955302710857842e-05, "loss": 3.2405, "step": 57770 }, { "epoch": 1.00468992, "grad_norm": 0.8565986156463623, "learning_rate": 1.8954944332417352e-05, "loss": 3.2124, "step": 57780 }, { "epoch": 1.00471552, "grad_norm": 0.7904036641120911, "learning_rate": 1.8954585895906435e-05, "loss": 3.2739, "step": 57790 }, { "epoch": 1.00474112, "grad_norm": 0.9072005152702332, "learning_rate": 1.8954227401327405e-05, "loss": 3.1718, "step": 57800 }, { "epoch": 1.00476672, "grad_norm": 1.1120959520339966, "learning_rate": 1.8953868848682592e-05, "loss": 3.5547, "step": 57810 }, { "epoch": 1.00479232, "grad_norm": 0.8152310848236084, "learning_rate": 1.895351023797432e-05, "loss": 3.4227, "step": 57820 }, { "epoch": 1.00481792, "grad_norm": 0.773758590221405, "learning_rate": 1.895315156920491e-05, "loss": 3.2103, "step": 57830 }, { "epoch": 1.00484352, "grad_norm": 0.8684396743774414, "learning_rate": 1.8952792842376696e-05, "loss": 3.3147, "step": 57840 }, { "epoch": 1.00486912, "grad_norm": 1.2373906373977661, "learning_rate": 1.8952434057491997e-05, "loss": 3.1115, "step": 57850 }, { "epoch": 1.00489472, "grad_norm": 1.9749830961227417, "learning_rate": 1.895207521455314e-05, "loss": 3.355, "step": 57860 }, { "epoch": 1.00492032, "grad_norm": 0.7364128828048706, "learning_rate": 1.8951716313562457e-05, "loss": 3.3754, "step": 57870 }, { "epoch": 1.00494592, "grad_norm": 0.8481390476226807, "learning_rate": 1.8951357354522274e-05, "loss": 3.0979, "step": 57880 }, { "epoch": 1.00497152, "grad_norm": 0.7622774839401245, "learning_rate": 1.8950998337434917e-05, "loss": 3.2945, "step": 57890 }, { "epoch": 1.00499712, "grad_norm": 0.7748206853866577, "learning_rate": 1.8950639262302713e-05, "loss": 3.1832, "step": 57900 }, { "epoch": 1.00502272, "grad_norm": 0.8852390646934509, "learning_rate": 1.8950280129127993e-05, "loss": 3.2398, "step": 57910 }, { "epoch": 1.00504832, "grad_norm": 0.75918048620224, "learning_rate": 1.8949920937913085e-05, "loss": 3.2363, "step": 57920 }, { "epoch": 1.00507392, "grad_norm": 0.7659853100776672, "learning_rate": 1.894956168866032e-05, "loss": 3.425, "step": 57930 }, { "epoch": 1.00509952, "grad_norm": 0.8076262474060059, "learning_rate": 1.8949202381372024e-05, "loss": 3.2984, "step": 57940 }, { "epoch": 1.00512512, "grad_norm": 0.7472148537635803, "learning_rate": 1.8948843016050527e-05, "loss": 3.3123, "step": 57950 }, { "epoch": 1.00515072, "grad_norm": 0.7305997610092163, "learning_rate": 1.8948483592698166e-05, "loss": 3.1853, "step": 57960 }, { "epoch": 1.00517632, "grad_norm": 0.7842053771018982, "learning_rate": 1.8948124111317264e-05, "loss": 3.2118, "step": 57970 }, { "epoch": 1.00520192, "grad_norm": 1.0214872360229492, "learning_rate": 1.8947764571910158e-05, "loss": 3.228, "step": 57980 }, { "epoch": 1.00522752, "grad_norm": 0.7801662683486938, "learning_rate": 1.8947404974479173e-05, "loss": 3.3234, "step": 57990 }, { "epoch": 1.00525312, "grad_norm": 0.7807407379150391, "learning_rate": 1.8947045319026648e-05, "loss": 3.1136, "step": 58000 }, { "epoch": 1.00527872, "grad_norm": 0.7564948797225952, "learning_rate": 1.8946685605554908e-05, "loss": 3.3288, "step": 58010 }, { "epoch": 1.00530432, "grad_norm": 0.788832426071167, "learning_rate": 1.8946325834066296e-05, "loss": 3.3731, "step": 58020 }, { "epoch": 1.00532992, "grad_norm": 0.9039220213890076, "learning_rate": 1.894596600456313e-05, "loss": 3.23, "step": 58030 }, { "epoch": 1.00535552, "grad_norm": 1.2698739767074585, "learning_rate": 1.894560611704776e-05, "loss": 3.3636, "step": 58040 }, { "epoch": 1.00538112, "grad_norm": 1.0137678384780884, "learning_rate": 1.894524617152251e-05, "loss": 3.2591, "step": 58050 }, { "epoch": 1.00540672, "grad_norm": 1.1279776096343994, "learning_rate": 1.8944886167989714e-05, "loss": 3.2394, "step": 58060 }, { "epoch": 1.00543232, "grad_norm": 0.8617894053459167, "learning_rate": 1.894452610645171e-05, "loss": 3.1355, "step": 58070 }, { "epoch": 1.00545792, "grad_norm": 0.7760174870491028, "learning_rate": 1.8944165986910828e-05, "loss": 3.3762, "step": 58080 }, { "epoch": 1.00548352, "grad_norm": 0.7729512453079224, "learning_rate": 1.8943805809369407e-05, "loss": 3.4311, "step": 58090 }, { "epoch": 1.00550912, "grad_norm": 0.7717087268829346, "learning_rate": 1.8943445573829787e-05, "loss": 3.2454, "step": 58100 }, { "epoch": 1.00553472, "grad_norm": 0.9149520397186279, "learning_rate": 1.89430852802943e-05, "loss": 3.4255, "step": 58110 }, { "epoch": 1.00556032, "grad_norm": 0.7650254964828491, "learning_rate": 1.8942724928765274e-05, "loss": 3.1452, "step": 58120 }, { "epoch": 1.00558592, "grad_norm": 0.7401390075683594, "learning_rate": 1.894236451924506e-05, "loss": 3.1186, "step": 58130 }, { "epoch": 1.00561152, "grad_norm": 0.7746999859809875, "learning_rate": 1.8942004051735987e-05, "loss": 3.7786, "step": 58140 }, { "epoch": 1.00563712, "grad_norm": 0.6779032349586487, "learning_rate": 1.8941643526240394e-05, "loss": 3.395, "step": 58150 }, { "epoch": 1.00566272, "grad_norm": 1.9015247821807861, "learning_rate": 1.8941282942760618e-05, "loss": 3.3949, "step": 58160 }, { "epoch": 1.00568832, "grad_norm": 2.251011371612549, "learning_rate": 1.8940922301298998e-05, "loss": 3.3416, "step": 58170 }, { "epoch": 1.00571392, "grad_norm": 0.9170510768890381, "learning_rate": 1.8940561601857878e-05, "loss": 3.1811, "step": 58180 }, { "epoch": 1.00573952, "grad_norm": 0.783301591873169, "learning_rate": 1.8940200844439585e-05, "loss": 3.0723, "step": 58190 }, { "epoch": 1.00576512, "grad_norm": 0.870514452457428, "learning_rate": 1.893984002904647e-05, "loss": 3.086, "step": 58200 }, { "epoch": 1.00579072, "grad_norm": 0.9488427639007568, "learning_rate": 1.8939479155680868e-05, "loss": 3.1596, "step": 58210 }, { "epoch": 1.00581632, "grad_norm": 0.8048334717750549, "learning_rate": 1.8939118224345118e-05, "loss": 3.2977, "step": 58220 }, { "epoch": 1.00584192, "grad_norm": 1.1895653009414673, "learning_rate": 1.8938757235041565e-05, "loss": 3.266, "step": 58230 }, { "epoch": 1.00586752, "grad_norm": 0.9130065441131592, "learning_rate": 1.8938396187772545e-05, "loss": 3.3123, "step": 58240 }, { "epoch": 1.00589312, "grad_norm": 0.7327914834022522, "learning_rate": 1.89380350825404e-05, "loss": 3.517, "step": 58250 }, { "epoch": 1.00591872, "grad_norm": 0.8614506125450134, "learning_rate": 1.8937673919347473e-05, "loss": 3.3418, "step": 58260 }, { "epoch": 1.00594432, "grad_norm": 0.8455150723457336, "learning_rate": 1.893731269819611e-05, "loss": 3.3863, "step": 58270 }, { "epoch": 1.00596992, "grad_norm": 0.776435136795044, "learning_rate": 1.8936951419088644e-05, "loss": 3.2565, "step": 58280 }, { "epoch": 1.00599552, "grad_norm": 1.0658682584762573, "learning_rate": 1.893659008202743e-05, "loss": 3.1715, "step": 58290 }, { "epoch": 1.00602112, "grad_norm": 0.7938683032989502, "learning_rate": 1.89362286870148e-05, "loss": 3.3414, "step": 58300 }, { "epoch": 1.00604672, "grad_norm": 0.7267460823059082, "learning_rate": 1.89358672340531e-05, "loss": 3.3479, "step": 58310 }, { "epoch": 1.00607232, "grad_norm": 0.7902297973632812, "learning_rate": 1.8935505723144677e-05, "loss": 3.3167, "step": 58320 }, { "epoch": 1.00609792, "grad_norm": 0.8134797215461731, "learning_rate": 1.8935144154291877e-05, "loss": 3.4434, "step": 58330 }, { "epoch": 1.00612352, "grad_norm": 0.7813754677772522, "learning_rate": 1.893478252749704e-05, "loss": 3.5655, "step": 58340 }, { "epoch": 1.00614912, "grad_norm": 0.9495521783828735, "learning_rate": 1.8934420842762513e-05, "loss": 3.226, "step": 58350 }, { "epoch": 1.00617472, "grad_norm": 0.7419652938842773, "learning_rate": 1.893405910009064e-05, "loss": 3.141, "step": 58360 }, { "epoch": 1.00620032, "grad_norm": 0.8375995755195618, "learning_rate": 1.8933697299483772e-05, "loss": 3.3293, "step": 58370 }, { "epoch": 1.00622592, "grad_norm": 0.7561655640602112, "learning_rate": 1.8933335440944247e-05, "loss": 3.3491, "step": 58380 }, { "epoch": 1.00625152, "grad_norm": 1.0687987804412842, "learning_rate": 1.8932973524474415e-05, "loss": 3.3228, "step": 58390 }, { "epoch": 1.00627712, "grad_norm": 0.8329102993011475, "learning_rate": 1.8932611550076626e-05, "loss": 3.3244, "step": 58400 }, { "epoch": 1.00630272, "grad_norm": 0.7295559644699097, "learning_rate": 1.8932249517753228e-05, "loss": 3.0834, "step": 58410 }, { "epoch": 1.00632832, "grad_norm": 0.8214912414550781, "learning_rate": 1.8931887427506563e-05, "loss": 3.2037, "step": 58420 }, { "epoch": 1.00635392, "grad_norm": 0.8390556573867798, "learning_rate": 1.8931525279338982e-05, "loss": 3.0777, "step": 58430 }, { "epoch": 1.00637952, "grad_norm": 0.9372396469116211, "learning_rate": 1.8931163073252834e-05, "loss": 3.2486, "step": 58440 }, { "epoch": 1.00640512, "grad_norm": 0.7781122326850891, "learning_rate": 1.893080080925047e-05, "loss": 3.4266, "step": 58450 }, { "epoch": 1.00643072, "grad_norm": 0.8086370229721069, "learning_rate": 1.8930438487334228e-05, "loss": 3.2808, "step": 58460 }, { "epoch": 1.00645632, "grad_norm": 0.7532852292060852, "learning_rate": 1.893007610750647e-05, "loss": 3.2415, "step": 58470 }, { "epoch": 1.00648192, "grad_norm": 0.7006362080574036, "learning_rate": 1.8929713669769546e-05, "loss": 3.1549, "step": 58480 }, { "epoch": 1.00650752, "grad_norm": 0.8173174858093262, "learning_rate": 1.8929351174125798e-05, "loss": 3.2571, "step": 58490 }, { "epoch": 1.00653312, "grad_norm": 1.123107671737671, "learning_rate": 1.892898862057758e-05, "loss": 3.267, "step": 58500 }, { "epoch": 1.00655872, "grad_norm": 0.7857993841171265, "learning_rate": 1.8928626009127246e-05, "loss": 3.3024, "step": 58510 }, { "epoch": 1.00658432, "grad_norm": 0.780623733997345, "learning_rate": 1.8928263339777145e-05, "loss": 3.2514, "step": 58520 }, { "epoch": 1.00660992, "grad_norm": 0.7422295212745667, "learning_rate": 1.8927900612529625e-05, "loss": 3.3814, "step": 58530 }, { "epoch": 1.00663552, "grad_norm": 0.8268074989318848, "learning_rate": 1.8927537827387047e-05, "loss": 3.265, "step": 58540 }, { "epoch": 1.00666112, "grad_norm": 0.9751174449920654, "learning_rate": 1.8927174984351756e-05, "loss": 3.4464, "step": 58550 }, { "epoch": 1.00668672, "grad_norm": 0.8084890842437744, "learning_rate": 1.892681208342611e-05, "loss": 3.1949, "step": 58560 }, { "epoch": 1.00671232, "grad_norm": 1.0881149768829346, "learning_rate": 1.892644912461246e-05, "loss": 3.3606, "step": 58570 }, { "epoch": 1.00673792, "grad_norm": 0.7673830986022949, "learning_rate": 1.892608610791316e-05, "loss": 3.2938, "step": 58580 }, { "epoch": 1.00676352, "grad_norm": 0.9419379234313965, "learning_rate": 1.892572303333056e-05, "loss": 3.333, "step": 58590 }, { "epoch": 1.00678912, "grad_norm": 0.7812657356262207, "learning_rate": 1.8925359900867022e-05, "loss": 3.2691, "step": 58600 }, { "epoch": 1.00681472, "grad_norm": 0.7571923136711121, "learning_rate": 1.8924996710524896e-05, "loss": 3.3655, "step": 58610 }, { "epoch": 1.00684032, "grad_norm": 0.7636945843696594, "learning_rate": 1.892463346230654e-05, "loss": 3.129, "step": 58620 }, { "epoch": 1.00686592, "grad_norm": 0.7412009835243225, "learning_rate": 1.8924270156214304e-05, "loss": 3.2619, "step": 58630 }, { "epoch": 1.00689152, "grad_norm": 0.7776269316673279, "learning_rate": 1.8923906792250553e-05, "loss": 3.188, "step": 58640 }, { "epoch": 1.00691712, "grad_norm": 1.0071979761123657, "learning_rate": 1.8923543370417635e-05, "loss": 3.2375, "step": 58650 }, { "epoch": 1.00694272, "grad_norm": 0.7976208925247192, "learning_rate": 1.892317989071791e-05, "loss": 3.4326, "step": 58660 }, { "epoch": 1.00696832, "grad_norm": 0.8167060017585754, "learning_rate": 1.8922816353153736e-05, "loss": 3.1762, "step": 58670 }, { "epoch": 1.00699392, "grad_norm": 1.2882726192474365, "learning_rate": 1.892245275772747e-05, "loss": 3.2398, "step": 58680 }, { "epoch": 1.00701952, "grad_norm": 0.9448082447052002, "learning_rate": 1.8922089104441466e-05, "loss": 3.2271, "step": 58690 }, { "epoch": 1.00704512, "grad_norm": 0.8329285383224487, "learning_rate": 1.892172539329809e-05, "loss": 3.1696, "step": 58700 }, { "epoch": 1.00707072, "grad_norm": 0.7440597414970398, "learning_rate": 1.8921361624299692e-05, "loss": 3.335, "step": 58710 }, { "epoch": 1.00709632, "grad_norm": 0.73825603723526, "learning_rate": 1.8920997797448634e-05, "loss": 3.3169, "step": 58720 }, { "epoch": 1.00712192, "grad_norm": 0.9824492931365967, "learning_rate": 1.892063391274728e-05, "loss": 3.1149, "step": 58730 }, { "epoch": 1.00714752, "grad_norm": 0.7771764397621155, "learning_rate": 1.8920269970197983e-05, "loss": 3.3269, "step": 58740 }, { "epoch": 1.00717312, "grad_norm": 0.7977641224861145, "learning_rate": 1.891990596980311e-05, "loss": 3.4043, "step": 58750 }, { "epoch": 1.00719872, "grad_norm": 0.7406814098358154, "learning_rate": 1.8919541911565012e-05, "loss": 3.1742, "step": 58760 }, { "epoch": 1.00722432, "grad_norm": 0.8753317594528198, "learning_rate": 1.891917779548606e-05, "loss": 3.3006, "step": 58770 }, { "epoch": 1.00724992, "grad_norm": 0.7762271165847778, "learning_rate": 1.8918813621568607e-05, "loss": 3.2891, "step": 58780 }, { "epoch": 1.00727552, "grad_norm": 0.9859418272972107, "learning_rate": 1.891844938981502e-05, "loss": 2.9817, "step": 58790 }, { "epoch": 1.00730112, "grad_norm": 0.7892842292785645, "learning_rate": 1.8918085100227657e-05, "loss": 3.2281, "step": 58800 }, { "epoch": 1.00732672, "grad_norm": 0.7780755758285522, "learning_rate": 1.8917720752808885e-05, "loss": 3.2197, "step": 58810 }, { "epoch": 1.00735232, "grad_norm": 0.7954865097999573, "learning_rate": 1.8917356347561063e-05, "loss": 3.3956, "step": 58820 }, { "epoch": 1.00737792, "grad_norm": 0.750461220741272, "learning_rate": 1.891699188448655e-05, "loss": 3.3405, "step": 58830 }, { "epoch": 1.00740352, "grad_norm": 0.8289059400558472, "learning_rate": 1.8916627363587722e-05, "loss": 3.1309, "step": 58840 }, { "epoch": 1.00742912, "grad_norm": 0.8422766923904419, "learning_rate": 1.8916262784866932e-05, "loss": 3.2946, "step": 58850 }, { "epoch": 1.00745472, "grad_norm": 0.7735729813575745, "learning_rate": 1.8915898148326546e-05, "loss": 3.2899, "step": 58860 }, { "epoch": 1.00748032, "grad_norm": 0.7700726985931396, "learning_rate": 1.8915533453968933e-05, "loss": 3.2571, "step": 58870 }, { "epoch": 1.00750592, "grad_norm": 0.7330123782157898, "learning_rate": 1.8915168701796454e-05, "loss": 3.2046, "step": 58880 }, { "epoch": 1.00753152, "grad_norm": 0.7655715942382812, "learning_rate": 1.8914803891811475e-05, "loss": 3.2288, "step": 58890 }, { "epoch": 1.00755712, "grad_norm": 0.8920489549636841, "learning_rate": 1.8914439024016362e-05, "loss": 3.3363, "step": 58900 }, { "epoch": 1.00758272, "grad_norm": 0.8156958818435669, "learning_rate": 1.8914074098413483e-05, "loss": 3.2359, "step": 58910 }, { "epoch": 1.00760832, "grad_norm": 0.7800896167755127, "learning_rate": 1.89137091150052e-05, "loss": 3.276, "step": 58920 }, { "epoch": 1.00763392, "grad_norm": 0.7701907753944397, "learning_rate": 1.891334407379388e-05, "loss": 3.2469, "step": 58930 }, { "epoch": 1.00765952, "grad_norm": 2.830580472946167, "learning_rate": 1.89129789747819e-05, "loss": 3.2887, "step": 58940 }, { "epoch": 1.00768512, "grad_norm": 0.7896952033042908, "learning_rate": 1.891261381797161e-05, "loss": 3.3233, "step": 58950 }, { "epoch": 1.00771072, "grad_norm": 0.922228217124939, "learning_rate": 1.8912248603365397e-05, "loss": 3.256, "step": 58960 }, { "epoch": 1.00773632, "grad_norm": 0.7548903226852417, "learning_rate": 1.8911883330965616e-05, "loss": 3.1345, "step": 58970 }, { "epoch": 1.00776192, "grad_norm": 0.725432813167572, "learning_rate": 1.891151800077464e-05, "loss": 3.2262, "step": 58980 }, { "epoch": 1.00778752, "grad_norm": 0.8040938377380371, "learning_rate": 1.8911152612794844e-05, "loss": 3.305, "step": 58990 }, { "epoch": 1.00781312, "grad_norm": 0.8262082934379578, "learning_rate": 1.8910787167028585e-05, "loss": 3.1508, "step": 59000 }, { "epoch": 1.00783872, "grad_norm": 0.9179583787918091, "learning_rate": 1.8910421663478243e-05, "loss": 3.184, "step": 59010 }, { "epoch": 1.00786432, "grad_norm": 0.8358368277549744, "learning_rate": 1.8910056102146183e-05, "loss": 3.1236, "step": 59020 }, { "epoch": 1.00788992, "grad_norm": 0.8814560770988464, "learning_rate": 1.8909690483034777e-05, "loss": 3.0942, "step": 59030 }, { "epoch": 1.00791552, "grad_norm": 0.8042178153991699, "learning_rate": 1.8909324806146397e-05, "loss": 3.2236, "step": 59040 }, { "epoch": 1.00794112, "grad_norm": 0.8311001062393188, "learning_rate": 1.8908959071483414e-05, "loss": 3.2742, "step": 59050 }, { "epoch": 1.00796672, "grad_norm": 0.8265188336372375, "learning_rate": 1.8908593279048196e-05, "loss": 3.2502, "step": 59060 }, { "epoch": 1.00799232, "grad_norm": 0.8020178079605103, "learning_rate": 1.890822742884312e-05, "loss": 3.275, "step": 59070 }, { "epoch": 1.00801792, "grad_norm": 0.7647243142127991, "learning_rate": 1.890786152087056e-05, "loss": 3.1182, "step": 59080 }, { "epoch": 1.00804352, "grad_norm": 0.8190531730651855, "learning_rate": 1.890749555513288e-05, "loss": 3.3345, "step": 59090 }, { "epoch": 1.00806912, "grad_norm": 0.7936152219772339, "learning_rate": 1.8907129531632464e-05, "loss": 3.4262, "step": 59100 }, { "epoch": 1.0080947199999999, "grad_norm": 0.6761417388916016, "learning_rate": 1.8906763450371676e-05, "loss": 3.1721, "step": 59110 }, { "epoch": 1.00812032, "grad_norm": 0.8663046360015869, "learning_rate": 1.8906397311352897e-05, "loss": 3.257, "step": 59120 }, { "epoch": 1.00814592, "grad_norm": 0.8223068118095398, "learning_rate": 1.8906031114578492e-05, "loss": 3.2077, "step": 59130 }, { "epoch": 1.00817152, "grad_norm": 0.6566177010536194, "learning_rate": 1.890566486005085e-05, "loss": 3.251, "step": 59140 }, { "epoch": 1.00819712, "grad_norm": 0.7160977125167847, "learning_rate": 1.8905298547772336e-05, "loss": 3.332, "step": 59150 }, { "epoch": 1.00822272, "grad_norm": 0.7575315833091736, "learning_rate": 1.8904932177745324e-05, "loss": 3.1808, "step": 59160 }, { "epoch": 1.00824832, "grad_norm": 0.9476910829544067, "learning_rate": 1.89045657499722e-05, "loss": 3.2747, "step": 59170 }, { "epoch": 1.00827392, "grad_norm": 0.8288655877113342, "learning_rate": 1.890419926445533e-05, "loss": 3.4858, "step": 59180 }, { "epoch": 1.00829952, "grad_norm": 0.9016755223274231, "learning_rate": 1.8903832721197097e-05, "loss": 3.2663, "step": 59190 }, { "epoch": 1.00832512, "grad_norm": 0.8890930414199829, "learning_rate": 1.890346612019987e-05, "loss": 3.4183, "step": 59200 }, { "epoch": 1.00835072, "grad_norm": 0.8487213253974915, "learning_rate": 1.8903099461466032e-05, "loss": 3.2871, "step": 59210 }, { "epoch": 1.00837632, "grad_norm": 0.7990339398384094, "learning_rate": 1.8902732744997965e-05, "loss": 3.3362, "step": 59220 }, { "epoch": 1.00840192, "grad_norm": 1.1640819311141968, "learning_rate": 1.890236597079804e-05, "loss": 3.4105, "step": 59230 }, { "epoch": 1.00842752, "grad_norm": 0.8031772971153259, "learning_rate": 1.8901999138868633e-05, "loss": 3.1679, "step": 59240 }, { "epoch": 1.00845312, "grad_norm": 0.821031928062439, "learning_rate": 1.8901632249212133e-05, "loss": 3.3442, "step": 59250 }, { "epoch": 1.00847872, "grad_norm": 1.155992865562439, "learning_rate": 1.8901265301830913e-05, "loss": 3.0673, "step": 59260 }, { "epoch": 1.00850432, "grad_norm": 0.7812164425849915, "learning_rate": 1.8900898296727353e-05, "loss": 3.1567, "step": 59270 }, { "epoch": 1.00852992, "grad_norm": 0.8389374017715454, "learning_rate": 1.890053123390383e-05, "loss": 3.3529, "step": 59280 }, { "epoch": 1.00855552, "grad_norm": 0.7883920073509216, "learning_rate": 1.890016411336273e-05, "loss": 3.4452, "step": 59290 }, { "epoch": 1.00858112, "grad_norm": 0.8196317553520203, "learning_rate": 1.889979693510643e-05, "loss": 3.0259, "step": 59300 }, { "epoch": 1.00860672, "grad_norm": 0.7887765765190125, "learning_rate": 1.8899429699137315e-05, "loss": 3.2913, "step": 59310 }, { "epoch": 1.00863232, "grad_norm": 0.7189851999282837, "learning_rate": 1.889906240545776e-05, "loss": 3.11, "step": 59320 }, { "epoch": 1.00865792, "grad_norm": 0.773853600025177, "learning_rate": 1.8898695054070154e-05, "loss": 3.2927, "step": 59330 }, { "epoch": 1.00868352, "grad_norm": 0.8073974251747131, "learning_rate": 1.8898327644976872e-05, "loss": 3.0794, "step": 59340 }, { "epoch": 1.00870912, "grad_norm": 0.732237696647644, "learning_rate": 1.8897960178180304e-05, "loss": 3.2985, "step": 59350 }, { "epoch": 1.00873472, "grad_norm": 0.9626346826553345, "learning_rate": 1.8897592653682825e-05, "loss": 3.3141, "step": 59360 }, { "epoch": 1.00876032, "grad_norm": 0.9535443782806396, "learning_rate": 1.8897225071486824e-05, "loss": 3.2712, "step": 59370 }, { "epoch": 1.00878592, "grad_norm": 0.7697577476501465, "learning_rate": 1.8896857431594685e-05, "loss": 3.1737, "step": 59380 }, { "epoch": 1.00881152, "grad_norm": 0.7710622549057007, "learning_rate": 1.889648973400879e-05, "loss": 3.1342, "step": 59390 }, { "epoch": 1.00883712, "grad_norm": 0.9539125561714172, "learning_rate": 1.8896121978731518e-05, "loss": 3.3323, "step": 59400 }, { "epoch": 1.00886272, "grad_norm": 0.9039134979248047, "learning_rate": 1.8895754165765263e-05, "loss": 3.1863, "step": 59410 }, { "epoch": 1.00888832, "grad_norm": 0.839591383934021, "learning_rate": 1.8895386295112405e-05, "loss": 3.3157, "step": 59420 }, { "epoch": 1.00891392, "grad_norm": 0.7987061738967896, "learning_rate": 1.8895018366775333e-05, "loss": 3.3313, "step": 59430 }, { "epoch": 1.00893952, "grad_norm": 0.89102703332901, "learning_rate": 1.889465038075643e-05, "loss": 3.3946, "step": 59440 }, { "epoch": 1.00896512, "grad_norm": 0.8308173418045044, "learning_rate": 1.8894282337058086e-05, "loss": 3.2313, "step": 59450 }, { "epoch": 1.00899072, "grad_norm": 0.8775300979614258, "learning_rate": 1.8893914235682683e-05, "loss": 3.156, "step": 59460 }, { "epoch": 1.00901632, "grad_norm": 1.2561942338943481, "learning_rate": 1.8893546076632606e-05, "loss": 3.2601, "step": 59470 }, { "epoch": 1.00904192, "grad_norm": 0.8866494297981262, "learning_rate": 1.889317785991025e-05, "loss": 3.305, "step": 59480 }, { "epoch": 1.00906752, "grad_norm": 0.8569106459617615, "learning_rate": 1.8892809585518e-05, "loss": 3.3068, "step": 59490 }, { "epoch": 1.00909312, "grad_norm": 0.7646617293357849, "learning_rate": 1.8892441253458242e-05, "loss": 3.0998, "step": 59500 }, { "epoch": 1.00911872, "grad_norm": 0.8746454119682312, "learning_rate": 1.8892072863733366e-05, "loss": 3.1117, "step": 59510 }, { "epoch": 1.00914432, "grad_norm": 0.7947604060173035, "learning_rate": 1.889170441634576e-05, "loss": 3.3764, "step": 59520 }, { "epoch": 1.00916992, "grad_norm": 0.8177546858787537, "learning_rate": 1.8891335911297814e-05, "loss": 3.1919, "step": 59530 }, { "epoch": 1.00919552, "grad_norm": 0.8523579835891724, "learning_rate": 1.889096734859192e-05, "loss": 3.3286, "step": 59540 }, { "epoch": 1.00922112, "grad_norm": 1.269184947013855, "learning_rate": 1.8890598728230464e-05, "loss": 3.5319, "step": 59550 }, { "epoch": 1.00924672, "grad_norm": 0.7524512410163879, "learning_rate": 1.8890230050215836e-05, "loss": 3.4586, "step": 59560 }, { "epoch": 1.00927232, "grad_norm": 0.7482761740684509, "learning_rate": 1.8889861314550432e-05, "loss": 3.5126, "step": 59570 }, { "epoch": 1.00929792, "grad_norm": 0.776736319065094, "learning_rate": 1.8889492521236636e-05, "loss": 3.3235, "step": 59580 }, { "epoch": 1.00932352, "grad_norm": 0.7040478587150574, "learning_rate": 1.8889123670276847e-05, "loss": 3.3133, "step": 59590 }, { "epoch": 1.00934912, "grad_norm": 0.8829094767570496, "learning_rate": 1.8888754761673455e-05, "loss": 3.1366, "step": 59600 }, { "epoch": 1.00937472, "grad_norm": 0.9505209922790527, "learning_rate": 1.8888385795428847e-05, "loss": 3.1186, "step": 59610 }, { "epoch": 1.00940032, "grad_norm": 0.9206326007843018, "learning_rate": 1.888801677154542e-05, "loss": 3.233, "step": 59620 }, { "epoch": 1.00942592, "grad_norm": 1.0773869752883911, "learning_rate": 1.888764769002557e-05, "loss": 3.097, "step": 59630 }, { "epoch": 1.00945152, "grad_norm": 1.4601722955703735, "learning_rate": 1.8887278550871683e-05, "loss": 3.1732, "step": 59640 }, { "epoch": 1.00947712, "grad_norm": 0.9982343912124634, "learning_rate": 1.8886909354086157e-05, "loss": 3.4853, "step": 59650 }, { "epoch": 1.00950272, "grad_norm": 0.7567480206489563, "learning_rate": 1.8886540099671386e-05, "loss": 3.3509, "step": 59660 }, { "epoch": 1.00952832, "grad_norm": 0.7430732250213623, "learning_rate": 1.8886170787629764e-05, "loss": 3.2324, "step": 59670 }, { "epoch": 1.00955392, "grad_norm": 0.7804284691810608, "learning_rate": 1.888580141796369e-05, "loss": 3.3771, "step": 59680 }, { "epoch": 1.00957952, "grad_norm": 0.8549275994300842, "learning_rate": 1.8885431990675548e-05, "loss": 3.0646, "step": 59690 }, { "epoch": 1.00960512, "grad_norm": 0.9578227996826172, "learning_rate": 1.8885062505767746e-05, "loss": 3.1166, "step": 59700 }, { "epoch": 1.00963072, "grad_norm": 0.716378927230835, "learning_rate": 1.8884692963242673e-05, "loss": 3.1853, "step": 59710 }, { "epoch": 1.00965632, "grad_norm": 0.8634372353553772, "learning_rate": 1.888432336310273e-05, "loss": 3.1175, "step": 59720 }, { "epoch": 1.00968192, "grad_norm": 0.9020065665245056, "learning_rate": 1.8883953705350307e-05, "loss": 3.2854, "step": 59730 }, { "epoch": 1.00970752, "grad_norm": 0.7892331480979919, "learning_rate": 1.888358398998781e-05, "loss": 3.4408, "step": 59740 }, { "epoch": 1.00973312, "grad_norm": 0.7958361506462097, "learning_rate": 1.8883214217017628e-05, "loss": 3.2668, "step": 59750 }, { "epoch": 1.00975872, "grad_norm": 0.7593075037002563, "learning_rate": 1.8882844386442162e-05, "loss": 3.3378, "step": 59760 }, { "epoch": 1.00978432, "grad_norm": 0.765743613243103, "learning_rate": 1.8882474498263814e-05, "loss": 3.2254, "step": 59770 }, { "epoch": 1.00980992, "grad_norm": 0.8359093070030212, "learning_rate": 1.8882104552484983e-05, "loss": 3.3916, "step": 59780 }, { "epoch": 1.00983552, "grad_norm": 0.7843884825706482, "learning_rate": 1.8881734549108057e-05, "loss": 3.1426, "step": 59790 }, { "epoch": 1.00986112, "grad_norm": 0.8925080895423889, "learning_rate": 1.8881364488135448e-05, "loss": 3.2419, "step": 59800 }, { "epoch": 1.00988672, "grad_norm": 1.3531800508499146, "learning_rate": 1.8880994369569546e-05, "loss": 3.2784, "step": 59810 }, { "epoch": 1.00991232, "grad_norm": 0.9485967755317688, "learning_rate": 1.888062419341276e-05, "loss": 3.2877, "step": 59820 }, { "epoch": 1.00993792, "grad_norm": 0.7721638083457947, "learning_rate": 1.8880253959667486e-05, "loss": 3.3554, "step": 59830 }, { "epoch": 1.00996352, "grad_norm": 0.7328392863273621, "learning_rate": 1.8879883668336126e-05, "loss": 3.2761, "step": 59840 }, { "epoch": 1.00998912, "grad_norm": 0.9777785539627075, "learning_rate": 1.8879513319421078e-05, "loss": 3.1999, "step": 59850 }, { "epoch": 1.01001472, "grad_norm": 0.7745660543441772, "learning_rate": 1.8879142912924747e-05, "loss": 3.2473, "step": 59860 }, { "epoch": 1.01004032, "grad_norm": 0.7482689619064331, "learning_rate": 1.887877244884953e-05, "loss": 3.3075, "step": 59870 }, { "epoch": 1.01006592, "grad_norm": 0.901996910572052, "learning_rate": 1.8878401927197838e-05, "loss": 3.1032, "step": 59880 }, { "epoch": 1.01009152, "grad_norm": 0.8352453708648682, "learning_rate": 1.8878031347972068e-05, "loss": 3.3276, "step": 59890 }, { "epoch": 1.01011712, "grad_norm": 0.8599341511726379, "learning_rate": 1.8877660711174623e-05, "loss": 3.3324, "step": 59900 }, { "epoch": 1.01014272, "grad_norm": 0.8980081081390381, "learning_rate": 1.887729001680791e-05, "loss": 3.4356, "step": 59910 }, { "epoch": 1.01016832, "grad_norm": 0.8331758975982666, "learning_rate": 1.887691926487433e-05, "loss": 3.1995, "step": 59920 }, { "epoch": 1.01019392, "grad_norm": 1.5328385829925537, "learning_rate": 1.8876548455376286e-05, "loss": 3.3344, "step": 59930 }, { "epoch": 1.01021952, "grad_norm": 0.9183111786842346, "learning_rate": 1.8876177588316187e-05, "loss": 3.3023, "step": 59940 }, { "epoch": 1.01024512, "grad_norm": 0.8521295189857483, "learning_rate": 1.8875806663696433e-05, "loss": 3.5388, "step": 59950 }, { "epoch": 1.01027072, "grad_norm": 0.8836315870285034, "learning_rate": 1.8875435681519434e-05, "loss": 3.2263, "step": 59960 }, { "epoch": 1.01029632, "grad_norm": 1.101244568824768, "learning_rate": 1.8875064641787594e-05, "loss": 3.1821, "step": 59970 }, { "epoch": 1.01032192, "grad_norm": 1.0417500734329224, "learning_rate": 1.8874693544503314e-05, "loss": 3.5702, "step": 59980 }, { "epoch": 1.01034752, "grad_norm": 0.96575528383255, "learning_rate": 1.8874322389669007e-05, "loss": 3.4983, "step": 59990 }, { "epoch": 1.01037312, "grad_norm": 0.7812644839286804, "learning_rate": 1.887395117728708e-05, "loss": 3.4627, "step": 60000 }, { "epoch": 1.01039872, "grad_norm": 0.8645031452178955, "learning_rate": 1.8873579907359936e-05, "loss": 3.18, "step": 60010 }, { "epoch": 1.01042432, "grad_norm": 0.83388751745224, "learning_rate": 1.8873208579889987e-05, "loss": 3.2266, "step": 60020 }, { "epoch": 1.01044992, "grad_norm": 0.7722791433334351, "learning_rate": 1.8872837194879633e-05, "loss": 3.3801, "step": 60030 }, { "epoch": 1.01047552, "grad_norm": 0.9700167179107666, "learning_rate": 1.8872465752331293e-05, "loss": 3.2338, "step": 60040 }, { "epoch": 1.01050112, "grad_norm": 0.7954868078231812, "learning_rate": 1.887209425224737e-05, "loss": 3.1854, "step": 60050 }, { "epoch": 1.01052672, "grad_norm": 0.7376927733421326, "learning_rate": 1.887172269463027e-05, "loss": 3.3256, "step": 60060 }, { "epoch": 1.01055232, "grad_norm": 0.7552865147590637, "learning_rate": 1.887135107948241e-05, "loss": 3.3208, "step": 60070 }, { "epoch": 1.01057792, "grad_norm": 1.0204321146011353, "learning_rate": 1.8870979406806196e-05, "loss": 3.309, "step": 60080 }, { "epoch": 1.01060352, "grad_norm": 0.7852098941802979, "learning_rate": 1.8870607676604034e-05, "loss": 3.1402, "step": 60090 }, { "epoch": 1.01062912, "grad_norm": 1.8295572996139526, "learning_rate": 1.8870235888878343e-05, "loss": 3.3131, "step": 60100 }, { "epoch": 1.01065472, "grad_norm": 0.99778813123703, "learning_rate": 1.886986404363153e-05, "loss": 4.0295, "step": 60110 }, { "epoch": 1.01068032, "grad_norm": 0.8546295166015625, "learning_rate": 1.8869492140866005e-05, "loss": 3.2581, "step": 60120 }, { "epoch": 1.01070592, "grad_norm": 1.4772453308105469, "learning_rate": 1.886912018058418e-05, "loss": 3.4579, "step": 60130 }, { "epoch": 1.01073152, "grad_norm": 2.057417869567871, "learning_rate": 1.8868748162788466e-05, "loss": 3.5766, "step": 60140 }, { "epoch": 1.01075712, "grad_norm": 0.7994017601013184, "learning_rate": 1.886837608748128e-05, "loss": 3.4768, "step": 60150 }, { "epoch": 1.01078272, "grad_norm": 1.0848783254623413, "learning_rate": 1.8868003954665032e-05, "loss": 3.0846, "step": 60160 }, { "epoch": 1.01080832, "grad_norm": 2.509446620941162, "learning_rate": 1.8867631764342135e-05, "loss": 3.4601, "step": 60170 }, { "epoch": 1.01083392, "grad_norm": 1.143480896949768, "learning_rate": 1.8867259516515002e-05, "loss": 3.5178, "step": 60180 }, { "epoch": 1.01085952, "grad_norm": 0.7631008625030518, "learning_rate": 1.8866887211186047e-05, "loss": 3.2128, "step": 60190 }, { "epoch": 1.01088512, "grad_norm": 0.8273354768753052, "learning_rate": 1.8866514848357688e-05, "loss": 3.3038, "step": 60200 }, { "epoch": 1.01091072, "grad_norm": 1.6989818811416626, "learning_rate": 1.886614242803233e-05, "loss": 3.8486, "step": 60210 }, { "epoch": 1.0109363199999999, "grad_norm": 0.8169388771057129, "learning_rate": 1.88657699502124e-05, "loss": 3.1144, "step": 60220 }, { "epoch": 1.01096192, "grad_norm": 0.7640743255615234, "learning_rate": 1.8865397414900308e-05, "loss": 3.2316, "step": 60230 }, { "epoch": 1.01098752, "grad_norm": 0.7896301746368408, "learning_rate": 1.886502482209847e-05, "loss": 3.297, "step": 60240 }, { "epoch": 1.01101312, "grad_norm": 0.8512036204338074, "learning_rate": 1.88646521718093e-05, "loss": 3.2797, "step": 60250 }, { "epoch": 1.01103872, "grad_norm": 0.6918354034423828, "learning_rate": 1.8864279464035217e-05, "loss": 3.1587, "step": 60260 }, { "epoch": 1.01106432, "grad_norm": 0.7849318385124207, "learning_rate": 1.886390669877864e-05, "loss": 3.3153, "step": 60270 }, { "epoch": 1.01108992, "grad_norm": 0.8240916132926941, "learning_rate": 1.8863533876041983e-05, "loss": 3.474, "step": 60280 }, { "epoch": 1.01111552, "grad_norm": 0.6507390141487122, "learning_rate": 1.8863160995827664e-05, "loss": 3.1114, "step": 60290 }, { "epoch": 1.01114112, "grad_norm": 0.7815951704978943, "learning_rate": 1.88627880581381e-05, "loss": 3.1635, "step": 60300 }, { "epoch": 1.01116672, "grad_norm": 1.0207829475402832, "learning_rate": 1.8862415062975712e-05, "loss": 3.1993, "step": 60310 }, { "epoch": 1.01119232, "grad_norm": 0.864100992679596, "learning_rate": 1.8862042010342918e-05, "loss": 3.3463, "step": 60320 }, { "epoch": 1.01121792, "grad_norm": 0.7916752099990845, "learning_rate": 1.8861668900242136e-05, "loss": 3.1814, "step": 60330 }, { "epoch": 1.01124352, "grad_norm": 1.357880711555481, "learning_rate": 1.8861295732675786e-05, "loss": 3.3226, "step": 60340 }, { "epoch": 1.01126912, "grad_norm": 0.8208373188972473, "learning_rate": 1.8860922507646288e-05, "loss": 3.2429, "step": 60350 }, { "epoch": 1.01129472, "grad_norm": 0.7387611269950867, "learning_rate": 1.8860549225156063e-05, "loss": 3.0362, "step": 60360 }, { "epoch": 1.01132032, "grad_norm": 2.3313992023468018, "learning_rate": 1.8860175885207534e-05, "loss": 3.159, "step": 60370 }, { "epoch": 1.01134592, "grad_norm": 0.7200655341148376, "learning_rate": 1.8859802487803117e-05, "loss": 3.0408, "step": 60380 }, { "epoch": 1.01137152, "grad_norm": 0.9528719186782837, "learning_rate": 1.8859429032945235e-05, "loss": 3.2774, "step": 60390 }, { "epoch": 1.01139712, "grad_norm": 0.9682828187942505, "learning_rate": 1.885905552063631e-05, "loss": 3.2418, "step": 60400 }, { "epoch": 1.01142272, "grad_norm": 0.8180437088012695, "learning_rate": 1.8858681950878764e-05, "loss": 3.1444, "step": 60410 }, { "epoch": 1.01144832, "grad_norm": 0.8130839467048645, "learning_rate": 1.885830832367502e-05, "loss": 3.5462, "step": 60420 }, { "epoch": 1.01147392, "grad_norm": 0.8483498096466064, "learning_rate": 1.88579346390275e-05, "loss": 3.3979, "step": 60430 }, { "epoch": 1.01149952, "grad_norm": 1.6182631254196167, "learning_rate": 1.8857560896938632e-05, "loss": 3.3503, "step": 60440 }, { "epoch": 1.01152512, "grad_norm": 0.8183635473251343, "learning_rate": 1.8857187097410833e-05, "loss": 3.2337, "step": 60450 }, { "epoch": 1.01155072, "grad_norm": 0.9367156028747559, "learning_rate": 1.885681324044653e-05, "loss": 3.4279, "step": 60460 }, { "epoch": 1.01157632, "grad_norm": 0.8754943609237671, "learning_rate": 1.885643932604815e-05, "loss": 3.3355, "step": 60470 }, { "epoch": 1.01160192, "grad_norm": 1.2628123760223389, "learning_rate": 1.8856065354218108e-05, "loss": 3.1805, "step": 60480 }, { "epoch": 1.01162752, "grad_norm": 0.8268959522247314, "learning_rate": 1.885569132495884e-05, "loss": 3.2672, "step": 60490 }, { "epoch": 1.01165312, "grad_norm": 0.7850597500801086, "learning_rate": 1.885531723827277e-05, "loss": 3.3222, "step": 60500 }, { "epoch": 1.01167872, "grad_norm": 0.9540174603462219, "learning_rate": 1.8854943094162314e-05, "loss": 3.2542, "step": 60510 }, { "epoch": 1.01170432, "grad_norm": 0.7878175973892212, "learning_rate": 1.885456889262991e-05, "loss": 3.3375, "step": 60520 }, { "epoch": 1.01172992, "grad_norm": 0.8429956436157227, "learning_rate": 1.885419463367798e-05, "loss": 3.2309, "step": 60530 }, { "epoch": 1.01175552, "grad_norm": 0.7510553598403931, "learning_rate": 1.885382031730895e-05, "loss": 3.4917, "step": 60540 }, { "epoch": 1.01178112, "grad_norm": 0.8120067715644836, "learning_rate": 1.885344594352525e-05, "loss": 3.2171, "step": 60550 }, { "epoch": 1.01180672, "grad_norm": 0.7264444231987, "learning_rate": 1.8853071512329303e-05, "loss": 3.3077, "step": 60560 }, { "epoch": 1.01183232, "grad_norm": 0.7954425811767578, "learning_rate": 1.885269702372354e-05, "loss": 3.3358, "step": 60570 }, { "epoch": 1.01185792, "grad_norm": 0.8115789890289307, "learning_rate": 1.8852322477710393e-05, "loss": 3.2159, "step": 60580 }, { "epoch": 1.01188352, "grad_norm": 0.7290059924125671, "learning_rate": 1.8851947874292287e-05, "loss": 3.173, "step": 60590 }, { "epoch": 1.01190912, "grad_norm": 0.7839084267616272, "learning_rate": 1.885157321347165e-05, "loss": 2.968, "step": 60600 }, { "epoch": 1.01193472, "grad_norm": 0.8391438722610474, "learning_rate": 1.8851198495250913e-05, "loss": 3.2871, "step": 60610 }, { "epoch": 1.01196032, "grad_norm": 0.8287951350212097, "learning_rate": 1.8850823719632506e-05, "loss": 3.158, "step": 60620 }, { "epoch": 1.01198592, "grad_norm": 0.7625424861907959, "learning_rate": 1.8850448886618862e-05, "loss": 3.2399, "step": 60630 }, { "epoch": 1.01201152, "grad_norm": 0.8382084965705872, "learning_rate": 1.8850073996212407e-05, "loss": 3.0996, "step": 60640 }, { "epoch": 1.01203712, "grad_norm": 0.8362663984298706, "learning_rate": 1.8849699048415574e-05, "loss": 3.0926, "step": 60650 }, { "epoch": 1.01206272, "grad_norm": 0.7830787301063538, "learning_rate": 1.8849324043230795e-05, "loss": 3.3326, "step": 60660 }, { "epoch": 1.01208832, "grad_norm": 0.800214946269989, "learning_rate": 1.8848948980660502e-05, "loss": 3.1854, "step": 60670 }, { "epoch": 1.01211392, "grad_norm": 0.7577104568481445, "learning_rate": 1.8848573860707126e-05, "loss": 3.0455, "step": 60680 }, { "epoch": 1.01213952, "grad_norm": 0.9738046526908875, "learning_rate": 1.8848198683373102e-05, "loss": 3.4875, "step": 60690 }, { "epoch": 1.01216512, "grad_norm": 0.9444986581802368, "learning_rate": 1.8847823448660856e-05, "loss": 3.0701, "step": 60700 }, { "epoch": 1.01219072, "grad_norm": 0.8125154972076416, "learning_rate": 1.884744815657283e-05, "loss": 3.1848, "step": 60710 }, { "epoch": 1.01221632, "grad_norm": 1.072321891784668, "learning_rate": 1.8847072807111452e-05, "loss": 3.3813, "step": 60720 }, { "epoch": 1.01224192, "grad_norm": 0.9091566205024719, "learning_rate": 1.884669740027916e-05, "loss": 3.1281, "step": 60730 }, { "epoch": 1.01226752, "grad_norm": 0.7663355469703674, "learning_rate": 1.8846321936078387e-05, "loss": 3.2309, "step": 60740 }, { "epoch": 1.01229312, "grad_norm": 0.7557840943336487, "learning_rate": 1.884594641451157e-05, "loss": 3.0128, "step": 60750 }, { "epoch": 1.01231872, "grad_norm": 1.7470595836639404, "learning_rate": 1.884557083558114e-05, "loss": 3.0422, "step": 60760 }, { "epoch": 1.01234432, "grad_norm": 0.9094883799552917, "learning_rate": 1.884519519928953e-05, "loss": 2.8975, "step": 60770 }, { "epoch": 1.01236992, "grad_norm": 0.799728512763977, "learning_rate": 1.884481950563918e-05, "loss": 3.3403, "step": 60780 }, { "epoch": 1.01239552, "grad_norm": 0.8274520635604858, "learning_rate": 1.8844443754632528e-05, "loss": 3.203, "step": 60790 }, { "epoch": 1.01242112, "grad_norm": 0.8142240047454834, "learning_rate": 1.884406794627201e-05, "loss": 3.1684, "step": 60800 }, { "epoch": 1.01244672, "grad_norm": 0.7169206142425537, "learning_rate": 1.8843692080560062e-05, "loss": 3.2694, "step": 60810 }, { "epoch": 1.01247232, "grad_norm": 0.8672749400138855, "learning_rate": 1.8843316157499122e-05, "loss": 3.2803, "step": 60820 }, { "epoch": 1.01249792, "grad_norm": 0.7955463528633118, "learning_rate": 1.8842940177091623e-05, "loss": 3.1988, "step": 60830 }, { "epoch": 1.01252352, "grad_norm": 0.7547717690467834, "learning_rate": 1.8842564139340013e-05, "loss": 3.3673, "step": 60840 }, { "epoch": 1.01254912, "grad_norm": 0.7705186009407043, "learning_rate": 1.884218804424672e-05, "loss": 3.1388, "step": 60850 }, { "epoch": 1.01257472, "grad_norm": 0.7570235729217529, "learning_rate": 1.8841811891814186e-05, "loss": 3.3421, "step": 60860 }, { "epoch": 1.01260032, "grad_norm": 0.8511719703674316, "learning_rate": 1.8841435682044856e-05, "loss": 3.2474, "step": 60870 }, { "epoch": 1.01262592, "grad_norm": 0.846892774105072, "learning_rate": 1.8841059414941162e-05, "loss": 3.3317, "step": 60880 }, { "epoch": 1.01265152, "grad_norm": 0.8828271627426147, "learning_rate": 1.884068309050555e-05, "loss": 3.4479, "step": 60890 }, { "epoch": 1.01267712, "grad_norm": 0.9626597166061401, "learning_rate": 1.8840306708740456e-05, "loss": 3.3016, "step": 60900 }, { "epoch": 1.01270272, "grad_norm": 0.8843138813972473, "learning_rate": 1.8839930269648322e-05, "loss": 3.2837, "step": 60910 }, { "epoch": 1.01272832, "grad_norm": 0.7522134780883789, "learning_rate": 1.8839553773231593e-05, "loss": 3.2958, "step": 60920 }, { "epoch": 1.01275392, "grad_norm": 0.9270429611206055, "learning_rate": 1.8839177219492706e-05, "loss": 3.2552, "step": 60930 }, { "epoch": 1.01277952, "grad_norm": 0.8626990914344788, "learning_rate": 1.88388006084341e-05, "loss": 3.2147, "step": 60940 }, { "epoch": 1.01280512, "grad_norm": 0.7736709713935852, "learning_rate": 1.8838423940058224e-05, "loss": 3.2921, "step": 60950 }, { "epoch": 1.01283072, "grad_norm": 0.8143059611320496, "learning_rate": 1.883804721436752e-05, "loss": 3.4198, "step": 60960 }, { "epoch": 1.01285632, "grad_norm": 1.272346019744873, "learning_rate": 1.8837670431364422e-05, "loss": 3.2952, "step": 60970 }, { "epoch": 1.01288192, "grad_norm": 1.2084827423095703, "learning_rate": 1.883729359105139e-05, "loss": 3.6667, "step": 60980 }, { "epoch": 1.01290752, "grad_norm": 0.7941725254058838, "learning_rate": 1.883691669343085e-05, "loss": 3.3109, "step": 60990 }, { "epoch": 1.01293312, "grad_norm": 0.6962313652038574, "learning_rate": 1.8836539738505253e-05, "loss": 3.0537, "step": 61000 }, { "epoch": 1.01295872, "grad_norm": 0.7802464962005615, "learning_rate": 1.883616272627705e-05, "loss": 3.0897, "step": 61010 }, { "epoch": 1.01298432, "grad_norm": 0.8899527788162231, "learning_rate": 1.8835785656748675e-05, "loss": 3.215, "step": 61020 }, { "epoch": 1.01300992, "grad_norm": 1.3363429307937622, "learning_rate": 1.8835408529922583e-05, "loss": 3.2929, "step": 61030 }, { "epoch": 1.01303552, "grad_norm": 0.7169592976570129, "learning_rate": 1.883503134580121e-05, "loss": 3.6413, "step": 61040 }, { "epoch": 1.01306112, "grad_norm": 0.8105068802833557, "learning_rate": 1.883465410438701e-05, "loss": 3.097, "step": 61050 }, { "epoch": 1.01308672, "grad_norm": 0.8287206888198853, "learning_rate": 1.8834276805682426e-05, "loss": 3.4626, "step": 61060 }, { "epoch": 1.01311232, "grad_norm": 0.8700898885726929, "learning_rate": 1.8833899449689903e-05, "loss": 3.3423, "step": 61070 }, { "epoch": 1.01313792, "grad_norm": 0.8197614550590515, "learning_rate": 1.8833522036411892e-05, "loss": 3.2542, "step": 61080 }, { "epoch": 1.01316352, "grad_norm": 0.7566291093826294, "learning_rate": 1.8833144565850838e-05, "loss": 3.2086, "step": 61090 }, { "epoch": 1.01318912, "grad_norm": 0.866557776927948, "learning_rate": 1.8832767038009184e-05, "loss": 3.1663, "step": 61100 }, { "epoch": 1.01321472, "grad_norm": 0.9603181481361389, "learning_rate": 1.883238945288939e-05, "loss": 3.3874, "step": 61110 }, { "epoch": 1.01324032, "grad_norm": 0.7386520504951477, "learning_rate": 1.8832011810493896e-05, "loss": 3.139, "step": 61120 }, { "epoch": 1.01326592, "grad_norm": 0.8316230773925781, "learning_rate": 1.883163411082515e-05, "loss": 3.5692, "step": 61130 }, { "epoch": 1.01329152, "grad_norm": 0.7682402729988098, "learning_rate": 1.883125635388561e-05, "loss": 3.3093, "step": 61140 }, { "epoch": 1.01331712, "grad_norm": 0.946972668170929, "learning_rate": 1.8830878539677714e-05, "loss": 3.1638, "step": 61150 }, { "epoch": 1.01334272, "grad_norm": 0.8382204174995422, "learning_rate": 1.8830500668203918e-05, "loss": 3.272, "step": 61160 }, { "epoch": 1.01336832, "grad_norm": 0.9955329895019531, "learning_rate": 1.8830122739466673e-05, "loss": 3.222, "step": 61170 }, { "epoch": 1.01339392, "grad_norm": 0.7898107767105103, "learning_rate": 1.8829744753468428e-05, "loss": 3.347, "step": 61180 }, { "epoch": 1.01341952, "grad_norm": 0.6598604917526245, "learning_rate": 1.8829366710211637e-05, "loss": 3.1831, "step": 61190 }, { "epoch": 1.01344512, "grad_norm": 0.8375228643417358, "learning_rate": 1.8828988609698746e-05, "loss": 3.4089, "step": 61200 }, { "epoch": 1.01347072, "grad_norm": 1.0298398733139038, "learning_rate": 1.8828610451932215e-05, "loss": 3.3984, "step": 61210 }, { "epoch": 1.01349632, "grad_norm": 1.0521379709243774, "learning_rate": 1.8828232236914486e-05, "loss": 3.1927, "step": 61220 }, { "epoch": 1.01352192, "grad_norm": 0.8483274579048157, "learning_rate": 1.8827853964648023e-05, "loss": 3.254, "step": 61230 }, { "epoch": 1.01354752, "grad_norm": 0.8890972137451172, "learning_rate": 1.882747563513527e-05, "loss": 3.2612, "step": 61240 }, { "epoch": 1.01357312, "grad_norm": 0.8232763409614563, "learning_rate": 1.8827097248378686e-05, "loss": 3.2846, "step": 61250 }, { "epoch": 1.01359872, "grad_norm": 0.8666127920150757, "learning_rate": 1.8826718804380718e-05, "loss": 3.9978, "step": 61260 }, { "epoch": 1.01362432, "grad_norm": 0.8099800944328308, "learning_rate": 1.8826340303143828e-05, "loss": 3.2535, "step": 61270 }, { "epoch": 1.01364992, "grad_norm": 0.9845234751701355, "learning_rate": 1.8825961744670468e-05, "loss": 3.1656, "step": 61280 }, { "epoch": 1.01367552, "grad_norm": 1.0766185522079468, "learning_rate": 1.882558312896309e-05, "loss": 3.3793, "step": 61290 }, { "epoch": 1.01370112, "grad_norm": 1.02122163772583, "learning_rate": 1.8825204456024146e-05, "loss": 3.1302, "step": 61300 }, { "epoch": 1.01372672, "grad_norm": 0.8271878361701965, "learning_rate": 1.88248257258561e-05, "loss": 3.1588, "step": 61310 }, { "epoch": 1.01375232, "grad_norm": 0.8037517070770264, "learning_rate": 1.8824446938461407e-05, "loss": 3.2756, "step": 61320 }, { "epoch": 1.01377792, "grad_norm": 1.0186809301376343, "learning_rate": 1.8824068093842523e-05, "loss": 3.1547, "step": 61330 }, { "epoch": 1.01380352, "grad_norm": 0.8187273740768433, "learning_rate": 1.88236891920019e-05, "loss": 3.127, "step": 61340 }, { "epoch": 1.01382912, "grad_norm": 0.7995762228965759, "learning_rate": 1.8823310232941993e-05, "loss": 3.2625, "step": 61350 }, { "epoch": 1.01385472, "grad_norm": 0.9159718751907349, "learning_rate": 1.882293121666527e-05, "loss": 3.3456, "step": 61360 }, { "epoch": 1.01388032, "grad_norm": 0.7905165553092957, "learning_rate": 1.882255214317418e-05, "loss": 3.0923, "step": 61370 }, { "epoch": 1.01390592, "grad_norm": 0.8081979155540466, "learning_rate": 1.8822173012471185e-05, "loss": 3.2895, "step": 61380 }, { "epoch": 1.01393152, "grad_norm": 0.9933127164840698, "learning_rate": 1.882179382455874e-05, "loss": 3.4487, "step": 61390 }, { "epoch": 1.01395712, "grad_norm": 0.8179569840431213, "learning_rate": 1.882141457943931e-05, "loss": 3.2248, "step": 61400 }, { "epoch": 1.01398272, "grad_norm": 0.9746119379997253, "learning_rate": 1.882103527711535e-05, "loss": 3.3253, "step": 61410 }, { "epoch": 1.01400832, "grad_norm": 0.8007475137710571, "learning_rate": 1.8820655917589322e-05, "loss": 3.2852, "step": 61420 }, { "epoch": 1.01403392, "grad_norm": 0.9096148014068604, "learning_rate": 1.882027650086368e-05, "loss": 3.2219, "step": 61430 }, { "epoch": 1.01405952, "grad_norm": 0.7902831435203552, "learning_rate": 1.8819897026940894e-05, "loss": 3.5052, "step": 61440 }, { "epoch": 1.01408512, "grad_norm": 0.8687134385108948, "learning_rate": 1.881951749582342e-05, "loss": 3.261, "step": 61450 }, { "epoch": 1.01411072, "grad_norm": 0.829496443271637, "learning_rate": 1.8819137907513717e-05, "loss": 3.5323, "step": 61460 }, { "epoch": 1.01413632, "grad_norm": 0.8609413504600525, "learning_rate": 1.881875826201425e-05, "loss": 3.3137, "step": 61470 }, { "epoch": 1.01416192, "grad_norm": 0.7818446159362793, "learning_rate": 1.8818378559327476e-05, "loss": 3.2841, "step": 61480 }, { "epoch": 1.01418752, "grad_norm": 0.9144800901412964, "learning_rate": 1.8817998799455862e-05, "loss": 3.3708, "step": 61490 }, { "epoch": 1.01421312, "grad_norm": 0.8640488982200623, "learning_rate": 1.8817618982401867e-05, "loss": 3.1336, "step": 61500 }, { "epoch": 1.01423872, "grad_norm": 0.9304916858673096, "learning_rate": 1.881723910816796e-05, "loss": 3.4018, "step": 61510 }, { "epoch": 1.01426432, "grad_norm": 0.8237600922584534, "learning_rate": 1.88168591767566e-05, "loss": 3.1285, "step": 61520 }, { "epoch": 1.01428992, "grad_norm": 0.9475694894790649, "learning_rate": 1.8816479188170254e-05, "loss": 3.3554, "step": 61530 }, { "epoch": 1.01431552, "grad_norm": 0.9020255208015442, "learning_rate": 1.881609914241138e-05, "loss": 3.2507, "step": 61540 }, { "epoch": 1.01434112, "grad_norm": 0.9559463858604431, "learning_rate": 1.8815719039482447e-05, "loss": 3.485, "step": 61550 }, { "epoch": 1.01436672, "grad_norm": 0.7836704850196838, "learning_rate": 1.881533887938592e-05, "loss": 3.2795, "step": 61560 }, { "epoch": 1.01439232, "grad_norm": 0.8446183800697327, "learning_rate": 1.881495866212426e-05, "loss": 3.202, "step": 61570 }, { "epoch": 1.01441792, "grad_norm": 0.7260257601737976, "learning_rate": 1.881457838769994e-05, "loss": 3.3183, "step": 61580 }, { "epoch": 1.01444352, "grad_norm": 0.9542235136032104, "learning_rate": 1.881419805611542e-05, "loss": 3.3052, "step": 61590 }, { "epoch": 1.01446912, "grad_norm": 0.7629637122154236, "learning_rate": 1.8813817667373165e-05, "loss": 3.2391, "step": 61600 }, { "epoch": 1.01449472, "grad_norm": 0.7718393206596375, "learning_rate": 1.8813437221475653e-05, "loss": 3.235, "step": 61610 }, { "epoch": 1.01452032, "grad_norm": 0.9854479432106018, "learning_rate": 1.8813056718425337e-05, "loss": 2.9521, "step": 61620 }, { "epoch": 1.01454592, "grad_norm": 0.8735564351081848, "learning_rate": 1.881267615822469e-05, "loss": 3.2077, "step": 61630 }, { "epoch": 1.01457152, "grad_norm": 0.7471902966499329, "learning_rate": 1.881229554087618e-05, "loss": 3.2424, "step": 61640 }, { "epoch": 1.01459712, "grad_norm": 0.8782669305801392, "learning_rate": 1.881191486638228e-05, "loss": 3.4747, "step": 61650 }, { "epoch": 1.01462272, "grad_norm": 0.8646526336669922, "learning_rate": 1.881153413474545e-05, "loss": 3.1049, "step": 61660 }, { "epoch": 1.01464832, "grad_norm": 0.7862977385520935, "learning_rate": 1.8811153345968166e-05, "loss": 3.401, "step": 61670 }, { "epoch": 1.01467392, "grad_norm": 0.8694393038749695, "learning_rate": 1.881077250005289e-05, "loss": 3.3977, "step": 61680 }, { "epoch": 1.01469952, "grad_norm": 0.6919945478439331, "learning_rate": 1.8810391597002097e-05, "loss": 3.35, "step": 61690 }, { "epoch": 1.01472512, "grad_norm": 0.8132972121238708, "learning_rate": 1.8810010636818257e-05, "loss": 3.2524, "step": 61700 }, { "epoch": 1.01475072, "grad_norm": 0.892093300819397, "learning_rate": 1.880962961950384e-05, "loss": 3.1192, "step": 61710 }, { "epoch": 1.01477632, "grad_norm": 0.7352201342582703, "learning_rate": 1.8809248545061315e-05, "loss": 3.2493, "step": 61720 }, { "epoch": 1.01480192, "grad_norm": 0.8790928721427917, "learning_rate": 1.880886741349316e-05, "loss": 3.2507, "step": 61730 }, { "epoch": 1.01482752, "grad_norm": 0.8561756014823914, "learning_rate": 1.8808486224801836e-05, "loss": 3.2607, "step": 61740 }, { "epoch": 1.01485312, "grad_norm": 0.8266725540161133, "learning_rate": 1.8808104978989816e-05, "loss": 3.281, "step": 61750 }, { "epoch": 1.01487872, "grad_norm": 0.7446742057800293, "learning_rate": 1.8807723676059583e-05, "loss": 3.1794, "step": 61760 }, { "epoch": 1.01490432, "grad_norm": 0.7549502849578857, "learning_rate": 1.88073423160136e-05, "loss": 3.3006, "step": 61770 }, { "epoch": 1.01492992, "grad_norm": 0.9199627637863159, "learning_rate": 1.8806960898854344e-05, "loss": 3.307, "step": 61780 }, { "epoch": 1.01495552, "grad_norm": 0.8196880221366882, "learning_rate": 1.8806579424584284e-05, "loss": 3.2592, "step": 61790 }, { "epoch": 1.01498112, "grad_norm": 0.951857328414917, "learning_rate": 1.8806197893205897e-05, "loss": 3.2577, "step": 61800 }, { "epoch": 1.01500672, "grad_norm": 1.1333396434783936, "learning_rate": 1.880581630472166e-05, "loss": 3.1224, "step": 61810 }, { "epoch": 1.01503232, "grad_norm": 0.727285623550415, "learning_rate": 1.8805434659134044e-05, "loss": 3.2542, "step": 61820 }, { "epoch": 1.01505792, "grad_norm": 0.843649685382843, "learning_rate": 1.8805052956445524e-05, "loss": 3.4291, "step": 61830 }, { "epoch": 1.01508352, "grad_norm": 0.7616875171661377, "learning_rate": 1.8804671196658577e-05, "loss": 3.2886, "step": 61840 }, { "epoch": 1.01510912, "grad_norm": 0.8390301465988159, "learning_rate": 1.8804289379775674e-05, "loss": 3.2259, "step": 61850 }, { "epoch": 1.01513472, "grad_norm": 0.7791866660118103, "learning_rate": 1.8803907505799298e-05, "loss": 3.291, "step": 61860 }, { "epoch": 1.01516032, "grad_norm": 0.7857422232627869, "learning_rate": 1.880352557473192e-05, "loss": 3.3882, "step": 61870 }, { "epoch": 1.01518592, "grad_norm": 1.1543627977371216, "learning_rate": 1.880314358657602e-05, "loss": 3.5665, "step": 61880 }, { "epoch": 1.01521152, "grad_norm": 1.3189125061035156, "learning_rate": 1.880276154133407e-05, "loss": 3.1985, "step": 61890 }, { "epoch": 1.01523712, "grad_norm": 0.777719259262085, "learning_rate": 1.880237943900855e-05, "loss": 3.4362, "step": 61900 }, { "epoch": 1.01526272, "grad_norm": 0.7346664071083069, "learning_rate": 1.8801997279601943e-05, "loss": 3.3222, "step": 61910 }, { "epoch": 1.01528832, "grad_norm": 0.8469340205192566, "learning_rate": 1.8801615063116722e-05, "loss": 2.9884, "step": 61920 }, { "epoch": 1.01531392, "grad_norm": 0.790084958076477, "learning_rate": 1.8801232789555365e-05, "loss": 3.1946, "step": 61930 }, { "epoch": 1.01533952, "grad_norm": 0.7595766186714172, "learning_rate": 1.8800850458920355e-05, "loss": 3.3239, "step": 61940 }, { "epoch": 1.01536512, "grad_norm": 0.9444803595542908, "learning_rate": 1.8800468071214163e-05, "loss": 3.4353, "step": 61950 }, { "epoch": 1.01539072, "grad_norm": 0.9150661826133728, "learning_rate": 1.8800085626439278e-05, "loss": 3.1609, "step": 61960 }, { "epoch": 1.01541632, "grad_norm": 0.7713222503662109, "learning_rate": 1.8799703124598178e-05, "loss": 3.2828, "step": 61970 }, { "epoch": 1.01544192, "grad_norm": 0.7905176281929016, "learning_rate": 1.8799320565693342e-05, "loss": 3.3067, "step": 61980 }, { "epoch": 1.01546752, "grad_norm": 0.782519519329071, "learning_rate": 1.879893794972725e-05, "loss": 3.145, "step": 61990 }, { "epoch": 2.00001024, "grad_norm": 0.8907527923583984, "learning_rate": 1.8798555276702384e-05, "loss": 3.728, "step": 62000 }, { "epoch": 2.00003584, "grad_norm": 0.9228608012199402, "learning_rate": 1.8798172546621222e-05, "loss": 3.2421, "step": 62010 }, { "epoch": 2.00006144, "grad_norm": 0.7554966807365417, "learning_rate": 1.8797789759486252e-05, "loss": 3.0511, "step": 62020 }, { "epoch": 2.00008704, "grad_norm": 0.7440978288650513, "learning_rate": 1.8797406915299958e-05, "loss": 2.9277, "step": 62030 }, { "epoch": 2.00011264, "grad_norm": 0.7105876207351685, "learning_rate": 1.879702401406481e-05, "loss": 3.0858, "step": 62040 }, { "epoch": 2.00013824, "grad_norm": 0.8009707927703857, "learning_rate": 1.8796641055783304e-05, "loss": 3.1919, "step": 62050 }, { "epoch": 2.00016384, "grad_norm": 0.9070486426353455, "learning_rate": 1.879625804045792e-05, "loss": 2.8964, "step": 62060 }, { "epoch": 2.00018944, "grad_norm": 0.7824996113777161, "learning_rate": 1.8795874968091133e-05, "loss": 3.2248, "step": 62070 }, { "epoch": 2.00021504, "grad_norm": 0.7542387247085571, "learning_rate": 1.8795491838685443e-05, "loss": 3.1785, "step": 62080 }, { "epoch": 2.00024064, "grad_norm": 0.8229371309280396, "learning_rate": 1.8795108652243322e-05, "loss": 3.1036, "step": 62090 }, { "epoch": 2.00026624, "grad_norm": 0.7214091420173645, "learning_rate": 1.879472540876726e-05, "loss": 3.2475, "step": 62100 }, { "epoch": 2.00029184, "grad_norm": 0.9639545679092407, "learning_rate": 1.879434210825974e-05, "loss": 3.3242, "step": 62110 }, { "epoch": 2.00031744, "grad_norm": 0.8349611163139343, "learning_rate": 1.879395875072325e-05, "loss": 3.1507, "step": 62120 }, { "epoch": 2.00034304, "grad_norm": 0.9150758385658264, "learning_rate": 1.8793575336160272e-05, "loss": 3.4204, "step": 62130 }, { "epoch": 2.00036864, "grad_norm": 0.7076103687286377, "learning_rate": 1.87931918645733e-05, "loss": 3.4058, "step": 62140 }, { "epoch": 2.00039424, "grad_norm": 0.7381229996681213, "learning_rate": 1.879280833596481e-05, "loss": 3.1827, "step": 62150 }, { "epoch": 2.00041984, "grad_norm": 0.9040524959564209, "learning_rate": 1.8792424750337298e-05, "loss": 2.951, "step": 62160 }, { "epoch": 2.00044544, "grad_norm": 0.7176164984703064, "learning_rate": 1.8792041107693247e-05, "loss": 3.236, "step": 62170 }, { "epoch": 2.00047104, "grad_norm": 0.8004394769668579, "learning_rate": 1.8791657408035148e-05, "loss": 3.0336, "step": 62180 }, { "epoch": 2.00049664, "grad_norm": 0.7562054991722107, "learning_rate": 1.8791273651365484e-05, "loss": 3.092, "step": 62190 }, { "epoch": 2.00052224, "grad_norm": 0.8293465971946716, "learning_rate": 1.8790889837686748e-05, "loss": 3.3794, "step": 62200 }, { "epoch": 2.00054784, "grad_norm": 0.8174023628234863, "learning_rate": 1.879050596700143e-05, "loss": 3.4759, "step": 62210 }, { "epoch": 2.00057344, "grad_norm": 0.7156073451042175, "learning_rate": 1.8790122039312014e-05, "loss": 2.9229, "step": 62220 }, { "epoch": 2.00059904, "grad_norm": 0.7753705978393555, "learning_rate": 1.8789738054620995e-05, "loss": 2.8299, "step": 62230 }, { "epoch": 2.00062464, "grad_norm": 1.0224988460540771, "learning_rate": 1.8789354012930863e-05, "loss": 3.2707, "step": 62240 }, { "epoch": 2.00065024, "grad_norm": 0.7725933194160461, "learning_rate": 1.87889699142441e-05, "loss": 2.9749, "step": 62250 }, { "epoch": 2.00067584, "grad_norm": 0.7611194849014282, "learning_rate": 1.8788585758563206e-05, "loss": 3.1072, "step": 62260 }, { "epoch": 2.00070144, "grad_norm": 0.8627264499664307, "learning_rate": 1.878820154589067e-05, "loss": 2.9138, "step": 62270 }, { "epoch": 2.00072704, "grad_norm": 0.974887490272522, "learning_rate": 1.878781727622898e-05, "loss": 3.0809, "step": 62280 }, { "epoch": 2.00075264, "grad_norm": 0.7956110239028931, "learning_rate": 1.8787432949580634e-05, "loss": 3.3415, "step": 62290 }, { "epoch": 2.00077824, "grad_norm": 0.7387296557426453, "learning_rate": 1.878704856594812e-05, "loss": 3.0611, "step": 62300 }, { "epoch": 2.00080384, "grad_norm": 0.8932182788848877, "learning_rate": 1.878666412533393e-05, "loss": 3.0049, "step": 62310 }, { "epoch": 2.00082944, "grad_norm": 0.8132904767990112, "learning_rate": 1.8786279627740558e-05, "loss": 3.0707, "step": 62320 }, { "epoch": 2.00085504, "grad_norm": 0.9053895473480225, "learning_rate": 1.87858950731705e-05, "loss": 3.238, "step": 62330 }, { "epoch": 2.00088064, "grad_norm": 0.7944598197937012, "learning_rate": 1.8785510461626243e-05, "loss": 3.2863, "step": 62340 }, { "epoch": 2.00090624, "grad_norm": 0.6676849126815796, "learning_rate": 1.8785125793110287e-05, "loss": 2.8227, "step": 62350 }, { "epoch": 2.00093184, "grad_norm": 0.7919907569885254, "learning_rate": 1.878474106762513e-05, "loss": 3.0464, "step": 62360 }, { "epoch": 2.00095744, "grad_norm": 0.7527852654457092, "learning_rate": 1.8784356285173258e-05, "loss": 3.3441, "step": 62370 }, { "epoch": 2.00098304, "grad_norm": 0.7932202816009521, "learning_rate": 1.878397144575717e-05, "loss": 2.9934, "step": 62380 }, { "epoch": 2.00100864, "grad_norm": 0.743740975856781, "learning_rate": 1.8783586549379362e-05, "loss": 3.2237, "step": 62390 }, { "epoch": 2.00103424, "grad_norm": 0.8161484003067017, "learning_rate": 1.8783201596042332e-05, "loss": 3.2, "step": 62400 }, { "epoch": 2.00105984, "grad_norm": 0.7693639397621155, "learning_rate": 1.8782816585748568e-05, "loss": 3.0955, "step": 62410 }, { "epoch": 2.00108544, "grad_norm": 0.8413819074630737, "learning_rate": 1.878243151850058e-05, "loss": 3.2007, "step": 62420 }, { "epoch": 2.00111104, "grad_norm": 0.823881983757019, "learning_rate": 1.8782046394300854e-05, "loss": 3.191, "step": 62430 }, { "epoch": 2.00113664, "grad_norm": 0.7566589117050171, "learning_rate": 1.878166121315189e-05, "loss": 3.1499, "step": 62440 }, { "epoch": 2.00116224, "grad_norm": 0.7555660605430603, "learning_rate": 1.8781275975056188e-05, "loss": 2.9141, "step": 62450 }, { "epoch": 2.00118784, "grad_norm": 0.766396164894104, "learning_rate": 1.8780890680016245e-05, "loss": 3.0849, "step": 62460 }, { "epoch": 2.00121344, "grad_norm": 0.7087743282318115, "learning_rate": 1.878050532803456e-05, "loss": 2.8561, "step": 62470 }, { "epoch": 2.00123904, "grad_norm": 0.7305258512496948, "learning_rate": 1.878011991911363e-05, "loss": 2.9676, "step": 62480 }, { "epoch": 2.00126464, "grad_norm": 0.877892017364502, "learning_rate": 1.877973445325596e-05, "loss": 3.1051, "step": 62490 }, { "epoch": 2.00129024, "grad_norm": 0.8282713890075684, "learning_rate": 1.8779348930464042e-05, "loss": 3.2135, "step": 62500 }, { "epoch": 2.00131584, "grad_norm": 0.7975897192955017, "learning_rate": 1.877896335074038e-05, "loss": 3.0891, "step": 62510 }, { "epoch": 2.00134144, "grad_norm": 0.7776709794998169, "learning_rate": 1.8778577714087474e-05, "loss": 3.0232, "step": 62520 }, { "epoch": 2.00136704, "grad_norm": 0.7692331671714783, "learning_rate": 1.877819202050782e-05, "loss": 3.1832, "step": 62530 }, { "epoch": 2.00139264, "grad_norm": 0.6942542791366577, "learning_rate": 1.877780627000393e-05, "loss": 3.2066, "step": 62540 }, { "epoch": 2.00141824, "grad_norm": 0.8251858949661255, "learning_rate": 1.8777420462578298e-05, "loss": 3.3446, "step": 62550 }, { "epoch": 2.00144384, "grad_norm": 0.7510063648223877, "learning_rate": 1.8777034598233427e-05, "loss": 3.0661, "step": 62560 }, { "epoch": 2.00146944, "grad_norm": 0.7915732860565186, "learning_rate": 1.877664867697182e-05, "loss": 3.0273, "step": 62570 }, { "epoch": 2.00149504, "grad_norm": 0.805571973323822, "learning_rate": 1.8776262698795976e-05, "loss": 2.8615, "step": 62580 }, { "epoch": 2.00152064, "grad_norm": 0.7338473796844482, "learning_rate": 1.8775876663708402e-05, "loss": 2.902, "step": 62590 }, { "epoch": 2.00154624, "grad_norm": 0.797220766544342, "learning_rate": 1.8775490571711602e-05, "loss": 3.2062, "step": 62600 }, { "epoch": 2.00157184, "grad_norm": 0.9013228416442871, "learning_rate": 1.8775104422808076e-05, "loss": 3.204, "step": 62610 }, { "epoch": 2.00159744, "grad_norm": 0.9164538979530334, "learning_rate": 1.8774718217000328e-05, "loss": 3.1045, "step": 62620 }, { "epoch": 2.00162304, "grad_norm": 0.8638635873794556, "learning_rate": 1.8774331954290868e-05, "loss": 3.117, "step": 62630 }, { "epoch": 2.00164864, "grad_norm": 0.7883939146995544, "learning_rate": 1.8773945634682197e-05, "loss": 3.0637, "step": 62640 }, { "epoch": 2.00167424, "grad_norm": 1.1031229496002197, "learning_rate": 1.877355925817682e-05, "loss": 2.9422, "step": 62650 }, { "epoch": 2.00169984, "grad_norm": 0.8189912438392639, "learning_rate": 1.8773172824777243e-05, "loss": 3.0558, "step": 62660 }, { "epoch": 2.00172544, "grad_norm": 0.7766791582107544, "learning_rate": 1.8772786334485968e-05, "loss": 3.237, "step": 62670 }, { "epoch": 2.00175104, "grad_norm": 0.9224056601524353, "learning_rate": 1.877239978730551e-05, "loss": 3.2449, "step": 62680 }, { "epoch": 2.00177664, "grad_norm": 0.7735202312469482, "learning_rate": 1.8772013183238368e-05, "loss": 3.0165, "step": 62690 }, { "epoch": 2.00180224, "grad_norm": 0.7821776270866394, "learning_rate": 1.8771626522287054e-05, "loss": 2.822, "step": 62700 }, { "epoch": 2.00182784, "grad_norm": 1.021988034248352, "learning_rate": 1.8771239804454072e-05, "loss": 3.2628, "step": 62710 }, { "epoch": 2.00185344, "grad_norm": 1.2877686023712158, "learning_rate": 1.877085302974193e-05, "loss": 3.0152, "step": 62720 }, { "epoch": 2.00187904, "grad_norm": 0.7448866367340088, "learning_rate": 1.8770466198153142e-05, "loss": 3.1183, "step": 62730 }, { "epoch": 2.00190464, "grad_norm": 0.7120988368988037, "learning_rate": 1.8770079309690205e-05, "loss": 3.243, "step": 62740 }, { "epoch": 2.00193024, "grad_norm": 0.7587783932685852, "learning_rate": 1.876969236435564e-05, "loss": 3.2941, "step": 62750 }, { "epoch": 2.00195584, "grad_norm": 0.784542441368103, "learning_rate": 1.876930536215195e-05, "loss": 3.0764, "step": 62760 }, { "epoch": 2.00198144, "grad_norm": 0.8558960556983948, "learning_rate": 1.876891830308164e-05, "loss": 3.2625, "step": 62770 }, { "epoch": 2.00200704, "grad_norm": 0.7867354154586792, "learning_rate": 1.876853118714723e-05, "loss": 3.2267, "step": 62780 }, { "epoch": 2.00203264, "grad_norm": 0.8825104236602783, "learning_rate": 1.8768144014351223e-05, "loss": 3.2785, "step": 62790 }, { "epoch": 2.00205824, "grad_norm": 1.1192338466644287, "learning_rate": 1.8767756784696133e-05, "loss": 3.0889, "step": 62800 }, { "epoch": 2.00208384, "grad_norm": 1.004063606262207, "learning_rate": 1.8767369498184472e-05, "loss": 3.0723, "step": 62810 }, { "epoch": 2.00210944, "grad_norm": 0.8396241068840027, "learning_rate": 1.876698215481875e-05, "loss": 3.2126, "step": 62820 }, { "epoch": 2.00213504, "grad_norm": 0.8381691575050354, "learning_rate": 1.876659475460148e-05, "loss": 3.3851, "step": 62830 }, { "epoch": 2.00216064, "grad_norm": 0.7813513278961182, "learning_rate": 1.876620729753517e-05, "loss": 3.2139, "step": 62840 }, { "epoch": 2.00218624, "grad_norm": 0.7810634970664978, "learning_rate": 1.8765819783622334e-05, "loss": 3.1427, "step": 62850 }, { "epoch": 2.00221184, "grad_norm": 0.7768933773040771, "learning_rate": 1.876543221286549e-05, "loss": 3.2715, "step": 62860 }, { "epoch": 2.00223744, "grad_norm": 0.7548379898071289, "learning_rate": 1.8765044585267147e-05, "loss": 3.2177, "step": 62870 }, { "epoch": 2.00226304, "grad_norm": 0.8016476035118103, "learning_rate": 1.876465690082982e-05, "loss": 3.134, "step": 62880 }, { "epoch": 2.00228864, "grad_norm": 0.8100965023040771, "learning_rate": 1.8764269159556018e-05, "loss": 3.2959, "step": 62890 }, { "epoch": 2.00231424, "grad_norm": 0.8502007722854614, "learning_rate": 1.8763881361448264e-05, "loss": 3.296, "step": 62900 }, { "epoch": 2.00233984, "grad_norm": 0.7655857801437378, "learning_rate": 1.876349350650907e-05, "loss": 3.0754, "step": 62910 }, { "epoch": 2.00236544, "grad_norm": 0.7959839701652527, "learning_rate": 1.8763105594740945e-05, "loss": 3.0845, "step": 62920 }, { "epoch": 2.00239104, "grad_norm": 0.7490748763084412, "learning_rate": 1.8762717626146408e-05, "loss": 2.8573, "step": 62930 }, { "epoch": 2.00241664, "grad_norm": 0.7710838317871094, "learning_rate": 1.876232960072798e-05, "loss": 3.2404, "step": 62940 }, { "epoch": 2.00244224, "grad_norm": 0.8073891997337341, "learning_rate": 1.8761941518488172e-05, "loss": 3.2254, "step": 62950 }, { "epoch": 2.00246784, "grad_norm": 0.8656654953956604, "learning_rate": 1.8761553379429505e-05, "loss": 3.0009, "step": 62960 }, { "epoch": 2.00249344, "grad_norm": 0.8328723311424255, "learning_rate": 1.876116518355449e-05, "loss": 3.2993, "step": 62970 }, { "epoch": 2.00251904, "grad_norm": 0.8066844940185547, "learning_rate": 1.8760776930865647e-05, "loss": 3.1909, "step": 62980 }, { "epoch": 2.00254464, "grad_norm": 0.7569288611412048, "learning_rate": 1.8760388621365492e-05, "loss": 3.1322, "step": 62990 }, { "epoch": 2.00257024, "grad_norm": 0.9012784361839294, "learning_rate": 1.876000025505655e-05, "loss": 3.175, "step": 63000 }, { "epoch": 2.00259584, "grad_norm": 0.9221194386482239, "learning_rate": 1.875961183194133e-05, "loss": 3.2863, "step": 63010 }, { "epoch": 2.00262144, "grad_norm": 0.8234881162643433, "learning_rate": 1.875922335202236e-05, "loss": 3.294, "step": 63020 }, { "epoch": 2.00264704, "grad_norm": 0.8187986016273499, "learning_rate": 1.8758834815302148e-05, "loss": 3.0839, "step": 63030 }, { "epoch": 2.00267264, "grad_norm": 0.7584266662597656, "learning_rate": 1.8758446221783226e-05, "loss": 3.0111, "step": 63040 }, { "epoch": 2.00269824, "grad_norm": 0.9227204918861389, "learning_rate": 1.8758057571468103e-05, "loss": 3.2002, "step": 63050 }, { "epoch": 2.00272384, "grad_norm": 0.8767001032829285, "learning_rate": 1.8757668864359312e-05, "loss": 3.3672, "step": 63060 }, { "epoch": 2.00274944, "grad_norm": 0.7940986156463623, "learning_rate": 1.8757280100459357e-05, "loss": 2.9469, "step": 63070 }, { "epoch": 2.00277504, "grad_norm": 0.8242813944816589, "learning_rate": 1.8756891279770773e-05, "loss": 3.3961, "step": 63080 }, { "epoch": 2.00280064, "grad_norm": 0.7903572916984558, "learning_rate": 1.8756502402296074e-05, "loss": 3.1716, "step": 63090 }, { "epoch": 2.00282624, "grad_norm": 0.7591491341590881, "learning_rate": 1.8756113468037786e-05, "loss": 3.066, "step": 63100 }, { "epoch": 2.00285184, "grad_norm": 0.7735969424247742, "learning_rate": 1.875572447699843e-05, "loss": 3.2048, "step": 63110 }, { "epoch": 2.00287744, "grad_norm": 0.8121340870857239, "learning_rate": 1.8755335429180524e-05, "loss": 3.1805, "step": 63120 }, { "epoch": 2.00290304, "grad_norm": 0.8341983556747437, "learning_rate": 1.8754946324586598e-05, "loss": 3.2564, "step": 63130 }, { "epoch": 2.00292864, "grad_norm": 0.9172292947769165, "learning_rate": 1.875455716321917e-05, "loss": 3.2172, "step": 63140 }, { "epoch": 2.00295424, "grad_norm": 0.8964648842811584, "learning_rate": 1.875416794508077e-05, "loss": 3.0867, "step": 63150 }, { "epoch": 2.00297984, "grad_norm": 0.8006200790405273, "learning_rate": 1.875377867017391e-05, "loss": 3.1661, "step": 63160 }, { "epoch": 2.00300544, "grad_norm": 0.8919657468795776, "learning_rate": 1.8753389338501128e-05, "loss": 3.133, "step": 63170 }, { "epoch": 2.00303104, "grad_norm": 0.9464545845985413, "learning_rate": 1.875299995006494e-05, "loss": 3.0863, "step": 63180 }, { "epoch": 2.00305664, "grad_norm": 0.814842700958252, "learning_rate": 1.8752610504867874e-05, "loss": 2.937, "step": 63190 }, { "epoch": 2.00308224, "grad_norm": 1.5110138654708862, "learning_rate": 1.8752221002912452e-05, "loss": 3.3555, "step": 63200 }, { "epoch": 2.00310784, "grad_norm": 0.8458154797554016, "learning_rate": 1.875183144420121e-05, "loss": 3.0837, "step": 63210 }, { "epoch": 2.00313344, "grad_norm": 0.8017703890800476, "learning_rate": 1.875144182873666e-05, "loss": 3.1709, "step": 63220 }, { "epoch": 2.00315904, "grad_norm": 0.9854212999343872, "learning_rate": 1.8751052156521337e-05, "loss": 3.2385, "step": 63230 }, { "epoch": 2.00318464, "grad_norm": 0.9334744215011597, "learning_rate": 1.875066242755777e-05, "loss": 3.608, "step": 63240 }, { "epoch": 2.00321024, "grad_norm": 0.957226574420929, "learning_rate": 1.8750272641848478e-05, "loss": 3.2372, "step": 63250 }, { "epoch": 2.00323584, "grad_norm": 0.8166417479515076, "learning_rate": 1.8749882799395996e-05, "loss": 3.1072, "step": 63260 }, { "epoch": 2.00326144, "grad_norm": 0.8100904822349548, "learning_rate": 1.874949290020285e-05, "loss": 3.2583, "step": 63270 }, { "epoch": 2.00328704, "grad_norm": 0.8098029494285583, "learning_rate": 1.874910294427157e-05, "loss": 3.2504, "step": 63280 }, { "epoch": 2.00331264, "grad_norm": 0.7212468981742859, "learning_rate": 1.874871293160468e-05, "loss": 3.0697, "step": 63290 }, { "epoch": 2.00333824, "grad_norm": 0.8478876948356628, "learning_rate": 1.8748322862204708e-05, "loss": 3.0827, "step": 63300 }, { "epoch": 2.00336384, "grad_norm": 0.743676483631134, "learning_rate": 1.8747932736074194e-05, "loss": 3.1077, "step": 63310 }, { "epoch": 2.00338944, "grad_norm": 0.7201043367385864, "learning_rate": 1.8747542553215657e-05, "loss": 3.1677, "step": 63320 }, { "epoch": 2.00341504, "grad_norm": 0.813799262046814, "learning_rate": 1.8747152313631632e-05, "loss": 3.1567, "step": 63330 }, { "epoch": 2.00344064, "grad_norm": 0.7352209687232971, "learning_rate": 1.874676201732465e-05, "loss": 3.0401, "step": 63340 }, { "epoch": 2.00346624, "grad_norm": 0.8520877361297607, "learning_rate": 1.874637166429724e-05, "loss": 3.2092, "step": 63350 }, { "epoch": 2.00349184, "grad_norm": 0.8204120993614197, "learning_rate": 1.8745981254551935e-05, "loss": 3.1432, "step": 63360 }, { "epoch": 2.00351744, "grad_norm": 0.9297643303871155, "learning_rate": 1.8745590788091267e-05, "loss": 3.2339, "step": 63370 }, { "epoch": 2.00354304, "grad_norm": 0.7793353199958801, "learning_rate": 1.8745200264917765e-05, "loss": 3.0996, "step": 63380 }, { "epoch": 2.00356864, "grad_norm": 1.6635102033615112, "learning_rate": 1.8744809685033964e-05, "loss": 3.1012, "step": 63390 }, { "epoch": 2.00359424, "grad_norm": 0.8586156368255615, "learning_rate": 1.8744419048442395e-05, "loss": 3.1144, "step": 63400 }, { "epoch": 2.00361984, "grad_norm": 0.7662979364395142, "learning_rate": 1.8744028355145592e-05, "loss": 3.2952, "step": 63410 }, { "epoch": 2.00364544, "grad_norm": 0.7930700182914734, "learning_rate": 1.8743637605146093e-05, "loss": 3.0737, "step": 63420 }, { "epoch": 2.00367104, "grad_norm": 0.9207538366317749, "learning_rate": 1.8743246798446425e-05, "loss": 3.099, "step": 63430 }, { "epoch": 2.00369664, "grad_norm": 0.6970914602279663, "learning_rate": 1.8742855935049127e-05, "loss": 3.1683, "step": 63440 }, { "epoch": 2.00372224, "grad_norm": 0.8680051565170288, "learning_rate": 1.874246501495673e-05, "loss": 3.1657, "step": 63450 }, { "epoch": 2.00374784, "grad_norm": 0.930910587310791, "learning_rate": 1.8742074038171774e-05, "loss": 3.1852, "step": 63460 }, { "epoch": 2.00377344, "grad_norm": 0.8491758704185486, "learning_rate": 1.8741683004696788e-05, "loss": 3.3263, "step": 63470 }, { "epoch": 2.00379904, "grad_norm": 0.9326974153518677, "learning_rate": 1.874129191453431e-05, "loss": 3.4018, "step": 63480 }, { "epoch": 2.00382464, "grad_norm": 0.9674433469772339, "learning_rate": 1.874090076768688e-05, "loss": 3.3398, "step": 63490 }, { "epoch": 2.00385024, "grad_norm": 0.8049038052558899, "learning_rate": 1.874050956415703e-05, "loss": 3.0309, "step": 63500 }, { "epoch": 2.00387584, "grad_norm": 0.8259103298187256, "learning_rate": 1.87401183039473e-05, "loss": 3.0881, "step": 63510 }, { "epoch": 2.00390144, "grad_norm": 0.8892590403556824, "learning_rate": 1.8739726987060223e-05, "loss": 3.2044, "step": 63520 }, { "epoch": 2.00392704, "grad_norm": 0.9874453544616699, "learning_rate": 1.873933561349834e-05, "loss": 3.2896, "step": 63530 }, { "epoch": 2.00395264, "grad_norm": 0.7676370739936829, "learning_rate": 1.8738944183264192e-05, "loss": 3.1468, "step": 63540 }, { "epoch": 2.00397824, "grad_norm": 0.7923542261123657, "learning_rate": 1.8738552696360307e-05, "loss": 2.9979, "step": 63550 }, { "epoch": 2.00400384, "grad_norm": 0.7867023348808289, "learning_rate": 1.8738161152789236e-05, "loss": 3.2242, "step": 63560 }, { "epoch": 2.00402944, "grad_norm": 0.7216726541519165, "learning_rate": 1.873776955255351e-05, "loss": 3.2018, "step": 63570 }, { "epoch": 2.00405504, "grad_norm": 1.3280125856399536, "learning_rate": 1.873737789565567e-05, "loss": 3.1043, "step": 63580 }, { "epoch": 2.00408064, "grad_norm": 0.7718921899795532, "learning_rate": 1.873698618209826e-05, "loss": 2.8858, "step": 63590 }, { "epoch": 2.00410624, "grad_norm": 0.8486248254776001, "learning_rate": 1.873659441188381e-05, "loss": 3.3166, "step": 63600 }, { "epoch": 2.00413184, "grad_norm": 0.8226185441017151, "learning_rate": 1.8736202585014873e-05, "loss": 3.2302, "step": 63610 }, { "epoch": 2.00415744, "grad_norm": 0.7579076886177063, "learning_rate": 1.8735810701493983e-05, "loss": 3.2977, "step": 63620 }, { "epoch": 2.00418304, "grad_norm": 0.8158546090126038, "learning_rate": 1.8735418761323682e-05, "loss": 3.214, "step": 63630 }, { "epoch": 2.00420864, "grad_norm": 0.8025495409965515, "learning_rate": 1.873502676450651e-05, "loss": 3.3287, "step": 63640 }, { "epoch": 2.00423424, "grad_norm": 0.86691814661026, "learning_rate": 1.873463471104501e-05, "loss": 3.2553, "step": 63650 }, { "epoch": 2.00425984, "grad_norm": 0.7647123336791992, "learning_rate": 1.8734242600941727e-05, "loss": 3.218, "step": 63660 }, { "epoch": 2.00428544, "grad_norm": 0.7830995917320251, "learning_rate": 1.8733850434199203e-05, "loss": 3.2078, "step": 63670 }, { "epoch": 2.00431104, "grad_norm": 0.748203456401825, "learning_rate": 1.8733458210819983e-05, "loss": 3.1844, "step": 63680 }, { "epoch": 2.00433664, "grad_norm": 0.8149401545524597, "learning_rate": 1.8733065930806603e-05, "loss": 3.1319, "step": 63690 }, { "epoch": 2.00436224, "grad_norm": 0.731502890586853, "learning_rate": 1.8732673594161612e-05, "loss": 3.2726, "step": 63700 }, { "epoch": 2.00438784, "grad_norm": 1.0591739416122437, "learning_rate": 1.8732281200887553e-05, "loss": 3.1249, "step": 63710 }, { "epoch": 2.00441344, "grad_norm": 0.7957193851470947, "learning_rate": 1.873188875098697e-05, "loss": 3.122, "step": 63720 }, { "epoch": 2.00443904, "grad_norm": 0.9431000351905823, "learning_rate": 1.8731496244462414e-05, "loss": 3.0929, "step": 63730 }, { "epoch": 2.00446464, "grad_norm": 0.927344799041748, "learning_rate": 1.8731103681316422e-05, "loss": 3.0923, "step": 63740 }, { "epoch": 2.00449024, "grad_norm": 0.8106002807617188, "learning_rate": 1.8730711061551547e-05, "loss": 3.0639, "step": 63750 }, { "epoch": 2.00451584, "grad_norm": 0.7742282152175903, "learning_rate": 1.8730318385170325e-05, "loss": 3.2813, "step": 63760 }, { "epoch": 2.00454144, "grad_norm": 0.7894890904426575, "learning_rate": 1.872992565217531e-05, "loss": 3.1051, "step": 63770 }, { "epoch": 2.00456704, "grad_norm": 0.765052855014801, "learning_rate": 1.8729532862569048e-05, "loss": 3.0737, "step": 63780 }, { "epoch": 2.00459264, "grad_norm": 0.726505696773529, "learning_rate": 1.8729140016354088e-05, "loss": 3.0406, "step": 63790 }, { "epoch": 2.00461824, "grad_norm": 0.9111708402633667, "learning_rate": 1.8728747113532974e-05, "loss": 3.319, "step": 63800 }, { "epoch": 2.00464384, "grad_norm": 0.8630850911140442, "learning_rate": 1.8728354154108252e-05, "loss": 3.3675, "step": 63810 }, { "epoch": 2.00466944, "grad_norm": 0.8534704446792603, "learning_rate": 1.8727961138082475e-05, "loss": 3.1214, "step": 63820 }, { "epoch": 2.00469504, "grad_norm": 0.7989508509635925, "learning_rate": 1.872756806545819e-05, "loss": 3.0836, "step": 63830 }, { "epoch": 2.00472064, "grad_norm": 0.8366269469261169, "learning_rate": 1.8727174936237946e-05, "loss": 3.2531, "step": 63840 }, { "epoch": 2.00474624, "grad_norm": 0.777751624584198, "learning_rate": 1.8726781750424292e-05, "loss": 3.045, "step": 63850 }, { "epoch": 2.00477184, "grad_norm": 0.8762006163597107, "learning_rate": 1.872638850801978e-05, "loss": 3.4487, "step": 63860 }, { "epoch": 2.00479744, "grad_norm": 0.8059254884719849, "learning_rate": 1.872599520902695e-05, "loss": 3.2645, "step": 63870 }, { "epoch": 2.00482304, "grad_norm": 0.8763244152069092, "learning_rate": 1.8725601853448366e-05, "loss": 3.1567, "step": 63880 }, { "epoch": 2.00484864, "grad_norm": 0.8151882290840149, "learning_rate": 1.872520844128657e-05, "loss": 3.2474, "step": 63890 }, { "epoch": 2.00487424, "grad_norm": 1.0388842821121216, "learning_rate": 1.872481497254412e-05, "loss": 2.882, "step": 63900 }, { "epoch": 2.00489984, "grad_norm": 0.768438994884491, "learning_rate": 1.872442144722356e-05, "loss": 3.3628, "step": 63910 }, { "epoch": 2.00492544, "grad_norm": 0.7324565649032593, "learning_rate": 1.8724027865327446e-05, "loss": 3.2146, "step": 63920 }, { "epoch": 2.00495104, "grad_norm": 0.8452079892158508, "learning_rate": 1.872363422685833e-05, "loss": 3.0284, "step": 63930 }, { "epoch": 2.00497664, "grad_norm": 1.3559702634811401, "learning_rate": 1.8723240531818767e-05, "loss": 3.1949, "step": 63940 }, { "epoch": 2.00500224, "grad_norm": 0.8548595905303955, "learning_rate": 1.8722846780211306e-05, "loss": 3.122, "step": 63950 }, { "epoch": 2.00502784, "grad_norm": 0.8327109813690186, "learning_rate": 1.87224529720385e-05, "loss": 3.1678, "step": 63960 }, { "epoch": 2.00505344, "grad_norm": 0.9150265455245972, "learning_rate": 1.872205910730291e-05, "loss": 3.1276, "step": 63970 }, { "epoch": 2.00507904, "grad_norm": 0.8518323302268982, "learning_rate": 1.8721665186007078e-05, "loss": 3.3185, "step": 63980 }, { "epoch": 2.00510464, "grad_norm": 0.778323233127594, "learning_rate": 1.872127120815357e-05, "loss": 3.1903, "step": 63990 }, { "epoch": 2.00513024, "grad_norm": 0.7824401259422302, "learning_rate": 1.8720877173744932e-05, "loss": 3.1025, "step": 64000 }, { "epoch": 2.00515584, "grad_norm": 0.7812429070472717, "learning_rate": 1.8720483082783727e-05, "loss": 3.1201, "step": 64010 }, { "epoch": 2.00518144, "grad_norm": 0.7881830930709839, "learning_rate": 1.8720088935272507e-05, "loss": 3.1382, "step": 64020 }, { "epoch": 2.00520704, "grad_norm": 0.9236249327659607, "learning_rate": 1.8719694731213826e-05, "loss": 3.157, "step": 64030 }, { "epoch": 2.00523264, "grad_norm": 0.7715539932250977, "learning_rate": 1.871930047061024e-05, "loss": 3.1287, "step": 64040 }, { "epoch": 2.00525824, "grad_norm": 0.7381432056427002, "learning_rate": 1.8718906153464312e-05, "loss": 3.0293, "step": 64050 }, { "epoch": 2.00528384, "grad_norm": 0.9310488104820251, "learning_rate": 1.8718511779778593e-05, "loss": 3.3105, "step": 64060 }, { "epoch": 2.00530944, "grad_norm": 1.0679312944412231, "learning_rate": 1.8718117349555643e-05, "loss": 3.2393, "step": 64070 }, { "epoch": 2.00533504, "grad_norm": 1.0677021741867065, "learning_rate": 1.8717722862798015e-05, "loss": 3.1409, "step": 64080 }, { "epoch": 2.00536064, "grad_norm": 0.7829196453094482, "learning_rate": 1.8717328319508276e-05, "loss": 3.1455, "step": 64090 }, { "epoch": 2.00538624, "grad_norm": 0.8565532565116882, "learning_rate": 1.8716933719688978e-05, "loss": 3.2912, "step": 64100 }, { "epoch": 2.00541184, "grad_norm": 0.8983452916145325, "learning_rate": 1.8716539063342682e-05, "loss": 3.112, "step": 64110 }, { "epoch": 2.00543744, "grad_norm": 0.8103291988372803, "learning_rate": 1.8716144350471946e-05, "loss": 3.0315, "step": 64120 }, { "epoch": 2.00546304, "grad_norm": 0.7558532953262329, "learning_rate": 1.871574958107933e-05, "loss": 3.2868, "step": 64130 }, { "epoch": 2.00548864, "grad_norm": 0.8227546215057373, "learning_rate": 1.8715354755167394e-05, "loss": 3.3268, "step": 64140 }, { "epoch": 2.00551424, "grad_norm": 1.009784460067749, "learning_rate": 1.8714959872738697e-05, "loss": 3.1472, "step": 64150 }, { "epoch": 2.00553984, "grad_norm": 0.8443902730941772, "learning_rate": 1.8714564933795806e-05, "loss": 3.219, "step": 64160 }, { "epoch": 2.00556544, "grad_norm": 0.7933541536331177, "learning_rate": 1.8714169938341276e-05, "loss": 3.1189, "step": 64170 }, { "epoch": 2.00559104, "grad_norm": 1.0417542457580566, "learning_rate": 1.8713774886377667e-05, "loss": 2.9861, "step": 64180 }, { "epoch": 2.00561664, "grad_norm": 0.7937814593315125, "learning_rate": 1.871337977790755e-05, "loss": 3.7071, "step": 64190 }, { "epoch": 2.00564224, "grad_norm": 0.8513811230659485, "learning_rate": 1.8712984612933476e-05, "loss": 3.2333, "step": 64200 }, { "epoch": 2.00566784, "grad_norm": 1.1416035890579224, "learning_rate": 1.8712589391458012e-05, "loss": 3.4013, "step": 64210 }, { "epoch": 2.00569344, "grad_norm": 0.854341983795166, "learning_rate": 1.8712194113483723e-05, "loss": 3.1065, "step": 64220 }, { "epoch": 2.00571904, "grad_norm": 0.8911032676696777, "learning_rate": 1.871179877901317e-05, "loss": 3.1027, "step": 64230 }, { "epoch": 2.00574464, "grad_norm": 0.7681540250778198, "learning_rate": 1.8711403388048917e-05, "loss": 2.9355, "step": 64240 }, { "epoch": 2.00577024, "grad_norm": 0.8102624416351318, "learning_rate": 1.871100794059353e-05, "loss": 2.9731, "step": 64250 }, { "epoch": 2.00579584, "grad_norm": 0.9253353476524353, "learning_rate": 1.871061243664957e-05, "loss": 3.1536, "step": 64260 }, { "epoch": 2.00582144, "grad_norm": 0.869793176651001, "learning_rate": 1.87102168762196e-05, "loss": 3.1117, "step": 64270 }, { "epoch": 2.00584704, "grad_norm": 1.0336774587631226, "learning_rate": 1.8709821259306197e-05, "loss": 3.1178, "step": 64280 }, { "epoch": 2.00587264, "grad_norm": 0.8397295475006104, "learning_rate": 1.870942558591191e-05, "loss": 3.3325, "step": 64290 }, { "epoch": 2.00589824, "grad_norm": 0.7569994330406189, "learning_rate": 1.8709029856039316e-05, "loss": 3.3567, "step": 64300 }, { "epoch": 2.00592384, "grad_norm": 0.9164832830429077, "learning_rate": 1.870863406969098e-05, "loss": 3.2789, "step": 64310 }, { "epoch": 2.00594944, "grad_norm": 0.7980831265449524, "learning_rate": 1.8708238226869463e-05, "loss": 3.2306, "step": 64320 }, { "epoch": 2.00597504, "grad_norm": 0.7710156440734863, "learning_rate": 1.8707842327577337e-05, "loss": 3.1619, "step": 64330 }, { "epoch": 2.00600064, "grad_norm": 0.9433761239051819, "learning_rate": 1.8707446371817166e-05, "loss": 3.1017, "step": 64340 }, { "epoch": 2.00602624, "grad_norm": 0.8654861450195312, "learning_rate": 1.870705035959152e-05, "loss": 3.2437, "step": 64350 }, { "epoch": 2.00605184, "grad_norm": 0.8500275611877441, "learning_rate": 1.8706654290902972e-05, "loss": 3.2805, "step": 64360 }, { "epoch": 2.00607744, "grad_norm": 0.8007543087005615, "learning_rate": 1.8706258165754078e-05, "loss": 3.1768, "step": 64370 }, { "epoch": 2.00610304, "grad_norm": 1.2316392660140991, "learning_rate": 1.8705861984147415e-05, "loss": 3.4053, "step": 64380 }, { "epoch": 2.00612864, "grad_norm": 0.7517049908638, "learning_rate": 1.8705465746085553e-05, "loss": 3.4254, "step": 64390 }, { "epoch": 2.00615424, "grad_norm": 0.8050990700721741, "learning_rate": 1.8705069451571055e-05, "loss": 2.9514, "step": 64400 }, { "epoch": 2.00617984, "grad_norm": 0.7908576130867004, "learning_rate": 1.8704673100606498e-05, "loss": 3.1882, "step": 64410 }, { "epoch": 2.00620544, "grad_norm": 0.8148719072341919, "learning_rate": 1.870427669319445e-05, "loss": 3.2038, "step": 64420 }, { "epoch": 2.00623104, "grad_norm": 0.8265300989151001, "learning_rate": 1.870388022933748e-05, "loss": 3.2808, "step": 64430 }, { "epoch": 2.00625664, "grad_norm": 1.006385087966919, "learning_rate": 1.8703483709038163e-05, "loss": 3.1969, "step": 64440 }, { "epoch": 2.00628224, "grad_norm": 0.8507032990455627, "learning_rate": 1.8703087132299064e-05, "loss": 3.1742, "step": 64450 }, { "epoch": 2.00630784, "grad_norm": 0.8202927112579346, "learning_rate": 1.870269049912276e-05, "loss": 3.0124, "step": 64460 }, { "epoch": 2.00633344, "grad_norm": 0.8264254927635193, "learning_rate": 1.8702293809511817e-05, "loss": 3.1127, "step": 64470 }, { "epoch": 2.00635904, "grad_norm": 0.9264494180679321, "learning_rate": 1.8701897063468815e-05, "loss": 2.9987, "step": 64480 }, { "epoch": 2.00638464, "grad_norm": 0.8431508541107178, "learning_rate": 1.870150026099632e-05, "loss": 3.1913, "step": 64490 }, { "epoch": 2.00641024, "grad_norm": 0.7739884853363037, "learning_rate": 1.8701103402096912e-05, "loss": 3.3026, "step": 64500 }, { "epoch": 2.00643584, "grad_norm": 0.8263424038887024, "learning_rate": 1.870070648677316e-05, "loss": 3.1473, "step": 64510 }, { "epoch": 2.00646144, "grad_norm": 0.9470038414001465, "learning_rate": 1.870030951502764e-05, "loss": 3.2124, "step": 64520 }, { "epoch": 2.00648704, "grad_norm": 1.0108439922332764, "learning_rate": 1.8699912486862923e-05, "loss": 2.9714, "step": 64530 }, { "epoch": 2.00651264, "grad_norm": 0.7629536986351013, "learning_rate": 1.869951540228159e-05, "loss": 3.182, "step": 64540 }, { "epoch": 2.00653824, "grad_norm": 0.9217499494552612, "learning_rate": 1.8699118261286206e-05, "loss": 3.2076, "step": 64550 }, { "epoch": 2.00656384, "grad_norm": 0.8132180571556091, "learning_rate": 1.869872106387935e-05, "loss": 3.2014, "step": 64560 }, { "epoch": 2.00658944, "grad_norm": 0.8570480942726135, "learning_rate": 1.8698323810063605e-05, "loss": 3.1564, "step": 64570 }, { "epoch": 2.00661504, "grad_norm": 0.843679666519165, "learning_rate": 1.8697926499841542e-05, "loss": 3.2929, "step": 64580 }, { "epoch": 2.00664064, "grad_norm": 0.7761173844337463, "learning_rate": 1.8697529133215732e-05, "loss": 3.1262, "step": 64590 }, { "epoch": 2.00666624, "grad_norm": 0.7881947755813599, "learning_rate": 1.8697131710188763e-05, "loss": 3.3924, "step": 64600 }, { "epoch": 2.00669184, "grad_norm": 0.8220365047454834, "learning_rate": 1.8696734230763203e-05, "loss": 3.0971, "step": 64610 }, { "epoch": 2.00671744, "grad_norm": 0.792378306388855, "learning_rate": 1.8696336694941633e-05, "loss": 3.222, "step": 64620 }, { "epoch": 2.00674304, "grad_norm": 0.7888547778129578, "learning_rate": 1.8695939102726633e-05, "loss": 3.2291, "step": 64630 }, { "epoch": 2.00676864, "grad_norm": 0.9016435146331787, "learning_rate": 1.8695541454120777e-05, "loss": 3.1973, "step": 64640 }, { "epoch": 2.00679424, "grad_norm": 0.7826635837554932, "learning_rate": 1.8695143749126646e-05, "loss": 3.1634, "step": 64650 }, { "epoch": 2.00681984, "grad_norm": 0.7922258377075195, "learning_rate": 1.869474598774682e-05, "loss": 3.2707, "step": 64660 }, { "epoch": 2.00684544, "grad_norm": 0.926552414894104, "learning_rate": 1.8694348169983876e-05, "loss": 3.0298, "step": 64670 }, { "epoch": 2.00687104, "grad_norm": 1.076894760131836, "learning_rate": 1.8693950295840396e-05, "loss": 3.1771, "step": 64680 }, { "epoch": 2.00689664, "grad_norm": 0.8093015551567078, "learning_rate": 1.8693552365318956e-05, "loss": 3.0848, "step": 64690 }, { "epoch": 2.00692224, "grad_norm": 0.8202661871910095, "learning_rate": 1.869315437842214e-05, "loss": 3.1604, "step": 64700 }, { "epoch": 2.00694784, "grad_norm": 0.8239847421646118, "learning_rate": 1.869275633515253e-05, "loss": 3.2888, "step": 64710 }, { "epoch": 2.00697344, "grad_norm": 0.7816072702407837, "learning_rate": 1.8692358235512706e-05, "loss": 3.1046, "step": 64720 }, { "epoch": 2.00699904, "grad_norm": 0.7724476456642151, "learning_rate": 1.8691960079505247e-05, "loss": 3.1338, "step": 64730 }, { "epoch": 2.00702464, "grad_norm": 0.7788202166557312, "learning_rate": 1.8691561867132737e-05, "loss": 3.1021, "step": 64740 }, { "epoch": 2.00705024, "grad_norm": 0.8522568345069885, "learning_rate": 1.869116359839776e-05, "loss": 3.1137, "step": 64750 }, { "epoch": 2.00707584, "grad_norm": 0.8294944167137146, "learning_rate": 1.8690765273302897e-05, "loss": 3.2369, "step": 64760 }, { "epoch": 2.00710144, "grad_norm": 0.7657561898231506, "learning_rate": 1.8690366891850728e-05, "loss": 3.2428, "step": 64770 }, { "epoch": 2.00712704, "grad_norm": 0.8993768692016602, "learning_rate": 1.868996845404384e-05, "loss": 2.9374, "step": 64780 }, { "epoch": 2.00715264, "grad_norm": 0.8771103024482727, "learning_rate": 1.868956995988482e-05, "loss": 3.3379, "step": 64790 }, { "epoch": 2.00717824, "grad_norm": 0.8507959842681885, "learning_rate": 1.8689171409376242e-05, "loss": 3.3064, "step": 64800 }, { "epoch": 2.00720384, "grad_norm": 0.900553286075592, "learning_rate": 1.86887728025207e-05, "loss": 3.0793, "step": 64810 }, { "epoch": 2.00722944, "grad_norm": 0.9199166893959045, "learning_rate": 1.8688374139320776e-05, "loss": 3.1885, "step": 64820 }, { "epoch": 2.00725504, "grad_norm": 0.7657848596572876, "learning_rate": 1.8687975419779056e-05, "loss": 3.1133, "step": 64830 }, { "epoch": 2.00728064, "grad_norm": 0.8539432883262634, "learning_rate": 1.868757664389812e-05, "loss": 3.0301, "step": 64840 }, { "epoch": 2.00730624, "grad_norm": 1.0322788953781128, "learning_rate": 1.8687177811680564e-05, "loss": 3.0618, "step": 64850 }, { "epoch": 2.00733184, "grad_norm": 0.7715433239936829, "learning_rate": 1.8686778923128966e-05, "loss": 3.0962, "step": 64860 }, { "epoch": 2.00735744, "grad_norm": 0.8773583769798279, "learning_rate": 1.8686379978245917e-05, "loss": 3.3174, "step": 64870 }, { "epoch": 2.00738304, "grad_norm": 1.0233559608459473, "learning_rate": 1.8685980977033998e-05, "loss": 3.1863, "step": 64880 }, { "epoch": 2.00740864, "grad_norm": 0.7864159345626831, "learning_rate": 1.8685581919495807e-05, "loss": 3.1417, "step": 64890 }, { "epoch": 2.00743424, "grad_norm": 0.8636047840118408, "learning_rate": 1.868518280563392e-05, "loss": 3.218, "step": 64900 }, { "epoch": 2.00745984, "grad_norm": 0.8812156319618225, "learning_rate": 1.8684783635450935e-05, "loss": 3.1068, "step": 64910 }, { "epoch": 2.00748544, "grad_norm": 0.816383421421051, "learning_rate": 1.868438440894943e-05, "loss": 3.2179, "step": 64920 }, { "epoch": 2.00751104, "grad_norm": 0.7794349193572998, "learning_rate": 1.8683985126132006e-05, "loss": 3.1048, "step": 64930 }, { "epoch": 2.00753664, "grad_norm": 0.7955616116523743, "learning_rate": 1.8683585787001246e-05, "loss": 3.1985, "step": 64940 }, { "epoch": 2.00756224, "grad_norm": 0.7637655138969421, "learning_rate": 1.8683186391559736e-05, "loss": 3.1864, "step": 64950 }, { "epoch": 2.00758784, "grad_norm": 0.8440675139427185, "learning_rate": 1.8682786939810074e-05, "loss": 3.1759, "step": 64960 }, { "epoch": 2.00761344, "grad_norm": 0.7832741141319275, "learning_rate": 1.8682387431754845e-05, "loss": 3.1509, "step": 64970 }, { "epoch": 2.00763904, "grad_norm": 0.843123733997345, "learning_rate": 1.8681987867396642e-05, "loss": 3.2321, "step": 64980 }, { "epoch": 2.00766464, "grad_norm": 0.8692936301231384, "learning_rate": 1.8681588246738053e-05, "loss": 3.1482, "step": 64990 }, { "epoch": 2.00769024, "grad_norm": 0.9032774567604065, "learning_rate": 1.8681188569781673e-05, "loss": 3.228, "step": 65000 }, { "epoch": 2.00771584, "grad_norm": 0.8005793690681458, "learning_rate": 1.868078883653009e-05, "loss": 3.1157, "step": 65010 }, { "epoch": 2.00774144, "grad_norm": 0.8812565803527832, "learning_rate": 1.86803890469859e-05, "loss": 3.0512, "step": 65020 }, { "epoch": 2.00776704, "grad_norm": 0.842902660369873, "learning_rate": 1.8679989201151694e-05, "loss": 3.1759, "step": 65030 }, { "epoch": 2.00779264, "grad_norm": 0.7798905372619629, "learning_rate": 1.8679589299030064e-05, "loss": 3.1018, "step": 65040 }, { "epoch": 2.00781824, "grad_norm": 0.8512728810310364, "learning_rate": 1.8679189340623606e-05, "loss": 3.156, "step": 65050 }, { "epoch": 2.00784384, "grad_norm": 0.8525206446647644, "learning_rate": 1.867878932593491e-05, "loss": 3.0604, "step": 65060 }, { "epoch": 2.00786944, "grad_norm": 2.2933595180511475, "learning_rate": 1.8678389254966577e-05, "loss": 3.0084, "step": 65070 }, { "epoch": 2.00789504, "grad_norm": 0.847817599773407, "learning_rate": 1.867798912772119e-05, "loss": 2.9898, "step": 65080 }, { "epoch": 2.00792064, "grad_norm": 0.9245830178260803, "learning_rate": 1.867758894420135e-05, "loss": 3.1388, "step": 65090 }, { "epoch": 2.00794624, "grad_norm": 0.8139448761940002, "learning_rate": 1.8677188704409657e-05, "loss": 3.1281, "step": 65100 }, { "epoch": 2.00797184, "grad_norm": 0.9740272164344788, "learning_rate": 1.8676788408348696e-05, "loss": 3.178, "step": 65110 }, { "epoch": 2.00799744, "grad_norm": 0.858241617679596, "learning_rate": 1.8676388056021072e-05, "loss": 3.2147, "step": 65120 }, { "epoch": 2.00802304, "grad_norm": 1.032709002494812, "learning_rate": 1.8675987647429376e-05, "loss": 2.9907, "step": 65130 }, { "epoch": 2.00804864, "grad_norm": 0.8740795254707336, "learning_rate": 1.8675587182576206e-05, "loss": 3.1928, "step": 65140 }, { "epoch": 2.00807424, "grad_norm": 0.817823052406311, "learning_rate": 1.8675186661464155e-05, "loss": 3.3798, "step": 65150 }, { "epoch": 2.00809984, "grad_norm": 0.8687276244163513, "learning_rate": 1.8674786084095828e-05, "loss": 3.0739, "step": 65160 }, { "epoch": 2.00812544, "grad_norm": 0.9489772915840149, "learning_rate": 1.867438545047382e-05, "loss": 3.1227, "step": 65170 }, { "epoch": 2.00815104, "grad_norm": 0.8653345108032227, "learning_rate": 1.8673984760600727e-05, "loss": 3.1189, "step": 65180 }, { "epoch": 2.00817664, "grad_norm": 0.7615789175033569, "learning_rate": 1.8673584014479144e-05, "loss": 3.166, "step": 65190 }, { "epoch": 2.00820224, "grad_norm": 0.8257099986076355, "learning_rate": 1.867318321211168e-05, "loss": 3.2608, "step": 65200 }, { "epoch": 2.00822784, "grad_norm": 0.8535352945327759, "learning_rate": 1.8672782353500923e-05, "loss": 3.0613, "step": 65210 }, { "epoch": 2.00825344, "grad_norm": 0.8091170787811279, "learning_rate": 1.867238143864948e-05, "loss": 3.2221, "step": 65220 }, { "epoch": 2.00827904, "grad_norm": 0.8395602703094482, "learning_rate": 1.8671980467559946e-05, "loss": 3.4113, "step": 65230 }, { "epoch": 2.00830464, "grad_norm": 0.8489948511123657, "learning_rate": 1.8671579440234927e-05, "loss": 3.2349, "step": 65240 }, { "epoch": 2.00833024, "grad_norm": 1.067165732383728, "learning_rate": 1.8671178356677014e-05, "loss": 3.1761, "step": 65250 }, { "epoch": 2.00835584, "grad_norm": 0.853294312953949, "learning_rate": 1.867077721688882e-05, "loss": 3.203, "step": 65260 }, { "epoch": 2.00838144, "grad_norm": 0.8318236470222473, "learning_rate": 1.867037602087294e-05, "loss": 3.2379, "step": 65270 }, { "epoch": 2.00840704, "grad_norm": 0.7536672353744507, "learning_rate": 1.8669974768631972e-05, "loss": 3.2946, "step": 65280 }, { "epoch": 2.00843264, "grad_norm": 0.7616162896156311, "learning_rate": 1.8669573460168522e-05, "loss": 3.0778, "step": 65290 }, { "epoch": 2.00845824, "grad_norm": 0.827164351940155, "learning_rate": 1.866917209548519e-05, "loss": 3.2343, "step": 65300 }, { "epoch": 2.00848384, "grad_norm": 0.8474356532096863, "learning_rate": 1.866877067458459e-05, "loss": 2.9582, "step": 65310 }, { "epoch": 2.00850944, "grad_norm": 0.7816720008850098, "learning_rate": 1.866836919746931e-05, "loss": 3.1249, "step": 65320 }, { "epoch": 2.00853504, "grad_norm": 0.9104251861572266, "learning_rate": 1.866796766414196e-05, "loss": 3.3068, "step": 65330 }, { "epoch": 2.00856064, "grad_norm": 0.892422616481781, "learning_rate": 1.866756607460514e-05, "loss": 3.1994, "step": 65340 }, { "epoch": 2.00858624, "grad_norm": 0.848405122756958, "learning_rate": 1.8667164428861463e-05, "loss": 2.9545, "step": 65350 }, { "epoch": 2.00861184, "grad_norm": 0.9807969927787781, "learning_rate": 1.8666762726913526e-05, "loss": 3.2037, "step": 65360 }, { "epoch": 2.00863744, "grad_norm": 0.9013963341712952, "learning_rate": 1.8666360968763937e-05, "loss": 3.0036, "step": 65370 }, { "epoch": 2.00866304, "grad_norm": 0.8168727159500122, "learning_rate": 1.86659591544153e-05, "loss": 3.2429, "step": 65380 }, { "epoch": 2.00868864, "grad_norm": 0.8114789724349976, "learning_rate": 1.866555728387022e-05, "loss": 3.0014, "step": 65390 }, { "epoch": 2.00871424, "grad_norm": 0.863135576248169, "learning_rate": 1.8665155357131306e-05, "loss": 3.1983, "step": 65400 }, { "epoch": 2.00873984, "grad_norm": 0.8297328948974609, "learning_rate": 1.8664753374201158e-05, "loss": 3.2537, "step": 65410 }, { "epoch": 2.00876544, "grad_norm": 0.8740454912185669, "learning_rate": 1.8664351335082395e-05, "loss": 3.1001, "step": 65420 }, { "epoch": 2.00879104, "grad_norm": 0.8236390352249146, "learning_rate": 1.866394923977761e-05, "loss": 3.0507, "step": 65430 }, { "epoch": 2.00881664, "grad_norm": 0.9313417077064514, "learning_rate": 1.866354708828942e-05, "loss": 3.0932, "step": 65440 }, { "epoch": 2.00884224, "grad_norm": 0.7898908257484436, "learning_rate": 1.8663144880620424e-05, "loss": 3.162, "step": 65450 }, { "epoch": 2.00886784, "grad_norm": 0.8370417952537537, "learning_rate": 1.8662742616773242e-05, "loss": 3.1042, "step": 65460 }, { "epoch": 2.00889344, "grad_norm": 0.8688897490501404, "learning_rate": 1.8662340296750473e-05, "loss": 3.2893, "step": 65470 }, { "epoch": 2.00891904, "grad_norm": 0.814591646194458, "learning_rate": 1.8661937920554732e-05, "loss": 3.1615, "step": 65480 }, { "epoch": 2.00894464, "grad_norm": 0.8272987604141235, "learning_rate": 1.866153548818862e-05, "loss": 3.343, "step": 65490 }, { "epoch": 2.00897024, "grad_norm": 0.8801150918006897, "learning_rate": 1.866113299965476e-05, "loss": 3.0855, "step": 65500 }, { "epoch": 2.00899584, "grad_norm": 0.8642297983169556, "learning_rate": 1.8660730454955753e-05, "loss": 3.0826, "step": 65510 }, { "epoch": 2.00902144, "grad_norm": 0.8274104595184326, "learning_rate": 1.8660327854094206e-05, "loss": 3.1588, "step": 65520 }, { "epoch": 2.00904704, "grad_norm": 1.0224767923355103, "learning_rate": 1.8659925197072738e-05, "loss": 3.1703, "step": 65530 }, { "epoch": 2.00907264, "grad_norm": 0.7535756826400757, "learning_rate": 1.8659522483893955e-05, "loss": 3.1916, "step": 65540 }, { "epoch": 2.00909824, "grad_norm": 0.7740877866744995, "learning_rate": 1.865911971456047e-05, "loss": 2.9596, "step": 65550 }, { "epoch": 2.00912384, "grad_norm": 0.8837729096412659, "learning_rate": 1.8658716889074898e-05, "loss": 3.1351, "step": 65560 }, { "epoch": 2.00914944, "grad_norm": 0.7394229769706726, "learning_rate": 1.8658314007439845e-05, "loss": 3.2056, "step": 65570 }, { "epoch": 2.00917504, "grad_norm": 0.9003626704216003, "learning_rate": 1.865791106965793e-05, "loss": 3.1194, "step": 65580 }, { "epoch": 2.00920064, "grad_norm": 0.8860389590263367, "learning_rate": 1.865750807573176e-05, "loss": 3.2182, "step": 65590 }, { "epoch": 2.00922624, "grad_norm": 1.0899581909179688, "learning_rate": 1.865710502566395e-05, "loss": 3.4437, "step": 65600 }, { "epoch": 2.00925184, "grad_norm": 0.8706648349761963, "learning_rate": 1.8656701919457117e-05, "loss": 3.3157, "step": 65610 }, { "epoch": 2.00927744, "grad_norm": 0.7833282947540283, "learning_rate": 1.8656298757113872e-05, "loss": 3.4737, "step": 65620 }, { "epoch": 2.00930304, "grad_norm": 0.9940776824951172, "learning_rate": 1.865589553863683e-05, "loss": 3.1996, "step": 65630 }, { "epoch": 2.00932864, "grad_norm": 0.8539883494377136, "learning_rate": 1.865549226402861e-05, "loss": 3.1528, "step": 65640 }, { "epoch": 2.00935424, "grad_norm": 1.2375682592391968, "learning_rate": 1.8655088933291815e-05, "loss": 3.0251, "step": 65650 }, { "epoch": 2.00937984, "grad_norm": 0.8645936846733093, "learning_rate": 1.865468554642907e-05, "loss": 3.0955, "step": 65660 }, { "epoch": 2.00940544, "grad_norm": 0.9347479939460754, "learning_rate": 1.8654282103442997e-05, "loss": 3.0881, "step": 65670 }, { "epoch": 2.00943104, "grad_norm": 0.9067532420158386, "learning_rate": 1.8653878604336196e-05, "loss": 3.0055, "step": 65680 }, { "epoch": 2.00945664, "grad_norm": 1.0078340768814087, "learning_rate": 1.8653475049111297e-05, "loss": 3.1633, "step": 65690 }, { "epoch": 2.00948224, "grad_norm": 0.8119580745697021, "learning_rate": 1.865307143777091e-05, "loss": 3.274, "step": 65700 }, { "epoch": 2.00950784, "grad_norm": 0.7996543645858765, "learning_rate": 1.8652667770317656e-05, "loss": 3.2288, "step": 65710 }, { "epoch": 2.00953344, "grad_norm": 0.7604486346244812, "learning_rate": 1.865226404675415e-05, "loss": 3.2212, "step": 65720 }, { "epoch": 2.00955904, "grad_norm": 0.8090474605560303, "learning_rate": 1.8651860267083014e-05, "loss": 3.2078, "step": 65730 }, { "epoch": 2.00958464, "grad_norm": 1.0850285291671753, "learning_rate": 1.8651456431306858e-05, "loss": 2.9024, "step": 65740 }, { "epoch": 2.00961024, "grad_norm": 1.012482762336731, "learning_rate": 1.865105253942831e-05, "loss": 3.0722, "step": 65750 }, { "epoch": 2.00963584, "grad_norm": 0.9508424997329712, "learning_rate": 1.8650648591449987e-05, "loss": 3.0798, "step": 65760 }, { "epoch": 2.00966144, "grad_norm": 0.7910489439964294, "learning_rate": 1.8650244587374503e-05, "loss": 3.0234, "step": 65770 }, { "epoch": 2.00968704, "grad_norm": 1.0791007280349731, "learning_rate": 1.8649840527204482e-05, "loss": 3.2694, "step": 65780 }, { "epoch": 2.00971264, "grad_norm": 1.1225658655166626, "learning_rate": 1.864943641094255e-05, "loss": 3.2658, "step": 65790 }, { "epoch": 2.00973824, "grad_norm": 0.8054215312004089, "learning_rate": 1.8649032238591314e-05, "loss": 3.1764, "step": 65800 }, { "epoch": 2.00976384, "grad_norm": 0.9090524911880493, "learning_rate": 1.864862801015341e-05, "loss": 3.2216, "step": 65810 }, { "epoch": 2.00978944, "grad_norm": 0.9340590238571167, "learning_rate": 1.8648223725631444e-05, "loss": 3.1385, "step": 65820 }, { "epoch": 2.00981504, "grad_norm": 0.8034808039665222, "learning_rate": 1.864781938502805e-05, "loss": 3.2402, "step": 65830 }, { "epoch": 2.00984064, "grad_norm": 0.7418292760848999, "learning_rate": 1.8647414988345845e-05, "loss": 3.0565, "step": 65840 }, { "epoch": 2.00986624, "grad_norm": 1.0031839609146118, "learning_rate": 1.8647010535587454e-05, "loss": 3.184, "step": 65850 }, { "epoch": 2.00989184, "grad_norm": 0.9395880699157715, "learning_rate": 1.8646606026755492e-05, "loss": 3.1314, "step": 65860 }, { "epoch": 2.00991744, "grad_norm": 0.883918046951294, "learning_rate": 1.8646201461852592e-05, "loss": 3.2062, "step": 65870 }, { "epoch": 2.00994304, "grad_norm": 0.9379035830497742, "learning_rate": 1.8645796840881373e-05, "loss": 3.296, "step": 65880 }, { "epoch": 2.00996864, "grad_norm": 0.7763227224349976, "learning_rate": 1.8645392163844464e-05, "loss": 3.1223, "step": 65890 }, { "epoch": 2.00999424, "grad_norm": 0.8717398643493652, "learning_rate": 1.8644987430744476e-05, "loss": 3.0897, "step": 65900 }, { "epoch": 2.01001984, "grad_norm": 0.9706661701202393, "learning_rate": 1.8644582641584047e-05, "loss": 3.1672, "step": 65910 }, { "epoch": 2.01004544, "grad_norm": 0.7979784607887268, "learning_rate": 1.8644177796365795e-05, "loss": 3.2057, "step": 65920 }, { "epoch": 2.01007104, "grad_norm": 0.8059325218200684, "learning_rate": 1.8643772895092346e-05, "loss": 3.0201, "step": 65930 }, { "epoch": 2.01009664, "grad_norm": 1.0774143934249878, "learning_rate": 1.8643367937766332e-05, "loss": 3.1558, "step": 65940 }, { "epoch": 2.01012224, "grad_norm": 0.8405349254608154, "learning_rate": 1.864296292439037e-05, "loss": 3.3237, "step": 65950 }, { "epoch": 2.01014784, "grad_norm": 0.7952497005462646, "learning_rate": 1.864255785496709e-05, "loss": 3.2889, "step": 65960 }, { "epoch": 2.01017344, "grad_norm": 1.2045466899871826, "learning_rate": 1.8642152729499124e-05, "loss": 3.1186, "step": 65970 }, { "epoch": 2.01019904, "grad_norm": 0.9700761437416077, "learning_rate": 1.864174754798909e-05, "loss": 3.2253, "step": 65980 }, { "epoch": 2.01022464, "grad_norm": 0.8435781002044678, "learning_rate": 1.8641342310439624e-05, "loss": 3.1477, "step": 65990 }, { "epoch": 2.01025024, "grad_norm": 0.8929880857467651, "learning_rate": 1.8640937016853344e-05, "loss": 3.4374, "step": 66000 }, { "epoch": 2.01027584, "grad_norm": 0.8048924803733826, "learning_rate": 1.8640531667232888e-05, "loss": 3.144, "step": 66010 }, { "epoch": 2.01030144, "grad_norm": 0.876971960067749, "learning_rate": 1.864012626158088e-05, "loss": 3.0816, "step": 66020 }, { "epoch": 2.01032704, "grad_norm": 1.2810282707214355, "learning_rate": 1.8639720799899946e-05, "loss": 3.6592, "step": 66030 }, { "epoch": 2.01035264, "grad_norm": 0.7774816155433655, "learning_rate": 1.8639315282192723e-05, "loss": 3.1443, "step": 66040 }, { "epoch": 2.01037824, "grad_norm": 0.9862901568412781, "learning_rate": 1.8638909708461837e-05, "loss": 3.3452, "step": 66050 }, { "epoch": 2.01040384, "grad_norm": 1.009138822555542, "learning_rate": 1.8638504078709912e-05, "loss": 3.0987, "step": 66060 }, { "epoch": 2.01042944, "grad_norm": 0.9124936461448669, "learning_rate": 1.863809839293959e-05, "loss": 3.0861, "step": 66070 }, { "epoch": 2.01045504, "grad_norm": 0.8188157081604004, "learning_rate": 1.8637692651153493e-05, "loss": 3.2871, "step": 66080 }, { "epoch": 2.01048064, "grad_norm": 0.9104998111724854, "learning_rate": 1.8637286853354256e-05, "loss": 3.1867, "step": 66090 }, { "epoch": 2.01050624, "grad_norm": 0.8372017741203308, "learning_rate": 1.8636880999544505e-05, "loss": 2.9994, "step": 66100 }, { "epoch": 2.01053184, "grad_norm": 1.0105587244033813, "learning_rate": 1.863647508972688e-05, "loss": 3.2921, "step": 66110 }, { "epoch": 2.01055744, "grad_norm": 0.9506620168685913, "learning_rate": 1.863606912390401e-05, "loss": 3.206, "step": 66120 }, { "epoch": 2.01058304, "grad_norm": 0.8818066716194153, "learning_rate": 1.8635663102078526e-05, "loss": 3.1599, "step": 66130 }, { "epoch": 2.01060864, "grad_norm": 0.8584057092666626, "learning_rate": 1.863525702425306e-05, "loss": 3.0151, "step": 66140 }, { "epoch": 2.01063424, "grad_norm": 1.7565280199050903, "learning_rate": 1.863485089043025e-05, "loss": 3.5335, "step": 66150 }, { "epoch": 2.01065984, "grad_norm": 0.8548204302787781, "learning_rate": 1.8634444700612724e-05, "loss": 3.582, "step": 66160 }, { "epoch": 2.01068544, "grad_norm": 0.8515326976776123, "learning_rate": 1.863403845480312e-05, "loss": 3.1593, "step": 66170 }, { "epoch": 2.01071104, "grad_norm": 1.039372444152832, "learning_rate": 1.8633632153004072e-05, "loss": 3.3604, "step": 66180 }, { "epoch": 2.01073664, "grad_norm": 1.8024717569351196, "learning_rate": 1.8633225795218215e-05, "loss": 3.5757, "step": 66190 }, { "epoch": 2.01076224, "grad_norm": 0.8001857995986938, "learning_rate": 1.863281938144818e-05, "loss": 3.1903, "step": 66200 }, { "epoch": 2.01078784, "grad_norm": 0.7994362115859985, "learning_rate": 1.863241291169661e-05, "loss": 3.01, "step": 66210 }, { "epoch": 2.01081344, "grad_norm": 1.5018986463546753, "learning_rate": 1.863200638596613e-05, "loss": 3.5037, "step": 66220 }, { "epoch": 2.01083904, "grad_norm": 0.8507755994796753, "learning_rate": 1.863159980425939e-05, "loss": 3.2495, "step": 66230 }, { "epoch": 2.01086464, "grad_norm": 0.8866749405860901, "learning_rate": 1.8631193166579014e-05, "loss": 3.0521, "step": 66240 }, { "epoch": 2.01089024, "grad_norm": 2.4925098419189453, "learning_rate": 1.8630786472927646e-05, "loss": 3.1729, "step": 66250 }, { "epoch": 2.01091584, "grad_norm": 2.3677773475646973, "learning_rate": 1.863037972330792e-05, "loss": 3.7291, "step": 66260 }, { "epoch": 2.01094144, "grad_norm": 0.8552407622337341, "learning_rate": 1.8629972917722478e-05, "loss": 3.0165, "step": 66270 }, { "epoch": 2.01096704, "grad_norm": 0.8280068039894104, "learning_rate": 1.8629566056173953e-05, "loss": 3.1171, "step": 66280 }, { "epoch": 2.01099264, "grad_norm": 0.824739933013916, "learning_rate": 1.8629159138664988e-05, "loss": 3.1977, "step": 66290 }, { "epoch": 2.01101824, "grad_norm": 0.79875248670578, "learning_rate": 1.8628752165198217e-05, "loss": 3.2177, "step": 66300 }, { "epoch": 2.01104384, "grad_norm": 0.8459076881408691, "learning_rate": 1.8628345135776284e-05, "loss": 3.0375, "step": 66310 }, { "epoch": 2.01106944, "grad_norm": 1.4998399019241333, "learning_rate": 1.862793805040182e-05, "loss": 3.2963, "step": 66320 }, { "epoch": 2.01109504, "grad_norm": 0.8854063153266907, "learning_rate": 1.8627530909077477e-05, "loss": 3.3812, "step": 66330 }, { "epoch": 2.01112064, "grad_norm": 0.7084937691688538, "learning_rate": 1.862712371180589e-05, "loss": 2.8377, "step": 66340 }, { "epoch": 2.01114624, "grad_norm": 0.7438392639160156, "learning_rate": 1.8626716458589692e-05, "loss": 3.1419, "step": 66350 }, { "epoch": 2.01117184, "grad_norm": 0.8641341328620911, "learning_rate": 1.8626309149431536e-05, "loss": 3.1346, "step": 66360 }, { "epoch": 2.01119744, "grad_norm": 0.7990108132362366, "learning_rate": 1.8625901784334054e-05, "loss": 3.175, "step": 66370 }, { "epoch": 2.01122304, "grad_norm": 0.7525598406791687, "learning_rate": 1.8625494363299895e-05, "loss": 3.0948, "step": 66380 }, { "epoch": 2.01124864, "grad_norm": 0.8374518752098083, "learning_rate": 1.8625086886331692e-05, "loss": 3.248, "step": 66390 }, { "epoch": 2.01127424, "grad_norm": 0.872906506061554, "learning_rate": 1.8624679353432097e-05, "loss": 3.1364, "step": 66400 }, { "epoch": 2.01129984, "grad_norm": 0.8857729434967041, "learning_rate": 1.862427176460375e-05, "loss": 2.8902, "step": 66410 }, { "epoch": 2.01132544, "grad_norm": 1.2522661685943604, "learning_rate": 1.862386411984929e-05, "loss": 2.9807, "step": 66420 }, { "epoch": 2.01135104, "grad_norm": 0.9026930332183838, "learning_rate": 1.8623456419171362e-05, "loss": 3.1007, "step": 66430 }, { "epoch": 2.01137664, "grad_norm": 0.8455812931060791, "learning_rate": 1.8623048662572616e-05, "loss": 3.127, "step": 66440 }, { "epoch": 2.01140224, "grad_norm": 0.8599061965942383, "learning_rate": 1.8622640850055686e-05, "loss": 3.12, "step": 66450 }, { "epoch": 2.01142784, "grad_norm": 0.9920203685760498, "learning_rate": 1.8622232981623223e-05, "loss": 3.1675, "step": 66460 }, { "epoch": 2.01145344, "grad_norm": 1.1961058378219604, "learning_rate": 1.8621825057277872e-05, "loss": 3.3008, "step": 66470 }, { "epoch": 2.01147904, "grad_norm": 0.8846496343612671, "learning_rate": 1.862141707702228e-05, "loss": 3.3286, "step": 66480 }, { "epoch": 2.01150464, "grad_norm": 0.8032433986663818, "learning_rate": 1.8621009040859085e-05, "loss": 3.1926, "step": 66490 }, { "epoch": 2.01153024, "grad_norm": 0.8440380096435547, "learning_rate": 1.8620600948790938e-05, "loss": 3.1299, "step": 66500 }, { "epoch": 2.01155584, "grad_norm": 0.8903041481971741, "learning_rate": 1.8620192800820486e-05, "loss": 3.347, "step": 66510 }, { "epoch": 2.01158144, "grad_norm": 0.861840546131134, "learning_rate": 1.861978459695037e-05, "loss": 3.2405, "step": 66520 }, { "epoch": 2.01160704, "grad_norm": 0.7938863039016724, "learning_rate": 1.8619376337183248e-05, "loss": 3.0177, "step": 66530 }, { "epoch": 2.01163264, "grad_norm": 0.8331494927406311, "learning_rate": 1.8618968021521757e-05, "loss": 3.298, "step": 66540 }, { "epoch": 2.01165824, "grad_norm": 1.1547712087631226, "learning_rate": 1.8618559649968553e-05, "loss": 3.0661, "step": 66550 }, { "epoch": 2.01168384, "grad_norm": 0.8886057138442993, "learning_rate": 1.861815122252628e-05, "loss": 3.2139, "step": 66560 }, { "epoch": 2.01170944, "grad_norm": 0.7984655499458313, "learning_rate": 1.861774273919758e-05, "loss": 3.1927, "step": 66570 }, { "epoch": 2.01173504, "grad_norm": 1.491188883781433, "learning_rate": 1.8617334199985114e-05, "loss": 3.1733, "step": 66580 }, { "epoch": 2.01176064, "grad_norm": 0.763908326625824, "learning_rate": 1.8616925604891526e-05, "loss": 3.2577, "step": 66590 }, { "epoch": 2.01178624, "grad_norm": 0.816683292388916, "learning_rate": 1.8616516953919467e-05, "loss": 3.1247, "step": 66600 }, { "epoch": 2.01181184, "grad_norm": 0.9077170491218567, "learning_rate": 1.8616108247071584e-05, "loss": 3.2093, "step": 66610 }, { "epoch": 2.01183744, "grad_norm": 0.7649826407432556, "learning_rate": 1.8615699484350528e-05, "loss": 3.2776, "step": 66620 }, { "epoch": 2.01186304, "grad_norm": 0.8179047703742981, "learning_rate": 1.861529066575895e-05, "loss": 3.0561, "step": 66630 }, { "epoch": 2.01188864, "grad_norm": 0.9087924361228943, "learning_rate": 1.8614881791299502e-05, "loss": 3.1471, "step": 66640 }, { "epoch": 2.01191424, "grad_norm": 0.7996589541435242, "learning_rate": 1.8614472860974838e-05, "loss": 2.9029, "step": 66650 }, { "epoch": 2.01193984, "grad_norm": 0.953025221824646, "learning_rate": 1.8614063874787605e-05, "loss": 3.1741, "step": 66660 }, { "epoch": 2.01196544, "grad_norm": 0.9509737491607666, "learning_rate": 1.8613654832740454e-05, "loss": 3.0817, "step": 66670 }, { "epoch": 2.01199104, "grad_norm": 0.7602091431617737, "learning_rate": 1.8613245734836044e-05, "loss": 3.0738, "step": 66680 }, { "epoch": 2.01201664, "grad_norm": 0.7864328622817993, "learning_rate": 1.8612836581077024e-05, "loss": 3.0782, "step": 66690 }, { "epoch": 2.01204224, "grad_norm": 0.883069634437561, "learning_rate": 1.8612427371466047e-05, "loss": 2.9793, "step": 66700 }, { "epoch": 2.01206784, "grad_norm": 0.8627247214317322, "learning_rate": 1.861201810600577e-05, "loss": 3.2985, "step": 66710 }, { "epoch": 2.01209344, "grad_norm": 0.7884185910224915, "learning_rate": 1.8611608784698842e-05, "loss": 3.0685, "step": 66720 }, { "epoch": 2.01211904, "grad_norm": 0.8546576499938965, "learning_rate": 1.861119940754792e-05, "loss": 2.935, "step": 66730 }, { "epoch": 2.01214464, "grad_norm": 0.8891527056694031, "learning_rate": 1.861078997455566e-05, "loss": 3.3976, "step": 66740 }, { "epoch": 2.01217024, "grad_norm": 1.417824149131775, "learning_rate": 1.8610380485724714e-05, "loss": 2.8866, "step": 66750 }, { "epoch": 2.01219584, "grad_norm": 0.9045745134353638, "learning_rate": 1.860997094105774e-05, "loss": 3.135, "step": 66760 }, { "epoch": 2.01222144, "grad_norm": 0.9966436624526978, "learning_rate": 1.860956134055739e-05, "loss": 3.3093, "step": 66770 }, { "epoch": 2.01224704, "grad_norm": 0.8185325264930725, "learning_rate": 1.8609151684226327e-05, "loss": 2.9639, "step": 66780 }, { "epoch": 2.01227264, "grad_norm": 0.7883582711219788, "learning_rate": 1.86087419720672e-05, "loss": 3.1017, "step": 66790 }, { "epoch": 2.01229824, "grad_norm": 0.8633955717086792, "learning_rate": 1.8608332204082672e-05, "loss": 2.9746, "step": 66800 }, { "epoch": 2.01232384, "grad_norm": 0.9662339687347412, "learning_rate": 1.8607922380275396e-05, "loss": 2.8894, "step": 66810 }, { "epoch": 2.01234944, "grad_norm": 0.8907463550567627, "learning_rate": 1.860751250064803e-05, "loss": 2.8548, "step": 66820 }, { "epoch": 2.01237504, "grad_norm": 0.846868097782135, "learning_rate": 1.8607102565203235e-05, "loss": 3.2569, "step": 66830 }, { "epoch": 2.01240064, "grad_norm": 0.9160833358764648, "learning_rate": 1.8606692573943664e-05, "loss": 3.0728, "step": 66840 }, { "epoch": 2.01242624, "grad_norm": 0.9094154834747314, "learning_rate": 1.8606282526871982e-05, "loss": 3.0999, "step": 66850 }, { "epoch": 2.01245184, "grad_norm": 0.9095665812492371, "learning_rate": 1.8605872423990844e-05, "loss": 3.2025, "step": 66860 }, { "epoch": 2.01247744, "grad_norm": 1.0071282386779785, "learning_rate": 1.8605462265302912e-05, "loss": 3.1631, "step": 66870 }, { "epoch": 2.01250304, "grad_norm": 0.8057489991188049, "learning_rate": 1.8605052050810845e-05, "loss": 3.0895, "step": 66880 }, { "epoch": 2.01252864, "grad_norm": 0.8923736214637756, "learning_rate": 1.86046417805173e-05, "loss": 3.2321, "step": 66890 }, { "epoch": 2.01255424, "grad_norm": 0.8676655888557434, "learning_rate": 1.8604231454424938e-05, "loss": 3.0471, "step": 66900 }, { "epoch": 2.01257984, "grad_norm": 0.9085206985473633, "learning_rate": 1.8603821072536426e-05, "loss": 3.2545, "step": 66910 }, { "epoch": 2.01260544, "grad_norm": 0.8331010937690735, "learning_rate": 1.860341063485442e-05, "loss": 3.1892, "step": 66920 }, { "epoch": 2.01263104, "grad_norm": 0.8256235122680664, "learning_rate": 1.860300014138158e-05, "loss": 3.2066, "step": 66930 }, { "epoch": 2.01265664, "grad_norm": 0.8743095993995667, "learning_rate": 1.860258959212057e-05, "loss": 3.3886, "step": 66940 }, { "epoch": 2.01268224, "grad_norm": 0.8527296185493469, "learning_rate": 1.860217898707406e-05, "loss": 3.1982, "step": 66950 }, { "epoch": 2.01270784, "grad_norm": 0.8771145343780518, "learning_rate": 1.86017683262447e-05, "loss": 3.1817, "step": 66960 }, { "epoch": 2.01273344, "grad_norm": 0.8089393377304077, "learning_rate": 1.860135760963516e-05, "loss": 3.2041, "step": 66970 }, { "epoch": 2.01275904, "grad_norm": 0.8418470621109009, "learning_rate": 1.86009468372481e-05, "loss": 3.1721, "step": 66980 }, { "epoch": 2.01278464, "grad_norm": 0.7584643363952637, "learning_rate": 1.8600536009086184e-05, "loss": 3.0352, "step": 66990 }, { "epoch": 2.01281024, "grad_norm": 0.8343415856361389, "learning_rate": 1.860012512515208e-05, "loss": 3.342, "step": 67000 }, { "epoch": 2.01283584, "grad_norm": 0.9077651500701904, "learning_rate": 1.859971418544845e-05, "loss": 3.2943, "step": 67010 }, { "epoch": 2.01286144, "grad_norm": 0.7939983010292053, "learning_rate": 1.8599303189977962e-05, "loss": 3.2765, "step": 67020 }, { "epoch": 2.01288704, "grad_norm": 0.8519248962402344, "learning_rate": 1.8598892138743276e-05, "loss": 3.4789, "step": 67030 }, { "epoch": 2.01291264, "grad_norm": 0.8847797513008118, "learning_rate": 1.859848103174706e-05, "loss": 3.0818, "step": 67040 }, { "epoch": 2.01293824, "grad_norm": 0.7662700414657593, "learning_rate": 1.8598069868991977e-05, "loss": 3.0277, "step": 67050 }, { "epoch": 2.01296384, "grad_norm": 1.1977691650390625, "learning_rate": 1.85976586504807e-05, "loss": 3.0211, "step": 67060 }, { "epoch": 2.01298944, "grad_norm": 0.832234799861908, "learning_rate": 1.8597247376215888e-05, "loss": 3.1197, "step": 67070 }, { "epoch": 2.01301504, "grad_norm": 1.4556928873062134, "learning_rate": 1.8596836046200212e-05, "loss": 3.3775, "step": 67080 }, { "epoch": 2.01304064, "grad_norm": 0.8779915571212769, "learning_rate": 1.8596424660436342e-05, "loss": 3.3133, "step": 67090 }, { "epoch": 2.01306624, "grad_norm": 1.338665246963501, "learning_rate": 1.859601321892694e-05, "loss": 3.1736, "step": 67100 }, { "epoch": 2.01309184, "grad_norm": 0.918336808681488, "learning_rate": 1.859560172167468e-05, "loss": 3.1708, "step": 67110 }, { "epoch": 2.01311744, "grad_norm": 0.8337840437889099, "learning_rate": 1.8595190168682224e-05, "loss": 3.204, "step": 67120 }, { "epoch": 2.01314304, "grad_norm": 1.0389567613601685, "learning_rate": 1.8594778559952248e-05, "loss": 3.1541, "step": 67130 }, { "epoch": 2.01316864, "grad_norm": 0.7899871468544006, "learning_rate": 1.8594366895487416e-05, "loss": 3.1471, "step": 67140 }, { "epoch": 2.01319424, "grad_norm": 0.8443084359169006, "learning_rate": 1.8593955175290394e-05, "loss": 3.122, "step": 67150 }, { "epoch": 2.01321984, "grad_norm": 0.8003730773925781, "learning_rate": 1.859354339936386e-05, "loss": 3.3178, "step": 67160 }, { "epoch": 2.01324544, "grad_norm": 0.8194665312767029, "learning_rate": 1.859313156771048e-05, "loss": 2.9358, "step": 67170 }, { "epoch": 2.01327104, "grad_norm": 0.8348137140274048, "learning_rate": 1.859271968033293e-05, "loss": 3.3925, "step": 67180 }, { "epoch": 2.01329664, "grad_norm": 0.8220705389976501, "learning_rate": 1.8592307737233873e-05, "loss": 3.2614, "step": 67190 }, { "epoch": 2.01332224, "grad_norm": 0.9539255499839783, "learning_rate": 1.859189573841598e-05, "loss": 3.1024, "step": 67200 }, { "epoch": 2.0133478399999998, "grad_norm": 0.8696038126945496, "learning_rate": 1.8591483683881933e-05, "loss": 3.1589, "step": 67210 }, { "epoch": 2.01337344, "grad_norm": 0.8741781711578369, "learning_rate": 1.8591071573634395e-05, "loss": 3.1195, "step": 67220 }, { "epoch": 2.01339904, "grad_norm": 0.8101542592048645, "learning_rate": 1.8590659407676042e-05, "loss": 3.2945, "step": 67230 }, { "epoch": 2.01342464, "grad_norm": 0.6846411824226379, "learning_rate": 1.8590247186009545e-05, "loss": 3.0073, "step": 67240 }, { "epoch": 2.01345024, "grad_norm": 0.9611539244651794, "learning_rate": 1.8589834908637578e-05, "loss": 3.3461, "step": 67250 }, { "epoch": 2.01347584, "grad_norm": 0.8884550333023071, "learning_rate": 1.8589422575562817e-05, "loss": 3.3257, "step": 67260 }, { "epoch": 2.01350144, "grad_norm": 1.2123292684555054, "learning_rate": 1.858901018678793e-05, "loss": 3.128, "step": 67270 }, { "epoch": 2.01352704, "grad_norm": 0.7839319109916687, "learning_rate": 1.8588597742315594e-05, "loss": 3.1334, "step": 67280 }, { "epoch": 2.01355264, "grad_norm": 0.8258579969406128, "learning_rate": 1.858818524214849e-05, "loss": 3.143, "step": 67290 }, { "epoch": 2.01357824, "grad_norm": 3.129499912261963, "learning_rate": 1.858777268628928e-05, "loss": 3.3867, "step": 67300 }, { "epoch": 2.01360384, "grad_norm": 0.8437666296958923, "learning_rate": 1.858736007474065e-05, "loss": 3.5149, "step": 67310 }, { "epoch": 2.01362944, "grad_norm": 0.9226922988891602, "learning_rate": 1.8586947407505273e-05, "loss": 3.1839, "step": 67320 }, { "epoch": 2.01365504, "grad_norm": 1.1420187950134277, "learning_rate": 1.8586534684585823e-05, "loss": 3.1156, "step": 67330 }, { "epoch": 2.01368064, "grad_norm": 0.9575901627540588, "learning_rate": 1.858612190598498e-05, "loss": 3.1824, "step": 67340 }, { "epoch": 2.01370624, "grad_norm": 0.8594197630882263, "learning_rate": 1.8585709071705413e-05, "loss": 3.0605, "step": 67350 }, { "epoch": 2.01373184, "grad_norm": 0.8789212107658386, "learning_rate": 1.8585296181749806e-05, "loss": 3.0965, "step": 67360 }, { "epoch": 2.01375744, "grad_norm": 1.1808675527572632, "learning_rate": 1.8584883236120838e-05, "loss": 3.062, "step": 67370 }, { "epoch": 2.01378304, "grad_norm": 0.8541171550750732, "learning_rate": 1.858447023482118e-05, "loss": 3.1471, "step": 67380 }, { "epoch": 2.01380864, "grad_norm": 0.8005569577217102, "learning_rate": 1.858405717785352e-05, "loss": 3.0087, "step": 67390 }, { "epoch": 2.01383424, "grad_norm": 0.842965304851532, "learning_rate": 1.8583644065220524e-05, "loss": 3.1448, "step": 67400 }, { "epoch": 2.01385984, "grad_norm": 0.8404271006584167, "learning_rate": 1.8583230896924883e-05, "loss": 3.218, "step": 67410 }, { "epoch": 2.01388544, "grad_norm": 1.8747918605804443, "learning_rate": 1.8582817672969267e-05, "loss": 3.008, "step": 67420 }, { "epoch": 2.01391104, "grad_norm": 1.000733733177185, "learning_rate": 1.858240439335636e-05, "loss": 3.334, "step": 67430 }, { "epoch": 2.01393664, "grad_norm": 0.9927657246589661, "learning_rate": 1.8581991058088843e-05, "loss": 3.2156, "step": 67440 }, { "epoch": 2.01396224, "grad_norm": 1.0680351257324219, "learning_rate": 1.8581577667169393e-05, "loss": 3.1253, "step": 67450 }, { "epoch": 2.01398784, "grad_norm": 0.7766662836074829, "learning_rate": 1.858116422060069e-05, "loss": 3.2199, "step": 67460 }, { "epoch": 2.01401344, "grad_norm": 0.8152205944061279, "learning_rate": 1.8580750718385424e-05, "loss": 3.1388, "step": 67470 }, { "epoch": 2.01403904, "grad_norm": 1.0384273529052734, "learning_rate": 1.8580337160526266e-05, "loss": 3.2237, "step": 67480 }, { "epoch": 2.01406464, "grad_norm": 0.765267014503479, "learning_rate": 1.8579923547025903e-05, "loss": 3.3224, "step": 67490 }, { "epoch": 2.01409024, "grad_norm": 0.8620066046714783, "learning_rate": 1.857950987788702e-05, "loss": 3.2061, "step": 67500 }, { "epoch": 2.01411584, "grad_norm": 1.0107592344284058, "learning_rate": 1.8579096153112287e-05, "loss": 3.4006, "step": 67510 }, { "epoch": 2.01414144, "grad_norm": 1.0980418920516968, "learning_rate": 1.8578682372704403e-05, "loss": 3.1749, "step": 67520 }, { "epoch": 2.01416704, "grad_norm": 0.8842694163322449, "learning_rate": 1.857826853666604e-05, "loss": 3.259, "step": 67530 }, { "epoch": 2.01419264, "grad_norm": 0.7356914281845093, "learning_rate": 1.8577854644999887e-05, "loss": 3.1905, "step": 67540 }, { "epoch": 2.01421824, "grad_norm": 0.8125368356704712, "learning_rate": 1.8577440697708623e-05, "loss": 3.0966, "step": 67550 }, { "epoch": 2.01424384, "grad_norm": 0.7796105146408081, "learning_rate": 1.857702669479494e-05, "loss": 3.2275, "step": 67560 }, { "epoch": 2.01426944, "grad_norm": 0.823276698589325, "learning_rate": 1.8576612636261514e-05, "loss": 3.0721, "step": 67570 }, { "epoch": 2.01429504, "grad_norm": 1.1629544496536255, "learning_rate": 1.857619852211104e-05, "loss": 3.3264, "step": 67580 }, { "epoch": 2.01432064, "grad_norm": 0.8039652109146118, "learning_rate": 1.8575784352346193e-05, "loss": 3.0871, "step": 67590 }, { "epoch": 2.01434624, "grad_norm": 0.8682482838630676, "learning_rate": 1.8575370126969666e-05, "loss": 3.4499, "step": 67600 }, { "epoch": 2.01437184, "grad_norm": 0.9662261605262756, "learning_rate": 1.8574955845984145e-05, "loss": 3.1154, "step": 67610 }, { "epoch": 2.01439744, "grad_norm": 0.973056972026825, "learning_rate": 1.857454150939231e-05, "loss": 3.1136, "step": 67620 }, { "epoch": 2.01442304, "grad_norm": 0.7748619318008423, "learning_rate": 1.857412711719685e-05, "loss": 3.1962, "step": 67630 }, { "epoch": 2.01444864, "grad_norm": 0.7815437316894531, "learning_rate": 1.8573712669400463e-05, "loss": 3.2525, "step": 67640 }, { "epoch": 2.01447424, "grad_norm": 0.7860746383666992, "learning_rate": 1.857329816600582e-05, "loss": 3.1033, "step": 67650 }, { "epoch": 2.01449984, "grad_norm": 0.9123473763465881, "learning_rate": 1.857288360701562e-05, "loss": 3.1953, "step": 67660 }, { "epoch": 2.01452544, "grad_norm": 0.8341509699821472, "learning_rate": 1.857246899243255e-05, "loss": 2.777, "step": 67670 }, { "epoch": 2.01455104, "grad_norm": 0.7988730669021606, "learning_rate": 1.8572054322259294e-05, "loss": 3.1808, "step": 67680 }, { "epoch": 2.01457664, "grad_norm": 0.7528575658798218, "learning_rate": 1.857163959649855e-05, "loss": 3.092, "step": 67690 }, { "epoch": 2.01460224, "grad_norm": 0.7640255093574524, "learning_rate": 1.8571224815152994e-05, "loss": 3.3996, "step": 67700 }, { "epoch": 2.01462784, "grad_norm": 0.8668977618217468, "learning_rate": 1.8570809978225323e-05, "loss": 3.0807, "step": 67710 }, { "epoch": 2.01465344, "grad_norm": 0.814024031162262, "learning_rate": 1.857039508571823e-05, "loss": 3.3182, "step": 67720 }, { "epoch": 2.01467904, "grad_norm": 0.9836511015892029, "learning_rate": 1.8569980137634403e-05, "loss": 3.2751, "step": 67730 }, { "epoch": 2.01470464, "grad_norm": 0.7888384461402893, "learning_rate": 1.856956513397653e-05, "loss": 3.2027, "step": 67740 }, { "epoch": 2.01473024, "grad_norm": 0.9307700395584106, "learning_rate": 1.8569150074747305e-05, "loss": 3.2328, "step": 67750 }, { "epoch": 2.01475584, "grad_norm": 0.847443163394928, "learning_rate": 1.8568734959949422e-05, "loss": 2.9545, "step": 67760 }, { "epoch": 2.01478144, "grad_norm": 0.7364823222160339, "learning_rate": 1.856831978958557e-05, "loss": 3.1324, "step": 67770 }, { "epoch": 2.01480704, "grad_norm": 0.7997447848320007, "learning_rate": 1.8567904563658437e-05, "loss": 3.2467, "step": 67780 }, { "epoch": 2.01483264, "grad_norm": 0.8781660199165344, "learning_rate": 1.856748928217072e-05, "loss": 3.1645, "step": 67790 }, { "epoch": 2.01485824, "grad_norm": 0.8833571076393127, "learning_rate": 1.856707394512512e-05, "loss": 3.0948, "step": 67800 }, { "epoch": 2.01488384, "grad_norm": 0.8116918802261353, "learning_rate": 1.8566658552524316e-05, "loss": 3.1288, "step": 67810 }, { "epoch": 2.01490944, "grad_norm": 0.7700566649436951, "learning_rate": 1.856624310437101e-05, "loss": 3.1841, "step": 67820 }, { "epoch": 2.01493504, "grad_norm": 0.8995378017425537, "learning_rate": 1.856582760066789e-05, "loss": 3.2554, "step": 67830 }, { "epoch": 2.01496064, "grad_norm": 0.838148832321167, "learning_rate": 1.8565412041417658e-05, "loss": 3.2189, "step": 67840 }, { "epoch": 2.01498624, "grad_norm": 0.8282201290130615, "learning_rate": 1.8564996426623006e-05, "loss": 3.0856, "step": 67850 }, { "epoch": 2.01501184, "grad_norm": 0.8410994410514832, "learning_rate": 1.856458075628663e-05, "loss": 3.0514, "step": 67860 }, { "epoch": 2.01503744, "grad_norm": 1.3200569152832031, "learning_rate": 1.8564165030411222e-05, "loss": 3.2207, "step": 67870 }, { "epoch": 2.01506304, "grad_norm": 0.970047652721405, "learning_rate": 1.8563749248999477e-05, "loss": 3.2621, "step": 67880 }, { "epoch": 2.01508864, "grad_norm": 0.8761866688728333, "learning_rate": 1.8563333412054098e-05, "loss": 3.0935, "step": 67890 }, { "epoch": 2.01511424, "grad_norm": 0.8341840505599976, "learning_rate": 1.8562917519577777e-05, "loss": 3.169, "step": 67900 }, { "epoch": 2.01513984, "grad_norm": 0.897890567779541, "learning_rate": 1.8562501571573208e-05, "loss": 3.2418, "step": 67910 }, { "epoch": 2.01516544, "grad_norm": 0.8040850758552551, "learning_rate": 1.8562085568043095e-05, "loss": 3.3593, "step": 67920 }, { "epoch": 2.01519104, "grad_norm": 1.0025699138641357, "learning_rate": 1.8561669508990135e-05, "loss": 3.4161, "step": 67930 }, { "epoch": 2.01521664, "grad_norm": 0.9160041213035583, "learning_rate": 1.856125339441702e-05, "loss": 3.0175, "step": 67940 }, { "epoch": 2.01524224, "grad_norm": 0.863086462020874, "learning_rate": 1.8560837224326452e-05, "loss": 3.4046, "step": 67950 }, { "epoch": 2.01526784, "grad_norm": 0.7952408194541931, "learning_rate": 1.856042099872113e-05, "loss": 3.1246, "step": 67960 }, { "epoch": 2.01529344, "grad_norm": 1.5497004985809326, "learning_rate": 1.8560004717603754e-05, "loss": 2.8911, "step": 67970 }, { "epoch": 2.01531904, "grad_norm": 1.1486026048660278, "learning_rate": 1.8559588380977024e-05, "loss": 3.1987, "step": 67980 }, { "epoch": 2.01534464, "grad_norm": 0.8476951122283936, "learning_rate": 1.8559171988843638e-05, "loss": 3.1925, "step": 67990 }, { "epoch": 2.01537024, "grad_norm": 0.8259848952293396, "learning_rate": 1.8558755541206294e-05, "loss": 3.2159, "step": 68000 }, { "epoch": 2.01539584, "grad_norm": 1.485584020614624, "learning_rate": 1.855833903806769e-05, "loss": 3.2218, "step": 68010 }, { "epoch": 2.01542144, "grad_norm": 0.8790751695632935, "learning_rate": 1.8557922479430542e-05, "loss": 3.1455, "step": 68020 }, { "epoch": 2.01544704, "grad_norm": 0.779892086982727, "learning_rate": 1.8557505865297536e-05, "loss": 3.2135, "step": 68030 }, { "epoch": 2.01547264, "grad_norm": 0.88133704662323, "learning_rate": 1.855708919567138e-05, "loss": 3.188, "step": 68040 }, { "epoch": 3.00001536, "grad_norm": 0.8856436014175415, "learning_rate": 1.8556672470554773e-05, "loss": 3.5483, "step": 68050 }, { "epoch": 3.00004096, "grad_norm": 0.794960618019104, "learning_rate": 1.8556255689950424e-05, "loss": 3.0009, "step": 68060 }, { "epoch": 3.00006656, "grad_norm": 0.8288582563400269, "learning_rate": 1.8555838853861022e-05, "loss": 3.0691, "step": 68070 }, { "epoch": 3.00009216, "grad_norm": 0.788782000541687, "learning_rate": 1.8555421962289286e-05, "loss": 2.8721, "step": 68080 }, { "epoch": 3.00011776, "grad_norm": 0.9096413254737854, "learning_rate": 1.855500501523791e-05, "loss": 3.0232, "step": 68090 }, { "epoch": 3.00014336, "grad_norm": 0.783536970615387, "learning_rate": 1.85545880127096e-05, "loss": 3.0514, "step": 68100 }, { "epoch": 3.00016896, "grad_norm": 0.8300952315330505, "learning_rate": 1.855417095470706e-05, "loss": 2.8493, "step": 68110 }, { "epoch": 3.00019456, "grad_norm": 0.8863765597343445, "learning_rate": 1.8553753841232997e-05, "loss": 3.1466, "step": 68120 }, { "epoch": 3.00022016, "grad_norm": 0.861284613609314, "learning_rate": 1.855333667229011e-05, "loss": 3.1191, "step": 68130 }, { "epoch": 3.00024576, "grad_norm": 0.8517141938209534, "learning_rate": 1.8552919447881114e-05, "loss": 2.9765, "step": 68140 }, { "epoch": 3.00027136, "grad_norm": 0.7559705376625061, "learning_rate": 1.8552502168008705e-05, "loss": 3.2063, "step": 68150 }, { "epoch": 3.00029696, "grad_norm": 0.8823150396347046, "learning_rate": 1.855208483267559e-05, "loss": 3.2594, "step": 68160 }, { "epoch": 3.00032256, "grad_norm": 0.8401793241500854, "learning_rate": 1.855166744188448e-05, "loss": 3.0645, "step": 68170 }, { "epoch": 3.00034816, "grad_norm": 0.8802188634872437, "learning_rate": 1.8551249995638078e-05, "loss": 3.5583, "step": 68180 }, { "epoch": 3.00037376, "grad_norm": 0.7535529136657715, "learning_rate": 1.8550832493939094e-05, "loss": 3.0495, "step": 68190 }, { "epoch": 3.00039936, "grad_norm": 0.7658653855323792, "learning_rate": 1.8550414936790236e-05, "loss": 3.0984, "step": 68200 }, { "epoch": 3.00042496, "grad_norm": 0.963451623916626, "learning_rate": 1.8549997324194205e-05, "loss": 2.9701, "step": 68210 }, { "epoch": 3.00045056, "grad_norm": 0.9249690175056458, "learning_rate": 1.8549579656153715e-05, "loss": 3.0963, "step": 68220 }, { "epoch": 3.00047616, "grad_norm": 0.794157087802887, "learning_rate": 1.8549161932671477e-05, "loss": 2.9561, "step": 68230 }, { "epoch": 3.00050176, "grad_norm": 0.8361192345619202, "learning_rate": 1.854874415375019e-05, "loss": 3.0392, "step": 68240 }, { "epoch": 3.00052736, "grad_norm": 0.8471079468727112, "learning_rate": 1.8548326319392575e-05, "loss": 3.2349, "step": 68250 }, { "epoch": 3.00055296, "grad_norm": 0.7953920364379883, "learning_rate": 1.854790842960133e-05, "loss": 3.3774, "step": 68260 }, { "epoch": 3.00057856, "grad_norm": 0.8682894110679626, "learning_rate": 1.8547490484379173e-05, "loss": 2.884, "step": 68270 }, { "epoch": 3.00060416, "grad_norm": 0.8511155843734741, "learning_rate": 1.8547072483728814e-05, "loss": 2.8535, "step": 68280 }, { "epoch": 3.00062976, "grad_norm": 0.8139959573745728, "learning_rate": 1.854665442765296e-05, "loss": 3.0206, "step": 68290 }, { "epoch": 3.00065536, "grad_norm": 0.7931597828865051, "learning_rate": 1.8546236316154323e-05, "loss": 2.933, "step": 68300 }, { "epoch": 3.00068096, "grad_norm": 0.7719200849533081, "learning_rate": 1.854581814923561e-05, "loss": 3.0315, "step": 68310 }, { "epoch": 3.00070656, "grad_norm": 0.8443838357925415, "learning_rate": 1.8545399926899546e-05, "loss": 2.855, "step": 68320 }, { "epoch": 3.00073216, "grad_norm": 0.8684695959091187, "learning_rate": 1.8544981649148833e-05, "loss": 3.1648, "step": 68330 }, { "epoch": 3.00075776, "grad_norm": 0.8118191361427307, "learning_rate": 1.854456331598618e-05, "loss": 3.1283, "step": 68340 }, { "epoch": 3.00078336, "grad_norm": 0.8762690424919128, "learning_rate": 1.854414492741431e-05, "loss": 3.0041, "step": 68350 }, { "epoch": 3.00080896, "grad_norm": 0.7895752191543579, "learning_rate": 1.8543726483435928e-05, "loss": 2.9608, "step": 68360 }, { "epoch": 3.00083456, "grad_norm": 0.809266209602356, "learning_rate": 1.854330798405375e-05, "loss": 2.9201, "step": 68370 }, { "epoch": 3.00086016, "grad_norm": 0.9333432912826538, "learning_rate": 1.854288942927049e-05, "loss": 3.2229, "step": 68380 }, { "epoch": 3.00088576, "grad_norm": 0.835319459438324, "learning_rate": 1.8542470819088865e-05, "loss": 3.1517, "step": 68390 }, { "epoch": 3.00091136, "grad_norm": 0.9425399899482727, "learning_rate": 1.8542052153511584e-05, "loss": 2.7404, "step": 68400 }, { "epoch": 2.56e-05, "grad_norm": 1.1275755167007446, "learning_rate": 1.8541633432541368e-05, "loss": 3.2477, "step": 68410 }, { "epoch": 5.12e-05, "grad_norm": 0.8466295599937439, "learning_rate": 1.8541214656180926e-05, "loss": 3.2291, "step": 68420 }, { "epoch": 7.68e-05, "grad_norm": 0.8824359178543091, "learning_rate": 1.8540795824432983e-05, "loss": 3.5027, "step": 68430 }, { "epoch": 0.0001024, "grad_norm": 0.9689075350761414, "learning_rate": 1.8540376937300243e-05, "loss": 3.1669, "step": 68440 }, { "epoch": 0.000128, "grad_norm": 1.178140640258789, "learning_rate": 1.853995799478543e-05, "loss": 3.3384, "step": 68450 }, { "epoch": 0.0001536, "grad_norm": 1.0721757411956787, "learning_rate": 1.853953899689126e-05, "loss": 3.1674, "step": 68460 }, { "epoch": 0.0001792, "grad_norm": 1.0485539436340332, "learning_rate": 1.853911994362045e-05, "loss": 2.8812, "step": 68470 }, { "epoch": 0.0002048, "grad_norm": 0.8290631175041199, "learning_rate": 1.8538700834975714e-05, "loss": 3.3571, "step": 68480 }, { "epoch": 0.0002304, "grad_norm": 0.8866173624992371, "learning_rate": 1.853828167095977e-05, "loss": 3.0206, "step": 68490 }, { "epoch": 0.000256, "grad_norm": 0.8432106375694275, "learning_rate": 1.8537862451575345e-05, "loss": 3.093, "step": 68500 }, { "epoch": 0.0002816, "grad_norm": 0.9326231479644775, "learning_rate": 1.853744317682515e-05, "loss": 3.299, "step": 68510 }, { "epoch": 0.0003072, "grad_norm": 0.8041993975639343, "learning_rate": 1.85370238467119e-05, "loss": 3.1285, "step": 68520 }, { "epoch": 0.0003328, "grad_norm": 0.8057230710983276, "learning_rate": 1.8536604461238323e-05, "loss": 3.253, "step": 68530 }, { "epoch": 0.0003584, "grad_norm": 0.9447476267814636, "learning_rate": 1.8536185020407137e-05, "loss": 3.1936, "step": 68540 }, { "epoch": 0.000384, "grad_norm": 0.9476444721221924, "learning_rate": 1.8535765524221055e-05, "loss": 3.1161, "step": 68550 }, { "epoch": 0.0004096, "grad_norm": 0.8332032561302185, "learning_rate": 1.8535345972682802e-05, "loss": 3.6528, "step": 68560 }, { "epoch": 0.0004352, "grad_norm": 0.8830991387367249, "learning_rate": 1.8534926365795104e-05, "loss": 3.1291, "step": 68570 }, { "epoch": 0.0004608, "grad_norm": 1.0197941064834595, "learning_rate": 1.8534506703560674e-05, "loss": 3.213, "step": 68580 }, { "epoch": 0.0004864, "grad_norm": 0.7987064123153687, "learning_rate": 1.8534086985982235e-05, "loss": 3.2199, "step": 68590 }, { "epoch": 0.000512, "grad_norm": 0.7167412638664246, "learning_rate": 1.853366721306251e-05, "loss": 3.2435, "step": 68600 }, { "epoch": 0.0005376, "grad_norm": 0.8939256072044373, "learning_rate": 1.853324738480422e-05, "loss": 3.0334, "step": 68610 }, { "epoch": 0.0005632, "grad_norm": 0.8333457708358765, "learning_rate": 1.853282750121009e-05, "loss": 3.2415, "step": 68620 }, { "epoch": 0.0005888, "grad_norm": 2.6308090686798096, "learning_rate": 1.8532407562282844e-05, "loss": 2.7084, "step": 68630 }, { "epoch": 0.0006144, "grad_norm": 0.7401320338249207, "learning_rate": 1.85319875680252e-05, "loss": 3.1341, "step": 68640 }, { "epoch": 0.00064, "grad_norm": 0.903962254524231, "learning_rate": 1.8531567518439885e-05, "loss": 3.4449, "step": 68650 }, { "epoch": 0.0006656, "grad_norm": 1.3972089290618896, "learning_rate": 1.853114741352962e-05, "loss": 3.2531, "step": 68660 }, { "epoch": 0.0006912, "grad_norm": 0.9488720297813416, "learning_rate": 1.8530727253297135e-05, "loss": 3.1262, "step": 68670 }, { "epoch": 0.0007168, "grad_norm": 0.8847355842590332, "learning_rate": 1.8530307037745147e-05, "loss": 3.1208, "step": 68680 }, { "epoch": 0.0007424, "grad_norm": 0.8788971900939941, "learning_rate": 1.852988676687639e-05, "loss": 3.0595, "step": 68690 }, { "epoch": 0.000768, "grad_norm": 0.6449944972991943, "learning_rate": 1.852946644069358e-05, "loss": 2.8489, "step": 68700 }, { "epoch": 0.0007936, "grad_norm": 0.7960837483406067, "learning_rate": 1.852904605919945e-05, "loss": 3.0255, "step": 68710 }, { "epoch": 0.0008192, "grad_norm": 0.9407640099525452, "learning_rate": 1.852862562239672e-05, "loss": 3.5427, "step": 68720 }, { "epoch": 0.0008448, "grad_norm": 0.9243564605712891, "learning_rate": 1.8528205130288118e-05, "loss": 3.1684, "step": 68730 }, { "epoch": 0.0008704, "grad_norm": 0.8382480144500732, "learning_rate": 1.8527784582876377e-05, "loss": 3.3619, "step": 68740 }, { "epoch": 0.000896, "grad_norm": 1.1333603858947754, "learning_rate": 1.852736398016422e-05, "loss": 3.2403, "step": 68750 }, { "epoch": 0.0009216, "grad_norm": 0.91749107837677, "learning_rate": 1.852694332215437e-05, "loss": 3.0898, "step": 68760 }, { "epoch": 0.0009472, "grad_norm": 0.8077456951141357, "learning_rate": 1.852652260884956e-05, "loss": 3.2226, "step": 68770 }, { "epoch": 0.0009728, "grad_norm": 1.0257495641708374, "learning_rate": 1.8526101840252514e-05, "loss": 3.2514, "step": 68780 }, { "epoch": 0.0009984, "grad_norm": 0.8529435992240906, "learning_rate": 1.852568101636597e-05, "loss": 3.2323, "step": 68790 }, { "epoch": 0.001024, "grad_norm": 0.6706376671791077, "learning_rate": 1.8525260137192647e-05, "loss": 3.1054, "step": 68800 }, { "epoch": 0.0010496, "grad_norm": 0.9679190516471863, "learning_rate": 1.8524839202735275e-05, "loss": 3.4427, "step": 68810 }, { "epoch": 0.0010752, "grad_norm": 0.9185637831687927, "learning_rate": 1.852441821299659e-05, "loss": 3.037, "step": 68820 }, { "epoch": 0.0011008, "grad_norm": 0.9077486991882324, "learning_rate": 1.8523997167979318e-05, "loss": 3.3066, "step": 68830 }, { "epoch": 0.0011264, "grad_norm": 0.754400908946991, "learning_rate": 1.852357606768619e-05, "loss": 3.1496, "step": 68840 }, { "epoch": 0.001152, "grad_norm": 0.914402425289154, "learning_rate": 1.8523154912119935e-05, "loss": 3.0169, "step": 68850 }, { "epoch": 0.0011776, "grad_norm": 1.073574423789978, "learning_rate": 1.852273370128329e-05, "loss": 3.1241, "step": 68860 }, { "epoch": 0.0012032, "grad_norm": 0.8624362945556641, "learning_rate": 1.852231243517898e-05, "loss": 3.0597, "step": 68870 }, { "epoch": 0.0012288, "grad_norm": 0.7145633101463318, "learning_rate": 1.852189111380973e-05, "loss": 2.9155, "step": 68880 }, { "epoch": 0.0012544, "grad_norm": 0.8737947940826416, "learning_rate": 1.852146973717829e-05, "loss": 2.9771, "step": 68890 }, { "epoch": 0.00128, "grad_norm": 0.7734243273735046, "learning_rate": 1.8521048305287383e-05, "loss": 3.164, "step": 68900 }, { "epoch": 0.0013056, "grad_norm": 0.8422999978065491, "learning_rate": 1.852062681813974e-05, "loss": 3.158, "step": 68910 }, { "epoch": 0.0013312, "grad_norm": 0.9439597129821777, "learning_rate": 1.8520205275738098e-05, "loss": 3.3113, "step": 68920 }, { "epoch": 0.0013568, "grad_norm": 0.7476864457130432, "learning_rate": 1.8519783678085187e-05, "loss": 3.1656, "step": 68930 }, { "epoch": 0.0013824, "grad_norm": 2.131420612335205, "learning_rate": 1.851936202518375e-05, "loss": 3.3366, "step": 68940 }, { "epoch": 0.001408, "grad_norm": 0.8903071284294128, "learning_rate": 1.8518940317036507e-05, "loss": 3.2968, "step": 68950 }, { "epoch": 0.0014336, "grad_norm": 0.7748425602912903, "learning_rate": 1.8518518553646204e-05, "loss": 3.1918, "step": 68960 }, { "epoch": 0.0014592, "grad_norm": 0.7999953031539917, "learning_rate": 1.8518096735015568e-05, "loss": 3.2328, "step": 68970 }, { "epoch": 0.0014848, "grad_norm": 0.6919881701469421, "learning_rate": 1.851767486114734e-05, "loss": 3.0454, "step": 68980 }, { "epoch": 0.0015104, "grad_norm": 0.7807772755622864, "learning_rate": 1.851725293204426e-05, "loss": 2.9313, "step": 68990 }, { "epoch": 0.001536, "grad_norm": 0.7372052669525146, "learning_rate": 1.851683094770905e-05, "loss": 3.0038, "step": 69000 }, { "epoch": 0.0015616, "grad_norm": 0.8041263818740845, "learning_rate": 1.851640890814446e-05, "loss": 3.175, "step": 69010 }, { "epoch": 0.0015872, "grad_norm": 0.7486966848373413, "learning_rate": 1.8515986813353222e-05, "loss": 3.1223, "step": 69020 }, { "epoch": 0.0016128, "grad_norm": 0.857748806476593, "learning_rate": 1.851556466333807e-05, "loss": 3.0392, "step": 69030 }, { "epoch": 0.0016384, "grad_norm": 0.8905967473983765, "learning_rate": 1.8515142458101744e-05, "loss": 3.181, "step": 69040 }, { "epoch": 0.001664, "grad_norm": 0.7175791263580322, "learning_rate": 1.8514720197646985e-05, "loss": 3.2253, "step": 69050 }, { "epoch": 0.0016896, "grad_norm": 0.8366844058036804, "learning_rate": 1.8514297881976528e-05, "loss": 3.0524, "step": 69060 }, { "epoch": 0.0017152, "grad_norm": 0.7939170598983765, "learning_rate": 1.851387551109311e-05, "loss": 3.0792, "step": 69070 }, { "epoch": 0.0017408, "grad_norm": 2.247758150100708, "learning_rate": 1.8513453084999475e-05, "loss": 3.1239, "step": 69080 }, { "epoch": 0.0017664, "grad_norm": 0.7993956208229065, "learning_rate": 1.8513030603698358e-05, "loss": 3.4083, "step": 69090 }, { "epoch": 0.001792, "grad_norm": 0.8828485012054443, "learning_rate": 1.8512608067192496e-05, "loss": 3.483, "step": 69100 }, { "epoch": 0.0018176, "grad_norm": 0.6675706505775452, "learning_rate": 1.851218547548464e-05, "loss": 2.9874, "step": 69110 }, { "epoch": 0.0018432, "grad_norm": 1.0287929773330688, "learning_rate": 1.8511762828577516e-05, "loss": 3.3548, "step": 69120 }, { "epoch": 0.0018688, "grad_norm": 0.9501246809959412, "learning_rate": 1.8511340126473877e-05, "loss": 2.8737, "step": 69130 }, { "epoch": 0.0018944, "grad_norm": 0.9368725419044495, "learning_rate": 1.851091736917646e-05, "loss": 3.2427, "step": 69140 }, { "epoch": 0.00192, "grad_norm": 0.7385638356208801, "learning_rate": 1.8510494556688002e-05, "loss": 3.2636, "step": 69150 }, { "epoch": 0.0019456, "grad_norm": 0.7526397109031677, "learning_rate": 1.8510071689011255e-05, "loss": 3.2905, "step": 69160 }, { "epoch": 0.0019712, "grad_norm": 0.8052219152450562, "learning_rate": 1.850964876614895e-05, "loss": 3.2557, "step": 69170 }, { "epoch": 0.0019968, "grad_norm": 0.8557917475700378, "learning_rate": 1.8509225788103836e-05, "loss": 3.2105, "step": 69180 }, { "epoch": 0.0020224, "grad_norm": 0.9468943476676941, "learning_rate": 1.8508802754878653e-05, "loss": 3.0802, "step": 69190 }, { "epoch": 0.002048, "grad_norm": 0.7194873690605164, "learning_rate": 1.8508379666476145e-05, "loss": 3.3026, "step": 69200 }, { "epoch": 0.0020736, "grad_norm": 0.7713773846626282, "learning_rate": 1.8507956522899055e-05, "loss": 3.2222, "step": 69210 }, { "epoch": 0.0020992, "grad_norm": 0.8241308331489563, "learning_rate": 1.850753332415013e-05, "loss": 3.3518, "step": 69220 }, { "epoch": 0.0021248, "grad_norm": 0.8248451352119446, "learning_rate": 1.8507110070232115e-05, "loss": 3.3916, "step": 69230 }, { "epoch": 0.0021504, "grad_norm": 1.0005232095718384, "learning_rate": 1.850668676114775e-05, "loss": 3.3571, "step": 69240 }, { "epoch": 0.002176, "grad_norm": 0.8176611065864563, "learning_rate": 1.8506263396899783e-05, "loss": 3.2323, "step": 69250 }, { "epoch": 0.0022016, "grad_norm": 0.7907357215881348, "learning_rate": 1.8505839977490954e-05, "loss": 3.2225, "step": 69260 }, { "epoch": 0.0022272, "grad_norm": 1.0390170812606812, "learning_rate": 1.8505416502924018e-05, "loss": 3.2123, "step": 69270 }, { "epoch": 0.0022528, "grad_norm": 0.8328372836112976, "learning_rate": 1.8504992973201715e-05, "loss": 3.1092, "step": 69280 }, { "epoch": 0.0022784, "grad_norm": 0.8555614352226257, "learning_rate": 1.8504569388326792e-05, "loss": 3.2946, "step": 69290 }, { "epoch": 0.002304, "grad_norm": 0.9117024540901184, "learning_rate": 1.8504145748302e-05, "loss": 3.3813, "step": 69300 }, { "epoch": 0.0023296, "grad_norm": 0.8446325063705444, "learning_rate": 1.8503722053130078e-05, "loss": 3.4068, "step": 69310 }, { "epoch": 0.0023552, "grad_norm": 0.7432315349578857, "learning_rate": 1.8503298302813785e-05, "loss": 3.3084, "step": 69320 }, { "epoch": 0.0023808, "grad_norm": 0.7014168500900269, "learning_rate": 1.8502874497355857e-05, "loss": 3.2158, "step": 69330 }, { "epoch": 0.0024064, "grad_norm": 0.8504374623298645, "learning_rate": 1.850245063675905e-05, "loss": 3.3419, "step": 69340 }, { "epoch": 0.002432, "grad_norm": 0.8168885707855225, "learning_rate": 1.8502026721026108e-05, "loss": 3.279, "step": 69350 }, { "epoch": 0.0024576, "grad_norm": 1.0696319341659546, "learning_rate": 1.8501602750159783e-05, "loss": 3.2723, "step": 69360 }, { "epoch": 0.0024832, "grad_norm": 0.9547381401062012, "learning_rate": 1.8501178724162824e-05, "loss": 3.4725, "step": 69370 }, { "epoch": 0.0025088, "grad_norm": 0.862190306186676, "learning_rate": 1.8500754643037983e-05, "loss": 3.0859, "step": 69380 }, { "epoch": 0.0025344, "grad_norm": 0.8106700778007507, "learning_rate": 1.8500330506788003e-05, "loss": 3.0589, "step": 69390 }, { "epoch": 0.00256, "grad_norm": 0.9053753614425659, "learning_rate": 1.849990631541564e-05, "loss": 3.3672, "step": 69400 }, { "epoch": 0.0025856, "grad_norm": 0.8969128131866455, "learning_rate": 1.8499482068923646e-05, "loss": 3.1076, "step": 69410 }, { "epoch": 0.0026112, "grad_norm": 1.026235818862915, "learning_rate": 1.849905776731477e-05, "loss": 3.3253, "step": 69420 }, { "epoch": 0.0026368, "grad_norm": 0.9583814144134521, "learning_rate": 1.8498633410591764e-05, "loss": 3.4266, "step": 69430 }, { "epoch": 0.0026624, "grad_norm": 0.7212189435958862, "learning_rate": 1.8498208998757375e-05, "loss": 3.1382, "step": 69440 }, { "epoch": 0.002688, "grad_norm": 0.7246344685554504, "learning_rate": 1.8497784531814356e-05, "loss": 2.8555, "step": 69450 }, { "epoch": 0.0027136, "grad_norm": 0.758539080619812, "learning_rate": 1.849736000976547e-05, "loss": 3.2059, "step": 69460 }, { "epoch": 0.0027392, "grad_norm": 0.9648528695106506, "learning_rate": 1.849693543261346e-05, "loss": 3.1546, "step": 69470 }, { "epoch": 0.0027648, "grad_norm": 0.7584167718887329, "learning_rate": 1.8496510800361082e-05, "loss": 3.3662, "step": 69480 }, { "epoch": 0.0027904, "grad_norm": 1.8250153064727783, "learning_rate": 1.849608611301109e-05, "loss": 3.3623, "step": 69490 }, { "epoch": 0.002816, "grad_norm": 0.8145759701728821, "learning_rate": 1.8495661370566238e-05, "loss": 3.2424, "step": 69500 }, { "epoch": 0.0028416, "grad_norm": 0.7872045040130615, "learning_rate": 1.849523657302928e-05, "loss": 3.2554, "step": 69510 }, { "epoch": 0.0028672, "grad_norm": 0.8868720531463623, "learning_rate": 1.8494811720402966e-05, "loss": 3.1912, "step": 69520 }, { "epoch": 0.0028928, "grad_norm": 0.703591525554657, "learning_rate": 1.849438681269006e-05, "loss": 3.2491, "step": 69530 }, { "epoch": 0.0029184, "grad_norm": 0.9696431159973145, "learning_rate": 1.8493961849893313e-05, "loss": 3.4784, "step": 69540 }, { "epoch": 0.002944, "grad_norm": 0.9066920876502991, "learning_rate": 1.849353683201548e-05, "loss": 3.3185, "step": 69550 }, { "epoch": 0.0029696, "grad_norm": 0.8990196585655212, "learning_rate": 1.849311175905932e-05, "loss": 3.5771, "step": 69560 }, { "epoch": 0.0029952, "grad_norm": 0.7520591020584106, "learning_rate": 1.8492686631027584e-05, "loss": 3.2542, "step": 69570 }, { "epoch": 0.0030208, "grad_norm": 0.7621356844902039, "learning_rate": 1.8492261447923036e-05, "loss": 3.1209, "step": 69580 }, { "epoch": 0.0030464, "grad_norm": 0.8702579736709595, "learning_rate": 1.8491836209748424e-05, "loss": 3.1512, "step": 69590 }, { "epoch": 0.003072, "grad_norm": 0.8033435940742493, "learning_rate": 1.8491410916506516e-05, "loss": 3.3963, "step": 69600 }, { "epoch": 0.0030976, "grad_norm": 0.7576177716255188, "learning_rate": 1.849098556820006e-05, "loss": 3.2068, "step": 69610 }, { "epoch": 0.0031232, "grad_norm": 0.7742358446121216, "learning_rate": 1.8490560164831825e-05, "loss": 3.3956, "step": 69620 }, { "epoch": 0.0031488, "grad_norm": 0.7926445007324219, "learning_rate": 1.849013470640456e-05, "loss": 3.1904, "step": 69630 }, { "epoch": 0.0031744, "grad_norm": 0.7856349945068359, "learning_rate": 1.8489709192921032e-05, "loss": 3.4599, "step": 69640 }, { "epoch": 0.0032, "grad_norm": 0.8342857956886292, "learning_rate": 1.848928362438399e-05, "loss": 3.2318, "step": 69650 }, { "epoch": 0.0032256, "grad_norm": 1.0840213298797607, "learning_rate": 1.8488858000796204e-05, "loss": 3.4992, "step": 69660 }, { "epoch": 0.0032512, "grad_norm": 0.7397967576980591, "learning_rate": 1.848843232216043e-05, "loss": 3.2773, "step": 69670 }, { "epoch": 0.0032768, "grad_norm": 0.8136853575706482, "learning_rate": 1.848800658847943e-05, "loss": 3.3476, "step": 69680 }, { "epoch": 0.0033024, "grad_norm": 0.8095155954360962, "learning_rate": 1.8487580799755957e-05, "loss": 3.3494, "step": 69690 }, { "epoch": 0.003328, "grad_norm": 0.8499106764793396, "learning_rate": 1.848715495599278e-05, "loss": 3.2914, "step": 69700 }, { "epoch": 0.0033536, "grad_norm": 0.7460806369781494, "learning_rate": 1.8486729057192664e-05, "loss": 3.0376, "step": 69710 }, { "epoch": 0.0033792, "grad_norm": 0.8524248600006104, "learning_rate": 1.848630310335836e-05, "loss": 3.1685, "step": 69720 }, { "epoch": 0.0034048, "grad_norm": 0.8730520009994507, "learning_rate": 1.848587709449264e-05, "loss": 3.3352, "step": 69730 }, { "epoch": 0.0034304, "grad_norm": 0.7811876535415649, "learning_rate": 1.848545103059826e-05, "loss": 3.196, "step": 69740 }, { "epoch": 0.003456, "grad_norm": 0.8580236434936523, "learning_rate": 1.8485024911677986e-05, "loss": 3.2567, "step": 69750 }, { "epoch": 0.0034816, "grad_norm": 0.7981710433959961, "learning_rate": 1.8484598737734577e-05, "loss": 3.3723, "step": 69760 }, { "epoch": 0.0035072, "grad_norm": 0.7371056079864502, "learning_rate": 1.8484172508770806e-05, "loss": 3.2755, "step": 69770 }, { "epoch": 0.0035328, "grad_norm": 0.7099502682685852, "learning_rate": 1.8483746224789425e-05, "loss": 3.2629, "step": 69780 }, { "epoch": 0.0035584, "grad_norm": 0.9464071393013, "learning_rate": 1.848331988579321e-05, "loss": 3.335, "step": 69790 }, { "epoch": 0.003584, "grad_norm": 1.050563097000122, "learning_rate": 1.8482893491784917e-05, "loss": 2.9918, "step": 69800 }, { "epoch": 0.0036096, "grad_norm": 1.105548620223999, "learning_rate": 1.8482467042767315e-05, "loss": 3.2537, "step": 69810 }, { "epoch": 0.0036352, "grad_norm": 0.7120277881622314, "learning_rate": 1.8482040538743166e-05, "loss": 3.5197, "step": 69820 }, { "epoch": 0.0036608, "grad_norm": 0.790337860584259, "learning_rate": 1.8481613979715242e-05, "loss": 3.2726, "step": 69830 }, { "epoch": 0.0036864, "grad_norm": 0.766021728515625, "learning_rate": 1.8481187365686305e-05, "loss": 3.2835, "step": 69840 }, { "epoch": 0.003712, "grad_norm": 0.9734716415405273, "learning_rate": 1.8480760696659122e-05, "loss": 3.2551, "step": 69850 }, { "epoch": 0.0037376, "grad_norm": 1.6507699489593506, "learning_rate": 1.8480333972636458e-05, "loss": 3.3969, "step": 69860 }, { "epoch": 0.0037632, "grad_norm": 0.7466112375259399, "learning_rate": 1.8479907193621084e-05, "loss": 3.5812, "step": 69870 }, { "epoch": 0.0037888, "grad_norm": 0.7073920369148254, "learning_rate": 1.8479480359615766e-05, "loss": 3.3127, "step": 69880 }, { "epoch": 0.0038144, "grad_norm": 0.999008059501648, "learning_rate": 1.8479053470623268e-05, "loss": 3.3355, "step": 69890 }, { "epoch": 0.00384, "grad_norm": 0.7662294507026672, "learning_rate": 1.8478626526646364e-05, "loss": 3.3343, "step": 69900 }, { "epoch": 0.0038656, "grad_norm": 0.8517425656318665, "learning_rate": 1.847819952768782e-05, "loss": 3.1122, "step": 69910 }, { "epoch": 0.0038912, "grad_norm": 0.8047366738319397, "learning_rate": 1.8477772473750408e-05, "loss": 3.2864, "step": 69920 }, { "epoch": 0.0039168, "grad_norm": 1.0919620990753174, "learning_rate": 1.847734536483689e-05, "loss": 3.4509, "step": 69930 }, { "epoch": 0.0039424, "grad_norm": 0.8218508362770081, "learning_rate": 1.8476918200950044e-05, "loss": 3.0935, "step": 69940 }, { "epoch": 0.003968, "grad_norm": 0.7097235321998596, "learning_rate": 1.8476490982092633e-05, "loss": 3.072, "step": 69950 }, { "epoch": 0.0039936, "grad_norm": 0.7673003077507019, "learning_rate": 1.8476063708267432e-05, "loss": 3.3936, "step": 69960 }, { "epoch": 0.0040192, "grad_norm": 0.9126924872398376, "learning_rate": 1.847563637947721e-05, "loss": 3.2126, "step": 69970 }, { "epoch": 0.0040448, "grad_norm": 0.7830830216407776, "learning_rate": 1.8475208995724744e-05, "loss": 3.2328, "step": 69980 }, { "epoch": 0.0040704, "grad_norm": 0.8173931837081909, "learning_rate": 1.8474781557012792e-05, "loss": 3.1563, "step": 69990 }, { "epoch": 0.004096, "grad_norm": 0.7509754300117493, "learning_rate": 1.8474354063344137e-05, "loss": 3.2402, "step": 70000 }, { "epoch": 0.0041216, "grad_norm": 0.8451940417289734, "learning_rate": 1.8473926514721553e-05, "loss": 3.3931, "step": 70010 }, { "epoch": 0.0041472, "grad_norm": 0.816677987575531, "learning_rate": 1.84734989111478e-05, "loss": 3.2949, "step": 70020 }, { "epoch": 0.0041728, "grad_norm": 1.0314418077468872, "learning_rate": 1.8473071252625664e-05, "loss": 3.3402, "step": 70030 }, { "epoch": 0.0041984, "grad_norm": 0.8262415528297424, "learning_rate": 1.8472643539157912e-05, "loss": 3.2752, "step": 70040 }, { "epoch": 0.004224, "grad_norm": 0.8937143087387085, "learning_rate": 1.8472215770747314e-05, "loss": 3.1793, "step": 70050 }, { "epoch": 0.0042496, "grad_norm": 0.8575237989425659, "learning_rate": 1.8471787947396654e-05, "loss": 3.3204, "step": 70060 }, { "epoch": 0.0042752, "grad_norm": 0.9386354684829712, "learning_rate": 1.8471360069108698e-05, "loss": 3.4641, "step": 70070 }, { "epoch": 0.0043008, "grad_norm": 1.2388969659805298, "learning_rate": 1.847093213588622e-05, "loss": 3.4167, "step": 70080 }, { "epoch": 0.0043264, "grad_norm": 1.2070999145507812, "learning_rate": 1.8470504147732005e-05, "loss": 3.7811, "step": 70090 }, { "epoch": 0.004352, "grad_norm": 0.8903472423553467, "learning_rate": 1.847007610464882e-05, "loss": 3.2809, "step": 70100 }, { "epoch": 0.0043776, "grad_norm": 0.863845705986023, "learning_rate": 1.846964800663944e-05, "loss": 3.3243, "step": 70110 }, { "epoch": 0.0044032, "grad_norm": 0.8031190037727356, "learning_rate": 1.8469219853706647e-05, "loss": 3.1961, "step": 70120 }, { "epoch": 0.0044288, "grad_norm": 0.9813411235809326, "learning_rate": 1.846879164585321e-05, "loss": 3.0593, "step": 70130 }, { "epoch": 0.0044544, "grad_norm": 0.8296841979026794, "learning_rate": 1.8468363383081912e-05, "loss": 3.294, "step": 70140 }, { "epoch": 0.00448, "grad_norm": 1.0339909791946411, "learning_rate": 1.8467935065395527e-05, "loss": 3.4457, "step": 70150 }, { "epoch": 0.0045056, "grad_norm": 0.703135073184967, "learning_rate": 1.8467506692796835e-05, "loss": 3.346, "step": 70160 }, { "epoch": 0.0045312, "grad_norm": 0.948402464389801, "learning_rate": 1.8467078265288613e-05, "loss": 3.3743, "step": 70170 }, { "epoch": 0.0045568, "grad_norm": 0.718004584312439, "learning_rate": 1.8466649782873638e-05, "loss": 3.4582, "step": 70180 }, { "epoch": 0.0045824, "grad_norm": 1.0943851470947266, "learning_rate": 1.846622124555469e-05, "loss": 3.033, "step": 70190 }, { "epoch": 0.004608, "grad_norm": 0.9082216620445251, "learning_rate": 1.8465792653334547e-05, "loss": 3.3795, "step": 70200 }, { "epoch": 0.0046336, "grad_norm": 0.8682243824005127, "learning_rate": 1.8465364006215988e-05, "loss": 3.1936, "step": 70210 }, { "epoch": 0.0046592, "grad_norm": 0.8466971516609192, "learning_rate": 1.8464935304201792e-05, "loss": 3.3444, "step": 70220 }, { "epoch": 0.0046848, "grad_norm": 0.7563896179199219, "learning_rate": 1.8464506547294742e-05, "loss": 3.1737, "step": 70230 }, { "epoch": 0.0047104, "grad_norm": 0.7707040309906006, "learning_rate": 1.8464077735497616e-05, "loss": 3.3231, "step": 70240 }, { "epoch": 0.004736, "grad_norm": 0.8743517994880676, "learning_rate": 1.8463648868813195e-05, "loss": 3.2409, "step": 70250 }, { "epoch": 0.0047616, "grad_norm": 0.9581217765808105, "learning_rate": 1.846321994724426e-05, "loss": 3.2409, "step": 70260 }, { "epoch": 0.0047872, "grad_norm": 0.876869261264801, "learning_rate": 1.8462790970793597e-05, "loss": 3.1063, "step": 70270 }, { "epoch": 0.0048128, "grad_norm": 0.8515183329582214, "learning_rate": 1.846236193946398e-05, "loss": 3.2309, "step": 70280 }, { "epoch": 0.0048384, "grad_norm": 1.0163205862045288, "learning_rate": 1.8461932853258197e-05, "loss": 3.2511, "step": 70290 }, { "epoch": 0.004864, "grad_norm": 0.7894362807273865, "learning_rate": 1.8461503712179025e-05, "loss": 3.3185, "step": 70300 }, { "epoch": 0.0048896, "grad_norm": 0.720415472984314, "learning_rate": 1.846107451622925e-05, "loss": 3.0969, "step": 70310 }, { "epoch": 0.0049152, "grad_norm": 1.2950632572174072, "learning_rate": 1.8460645265411662e-05, "loss": 3.2046, "step": 70320 }, { "epoch": 0.0049408, "grad_norm": 0.7073477506637573, "learning_rate": 1.8460215959729032e-05, "loss": 3.2162, "step": 70330 }, { "epoch": 0.0049664, "grad_norm": 0.8712908625602722, "learning_rate": 1.8459786599184152e-05, "loss": 3.0949, "step": 70340 }, { "epoch": 0.004992, "grad_norm": 0.7294749617576599, "learning_rate": 1.8459357183779804e-05, "loss": 3.1822, "step": 70350 }, { "epoch": 0.0050176, "grad_norm": 0.8452627062797546, "learning_rate": 1.845892771351877e-05, "loss": 3.5466, "step": 70360 }, { "epoch": 0.0050432, "grad_norm": 0.7718753814697266, "learning_rate": 1.8458498188403842e-05, "loss": 3.1315, "step": 70370 }, { "epoch": 0.0050688, "grad_norm": 0.8293849229812622, "learning_rate": 1.84580686084378e-05, "loss": 3.2801, "step": 70380 }, { "epoch": 0.0050944, "grad_norm": 1.9005035161972046, "learning_rate": 1.8457638973623432e-05, "loss": 3.1948, "step": 70390 }, { "epoch": 0.00512, "grad_norm": 0.9640786647796631, "learning_rate": 1.845720928396352e-05, "loss": 3.366, "step": 70400 }, { "epoch": 0.0051456, "grad_norm": 1.2535552978515625, "learning_rate": 1.8456779539460856e-05, "loss": 3.4049, "step": 70410 }, { "epoch": 0.0051712, "grad_norm": 0.7608327269554138, "learning_rate": 1.845634974011822e-05, "loss": 3.3765, "step": 70420 }, { "epoch": 0.0051968, "grad_norm": 0.8438712358474731, "learning_rate": 1.845591988593841e-05, "loss": 3.2875, "step": 70430 }, { "epoch": 0.0052224, "grad_norm": 0.9291661977767944, "learning_rate": 1.8455489976924203e-05, "loss": 3.1461, "step": 70440 }, { "epoch": 0.005248, "grad_norm": 1.0552173852920532, "learning_rate": 1.845506001307839e-05, "loss": 3.3072, "step": 70450 }, { "epoch": 0.0052736, "grad_norm": 0.950257420539856, "learning_rate": 1.845462999440376e-05, "loss": 3.1994, "step": 70460 }, { "epoch": 0.0052992, "grad_norm": 0.8778652548789978, "learning_rate": 1.8454199920903105e-05, "loss": 3.024, "step": 70470 }, { "epoch": 0.0053248, "grad_norm": 0.996397852897644, "learning_rate": 1.845376979257921e-05, "loss": 3.2575, "step": 70480 }, { "epoch": 0.0053504, "grad_norm": 0.9165460467338562, "learning_rate": 1.845333960943486e-05, "loss": 3.2379, "step": 70490 }, { "epoch": 0.005376, "grad_norm": 1.077330231666565, "learning_rate": 1.8452909371472854e-05, "loss": 3.3539, "step": 70500 }, { "epoch": 0.0054016, "grad_norm": 0.8669360876083374, "learning_rate": 1.8452479078695974e-05, "loss": 3.1772, "step": 70510 }, { "epoch": 0.0054272, "grad_norm": 0.8885001540184021, "learning_rate": 1.8452048731107015e-05, "loss": 3.2586, "step": 70520 }, { "epoch": 0.0054528, "grad_norm": 0.844397246837616, "learning_rate": 1.8451618328708768e-05, "loss": 3.2421, "step": 70530 }, { "epoch": 0.0054784, "grad_norm": 0.7651969194412231, "learning_rate": 1.845118787150402e-05, "loss": 3.1822, "step": 70540 }, { "epoch": 0.005504, "grad_norm": 0.7750743627548218, "learning_rate": 1.8450757359495567e-05, "loss": 3.2799, "step": 70550 }, { "epoch": 0.0055296, "grad_norm": 0.9914882183074951, "learning_rate": 1.8450326792686195e-05, "loss": 3.0585, "step": 70560 }, { "epoch": 0.0055552, "grad_norm": 0.9832212924957275, "learning_rate": 1.8449896171078697e-05, "loss": 3.1942, "step": 70570 }, { "epoch": 0.0055808, "grad_norm": 0.907675564289093, "learning_rate": 1.8449465494675875e-05, "loss": 3.5586, "step": 70580 }, { "epoch": 0.0056064, "grad_norm": 0.9228242039680481, "learning_rate": 1.8449034763480515e-05, "loss": 3.4248, "step": 70590 }, { "epoch": 0.005632, "grad_norm": 0.8086884021759033, "learning_rate": 1.8448603977495404e-05, "loss": 3.2196, "step": 70600 }, { "epoch": 0.0056576, "grad_norm": 0.9583324790000916, "learning_rate": 1.8448173136723346e-05, "loss": 3.319, "step": 70610 }, { "epoch": 0.0056832, "grad_norm": 2.2052857875823975, "learning_rate": 1.8447742241167127e-05, "loss": 3.3588, "step": 70620 }, { "epoch": 0.0057088, "grad_norm": 1.0087544918060303, "learning_rate": 1.8447311290829547e-05, "loss": 3.3409, "step": 70630 }, { "epoch": 0.0057344, "grad_norm": 0.7766804695129395, "learning_rate": 1.8446880285713398e-05, "loss": 3.2269, "step": 70640 }, { "epoch": 0.00576, "grad_norm": 0.8458024263381958, "learning_rate": 1.8446449225821478e-05, "loss": 3.4492, "step": 70650 }, { "epoch": 0.0057856, "grad_norm": 0.9295803904533386, "learning_rate": 1.8446018111156573e-05, "loss": 3.0202, "step": 70660 }, { "epoch": 0.0058112, "grad_norm": 1.358441948890686, "learning_rate": 1.8445586941721493e-05, "loss": 3.1725, "step": 70670 }, { "epoch": 0.0058368, "grad_norm": 0.8147319555282593, "learning_rate": 1.8445155717519022e-05, "loss": 3.2947, "step": 70680 }, { "epoch": 0.0058624, "grad_norm": 0.7574815154075623, "learning_rate": 1.844472443855196e-05, "loss": 3.2518, "step": 70690 }, { "epoch": 0.005888, "grad_norm": 0.8195670247077942, "learning_rate": 1.8444293104823105e-05, "loss": 2.9613, "step": 70700 }, { "epoch": 0.0059136, "grad_norm": 0.7762078046798706, "learning_rate": 1.8443861716335253e-05, "loss": 3.2455, "step": 70710 }, { "epoch": 0.0059392, "grad_norm": 0.7724210619926453, "learning_rate": 1.8443430273091204e-05, "loss": 3.4022, "step": 70720 }, { "epoch": 0.0059648, "grad_norm": 0.915790319442749, "learning_rate": 1.8442998775093755e-05, "loss": 3.1748, "step": 70730 }, { "epoch": 0.0059904, "grad_norm": 0.835137128829956, "learning_rate": 1.8442567222345696e-05, "loss": 3.1584, "step": 70740 }, { "epoch": 0.006016, "grad_norm": 0.8854544758796692, "learning_rate": 1.844213561484984e-05, "loss": 3.2844, "step": 70750 }, { "epoch": 0.0060416, "grad_norm": 0.8083662390708923, "learning_rate": 1.8441703952608975e-05, "loss": 3.4662, "step": 70760 }, { "epoch": 0.0060672, "grad_norm": 0.7867432832717896, "learning_rate": 1.8441272235625903e-05, "loss": 3.3226, "step": 70770 }, { "epoch": 0.0060928, "grad_norm": 0.7723312377929688, "learning_rate": 1.8440840463903422e-05, "loss": 3.255, "step": 70780 }, { "epoch": 0.0061184, "grad_norm": 0.8166115283966064, "learning_rate": 1.8440408637444337e-05, "loss": 3.2128, "step": 70790 }, { "epoch": 0.006144, "grad_norm": 1.1230679750442505, "learning_rate": 1.8439976756251443e-05, "loss": 3.4022, "step": 70800 }, { "epoch": 0.0061696, "grad_norm": 0.69907546043396, "learning_rate": 1.8439544820327542e-05, "loss": 3.4225, "step": 70810 }, { "epoch": 0.0061952, "grad_norm": 0.8555886745452881, "learning_rate": 1.843911282967544e-05, "loss": 3.2767, "step": 70820 }, { "epoch": 0.0062208, "grad_norm": 0.7494130730628967, "learning_rate": 1.8438680784297932e-05, "loss": 3.2201, "step": 70830 }, { "epoch": 0.0062464, "grad_norm": 0.7402685284614563, "learning_rate": 1.8438248684197823e-05, "loss": 3.4229, "step": 70840 }, { "epoch": 0.006272, "grad_norm": 0.7134379148483276, "learning_rate": 1.843781652937791e-05, "loss": 3.1795, "step": 70850 }, { "epoch": 0.0062976, "grad_norm": 0.8107073903083801, "learning_rate": 1.8437384319841007e-05, "loss": 3.2334, "step": 70860 }, { "epoch": 0.0063232, "grad_norm": 0.7546395659446716, "learning_rate": 1.84369520555899e-05, "loss": 3.2402, "step": 70870 }, { "epoch": 0.0063488, "grad_norm": 0.8856091499328613, "learning_rate": 1.8436519736627407e-05, "loss": 3.2116, "step": 70880 }, { "epoch": 0.0063744, "grad_norm": 0.6944446563720703, "learning_rate": 1.8436087362956324e-05, "loss": 3.2756, "step": 70890 }, { "epoch": 0.0064, "grad_norm": 1.0955140590667725, "learning_rate": 1.8435654934579455e-05, "loss": 3.2646, "step": 70900 }, { "epoch": 0.0064256, "grad_norm": 0.9934340715408325, "learning_rate": 1.843522245149961e-05, "loss": 3.3972, "step": 70910 }, { "epoch": 0.0064512, "grad_norm": 0.8156650066375732, "learning_rate": 1.8434789913719585e-05, "loss": 3.2884, "step": 70920 }, { "epoch": 0.0064768, "grad_norm": 0.7875573039054871, "learning_rate": 1.8434357321242192e-05, "loss": 3.2703, "step": 70930 }, { "epoch": 0.0065024, "grad_norm": 0.7551320195198059, "learning_rate": 1.843392467407023e-05, "loss": 3.3448, "step": 70940 }, { "epoch": 0.006528, "grad_norm": 1.052786111831665, "learning_rate": 1.8433491972206508e-05, "loss": 3.3292, "step": 70950 }, { "epoch": 0.0065536, "grad_norm": 0.753853976726532, "learning_rate": 1.8433059215653835e-05, "loss": 3.3518, "step": 70960 }, { "epoch": 0.0065792, "grad_norm": 0.7947431802749634, "learning_rate": 1.8432626404415015e-05, "loss": 3.3277, "step": 70970 }, { "epoch": 0.0066048, "grad_norm": 0.9517746567726135, "learning_rate": 1.843219353849285e-05, "loss": 3.2235, "step": 70980 }, { "epoch": 0.0066304, "grad_norm": 1.055096983909607, "learning_rate": 1.8431760617890155e-05, "loss": 3.2401, "step": 70990 }, { "epoch": 0.006656, "grad_norm": 0.8365855813026428, "learning_rate": 1.843132764260973e-05, "loss": 3.1484, "step": 71000 }, { "epoch": 0.0066816, "grad_norm": 1.0947476625442505, "learning_rate": 1.8430894612654388e-05, "loss": 3.3061, "step": 71010 }, { "epoch": 0.0067072, "grad_norm": 0.957740068435669, "learning_rate": 1.843046152802693e-05, "loss": 3.3136, "step": 71020 }, { "epoch": 0.0067328, "grad_norm": 0.7284828424453735, "learning_rate": 1.8430028388730177e-05, "loss": 3.3836, "step": 71030 }, { "epoch": 0.0067584, "grad_norm": 0.7598835229873657, "learning_rate": 1.8429595194766925e-05, "loss": 3.3644, "step": 71040 }, { "epoch": 0.006784, "grad_norm": 0.8221083283424377, "learning_rate": 1.842916194613999e-05, "loss": 3.1408, "step": 71050 }, { "epoch": 0.0068096, "grad_norm": 0.9917402267456055, "learning_rate": 1.842872864285218e-05, "loss": 3.1151, "step": 71060 }, { "epoch": 0.0068352, "grad_norm": 0.7455815076828003, "learning_rate": 1.8428295284906305e-05, "loss": 3.2159, "step": 71070 }, { "epoch": 0.0068608, "grad_norm": 0.8491024374961853, "learning_rate": 1.842786187230518e-05, "loss": 3.2864, "step": 71080 }, { "epoch": 0.0068864, "grad_norm": 0.6852574944496155, "learning_rate": 1.8427428405051605e-05, "loss": 3.1536, "step": 71090 }, { "epoch": 0.006912, "grad_norm": 0.7640900015830994, "learning_rate": 1.8426994883148394e-05, "loss": 3.2211, "step": 71100 }, { "epoch": 0.0069376, "grad_norm": 0.7873950600624084, "learning_rate": 1.842656130659837e-05, "loss": 3.3729, "step": 71110 }, { "epoch": 0.0069632, "grad_norm": 0.7721417546272278, "learning_rate": 1.8426127675404326e-05, "loss": 3.2204, "step": 71120 }, { "epoch": 0.0069888, "grad_norm": 0.8799095153808594, "learning_rate": 1.8425693989569086e-05, "loss": 3.3125, "step": 71130 }, { "epoch": 0.0070144, "grad_norm": 0.7996084690093994, "learning_rate": 1.842526024909546e-05, "loss": 3.0508, "step": 71140 }, { "epoch": 0.00704, "grad_norm": 0.8659098148345947, "learning_rate": 1.8424826453986263e-05, "loss": 3.1588, "step": 71150 }, { "epoch": 0.0070656, "grad_norm": 0.7844110131263733, "learning_rate": 1.8424392604244303e-05, "loss": 3.2359, "step": 71160 }, { "epoch": 0.0070912, "grad_norm": 0.8964323401451111, "learning_rate": 1.84239586998724e-05, "loss": 3.3133, "step": 71170 }, { "epoch": 0.0071168, "grad_norm": 0.7162593007087708, "learning_rate": 1.842352474087336e-05, "loss": 3.3333, "step": 71180 }, { "epoch": 0.0071424, "grad_norm": 0.8187772035598755, "learning_rate": 1.8423090727250002e-05, "loss": 3.3172, "step": 71190 }, { "epoch": 0.007168, "grad_norm": 0.8512953519821167, "learning_rate": 1.8422656659005137e-05, "loss": 3.1421, "step": 71200 }, { "epoch": 0.0071936, "grad_norm": 0.940229594707489, "learning_rate": 1.8422222536141584e-05, "loss": 3.3334, "step": 71210 }, { "epoch": 0.0072192, "grad_norm": 0.8220812082290649, "learning_rate": 1.8421788358662157e-05, "loss": 3.2142, "step": 71220 }, { "epoch": 0.0072448, "grad_norm": 0.7724579572677612, "learning_rate": 1.8421354126569667e-05, "loss": 3.2011, "step": 71230 }, { "epoch": 0.0072704, "grad_norm": 1.5443936586380005, "learning_rate": 1.8420919839866934e-05, "loss": 3.2905, "step": 71240 }, { "epoch": 0.007296, "grad_norm": 0.8516893982887268, "learning_rate": 1.8420485498556776e-05, "loss": 3.5045, "step": 71250 }, { "epoch": 0.0073216, "grad_norm": 0.8151030540466309, "learning_rate": 1.842005110264201e-05, "loss": 3.0991, "step": 71260 }, { "epoch": 0.0073472, "grad_norm": 0.7976556420326233, "learning_rate": 1.8419616652125448e-05, "loss": 3.2241, "step": 71270 }, { "epoch": 0.0073728, "grad_norm": 0.8292360305786133, "learning_rate": 1.841918214700991e-05, "loss": 3.3822, "step": 71280 }, { "epoch": 0.0073984, "grad_norm": 1.075291395187378, "learning_rate": 1.841874758729821e-05, "loss": 3.2409, "step": 71290 }, { "epoch": 0.007424, "grad_norm": 0.8860562443733215, "learning_rate": 1.8418312972993173e-05, "loss": 3.1903, "step": 71300 }, { "epoch": 0.0074496, "grad_norm": 0.6723232269287109, "learning_rate": 1.8417878304097613e-05, "loss": 3.17, "step": 71310 }, { "epoch": 0.0074752, "grad_norm": 0.6889328956604004, "learning_rate": 1.8417443580614348e-05, "loss": 3.1928, "step": 71320 }, { "epoch": 0.0075008, "grad_norm": 3.439323902130127, "learning_rate": 1.84170088025462e-05, "loss": 3.0919, "step": 71330 }, { "epoch": 0.0075264, "grad_norm": 0.828589916229248, "learning_rate": 1.8416573969895986e-05, "loss": 3.1227, "step": 71340 }, { "epoch": 0.007552, "grad_norm": 0.7625077962875366, "learning_rate": 1.8416139082666528e-05, "loss": 3.4111, "step": 71350 }, { "epoch": 0.0075776, "grad_norm": 0.7048319578170776, "learning_rate": 1.8415704140860646e-05, "loss": 3.2109, "step": 71360 }, { "epoch": 0.0076032, "grad_norm": 1.1028257608413696, "learning_rate": 1.8415269144481156e-05, "loss": 3.1861, "step": 71370 }, { "epoch": 0.0076288, "grad_norm": 0.7460927963256836, "learning_rate": 1.841483409353088e-05, "loss": 3.3184, "step": 71380 }, { "epoch": 0.0076544, "grad_norm": 0.791642427444458, "learning_rate": 1.8414398988012646e-05, "loss": 3.1286, "step": 71390 }, { "epoch": 0.00768, "grad_norm": 0.7221861481666565, "learning_rate": 1.841396382792927e-05, "loss": 3.2305, "step": 71400 }, { "epoch": 0.0077056, "grad_norm": 0.8302741050720215, "learning_rate": 1.8413528613283573e-05, "loss": 3.3428, "step": 71410 }, { "epoch": 0.0077312, "grad_norm": 0.9148625135421753, "learning_rate": 1.8413093344078382e-05, "loss": 3.2762, "step": 71420 }, { "epoch": 0.0077568, "grad_norm": 0.7578990459442139, "learning_rate": 1.8412658020316515e-05, "loss": 3.0905, "step": 71430 }, { "epoch": 0.0077824, "grad_norm": 0.852791428565979, "learning_rate": 1.8412222642000796e-05, "loss": 3.2022, "step": 71440 }, { "epoch": 0.007808, "grad_norm": 1.3616275787353516, "learning_rate": 1.8411787209134047e-05, "loss": 3.6032, "step": 71450 }, { "epoch": 0.0078336, "grad_norm": 0.8247151374816895, "learning_rate": 1.84113517217191e-05, "loss": 3.2089, "step": 71460 }, { "epoch": 0.0078592, "grad_norm": 0.8402827978134155, "learning_rate": 1.8410916179758766e-05, "loss": 3.2655, "step": 71470 }, { "epoch": 0.0078848, "grad_norm": 0.7307063937187195, "learning_rate": 1.8410480583255877e-05, "loss": 3.2865, "step": 71480 }, { "epoch": 0.0079104, "grad_norm": 0.7411314845085144, "learning_rate": 1.841004493221326e-05, "loss": 3.2538, "step": 71490 }, { "epoch": 0.007936, "grad_norm": 0.9171808958053589, "learning_rate": 1.8409609226633733e-05, "loss": 3.1252, "step": 71500 }, { "epoch": 0.0079616, "grad_norm": 0.8764268755912781, "learning_rate": 1.8409173466520126e-05, "loss": 3.4365, "step": 71510 }, { "epoch": 0.0079872, "grad_norm": 0.8540515303611755, "learning_rate": 1.8408737651875265e-05, "loss": 3.2871, "step": 71520 }, { "epoch": 0.0080128, "grad_norm": 0.7729203104972839, "learning_rate": 1.8408301782701975e-05, "loss": 3.4381, "step": 71530 }, { "epoch": 0.0080384, "grad_norm": 1.3525208234786987, "learning_rate": 1.8407865859003086e-05, "loss": 3.5028, "step": 71540 }, { "epoch": 0.008064, "grad_norm": 1.7279223203659058, "learning_rate": 1.8407429880781415e-05, "loss": 3.4365, "step": 71550 }, { "epoch": 0.0080896, "grad_norm": 0.9929042458534241, "learning_rate": 1.8406993848039798e-05, "loss": 3.2208, "step": 71560 }, { "epoch": 0.0081152, "grad_norm": 1.1229631900787354, "learning_rate": 1.8406557760781063e-05, "loss": 3.4505, "step": 71570 }, { "epoch": 0.0081408, "grad_norm": 0.8932526111602783, "learning_rate": 1.8406121619008033e-05, "loss": 3.0851, "step": 71580 }, { "epoch": 0.0081664, "grad_norm": 0.7592490315437317, "learning_rate": 1.8405685422723537e-05, "loss": 3.3691, "step": 71590 }, { "epoch": 0.008192, "grad_norm": 0.8406972885131836, "learning_rate": 1.8405249171930407e-05, "loss": 3.2083, "step": 71600 }, { "epoch": 0.0082176, "grad_norm": 0.7736629843711853, "learning_rate": 1.8404812866631467e-05, "loss": 3.3201, "step": 71610 }, { "epoch": 0.0082432, "grad_norm": 0.8640697002410889, "learning_rate": 1.8404376506829554e-05, "loss": 3.1657, "step": 71620 }, { "epoch": 0.0082688, "grad_norm": 0.7916311621665955, "learning_rate": 1.8403940092527492e-05, "loss": 3.2084, "step": 71630 }, { "epoch": 0.0082944, "grad_norm": 0.7691082954406738, "learning_rate": 1.840350362372811e-05, "loss": 3.299, "step": 71640 }, { "epoch": 0.00832, "grad_norm": 0.8645432591438293, "learning_rate": 1.8403067100434238e-05, "loss": 3.4499, "step": 71650 }, { "epoch": 0.0083456, "grad_norm": 0.7844204306602478, "learning_rate": 1.840263052264871e-05, "loss": 3.437, "step": 71660 }, { "epoch": 0.0083712, "grad_norm": 0.8297649621963501, "learning_rate": 1.8402193890374362e-05, "loss": 3.4401, "step": 71670 }, { "epoch": 0.0083968, "grad_norm": 0.8243592977523804, "learning_rate": 1.8401757203614015e-05, "loss": 3.2428, "step": 71680 }, { "epoch": 0.0084224, "grad_norm": 0.8587361574172974, "learning_rate": 1.8401320462370506e-05, "loss": 3.3211, "step": 71690 }, { "epoch": 0.008448, "grad_norm": 0.7695870995521545, "learning_rate": 1.8400883666646668e-05, "loss": 3.2356, "step": 71700 }, { "epoch": 0.0084736, "grad_norm": 0.7415774464607239, "learning_rate": 1.840044681644533e-05, "loss": 3.2516, "step": 71710 }, { "epoch": 0.0084992, "grad_norm": 1.0383678674697876, "learning_rate": 1.840000991176933e-05, "loss": 3.245, "step": 71720 }, { "epoch": 0.0085248, "grad_norm": 1.3549238443374634, "learning_rate": 1.8399572952621495e-05, "loss": 3.1023, "step": 71730 }, { "epoch": 0.0085504, "grad_norm": 0.771801233291626, "learning_rate": 1.8399135939004663e-05, "loss": 3.2887, "step": 71740 }, { "epoch": 0.008576, "grad_norm": 0.7361639142036438, "learning_rate": 1.8398698870921667e-05, "loss": 3.4189, "step": 71750 }, { "epoch": 0.0086016, "grad_norm": 0.7345583438873291, "learning_rate": 1.839826174837534e-05, "loss": 3.3107, "step": 71760 }, { "epoch": 0.0086272, "grad_norm": 0.8862747550010681, "learning_rate": 1.8397824571368515e-05, "loss": 3.221, "step": 71770 }, { "epoch": 0.0086528, "grad_norm": 0.7248005270957947, "learning_rate": 1.8397387339904034e-05, "loss": 3.2359, "step": 71780 }, { "epoch": 0.0086784, "grad_norm": 1.9460426568984985, "learning_rate": 1.8396950053984727e-05, "loss": 3.2112, "step": 71790 }, { "epoch": 0.008704, "grad_norm": 0.7566145062446594, "learning_rate": 1.839651271361343e-05, "loss": 3.3181, "step": 71800 }, { "epoch": 0.0087296, "grad_norm": 0.9219651222229004, "learning_rate": 1.8396075318792982e-05, "loss": 3.4129, "step": 71810 }, { "epoch": 0.0087552, "grad_norm": 0.8102073073387146, "learning_rate": 1.839563786952621e-05, "loss": 3.146, "step": 71820 }, { "epoch": 0.0087808, "grad_norm": 2.5956366062164307, "learning_rate": 1.8395200365815965e-05, "loss": 3.3404, "step": 71830 }, { "epoch": 0.0088064, "grad_norm": 2.1223151683807373, "learning_rate": 1.8394762807665073e-05, "loss": 3.9479, "step": 71840 }, { "epoch": 0.008832, "grad_norm": 0.9164388179779053, "learning_rate": 1.8394325195076376e-05, "loss": 3.4024, "step": 71850 }, { "epoch": 0.0088576, "grad_norm": 0.9052456021308899, "learning_rate": 1.839388752805271e-05, "loss": 3.2371, "step": 71860 }, { "epoch": 0.0088832, "grad_norm": 0.8652441501617432, "learning_rate": 1.8393449806596916e-05, "loss": 3.185, "step": 71870 }, { "epoch": 0.0089088, "grad_norm": 0.7835156917572021, "learning_rate": 1.8393012030711834e-05, "loss": 3.3307, "step": 71880 }, { "epoch": 0.0089344, "grad_norm": 0.8175968527793884, "learning_rate": 1.8392574200400294e-05, "loss": 3.192, "step": 71890 }, { "epoch": 0.00896, "grad_norm": 0.7400397062301636, "learning_rate": 1.8392136315665142e-05, "loss": 3.2945, "step": 71900 }, { "epoch": 0.0089856, "grad_norm": 0.7460818886756897, "learning_rate": 1.8391698376509218e-05, "loss": 3.5992, "step": 71910 }, { "epoch": 0.0090112, "grad_norm": 0.8466747999191284, "learning_rate": 1.8391260382935357e-05, "loss": 3.2241, "step": 71920 }, { "epoch": 0.0090368, "grad_norm": 0.7364969849586487, "learning_rate": 1.8390822334946406e-05, "loss": 3.2538, "step": 71930 }, { "epoch": 0.0090624, "grad_norm": 0.7488901615142822, "learning_rate": 1.83903842325452e-05, "loss": 3.031, "step": 71940 }, { "epoch": 0.009088, "grad_norm": 0.8642361164093018, "learning_rate": 1.8389946075734585e-05, "loss": 3.2878, "step": 71950 }, { "epoch": 0.0091136, "grad_norm": 0.9940870404243469, "learning_rate": 1.8389507864517394e-05, "loss": 3.2969, "step": 71960 }, { "epoch": 0.0091392, "grad_norm": 0.8323397040367126, "learning_rate": 1.838906959889648e-05, "loss": 3.5424, "step": 71970 }, { "epoch": 0.0091648, "grad_norm": 1.0758451223373413, "learning_rate": 1.8388631278874678e-05, "loss": 3.2643, "step": 71980 }, { "epoch": 0.0091904, "grad_norm": 0.8183812499046326, "learning_rate": 1.8388192904454828e-05, "loss": 3.2708, "step": 71990 }, { "epoch": 0.009216, "grad_norm": 1.0096352100372314, "learning_rate": 1.8387754475639782e-05, "loss": 3.305, "step": 72000 }, { "epoch": 0.0092416, "grad_norm": 1.6417300701141357, "learning_rate": 1.8387315992432374e-05, "loss": 3.6126, "step": 72010 }, { "epoch": 0.0092672, "grad_norm": 0.8581698536872864, "learning_rate": 1.838687745483545e-05, "loss": 3.3136, "step": 72020 }, { "epoch": 0.0092928, "grad_norm": 1.1006673574447632, "learning_rate": 1.8386438862851858e-05, "loss": 3.4088, "step": 72030 }, { "epoch": 0.0093184, "grad_norm": 1.7084941864013672, "learning_rate": 1.8386000216484433e-05, "loss": 3.2172, "step": 72040 }, { "epoch": 0.009344, "grad_norm": 1.2037993669509888, "learning_rate": 1.838556151573603e-05, "loss": 3.2317, "step": 72050 }, { "epoch": 0.0093696, "grad_norm": 0.9111135005950928, "learning_rate": 1.838512276060949e-05, "loss": 3.1907, "step": 72060 }, { "epoch": 0.0093952, "grad_norm": 0.8336208462715149, "learning_rate": 1.8384683951107652e-05, "loss": 3.2094, "step": 72070 }, { "epoch": 0.0094208, "grad_norm": 0.8045077323913574, "learning_rate": 1.838424508723337e-05, "loss": 3.2206, "step": 72080 }, { "epoch": 0.0094464, "grad_norm": 0.7430946230888367, "learning_rate": 1.838380616898949e-05, "loss": 3.2795, "step": 72090 }, { "epoch": 0.009472, "grad_norm": 0.8559566140174866, "learning_rate": 1.8383367196378853e-05, "loss": 3.3463, "step": 72100 }, { "epoch": 0.0094976, "grad_norm": 0.9363788366317749, "learning_rate": 1.8382928169404306e-05, "loss": 3.1232, "step": 72110 }, { "epoch": 0.0095232, "grad_norm": 0.9242529273033142, "learning_rate": 1.83824890880687e-05, "loss": 3.2816, "step": 72120 }, { "epoch": 0.0095488, "grad_norm": 0.7474561333656311, "learning_rate": 1.838204995237488e-05, "loss": 3.1439, "step": 72130 }, { "epoch": 0.0095744, "grad_norm": 0.9186493158340454, "learning_rate": 1.8381610762325693e-05, "loss": 3.2135, "step": 72140 }, { "epoch": 0.0096, "grad_norm": 0.8162282109260559, "learning_rate": 1.8381171517923988e-05, "loss": 3.1221, "step": 72150 }, { "epoch": 0.0096256, "grad_norm": 0.7356681227684021, "learning_rate": 1.8380732219172614e-05, "loss": 3.0391, "step": 72160 }, { "epoch": 0.0096512, "grad_norm": 0.7824650406837463, "learning_rate": 1.8380292866074416e-05, "loss": 3.2983, "step": 72170 }, { "epoch": 0.0096768, "grad_norm": 0.8108005523681641, "learning_rate": 1.8379853458632245e-05, "loss": 3.3275, "step": 72180 }, { "epoch": 0.0097024, "grad_norm": 0.8124372959136963, "learning_rate": 1.8379413996848957e-05, "loss": 3.2177, "step": 72190 }, { "epoch": 0.009728, "grad_norm": 0.8176568150520325, "learning_rate": 1.8378974480727393e-05, "loss": 3.2825, "step": 72200 }, { "epoch": 0.0097536, "grad_norm": 0.852527916431427, "learning_rate": 1.8378534910270404e-05, "loss": 3.5084, "step": 72210 }, { "epoch": 0.0097792, "grad_norm": 0.8286017775535583, "learning_rate": 1.8378095285480846e-05, "loss": 3.2945, "step": 72220 }, { "epoch": 0.0098048, "grad_norm": 0.7835816740989685, "learning_rate": 1.8377655606361568e-05, "loss": 3.2786, "step": 72230 }, { "epoch": 0.0098304, "grad_norm": 0.7725091576576233, "learning_rate": 1.837721587291542e-05, "loss": 3.3549, "step": 72240 }, { "epoch": 0.009856, "grad_norm": 0.6484479308128357, "learning_rate": 1.837677608514525e-05, "loss": 3.3591, "step": 72250 }, { "epoch": 0.0098816, "grad_norm": 1.0268257856369019, "learning_rate": 1.8376336243053913e-05, "loss": 3.283, "step": 72260 }, { "epoch": 0.0099072, "grad_norm": 0.8655868172645569, "learning_rate": 1.8375896346644263e-05, "loss": 3.3506, "step": 72270 }, { "epoch": 0.0099328, "grad_norm": 0.8583236932754517, "learning_rate": 1.8375456395919154e-05, "loss": 3.2225, "step": 72280 }, { "epoch": 0.0099584, "grad_norm": 0.9387477040290833, "learning_rate": 1.8375016390881436e-05, "loss": 3.355, "step": 72290 }, { "epoch": 0.009984, "grad_norm": 0.924115777015686, "learning_rate": 1.8374576331533957e-05, "loss": 3.384, "step": 72300 }, { "epoch": 0.0100096, "grad_norm": 0.916127622127533, "learning_rate": 1.837413621787958e-05, "loss": 3.1958, "step": 72310 }, { "epoch": 0.0100352, "grad_norm": 1.1816624402999878, "learning_rate": 1.8373696049921156e-05, "loss": 3.395, "step": 72320 }, { "epoch": 0.0100608, "grad_norm": 0.7578452229499817, "learning_rate": 1.8373255827661537e-05, "loss": 3.3948, "step": 72330 }, { "epoch": 0.0100864, "grad_norm": 0.978600025177002, "learning_rate": 1.837281555110358e-05, "loss": 3.2248, "step": 72340 }, { "epoch": 0.010112, "grad_norm": 0.7973002791404724, "learning_rate": 1.8372375220250137e-05, "loss": 3.0442, "step": 72350 }, { "epoch": 0.0101376, "grad_norm": 0.8786894679069519, "learning_rate": 1.837193483510407e-05, "loss": 3.2331, "step": 72360 }, { "epoch": 0.0101632, "grad_norm": 0.7652470469474792, "learning_rate": 1.8371494395668228e-05, "loss": 3.3102, "step": 72370 }, { "epoch": 0.0101888, "grad_norm": 0.724944531917572, "learning_rate": 1.8371053901945472e-05, "loss": 3.0194, "step": 72380 }, { "epoch": 0.0102144, "grad_norm": 0.7439212203025818, "learning_rate": 1.8370613353938652e-05, "loss": 3.2542, "step": 72390 }, { "epoch": 0.01024, "grad_norm": 0.8595184087753296, "learning_rate": 1.8370172751650633e-05, "loss": 3.3018, "step": 72400 }, { "epoch": 2.56e-05, "grad_norm": 0.7517445683479309, "learning_rate": 1.836973209508427e-05, "loss": 3.2424, "step": 72410 }, { "epoch": 5.12e-05, "grad_norm": 0.974805474281311, "learning_rate": 1.836929138424241e-05, "loss": 3.2855, "step": 72420 }, { "epoch": 7.68e-05, "grad_norm": 0.682415783405304, "learning_rate": 1.8368850619127928e-05, "loss": 3.1858, "step": 72430 }, { "epoch": 0.0001024, "grad_norm": 0.7964872121810913, "learning_rate": 1.836840979974367e-05, "loss": 3.2037, "step": 72440 }, { "epoch": 0.000128, "grad_norm": 0.7409025430679321, "learning_rate": 1.83679689260925e-05, "loss": 3.0784, "step": 72450 }, { "epoch": 0.0001536, "grad_norm": 0.9074993133544922, "learning_rate": 1.8367527998177276e-05, "loss": 2.8787, "step": 72460 }, { "epoch": 0.0001792, "grad_norm": 0.8943837285041809, "learning_rate": 1.8367087016000856e-05, "loss": 3.4543, "step": 72470 }, { "epoch": 0.0002048, "grad_norm": 1.0407544374465942, "learning_rate": 1.83666459795661e-05, "loss": 3.0177, "step": 72480 }, { "epoch": 0.0002304, "grad_norm": 0.9619806408882141, "learning_rate": 1.8366204888875868e-05, "loss": 2.9846, "step": 72490 }, { "epoch": 0.000256, "grad_norm": 0.7746164798736572, "learning_rate": 1.836576374393302e-05, "loss": 3.1824, "step": 72500 }, { "epoch": 0.0002816, "grad_norm": 0.7065103650093079, "learning_rate": 1.836532254474042e-05, "loss": 3.2268, "step": 72510 }, { "epoch": 0.0003072, "grad_norm": 0.8053888082504272, "learning_rate": 1.8364881291300928e-05, "loss": 3.469, "step": 72520 }, { "epoch": 0.0003328, "grad_norm": 1.402242660522461, "learning_rate": 1.8364439983617396e-05, "loss": 3.3265, "step": 72530 }, { "epoch": 0.0003584, "grad_norm": 0.8605962991714478, "learning_rate": 1.8363998621692703e-05, "loss": 3.142, "step": 72540 }, { "epoch": 0.000384, "grad_norm": 0.6981706619262695, "learning_rate": 1.8363557205529696e-05, "loss": 3.0778, "step": 72550 }, { "epoch": 0.0004096, "grad_norm": 0.9002684354782104, "learning_rate": 1.8363115735131246e-05, "loss": 3.4933, "step": 72560 }, { "epoch": 0.0004352, "grad_norm": 0.7329782843589783, "learning_rate": 1.8362674210500212e-05, "loss": 3.0509, "step": 72570 }, { "epoch": 0.0004608, "grad_norm": 1.0808359384536743, "learning_rate": 1.8362232631639458e-05, "loss": 3.0997, "step": 72580 }, { "epoch": 0.0004864, "grad_norm": 0.7085798382759094, "learning_rate": 1.8361790998551848e-05, "loss": 3.1868, "step": 72590 }, { "epoch": 0.000512, "grad_norm": 0.895729124546051, "learning_rate": 1.8361349311240248e-05, "loss": 3.1756, "step": 72600 }, { "epoch": 0.0005376, "grad_norm": 0.8788576722145081, "learning_rate": 1.8360907569707514e-05, "loss": 3.1841, "step": 72610 }, { "epoch": 0.0005632, "grad_norm": 0.7369155883789062, "learning_rate": 1.8360465773956524e-05, "loss": 3.1426, "step": 72620 }, { "epoch": 0.0005888, "grad_norm": 0.7963088154792786, "learning_rate": 1.8360023923990132e-05, "loss": 3.1211, "step": 72630 }, { "epoch": 0.0006144, "grad_norm": 0.8023513555526733, "learning_rate": 1.8359582019811206e-05, "loss": 3.8155, "step": 72640 }, { "epoch": 0.00064, "grad_norm": 1.142617106437683, "learning_rate": 1.8359140061422612e-05, "loss": 3.151, "step": 72650 }, { "epoch": 0.0006656, "grad_norm": 0.8430301547050476, "learning_rate": 1.8358698048827216e-05, "loss": 3.193, "step": 72660 }, { "epoch": 0.0006912, "grad_norm": 0.98072350025177, "learning_rate": 1.8358255982027885e-05, "loss": 3.0253, "step": 72670 }, { "epoch": 0.0007168, "grad_norm": 0.9792440533638, "learning_rate": 1.8357813861027488e-05, "loss": 3.0529, "step": 72680 }, { "epoch": 0.0007424, "grad_norm": 1.1320866346359253, "learning_rate": 1.8357371685828885e-05, "loss": 3.4581, "step": 72690 }, { "epoch": 0.000768, "grad_norm": 0.6905561089515686, "learning_rate": 1.8356929456434952e-05, "loss": 3.1202, "step": 72700 }, { "epoch": 0.0007936, "grad_norm": 0.7281345725059509, "learning_rate": 1.8356487172848553e-05, "loss": 3.2594, "step": 72710 }, { "epoch": 0.0008192, "grad_norm": 0.8384621739387512, "learning_rate": 1.8356044835072553e-05, "loss": 3.1542, "step": 72720 }, { "epoch": 0.0008448, "grad_norm": 0.7574378848075867, "learning_rate": 1.8355602443109823e-05, "loss": 3.3049, "step": 72730 }, { "epoch": 0.0008704, "grad_norm": 0.9336187839508057, "learning_rate": 1.835515999696323e-05, "loss": 3.3425, "step": 72740 }, { "epoch": 0.000896, "grad_norm": 0.8579084873199463, "learning_rate": 1.8354717496635646e-05, "loss": 3.2674, "step": 72750 }, { "epoch": 0.0009216, "grad_norm": 0.9884093999862671, "learning_rate": 1.8354274942129943e-05, "loss": 3.1685, "step": 72760 }, { "epoch": 0.0009472, "grad_norm": 1.012017011642456, "learning_rate": 1.8353832333448985e-05, "loss": 3.2097, "step": 72770 }, { "epoch": 0.0009728, "grad_norm": 0.9571955800056458, "learning_rate": 1.835338967059564e-05, "loss": 3.1874, "step": 72780 }, { "epoch": 0.0009984, "grad_norm": 0.7699541449546814, "learning_rate": 1.835294695357279e-05, "loss": 3.0815, "step": 72790 }, { "epoch": 0.001024, "grad_norm": 0.7201731204986572, "learning_rate": 1.83525041823833e-05, "loss": 3.1686, "step": 72800 }, { "epoch": 0.0010496, "grad_norm": 0.8173952102661133, "learning_rate": 1.8352061357030032e-05, "loss": 3.2131, "step": 72810 }, { "epoch": 0.0010752, "grad_norm": 0.8069842457771301, "learning_rate": 1.8351618477515872e-05, "loss": 3.1373, "step": 72820 }, { "epoch": 0.0011008, "grad_norm": 0.7675561308860779, "learning_rate": 1.8351175543843683e-05, "loss": 3.1104, "step": 72830 }, { "epoch": 0.0011264, "grad_norm": 0.9337911009788513, "learning_rate": 1.835073255601634e-05, "loss": 3.066, "step": 72840 }, { "epoch": 0.001152, "grad_norm": 0.8938767910003662, "learning_rate": 1.835028951403672e-05, "loss": 3.1439, "step": 72850 }, { "epoch": 0.0011776, "grad_norm": 0.7984544634819031, "learning_rate": 1.8349846417907687e-05, "loss": 2.8712, "step": 72860 }, { "epoch": 0.0012032, "grad_norm": 0.857464611530304, "learning_rate": 1.834940326763212e-05, "loss": 3.0338, "step": 72870 }, { "epoch": 0.0012288, "grad_norm": 1.0030982494354248, "learning_rate": 1.8348960063212894e-05, "loss": 3.3147, "step": 72880 }, { "epoch": 0.0012544, "grad_norm": 0.954794704914093, "learning_rate": 1.834851680465288e-05, "loss": 3.0793, "step": 72890 }, { "epoch": 0.00128, "grad_norm": 0.6708449125289917, "learning_rate": 1.834807349195495e-05, "loss": 3.1155, "step": 72900 }, { "epoch": 0.0013056, "grad_norm": 0.8418651819229126, "learning_rate": 1.8347630125121985e-05, "loss": 3.2567, "step": 72910 }, { "epoch": 0.0013312, "grad_norm": 0.8861457705497742, "learning_rate": 1.8347186704156857e-05, "loss": 3.26, "step": 72920 }, { "epoch": 0.0013568, "grad_norm": 0.8486770391464233, "learning_rate": 1.834674322906244e-05, "loss": 3.2367, "step": 72930 }, { "epoch": 0.0013824, "grad_norm": 0.7797120809555054, "learning_rate": 1.8346299699841614e-05, "loss": 3.1883, "step": 72940 }, { "epoch": 0.001408, "grad_norm": 0.949970006942749, "learning_rate": 1.8345856116497255e-05, "loss": 3.1351, "step": 72950 }, { "epoch": 0.0014336, "grad_norm": 0.6991974711418152, "learning_rate": 1.834541247903223e-05, "loss": 3.2299, "step": 72960 }, { "epoch": 0.0014592, "grad_norm": 1.0755115747451782, "learning_rate": 1.8344968787449428e-05, "loss": 2.9538, "step": 72970 }, { "epoch": 0.0014848, "grad_norm": 1.4795939922332764, "learning_rate": 1.834452504175172e-05, "loss": 2.8985, "step": 72980 }, { "epoch": 0.0015104, "grad_norm": 0.810732364654541, "learning_rate": 1.8344081241941986e-05, "loss": 2.8585, "step": 72990 }, { "epoch": 0.001536, "grad_norm": 0.7135228514671326, "learning_rate": 1.83436373880231e-05, "loss": 3.2255, "step": 73000 }, { "epoch": 0.0015616, "grad_norm": 2.0523109436035156, "learning_rate": 1.8343193479997943e-05, "loss": 3.3224, "step": 73010 }, { "epoch": 0.0015872, "grad_norm": 0.7256774306297302, "learning_rate": 1.83427495178694e-05, "loss": 3.5344, "step": 73020 }, { "epoch": 0.0016128, "grad_norm": 0.7970422506332397, "learning_rate": 1.8342305501640337e-05, "loss": 3.2296, "step": 73030 }, { "epoch": 0.0016384, "grad_norm": 0.7376766800880432, "learning_rate": 1.834186143131364e-05, "loss": 3.1808, "step": 73040 }, { "epoch": 0.001664, "grad_norm": 0.8476969003677368, "learning_rate": 1.834141730689219e-05, "loss": 3.1198, "step": 73050 }, { "epoch": 0.0016896, "grad_norm": 0.7706693410873413, "learning_rate": 1.8340973128378867e-05, "loss": 3.1835, "step": 73060 }, { "epoch": 0.0017152, "grad_norm": 0.9767358899116516, "learning_rate": 1.834052889577655e-05, "loss": 3.1662, "step": 73070 }, { "epoch": 0.0017408, "grad_norm": 0.7435544729232788, "learning_rate": 1.8340084609088118e-05, "loss": 3.1358, "step": 73080 }, { "epoch": 0.0017664, "grad_norm": 0.8727687001228333, "learning_rate": 1.8339640268316455e-05, "loss": 3.3673, "step": 73090 }, { "epoch": 0.001792, "grad_norm": 0.9309049844741821, "learning_rate": 1.833919587346444e-05, "loss": 3.1894, "step": 73100 }, { "epoch": 0.0018176, "grad_norm": 0.7547355890274048, "learning_rate": 1.8338751424534957e-05, "loss": 3.387, "step": 73110 }, { "epoch": 0.0018432, "grad_norm": 0.8297240734100342, "learning_rate": 1.8338306921530887e-05, "loss": 3.0946, "step": 73120 }, { "epoch": 0.0018688, "grad_norm": 0.8272106647491455, "learning_rate": 1.833786236445511e-05, "loss": 3.236, "step": 73130 }, { "epoch": 0.0018944, "grad_norm": 0.7459842562675476, "learning_rate": 1.833741775331052e-05, "loss": 2.7413, "step": 73140 }, { "epoch": 0.00192, "grad_norm": 0.7770636677742004, "learning_rate": 1.8336973088099985e-05, "loss": 3.0789, "step": 73150 }, { "epoch": 0.0019456, "grad_norm": 0.7092035412788391, "learning_rate": 1.8336528368826395e-05, "loss": 3.1901, "step": 73160 }, { "epoch": 0.0019712, "grad_norm": 0.7145258188247681, "learning_rate": 1.8336083595492635e-05, "loss": 3.241, "step": 73170 }, { "epoch": 0.0019968, "grad_norm": 0.920287549495697, "learning_rate": 1.8335638768101587e-05, "loss": 3.201, "step": 73180 }, { "epoch": 0.0020224, "grad_norm": 0.7539073824882507, "learning_rate": 1.833519388665614e-05, "loss": 3.207, "step": 73190 }, { "epoch": 0.002048, "grad_norm": 0.7335073947906494, "learning_rate": 1.8334748951159172e-05, "loss": 3.2101, "step": 73200 }, { "epoch": 0.0020736, "grad_norm": 0.781876802444458, "learning_rate": 1.8334303961613577e-05, "loss": 3.1558, "step": 73210 }, { "epoch": 0.0020992, "grad_norm": 1.1454533338546753, "learning_rate": 1.8333858918022232e-05, "loss": 3.3962, "step": 73220 }, { "epoch": 0.0021248, "grad_norm": 0.7576050758361816, "learning_rate": 1.833341382038803e-05, "loss": 3.2801, "step": 73230 }, { "epoch": 0.0021504, "grad_norm": 1.91326904296875, "learning_rate": 1.8332968668713855e-05, "loss": 3.3896, "step": 73240 }, { "epoch": 0.002176, "grad_norm": 0.7399184107780457, "learning_rate": 1.833252346300259e-05, "loss": 3.3302, "step": 73250 }, { "epoch": 0.0022016, "grad_norm": 1.1291292905807495, "learning_rate": 1.8332078203257123e-05, "loss": 3.1795, "step": 73260 }, { "epoch": 0.0022272, "grad_norm": 0.7970919609069824, "learning_rate": 1.8331632889480343e-05, "loss": 3.3265, "step": 73270 }, { "epoch": 0.0022528, "grad_norm": 0.8124542832374573, "learning_rate": 1.8331187521675144e-05, "loss": 3.3071, "step": 73280 }, { "epoch": 0.0022784, "grad_norm": 0.724261999130249, "learning_rate": 1.8330742099844404e-05, "loss": 3.0164, "step": 73290 }, { "epoch": 0.002304, "grad_norm": 0.8312084674835205, "learning_rate": 1.8330296623991014e-05, "loss": 3.2174, "step": 73300 }, { "epoch": 0.0023296, "grad_norm": 0.9822275638580322, "learning_rate": 1.8329851094117867e-05, "loss": 3.1209, "step": 73310 }, { "epoch": 0.0023552, "grad_norm": 0.9135428071022034, "learning_rate": 1.8329405510227848e-05, "loss": 3.2073, "step": 73320 }, { "epoch": 0.0023808, "grad_norm": 1.4572559595108032, "learning_rate": 1.832895987232385e-05, "loss": 3.1927, "step": 73330 }, { "epoch": 0.0024064, "grad_norm": 1.6813732385635376, "learning_rate": 1.8328514180408762e-05, "loss": 3.3374, "step": 73340 }, { "epoch": 0.002432, "grad_norm": 0.7471562027931213, "learning_rate": 1.832806843448547e-05, "loss": 3.0874, "step": 73350 }, { "epoch": 0.0024576, "grad_norm": 0.7340925931930542, "learning_rate": 1.8327622634556864e-05, "loss": 3.5509, "step": 73360 }, { "epoch": 0.0024832, "grad_norm": 0.8083986639976501, "learning_rate": 1.8327176780625843e-05, "loss": 3.262, "step": 73370 }, { "epoch": 0.0025088, "grad_norm": 1.0565468072891235, "learning_rate": 1.832673087269529e-05, "loss": 3.2278, "step": 73380 }, { "epoch": 0.0025344, "grad_norm": 0.869217574596405, "learning_rate": 1.8326284910768105e-05, "loss": 3.4261, "step": 73390 }, { "epoch": 0.00256, "grad_norm": 0.749905526638031, "learning_rate": 1.8325838894847175e-05, "loss": 3.3687, "step": 73400 }, { "epoch": 0.0025856, "grad_norm": 1.1586652994155884, "learning_rate": 1.8325392824935393e-05, "loss": 3.5291, "step": 73410 }, { "epoch": 0.0026112, "grad_norm": 0.765252411365509, "learning_rate": 1.832494670103565e-05, "loss": 3.1305, "step": 73420 }, { "epoch": 0.0026368, "grad_norm": 0.7530612349510193, "learning_rate": 1.832450052315084e-05, "loss": 3.2324, "step": 73430 }, { "epoch": 0.0026624, "grad_norm": 0.7355372905731201, "learning_rate": 1.8324054291283853e-05, "loss": 3.3708, "step": 73440 }, { "epoch": 0.002688, "grad_norm": 0.7347067594528198, "learning_rate": 1.832360800543759e-05, "loss": 3.2667, "step": 73450 }, { "epoch": 0.0027136, "grad_norm": 0.7792952656745911, "learning_rate": 1.832316166561494e-05, "loss": 3.3408, "step": 73460 }, { "epoch": 0.0027392, "grad_norm": 0.690035879611969, "learning_rate": 1.83227152718188e-05, "loss": 3.238, "step": 73470 }, { "epoch": 0.0027648, "grad_norm": 0.8005167841911316, "learning_rate": 1.8322268824052064e-05, "loss": 3.3448, "step": 73480 }, { "epoch": 0.0027904, "grad_norm": 0.8656469583511353, "learning_rate": 1.8321822322317624e-05, "loss": 2.9787, "step": 73490 }, { "epoch": 0.002816, "grad_norm": 0.7014950513839722, "learning_rate": 1.832137576661838e-05, "loss": 3.0463, "step": 73500 }, { "epoch": 0.0028416, "grad_norm": 1.3018656969070435, "learning_rate": 1.8320929156957223e-05, "loss": 3.1573, "step": 73510 }, { "epoch": 0.0028672, "grad_norm": 1.4308785200119019, "learning_rate": 1.8320482493337053e-05, "loss": 3.4722, "step": 73520 }, { "epoch": 0.0028928, "grad_norm": 0.7680463194847107, "learning_rate": 1.8320035775760766e-05, "loss": 3.3191, "step": 73530 }, { "epoch": 0.0029184, "grad_norm": 0.7986927032470703, "learning_rate": 1.8319589004231258e-05, "loss": 3.3732, "step": 73540 }, { "epoch": 0.002944, "grad_norm": 0.7177121639251709, "learning_rate": 1.8319142178751423e-05, "loss": 3.2526, "step": 73550 }, { "epoch": 0.0029696, "grad_norm": 1.0003077983856201, "learning_rate": 1.831869529932417e-05, "loss": 3.4202, "step": 73560 }, { "epoch": 0.0029952, "grad_norm": 0.8774159550666809, "learning_rate": 1.831824836595238e-05, "loss": 3.2794, "step": 73570 }, { "epoch": 0.0030208, "grad_norm": 1.3311340808868408, "learning_rate": 1.8317801378638966e-05, "loss": 3.3063, "step": 73580 }, { "epoch": 0.0030464, "grad_norm": 1.1073379516601562, "learning_rate": 1.8317354337386815e-05, "loss": 3.1744, "step": 73590 }, { "epoch": 0.003072, "grad_norm": 0.7672441005706787, "learning_rate": 1.831690724219884e-05, "loss": 3.0012, "step": 73600 }, { "epoch": 0.0030976, "grad_norm": 0.8485337495803833, "learning_rate": 1.8316460093077927e-05, "loss": 3.2166, "step": 73610 }, { "epoch": 0.0031232, "grad_norm": 0.8369320034980774, "learning_rate": 1.831601289002698e-05, "loss": 3.0736, "step": 73620 }, { "epoch": 0.0031488, "grad_norm": 0.7407512068748474, "learning_rate": 1.83155656330489e-05, "loss": 3.0716, "step": 73630 }, { "epoch": 0.0031744, "grad_norm": 0.7828805446624756, "learning_rate": 1.8315118322146585e-05, "loss": 3.2622, "step": 73640 }, { "epoch": 0.0032, "grad_norm": 0.7968101501464844, "learning_rate": 1.8314670957322943e-05, "loss": 3.3723, "step": 73650 }, { "epoch": 0.0032256, "grad_norm": 0.930447518825531, "learning_rate": 1.8314223538580863e-05, "loss": 3.1706, "step": 73660 }, { "epoch": 0.0032512, "grad_norm": 0.9351148009300232, "learning_rate": 1.8313776065923258e-05, "loss": 3.0541, "step": 73670 }, { "epoch": 0.0032768, "grad_norm": 1.029669165611267, "learning_rate": 1.8313328539353023e-05, "loss": 3.2094, "step": 73680 }, { "epoch": 0.0033024, "grad_norm": 0.7823483347892761, "learning_rate": 1.8312880958873064e-05, "loss": 3.2489, "step": 73690 }, { "epoch": 0.003328, "grad_norm": 0.8509366512298584, "learning_rate": 1.8312433324486275e-05, "loss": 3.4774, "step": 73700 }, { "epoch": 0.0033536, "grad_norm": 0.876122236251831, "learning_rate": 1.831198563619557e-05, "loss": 3.1867, "step": 73710 }, { "epoch": 0.0033792, "grad_norm": 0.8596144318580627, "learning_rate": 1.831153789400385e-05, "loss": 3.3083, "step": 73720 }, { "epoch": 0.0034048, "grad_norm": 0.8187572360038757, "learning_rate": 1.831109009791401e-05, "loss": 3.2287, "step": 73730 }, { "epoch": 0.0034304, "grad_norm": 1.4259659051895142, "learning_rate": 1.831064224792896e-05, "loss": 3.435, "step": 73740 }, { "epoch": 0.003456, "grad_norm": 0.9784606099128723, "learning_rate": 1.831019434405161e-05, "loss": 3.2784, "step": 73750 }, { "epoch": 0.0034816, "grad_norm": 0.805188000202179, "learning_rate": 1.8309746386284854e-05, "loss": 3.4066, "step": 73760 }, { "epoch": 0.0035072, "grad_norm": 0.7130529284477234, "learning_rate": 1.8309298374631602e-05, "loss": 3.1865, "step": 73770 }, { "epoch": 0.0035328, "grad_norm": 0.8597557544708252, "learning_rate": 1.8308850309094754e-05, "loss": 3.1425, "step": 73780 }, { "epoch": 0.0035584, "grad_norm": 0.778598427772522, "learning_rate": 1.8308402189677224e-05, "loss": 3.41, "step": 73790 }, { "epoch": 0.003584, "grad_norm": 0.6675491333007812, "learning_rate": 1.830795401638191e-05, "loss": 3.2991, "step": 73800 }, { "epoch": 0.0036096, "grad_norm": 0.834248960018158, "learning_rate": 1.8307505789211728e-05, "loss": 3.0477, "step": 73810 }, { "epoch": 0.0036352, "grad_norm": 0.7606270909309387, "learning_rate": 1.8307057508169575e-05, "loss": 3.0767, "step": 73820 }, { "epoch": 0.0036608, "grad_norm": 0.6834601759910583, "learning_rate": 1.8306609173258363e-05, "loss": 3.1679, "step": 73830 }, { "epoch": 0.0036864, "grad_norm": 0.7769909501075745, "learning_rate": 1.8306160784481e-05, "loss": 3.2507, "step": 73840 }, { "epoch": 0.003712, "grad_norm": 0.8151127099990845, "learning_rate": 1.830571234184039e-05, "loss": 3.0259, "step": 73850 }, { "epoch": 0.0037376, "grad_norm": 0.7844651937484741, "learning_rate": 1.830526384533944e-05, "loss": 3.2756, "step": 73860 }, { "epoch": 0.0037632, "grad_norm": 0.7828106880187988, "learning_rate": 1.8304815294981066e-05, "loss": 3.1279, "step": 73870 }, { "epoch": 0.0037888, "grad_norm": 1.0136643648147583, "learning_rate": 1.830436669076817e-05, "loss": 3.3766, "step": 73880 }, { "epoch": 0.0038144, "grad_norm": 0.9610869288444519, "learning_rate": 1.8303918032703666e-05, "loss": 3.4761, "step": 73890 }, { "epoch": 0.00384, "grad_norm": 0.848628580570221, "learning_rate": 1.8303469320790456e-05, "loss": 3.4144, "step": 73900 }, { "epoch": 0.0038656, "grad_norm": 0.7949399948120117, "learning_rate": 1.8303020555031453e-05, "loss": 3.0315, "step": 73910 }, { "epoch": 0.0038912, "grad_norm": 0.85963374376297, "learning_rate": 1.8302571735429574e-05, "loss": 3.2348, "step": 73920 }, { "epoch": 0.0039168, "grad_norm": 0.7291013598442078, "learning_rate": 1.830212286198772e-05, "loss": 3.1824, "step": 73930 }, { "epoch": 0.0039424, "grad_norm": 0.7669106125831604, "learning_rate": 1.8301673934708807e-05, "loss": 3.2078, "step": 73940 }, { "epoch": 0.003968, "grad_norm": 0.6470391750335693, "learning_rate": 1.8301224953595746e-05, "loss": 3.0041, "step": 73950 }, { "epoch": 0.0039936, "grad_norm": 1.0615177154541016, "learning_rate": 1.8300775918651444e-05, "loss": 3.1604, "step": 73960 }, { "epoch": 0.0040192, "grad_norm": 0.8098152875900269, "learning_rate": 1.8300326829878822e-05, "loss": 3.2336, "step": 73970 }, { "epoch": 0.0040448, "grad_norm": 0.7097730040550232, "learning_rate": 1.829987768728078e-05, "loss": 3.507, "step": 73980 }, { "epoch": 0.0040704, "grad_norm": 0.8331212997436523, "learning_rate": 1.829942849086024e-05, "loss": 3.2671, "step": 73990 }, { "epoch": 0.004096, "grad_norm": 1.2050379514694214, "learning_rate": 1.829897924062011e-05, "loss": 3.3034, "step": 74000 }, { "epoch": 0.0041216, "grad_norm": 0.7728243470191956, "learning_rate": 1.829852993656331e-05, "loss": 3.4628, "step": 74010 }, { "epoch": 0.0041472, "grad_norm": 0.8379797339439392, "learning_rate": 1.8298080578692742e-05, "loss": 3.2437, "step": 74020 }, { "epoch": 0.0041728, "grad_norm": 0.8101377487182617, "learning_rate": 1.829763116701133e-05, "loss": 3.2045, "step": 74030 }, { "epoch": 0.0041984, "grad_norm": 0.6533898711204529, "learning_rate": 1.829718170152199e-05, "loss": 3.2614, "step": 74040 }, { "epoch": 0.004224, "grad_norm": 0.7750204205513, "learning_rate": 1.8296732182227627e-05, "loss": 3.2793, "step": 74050 }, { "epoch": 0.0042496, "grad_norm": 0.7785443663597107, "learning_rate": 1.829628260913116e-05, "loss": 3.2896, "step": 74060 }, { "epoch": 0.0042752, "grad_norm": 0.8625756502151489, "learning_rate": 1.829583298223551e-05, "loss": 3.3334, "step": 74070 }, { "epoch": 0.0043008, "grad_norm": 0.8488019704818726, "learning_rate": 1.829538330154358e-05, "loss": 3.283, "step": 74080 }, { "epoch": 0.0043264, "grad_norm": 0.9351732730865479, "learning_rate": 1.82949335670583e-05, "loss": 3.2852, "step": 74090 }, { "epoch": 0.004352, "grad_norm": 0.7516289353370667, "learning_rate": 1.829448377878258e-05, "loss": 3.0743, "step": 74100 }, { "epoch": 2.56e-05, "grad_norm": 0.7988086342811584, "learning_rate": 1.8294033936719335e-05, "loss": 3.1219, "step": 74110 }, { "epoch": 5.12e-05, "grad_norm": 0.9086219072341919, "learning_rate": 1.8293584040871484e-05, "loss": 3.2024, "step": 74120 }, { "epoch": 7.68e-05, "grad_norm": 0.778191328048706, "learning_rate": 1.8293134091241948e-05, "loss": 3.1165, "step": 74130 }, { "epoch": 0.0001024, "grad_norm": 0.8641232252120972, "learning_rate": 1.8292684087833635e-05, "loss": 3.0737, "step": 74140 }, { "epoch": 0.000128, "grad_norm": 0.7319664359092712, "learning_rate": 1.8292234030649478e-05, "loss": 3.0597, "step": 74150 }, { "epoch": 0.0001536, "grad_norm": 0.7273725271224976, "learning_rate": 1.8291783919692382e-05, "loss": 3.1338, "step": 74160 }, { "epoch": 0.0001792, "grad_norm": 0.8088834881782532, "learning_rate": 1.829133375496527e-05, "loss": 3.1358, "step": 74170 }, { "epoch": 0.0002048, "grad_norm": 0.7310231328010559, "learning_rate": 1.8290883536471063e-05, "loss": 3.4178, "step": 74180 }, { "epoch": 0.0002304, "grad_norm": 0.7937026619911194, "learning_rate": 1.829043326421268e-05, "loss": 2.9135, "step": 74190 }, { "epoch": 0.000256, "grad_norm": 0.8080084919929504, "learning_rate": 1.8289982938193046e-05, "loss": 2.8194, "step": 74200 }, { "epoch": 0.0002816, "grad_norm": 0.8259453177452087, "learning_rate": 1.828953255841507e-05, "loss": 3.3125, "step": 74210 }, { "epoch": 0.0003072, "grad_norm": 0.7728373408317566, "learning_rate": 1.8289082124881677e-05, "loss": 3.3062, "step": 74220 }, { "epoch": 0.0003328, "grad_norm": 1.039847493171692, "learning_rate": 1.8288631637595795e-05, "loss": 3.2433, "step": 74230 }, { "epoch": 0.0003584, "grad_norm": 0.8606554269790649, "learning_rate": 1.8288181096560333e-05, "loss": 3.2139, "step": 74240 }, { "epoch": 0.000384, "grad_norm": 0.6366021633148193, "learning_rate": 1.8287730501778225e-05, "loss": 3.0864, "step": 74250 }, { "epoch": 0.0004096, "grad_norm": 0.8837344646453857, "learning_rate": 1.8287279853252388e-05, "loss": 3.1534, "step": 74260 }, { "epoch": 0.0004352, "grad_norm": 0.7964410781860352, "learning_rate": 1.828682915098574e-05, "loss": 3.1192, "step": 74270 }, { "epoch": 0.0004608, "grad_norm": 0.7130581736564636, "learning_rate": 1.8286378394981206e-05, "loss": 3.1861, "step": 74280 }, { "epoch": 0.0004864, "grad_norm": 0.7417377233505249, "learning_rate": 1.8285927585241715e-05, "loss": 2.9161, "step": 74290 }, { "epoch": 0.000512, "grad_norm": 0.8739587068557739, "learning_rate": 1.8285476721770182e-05, "loss": 3.4388, "step": 74300 }, { "epoch": 0.0005376, "grad_norm": 0.8799073100090027, "learning_rate": 1.8285025804569536e-05, "loss": 3.5573, "step": 74310 }, { "epoch": 0.0005632, "grad_norm": 0.9561094045639038, "learning_rate": 1.8284574833642698e-05, "loss": 3.3227, "step": 74320 }, { "epoch": 0.0005888, "grad_norm": 0.8388364315032959, "learning_rate": 1.8284123808992596e-05, "loss": 3.2326, "step": 74330 }, { "epoch": 0.0006144, "grad_norm": 0.7970450520515442, "learning_rate": 1.8283672730622152e-05, "loss": 3.1412, "step": 74340 }, { "epoch": 0.00064, "grad_norm": 0.7734302878379822, "learning_rate": 1.8283221598534293e-05, "loss": 2.9651, "step": 74350 }, { "epoch": 0.0006656, "grad_norm": 0.8666585683822632, "learning_rate": 1.8282770412731944e-05, "loss": 3.1461, "step": 74360 }, { "epoch": 0.0006912, "grad_norm": 0.7354971766471863, "learning_rate": 1.828231917321803e-05, "loss": 3.3163, "step": 74370 }, { "epoch": 0.0007168, "grad_norm": 0.8417108654975891, "learning_rate": 1.8281867879995473e-05, "loss": 3.1026, "step": 74380 }, { "epoch": 0.0007424, "grad_norm": 0.7650855183601379, "learning_rate": 1.8281416533067207e-05, "loss": 3.1743, "step": 74390 }, { "epoch": 0.000768, "grad_norm": 0.9270594120025635, "learning_rate": 1.8280965132436153e-05, "loss": 3.1242, "step": 74400 }, { "epoch": 0.0007936, "grad_norm": 0.8016295433044434, "learning_rate": 1.8280513678105244e-05, "loss": 3.0363, "step": 74410 }, { "epoch": 0.0008192, "grad_norm": 1.330177903175354, "learning_rate": 1.8280062170077406e-05, "loss": 3.0626, "step": 74420 }, { "epoch": 0.0008448, "grad_norm": 0.9519163370132446, "learning_rate": 1.8279610608355564e-05, "loss": 3.388, "step": 74430 }, { "epoch": 0.0008704, "grad_norm": 0.6997511386871338, "learning_rate": 1.8279158992942644e-05, "loss": 3.162, "step": 74440 }, { "epoch": 0.000896, "grad_norm": 0.7269542217254639, "learning_rate": 1.827870732384158e-05, "loss": 3.0239, "step": 74450 }, { "epoch": 0.0009216, "grad_norm": 1.2633030414581299, "learning_rate": 1.8278255601055302e-05, "loss": 3.1313, "step": 74460 }, { "epoch": 0.0009472, "grad_norm": 0.8491451740264893, "learning_rate": 1.8277803824586738e-05, "loss": 3.1143, "step": 74470 }, { "epoch": 0.0009728, "grad_norm": 0.9868881106376648, "learning_rate": 1.8277351994438813e-05, "loss": 3.296, "step": 74480 }, { "epoch": 0.0009984, "grad_norm": 0.83919757604599, "learning_rate": 1.827690011061446e-05, "loss": 3.4305, "step": 74490 }, { "epoch": 0.001024, "grad_norm": 0.7888506650924683, "learning_rate": 1.827644817311661e-05, "loss": 3.2496, "step": 74500 }, { "epoch": 0.0010496, "grad_norm": 1.4184108972549438, "learning_rate": 1.827599618194819e-05, "loss": 3.1789, "step": 74510 }, { "epoch": 0.0010752, "grad_norm": 1.9290130138397217, "learning_rate": 1.827554413711214e-05, "loss": 2.9721, "step": 74520 }, { "epoch": 0.0011008, "grad_norm": 0.7237392067909241, "learning_rate": 1.8275092038611382e-05, "loss": 3.1524, "step": 74530 }, { "epoch": 0.0011264, "grad_norm": 0.7804029583930969, "learning_rate": 1.8274639886448852e-05, "loss": 3.3052, "step": 74540 }, { "epoch": 0.001152, "grad_norm": 0.9217016100883484, "learning_rate": 1.8274187680627483e-05, "loss": 3.146, "step": 74550 }, { "epoch": 0.0011776, "grad_norm": 0.7757906317710876, "learning_rate": 1.8273735421150206e-05, "loss": 2.979, "step": 74560 }, { "epoch": 0.0012032, "grad_norm": 0.8702149391174316, "learning_rate": 1.827328310801995e-05, "loss": 2.8492, "step": 74570 }, { "epoch": 0.0012288, "grad_norm": 0.6737253069877625, "learning_rate": 1.8272830741239655e-05, "loss": 2.9359, "step": 74580 }, { "epoch": 0.0012544, "grad_norm": 0.8184847235679626, "learning_rate": 1.8272378320812253e-05, "loss": 2.9886, "step": 74590 }, { "epoch": 0.00128, "grad_norm": 0.8317732214927673, "learning_rate": 1.8271925846740672e-05, "loss": 3.1061, "step": 74600 }, { "epoch": 0.0013056, "grad_norm": 0.7985703349113464, "learning_rate": 1.827147331902785e-05, "loss": 3.2471, "step": 74610 }, { "epoch": 0.0013312, "grad_norm": 0.8242189884185791, "learning_rate": 1.8271020737676724e-05, "loss": 3.3571, "step": 74620 }, { "epoch": 0.0013568, "grad_norm": 0.8230469822883606, "learning_rate": 1.827056810269023e-05, "loss": 3.0907, "step": 74630 }, { "epoch": 0.0013824, "grad_norm": 0.758531928062439, "learning_rate": 1.8270115414071295e-05, "loss": 3.5339, "step": 74640 }, { "epoch": 0.001408, "grad_norm": 0.9237262606620789, "learning_rate": 1.826966267182286e-05, "loss": 3.3433, "step": 74650 }, { "epoch": 0.0014336, "grad_norm": 0.7629753351211548, "learning_rate": 1.8269209875947863e-05, "loss": 2.9407, "step": 74660 }, { "epoch": 0.0014592, "grad_norm": 0.759842038154602, "learning_rate": 1.8268757026449237e-05, "loss": 3.0304, "step": 74670 }, { "epoch": 0.0014848, "grad_norm": 1.0840086936950684, "learning_rate": 1.8268304123329915e-05, "loss": 2.8087, "step": 74680 }, { "epoch": 0.0015104, "grad_norm": 1.0886108875274658, "learning_rate": 1.8267851166592842e-05, "loss": 3.0837, "step": 74690 }, { "epoch": 0.001536, "grad_norm": 0.6905674934387207, "learning_rate": 1.8267398156240948e-05, "loss": 3.0263, "step": 74700 }, { "epoch": 0.0015616, "grad_norm": 0.7907755970954895, "learning_rate": 1.826694509227718e-05, "loss": 3.2611, "step": 74710 }, { "epoch": 0.0015872, "grad_norm": 0.925965428352356, "learning_rate": 1.826649197470447e-05, "loss": 3.0596, "step": 74720 }, { "epoch": 0.0016128, "grad_norm": 0.7702814936637878, "learning_rate": 1.8266038803525752e-05, "loss": 3.0201, "step": 74730 }, { "epoch": 0.0016384, "grad_norm": 0.6825237274169922, "learning_rate": 1.826558557874397e-05, "loss": 3.1937, "step": 74740 }, { "epoch": 0.001664, "grad_norm": 0.750316858291626, "learning_rate": 1.8265132300362064e-05, "loss": 3.0345, "step": 74750 }, { "epoch": 0.0016896, "grad_norm": 0.7700313925743103, "learning_rate": 1.826467896838297e-05, "loss": 3.198, "step": 74760 }, { "epoch": 0.0017152, "grad_norm": 0.72023606300354, "learning_rate": 1.826422558280963e-05, "loss": 3.2945, "step": 74770 }, { "epoch": 0.0017408, "grad_norm": 0.7081253528594971, "learning_rate": 1.8263772143644983e-05, "loss": 3.2656, "step": 74780 }, { "epoch": 0.0017664, "grad_norm": 0.8046950101852417, "learning_rate": 1.8263318650891973e-05, "loss": 2.95, "step": 74790 }, { "epoch": 0.001792, "grad_norm": 0.6966399550437927, "learning_rate": 1.8262865104553534e-05, "loss": 2.7911, "step": 74800 }, { "epoch": 2.56e-05, "grad_norm": 0.8088213801383972, "learning_rate": 1.826241150463261e-05, "loss": 3.0456, "step": 74810 }, { "epoch": 5.12e-05, "grad_norm": 0.9059620499610901, "learning_rate": 1.8261957851132148e-05, "loss": 3.1095, "step": 74820 }, { "epoch": 7.68e-05, "grad_norm": 0.6980662941932678, "learning_rate": 1.8261504144055082e-05, "loss": 3.1008, "step": 74830 }, { "epoch": 0.0001024, "grad_norm": 0.7694466710090637, "learning_rate": 1.8261050383404358e-05, "loss": 3.0962, "step": 74840 }, { "epoch": 0.000128, "grad_norm": 0.7751249670982361, "learning_rate": 1.8260596569182917e-05, "loss": 3.0557, "step": 74850 }, { "epoch": 0.0001536, "grad_norm": 0.7152256369590759, "learning_rate": 1.8260142701393702e-05, "loss": 3.0299, "step": 74860 }, { "epoch": 0.0001792, "grad_norm": 1.656860589981079, "learning_rate": 1.825968878003966e-05, "loss": 3.1244, "step": 74870 }, { "epoch": 0.0002048, "grad_norm": 0.953921914100647, "learning_rate": 1.8259234805123728e-05, "loss": 3.1873, "step": 74880 }, { "epoch": 0.0002304, "grad_norm": 0.9248693585395813, "learning_rate": 1.8258780776648857e-05, "loss": 3.0198, "step": 74890 }, { "epoch": 0.000256, "grad_norm": 0.7219067811965942, "learning_rate": 1.8258326694617983e-05, "loss": 2.8935, "step": 74900 }, { "epoch": 0.0002816, "grad_norm": 0.8435797095298767, "learning_rate": 1.8257872559034058e-05, "loss": 3.4893, "step": 74910 }, { "epoch": 0.0003072, "grad_norm": 0.9260576963424683, "learning_rate": 1.8257418369900025e-05, "loss": 3.0745, "step": 74920 }, { "epoch": 0.0003328, "grad_norm": 0.906243622303009, "learning_rate": 1.825696412721883e-05, "loss": 3.0189, "step": 74930 }, { "epoch": 0.0003584, "grad_norm": 0.7700140476226807, "learning_rate": 1.825650983099341e-05, "loss": 2.8842, "step": 74940 }, { "epoch": 0.000384, "grad_norm": 0.7472595572471619, "learning_rate": 1.8256055481226722e-05, "loss": 2.8697, "step": 74950 }, { "epoch": 0.0004096, "grad_norm": 0.6617818474769592, "learning_rate": 1.825560107792171e-05, "loss": 3.2793, "step": 74960 }, { "epoch": 0.0004352, "grad_norm": 0.8075375556945801, "learning_rate": 1.8255146621081318e-05, "loss": 3.2466, "step": 74970 }, { "epoch": 0.0004608, "grad_norm": 0.8748890161514282, "learning_rate": 1.8254692110708495e-05, "loss": 3.0702, "step": 74980 }, { "epoch": 0.0004864, "grad_norm": 0.8584890961647034, "learning_rate": 1.8254237546806185e-05, "loss": 3.0418, "step": 74990 }, { "epoch": 0.000512, "grad_norm": 0.7017234563827515, "learning_rate": 1.825378292937734e-05, "loss": 3.1392, "step": 75000 }, { "epoch": 0.0005376, "grad_norm": 0.7879912257194519, "learning_rate": 1.8253328258424904e-05, "loss": 3.1812, "step": 75010 }, { "epoch": 0.0005632, "grad_norm": 0.8942275643348694, "learning_rate": 1.8252873533951832e-05, "loss": 2.9354, "step": 75020 }, { "epoch": 0.0005888, "grad_norm": 0.6918119788169861, "learning_rate": 1.8252418755961067e-05, "loss": 2.9634, "step": 75030 }, { "epoch": 0.0006144, "grad_norm": 0.7644297480583191, "learning_rate": 1.825196392445556e-05, "loss": 3.0379, "step": 75040 }, { "epoch": 0.00064, "grad_norm": 0.8586354851722717, "learning_rate": 1.8251509039438255e-05, "loss": 3.1353, "step": 75050 }, { "epoch": 0.0006656, "grad_norm": 0.8252182006835938, "learning_rate": 1.825105410091211e-05, "loss": 3.1043, "step": 75060 }, { "epoch": 0.0006912, "grad_norm": 0.8007291555404663, "learning_rate": 1.8250599108880073e-05, "loss": 2.9829, "step": 75070 }, { "epoch": 0.0007168, "grad_norm": 1.5977078676223755, "learning_rate": 1.825014406334509e-05, "loss": 2.9527, "step": 75080 }, { "epoch": 0.0007424, "grad_norm": 0.9643645286560059, "learning_rate": 1.824968896431012e-05, "loss": 3.3868, "step": 75090 }, { "epoch": 0.000768, "grad_norm": 1.505159616470337, "learning_rate": 1.824923381177811e-05, "loss": 3.1902, "step": 75100 }, { "epoch": 0.0007936, "grad_norm": 0.9664756059646606, "learning_rate": 1.8248778605752008e-05, "loss": 3.1108, "step": 75110 }, { "epoch": 0.0008192, "grad_norm": 1.1528621912002563, "learning_rate": 1.8248323346234772e-05, "loss": 3.2829, "step": 75120 }, { "epoch": 0.0008448, "grad_norm": 0.9156306385993958, "learning_rate": 1.824786803322935e-05, "loss": 3.3343, "step": 75130 }, { "epoch": 0.0008704, "grad_norm": 0.9788336157798767, "learning_rate": 1.824741266673869e-05, "loss": 3.1961, "step": 75140 }, { "epoch": 0.000896, "grad_norm": 0.671384334564209, "learning_rate": 1.824695724676576e-05, "loss": 3.0358, "step": 75150 }, { "epoch": 0.0009216, "grad_norm": 0.795287013053894, "learning_rate": 1.82465017733135e-05, "loss": 2.7701, "step": 75160 }, { "epoch": 0.0009472, "grad_norm": 1.2359514236450195, "learning_rate": 1.8246046246384866e-05, "loss": 3.1302, "step": 75170 }, { "epoch": 0.0009728, "grad_norm": 0.7935880422592163, "learning_rate": 1.824559066598282e-05, "loss": 3.8476, "step": 75180 }, { "epoch": 0.0009984, "grad_norm": 0.6947766542434692, "learning_rate": 1.8245135032110302e-05, "loss": 3.101, "step": 75190 }, { "epoch": 0.001024, "grad_norm": 0.7782139182090759, "learning_rate": 1.8244679344770278e-05, "loss": 2.914, "step": 75200 }, { "epoch": 0.0010496, "grad_norm": 0.6906655430793762, "learning_rate": 1.82442236039657e-05, "loss": 3.0022, "step": 75210 }, { "epoch": 0.0010752, "grad_norm": 1.7032614946365356, "learning_rate": 1.8243767809699526e-05, "loss": 2.8605, "step": 75220 }, { "epoch": 0.0011008, "grad_norm": 0.7198454141616821, "learning_rate": 1.82433119619747e-05, "loss": 2.9591, "step": 75230 }, { "epoch": 0.0011264, "grad_norm": 0.8067780137062073, "learning_rate": 1.8242856060794195e-05, "loss": 2.9964, "step": 75240 }, { "epoch": 0.001152, "grad_norm": 0.8867791891098022, "learning_rate": 1.8242400106160953e-05, "loss": 3.0109, "step": 75250 }, { "epoch": 0.0011776, "grad_norm": 0.7071640491485596, "learning_rate": 1.824194409807794e-05, "loss": 3.0709, "step": 75260 }, { "epoch": 0.0012032, "grad_norm": 0.7002399563789368, "learning_rate": 1.824148803654811e-05, "loss": 2.8878, "step": 75270 }, { "epoch": 0.0012288, "grad_norm": 0.6641512513160706, "learning_rate": 1.824103192157442e-05, "loss": 3.0056, "step": 75280 }, { "epoch": 0.0012544, "grad_norm": 0.779098391532898, "learning_rate": 1.824057575315983e-05, "loss": 3.0445, "step": 75290 }, { "epoch": 0.00128, "grad_norm": 0.7851731777191162, "learning_rate": 1.8240119531307295e-05, "loss": 3.0552, "step": 75300 }, { "epoch": 0.0013056, "grad_norm": 0.7762991786003113, "learning_rate": 1.8239663256019774e-05, "loss": 3.2872, "step": 75310 }, { "epoch": 0.0013312, "grad_norm": 0.8250232934951782, "learning_rate": 1.823920692730023e-05, "loss": 3.0567, "step": 75320 }, { "epoch": 0.0013568, "grad_norm": 0.7863538265228271, "learning_rate": 1.8238750545151617e-05, "loss": 3.1808, "step": 75330 }, { "epoch": 0.0013824, "grad_norm": 0.785383403301239, "learning_rate": 1.8238294109576893e-05, "loss": 3.2359, "step": 75340 }, { "epoch": 0.001408, "grad_norm": 1.1068265438079834, "learning_rate": 1.8237837620579026e-05, "loss": 3.1713, "step": 75350 }, { "epoch": 0.0014336, "grad_norm": 0.7968222498893738, "learning_rate": 1.823738107816097e-05, "loss": 2.9646, "step": 75360 }, { "epoch": 0.0014592, "grad_norm": 0.7733787894248962, "learning_rate": 1.823692448232568e-05, "loss": 2.9877, "step": 75370 }, { "epoch": 0.0014848, "grad_norm": 0.6605434417724609, "learning_rate": 1.823646783307613e-05, "loss": 2.817, "step": 75380 }, { "epoch": 0.0015104, "grad_norm": 1.048447608947754, "learning_rate": 1.8236011130415277e-05, "loss": 3.1697, "step": 75390 }, { "epoch": 0.001536, "grad_norm": 0.7933422923088074, "learning_rate": 1.823555437434608e-05, "loss": 3.2392, "step": 75400 }, { "epoch": 0.0015616, "grad_norm": 0.6329424977302551, "learning_rate": 1.8235097564871496e-05, "loss": 3.0374, "step": 75410 }, { "epoch": 0.0015872, "grad_norm": 0.8555164933204651, "learning_rate": 1.82346407019945e-05, "loss": 3.0768, "step": 75420 }, { "epoch": 0.0016128, "grad_norm": 0.6910673379898071, "learning_rate": 1.8234183785718044e-05, "loss": 3.0741, "step": 75430 }, { "epoch": 0.0016384, "grad_norm": 1.3333646059036255, "learning_rate": 1.8233726816045093e-05, "loss": 3.1145, "step": 75440 }, { "epoch": 0.001664, "grad_norm": 0.6467366218566895, "learning_rate": 1.8233269792978617e-05, "loss": 2.7481, "step": 75450 }, { "epoch": 0.0016896, "grad_norm": 0.949395477771759, "learning_rate": 1.823281271652157e-05, "loss": 3.0705, "step": 75460 }, { "epoch": 0.0017152, "grad_norm": 0.7314559817314148, "learning_rate": 1.8232355586676922e-05, "loss": 3.1934, "step": 75470 }, { "epoch": 0.0017408, "grad_norm": 0.7551335692405701, "learning_rate": 1.8231898403447636e-05, "loss": 3.2591, "step": 75480 }, { "epoch": 0.0017664, "grad_norm": 1.0511244535446167, "learning_rate": 1.823144116683668e-05, "loss": 3.2833, "step": 75490 }, { "epoch": 0.001792, "grad_norm": 0.7617582678794861, "learning_rate": 1.8230983876847012e-05, "loss": 3.2051, "step": 75500 }, { "epoch": 2.56e-05, "grad_norm": 0.7560384273529053, "learning_rate": 1.82305265334816e-05, "loss": 2.9654, "step": 75510 }, { "epoch": 5.12e-05, "grad_norm": 1.0329474210739136, "learning_rate": 1.8230069136743412e-05, "loss": 3.2025, "step": 75520 }, { "epoch": 7.68e-05, "grad_norm": 0.6753581762313843, "learning_rate": 1.8229611686635413e-05, "loss": 3.0291, "step": 75530 }, { "epoch": 0.0001024, "grad_norm": 0.879427433013916, "learning_rate": 1.822915418316057e-05, "loss": 3.1192, "step": 75540 }, { "epoch": 0.000128, "grad_norm": 0.7548391222953796, "learning_rate": 1.8228696626321846e-05, "loss": 3.1979, "step": 75550 }, { "epoch": 0.0001536, "grad_norm": 0.7970843315124512, "learning_rate": 1.8228239016122214e-05, "loss": 3.2103, "step": 75560 }, { "epoch": 0.0001792, "grad_norm": 0.8800549507141113, "learning_rate": 1.822778135256464e-05, "loss": 3.082, "step": 75570 }, { "epoch": 0.0002048, "grad_norm": 0.771321713924408, "learning_rate": 1.822732363565209e-05, "loss": 3.1897, "step": 75580 }, { "epoch": 0.0002304, "grad_norm": 0.8042849898338318, "learning_rate": 1.8226865865387532e-05, "loss": 3.0073, "step": 75590 }, { "epoch": 0.000256, "grad_norm": 0.9563852548599243, "learning_rate": 1.8226408041773934e-05, "loss": 3.0903, "step": 75600 }, { "epoch": 2.56e-05, "grad_norm": 0.8130218386650085, "learning_rate": 1.822595016481427e-05, "loss": 3.0507, "step": 75610 }, { "epoch": 5.12e-05, "grad_norm": 0.8872909545898438, "learning_rate": 1.8225492234511502e-05, "loss": 3.162, "step": 75620 }, { "epoch": 7.68e-05, "grad_norm": 0.8012165427207947, "learning_rate": 1.8225034250868606e-05, "loss": 3.154, "step": 75630 }, { "epoch": 0.0001024, "grad_norm": 0.762638509273529, "learning_rate": 1.8224576213888548e-05, "loss": 2.9919, "step": 75640 }, { "epoch": 0.000128, "grad_norm": 0.6734988689422607, "learning_rate": 1.8224118123574297e-05, "loss": 3.0528, "step": 75650 }, { "epoch": 0.0001536, "grad_norm": 0.7880064249038696, "learning_rate": 1.8223659979928826e-05, "loss": 3.0785, "step": 75660 }, { "epoch": 0.0001792, "grad_norm": 1.1051976680755615, "learning_rate": 1.8223201782955106e-05, "loss": 3.1799, "step": 75670 }, { "epoch": 0.0002048, "grad_norm": 1.5742188692092896, "learning_rate": 1.822274353265611e-05, "loss": 3.2423, "step": 75680 }, { "epoch": 0.0002304, "grad_norm": 0.8714513778686523, "learning_rate": 1.82222852290348e-05, "loss": 3.0484, "step": 75690 }, { "epoch": 0.000256, "grad_norm": 0.6855962872505188, "learning_rate": 1.8221826872094165e-05, "loss": 3.079, "step": 75700 }, { "epoch": 0.0002816, "grad_norm": 1.1724653244018555, "learning_rate": 1.8221368461837163e-05, "loss": 3.222, "step": 75710 }, { "epoch": 0.0003072, "grad_norm": 0.923553466796875, "learning_rate": 1.8220909998266775e-05, "loss": 3.2252, "step": 75720 }, { "epoch": 0.0003328, "grad_norm": 1.4113997220993042, "learning_rate": 1.8220451481385967e-05, "loss": 3.0861, "step": 75730 }, { "epoch": 0.0003584, "grad_norm": 0.7902104258537292, "learning_rate": 1.8219992911197716e-05, "loss": 3.2789, "step": 75740 }, { "epoch": 0.000384, "grad_norm": 0.9656028151512146, "learning_rate": 1.8219534287704997e-05, "loss": 2.9316, "step": 75750 }, { "epoch": 0.0004096, "grad_norm": 0.936703085899353, "learning_rate": 1.821907561091078e-05, "loss": 3.67, "step": 75760 }, { "epoch": 0.0004352, "grad_norm": 0.8212190270423889, "learning_rate": 1.8218616880818045e-05, "loss": 3.3463, "step": 75770 }, { "epoch": 0.0004608, "grad_norm": 0.9003755450248718, "learning_rate": 1.8218158097429764e-05, "loss": 3.3272, "step": 75780 }, { "epoch": 0.0004864, "grad_norm": 0.8351710438728333, "learning_rate": 1.821769926074891e-05, "loss": 3.0284, "step": 75790 }, { "epoch": 0.000512, "grad_norm": 1.2082324028015137, "learning_rate": 1.8217240370778463e-05, "loss": 2.863, "step": 75800 }, { "epoch": 2.56e-05, "grad_norm": 0.9235053062438965, "learning_rate": 1.8216781427521394e-05, "loss": 3.1224, "step": 75810 }, { "epoch": 5.12e-05, "grad_norm": 1.1136929988861084, "learning_rate": 1.821632243098068e-05, "loss": 3.1118, "step": 75820 }, { "epoch": 7.68e-05, "grad_norm": 0.8426021337509155, "learning_rate": 1.82158633811593e-05, "loss": 3.2135, "step": 75830 }, { "epoch": 0.0001024, "grad_norm": 0.7384558320045471, "learning_rate": 1.8215404278060228e-05, "loss": 3.0732, "step": 75840 }, { "epoch": 0.000128, "grad_norm": 0.8193477988243103, "learning_rate": 1.8214945121686444e-05, "loss": 3.2157, "step": 75850 }, { "epoch": 0.0001536, "grad_norm": 0.6541699171066284, "learning_rate": 1.8214485912040926e-05, "loss": 2.8573, "step": 75860 }, { "epoch": 0.0001792, "grad_norm": 0.8866637349128723, "learning_rate": 1.8214026649126647e-05, "loss": 3.1499, "step": 75870 }, { "epoch": 0.0002048, "grad_norm": 0.9279373288154602, "learning_rate": 1.8213567332946592e-05, "loss": 3.1005, "step": 75880 }, { "epoch": 0.0002304, "grad_norm": 0.8993390798568726, "learning_rate": 1.8213107963503734e-05, "loss": 3.176, "step": 75890 }, { "epoch": 0.000256, "grad_norm": 0.7886168360710144, "learning_rate": 1.8212648540801054e-05, "loss": 3.1173, "step": 75900 }, { "epoch": 2.56e-05, "grad_norm": 0.9589198231697083, "learning_rate": 1.821218906484153e-05, "loss": 2.8011, "step": 75910 }, { "epoch": 5.12e-05, "grad_norm": 0.8026543855667114, "learning_rate": 1.8211729535628138e-05, "loss": 3.1221, "step": 75920 }, { "epoch": 7.68e-05, "grad_norm": 2.997891426086426, "learning_rate": 1.821126995316387e-05, "loss": 3.3188, "step": 75930 }, { "epoch": 0.0001024, "grad_norm": 0.6805946230888367, "learning_rate": 1.821081031745169e-05, "loss": 3.3929, "step": 75940 }, { "epoch": 0.000128, "grad_norm": 0.7869473099708557, "learning_rate": 1.82103506284946e-05, "loss": 2.9759, "step": 75950 }, { "epoch": 0.0001536, "grad_norm": 0.6854243874549866, "learning_rate": 1.8209890886295556e-05, "loss": 2.9531, "step": 75960 }, { "epoch": 0.0001792, "grad_norm": 0.8536023497581482, "learning_rate": 1.8209431090857555e-05, "loss": 2.8214, "step": 75970 }, { "epoch": 0.0002048, "grad_norm": 0.8321555256843567, "learning_rate": 1.820897124218358e-05, "loss": 3.1999, "step": 75980 }, { "epoch": 0.0002304, "grad_norm": 0.8657951951026917, "learning_rate": 1.82085113402766e-05, "loss": 3.0747, "step": 75990 }, { "epoch": 0.000256, "grad_norm": 0.8169382810592651, "learning_rate": 1.8208051385139615e-05, "loss": 3.0935, "step": 76000 }, { "epoch": 0.0002816, "grad_norm": 0.850527822971344, "learning_rate": 1.820759137677559e-05, "loss": 3.1718, "step": 76010 }, { "epoch": 0.0003072, "grad_norm": 0.7942628860473633, "learning_rate": 1.820713131518752e-05, "loss": 3.4323, "step": 76020 }, { "epoch": 0.0003328, "grad_norm": 0.8327189683914185, "learning_rate": 1.8206671200378387e-05, "loss": 3.3534, "step": 76030 }, { "epoch": 0.0003584, "grad_norm": 0.7363366484642029, "learning_rate": 1.8206211032351165e-05, "loss": 3.0102, "step": 76040 }, { "epoch": 0.000384, "grad_norm": 0.7300564646720886, "learning_rate": 1.820575081110885e-05, "loss": 3.0402, "step": 76050 }, { "epoch": 0.0004096, "grad_norm": 0.7781769633293152, "learning_rate": 1.8205290536654427e-05, "loss": 3.3748, "step": 76060 }, { "epoch": 0.0004352, "grad_norm": 0.8287456035614014, "learning_rate": 1.820483020899087e-05, "loss": 3.2569, "step": 76070 }, { "epoch": 0.0004608, "grad_norm": 0.7185672521591187, "learning_rate": 1.8204369828121168e-05, "loss": 3.0982, "step": 76080 }, { "epoch": 0.0004864, "grad_norm": 0.7068267464637756, "learning_rate": 1.820390939404831e-05, "loss": 3.4818, "step": 76090 }, { "epoch": 0.000512, "grad_norm": 0.7697793245315552, "learning_rate": 1.820344890677528e-05, "loss": 3.0904, "step": 76100 }, { "epoch": 0.0005376, "grad_norm": 0.9493541717529297, "learning_rate": 1.8202988366305065e-05, "loss": 3.0945, "step": 76110 }, { "epoch": 0.0005632, "grad_norm": 1.3595187664031982, "learning_rate": 1.8202527772640647e-05, "loss": 3.029, "step": 76120 }, { "epoch": 0.0005888, "grad_norm": 0.8167591691017151, "learning_rate": 1.8202067125785022e-05, "loss": 3.0861, "step": 76130 }, { "epoch": 0.0006144, "grad_norm": 0.6795468330383301, "learning_rate": 1.8201606425741167e-05, "loss": 3.097, "step": 76140 }, { "epoch": 0.00064, "grad_norm": 0.7354022264480591, "learning_rate": 1.820114567251208e-05, "loss": 3.117, "step": 76150 }, { "epoch": 0.0006656, "grad_norm": 0.8482758402824402, "learning_rate": 1.8200684866100736e-05, "loss": 3.0872, "step": 76160 }, { "epoch": 0.0006912, "grad_norm": 0.7737576365470886, "learning_rate": 1.820022400651013e-05, "loss": 3.1503, "step": 76170 }, { "epoch": 0.0007168, "grad_norm": 0.7890433669090271, "learning_rate": 1.8199763093743254e-05, "loss": 3.0641, "step": 76180 }, { "epoch": 0.0007424, "grad_norm": 0.9610941410064697, "learning_rate": 1.8199302127803095e-05, "loss": 3.3017, "step": 76190 }, { "epoch": 0.000768, "grad_norm": 0.7859728932380676, "learning_rate": 1.8198841108692638e-05, "loss": 3.2366, "step": 76200 }, { "epoch": 0.0007936, "grad_norm": 0.7023124694824219, "learning_rate": 1.8198380036414875e-05, "loss": 3.1687, "step": 76210 }, { "epoch": 0.0008192, "grad_norm": 0.7709266543388367, "learning_rate": 1.8197918910972798e-05, "loss": 3.0952, "step": 76220 }, { "epoch": 0.0008448, "grad_norm": 0.6536075472831726, "learning_rate": 1.8197457732369398e-05, "loss": 2.9417, "step": 76230 }, { "epoch": 0.0008704, "grad_norm": 0.9477436542510986, "learning_rate": 1.8196996500607657e-05, "loss": 2.9585, "step": 76240 }, { "epoch": 0.000896, "grad_norm": 0.6916930675506592, "learning_rate": 1.8196535215690577e-05, "loss": 3.2668, "step": 76250 }, { "epoch": 0.0009216, "grad_norm": 0.6945897936820984, "learning_rate": 1.819607387762114e-05, "loss": 2.9537, "step": 76260 }, { "epoch": 0.0009472, "grad_norm": 0.7274574041366577, "learning_rate": 1.8195612486402346e-05, "loss": 3.3257, "step": 76270 }, { "epoch": 0.0009728, "grad_norm": 0.7636537551879883, "learning_rate": 1.8195151042037182e-05, "loss": 3.051, "step": 76280 }, { "epoch": 0.0009984, "grad_norm": 0.7824432849884033, "learning_rate": 1.8194689544528646e-05, "loss": 3.1308, "step": 76290 }, { "epoch": 0.001024, "grad_norm": 0.8453372120857239, "learning_rate": 1.819422799387972e-05, "loss": 3.1926, "step": 76300 }, { "epoch": 0.0010496, "grad_norm": 0.7457256317138672, "learning_rate": 1.8193766390093403e-05, "loss": 3.2749, "step": 76310 }, { "epoch": 0.0010752, "grad_norm": 0.7106383442878723, "learning_rate": 1.8193304733172693e-05, "loss": 3.0739, "step": 76320 }, { "epoch": 0.0011008, "grad_norm": 0.7073608040809631, "learning_rate": 1.8192843023120578e-05, "loss": 2.9383, "step": 76330 }, { "epoch": 0.0011264, "grad_norm": 0.729028046131134, "learning_rate": 1.819238125994005e-05, "loss": 3.081, "step": 76340 }, { "epoch": 0.001152, "grad_norm": 1.0167568922042847, "learning_rate": 1.8191919443634112e-05, "loss": 3.0828, "step": 76350 }, { "epoch": 0.0011776, "grad_norm": 1.2205560207366943, "learning_rate": 1.8191457574205752e-05, "loss": 3.0535, "step": 76360 }, { "epoch": 0.0012032, "grad_norm": 0.7050920128822327, "learning_rate": 1.8190995651657963e-05, "loss": 3.1906, "step": 76370 }, { "epoch": 0.0012288, "grad_norm": 0.8367283940315247, "learning_rate": 1.8190533675993748e-05, "loss": 3.129, "step": 76380 }, { "epoch": 0.0012544, "grad_norm": 0.7491809129714966, "learning_rate": 1.8190071647216097e-05, "loss": 3.1233, "step": 76390 }, { "epoch": 0.00128, "grad_norm": 0.7513609528541565, "learning_rate": 1.8189609565328008e-05, "loss": 3.2073, "step": 76400 }, { "epoch": 0.0013056, "grad_norm": 0.6944378614425659, "learning_rate": 1.818914743033248e-05, "loss": 3.1116, "step": 76410 }, { "epoch": 0.0013312, "grad_norm": 0.965051531791687, "learning_rate": 1.8188685242232503e-05, "loss": 3.1272, "step": 76420 }, { "epoch": 0.0013568, "grad_norm": 1.0586795806884766, "learning_rate": 1.8188223001031082e-05, "loss": 3.0259, "step": 76430 }, { "epoch": 0.0013824, "grad_norm": 0.838370144367218, "learning_rate": 1.818776070673121e-05, "loss": 3.3703, "step": 76440 }, { "epoch": 0.001408, "grad_norm": 1.0643025636672974, "learning_rate": 1.8187298359335885e-05, "loss": 3.026, "step": 76450 }, { "epoch": 0.0014336, "grad_norm": 0.7935078740119934, "learning_rate": 1.818683595884811e-05, "loss": 3.2486, "step": 76460 }, { "epoch": 0.0014592, "grad_norm": 0.8288565874099731, "learning_rate": 1.8186373505270873e-05, "loss": 3.2553, "step": 76470 }, { "epoch": 0.0014848, "grad_norm": 0.7512465119361877, "learning_rate": 1.8185910998607185e-05, "loss": 2.9194, "step": 76480 }, { "epoch": 0.0015104, "grad_norm": 0.8731803297996521, "learning_rate": 1.8185448438860037e-05, "loss": 2.9927, "step": 76490 }, { "epoch": 0.001536, "grad_norm": 0.7147346138954163, "learning_rate": 1.818498582603243e-05, "loss": 3.0384, "step": 76500 }, { "epoch": 2.56e-05, "grad_norm": 0.8405245542526245, "learning_rate": 1.818452316012737e-05, "loss": 2.7613, "step": 76510 }, { "epoch": 5.12e-05, "grad_norm": 0.911839485168457, "learning_rate": 1.818406044114785e-05, "loss": 3.2376, "step": 76520 }, { "epoch": 7.68e-05, "grad_norm": 0.7014979124069214, "learning_rate": 1.818359766909687e-05, "loss": 3.2666, "step": 76530 }, { "epoch": 0.0001024, "grad_norm": 0.7653788924217224, "learning_rate": 1.8183134843977437e-05, "loss": 3.0434, "step": 76540 }, { "epoch": 0.000128, "grad_norm": 0.6791864037513733, "learning_rate": 1.8182671965792552e-05, "loss": 3.1301, "step": 76550 }, { "epoch": 0.0001536, "grad_norm": 0.7697785496711731, "learning_rate": 1.818220903454521e-05, "loss": 3.2422, "step": 76560 }, { "epoch": 0.0001792, "grad_norm": 0.6908467411994934, "learning_rate": 1.8181746050238417e-05, "loss": 3.1325, "step": 76570 }, { "epoch": 0.0002048, "grad_norm": 0.8852611184120178, "learning_rate": 1.8181283012875177e-05, "loss": 3.1161, "step": 76580 }, { "epoch": 0.0002304, "grad_norm": 0.7623065114021301, "learning_rate": 1.818081992245849e-05, "loss": 3.1835, "step": 76590 }, { "epoch": 0.000256, "grad_norm": 0.7407328486442566, "learning_rate": 1.8180356778991362e-05, "loss": 2.9234, "step": 76600 }, { "epoch": 0.0002816, "grad_norm": 0.9997348785400391, "learning_rate": 1.817989358247679e-05, "loss": 3.1895, "step": 76610 }, { "epoch": 0.0003072, "grad_norm": 1.3788602352142334, "learning_rate": 1.817943033291779e-05, "loss": 3.0934, "step": 76620 }, { "epoch": 0.0003328, "grad_norm": 0.9454925060272217, "learning_rate": 1.817896703031735e-05, "loss": 3.1307, "step": 76630 }, { "epoch": 0.0003584, "grad_norm": 0.704347550868988, "learning_rate": 1.8178503674678487e-05, "loss": 3.0292, "step": 76640 }, { "epoch": 0.000384, "grad_norm": 0.9454473853111267, "learning_rate": 1.81780402660042e-05, "loss": 3.306, "step": 76650 }, { "epoch": 0.0004096, "grad_norm": 1.10232412815094, "learning_rate": 1.8177576804297494e-05, "loss": 2.7443, "step": 76660 }, { "epoch": 0.0004352, "grad_norm": 0.9428669810295105, "learning_rate": 1.8177113289561376e-05, "loss": 3.0461, "step": 76670 }, { "epoch": 0.0004608, "grad_norm": 0.8286260366439819, "learning_rate": 1.8176649721798854e-05, "loss": 3.0466, "step": 76680 }, { "epoch": 0.0004864, "grad_norm": 0.7516274452209473, "learning_rate": 1.817618610101293e-05, "loss": 2.9719, "step": 76690 }, { "epoch": 0.000512, "grad_norm": 0.7848268747329712, "learning_rate": 1.8175722427206613e-05, "loss": 3.1316, "step": 76700 }, { "epoch": 0.0005376, "grad_norm": 0.7520039677619934, "learning_rate": 1.8175258700382908e-05, "loss": 3.6015, "step": 76710 }, { "epoch": 0.0005632, "grad_norm": 0.7217428684234619, "learning_rate": 1.8174794920544824e-05, "loss": 2.971, "step": 76720 }, { "epoch": 0.0005888, "grad_norm": 0.747959554195404, "learning_rate": 1.8174331087695365e-05, "loss": 3.1322, "step": 76730 }, { "epoch": 0.0006144, "grad_norm": 0.7117457389831543, "learning_rate": 1.817386720183754e-05, "loss": 3.0885, "step": 76740 }, { "epoch": 0.00064, "grad_norm": 0.7567434310913086, "learning_rate": 1.8173403262974365e-05, "loss": 2.897, "step": 76750 }, { "epoch": 0.0006656, "grad_norm": 0.6710410118103027, "learning_rate": 1.817293927110884e-05, "loss": 3.0553, "step": 76760 }, { "epoch": 0.0006912, "grad_norm": 0.7272279858589172, "learning_rate": 1.817247522624397e-05, "loss": 2.975, "step": 76770 }, { "epoch": 0.0007168, "grad_norm": 1.1380205154418945, "learning_rate": 1.817201112838278e-05, "loss": 3.0523, "step": 76780 }, { "epoch": 0.0007424, "grad_norm": 0.6703000068664551, "learning_rate": 1.8171546977528262e-05, "loss": 3.1987, "step": 76790 }, { "epoch": 0.000768, "grad_norm": 0.7583841681480408, "learning_rate": 1.817108277368344e-05, "loss": 3.0444, "step": 76800 }, { "epoch": 0.0007936, "grad_norm": 0.6473417282104492, "learning_rate": 1.817061851685131e-05, "loss": 3.2354, "step": 76810 }, { "epoch": 0.0008192, "grad_norm": 0.9906491041183472, "learning_rate": 1.8170154207034895e-05, "loss": 3.2724, "step": 76820 }, { "epoch": 0.0008448, "grad_norm": 0.7084977030754089, "learning_rate": 1.8169689844237203e-05, "loss": 3.0334, "step": 76830 }, { "epoch": 0.0008704, "grad_norm": 1.6706693172454834, "learning_rate": 1.816922542846124e-05, "loss": 3.1262, "step": 76840 }, { "epoch": 0.000896, "grad_norm": 0.8754366636276245, "learning_rate": 1.8168760959710022e-05, "loss": 2.9925, "step": 76850 }, { "epoch": 0.0009216, "grad_norm": 0.7152764201164246, "learning_rate": 1.816829643798656e-05, "loss": 3.2293, "step": 76860 }, { "epoch": 0.0009472, "grad_norm": 1.1080024242401123, "learning_rate": 1.8167831863293866e-05, "loss": 3.3948, "step": 76870 }, { "epoch": 0.0009728, "grad_norm": 0.6845343112945557, "learning_rate": 1.8167367235634955e-05, "loss": 3.2243, "step": 76880 }, { "epoch": 0.0009984, "grad_norm": 0.7141168117523193, "learning_rate": 1.816690255501284e-05, "loss": 3.0676, "step": 76890 }, { "epoch": 0.001024, "grad_norm": 0.9155321717262268, "learning_rate": 1.8166437821430527e-05, "loss": 3.2874, "step": 76900 }, { "epoch": 2.56e-05, "grad_norm": 0.7388768792152405, "learning_rate": 1.816597303489104e-05, "loss": 2.7151, "step": 76910 }, { "epoch": 5.12e-05, "grad_norm": 0.7241506576538086, "learning_rate": 1.8165508195397385e-05, "loss": 3.1516, "step": 76920 }, { "epoch": 7.68e-05, "grad_norm": 0.885039210319519, "learning_rate": 1.816504330295258e-05, "loss": 3.0738, "step": 76930 }, { "epoch": 0.0001024, "grad_norm": 0.7575026750564575, "learning_rate": 1.816457835755964e-05, "loss": 3.0682, "step": 76940 }, { "epoch": 0.000128, "grad_norm": 0.7118191123008728, "learning_rate": 1.8164113359221577e-05, "loss": 2.8863, "step": 76950 }, { "epoch": 0.0001536, "grad_norm": 1.2193149328231812, "learning_rate": 1.816364830794141e-05, "loss": 3.1344, "step": 76960 }, { "epoch": 0.0001792, "grad_norm": 0.7268720269203186, "learning_rate": 1.8163183203722157e-05, "loss": 2.9669, "step": 76970 }, { "epoch": 0.0002048, "grad_norm": 0.8964910507202148, "learning_rate": 1.8162718046566826e-05, "loss": 3.0384, "step": 76980 }, { "epoch": 0.0002304, "grad_norm": 0.6859027147293091, "learning_rate": 1.816225283647844e-05, "loss": 2.942, "step": 76990 }, { "epoch": 0.000256, "grad_norm": 0.7248656749725342, "learning_rate": 1.8161787573460012e-05, "loss": 2.6912, "step": 77000 }, { "epoch": 0.0002816, "grad_norm": 0.8219686150550842, "learning_rate": 1.8161322257514562e-05, "loss": 3.2023, "step": 77010 }, { "epoch": 0.0003072, "grad_norm": 0.7240675091743469, "learning_rate": 1.8160856888645105e-05, "loss": 3.162, "step": 77020 }, { "epoch": 0.0003328, "grad_norm": 0.7519275546073914, "learning_rate": 1.8160391466854662e-05, "loss": 3.2543, "step": 77030 }, { "epoch": 0.0003584, "grad_norm": 0.7236683964729309, "learning_rate": 1.815992599214625e-05, "loss": 3.1384, "step": 77040 }, { "epoch": 0.000384, "grad_norm": 0.7547347545623779, "learning_rate": 1.8159460464522882e-05, "loss": 3.1845, "step": 77050 }, { "epoch": 0.0004096, "grad_norm": 0.7115366458892822, "learning_rate": 1.8158994883987586e-05, "loss": 3.1391, "step": 77060 }, { "epoch": 0.0004352, "grad_norm": 0.9399713277816772, "learning_rate": 1.8158529250543372e-05, "loss": 2.7647, "step": 77070 }, { "epoch": 0.0004608, "grad_norm": 0.8022136092185974, "learning_rate": 1.8158063564193268e-05, "loss": 3.2277, "step": 77080 }, { "epoch": 0.0004864, "grad_norm": 0.7663134932518005, "learning_rate": 1.8157597824940287e-05, "loss": 3.0058, "step": 77090 }, { "epoch": 0.000512, "grad_norm": 0.6704191565513611, "learning_rate": 1.8157132032787453e-05, "loss": 3.0078, "step": 77100 }, { "epoch": 0.0005376, "grad_norm": 0.8064798712730408, "learning_rate": 1.8156666187737787e-05, "loss": 3.0841, "step": 77110 }, { "epoch": 0.0005632, "grad_norm": 0.7790470123291016, "learning_rate": 1.815620028979431e-05, "loss": 3.4508, "step": 77120 }, { "epoch": 0.0005888, "grad_norm": 0.7727864384651184, "learning_rate": 1.8155734338960037e-05, "loss": 3.073, "step": 77130 }, { "epoch": 0.0006144, "grad_norm": 0.6982897520065308, "learning_rate": 1.8155268335237996e-05, "loss": 3.0469, "step": 77140 }, { "epoch": 0.00064, "grad_norm": 0.7630974650382996, "learning_rate": 1.815480227863121e-05, "loss": 3.0304, "step": 77150 }, { "epoch": 0.0006656, "grad_norm": 0.696934700012207, "learning_rate": 1.8154336169142696e-05, "loss": 3.1833, "step": 77160 }, { "epoch": 0.0006912, "grad_norm": 0.8724846839904785, "learning_rate": 1.815387000677548e-05, "loss": 3.1119, "step": 77170 }, { "epoch": 0.0007168, "grad_norm": 0.8104130029678345, "learning_rate": 1.8153403791532583e-05, "loss": 2.982, "step": 77180 }, { "epoch": 0.0007424, "grad_norm": 0.8294669985771179, "learning_rate": 1.815293752341703e-05, "loss": 3.0201, "step": 77190 }, { "epoch": 0.000768, "grad_norm": 0.7560248970985413, "learning_rate": 1.8152471202431845e-05, "loss": 3.2128, "step": 77200 }, { "epoch": 0.0007936, "grad_norm": 0.687578022480011, "learning_rate": 1.8152004828580053e-05, "loss": 3.1397, "step": 77210 }, { "epoch": 0.0008192, "grad_norm": 0.7274819612503052, "learning_rate": 1.815153840186467e-05, "loss": 3.1736, "step": 77220 }, { "epoch": 0.0008448, "grad_norm": 0.8984965682029724, "learning_rate": 1.815107192228873e-05, "loss": 3.329, "step": 77230 }, { "epoch": 0.0008704, "grad_norm": 0.8674660921096802, "learning_rate": 1.8150605389855255e-05, "loss": 3.3436, "step": 77240 }, { "epoch": 0.000896, "grad_norm": 0.8979165554046631, "learning_rate": 1.815013880456727e-05, "loss": 3.2241, "step": 77250 }, { "epoch": 0.0009216, "grad_norm": 0.7159909605979919, "learning_rate": 1.8149672166427802e-05, "loss": 3.0681, "step": 77260 }, { "epoch": 0.0009472, "grad_norm": 0.7217122912406921, "learning_rate": 1.8149205475439874e-05, "loss": 3.1545, "step": 77270 }, { "epoch": 0.0009728, "grad_norm": 0.8143226504325867, "learning_rate": 1.8148738731606516e-05, "loss": 3.2338, "step": 77280 }, { "epoch": 0.0009984, "grad_norm": 0.7240630984306335, "learning_rate": 1.8148271934930753e-05, "loss": 3.4315, "step": 77290 }, { "epoch": 0.001024, "grad_norm": 0.709406852722168, "learning_rate": 1.8147805085415612e-05, "loss": 3.0799, "step": 77300 }, { "epoch": 0.0010496, "grad_norm": 0.8632813692092896, "learning_rate": 1.814733818306412e-05, "loss": 3.4423, "step": 77310 }, { "epoch": 0.0010752, "grad_norm": 0.815156877040863, "learning_rate": 1.8146871227879305e-05, "loss": 3.1346, "step": 77320 }, { "epoch": 0.0011008, "grad_norm": 1.2938158512115479, "learning_rate": 1.8146404219864197e-05, "loss": 3.1109, "step": 77330 }, { "epoch": 0.0011264, "grad_norm": 0.7732895612716675, "learning_rate": 1.814593715902182e-05, "loss": 3.0853, "step": 77340 }, { "epoch": 0.001152, "grad_norm": 1.226667046546936, "learning_rate": 1.8145470045355205e-05, "loss": 2.92, "step": 77350 }, { "epoch": 0.0011776, "grad_norm": 0.7604780197143555, "learning_rate": 1.8145002878867385e-05, "loss": 3.083, "step": 77360 }, { "epoch": 0.0012032, "grad_norm": 0.7303622364997864, "learning_rate": 1.8144535659561382e-05, "loss": 2.8729, "step": 77370 }, { "epoch": 0.0012288, "grad_norm": 1.0412346124649048, "learning_rate": 1.8144068387440233e-05, "loss": 3.0468, "step": 77380 }, { "epoch": 0.0012544, "grad_norm": 0.6884446740150452, "learning_rate": 1.8143601062506962e-05, "loss": 3.1246, "step": 77390 }, { "epoch": 0.00128, "grad_norm": 0.7708778381347656, "learning_rate": 1.8143133684764605e-05, "loss": 3.1974, "step": 77400 }, { "epoch": 0.0013056, "grad_norm": 0.7303810119628906, "learning_rate": 1.814266625421619e-05, "loss": 3.0844, "step": 77410 }, { "epoch": 0.0013312, "grad_norm": 1.3102421760559082, "learning_rate": 1.8142198770864748e-05, "loss": 3.1522, "step": 77420 }, { "epoch": 0.0013568, "grad_norm": 1.0226449966430664, "learning_rate": 1.814173123471331e-05, "loss": 3.0163, "step": 77430 }, { "epoch": 0.0013824, "grad_norm": 0.6807191967964172, "learning_rate": 1.814126364576491e-05, "loss": 3.2774, "step": 77440 }, { "epoch": 0.001408, "grad_norm": 0.7382615208625793, "learning_rate": 1.814079600402258e-05, "loss": 3.1543, "step": 77450 }, { "epoch": 0.0014336, "grad_norm": 0.7219125032424927, "learning_rate": 1.814032830948935e-05, "loss": 3.0856, "step": 77460 }, { "epoch": 0.0014592, "grad_norm": 0.7305036187171936, "learning_rate": 1.8139860562168252e-05, "loss": 2.8693, "step": 77470 }, { "epoch": 0.0014848, "grad_norm": 0.8011326193809509, "learning_rate": 1.813939276206232e-05, "loss": 2.9478, "step": 77480 }, { "epoch": 0.0015104, "grad_norm": 0.8201963305473328, "learning_rate": 1.8138924909174597e-05, "loss": 3.2729, "step": 77490 }, { "epoch": 0.001536, "grad_norm": 0.6807054877281189, "learning_rate": 1.8138457003508104e-05, "loss": 3.2675, "step": 77500 }, { "epoch": 0.0015616, "grad_norm": 0.7243641018867493, "learning_rate": 1.813798904506588e-05, "loss": 3.1283, "step": 77510 }, { "epoch": 0.0015872, "grad_norm": 1.032050371170044, "learning_rate": 1.813752103385096e-05, "loss": 3.0315, "step": 77520 }, { "epoch": 0.0016128, "grad_norm": 0.7952753901481628, "learning_rate": 1.8137052969866384e-05, "loss": 2.9905, "step": 77530 }, { "epoch": 0.0016384, "grad_norm": 0.6877267956733704, "learning_rate": 1.8136584853115177e-05, "loss": 3.3099, "step": 77540 }, { "epoch": 0.001664, "grad_norm": 0.837141215801239, "learning_rate": 1.813611668360038e-05, "loss": 3.0973, "step": 77550 }, { "epoch": 0.0016896, "grad_norm": 0.9303866028785706, "learning_rate": 1.813564846132503e-05, "loss": 3.1177, "step": 77560 }, { "epoch": 0.0017152, "grad_norm": 0.7406547665596008, "learning_rate": 1.813518018629216e-05, "loss": 3.2898, "step": 77570 }, { "epoch": 0.0017408, "grad_norm": 0.8254749178886414, "learning_rate": 1.8134711858504814e-05, "loss": 3.2078, "step": 77580 }, { "epoch": 0.0017664, "grad_norm": 0.690893828868866, "learning_rate": 1.813424347796602e-05, "loss": 3.2738, "step": 77590 }, { "epoch": 0.001792, "grad_norm": 0.8583072423934937, "learning_rate": 1.8133775044678813e-05, "loss": 3.1749, "step": 77600 }, { "epoch": 0.0018176, "grad_norm": 0.949697732925415, "learning_rate": 1.8133306558646243e-05, "loss": 3.2442, "step": 77610 }, { "epoch": 0.0018432, "grad_norm": 0.8814584612846375, "learning_rate": 1.813283801987134e-05, "loss": 3.1333, "step": 77620 }, { "epoch": 0.0018688, "grad_norm": 0.7742825150489807, "learning_rate": 1.8132369428357143e-05, "loss": 3.0386, "step": 77630 }, { "epoch": 0.0018944, "grad_norm": 0.940916121006012, "learning_rate": 1.8131900784106695e-05, "loss": 3.1214, "step": 77640 }, { "epoch": 0.00192, "grad_norm": 2.529768705368042, "learning_rate": 1.813143208712303e-05, "loss": 3.0507, "step": 77650 }, { "epoch": 0.0019456, "grad_norm": 0.9643287658691406, "learning_rate": 1.813096333740919e-05, "loss": 3.2518, "step": 77660 }, { "epoch": 0.0019712, "grad_norm": 0.6961191892623901, "learning_rate": 1.813049453496821e-05, "loss": 2.9705, "step": 77670 }, { "epoch": 0.0019968, "grad_norm": 0.7415751814842224, "learning_rate": 1.8130025679803137e-05, "loss": 3.0365, "step": 77680 }, { "epoch": 0.0020224, "grad_norm": 0.7011373043060303, "learning_rate": 1.8129556771917005e-05, "loss": 3.2497, "step": 77690 }, { "epoch": 0.002048, "grad_norm": 1.0786762237548828, "learning_rate": 1.812908781131286e-05, "loss": 3.069, "step": 77700 }, { "epoch": 0.0020736, "grad_norm": 0.7546191811561584, "learning_rate": 1.812861879799374e-05, "loss": 3.178, "step": 77710 }, { "epoch": 0.0020992, "grad_norm": 0.8112925291061401, "learning_rate": 1.812814973196269e-05, "loss": 3.2069, "step": 77720 }, { "epoch": 0.0021248, "grad_norm": 0.9793450832366943, "learning_rate": 1.8127680613222748e-05, "loss": 3.367, "step": 77730 }, { "epoch": 0.0021504, "grad_norm": 0.8646671175956726, "learning_rate": 1.8127211441776954e-05, "loss": 3.376, "step": 77740 }, { "epoch": 0.002176, "grad_norm": 0.7921998500823975, "learning_rate": 1.8126742217628357e-05, "loss": 3.3658, "step": 77750 }, { "epoch": 0.0022016, "grad_norm": 0.7492125630378723, "learning_rate": 1.812627294078e-05, "loss": 3.3249, "step": 77760 }, { "epoch": 0.0022272, "grad_norm": 0.817894697189331, "learning_rate": 1.812580361123492e-05, "loss": 3.2003, "step": 77770 }, { "epoch": 0.0022528, "grad_norm": 0.7435522079467773, "learning_rate": 1.8125334228996157e-05, "loss": 3.1404, "step": 77780 }, { "epoch": 0.0022784, "grad_norm": 0.8172804117202759, "learning_rate": 1.8124864794066766e-05, "loss": 3.1197, "step": 77790 }, { "epoch": 0.002304, "grad_norm": 1.2501115798950195, "learning_rate": 1.8124395306449787e-05, "loss": 3.3186, "step": 77800 }, { "epoch": 0.0023296, "grad_norm": 1.165762186050415, "learning_rate": 1.8123925766148263e-05, "loss": 3.3785, "step": 77810 }, { "epoch": 0.0023552, "grad_norm": 0.7896052598953247, "learning_rate": 1.812345617316524e-05, "loss": 3.1712, "step": 77820 }, { "epoch": 0.0023808, "grad_norm": 0.681891918182373, "learning_rate": 1.8122986527503762e-05, "loss": 3.2288, "step": 77830 }, { "epoch": 0.0024064, "grad_norm": 0.6553117036819458, "learning_rate": 1.8122516829166873e-05, "loss": 3.2822, "step": 77840 }, { "epoch": 0.002432, "grad_norm": 0.7764378786087036, "learning_rate": 1.8122047078157626e-05, "loss": 3.3099, "step": 77850 }, { "epoch": 0.0024576, "grad_norm": 0.8324938416481018, "learning_rate": 1.8121577274479058e-05, "loss": 3.1993, "step": 77860 }, { "epoch": 0.0024832, "grad_norm": 0.8403379917144775, "learning_rate": 1.812110741813422e-05, "loss": 3.3227, "step": 77870 }, { "epoch": 0.0025088, "grad_norm": 0.8359043598175049, "learning_rate": 1.812063750912616e-05, "loss": 3.1628, "step": 77880 }, { "epoch": 0.0025344, "grad_norm": 0.8098811507225037, "learning_rate": 1.8120167547457925e-05, "loss": 3.2949, "step": 77890 }, { "epoch": 0.00256, "grad_norm": 0.87090665102005, "learning_rate": 1.811969753313256e-05, "loss": 3.435, "step": 77900 }, { "epoch": 0.0025856, "grad_norm": 0.7308509945869446, "learning_rate": 1.8119227466153115e-05, "loss": 3.3197, "step": 77910 }, { "epoch": 0.0026112, "grad_norm": 0.8808104395866394, "learning_rate": 1.8118757346522638e-05, "loss": 3.0888, "step": 77920 }, { "epoch": 0.0026368, "grad_norm": 0.7156960964202881, "learning_rate": 1.8118287174244174e-05, "loss": 3.1084, "step": 77930 }, { "epoch": 0.0026624, "grad_norm": 0.7224061489105225, "learning_rate": 1.811781694932078e-05, "loss": 3.2696, "step": 77940 }, { "epoch": 0.002688, "grad_norm": 0.8596018552780151, "learning_rate": 1.81173466717555e-05, "loss": 3.2465, "step": 77950 }, { "epoch": 0.0027136, "grad_norm": 0.7550045847892761, "learning_rate": 1.8116876341551383e-05, "loss": 2.9442, "step": 77960 }, { "epoch": 0.0027392, "grad_norm": 0.6828597784042358, "learning_rate": 1.8116405958711482e-05, "loss": 3.3015, "step": 77970 }, { "epoch": 0.0027648, "grad_norm": 0.736884355545044, "learning_rate": 1.811593552323884e-05, "loss": 3.3019, "step": 77980 }, { "epoch": 0.0027904, "grad_norm": 0.8338313698768616, "learning_rate": 1.8115465035136516e-05, "loss": 3.2747, "step": 77990 }, { "epoch": 0.002816, "grad_norm": 0.7102814316749573, "learning_rate": 1.811499449440756e-05, "loss": 3.1285, "step": 78000 }, { "epoch": 0.0028416, "grad_norm": 0.7246792316436768, "learning_rate": 1.8114523901055022e-05, "loss": 3.302, "step": 78010 }, { "epoch": 0.0028672, "grad_norm": 0.6799458861351013, "learning_rate": 1.811405325508195e-05, "loss": 3.3259, "step": 78020 }, { "epoch": 0.0028928, "grad_norm": 0.8936602473258972, "learning_rate": 1.81135825564914e-05, "loss": 3.3735, "step": 78030 }, { "epoch": 0.0029184, "grad_norm": 0.7565746307373047, "learning_rate": 1.811311180528642e-05, "loss": 3.0961, "step": 78040 }, { "epoch": 0.002944, "grad_norm": 0.8290198445320129, "learning_rate": 1.8112641001470072e-05, "loss": 3.3431, "step": 78050 }, { "epoch": 0.0029696, "grad_norm": 0.8346707820892334, "learning_rate": 1.8112170145045404e-05, "loss": 3.3272, "step": 78060 }, { "epoch": 0.0029952, "grad_norm": 0.7245383262634277, "learning_rate": 1.8111699236015467e-05, "loss": 3.3427, "step": 78070 }, { "epoch": 0.0030208, "grad_norm": 0.7426453828811646, "learning_rate": 1.8111228274383313e-05, "loss": 2.9965, "step": 78080 }, { "epoch": 0.0030464, "grad_norm": 0.7395073771476746, "learning_rate": 1.8110757260152002e-05, "loss": 3.2182, "step": 78090 }, { "epoch": 0.003072, "grad_norm": 0.7740765810012817, "learning_rate": 1.8110286193324584e-05, "loss": 3.4227, "step": 78100 }, { "epoch": 0.0030976, "grad_norm": 0.7889322638511658, "learning_rate": 1.8109815073904114e-05, "loss": 3.3202, "step": 78110 }, { "epoch": 0.0031232, "grad_norm": 0.7153568863868713, "learning_rate": 1.8109343901893656e-05, "loss": 3.0667, "step": 78120 }, { "epoch": 0.0031488, "grad_norm": 1.018573522567749, "learning_rate": 1.810887267729625e-05, "loss": 3.1549, "step": 78130 }, { "epoch": 0.0031744, "grad_norm": 0.6982341408729553, "learning_rate": 1.8108401400114963e-05, "loss": 3.3158, "step": 78140 }, { "epoch": 0.0032, "grad_norm": 0.889559805393219, "learning_rate": 1.8107930070352848e-05, "loss": 3.4411, "step": 78150 }, { "epoch": 0.0032256, "grad_norm": 0.7595884203910828, "learning_rate": 1.810745868801296e-05, "loss": 3.2066, "step": 78160 }, { "epoch": 0.0032512, "grad_norm": 0.8306808471679688, "learning_rate": 1.8106987253098356e-05, "loss": 3.3189, "step": 78170 }, { "epoch": 0.0032768, "grad_norm": 0.8221988081932068, "learning_rate": 1.81065157656121e-05, "loss": 3.2535, "step": 78180 }, { "epoch": 0.0033024, "grad_norm": 0.8209846019744873, "learning_rate": 1.810604422555724e-05, "loss": 3.2505, "step": 78190 }, { "epoch": 0.003328, "grad_norm": 0.9669719934463501, "learning_rate": 1.8105572632936837e-05, "loss": 3.3678, "step": 78200 }, { "epoch": 0.0033536, "grad_norm": 0.8984742760658264, "learning_rate": 1.810510098775395e-05, "loss": 3.2139, "step": 78210 }, { "epoch": 0.0033792, "grad_norm": 0.7657219767570496, "learning_rate": 1.810462929001164e-05, "loss": 3.1987, "step": 78220 }, { "epoch": 0.0034048, "grad_norm": 0.8118120431900024, "learning_rate": 1.810415753971296e-05, "loss": 3.4275, "step": 78230 }, { "epoch": 0.0034304, "grad_norm": 0.7232306599617004, "learning_rate": 1.810368573686097e-05, "loss": 3.3499, "step": 78240 }, { "epoch": 0.003456, "grad_norm": 0.7120094895362854, "learning_rate": 1.8103213881458737e-05, "loss": 3.4797, "step": 78250 }, { "epoch": 0.0034816, "grad_norm": 0.9263100028038025, "learning_rate": 1.8102741973509312e-05, "loss": 3.3999, "step": 78260 }, { "epoch": 0.0035072, "grad_norm": 0.7697084546089172, "learning_rate": 1.8102270013015758e-05, "loss": 3.1606, "step": 78270 }, { "epoch": 0.0035328, "grad_norm": 0.8225441575050354, "learning_rate": 1.810179799998114e-05, "loss": 3.3145, "step": 78280 }, { "epoch": 0.0035584, "grad_norm": 0.7946455478668213, "learning_rate": 1.810132593440851e-05, "loss": 3.3995, "step": 78290 }, { "epoch": 0.003584, "grad_norm": 0.842926561832428, "learning_rate": 1.810085381630094e-05, "loss": 3.2103, "step": 78300 }, { "epoch": 0.0036096, "grad_norm": 0.8318212628364563, "learning_rate": 1.8100381645661484e-05, "loss": 3.4009, "step": 78310 }, { "epoch": 0.0036352, "grad_norm": 0.7556071281433105, "learning_rate": 1.8099909422493204e-05, "loss": 3.0917, "step": 78320 }, { "epoch": 0.0036608, "grad_norm": 0.6953620910644531, "learning_rate": 1.8099437146799168e-05, "loss": 3.2883, "step": 78330 }, { "epoch": 0.0036864, "grad_norm": 0.815588116645813, "learning_rate": 1.809896481858243e-05, "loss": 3.4696, "step": 78340 }, { "epoch": 0.003712, "grad_norm": 0.9005770087242126, "learning_rate": 1.809849243784606e-05, "loss": 3.2813, "step": 78350 }, { "epoch": 0.0037376, "grad_norm": 0.7790788412094116, "learning_rate": 1.8098020004593117e-05, "loss": 3.3757, "step": 78360 }, { "epoch": 0.0037632, "grad_norm": 0.8090675473213196, "learning_rate": 1.809754751882667e-05, "loss": 3.3002, "step": 78370 }, { "epoch": 0.0037888, "grad_norm": 0.80632483959198, "learning_rate": 1.8097074980549777e-05, "loss": 3.1486, "step": 78380 }, { "epoch": 0.0038144, "grad_norm": 1.1009072065353394, "learning_rate": 1.8096602389765505e-05, "loss": 3.4968, "step": 78390 }, { "epoch": 0.00384, "grad_norm": 0.8787556290626526, "learning_rate": 1.809612974647692e-05, "loss": 3.2224, "step": 78400 }, { "epoch": 0.0038656, "grad_norm": 0.8211196064949036, "learning_rate": 1.8095657050687084e-05, "loss": 3.4357, "step": 78410 }, { "epoch": 0.0038912, "grad_norm": 0.8187230825424194, "learning_rate": 1.8095184302399062e-05, "loss": 3.2143, "step": 78420 }, { "epoch": 0.0039168, "grad_norm": 0.7659791111946106, "learning_rate": 1.809471150161592e-05, "loss": 3.0561, "step": 78430 }, { "epoch": 0.0039424, "grad_norm": 0.6952850222587585, "learning_rate": 1.809423864834073e-05, "loss": 3.1356, "step": 78440 }, { "epoch": 0.003968, "grad_norm": 0.8099113702774048, "learning_rate": 1.8093765742576553e-05, "loss": 3.1672, "step": 78450 }, { "epoch": 0.0039936, "grad_norm": 0.6825242638587952, "learning_rate": 1.8093292784326454e-05, "loss": 3.308, "step": 78460 }, { "epoch": 0.0040192, "grad_norm": 0.6658618450164795, "learning_rate": 1.8092819773593503e-05, "loss": 3.2544, "step": 78470 }, { "epoch": 0.0040448, "grad_norm": 0.7316429615020752, "learning_rate": 1.8092346710380765e-05, "loss": 3.0914, "step": 78480 }, { "epoch": 0.0040704, "grad_norm": 0.8834225535392761, "learning_rate": 1.809187359469131e-05, "loss": 3.0724, "step": 78490 }, { "epoch": 0.004096, "grad_norm": 0.7333146929740906, "learning_rate": 1.809140042652821e-05, "loss": 3.1343, "step": 78500 }, { "epoch": 0.0041216, "grad_norm": 0.8971707820892334, "learning_rate": 1.8090927205894523e-05, "loss": 3.3104, "step": 78510 }, { "epoch": 0.0041472, "grad_norm": 0.6763778924942017, "learning_rate": 1.8090453932793323e-05, "loss": 3.2175, "step": 78520 }, { "epoch": 0.0041728, "grad_norm": 0.9081328511238098, "learning_rate": 1.808998060722768e-05, "loss": 3.1714, "step": 78530 }, { "epoch": 0.0041984, "grad_norm": 1.4965394735336304, "learning_rate": 1.8089507229200667e-05, "loss": 3.4315, "step": 78540 }, { "epoch": 0.004224, "grad_norm": 0.7892059087753296, "learning_rate": 1.8089033798715348e-05, "loss": 3.0634, "step": 78550 }, { "epoch": 0.0042496, "grad_norm": 0.7466686964035034, "learning_rate": 1.8088560315774793e-05, "loss": 3.09, "step": 78560 }, { "epoch": 0.0042752, "grad_norm": 0.7016780376434326, "learning_rate": 1.808808678038207e-05, "loss": 2.9664, "step": 78570 }, { "epoch": 0.0043008, "grad_norm": 0.7258302569389343, "learning_rate": 1.808761319254026e-05, "loss": 3.0365, "step": 78580 }, { "epoch": 0.0043264, "grad_norm": 0.7194721698760986, "learning_rate": 1.8087139552252425e-05, "loss": 3.0732, "step": 78590 }, { "epoch": 0.004352, "grad_norm": 0.8691702485084534, "learning_rate": 1.8086665859521643e-05, "loss": 3.1498, "step": 78600 }, { "epoch": 0.0043776, "grad_norm": 0.7082597613334656, "learning_rate": 1.808619211435098e-05, "loss": 3.1409, "step": 78610 }, { "epoch": 0.0044032, "grad_norm": 0.9803881049156189, "learning_rate": 1.808571831674351e-05, "loss": 3.1309, "step": 78620 }, { "epoch": 0.0044288, "grad_norm": 0.7418289184570312, "learning_rate": 1.8085244466702304e-05, "loss": 3.1505, "step": 78630 }, { "epoch": 0.0044544, "grad_norm": 0.8754920363426208, "learning_rate": 1.808477056423044e-05, "loss": 3.2541, "step": 78640 }, { "epoch": 0.00448, "grad_norm": 1.0960808992385864, "learning_rate": 1.8084296609330984e-05, "loss": 3.3335, "step": 78650 }, { "epoch": 0.0045056, "grad_norm": 0.7262635827064514, "learning_rate": 1.8083822602007015e-05, "loss": 3.1603, "step": 78660 }, { "epoch": 0.0045312, "grad_norm": 0.9251872301101685, "learning_rate": 1.8083348542261604e-05, "loss": 2.99, "step": 78670 }, { "epoch": 0.0045568, "grad_norm": 0.9890924096107483, "learning_rate": 1.8082874430097826e-05, "loss": 3.038, "step": 78680 }, { "epoch": 0.0045824, "grad_norm": 0.6883385181427002, "learning_rate": 1.8082400265518755e-05, "loss": 3.3056, "step": 78690 }, { "epoch": 0.004608, "grad_norm": 0.7787701487541199, "learning_rate": 1.8081926048527468e-05, "loss": 3.1135, "step": 78700 }, { "epoch": 0.0046336, "grad_norm": 0.7672309875488281, "learning_rate": 1.8081451779127036e-05, "loss": 3.1134, "step": 78710 }, { "epoch": 0.0046592, "grad_norm": 0.7878841757774353, "learning_rate": 1.808097745732054e-05, "loss": 3.3587, "step": 78720 }, { "epoch": 0.0046848, "grad_norm": 0.6911520957946777, "learning_rate": 1.808050308311105e-05, "loss": 3.1124, "step": 78730 }, { "epoch": 0.0047104, "grad_norm": 0.8635269403457642, "learning_rate": 1.8080028656501647e-05, "loss": 3.1183, "step": 78740 }, { "epoch": 0.004736, "grad_norm": 0.7365238070487976, "learning_rate": 1.8079554177495405e-05, "loss": 3.2266, "step": 78750 }, { "epoch": 0.0047616, "grad_norm": 0.7777373790740967, "learning_rate": 1.8079079646095404e-05, "loss": 3.2712, "step": 78760 }, { "epoch": 0.0047872, "grad_norm": 0.6721335053443909, "learning_rate": 1.8078605062304715e-05, "loss": 3.1519, "step": 78770 }, { "epoch": 0.0048128, "grad_norm": 0.8112261295318604, "learning_rate": 1.807813042612642e-05, "loss": 3.25, "step": 78780 }, { "epoch": 0.0048384, "grad_norm": 0.7122942209243774, "learning_rate": 1.8077655737563597e-05, "loss": 3.13, "step": 78790 }, { "epoch": 0.004864, "grad_norm": 0.9090851545333862, "learning_rate": 1.807718099661932e-05, "loss": 3.352, "step": 78800 }, { "epoch": 0.0048896, "grad_norm": 0.9800135493278503, "learning_rate": 1.8076706203296678e-05, "loss": 3.2521, "step": 78810 }, { "epoch": 0.0049152, "grad_norm": 0.7977496385574341, "learning_rate": 1.8076231357598736e-05, "loss": 3.3759, "step": 78820 }, { "epoch": 0.0049408, "grad_norm": 0.7611898183822632, "learning_rate": 1.8075756459528584e-05, "loss": 3.1144, "step": 78830 }, { "epoch": 0.0049664, "grad_norm": 0.7842018604278564, "learning_rate": 1.8075281509089296e-05, "loss": 3.1542, "step": 78840 }, { "epoch": 0.004992, "grad_norm": 0.7904927134513855, "learning_rate": 1.8074806506283954e-05, "loss": 3.1766, "step": 78850 }, { "epoch": 0.0050176, "grad_norm": 0.7183811068534851, "learning_rate": 1.8074331451115635e-05, "loss": 3.1667, "step": 78860 }, { "epoch": 0.0050432, "grad_norm": 0.809138298034668, "learning_rate": 1.8073856343587425e-05, "loss": 3.2571, "step": 78870 }, { "epoch": 0.0050688, "grad_norm": 0.7861404418945312, "learning_rate": 1.8073381183702405e-05, "loss": 3.431, "step": 78880 }, { "epoch": 0.0050944, "grad_norm": 0.9383975267410278, "learning_rate": 1.807290597146365e-05, "loss": 3.1659, "step": 78890 }, { "epoch": 0.00512, "grad_norm": 0.7312339544296265, "learning_rate": 1.8072430706874248e-05, "loss": 3.1461, "step": 78900 }, { "epoch": 0.0051456, "grad_norm": 0.7854607701301575, "learning_rate": 1.8071955389937278e-05, "loss": 3.2795, "step": 78910 }, { "epoch": 0.0051712, "grad_norm": 0.7560818791389465, "learning_rate": 1.807148002065582e-05, "loss": 3.0729, "step": 78920 }, { "epoch": 0.0051968, "grad_norm": 0.7178401350975037, "learning_rate": 1.8071004599032957e-05, "loss": 3.2943, "step": 78930 }, { "epoch": 0.0052224, "grad_norm": 0.8218313455581665, "learning_rate": 1.8070529125071782e-05, "loss": 3.0471, "step": 78940 }, { "epoch": 0.005248, "grad_norm": 0.7470404505729675, "learning_rate": 1.8070053598775363e-05, "loss": 3.1658, "step": 78950 }, { "epoch": 0.0052736, "grad_norm": 0.72221440076828, "learning_rate": 1.8069578020146794e-05, "loss": 3.1612, "step": 78960 }, { "epoch": 0.0052992, "grad_norm": 0.9901297092437744, "learning_rate": 1.8069102389189157e-05, "loss": 3.2385, "step": 78970 }, { "epoch": 0.0053248, "grad_norm": 0.7557850480079651, "learning_rate": 1.8068626705905535e-05, "loss": 3.1741, "step": 78980 }, { "epoch": 0.0053504, "grad_norm": 0.6974213123321533, "learning_rate": 1.8068150970299014e-05, "loss": 3.2547, "step": 78990 }, { "epoch": 0.005376, "grad_norm": 0.8447107672691345, "learning_rate": 1.8067675182372677e-05, "loss": 3.1266, "step": 79000 }, { "epoch": 0.0054016, "grad_norm": 0.6550012230873108, "learning_rate": 1.806719934212961e-05, "loss": 3.0248, "step": 79010 }, { "epoch": 0.0054272, "grad_norm": 0.8216024041175842, "learning_rate": 1.8066723449572902e-05, "loss": 3.2694, "step": 79020 }, { "epoch": 0.0054528, "grad_norm": 0.7057666182518005, "learning_rate": 1.8066247504705634e-05, "loss": 3.0212, "step": 79030 }, { "epoch": 0.0054784, "grad_norm": 0.866826593875885, "learning_rate": 1.8065771507530897e-05, "loss": 3.167, "step": 79040 }, { "epoch": 0.005504, "grad_norm": 0.7989551424980164, "learning_rate": 1.806529545805177e-05, "loss": 3.0988, "step": 79050 }, { "epoch": 0.0055296, "grad_norm": 0.8601652383804321, "learning_rate": 1.806481935627135e-05, "loss": 3.2403, "step": 79060 }, { "epoch": 0.0055552, "grad_norm": 0.7273042798042297, "learning_rate": 1.806434320219272e-05, "loss": 3.1551, "step": 79070 }, { "epoch": 0.0055808, "grad_norm": 2.14884090423584, "learning_rate": 1.8063866995818965e-05, "loss": 3.0753, "step": 79080 }, { "epoch": 0.0056064, "grad_norm": 0.6707547903060913, "learning_rate": 1.806339073715318e-05, "loss": 3.0551, "step": 79090 }, { "epoch": 0.005632, "grad_norm": 0.717045247554779, "learning_rate": 1.8062914426198442e-05, "loss": 3.3678, "step": 79100 }, { "epoch": 0.0056576, "grad_norm": 0.8092866539955139, "learning_rate": 1.806243806295785e-05, "loss": 3.4368, "step": 79110 }, { "epoch": 0.0056832, "grad_norm": 0.772686779499054, "learning_rate": 1.8061961647434493e-05, "loss": 3.1589, "step": 79120 }, { "epoch": 0.0057088, "grad_norm": 0.9331104755401611, "learning_rate": 1.806148517963146e-05, "loss": 3.217, "step": 79130 }, { "epoch": 0.0057344, "grad_norm": 0.8394726514816284, "learning_rate": 1.806100865955183e-05, "loss": 3.329, "step": 79140 }, { "epoch": 0.00576, "grad_norm": 0.8509762287139893, "learning_rate": 1.8060532087198703e-05, "loss": 3.3372, "step": 79150 }, { "epoch": 0.0057856, "grad_norm": 0.7187826037406921, "learning_rate": 1.806005546257517e-05, "loss": 3.3815, "step": 79160 }, { "epoch": 0.0058112, "grad_norm": 0.8921868801116943, "learning_rate": 1.8059578785684318e-05, "loss": 3.2718, "step": 79170 }, { "epoch": 0.0058368, "grad_norm": 0.8657686710357666, "learning_rate": 1.805910205652924e-05, "loss": 3.2428, "step": 79180 }, { "epoch": 0.0058624, "grad_norm": 0.8482664227485657, "learning_rate": 1.805862527511303e-05, "loss": 3.3652, "step": 79190 }, { "epoch": 0.005888, "grad_norm": 0.6740627884864807, "learning_rate": 1.8058148441438773e-05, "loss": 3.226, "step": 79200 }, { "epoch": 0.0059136, "grad_norm": 0.6931639909744263, "learning_rate": 1.8057671555509566e-05, "loss": 3.1015, "step": 79210 }, { "epoch": 0.0059392, "grad_norm": 0.8779779672622681, "learning_rate": 1.80571946173285e-05, "loss": 3.2387, "step": 79220 }, { "epoch": 0.0059648, "grad_norm": 0.7726579308509827, "learning_rate": 1.805671762689867e-05, "loss": 3.1928, "step": 79230 }, { "epoch": 0.0059904, "grad_norm": 0.7989529371261597, "learning_rate": 1.8056240584223165e-05, "loss": 3.3329, "step": 79240 }, { "epoch": 0.006016, "grad_norm": 0.9479550719261169, "learning_rate": 1.8055763489305082e-05, "loss": 3.2888, "step": 79250 }, { "epoch": 0.0060416, "grad_norm": 1.222110390663147, "learning_rate": 1.8055286342147512e-05, "loss": 3.1845, "step": 79260 }, { "epoch": 0.0060672, "grad_norm": 0.8439730405807495, "learning_rate": 1.8054809142753553e-05, "loss": 3.3686, "step": 79270 }, { "epoch": 0.0060928, "grad_norm": 0.7403257489204407, "learning_rate": 1.8054331891126297e-05, "loss": 3.2639, "step": 79280 }, { "epoch": 0.0061184, "grad_norm": 0.7227660417556763, "learning_rate": 1.805385458726884e-05, "loss": 3.2421, "step": 79290 }, { "epoch": 0.006144, "grad_norm": 0.7242393493652344, "learning_rate": 1.8053377231184276e-05, "loss": 3.2736, "step": 79300 }, { "epoch": 2.56e-05, "grad_norm": 0.7579813599586487, "learning_rate": 1.80528998228757e-05, "loss": 2.8749, "step": 79310 }, { "epoch": 5.12e-05, "grad_norm": 1.0250256061553955, "learning_rate": 1.805242236234621e-05, "loss": 3.2084, "step": 79320 }, { "epoch": 7.68e-05, "grad_norm": 0.667725682258606, "learning_rate": 1.8051944849598897e-05, "loss": 2.9959, "step": 79330 }, { "epoch": 0.0001024, "grad_norm": 0.9182922840118408, "learning_rate": 1.805146728463687e-05, "loss": 3.0164, "step": 79340 }, { "epoch": 0.000128, "grad_norm": 0.691032886505127, "learning_rate": 1.8050989667463213e-05, "loss": 2.957, "step": 79350 }, { "epoch": 0.0001536, "grad_norm": 0.7666993737220764, "learning_rate": 1.8050511998081025e-05, "loss": 3.0772, "step": 79360 }, { "epoch": 0.0001792, "grad_norm": 0.7661208510398865, "learning_rate": 1.8050034276493412e-05, "loss": 3.0395, "step": 79370 }, { "epoch": 0.0002048, "grad_norm": 1.0972230434417725, "learning_rate": 1.804955650270346e-05, "loss": 3.1115, "step": 79380 }, { "epoch": 0.0002304, "grad_norm": 0.7291466593742371, "learning_rate": 1.804907867671428e-05, "loss": 2.9478, "step": 79390 }, { "epoch": 0.000256, "grad_norm": 0.7090075612068176, "learning_rate": 1.8048600798528956e-05, "loss": 3.2148, "step": 79400 }, { "epoch": 2.56e-05, "grad_norm": 0.8213580846786499, "learning_rate": 1.8048122868150598e-05, "loss": 2.7421, "step": 79410 }, { "epoch": 5.12e-05, "grad_norm": 0.9459704756736755, "learning_rate": 1.8047644885582305e-05, "loss": 3.1447, "step": 79420 }, { "epoch": 7.68e-05, "grad_norm": 0.7258463501930237, "learning_rate": 1.8047166850827173e-05, "loss": 2.8392, "step": 79430 }, { "epoch": 0.0001024, "grad_norm": 0.7708075642585754, "learning_rate": 1.80466887638883e-05, "loss": 2.9672, "step": 79440 }, { "epoch": 0.000128, "grad_norm": 0.747331440448761, "learning_rate": 1.8046210624768792e-05, "loss": 2.908, "step": 79450 }, { "epoch": 0.0001536, "grad_norm": 0.8705341219902039, "learning_rate": 1.8045732433471747e-05, "loss": 3.1585, "step": 79460 }, { "epoch": 0.0001792, "grad_norm": 0.7849641442298889, "learning_rate": 1.8045254190000263e-05, "loss": 3.1407, "step": 79470 }, { "epoch": 0.0002048, "grad_norm": 0.7315782308578491, "learning_rate": 1.8044775894357442e-05, "loss": 3.0958, "step": 79480 }, { "epoch": 0.0002304, "grad_norm": 0.7252669930458069, "learning_rate": 1.804429754654639e-05, "loss": 2.8176, "step": 79490 }, { "epoch": 0.000256, "grad_norm": 0.8184401988983154, "learning_rate": 1.804381914657021e-05, "loss": 3.0925, "step": 79500 }, { "epoch": 0.0002816, "grad_norm": 1.076456069946289, "learning_rate": 1.8043340694431993e-05, "loss": 3.2329, "step": 79510 }, { "epoch": 0.0003072, "grad_norm": 0.8045795559883118, "learning_rate": 1.8042862190134854e-05, "loss": 3.2543, "step": 79520 }, { "epoch": 0.0003328, "grad_norm": 0.7247991561889648, "learning_rate": 1.8042383633681887e-05, "loss": 3.2426, "step": 79530 }, { "epoch": 0.0003584, "grad_norm": 0.7022386789321899, "learning_rate": 1.8041905025076205e-05, "loss": 3.2161, "step": 79540 }, { "epoch": 0.000384, "grad_norm": 0.6917110085487366, "learning_rate": 1.8041426364320903e-05, "loss": 3.0466, "step": 79550 }, { "epoch": 0.0004096, "grad_norm": 0.7370448112487793, "learning_rate": 1.8040947651419088e-05, "loss": 2.9913, "step": 79560 }, { "epoch": 0.0004352, "grad_norm": 0.718894898891449, "learning_rate": 1.8040468886373867e-05, "loss": 2.9094, "step": 79570 }, { "epoch": 0.0004608, "grad_norm": 0.7777872681617737, "learning_rate": 1.803999006918834e-05, "loss": 2.9643, "step": 79580 }, { "epoch": 0.0004864, "grad_norm": 0.7030222415924072, "learning_rate": 1.8039511199865615e-05, "loss": 2.8341, "step": 79590 }, { "epoch": 0.000512, "grad_norm": 0.8451042175292969, "learning_rate": 1.80390322784088e-05, "loss": 2.8669, "step": 79600 }, { "epoch": 0.0005376, "grad_norm": 0.853699266910553, "learning_rate": 1.8038553304820993e-05, "loss": 3.2503, "step": 79610 }, { "epoch": 0.0005632, "grad_norm": 0.7282907366752625, "learning_rate": 1.8038074279105303e-05, "loss": 3.241, "step": 79620 }, { "epoch": 0.0005888, "grad_norm": 0.7993589639663696, "learning_rate": 1.8037595201264838e-05, "loss": 2.9905, "step": 79630 }, { "epoch": 0.0006144, "grad_norm": 0.6700794696807861, "learning_rate": 1.8037116071302706e-05, "loss": 3.1176, "step": 79640 }, { "epoch": 0.00064, "grad_norm": 0.7890006303787231, "learning_rate": 1.8036636889222013e-05, "loss": 3.0615, "step": 79650 }, { "epoch": 0.0006656, "grad_norm": 0.6938797831535339, "learning_rate": 1.8036157655025866e-05, "loss": 3.116, "step": 79660 }, { "epoch": 0.0006912, "grad_norm": 0.6498193144798279, "learning_rate": 1.803567836871737e-05, "loss": 3.0741, "step": 79670 }, { "epoch": 0.0007168, "grad_norm": 0.921036958694458, "learning_rate": 1.8035199030299636e-05, "loss": 2.9295, "step": 79680 }, { "epoch": 0.0007424, "grad_norm": 0.7748410701751709, "learning_rate": 1.803471963977577e-05, "loss": 3.1092, "step": 79690 }, { "epoch": 0.000768, "grad_norm": 0.9247397184371948, "learning_rate": 1.8034240197148887e-05, "loss": 2.9555, "step": 79700 }, { "epoch": 0.0007936, "grad_norm": 1.0266952514648438, "learning_rate": 1.8033760702422086e-05, "loss": 3.1057, "step": 79710 }, { "epoch": 0.0008192, "grad_norm": 0.7841185927391052, "learning_rate": 1.803328115559849e-05, "loss": 2.9272, "step": 79720 }, { "epoch": 0.0008448, "grad_norm": 0.9896641969680786, "learning_rate": 1.8032801556681194e-05, "loss": 3.1276, "step": 79730 }, { "epoch": 0.0008704, "grad_norm": 0.7544136643409729, "learning_rate": 1.8032321905673317e-05, "loss": 3.1266, "step": 79740 }, { "epoch": 0.000896, "grad_norm": 0.7184849977493286, "learning_rate": 1.8031842202577966e-05, "loss": 2.8946, "step": 79750 }, { "epoch": 0.0009216, "grad_norm": 0.8497905731201172, "learning_rate": 1.8031362447398253e-05, "loss": 3.1149, "step": 79760 }, { "epoch": 0.0009472, "grad_norm": 0.9564265608787537, "learning_rate": 1.803088264013729e-05, "loss": 3.176, "step": 79770 }, { "epoch": 0.0009728, "grad_norm": 0.8056663274765015, "learning_rate": 1.8030402780798188e-05, "loss": 3.2399, "step": 79780 }, { "epoch": 0.0009984, "grad_norm": 0.7029332518577576, "learning_rate": 1.8029922869384058e-05, "loss": 3.1594, "step": 79790 }, { "epoch": 0.001024, "grad_norm": 1.0080798864364624, "learning_rate": 1.8029442905898012e-05, "loss": 3.0969, "step": 79800 }, { "epoch": 0.0010496, "grad_norm": 0.6833388209342957, "learning_rate": 1.8028962890343164e-05, "loss": 3.0982, "step": 79810 }, { "epoch": 0.0010752, "grad_norm": 0.7617015242576599, "learning_rate": 1.8028482822722623e-05, "loss": 2.9746, "step": 79820 }, { "epoch": 0.0011008, "grad_norm": 0.6292052865028381, "learning_rate": 1.8028002703039504e-05, "loss": 2.9564, "step": 79830 }, { "epoch": 0.0011264, "grad_norm": 0.7694786787033081, "learning_rate": 1.802752253129692e-05, "loss": 2.8564, "step": 79840 }, { "epoch": 0.001152, "grad_norm": 0.9180823564529419, "learning_rate": 1.802704230749799e-05, "loss": 3.1449, "step": 79850 }, { "epoch": 0.0011776, "grad_norm": 0.7458558678627014, "learning_rate": 1.8026562031645826e-05, "loss": 2.9148, "step": 79860 }, { "epoch": 0.0012032, "grad_norm": 0.7282283902168274, "learning_rate": 1.8026081703743533e-05, "loss": 2.9494, "step": 79870 }, { "epoch": 0.0012288, "grad_norm": 0.6600224375724792, "learning_rate": 1.8025601323794236e-05, "loss": 3.074, "step": 79880 }, { "epoch": 0.0012544, "grad_norm": 0.7348592877388, "learning_rate": 1.802512089180105e-05, "loss": 3.1559, "step": 79890 }, { "epoch": 0.00128, "grad_norm": 0.7764055132865906, "learning_rate": 1.8024640407767088e-05, "loss": 3.1503, "step": 79900 }, { "epoch": 0.0013056, "grad_norm": 0.7108672261238098, "learning_rate": 1.802415987169546e-05, "loss": 3.2916, "step": 79910 }, { "epoch": 0.0013312, "grad_norm": 0.9560642838478088, "learning_rate": 1.802367928358929e-05, "loss": 2.9998, "step": 79920 }, { "epoch": 0.0013568, "grad_norm": 0.8885350823402405, "learning_rate": 1.8023198643451697e-05, "loss": 2.8846, "step": 79930 }, { "epoch": 0.0013824, "grad_norm": 0.6822604537010193, "learning_rate": 1.8022717951285792e-05, "loss": 3.1554, "step": 79940 }, { "epoch": 0.001408, "grad_norm": 0.7757704854011536, "learning_rate": 1.802223720709469e-05, "loss": 2.8994, "step": 79950 }, { "epoch": 0.0014336, "grad_norm": 0.7071189284324646, "learning_rate": 1.8021756410881514e-05, "loss": 2.9011, "step": 79960 }, { "epoch": 0.0014592, "grad_norm": 0.8505736589431763, "learning_rate": 1.802127556264938e-05, "loss": 3.0443, "step": 79970 }, { "epoch": 0.0014848, "grad_norm": 0.7235031127929688, "learning_rate": 1.8020794662401404e-05, "loss": 2.9584, "step": 79980 }, { "epoch": 0.0015104, "grad_norm": 0.7291005253791809, "learning_rate": 1.8020313710140707e-05, "loss": 3.0717, "step": 79990 }, { "epoch": 0.001536, "grad_norm": 1.2141950130462646, "learning_rate": 1.8019832705870404e-05, "loss": 3.1395, "step": 80000 }, { "epoch": 0.0015616, "grad_norm": 0.8754127621650696, "learning_rate": 1.801935164959362e-05, "loss": 3.2242, "step": 80010 }, { "epoch": 0.0015872, "grad_norm": 1.077027440071106, "learning_rate": 1.8018870541313473e-05, "loss": 2.9852, "step": 80020 }, { "epoch": 0.0016128, "grad_norm": 0.7045833468437195, "learning_rate": 1.8018389381033082e-05, "loss": 2.9283, "step": 80030 }, { "epoch": 0.0016384, "grad_norm": 0.6850091218948364, "learning_rate": 1.8017908168755568e-05, "loss": 3.1309, "step": 80040 }, { "epoch": 0.001664, "grad_norm": 0.7144878506660461, "learning_rate": 1.8017426904484044e-05, "loss": 2.8325, "step": 80050 }, { "epoch": 0.0016896, "grad_norm": 0.9212949872016907, "learning_rate": 1.801694558822164e-05, "loss": 3.1898, "step": 80060 }, { "epoch": 0.0017152, "grad_norm": 1.05116868019104, "learning_rate": 1.801646421997148e-05, "loss": 2.8764, "step": 80070 }, { "epoch": 0.0017408, "grad_norm": 0.8075961470603943, "learning_rate": 1.8015982799736673e-05, "loss": 3.4303, "step": 80080 }, { "epoch": 0.0017664, "grad_norm": 0.7775629162788391, "learning_rate": 1.801550132752035e-05, "loss": 2.9306, "step": 80090 }, { "epoch": 0.001792, "grad_norm": 1.3723477125167847, "learning_rate": 1.8015019803325632e-05, "loss": 2.981, "step": 80100 }, { "epoch": 0.0018176, "grad_norm": 0.690559446811676, "learning_rate": 1.801453822715564e-05, "loss": 3.4091, "step": 80110 }, { "epoch": 0.0018432, "grad_norm": 0.621709406375885, "learning_rate": 1.80140565990135e-05, "loss": 3.0962, "step": 80120 }, { "epoch": 0.0018688, "grad_norm": 0.7618443369865417, "learning_rate": 1.8013574918902333e-05, "loss": 3.2366, "step": 80130 }, { "epoch": 0.0018944, "grad_norm": 1.1940237283706665, "learning_rate": 1.8013093186825263e-05, "loss": 2.829, "step": 80140 }, { "epoch": 0.00192, "grad_norm": 0.6412953734397888, "learning_rate": 1.8012611402785408e-05, "loss": 3.1088, "step": 80150 }, { "epoch": 0.0019456, "grad_norm": 0.8176347017288208, "learning_rate": 1.8012129566785902e-05, "loss": 3.4892, "step": 80160 }, { "epoch": 0.0019712, "grad_norm": 0.8082502484321594, "learning_rate": 1.8011647678829864e-05, "loss": 3.2646, "step": 80170 }, { "epoch": 0.0019968, "grad_norm": 0.7464001774787903, "learning_rate": 1.801116573892042e-05, "loss": 3.2813, "step": 80180 }, { "epoch": 0.0020224, "grad_norm": 0.7321267127990723, "learning_rate": 1.8010683747060695e-05, "loss": 3.3191, "step": 80190 }, { "epoch": 0.002048, "grad_norm": 0.8157600164413452, "learning_rate": 1.8010201703253812e-05, "loss": 2.9791, "step": 80200 }, { "epoch": 0.0020736, "grad_norm": 0.7530309557914734, "learning_rate": 1.8009719607502902e-05, "loss": 3.0039, "step": 80210 }, { "epoch": 0.0020992, "grad_norm": 0.7248949408531189, "learning_rate": 1.800923745981109e-05, "loss": 3.1007, "step": 80220 }, { "epoch": 0.0021248, "grad_norm": 0.7539255023002625, "learning_rate": 1.80087552601815e-05, "loss": 3.1972, "step": 80230 }, { "epoch": 0.0021504, "grad_norm": 0.9372804164886475, "learning_rate": 1.8008273008617264e-05, "loss": 3.2906, "step": 80240 }, { "epoch": 0.002176, "grad_norm": 0.8636125326156616, "learning_rate": 1.8007790705121505e-05, "loss": 3.2287, "step": 80250 }, { "epoch": 0.0022016, "grad_norm": 0.7586907148361206, "learning_rate": 1.800730834969735e-05, "loss": 3.3989, "step": 80260 }, { "epoch": 0.0022272, "grad_norm": 0.8876907229423523, "learning_rate": 1.800682594234793e-05, "loss": 3.3186, "step": 80270 }, { "epoch": 0.0022528, "grad_norm": 0.838699996471405, "learning_rate": 1.8006343483076366e-05, "loss": 3.0425, "step": 80280 }, { "epoch": 0.0022784, "grad_norm": 0.8913573026657104, "learning_rate": 1.80058609718858e-05, "loss": 3.4049, "step": 80290 }, { "epoch": 0.002304, "grad_norm": 1.04286789894104, "learning_rate": 1.800537840877935e-05, "loss": 3.0905, "step": 80300 }, { "epoch": 0.0023296, "grad_norm": 0.7618810534477234, "learning_rate": 1.800489579376015e-05, "loss": 3.2992, "step": 80310 }, { "epoch": 0.0023552, "grad_norm": 0.7431873679161072, "learning_rate": 1.8004413126831328e-05, "loss": 2.8951, "step": 80320 }, { "epoch": 0.0023808, "grad_norm": 0.8307173252105713, "learning_rate": 1.8003930407996012e-05, "loss": 3.1798, "step": 80330 }, { "epoch": 0.0024064, "grad_norm": 0.6780831217765808, "learning_rate": 1.8003447637257333e-05, "loss": 3.2726, "step": 80340 }, { "epoch": 0.002432, "grad_norm": 0.8236791491508484, "learning_rate": 1.8002964814618428e-05, "loss": 3.158, "step": 80350 }, { "epoch": 0.0024576, "grad_norm": 0.836181104183197, "learning_rate": 1.800248194008242e-05, "loss": 3.2105, "step": 80360 }, { "epoch": 0.0024832, "grad_norm": 0.8021985292434692, "learning_rate": 1.8001999013652445e-05, "loss": 3.1647, "step": 80370 }, { "epoch": 0.0025088, "grad_norm": 0.7797718644142151, "learning_rate": 1.8001516035331632e-05, "loss": 3.2221, "step": 80380 }, { "epoch": 0.0025344, "grad_norm": 0.7257313132286072, "learning_rate": 1.8001033005123116e-05, "loss": 3.1459, "step": 80390 }, { "epoch": 0.00256, "grad_norm": 0.787312388420105, "learning_rate": 1.8000549923030023e-05, "loss": 3.2281, "step": 80400 }, { "epoch": 0.0025856, "grad_norm": 0.819769561290741, "learning_rate": 1.8000066789055495e-05, "loss": 3.4883, "step": 80410 }, { "epoch": 0.0026112, "grad_norm": 0.7261618971824646, "learning_rate": 1.7999583603202658e-05, "loss": 3.1, "step": 80420 }, { "epoch": 0.0026368, "grad_norm": 0.6992372870445251, "learning_rate": 1.7999100365474647e-05, "loss": 3.2196, "step": 80430 }, { "epoch": 0.0026624, "grad_norm": 0.9660477638244629, "learning_rate": 1.7998617075874593e-05, "loss": 3.1589, "step": 80440 }, { "epoch": 0.002688, "grad_norm": 0.7509732842445374, "learning_rate": 1.799813373440564e-05, "loss": 3.2659, "step": 80450 }, { "epoch": 0.0027136, "grad_norm": 0.8879984617233276, "learning_rate": 1.799765034107091e-05, "loss": 3.2518, "step": 80460 }, { "epoch": 0.0027392, "grad_norm": 0.7065734267234802, "learning_rate": 1.7997166895873546e-05, "loss": 3.2028, "step": 80470 }, { "epoch": 0.0027648, "grad_norm": 1.1324504613876343, "learning_rate": 1.7996683398816678e-05, "loss": 3.1758, "step": 80480 }, { "epoch": 0.0027904, "grad_norm": 0.7548165917396545, "learning_rate": 1.7996199849903443e-05, "loss": 3.1537, "step": 80490 }, { "epoch": 0.002816, "grad_norm": 0.6922405362129211, "learning_rate": 1.799571624913698e-05, "loss": 3.1495, "step": 80500 }, { "epoch": 0.0028416, "grad_norm": 0.726028561592102, "learning_rate": 1.799523259652042e-05, "loss": 3.184, "step": 80510 }, { "epoch": 0.0028672, "grad_norm": 0.7527893781661987, "learning_rate": 1.7994748892056903e-05, "loss": 3.2365, "step": 80520 }, { "epoch": 0.0028928, "grad_norm": 0.7776114344596863, "learning_rate": 1.799426513574956e-05, "loss": 3.1572, "step": 80530 }, { "epoch": 0.0029184, "grad_norm": 0.7399207353591919, "learning_rate": 1.7993781327601538e-05, "loss": 3.0233, "step": 80540 }, { "epoch": 0.002944, "grad_norm": 0.7628480195999146, "learning_rate": 1.7993297467615968e-05, "loss": 3.2152, "step": 80550 }, { "epoch": 0.0029696, "grad_norm": 1.1908140182495117, "learning_rate": 1.7992813555795987e-05, "loss": 3.2607, "step": 80560 }, { "epoch": 0.0029952, "grad_norm": 0.7142509818077087, "learning_rate": 1.799232959214473e-05, "loss": 3.3303, "step": 80570 }, { "epoch": 0.0030208, "grad_norm": 0.7268750667572021, "learning_rate": 1.7991845576665342e-05, "loss": 2.9888, "step": 80580 }, { "epoch": 0.0030464, "grad_norm": 1.1284877061843872, "learning_rate": 1.7991361509360964e-05, "loss": 3.1638, "step": 80590 }, { "epoch": 0.003072, "grad_norm": 0.69601970911026, "learning_rate": 1.7990877390234728e-05, "loss": 3.0755, "step": 80600 }, { "epoch": 0.0030976, "grad_norm": 0.7092011570930481, "learning_rate": 1.7990393219289773e-05, "loss": 3.3211, "step": 80610 }, { "epoch": 0.0031232, "grad_norm": 0.7395638227462769, "learning_rate": 1.7989908996529245e-05, "loss": 3.1163, "step": 80620 }, { "epoch": 0.0031488, "grad_norm": 0.9451331496238708, "learning_rate": 1.7989424721956278e-05, "loss": 3.3434, "step": 80630 }, { "epoch": 0.0031744, "grad_norm": 1.0088248252868652, "learning_rate": 1.7988940395574018e-05, "loss": 3.2321, "step": 80640 }, { "epoch": 0.0032, "grad_norm": 0.7991607189178467, "learning_rate": 1.79884560173856e-05, "loss": 3.1079, "step": 80650 }, { "epoch": 0.0032256, "grad_norm": 0.6717747449874878, "learning_rate": 1.798797158739417e-05, "loss": 3.2998, "step": 80660 }, { "epoch": 0.0032512, "grad_norm": 0.8005548715591431, "learning_rate": 1.7987487105602865e-05, "loss": 3.3155, "step": 80670 }, { "epoch": 0.0032768, "grad_norm": 0.8730759620666504, "learning_rate": 1.798700257201483e-05, "loss": 3.3158, "step": 80680 }, { "epoch": 0.0033024, "grad_norm": 0.8134230971336365, "learning_rate": 1.7986517986633204e-05, "loss": 3.0885, "step": 80690 }, { "epoch": 0.003328, "grad_norm": 0.9862978458404541, "learning_rate": 1.7986033349461136e-05, "loss": 3.257, "step": 80700 }, { "epoch": 2.56e-05, "grad_norm": 0.7826172709465027, "learning_rate": 1.7985548660501758e-05, "loss": 2.6032, "step": 80710 }, { "epoch": 5.12e-05, "grad_norm": 0.7780932188034058, "learning_rate": 1.7985063919758224e-05, "loss": 3.1746, "step": 80720 }, { "epoch": 7.68e-05, "grad_norm": 1.7340834140777588, "learning_rate": 1.7984579127233673e-05, "loss": 3.8007, "step": 80730 }, { "epoch": 0.0001024, "grad_norm": 0.7117751240730286, "learning_rate": 1.7984094282931247e-05, "loss": 2.8677, "step": 80740 }, { "epoch": 0.000128, "grad_norm": 0.7959073781967163, "learning_rate": 1.7983609386854092e-05, "loss": 3.2633, "step": 80750 }, { "epoch": 0.0001536, "grad_norm": 0.8110343217849731, "learning_rate": 1.7983124439005348e-05, "loss": 3.2088, "step": 80760 }, { "epoch": 0.0001792, "grad_norm": 1.1512764692306519, "learning_rate": 1.798263943938817e-05, "loss": 3.1272, "step": 80770 }, { "epoch": 0.0002048, "grad_norm": 0.7421654462814331, "learning_rate": 1.798215438800569e-05, "loss": 3.1244, "step": 80780 }, { "epoch": 0.0002304, "grad_norm": 0.7742581963539124, "learning_rate": 1.7981669284861066e-05, "loss": 3.1334, "step": 80790 }, { "epoch": 0.000256, "grad_norm": 0.6662931442260742, "learning_rate": 1.7981184129957435e-05, "loss": 3.0193, "step": 80800 }, { "epoch": 0.0002816, "grad_norm": 0.8519142866134644, "learning_rate": 1.7980698923297947e-05, "loss": 3.1843, "step": 80810 }, { "epoch": 0.0003072, "grad_norm": 0.9113881587982178, "learning_rate": 1.798021366488575e-05, "loss": 3.0315, "step": 80820 }, { "epoch": 0.0003328, "grad_norm": 0.8259074091911316, "learning_rate": 1.7979728354723985e-05, "loss": 3.0401, "step": 80830 }, { "epoch": 0.0003584, "grad_norm": 0.8002187609672546, "learning_rate": 1.7979242992815802e-05, "loss": 2.8247, "step": 80840 }, { "epoch": 0.000384, "grad_norm": 0.7389005422592163, "learning_rate": 1.7978757579164348e-05, "loss": 3.1877, "step": 80850 }, { "epoch": 0.0004096, "grad_norm": 0.6691752076148987, "learning_rate": 1.7978272113772775e-05, "loss": 3.2277, "step": 80860 }, { "epoch": 0.0004352, "grad_norm": 0.7553240656852722, "learning_rate": 1.7977786596644225e-05, "loss": 3.1309, "step": 80870 }, { "epoch": 0.0004608, "grad_norm": 0.8018867373466492, "learning_rate": 1.797730102778185e-05, "loss": 2.7355, "step": 80880 }, { "epoch": 0.0004864, "grad_norm": 0.6714013814926147, "learning_rate": 1.79768154071888e-05, "loss": 2.8737, "step": 80890 }, { "epoch": 0.000512, "grad_norm": 0.7771908044815063, "learning_rate": 1.797632973486822e-05, "loss": 2.9668, "step": 80900 }, { "epoch": 0.0005376, "grad_norm": 0.7193384766578674, "learning_rate": 1.7975844010823264e-05, "loss": 3.1355, "step": 80910 }, { "epoch": 0.0005632, "grad_norm": 0.6738455891609192, "learning_rate": 1.7975358235057077e-05, "loss": 3.1092, "step": 80920 }, { "epoch": 0.0005888, "grad_norm": 0.6689484119415283, "learning_rate": 1.7974872407572813e-05, "loss": 3.0872, "step": 80930 }, { "epoch": 0.0006144, "grad_norm": 1.1003506183624268, "learning_rate": 1.7974386528373617e-05, "loss": 3.1694, "step": 80940 }, { "epoch": 0.00064, "grad_norm": 0.8390071392059326, "learning_rate": 1.7973900597462646e-05, "loss": 3.2099, "step": 80950 }, { "epoch": 0.0006656, "grad_norm": 0.7597544193267822, "learning_rate": 1.797341461484305e-05, "loss": 2.8961, "step": 80960 }, { "epoch": 0.0006912, "grad_norm": 0.8641823530197144, "learning_rate": 1.797292858051798e-05, "loss": 3.1682, "step": 80970 }, { "epoch": 0.0007168, "grad_norm": 1.0092719793319702, "learning_rate": 1.7972442494490583e-05, "loss": 2.8119, "step": 80980 }, { "epoch": 0.0007424, "grad_norm": 0.7545005679130554, "learning_rate": 1.797195635676402e-05, "loss": 3.2404, "step": 80990 }, { "epoch": 0.000768, "grad_norm": 0.9080180525779724, "learning_rate": 1.7971470167341435e-05, "loss": 3.3154, "step": 81000 }, { "epoch": 0.0007936, "grad_norm": 0.7940845489501953, "learning_rate": 1.7970983926225986e-05, "loss": 3.1698, "step": 81010 }, { "epoch": 0.0008192, "grad_norm": 0.9338691830635071, "learning_rate": 1.7970497633420825e-05, "loss": 3.2301, "step": 81020 }, { "epoch": 0.0008448, "grad_norm": 0.7068079113960266, "learning_rate": 1.7970011288929107e-05, "loss": 3.0797, "step": 81030 }, { "epoch": 0.0008704, "grad_norm": 0.719399094581604, "learning_rate": 1.7969524892753977e-05, "loss": 3.0774, "step": 81040 }, { "epoch": 0.000896, "grad_norm": 0.7434094548225403, "learning_rate": 1.7969038444898605e-05, "loss": 2.9987, "step": 81050 }, { "epoch": 0.0009216, "grad_norm": 0.9010665416717529, "learning_rate": 1.796855194536613e-05, "loss": 3.0272, "step": 81060 }, { "epoch": 0.0009472, "grad_norm": 1.1341696977615356, "learning_rate": 1.7968065394159717e-05, "loss": 3.0143, "step": 81070 }, { "epoch": 0.0009728, "grad_norm": 0.6925312876701355, "learning_rate": 1.7967578791282516e-05, "loss": 3.2247, "step": 81080 }, { "epoch": 0.0009984, "grad_norm": 0.7132112979888916, "learning_rate": 1.7967092136737685e-05, "loss": 3.0336, "step": 81090 }, { "epoch": 0.001024, "grad_norm": 0.8559739589691162, "learning_rate": 1.7966605430528377e-05, "loss": 2.9665, "step": 81100 }, { "epoch": 0.0010496, "grad_norm": 0.7248346209526062, "learning_rate": 1.7966118672657754e-05, "loss": 3.1143, "step": 81110 }, { "epoch": 0.0010752, "grad_norm": 0.8059195280075073, "learning_rate": 1.7965631863128963e-05, "loss": 2.9401, "step": 81120 }, { "epoch": 0.0011008, "grad_norm": 0.7489394545555115, "learning_rate": 1.796514500194517e-05, "loss": 3.0743, "step": 81130 }, { "epoch": 0.0011264, "grad_norm": 0.6896579265594482, "learning_rate": 1.796465808910953e-05, "loss": 3.0163, "step": 81140 }, { "epoch": 0.001152, "grad_norm": 0.7707568407058716, "learning_rate": 1.7964171124625195e-05, "loss": 3.2891, "step": 81150 }, { "epoch": 0.0011776, "grad_norm": 0.8174881935119629, "learning_rate": 1.796368410849533e-05, "loss": 2.9564, "step": 81160 }, { "epoch": 0.0012032, "grad_norm": 0.7011561989784241, "learning_rate": 1.7963197040723087e-05, "loss": 2.8875, "step": 81170 }, { "epoch": 0.0012288, "grad_norm": 0.8832111358642578, "learning_rate": 1.796270992131163e-05, "loss": 2.9772, "step": 81180 }, { "epoch": 0.0012544, "grad_norm": 0.7555426955223083, "learning_rate": 1.7962222750264114e-05, "loss": 3.0104, "step": 81190 }, { "epoch": 0.00128, "grad_norm": 0.6734532713890076, "learning_rate": 1.79617355275837e-05, "loss": 3.0397, "step": 81200 }, { "epoch": 2.56e-05, "grad_norm": 0.7699391841888428, "learning_rate": 1.796124825327355e-05, "loss": 2.5389, "step": 81210 }, { "epoch": 5.12e-05, "grad_norm": 0.7929810881614685, "learning_rate": 1.7960760927336817e-05, "loss": 3.2376, "step": 81220 }, { "epoch": 7.68e-05, "grad_norm": 1.7599365711212158, "learning_rate": 1.796027354977667e-05, "loss": 3.4243, "step": 81230 }, { "epoch": 0.0001024, "grad_norm": 0.7623293995857239, "learning_rate": 1.7959786120596257e-05, "loss": 2.9972, "step": 81240 }, { "epoch": 0.000128, "grad_norm": 0.7060447335243225, "learning_rate": 1.795929863979875e-05, "loss": 2.9199, "step": 81250 }, { "epoch": 0.0001536, "grad_norm": 0.7910948991775513, "learning_rate": 1.795881110738731e-05, "loss": 2.8919, "step": 81260 }, { "epoch": 0.0001792, "grad_norm": 1.477159857749939, "learning_rate": 1.7958323523365094e-05, "loss": 3.1433, "step": 81270 }, { "epoch": 0.0002048, "grad_norm": 0.9399178624153137, "learning_rate": 1.795783588773526e-05, "loss": 3.0612, "step": 81280 }, { "epoch": 0.0002304, "grad_norm": 0.747843325138092, "learning_rate": 1.7957348200500982e-05, "loss": 3.1236, "step": 81290 }, { "epoch": 0.000256, "grad_norm": 0.6676900386810303, "learning_rate": 1.795686046166541e-05, "loss": 2.7209, "step": 81300 }, { "epoch": 0.0002816, "grad_norm": 0.8918858170509338, "learning_rate": 1.7956372671231713e-05, "loss": 3.1896, "step": 81310 }, { "epoch": 0.0003072, "grad_norm": 0.7383655309677124, "learning_rate": 1.7955884829203058e-05, "loss": 3.4499, "step": 81320 }, { "epoch": 0.0003328, "grad_norm": 0.9208446145057678, "learning_rate": 1.79553969355826e-05, "loss": 2.9142, "step": 81330 }, { "epoch": 0.0003584, "grad_norm": 0.6930051445960999, "learning_rate": 1.7954908990373512e-05, "loss": 2.7045, "step": 81340 }, { "epoch": 0.000384, "grad_norm": 0.7463699579238892, "learning_rate": 1.7954420993578948e-05, "loss": 3.1038, "step": 81350 }, { "epoch": 0.0004096, "grad_norm": 0.7773091197013855, "learning_rate": 1.7953932945202082e-05, "loss": 3.3455, "step": 81360 }, { "epoch": 0.0004352, "grad_norm": 0.7597787976264954, "learning_rate": 1.7953444845246074e-05, "loss": 2.9709, "step": 81370 }, { "epoch": 0.0004608, "grad_norm": 0.8037691712379456, "learning_rate": 1.795295669371409e-05, "loss": 2.8925, "step": 81380 }, { "epoch": 0.0004864, "grad_norm": 0.7351424098014832, "learning_rate": 1.7952468490609293e-05, "loss": 3.0623, "step": 81390 }, { "epoch": 0.000512, "grad_norm": 0.6426417231559753, "learning_rate": 1.7951980235934853e-05, "loss": 3.0047, "step": 81400 }, { "epoch": 0.0005376, "grad_norm": 0.7682191133499146, "learning_rate": 1.7951491929693935e-05, "loss": 3.0808, "step": 81410 }, { "epoch": 0.0005632, "grad_norm": 0.8017024993896484, "learning_rate": 1.7951003571889703e-05, "loss": 3.0912, "step": 81420 }, { "epoch": 0.0005888, "grad_norm": 0.7480160593986511, "learning_rate": 1.795051516252533e-05, "loss": 3.2071, "step": 81430 }, { "epoch": 0.0006144, "grad_norm": 0.8009031414985657, "learning_rate": 1.7950026701603977e-05, "loss": 3.2546, "step": 81440 }, { "epoch": 0.00064, "grad_norm": 0.9484410881996155, "learning_rate": 1.7949538189128812e-05, "loss": 2.853, "step": 81450 }, { "epoch": 0.0006656, "grad_norm": 1.461549162864685, "learning_rate": 1.7949049625103008e-05, "loss": 2.7667, "step": 81460 }, { "epoch": 0.0006912, "grad_norm": 0.708010196685791, "learning_rate": 1.7948561009529727e-05, "loss": 3.1347, "step": 81470 }, { "epoch": 0.0007168, "grad_norm": 0.8872668147087097, "learning_rate": 1.7948072342412143e-05, "loss": 3.0631, "step": 81480 }, { "epoch": 0.0007424, "grad_norm": 0.7440265417098999, "learning_rate": 1.7947583623753417e-05, "loss": 3.2416, "step": 81490 }, { "epoch": 0.000768, "grad_norm": 0.7644259929656982, "learning_rate": 1.7947094853556727e-05, "loss": 4.0224, "step": 81500 }, { "epoch": 0.0007936, "grad_norm": 0.6592097282409668, "learning_rate": 1.794660603182524e-05, "loss": 3.1436, "step": 81510 }, { "epoch": 0.0008192, "grad_norm": 0.781905472278595, "learning_rate": 1.7946117158562126e-05, "loss": 3.0627, "step": 81520 }, { "epoch": 0.0008448, "grad_norm": 0.7072577476501465, "learning_rate": 1.794562823377055e-05, "loss": 3.0108, "step": 81530 }, { "epoch": 0.0008704, "grad_norm": 0.7967629432678223, "learning_rate": 1.7945139257453688e-05, "loss": 3.2299, "step": 81540 }, { "epoch": 0.000896, "grad_norm": 0.7317695617675781, "learning_rate": 1.794465022961471e-05, "loss": 3.0353, "step": 81550 }, { "epoch": 0.0009216, "grad_norm": 0.8051439523696899, "learning_rate": 1.7944161150256787e-05, "loss": 2.9575, "step": 81560 }, { "epoch": 0.0009472, "grad_norm": 0.7241190671920776, "learning_rate": 1.7943672019383088e-05, "loss": 3.0013, "step": 81570 }, { "epoch": 0.0009728, "grad_norm": 1.0141757726669312, "learning_rate": 1.794318283699679e-05, "loss": 3.2158, "step": 81580 }, { "epoch": 0.0009984, "grad_norm": 0.8973027467727661, "learning_rate": 1.7942693603101063e-05, "loss": 2.9973, "step": 81590 }, { "epoch": 0.001024, "grad_norm": 0.7197697758674622, "learning_rate": 1.7942204317699073e-05, "loss": 2.9905, "step": 81600 }, { "epoch": 0.0010496, "grad_norm": 0.7174480557441711, "learning_rate": 1.7941714980794005e-05, "loss": 3.1811, "step": 81610 }, { "epoch": 0.0010752, "grad_norm": 1.017283320426941, "learning_rate": 1.794122559238902e-05, "loss": 2.9251, "step": 81620 }, { "epoch": 0.0011008, "grad_norm": 0.8065818548202515, "learning_rate": 1.7940736152487304e-05, "loss": 2.9135, "step": 81630 }, { "epoch": 0.0011264, "grad_norm": 0.8495956659317017, "learning_rate": 1.7940246661092025e-05, "loss": 3.2542, "step": 81640 }, { "epoch": 0.001152, "grad_norm": 0.9039235711097717, "learning_rate": 1.793975711820635e-05, "loss": 3.033, "step": 81650 }, { "epoch": 0.0011776, "grad_norm": 0.7020972371101379, "learning_rate": 1.7939267523833465e-05, "loss": 3.0393, "step": 81660 }, { "epoch": 0.0012032, "grad_norm": 0.8525098562240601, "learning_rate": 1.793877787797654e-05, "loss": 2.9014, "step": 81670 }, { "epoch": 0.0012288, "grad_norm": 0.7093810439109802, "learning_rate": 1.793828818063875e-05, "loss": 2.6695, "step": 81680 }, { "epoch": 0.0012544, "grad_norm": 0.6992764472961426, "learning_rate": 1.793779843182327e-05, "loss": 3.1172, "step": 81690 }, { "epoch": 0.00128, "grad_norm": 0.9950869083404541, "learning_rate": 1.7937308631533278e-05, "loss": 3.1768, "step": 81700 }, { "epoch": 0.0013056, "grad_norm": 0.6503626108169556, "learning_rate": 1.7936818779771945e-05, "loss": 3.0536, "step": 81710 }, { "epoch": 0.0013312, "grad_norm": 1.9486854076385498, "learning_rate": 1.7936328876542453e-05, "loss": 3.1182, "step": 81720 }, { "epoch": 0.0013568, "grad_norm": 0.841249942779541, "learning_rate": 1.7935838921847982e-05, "loss": 3.2454, "step": 81730 }, { "epoch": 0.0013824, "grad_norm": 0.7320167422294617, "learning_rate": 1.79353489156917e-05, "loss": 3.0267, "step": 81740 }, { "epoch": 0.001408, "grad_norm": 1.0017842054367065, "learning_rate": 1.793485885807679e-05, "loss": 3.1782, "step": 81750 }, { "epoch": 0.0014336, "grad_norm": 0.8255199790000916, "learning_rate": 1.793436874900643e-05, "loss": 2.9553, "step": 81760 }, { "epoch": 0.0014592, "grad_norm": 0.7523549795150757, "learning_rate": 1.79338785884838e-05, "loss": 3.0105, "step": 81770 }, { "epoch": 0.0014848, "grad_norm": 0.6352896094322205, "learning_rate": 1.793338837651207e-05, "loss": 2.9395, "step": 81780 }, { "epoch": 0.0015104, "grad_norm": 0.644936203956604, "learning_rate": 1.793289811309443e-05, "loss": 3.0641, "step": 81790 }, { "epoch": 0.001536, "grad_norm": 0.7142049074172974, "learning_rate": 1.793240779823405e-05, "loss": 2.9342, "step": 81800 }, { "epoch": 0.0015616, "grad_norm": 0.6547462940216064, "learning_rate": 1.7931917431934118e-05, "loss": 3.0841, "step": 81810 }, { "epoch": 0.0015872, "grad_norm": 0.6741552352905273, "learning_rate": 1.7931427014197807e-05, "loss": 2.756, "step": 81820 }, { "epoch": 0.0016128, "grad_norm": 0.9940823912620544, "learning_rate": 1.79309365450283e-05, "loss": 3.1813, "step": 81830 }, { "epoch": 0.0016384, "grad_norm": 0.7157277464866638, "learning_rate": 1.7930446024428775e-05, "loss": 3.0135, "step": 81840 }, { "epoch": 0.001664, "grad_norm": 0.8033466935157776, "learning_rate": 1.7929955452402418e-05, "loss": 2.888, "step": 81850 }, { "epoch": 0.0016896, "grad_norm": 0.915857195854187, "learning_rate": 1.792946482895241e-05, "loss": 3.0483, "step": 81860 }, { "epoch": 0.0017152, "grad_norm": 0.7613071799278259, "learning_rate": 1.7928974154081927e-05, "loss": 3.0992, "step": 81870 }, { "epoch": 0.0017408, "grad_norm": 1.0329641103744507, "learning_rate": 1.792848342779415e-05, "loss": 3.3326, "step": 81880 }, { "epoch": 0.0017664, "grad_norm": 0.8480429649353027, "learning_rate": 1.792799265009227e-05, "loss": 3.1791, "step": 81890 }, { "epoch": 0.001792, "grad_norm": 1.5434986352920532, "learning_rate": 1.7927501820979464e-05, "loss": 2.9774, "step": 81900 }, { "epoch": 0.0018176, "grad_norm": 0.7254869341850281, "learning_rate": 1.7927010940458916e-05, "loss": 3.3784, "step": 81910 }, { "epoch": 0.0018432, "grad_norm": 0.7302607893943787, "learning_rate": 1.7926520008533806e-05, "loss": 3.2133, "step": 81920 }, { "epoch": 0.0018688, "grad_norm": 0.8797802329063416, "learning_rate": 1.7926029025207325e-05, "loss": 2.8327, "step": 81930 }, { "epoch": 0.0018944, "grad_norm": 0.728505551815033, "learning_rate": 1.792553799048265e-05, "loss": 3.1351, "step": 81940 }, { "epoch": 0.00192, "grad_norm": 0.8320403695106506, "learning_rate": 1.7925046904362967e-05, "loss": 3.0623, "step": 81950 }, { "epoch": 0.0019456, "grad_norm": 0.7180642485618591, "learning_rate": 1.7924555766851458e-05, "loss": 3.1604, "step": 81960 }, { "epoch": 0.0019712, "grad_norm": 0.6884340643882751, "learning_rate": 1.7924064577951313e-05, "loss": 3.1324, "step": 81970 }, { "epoch": 0.0019968, "grad_norm": 0.656367838382721, "learning_rate": 1.792357333766572e-05, "loss": 3.1642, "step": 81980 }, { "epoch": 0.0020224, "grad_norm": 0.9320554733276367, "learning_rate": 1.7923082045997855e-05, "loss": 3.3894, "step": 81990 }, { "epoch": 0.002048, "grad_norm": 1.1053394079208374, "learning_rate": 1.7922590702950908e-05, "loss": 3.1332, "step": 82000 }, { "epoch": 0.0020736, "grad_norm": 0.9270744323730469, "learning_rate": 1.7922099308528068e-05, "loss": 3.4783, "step": 82010 }, { "epoch": 0.0020992, "grad_norm": 0.7804674506187439, "learning_rate": 1.792160786273252e-05, "loss": 2.9055, "step": 82020 }, { "epoch": 0.0021248, "grad_norm": 0.7452394366264343, "learning_rate": 1.792111636556745e-05, "loss": 3.2546, "step": 82030 }, { "epoch": 0.0021504, "grad_norm": 0.8409038782119751, "learning_rate": 1.7920624817036043e-05, "loss": 3.0829, "step": 82040 }, { "epoch": 0.002176, "grad_norm": 0.8793091773986816, "learning_rate": 1.7920133217141492e-05, "loss": 3.4809, "step": 82050 }, { "epoch": 0.0022016, "grad_norm": 1.1483219861984253, "learning_rate": 1.7919641565886977e-05, "loss": 3.1408, "step": 82060 }, { "epoch": 0.0022272, "grad_norm": 0.6999601721763611, "learning_rate": 1.79191498632757e-05, "loss": 3.1973, "step": 82070 }, { "epoch": 0.0022528, "grad_norm": 0.7865033149719238, "learning_rate": 1.7918658109310832e-05, "loss": 3.2056, "step": 82080 }, { "epoch": 0.0022784, "grad_norm": 0.7593871355056763, "learning_rate": 1.7918166303995576e-05, "loss": 3.145, "step": 82090 }, { "epoch": 0.002304, "grad_norm": 0.7158198356628418, "learning_rate": 1.7917674447333114e-05, "loss": 3.0705, "step": 82100 }, { "epoch": 0.0023296, "grad_norm": 1.9093431234359741, "learning_rate": 1.791718253932664e-05, "loss": 3.3191, "step": 82110 }, { "epoch": 0.0023552, "grad_norm": 1.6225661039352417, "learning_rate": 1.791669057997934e-05, "loss": 3.2119, "step": 82120 }, { "epoch": 0.0023808, "grad_norm": 0.7530732154846191, "learning_rate": 1.7916198569294402e-05, "loss": 3.2186, "step": 82130 }, { "epoch": 0.0024064, "grad_norm": 0.8230826258659363, "learning_rate": 1.7915706507275023e-05, "loss": 2.9338, "step": 82140 }, { "epoch": 0.002432, "grad_norm": 0.9479503631591797, "learning_rate": 1.791521439392439e-05, "loss": 3.1788, "step": 82150 }, { "epoch": 0.0024576, "grad_norm": 0.7421655058860779, "learning_rate": 1.79147222292457e-05, "loss": 3.183, "step": 82160 }, { "epoch": 0.0024832, "grad_norm": 0.8855109810829163, "learning_rate": 1.7914230013242134e-05, "loss": 3.2203, "step": 82170 }, { "epoch": 0.0025088, "grad_norm": 0.8210006952285767, "learning_rate": 1.7913737745916894e-05, "loss": 3.2036, "step": 82180 }, { "epoch": 0.0025344, "grad_norm": 2.843090295791626, "learning_rate": 1.7913245427273166e-05, "loss": 3.1412, "step": 82190 }, { "epoch": 0.00256, "grad_norm": 0.7066723108291626, "learning_rate": 1.7912753057314143e-05, "loss": 3.3102, "step": 82200 }, { "epoch": 0.0025856, "grad_norm": 0.7978180646896362, "learning_rate": 1.791226063604302e-05, "loss": 3.3239, "step": 82210 }, { "epoch": 0.0026112, "grad_norm": 0.8960127830505371, "learning_rate": 1.791176816346299e-05, "loss": 3.3007, "step": 82220 }, { "epoch": 0.0026368, "grad_norm": 0.7772094011306763, "learning_rate": 1.7911275639577246e-05, "loss": 3.2968, "step": 82230 }, { "epoch": 0.0026624, "grad_norm": 0.738567054271698, "learning_rate": 1.7910783064388983e-05, "loss": 3.1951, "step": 82240 }, { "epoch": 0.002688, "grad_norm": 0.6693858504295349, "learning_rate": 1.7910290437901393e-05, "loss": 3.2205, "step": 82250 }, { "epoch": 0.0027136, "grad_norm": 0.7648359537124634, "learning_rate": 1.7909797760117674e-05, "loss": 3.0819, "step": 82260 }, { "epoch": 0.0027392, "grad_norm": 0.852625846862793, "learning_rate": 1.7909305031041015e-05, "loss": 3.278, "step": 82270 }, { "epoch": 0.0027648, "grad_norm": 0.8706238865852356, "learning_rate": 1.790881225067462e-05, "loss": 3.3228, "step": 82280 }, { "epoch": 0.0027904, "grad_norm": 0.7275105118751526, "learning_rate": 1.7908319419021674e-05, "loss": 3.1739, "step": 82290 }, { "epoch": 0.002816, "grad_norm": 0.7306494116783142, "learning_rate": 1.7907826536085383e-05, "loss": 3.1472, "step": 82300 }, { "epoch": 0.0028416, "grad_norm": 0.9107081890106201, "learning_rate": 1.7907333601868938e-05, "loss": 3.245, "step": 82310 }, { "epoch": 0.0028672, "grad_norm": 0.8453983664512634, "learning_rate": 1.790684061637553e-05, "loss": 3.2779, "step": 82320 }, { "epoch": 0.0028928, "grad_norm": 1.1364837884902954, "learning_rate": 1.790634757960837e-05, "loss": 3.3638, "step": 82330 }, { "epoch": 0.0029184, "grad_norm": 0.7538272738456726, "learning_rate": 1.7905854491570643e-05, "loss": 3.1622, "step": 82340 }, { "epoch": 0.002944, "grad_norm": 0.7605088353157043, "learning_rate": 1.7905361352265553e-05, "loss": 3.3855, "step": 82350 }, { "epoch": 0.0029696, "grad_norm": 0.7429643869400024, "learning_rate": 1.7904868161696294e-05, "loss": 3.3894, "step": 82360 }, { "epoch": 0.0029952, "grad_norm": 0.7222422361373901, "learning_rate": 1.7904374919866064e-05, "loss": 3.1538, "step": 82370 }, { "epoch": 0.0030208, "grad_norm": 0.7483829855918884, "learning_rate": 1.7903881626778067e-05, "loss": 3.3227, "step": 82380 }, { "epoch": 0.0030464, "grad_norm": 0.7731891870498657, "learning_rate": 1.7903388282435495e-05, "loss": 3.1546, "step": 82390 }, { "epoch": 0.003072, "grad_norm": 0.7016053795814514, "learning_rate": 1.7902894886841556e-05, "loss": 3.1173, "step": 82400 }, { "epoch": 0.0030976, "grad_norm": 0.7829409241676331, "learning_rate": 1.790240143999944e-05, "loss": 3.1689, "step": 82410 }, { "epoch": 0.0031232, "grad_norm": 0.703357458114624, "learning_rate": 1.790190794191235e-05, "loss": 3.2034, "step": 82420 }, { "epoch": 0.0031488, "grad_norm": 0.685565173625946, "learning_rate": 1.7901414392583487e-05, "loss": 3.3336, "step": 82430 }, { "epoch": 0.0031744, "grad_norm": 0.6765757203102112, "learning_rate": 1.7900920792016055e-05, "loss": 3.2237, "step": 82440 }, { "epoch": 0.0032, "grad_norm": 0.7607923150062561, "learning_rate": 1.7900427140213246e-05, "loss": 3.1061, "step": 82450 }, { "epoch": 0.0032256, "grad_norm": 0.6999791264533997, "learning_rate": 1.7899933437178273e-05, "loss": 3.2954, "step": 82460 }, { "epoch": 0.0032512, "grad_norm": 0.8694750070571899, "learning_rate": 1.7899439682914327e-05, "loss": 3.3523, "step": 82470 }, { "epoch": 0.0032768, "grad_norm": 0.6864495277404785, "learning_rate": 1.7898945877424614e-05, "loss": 3.1269, "step": 82480 }, { "epoch": 0.0033024, "grad_norm": 0.8948370814323425, "learning_rate": 1.7898452020712338e-05, "loss": 3.298, "step": 82490 }, { "epoch": 0.003328, "grad_norm": 0.672734260559082, "learning_rate": 1.78979581127807e-05, "loss": 3.1067, "step": 82500 }, { "epoch": 0.0033536, "grad_norm": 0.833482563495636, "learning_rate": 1.7897464153632902e-05, "loss": 3.3109, "step": 82510 }, { "epoch": 0.0033792, "grad_norm": 0.6983487606048584, "learning_rate": 1.7896970143272146e-05, "loss": 3.2738, "step": 82520 }, { "epoch": 0.0034048, "grad_norm": 0.6974329352378845, "learning_rate": 1.789647608170164e-05, "loss": 3.2864, "step": 82530 }, { "epoch": 0.0034304, "grad_norm": 0.7275137305259705, "learning_rate": 1.7895981968924584e-05, "loss": 3.0476, "step": 82540 }, { "epoch": 0.003456, "grad_norm": 0.7444261908531189, "learning_rate": 1.7895487804944184e-05, "loss": 3.1174, "step": 82550 }, { "epoch": 0.0034816, "grad_norm": 0.7882407903671265, "learning_rate": 1.7894993589763645e-05, "loss": 3.1849, "step": 82560 }, { "epoch": 0.0035072, "grad_norm": 0.7405224442481995, "learning_rate": 1.7894499323386168e-05, "loss": 3.119, "step": 82570 }, { "epoch": 0.0035328, "grad_norm": 0.8099742531776428, "learning_rate": 1.789400500581496e-05, "loss": 3.1023, "step": 82580 }, { "epoch": 0.0035584, "grad_norm": 0.6424114108085632, "learning_rate": 1.7893510637053234e-05, "loss": 2.9684, "step": 82590 }, { "epoch": 0.003584, "grad_norm": 0.715383768081665, "learning_rate": 1.7893016217104184e-05, "loss": 3.1925, "step": 82600 }, { "epoch": 0.0036096, "grad_norm": 0.742876410484314, "learning_rate": 1.7892521745971027e-05, "loss": 3.0711, "step": 82610 }, { "epoch": 0.0036352, "grad_norm": 0.7966562509536743, "learning_rate": 1.7892027223656957e-05, "loss": 3.4055, "step": 82620 }, { "epoch": 0.0036608, "grad_norm": 0.6904230117797852, "learning_rate": 1.7891532650165188e-05, "loss": 3.2184, "step": 82630 }, { "epoch": 0.0036864, "grad_norm": 2.290555000305176, "learning_rate": 1.7891038025498933e-05, "loss": 3.4356, "step": 82640 }, { "epoch": 0.003712, "grad_norm": 0.7019874453544617, "learning_rate": 1.7890543349661393e-05, "loss": 3.134, "step": 82650 }, { "epoch": 0.0037376, "grad_norm": 0.907810628414154, "learning_rate": 1.789004862265577e-05, "loss": 3.6062, "step": 82660 }, { "epoch": 0.0037632, "grad_norm": 0.676842987537384, "learning_rate": 1.7889553844485283e-05, "loss": 3.3331, "step": 82670 }, { "epoch": 0.0037888, "grad_norm": 0.8310084939002991, "learning_rate": 1.7889059015153137e-05, "loss": 3.2329, "step": 82680 }, { "epoch": 0.0038144, "grad_norm": 0.9395591020584106, "learning_rate": 1.788856413466254e-05, "loss": 3.2229, "step": 82690 }, { "epoch": 0.00384, "grad_norm": 0.6592369079589844, "learning_rate": 1.78880692030167e-05, "loss": 3.0872, "step": 82700 }, { "epoch": 0.0038656, "grad_norm": 0.7893472909927368, "learning_rate": 1.7887574220218826e-05, "loss": 3.4021, "step": 82710 }, { "epoch": 0.0038912, "grad_norm": 0.7171511650085449, "learning_rate": 1.7887079186272133e-05, "loss": 3.1859, "step": 82720 }, { "epoch": 0.0039168, "grad_norm": 0.7484754323959351, "learning_rate": 1.7886584101179822e-05, "loss": 3.1244, "step": 82730 }, { "epoch": 0.0039424, "grad_norm": 0.6503996253013611, "learning_rate": 1.7886088964945113e-05, "loss": 2.9734, "step": 82740 }, { "epoch": 0.003968, "grad_norm": 0.8342960476875305, "learning_rate": 1.7885593777571212e-05, "loss": 3.3053, "step": 82750 }, { "epoch": 0.0039936, "grad_norm": 1.1802834272384644, "learning_rate": 1.7885098539061333e-05, "loss": 3.1499, "step": 82760 }, { "epoch": 0.0040192, "grad_norm": 0.7891044020652771, "learning_rate": 1.7884603249418683e-05, "loss": 2.9994, "step": 82770 }, { "epoch": 0.0040448, "grad_norm": 0.850142240524292, "learning_rate": 1.788410790864648e-05, "loss": 3.1923, "step": 82780 }, { "epoch": 0.0040704, "grad_norm": 0.8257438540458679, "learning_rate": 1.788361251674793e-05, "loss": 3.2183, "step": 82790 }, { "epoch": 0.004096, "grad_norm": 2.107185125350952, "learning_rate": 1.7883117073726247e-05, "loss": 3.4042, "step": 82800 }, { "epoch": 0.0041216, "grad_norm": 0.7145223021507263, "learning_rate": 1.7882621579584647e-05, "loss": 3.0134, "step": 82810 }, { "epoch": 0.0041472, "grad_norm": 0.8705649971961975, "learning_rate": 1.788212603432634e-05, "loss": 3.1869, "step": 82820 }, { "epoch": 0.0041728, "grad_norm": 0.9002379775047302, "learning_rate": 1.788163043795454e-05, "loss": 3.3544, "step": 82830 }, { "epoch": 0.0041984, "grad_norm": 0.7043635845184326, "learning_rate": 1.788113479047246e-05, "loss": 3.3192, "step": 82840 }, { "epoch": 0.004224, "grad_norm": 0.8256565928459167, "learning_rate": 1.7880639091883318e-05, "loss": 3.0189, "step": 82850 }, { "epoch": 0.0042496, "grad_norm": 0.8251588940620422, "learning_rate": 1.7880143342190326e-05, "loss": 3.2074, "step": 82860 }, { "epoch": 0.0042752, "grad_norm": 0.9777547121047974, "learning_rate": 1.7879647541396696e-05, "loss": 3.4157, "step": 82870 }, { "epoch": 0.0043008, "grad_norm": 1.521761417388916, "learning_rate": 1.787915168950565e-05, "loss": 3.2303, "step": 82880 }, { "epoch": 0.0043264, "grad_norm": 0.7792930603027344, "learning_rate": 1.7878655786520398e-05, "loss": 3.1603, "step": 82890 }, { "epoch": 0.004352, "grad_norm": 0.6986308097839355, "learning_rate": 1.7878159832444155e-05, "loss": 3.22, "step": 82900 }, { "epoch": 0.0043776, "grad_norm": 0.9190073013305664, "learning_rate": 1.787766382728014e-05, "loss": 3.3527, "step": 82910 }, { "epoch": 0.0044032, "grad_norm": 0.846754789352417, "learning_rate": 1.787716777103157e-05, "loss": 3.3324, "step": 82920 }, { "epoch": 0.0044288, "grad_norm": 0.725064754486084, "learning_rate": 1.787667166370166e-05, "loss": 3.1447, "step": 82930 }, { "epoch": 0.0044544, "grad_norm": 0.7237928509712219, "learning_rate": 1.7876175505293627e-05, "loss": 3.3142, "step": 82940 }, { "epoch": 0.00448, "grad_norm": 0.8903592824935913, "learning_rate": 1.787567929581069e-05, "loss": 3.2176, "step": 82950 }, { "epoch": 0.0045056, "grad_norm": 0.8183878660202026, "learning_rate": 1.7875183035256064e-05, "loss": 3.415, "step": 82960 }, { "epoch": 0.0045312, "grad_norm": 0.6858823299407959, "learning_rate": 1.787468672363297e-05, "loss": 3.0411, "step": 82970 }, { "epoch": 0.0045568, "grad_norm": 0.7165356278419495, "learning_rate": 1.7874190360944627e-05, "loss": 3.1881, "step": 82980 }, { "epoch": 0.0045824, "grad_norm": 0.8097560405731201, "learning_rate": 1.787369394719425e-05, "loss": 3.1298, "step": 82990 }, { "epoch": 0.004608, "grad_norm": 0.7645831108093262, "learning_rate": 1.7873197482385062e-05, "loss": 3.143, "step": 83000 }, { "epoch": 0.0046336, "grad_norm": 0.800488293170929, "learning_rate": 1.787270096652028e-05, "loss": 3.2054, "step": 83010 }, { "epoch": 0.0046592, "grad_norm": 0.7477725148200989, "learning_rate": 1.7872204399603123e-05, "loss": 3.2597, "step": 83020 }, { "epoch": 0.0046848, "grad_norm": 0.7768265604972839, "learning_rate": 1.7871707781636816e-05, "loss": 3.0631, "step": 83030 }, { "epoch": 0.0047104, "grad_norm": 0.7537487149238586, "learning_rate": 1.7871211112624576e-05, "loss": 3.2747, "step": 83040 }, { "epoch": 0.004736, "grad_norm": 0.7638651132583618, "learning_rate": 1.787071439256962e-05, "loss": 3.1563, "step": 83050 }, { "epoch": 0.0047616, "grad_norm": 0.8830761909484863, "learning_rate": 1.7870217621475175e-05, "loss": 3.0214, "step": 83060 }, { "epoch": 0.0047872, "grad_norm": 0.7271614074707031, "learning_rate": 1.786972079934446e-05, "loss": 3.2781, "step": 83070 }, { "epoch": 0.0048128, "grad_norm": 0.7513790130615234, "learning_rate": 1.78692239261807e-05, "loss": 2.9921, "step": 83080 }, { "epoch": 0.0048384, "grad_norm": 2.323984146118164, "learning_rate": 1.786872700198711e-05, "loss": 3.103, "step": 83090 }, { "epoch": 0.004864, "grad_norm": 0.6975104212760925, "learning_rate": 1.786823002676692e-05, "loss": 3.1057, "step": 83100 }, { "epoch": 0.0048896, "grad_norm": 0.8035448789596558, "learning_rate": 1.7867733000523346e-05, "loss": 3.1229, "step": 83110 }, { "epoch": 0.0049152, "grad_norm": 0.7211892008781433, "learning_rate": 1.7867235923259617e-05, "loss": 3.0893, "step": 83120 }, { "epoch": 0.0049408, "grad_norm": 0.664517879486084, "learning_rate": 1.7866738794978954e-05, "loss": 3.1324, "step": 83130 }, { "epoch": 0.0049664, "grad_norm": 0.6887834072113037, "learning_rate": 1.7866241615684577e-05, "loss": 2.9461, "step": 83140 }, { "epoch": 0.004992, "grad_norm": 0.7386434674263, "learning_rate": 1.7865744385379718e-05, "loss": 3.1007, "step": 83150 }, { "epoch": 0.0050176, "grad_norm": 0.7195404767990112, "learning_rate": 1.7865247104067598e-05, "loss": 3.2321, "step": 83160 }, { "epoch": 0.0050432, "grad_norm": 0.7083881497383118, "learning_rate": 1.7864749771751436e-05, "loss": 3.1495, "step": 83170 }, { "epoch": 0.0050688, "grad_norm": 0.6898569464683533, "learning_rate": 1.7864252388434466e-05, "loss": 3.087, "step": 83180 }, { "epoch": 0.0050944, "grad_norm": 0.9411585927009583, "learning_rate": 1.786375495411991e-05, "loss": 3.3237, "step": 83190 }, { "epoch": 0.00512, "grad_norm": 0.7057610154151917, "learning_rate": 1.786325746881099e-05, "loss": 3.1061, "step": 83200 }, { "epoch": 0.0051456, "grad_norm": 0.7484002709388733, "learning_rate": 1.7862759932510934e-05, "loss": 3.2181, "step": 83210 }, { "epoch": 0.0051712, "grad_norm": 0.8095123767852783, "learning_rate": 1.786226234522297e-05, "loss": 3.1603, "step": 83220 }, { "epoch": 0.0051968, "grad_norm": 0.7042085528373718, "learning_rate": 1.7861764706950327e-05, "loss": 3.1646, "step": 83230 }, { "epoch": 0.0052224, "grad_norm": 0.7342644333839417, "learning_rate": 1.786126701769623e-05, "loss": 3.4257, "step": 83240 }, { "epoch": 0.005248, "grad_norm": 0.8128789663314819, "learning_rate": 1.7860769277463903e-05, "loss": 3.15, "step": 83250 }, { "epoch": 0.0052736, "grad_norm": 0.719603955745697, "learning_rate": 1.7860271486256576e-05, "loss": 3.1717, "step": 83260 }, { "epoch": 0.0052992, "grad_norm": 0.676590621471405, "learning_rate": 1.785977364407748e-05, "loss": 3.2041, "step": 83270 }, { "epoch": 0.0053248, "grad_norm": 0.7754819393157959, "learning_rate": 1.7859275750929838e-05, "loss": 3.2533, "step": 83280 }, { "epoch": 0.0053504, "grad_norm": 0.8830581307411194, "learning_rate": 1.7858777806816886e-05, "loss": 3.0249, "step": 83290 }, { "epoch": 0.005376, "grad_norm": 0.717320442199707, "learning_rate": 1.7858279811741842e-05, "loss": 3.1352, "step": 83300 }, { "epoch": 0.0054016, "grad_norm": 0.7933763265609741, "learning_rate": 1.7857781765707947e-05, "loss": 3.2162, "step": 83310 }, { "epoch": 0.0054272, "grad_norm": 1.8028072118759155, "learning_rate": 1.7857283668718423e-05, "loss": 3.1256, "step": 83320 }, { "epoch": 0.0054528, "grad_norm": 0.8138003349304199, "learning_rate": 1.7856785520776504e-05, "loss": 3.2703, "step": 83330 }, { "epoch": 0.0054784, "grad_norm": 0.8161110281944275, "learning_rate": 1.785628732188542e-05, "loss": 3.2984, "step": 83340 }, { "epoch": 0.005504, "grad_norm": 0.9189378619194031, "learning_rate": 1.7855789072048398e-05, "loss": 2.9792, "step": 83350 }, { "epoch": 0.0055296, "grad_norm": 0.7331528663635254, "learning_rate": 1.7855290771268673e-05, "loss": 2.8624, "step": 83360 }, { "epoch": 0.0055552, "grad_norm": 1.7005866765975952, "learning_rate": 1.7854792419549475e-05, "loss": 2.9481, "step": 83370 }, { "epoch": 0.0055808, "grad_norm": 0.9867343306541443, "learning_rate": 1.7854294016894035e-05, "loss": 3.121, "step": 83380 }, { "epoch": 0.0056064, "grad_norm": 0.6708330512046814, "learning_rate": 1.7853795563305585e-05, "loss": 3.0321, "step": 83390 }, { "epoch": 0.005632, "grad_norm": 1.3649091720581055, "learning_rate": 1.785329705878736e-05, "loss": 3.2151, "step": 83400 }, { "epoch": 0.0056576, "grad_norm": 0.7302385568618774, "learning_rate": 1.785279850334259e-05, "loss": 3.4067, "step": 83410 }, { "epoch": 0.0056832, "grad_norm": 0.7115927934646606, "learning_rate": 1.785229989697451e-05, "loss": 3.2201, "step": 83420 }, { "epoch": 0.0057088, "grad_norm": 0.8174828290939331, "learning_rate": 1.785180123968635e-05, "loss": 3.3287, "step": 83430 }, { "epoch": 0.0057344, "grad_norm": 0.7378336787223816, "learning_rate": 1.7851302531481348e-05, "loss": 3.2478, "step": 83440 }, { "epoch": 0.00576, "grad_norm": 0.8909204602241516, "learning_rate": 1.7850803772362734e-05, "loss": 3.0875, "step": 83450 }, { "epoch": 0.0057856, "grad_norm": 1.0252165794372559, "learning_rate": 1.7850304962333744e-05, "loss": 3.2436, "step": 83460 }, { "epoch": 0.0058112, "grad_norm": 0.7192528247833252, "learning_rate": 1.7849806101397612e-05, "loss": 3.1615, "step": 83470 }, { "epoch": 0.0058368, "grad_norm": 0.6867434978485107, "learning_rate": 1.7849307189557576e-05, "loss": 3.2806, "step": 83480 }, { "epoch": 0.0058624, "grad_norm": 0.8641122579574585, "learning_rate": 1.7848808226816866e-05, "loss": 3.2986, "step": 83490 }, { "epoch": 0.005888, "grad_norm": 0.9439052939414978, "learning_rate": 1.7848309213178722e-05, "loss": 3.0289, "step": 83500 }, { "epoch": 0.0059136, "grad_norm": 0.8056607246398926, "learning_rate": 1.784781014864638e-05, "loss": 3.2839, "step": 83510 }, { "epoch": 0.0059392, "grad_norm": 0.7714587450027466, "learning_rate": 1.7847311033223076e-05, "loss": 3.1365, "step": 83520 }, { "epoch": 0.0059648, "grad_norm": 0.7538530230522156, "learning_rate": 1.784681186691204e-05, "loss": 3.1133, "step": 83530 }, { "epoch": 0.0059904, "grad_norm": 0.7306691408157349, "learning_rate": 1.784631264971652e-05, "loss": 3.0971, "step": 83540 }, { "epoch": 0.006016, "grad_norm": 0.9377347230911255, "learning_rate": 1.7845813381639744e-05, "loss": 3.2606, "step": 83550 }, { "epoch": 0.0060416, "grad_norm": 0.7881490588188171, "learning_rate": 1.7845314062684956e-05, "loss": 3.1667, "step": 83560 }, { "epoch": 0.0060672, "grad_norm": 0.8442240357398987, "learning_rate": 1.784481469285539e-05, "loss": 3.3113, "step": 83570 }, { "epoch": 0.0060928, "grad_norm": 0.8689817786216736, "learning_rate": 1.7844315272154284e-05, "loss": 3.1585, "step": 83580 }, { "epoch": 0.0061184, "grad_norm": 0.6953778266906738, "learning_rate": 1.7843815800584884e-05, "loss": 3.2639, "step": 83590 }, { "epoch": 0.006144, "grad_norm": 0.7295706272125244, "learning_rate": 1.7843316278150417e-05, "loss": 3.2102, "step": 83600 }, { "epoch": 0.0061696, "grad_norm": 0.8004724979400635, "learning_rate": 1.7842816704854132e-05, "loss": 3.8677, "step": 83610 }, { "epoch": 0.0061952, "grad_norm": 0.7511700391769409, "learning_rate": 1.784231708069926e-05, "loss": 3.1648, "step": 83620 }, { "epoch": 0.0062208, "grad_norm": 0.8713441491127014, "learning_rate": 1.784181740568905e-05, "loss": 3.2648, "step": 83630 }, { "epoch": 0.0062464, "grad_norm": 0.7034662365913391, "learning_rate": 1.784131767982674e-05, "loss": 3.037, "step": 83640 }, { "epoch": 0.006272, "grad_norm": 0.7794750928878784, "learning_rate": 1.7840817903115568e-05, "loss": 3.2513, "step": 83650 }, { "epoch": 0.0062976, "grad_norm": 0.7314984202384949, "learning_rate": 1.7840318075558773e-05, "loss": 3.1719, "step": 83660 }, { "epoch": 0.0063232, "grad_norm": 0.8473241329193115, "learning_rate": 1.78398181971596e-05, "loss": 3.1484, "step": 83670 }, { "epoch": 0.0063488, "grad_norm": 0.7567756772041321, "learning_rate": 1.783931826792129e-05, "loss": 3.2486, "step": 83680 }, { "epoch": 0.0063744, "grad_norm": 0.8887678980827332, "learning_rate": 1.7838818287847086e-05, "loss": 3.0646, "step": 83690 }, { "epoch": 0.0064, "grad_norm": 0.7009726762771606, "learning_rate": 1.7838318256940225e-05, "loss": 3.2521, "step": 83700 }, { "epoch": 0.0064256, "grad_norm": 0.7297385931015015, "learning_rate": 1.7837818175203957e-05, "loss": 3.2612, "step": 83710 }, { "epoch": 0.0064512, "grad_norm": 0.7066751718521118, "learning_rate": 1.7837318042641518e-05, "loss": 3.3154, "step": 83720 }, { "epoch": 0.0064768, "grad_norm": 0.7284857630729675, "learning_rate": 1.783681785925616e-05, "loss": 3.204, "step": 83730 }, { "epoch": 0.0065024, "grad_norm": 0.7319075465202332, "learning_rate": 1.7836317625051115e-05, "loss": 3.2773, "step": 83740 }, { "epoch": 0.006528, "grad_norm": 0.7199162840843201, "learning_rate": 1.7835817340029633e-05, "loss": 3.3048, "step": 83750 }, { "epoch": 0.0065536, "grad_norm": 0.7815208435058594, "learning_rate": 1.7835317004194963e-05, "loss": 3.2812, "step": 83760 }, { "epoch": 0.0065792, "grad_norm": 0.7458821535110474, "learning_rate": 1.783481661755034e-05, "loss": 3.2973, "step": 83770 }, { "epoch": 0.0066048, "grad_norm": 0.8295653462409973, "learning_rate": 1.7834316180099015e-05, "loss": 3.268, "step": 83780 }, { "epoch": 0.0066304, "grad_norm": 0.7037608623504639, "learning_rate": 1.7833815691844232e-05, "loss": 3.1954, "step": 83790 }, { "epoch": 0.006656, "grad_norm": 0.8832063674926758, "learning_rate": 1.7833315152789238e-05, "loss": 3.2427, "step": 83800 }, { "epoch": 0.0066816, "grad_norm": 0.688961923122406, "learning_rate": 1.783281456293727e-05, "loss": 3.1233, "step": 83810 }, { "epoch": 0.0067072, "grad_norm": 0.6931586861610413, "learning_rate": 1.783231392229159e-05, "loss": 3.0619, "step": 83820 }, { "epoch": 0.0067328, "grad_norm": 0.731620728969574, "learning_rate": 1.783181323085543e-05, "loss": 3.0506, "step": 83830 }, { "epoch": 0.0067584, "grad_norm": 0.7011672854423523, "learning_rate": 1.7831312488632046e-05, "loss": 3.1593, "step": 83840 }, { "epoch": 0.006784, "grad_norm": 0.6400138139724731, "learning_rate": 1.783081169562468e-05, "loss": 2.9986, "step": 83850 }, { "epoch": 0.0068096, "grad_norm": 0.7306473255157471, "learning_rate": 1.7830310851836585e-05, "loss": 3.1423, "step": 83860 }, { "epoch": 0.0068352, "grad_norm": 0.7251800298690796, "learning_rate": 1.7829809957270997e-05, "loss": 3.2944, "step": 83870 }, { "epoch": 0.0068608, "grad_norm": 0.7010616064071655, "learning_rate": 1.782930901193118e-05, "loss": 3.3494, "step": 83880 }, { "epoch": 0.0068864, "grad_norm": 0.7256924510002136, "learning_rate": 1.7828808015820373e-05, "loss": 3.1787, "step": 83890 }, { "epoch": 0.006912, "grad_norm": 0.7672321200370789, "learning_rate": 1.7828306968941825e-05, "loss": 3.2512, "step": 83900 }, { "epoch": 0.0069376, "grad_norm": 0.72486811876297, "learning_rate": 1.782780587129879e-05, "loss": 3.208, "step": 83910 }, { "epoch": 0.0069632, "grad_norm": 1.2128115892410278, "learning_rate": 1.782730472289451e-05, "loss": 3.1001, "step": 83920 }, { "epoch": 0.0069888, "grad_norm": 0.6802054643630981, "learning_rate": 1.782680352373224e-05, "loss": 3.1153, "step": 83930 }, { "epoch": 0.0070144, "grad_norm": 0.7495365142822266, "learning_rate": 1.7826302273815233e-05, "loss": 3.1346, "step": 83940 }, { "epoch": 0.00704, "grad_norm": 0.817084789276123, "learning_rate": 1.7825800973146732e-05, "loss": 3.1253, "step": 83950 }, { "epoch": 0.0070656, "grad_norm": 0.7705501914024353, "learning_rate": 1.7825299621729994e-05, "loss": 3.298, "step": 83960 }, { "epoch": 0.0070912, "grad_norm": 0.7475550770759583, "learning_rate": 1.7824798219568264e-05, "loss": 3.175, "step": 83970 }, { "epoch": 0.0071168, "grad_norm": 0.9322152137756348, "learning_rate": 1.7824296766664798e-05, "loss": 3.1376, "step": 83980 }, { "epoch": 0.0071424, "grad_norm": 1.6501811742782593, "learning_rate": 1.7823795263022853e-05, "loss": 3.136, "step": 83990 }, { "epoch": 0.007168, "grad_norm": 1.2838549613952637, "learning_rate": 1.7823293708645667e-05, "loss": 3.2398, "step": 84000 }, { "epoch": 0.0071936, "grad_norm": 0.8420371413230896, "learning_rate": 1.7822792103536506e-05, "loss": 3.2082, "step": 84010 }, { "epoch": 0.0072192, "grad_norm": 0.7540070414543152, "learning_rate": 1.7822290447698613e-05, "loss": 3.0235, "step": 84020 }, { "epoch": 0.0072448, "grad_norm": 0.7136783599853516, "learning_rate": 1.7821788741135248e-05, "loss": 3.2853, "step": 84030 }, { "epoch": 0.0072704, "grad_norm": 0.832988977432251, "learning_rate": 1.7821286983849663e-05, "loss": 3.2321, "step": 84040 }, { "epoch": 0.007296, "grad_norm": 0.6888686418533325, "learning_rate": 1.7820785175845107e-05, "loss": 3.188, "step": 84050 }, { "epoch": 0.0073216, "grad_norm": 0.7963542342185974, "learning_rate": 1.782028331712484e-05, "loss": 3.1085, "step": 84060 }, { "epoch": 0.0073472, "grad_norm": 0.7020915150642395, "learning_rate": 1.7819781407692112e-05, "loss": 3.1891, "step": 84070 }, { "epoch": 0.0073728, "grad_norm": 0.6537467837333679, "learning_rate": 1.7819279447550183e-05, "loss": 3.1549, "step": 84080 }, { "epoch": 0.0073984, "grad_norm": 0.8241604566574097, "learning_rate": 1.7818777436702303e-05, "loss": 3.2533, "step": 84090 }, { "epoch": 0.007424, "grad_norm": 1.9733812808990479, "learning_rate": 1.781827537515173e-05, "loss": 2.9162, "step": 84100 }, { "epoch": 0.0074496, "grad_norm": 0.6528661251068115, "learning_rate": 1.7817773262901717e-05, "loss": 3.1691, "step": 84110 }, { "epoch": 0.0074752, "grad_norm": 0.7255703210830688, "learning_rate": 1.7817271099955526e-05, "loss": 3.1515, "step": 84120 }, { "epoch": 0.0075008, "grad_norm": 0.7791467308998108, "learning_rate": 1.7816768886316405e-05, "loss": 3.0868, "step": 84130 }, { "epoch": 0.0075264, "grad_norm": 0.7121562957763672, "learning_rate": 1.7816266621987617e-05, "loss": 3.2164, "step": 84140 }, { "epoch": 0.007552, "grad_norm": 0.7664094567298889, "learning_rate": 1.781576430697242e-05, "loss": 3.2487, "step": 84150 }, { "epoch": 0.0075776, "grad_norm": 0.6534966230392456, "learning_rate": 1.7815261941274065e-05, "loss": 2.9886, "step": 84160 }, { "epoch": 0.0076032, "grad_norm": 0.7196030616760254, "learning_rate": 1.7814759524895816e-05, "loss": 3.2122, "step": 84170 }, { "epoch": 0.0076288, "grad_norm": 0.7196028828620911, "learning_rate": 1.7814257057840928e-05, "loss": 3.2958, "step": 84180 }, { "epoch": 0.0076544, "grad_norm": 0.6949611306190491, "learning_rate": 1.781375454011266e-05, "loss": 3.1319, "step": 84190 }, { "epoch": 0.00768, "grad_norm": 0.7902724742889404, "learning_rate": 1.7813251971714266e-05, "loss": 3.163, "step": 84200 }, { "epoch": 0.0077056, "grad_norm": 0.7420744299888611, "learning_rate": 1.7812749352649015e-05, "loss": 3.3004, "step": 84210 }, { "epoch": 0.0077312, "grad_norm": 0.7246090173721313, "learning_rate": 1.781224668292016e-05, "loss": 3.2597, "step": 84220 }, { "epoch": 0.0077568, "grad_norm": 1.0378881692886353, "learning_rate": 1.781174396253096e-05, "loss": 3.2019, "step": 84230 }, { "epoch": 0.0077824, "grad_norm": 0.8457086086273193, "learning_rate": 1.7811241191484675e-05, "loss": 3.3367, "step": 84240 }, { "epoch": 0.007808, "grad_norm": 0.8609301447868347, "learning_rate": 1.781073836978457e-05, "loss": 3.1451, "step": 84250 }, { "epoch": 0.0078336, "grad_norm": 0.7682400941848755, "learning_rate": 1.7810235497433898e-05, "loss": 3.1856, "step": 84260 }, { "epoch": 0.0078592, "grad_norm": 0.809006929397583, "learning_rate": 1.780973257443593e-05, "loss": 3.1572, "step": 84270 }, { "epoch": 0.0078848, "grad_norm": 0.9478552937507629, "learning_rate": 1.7809229600793922e-05, "loss": 3.2995, "step": 84280 }, { "epoch": 0.0079104, "grad_norm": 0.8436858654022217, "learning_rate": 1.780872657651113e-05, "loss": 3.1467, "step": 84290 }, { "epoch": 0.007936, "grad_norm": 0.9827684164047241, "learning_rate": 1.7808223501590824e-05, "loss": 3.1734, "step": 84300 }, { "epoch": 0.0079616, "grad_norm": 0.7752622365951538, "learning_rate": 1.7807720376036262e-05, "loss": 3.0264, "step": 84310 }, { "epoch": 0.0079872, "grad_norm": 1.0379178524017334, "learning_rate": 1.7807217199850713e-05, "loss": 2.9768, "step": 84320 }, { "epoch": 0.0080128, "grad_norm": 0.6506549715995789, "learning_rate": 1.7806713973037432e-05, "loss": 3.0392, "step": 84330 }, { "epoch": 0.0080384, "grad_norm": 0.7847203612327576, "learning_rate": 1.7806210695599685e-05, "loss": 3.5149, "step": 84340 }, { "epoch": 0.008064, "grad_norm": 0.8124104738235474, "learning_rate": 1.780570736754074e-05, "loss": 3.3555, "step": 84350 }, { "epoch": 0.0080896, "grad_norm": 0.8079211711883545, "learning_rate": 1.7805203988863856e-05, "loss": 3.1328, "step": 84360 }, { "epoch": 0.0081152, "grad_norm": 0.8277809619903564, "learning_rate": 1.7804700559572295e-05, "loss": 3.1671, "step": 84370 }, { "epoch": 0.0081408, "grad_norm": 0.7223030924797058, "learning_rate": 1.780419707966933e-05, "loss": 2.9503, "step": 84380 }, { "epoch": 0.0081664, "grad_norm": 0.7295430898666382, "learning_rate": 1.7803693549158217e-05, "loss": 3.0934, "step": 84390 }, { "epoch": 0.008192, "grad_norm": 0.8168074488639832, "learning_rate": 1.7803189968042228e-05, "loss": 3.5326, "step": 84400 }, { "epoch": 0.0082176, "grad_norm": 0.9441348910331726, "learning_rate": 1.7802686336324626e-05, "loss": 3.1304, "step": 84410 }, { "epoch": 0.0082432, "grad_norm": 0.9180914163589478, "learning_rate": 1.7802182654008675e-05, "loss": 3.2002, "step": 84420 }, { "epoch": 0.0082688, "grad_norm": 0.9115668535232544, "learning_rate": 1.7801678921097646e-05, "loss": 3.3262, "step": 84430 }, { "epoch": 0.0082944, "grad_norm": 0.890538215637207, "learning_rate": 1.7801175137594797e-05, "loss": 3.2299, "step": 84440 }, { "epoch": 0.00832, "grad_norm": 0.6725094318389893, "learning_rate": 1.780067130350341e-05, "loss": 3.2144, "step": 84450 }, { "epoch": 0.0083456, "grad_norm": 0.7880932688713074, "learning_rate": 1.7800167418826733e-05, "loss": 3.1322, "step": 84460 }, { "epoch": 0.0083712, "grad_norm": 0.6649607419967651, "learning_rate": 1.7799663483568047e-05, "loss": 3.2221, "step": 84470 }, { "epoch": 0.0083968, "grad_norm": 0.8926787972450256, "learning_rate": 1.7799159497730616e-05, "loss": 3.2798, "step": 84480 }, { "epoch": 0.0084224, "grad_norm": 0.7987045645713806, "learning_rate": 1.7798655461317707e-05, "loss": 3.1934, "step": 84490 }, { "epoch": 0.008448, "grad_norm": 0.7311214804649353, "learning_rate": 1.7798151374332592e-05, "loss": 3.119, "step": 84500 }, { "epoch": 0.0084736, "grad_norm": 0.7538623809814453, "learning_rate": 1.779764723677854e-05, "loss": 3.1722, "step": 84510 }, { "epoch": 0.0084992, "grad_norm": 0.9030731320381165, "learning_rate": 1.7797143048658814e-05, "loss": 3.1825, "step": 84520 }, { "epoch": 0.0085248, "grad_norm": 0.6963644027709961, "learning_rate": 1.7796638809976688e-05, "loss": 3.1621, "step": 84530 }, { "epoch": 0.0085504, "grad_norm": 0.7893673181533813, "learning_rate": 1.7796134520735436e-05, "loss": 3.2476, "step": 84540 }, { "epoch": 0.008576, "grad_norm": 0.7062170505523682, "learning_rate": 1.7795630180938318e-05, "loss": 3.2698, "step": 84550 }, { "epoch": 0.0086016, "grad_norm": 0.7472831010818481, "learning_rate": 1.7795125790588613e-05, "loss": 3.2522, "step": 84560 }, { "epoch": 0.0086272, "grad_norm": 0.7608307600021362, "learning_rate": 1.7794621349689588e-05, "loss": 3.3769, "step": 84570 }, { "epoch": 0.0086528, "grad_norm": 0.9880807995796204, "learning_rate": 1.7794116858244516e-05, "loss": 3.1304, "step": 84580 }, { "epoch": 0.0086784, "grad_norm": 0.8718721270561218, "learning_rate": 1.7793612316256666e-05, "loss": 3.0181, "step": 84590 }, { "epoch": 0.008704, "grad_norm": 0.7275211811065674, "learning_rate": 1.7793107723729314e-05, "loss": 3.3419, "step": 84600 }, { "epoch": 0.0087296, "grad_norm": 0.8000180721282959, "learning_rate": 1.7792603080665727e-05, "loss": 3.4708, "step": 84610 }, { "epoch": 0.0087552, "grad_norm": 0.8722391724586487, "learning_rate": 1.779209838706918e-05, "loss": 3.3544, "step": 84620 }, { "epoch": 0.0087808, "grad_norm": 2.350494146347046, "learning_rate": 1.779159364294295e-05, "loss": 3.2897, "step": 84630 }, { "epoch": 0.0088064, "grad_norm": 1.1722723245620728, "learning_rate": 1.77910888482903e-05, "loss": 7.0137, "step": 84640 }, { "epoch": 0.008832, "grad_norm": 0.7415908575057983, "learning_rate": 1.779058400311451e-05, "loss": 3.2841, "step": 84650 }, { "epoch": 0.0088576, "grad_norm": 0.693681001663208, "learning_rate": 1.7790079107418855e-05, "loss": 3.1614, "step": 84660 }, { "epoch": 0.0088832, "grad_norm": 0.688596248626709, "learning_rate": 1.7789574161206608e-05, "loss": 3.2575, "step": 84670 }, { "epoch": 0.0089088, "grad_norm": 0.7769680023193359, "learning_rate": 1.7789069164481046e-05, "loss": 3.1863, "step": 84680 }, { "epoch": 0.0089344, "grad_norm": 0.681415319442749, "learning_rate": 1.7788564117245436e-05, "loss": 3.2408, "step": 84690 }, { "epoch": 0.00896, "grad_norm": 1.0043747425079346, "learning_rate": 1.778805901950306e-05, "loss": 3.0301, "step": 84700 }, { "epoch": 0.0089856, "grad_norm": 0.8182374238967896, "learning_rate": 1.778755387125719e-05, "loss": 3.0283, "step": 84710 }, { "epoch": 0.0090112, "grad_norm": 0.7056345343589783, "learning_rate": 1.77870486725111e-05, "loss": 3.2313, "step": 84720 }, { "epoch": 0.0090368, "grad_norm": 0.7840628623962402, "learning_rate": 1.7786543423268074e-05, "loss": 3.0924, "step": 84730 }, { "epoch": 0.0090624, "grad_norm": 0.964735746383667, "learning_rate": 1.778603812353138e-05, "loss": 3.1433, "step": 84740 }, { "epoch": 0.009088, "grad_norm": 0.7786731719970703, "learning_rate": 1.7785532773304298e-05, "loss": 3.1822, "step": 84750 }, { "epoch": 0.0091136, "grad_norm": 0.6986286044120789, "learning_rate": 1.7785027372590106e-05, "loss": 3.3273, "step": 84760 }, { "epoch": 0.0091392, "grad_norm": 0.7802546620368958, "learning_rate": 1.7784521921392082e-05, "loss": 3.1102, "step": 84770 }, { "epoch": 0.0091648, "grad_norm": 0.7760013341903687, "learning_rate": 1.77840164197135e-05, "loss": 3.5255, "step": 84780 }, { "epoch": 0.0091904, "grad_norm": 0.8992941975593567, "learning_rate": 1.778351086755764e-05, "loss": 3.1666, "step": 84790 }, { "epoch": 0.009216, "grad_norm": 0.8548614978790283, "learning_rate": 1.7783005264927782e-05, "loss": 3.3077, "step": 84800 }, { "epoch": 0.0092416, "grad_norm": 1.1103311777114868, "learning_rate": 1.7782499611827204e-05, "loss": 3.3036, "step": 84810 }, { "epoch": 0.0092672, "grad_norm": 0.9433510303497314, "learning_rate": 1.7781993908259183e-05, "loss": 3.1736, "step": 84820 }, { "epoch": 0.0092928, "grad_norm": 0.7270353436470032, "learning_rate": 1.7781488154227e-05, "loss": 3.2309, "step": 84830 }, { "epoch": 0.0093184, "grad_norm": 0.923827588558197, "learning_rate": 1.7780982349733936e-05, "loss": 3.1882, "step": 84840 }, { "epoch": 0.009344, "grad_norm": 0.6551488041877747, "learning_rate": 1.778047649478327e-05, "loss": 3.2147, "step": 84850 }, { "epoch": 0.0093696, "grad_norm": 0.7229903936386108, "learning_rate": 1.7779970589378278e-05, "loss": 3.1154, "step": 84860 }, { "epoch": 0.0093952, "grad_norm": 0.7821279764175415, "learning_rate": 1.7779464633522243e-05, "loss": 3.0083, "step": 84870 }, { "epoch": 0.0094208, "grad_norm": 0.7285717725753784, "learning_rate": 1.7778958627218454e-05, "loss": 3.0923, "step": 84880 }, { "epoch": 0.0094464, "grad_norm": 0.7968716621398926, "learning_rate": 1.777845257047018e-05, "loss": 3.0987, "step": 84890 }, { "epoch": 0.009472, "grad_norm": 0.6947049498558044, "learning_rate": 1.777794646328071e-05, "loss": 3.0252, "step": 84900 }, { "epoch": 0.0094976, "grad_norm": 0.7409247756004333, "learning_rate": 1.7777440305653323e-05, "loss": 3.2129, "step": 84910 }, { "epoch": 0.0095232, "grad_norm": 0.9307373762130737, "learning_rate": 1.7776934097591306e-05, "loss": 3.0914, "step": 84920 }, { "epoch": 0.0095488, "grad_norm": 2.4571588039398193, "learning_rate": 1.7776427839097938e-05, "loss": 3.1747, "step": 84930 }, { "epoch": 0.0095744, "grad_norm": 0.9248173236846924, "learning_rate": 1.77759215301765e-05, "loss": 3.4891, "step": 84940 }, { "epoch": 0.0096, "grad_norm": 0.6491432785987854, "learning_rate": 1.777541517083028e-05, "loss": 3.0613, "step": 84950 }, { "epoch": 0.0096256, "grad_norm": 0.8072100877761841, "learning_rate": 1.7774908761062556e-05, "loss": 2.9559, "step": 84960 }, { "epoch": 0.0096512, "grad_norm": 0.7664884924888611, "learning_rate": 1.7774402300876615e-05, "loss": 3.2842, "step": 84970 }, { "epoch": 0.0096768, "grad_norm": 1.4422284364700317, "learning_rate": 1.7773895790275745e-05, "loss": 3.0545, "step": 84980 }, { "epoch": 0.0097024, "grad_norm": 0.7410898208618164, "learning_rate": 1.777338922926322e-05, "loss": 3.1174, "step": 84990 }, { "epoch": 0.009728, "grad_norm": 0.7437371015548706, "learning_rate": 1.777288261784234e-05, "loss": 3.0646, "step": 85000 }, { "epoch": 0.0097536, "grad_norm": 0.814927875995636, "learning_rate": 1.7772375956016378e-05, "loss": 3.1996, "step": 85010 }, { "epoch": 0.0097792, "grad_norm": 0.8044995069503784, "learning_rate": 1.7771869243788625e-05, "loss": 3.0601, "step": 85020 }, { "epoch": 0.0098048, "grad_norm": 0.6742279529571533, "learning_rate": 1.777136248116236e-05, "loss": 3.3547, "step": 85030 }, { "epoch": 0.0098304, "grad_norm": 0.8719801902770996, "learning_rate": 1.7770855668140883e-05, "loss": 3.1094, "step": 85040 }, { "epoch": 0.009856, "grad_norm": 0.8188557028770447, "learning_rate": 1.7770348804727466e-05, "loss": 3.187, "step": 85050 }, { "epoch": 0.0098816, "grad_norm": 1.1897315979003906, "learning_rate": 1.7769841890925404e-05, "loss": 3.5512, "step": 85060 }, { "epoch": 0.0099072, "grad_norm": 1.3451931476593018, "learning_rate": 1.7769334926737984e-05, "loss": 3.4786, "step": 85070 }, { "epoch": 0.0099328, "grad_norm": 1.0826619863510132, "learning_rate": 1.7768827912168492e-05, "loss": 3.3913, "step": 85080 }, { "epoch": 0.0099584, "grad_norm": 0.7003690600395203, "learning_rate": 1.776832084722021e-05, "loss": 3.1033, "step": 85090 }, { "epoch": 0.009984, "grad_norm": 0.8789432048797607, "learning_rate": 1.7767813731896438e-05, "loss": 3.2615, "step": 85100 }, { "epoch": 0.0100096, "grad_norm": 0.9439405202865601, "learning_rate": 1.7767306566200456e-05, "loss": 3.188, "step": 85110 }, { "epoch": 0.0100352, "grad_norm": 0.6917063593864441, "learning_rate": 1.7766799350135558e-05, "loss": 3.2649, "step": 85120 }, { "epoch": 0.0100608, "grad_norm": 0.8018799424171448, "learning_rate": 1.7766292083705027e-05, "loss": 3.3069, "step": 85130 }, { "epoch": 0.0100864, "grad_norm": 0.693027913570404, "learning_rate": 1.7765784766912156e-05, "loss": 3.1827, "step": 85140 }, { "epoch": 0.010112, "grad_norm": 0.8248261213302612, "learning_rate": 1.7765277399760236e-05, "loss": 3.0058, "step": 85150 }, { "epoch": 0.0101376, "grad_norm": 0.8851009607315063, "learning_rate": 1.7764769982252557e-05, "loss": 3.1781, "step": 85160 }, { "epoch": 0.0101632, "grad_norm": 0.9861998558044434, "learning_rate": 1.7764262514392404e-05, "loss": 3.3615, "step": 85170 }, { "epoch": 0.0101888, "grad_norm": 1.4853113889694214, "learning_rate": 1.7763754996183074e-05, "loss": 3.2729, "step": 85180 }, { "epoch": 0.0102144, "grad_norm": 0.8522446155548096, "learning_rate": 1.7763247427627858e-05, "loss": 3.3691, "step": 85190 }, { "epoch": 0.01024, "grad_norm": 1.755995512008667, "learning_rate": 1.7762739808730045e-05, "loss": 3.3997, "step": 85200 }, { "epoch": 0.0102656, "grad_norm": 3.878840208053589, "learning_rate": 1.776223213949293e-05, "loss": 3.1894, "step": 85210 }, { "epoch": 0.0102912, "grad_norm": 0.717476487159729, "learning_rate": 1.7761724419919798e-05, "loss": 3.3114, "step": 85220 }, { "epoch": 0.0103168, "grad_norm": 0.8293439149856567, "learning_rate": 1.7761216650013947e-05, "loss": 3.2464, "step": 85230 }, { "epoch": 0.0103424, "grad_norm": 0.7636751532554626, "learning_rate": 1.776070882977867e-05, "loss": 3.3187, "step": 85240 }, { "epoch": 0.010368, "grad_norm": 1.0033780336380005, "learning_rate": 1.7760200959217257e-05, "loss": 2.9494, "step": 85250 }, { "epoch": 0.0103936, "grad_norm": 0.7227867841720581, "learning_rate": 1.7759693038333004e-05, "loss": 3.0345, "step": 85260 }, { "epoch": 0.0104192, "grad_norm": 0.6740532517433167, "learning_rate": 1.7759185067129204e-05, "loss": 3.197, "step": 85270 }, { "epoch": 0.0104448, "grad_norm": 1.0932906866073608, "learning_rate": 1.7758677045609153e-05, "loss": 3.3665, "step": 85280 }, { "epoch": 0.0104704, "grad_norm": 0.7314775586128235, "learning_rate": 1.7758168973776142e-05, "loss": 3.0369, "step": 85290 }, { "epoch": 0.010496, "grad_norm": 0.9896077513694763, "learning_rate": 1.7757660851633466e-05, "loss": 3.1317, "step": 85300 }, { "epoch": 0.0105216, "grad_norm": 1.353715181350708, "learning_rate": 1.7757152679184423e-05, "loss": 3.2067, "step": 85310 }, { "epoch": 0.0105472, "grad_norm": 0.7971328496932983, "learning_rate": 1.7756644456432305e-05, "loss": 3.1307, "step": 85320 }, { "epoch": 0.0105728, "grad_norm": 0.8691307902336121, "learning_rate": 1.7756136183380407e-05, "loss": 3.207, "step": 85330 }, { "epoch": 0.0105984, "grad_norm": 0.8198398947715759, "learning_rate": 1.7755627860032034e-05, "loss": 3.1765, "step": 85340 }, { "epoch": 0.010624, "grad_norm": 0.6731681227684021, "learning_rate": 1.775511948639047e-05, "loss": 3.1222, "step": 85350 }, { "epoch": 0.0106496, "grad_norm": 0.7129675149917603, "learning_rate": 1.7754611062459015e-05, "loss": 3.0234, "step": 85360 }, { "epoch": 0.0106752, "grad_norm": 0.6902815699577332, "learning_rate": 1.7754102588240972e-05, "loss": 3.2432, "step": 85370 }, { "epoch": 0.0107008, "grad_norm": 0.8084467649459839, "learning_rate": 1.7753594063739635e-05, "loss": 3.5915, "step": 85380 }, { "epoch": 0.0107264, "grad_norm": 0.7683323621749878, "learning_rate": 1.77530854889583e-05, "loss": 3.3624, "step": 85390 }, { "epoch": 0.010752, "grad_norm": 0.9171941876411438, "learning_rate": 1.7752576863900264e-05, "loss": 3.1938, "step": 85400 }, { "epoch": 0.0107776, "grad_norm": 0.739820122718811, "learning_rate": 1.775206818856883e-05, "loss": 3.3288, "step": 85410 }, { "epoch": 0.0108032, "grad_norm": 0.9810587167739868, "learning_rate": 1.77515594629673e-05, "loss": 3.1573, "step": 85420 }, { "epoch": 0.0108288, "grad_norm": 0.690439760684967, "learning_rate": 1.7751050687098958e-05, "loss": 3.0335, "step": 85430 }, { "epoch": 0.0108544, "grad_norm": 2.265007734298706, "learning_rate": 1.7750541860967115e-05, "loss": 3.7126, "step": 85440 }, { "epoch": 0.01088, "grad_norm": 0.830422580242157, "learning_rate": 1.7750032984575067e-05, "loss": 3.1866, "step": 85450 }, { "epoch": 0.0109056, "grad_norm": 1.0546425580978394, "learning_rate": 1.7749524057926116e-05, "loss": 3.3463, "step": 85460 }, { "epoch": 0.0109312, "grad_norm": 1.2628676891326904, "learning_rate": 1.7749015081023563e-05, "loss": 3.2941, "step": 85470 }, { "epoch": 0.0109568, "grad_norm": 1.471111536026001, "learning_rate": 1.7748506053870704e-05, "loss": 3.2166, "step": 85480 }, { "epoch": 0.0109824, "grad_norm": 1.1128631830215454, "learning_rate": 1.774799697647084e-05, "loss": 3.1922, "step": 85490 }, { "epoch": 0.011008, "grad_norm": 0.8674927353858948, "learning_rate": 1.774748784882728e-05, "loss": 3.434, "step": 85500 }, { "epoch": 2.56e-05, "grad_norm": 0.8865966796875, "learning_rate": 1.7746978670943316e-05, "loss": 2.7577, "step": 85510 }, { "epoch": 5.12e-05, "grad_norm": 1.1015363931655884, "learning_rate": 1.7746469442822257e-05, "loss": 3.1253, "step": 85520 }, { "epoch": 7.68e-05, "grad_norm": 0.6334646344184875, "learning_rate": 1.77459601644674e-05, "loss": 2.8758, "step": 85530 }, { "epoch": 0.0001024, "grad_norm": 0.7993218302726746, "learning_rate": 1.7745450835882055e-05, "loss": 2.9925, "step": 85540 }, { "epoch": 0.000128, "grad_norm": 0.8057848215103149, "learning_rate": 1.7744941457069514e-05, "loss": 3.0808, "step": 85550 }, { "epoch": 0.0001536, "grad_norm": 0.7371962070465088, "learning_rate": 1.774443202803309e-05, "loss": 3.2216, "step": 85560 }, { "epoch": 0.0001792, "grad_norm": 0.6878194212913513, "learning_rate": 1.7743922548776083e-05, "loss": 3.1582, "step": 85570 }, { "epoch": 0.0002048, "grad_norm": 0.9298004508018494, "learning_rate": 1.7743413019301793e-05, "loss": 2.953, "step": 85580 }, { "epoch": 0.0002304, "grad_norm": 0.760863721370697, "learning_rate": 1.774290343961353e-05, "loss": 3.0782, "step": 85590 }, { "epoch": 0.000256, "grad_norm": 1.0783452987670898, "learning_rate": 1.7742393809714592e-05, "loss": 3.0935, "step": 85600 }, { "epoch": 0.0002816, "grad_norm": 0.7652842402458191, "learning_rate": 1.7741884129608292e-05, "loss": 3.1615, "step": 85610 }, { "epoch": 0.0003072, "grad_norm": 0.8976060748100281, "learning_rate": 1.774137439929793e-05, "loss": 3.0722, "step": 85620 }, { "epoch": 0.0003328, "grad_norm": 0.7002584934234619, "learning_rate": 1.7740864618786814e-05, "loss": 3.0396, "step": 85630 }, { "epoch": 0.0003584, "grad_norm": 0.7916409373283386, "learning_rate": 1.7740354788078246e-05, "loss": 3.1128, "step": 85640 }, { "epoch": 0.000384, "grad_norm": 0.8791379332542419, "learning_rate": 1.7739844907175535e-05, "loss": 3.361, "step": 85650 }, { "epoch": 0.0004096, "grad_norm": 0.7594455480575562, "learning_rate": 1.7739334976081982e-05, "loss": 2.8182, "step": 85660 }, { "epoch": 0.0004352, "grad_norm": 0.6588128209114075, "learning_rate": 1.7738824994800907e-05, "loss": 3.0099, "step": 85670 }, { "epoch": 0.0004608, "grad_norm": 0.7201749682426453, "learning_rate": 1.7738314963335603e-05, "loss": 3.1475, "step": 85680 }, { "epoch": 0.0004864, "grad_norm": 1.0985052585601807, "learning_rate": 1.7737804881689387e-05, "loss": 3.2234, "step": 85690 }, { "epoch": 0.000512, "grad_norm": 0.6720535755157471, "learning_rate": 1.773729474986556e-05, "loss": 2.9148, "step": 85700 }, { "epoch": 0.0005376, "grad_norm": 0.9206758141517639, "learning_rate": 1.773678456786743e-05, "loss": 3.0033, "step": 85710 }, { "epoch": 0.0005632, "grad_norm": 0.6746699213981628, "learning_rate": 1.7736274335698317e-05, "loss": 3.0752, "step": 85720 }, { "epoch": 0.0005888, "grad_norm": 0.7327659130096436, "learning_rate": 1.7735764053361512e-05, "loss": 3.0566, "step": 85730 }, { "epoch": 0.0006144, "grad_norm": 0.7405019998550415, "learning_rate": 1.773525372086034e-05, "loss": 3.2904, "step": 85740 }, { "epoch": 0.00064, "grad_norm": 0.6941512227058411, "learning_rate": 1.77347433381981e-05, "loss": 2.8824, "step": 85750 }, { "epoch": 0.0006656, "grad_norm": 0.8975553512573242, "learning_rate": 1.7734232905378103e-05, "loss": 2.8766, "step": 85760 }, { "epoch": 0.0006912, "grad_norm": 0.837807834148407, "learning_rate": 1.773372242240366e-05, "loss": 3.2182, "step": 85770 }, { "epoch": 0.0007168, "grad_norm": 0.7396982312202454, "learning_rate": 1.7733211889278087e-05, "loss": 2.9259, "step": 85780 }, { "epoch": 0.0007424, "grad_norm": 0.8224688768386841, "learning_rate": 1.7732701306004692e-05, "loss": 3.1987, "step": 85790 }, { "epoch": 0.000768, "grad_norm": 0.6894724369049072, "learning_rate": 1.773219067258678e-05, "loss": 2.9413, "step": 85800 }, { "epoch": 0.0007936, "grad_norm": 0.6797929406166077, "learning_rate": 1.7731679989027668e-05, "loss": 2.9471, "step": 85810 }, { "epoch": 0.0008192, "grad_norm": 1.1049962043762207, "learning_rate": 1.7731169255330668e-05, "loss": 3.2034, "step": 85820 }, { "epoch": 0.0008448, "grad_norm": 0.6772618293762207, "learning_rate": 1.7730658471499086e-05, "loss": 3.0846, "step": 85830 }, { "epoch": 0.0008704, "grad_norm": 0.6970417499542236, "learning_rate": 1.7730147637536238e-05, "loss": 3.0473, "step": 85840 }, { "epoch": 0.000896, "grad_norm": 0.975458562374115, "learning_rate": 1.772963675344544e-05, "loss": 3.3051, "step": 85850 }, { "epoch": 0.0009216, "grad_norm": 0.7514699101448059, "learning_rate": 1.7729125819230003e-05, "loss": 3.079, "step": 85860 }, { "epoch": 0.0009472, "grad_norm": 0.6777455806732178, "learning_rate": 1.7728614834893237e-05, "loss": 2.9322, "step": 85870 }, { "epoch": 0.0009728, "grad_norm": 0.7242442965507507, "learning_rate": 1.7728103800438456e-05, "loss": 3.1162, "step": 85880 }, { "epoch": 0.0009984, "grad_norm": 0.6677154302597046, "learning_rate": 1.7727592715868978e-05, "loss": 3.0885, "step": 85890 }, { "epoch": 0.001024, "grad_norm": 0.6764902472496033, "learning_rate": 1.7727081581188116e-05, "loss": 2.8722, "step": 85900 }, { "epoch": 0.0010496, "grad_norm": 0.7944416403770447, "learning_rate": 1.7726570396399183e-05, "loss": 3.0606, "step": 85910 }, { "epoch": 0.0010752, "grad_norm": 0.828618049621582, "learning_rate": 1.7726059161505494e-05, "loss": 2.8497, "step": 85920 }, { "epoch": 0.0011008, "grad_norm": 0.9540845155715942, "learning_rate": 1.7725547876510365e-05, "loss": 3.1261, "step": 85930 }, { "epoch": 0.0011264, "grad_norm": 0.7214516997337341, "learning_rate": 1.772503654141711e-05, "loss": 3.1638, "step": 85940 }, { "epoch": 0.001152, "grad_norm": 0.7697851061820984, "learning_rate": 1.7724525156229045e-05, "loss": 3.1774, "step": 85950 }, { "epoch": 0.0011776, "grad_norm": 0.7370977997779846, "learning_rate": 1.772401372094949e-05, "loss": 3.0927, "step": 85960 }, { "epoch": 0.0012032, "grad_norm": 0.794946014881134, "learning_rate": 1.772350223558176e-05, "loss": 2.8901, "step": 85970 }, { "epoch": 0.0012288, "grad_norm": 0.7401305437088013, "learning_rate": 1.772299070012917e-05, "loss": 2.7589, "step": 85980 }, { "epoch": 0.0012544, "grad_norm": 0.8333523869514465, "learning_rate": 1.772247911459503e-05, "loss": 2.959, "step": 85990 }, { "epoch": 0.00128, "grad_norm": 0.7131878137588501, "learning_rate": 1.7721967478982675e-05, "loss": 2.9227, "step": 86000 }, { "epoch": 2.56e-05, "grad_norm": 0.7697532773017883, "learning_rate": 1.7721455793295412e-05, "loss": 2.7353, "step": 86010 }, { "epoch": 5.12e-05, "grad_norm": 0.8083232641220093, "learning_rate": 1.7720944057536558e-05, "loss": 3.0759, "step": 86020 }, { "epoch": 7.68e-05, "grad_norm": 0.9355031847953796, "learning_rate": 1.772043227170943e-05, "loss": 2.9258, "step": 86030 }, { "epoch": 0.0001024, "grad_norm": 0.8117826581001282, "learning_rate": 1.7719920435817358e-05, "loss": 3.0928, "step": 86040 }, { "epoch": 0.000128, "grad_norm": 0.8258039355278015, "learning_rate": 1.771940854986365e-05, "loss": 2.9313, "step": 86050 }, { "epoch": 0.0001536, "grad_norm": 0.8651719689369202, "learning_rate": 1.7718896613851627e-05, "loss": 3.1159, "step": 86060 }, { "epoch": 0.0001792, "grad_norm": 0.7016631960868835, "learning_rate": 1.7718384627784614e-05, "loss": 3.0752, "step": 86070 }, { "epoch": 0.0002048, "grad_norm": 0.7115366458892822, "learning_rate": 1.771787259166593e-05, "loss": 3.0235, "step": 86080 }, { "epoch": 0.0002304, "grad_norm": 0.705021321773529, "learning_rate": 1.771736050549889e-05, "loss": 2.7655, "step": 86090 }, { "epoch": 0.000256, "grad_norm": 1.9269770383834839, "learning_rate": 1.7716848369286816e-05, "loss": 3.0848, "step": 86100 }, { "epoch": 0.0002816, "grad_norm": 1.270746111869812, "learning_rate": 1.7716336183033032e-05, "loss": 3.0694, "step": 86110 }, { "epoch": 0.0003072, "grad_norm": 0.7892929911613464, "learning_rate": 1.7715823946740865e-05, "loss": 3.0834, "step": 86120 }, { "epoch": 0.0003328, "grad_norm": 0.769841730594635, "learning_rate": 1.7715311660413625e-05, "loss": 2.9271, "step": 86130 }, { "epoch": 0.0003584, "grad_norm": 0.7683727145195007, "learning_rate": 1.771479932405464e-05, "loss": 2.9993, "step": 86140 }, { "epoch": 0.000384, "grad_norm": 0.7043405771255493, "learning_rate": 1.7714286937667234e-05, "loss": 2.9039, "step": 86150 }, { "epoch": 0.0004096, "grad_norm": 0.6873126029968262, "learning_rate": 1.7713774501254723e-05, "loss": 2.8345, "step": 86160 }, { "epoch": 0.0004352, "grad_norm": 0.9329010248184204, "learning_rate": 1.771326201482044e-05, "loss": 2.9738, "step": 86170 }, { "epoch": 0.0004608, "grad_norm": 0.73539137840271, "learning_rate": 1.7712749478367696e-05, "loss": 3.0232, "step": 86180 }, { "epoch": 0.0004864, "grad_norm": 0.6592492461204529, "learning_rate": 1.7712236891899826e-05, "loss": 3.0792, "step": 86190 }, { "epoch": 0.000512, "grad_norm": 0.9024415612220764, "learning_rate": 1.771172425542015e-05, "loss": 3.2201, "step": 86200 }, { "epoch": 2.56e-05, "grad_norm": 0.7434532046318054, "learning_rate": 1.771121156893199e-05, "loss": 2.5331, "step": 86210 }, { "epoch": 5.12e-05, "grad_norm": 0.8827967643737793, "learning_rate": 1.7710698832438672e-05, "loss": 3.1944, "step": 86220 }, { "epoch": 7.68e-05, "grad_norm": 0.78532475233078, "learning_rate": 1.771018604594352e-05, "loss": 3.2502, "step": 86230 }, { "epoch": 0.0001024, "grad_norm": 0.7043386101722717, "learning_rate": 1.7709673209449866e-05, "loss": 3.1886, "step": 86240 }, { "epoch": 0.000128, "grad_norm": 0.7061363458633423, "learning_rate": 1.7709160322961025e-05, "loss": 3.2014, "step": 86250 }, { "epoch": 0.0001536, "grad_norm": 0.98736971616745, "learning_rate": 1.770864738648033e-05, "loss": 3.0241, "step": 86260 }, { "epoch": 0.0001792, "grad_norm": 0.7582751512527466, "learning_rate": 1.7708134400011106e-05, "loss": 2.8814, "step": 86270 }, { "epoch": 0.0002048, "grad_norm": 1.329592227935791, "learning_rate": 1.770762136355668e-05, "loss": 2.9319, "step": 86280 }, { "epoch": 0.0002304, "grad_norm": 0.6734366416931152, "learning_rate": 1.7707108277120375e-05, "loss": 3.1856, "step": 86290 }, { "epoch": 0.000256, "grad_norm": 0.6978035569190979, "learning_rate": 1.7706595140705522e-05, "loss": 3.1593, "step": 86300 }, { "epoch": 0.0002816, "grad_norm": 0.6689807772636414, "learning_rate": 1.7706081954315448e-05, "loss": 3.0815, "step": 86310 }, { "epoch": 0.0003072, "grad_norm": 0.7019005417823792, "learning_rate": 1.7705568717953482e-05, "loss": 3.0081, "step": 86320 }, { "epoch": 0.0003328, "grad_norm": 0.7770692706108093, "learning_rate": 1.770505543162295e-05, "loss": 3.1489, "step": 86330 }, { "epoch": 0.0003584, "grad_norm": 0.7650564908981323, "learning_rate": 1.770454209532718e-05, "loss": 3.0576, "step": 86340 }, { "epoch": 0.000384, "grad_norm": 0.874925434589386, "learning_rate": 1.7704028709069503e-05, "loss": 3.2695, "step": 86350 }, { "epoch": 0.0004096, "grad_norm": 0.7756960391998291, "learning_rate": 1.7703515272853248e-05, "loss": 2.988, "step": 86360 }, { "epoch": 0.0004352, "grad_norm": 0.7877277135848999, "learning_rate": 1.770300178668174e-05, "loss": 2.9224, "step": 86370 }, { "epoch": 0.0004608, "grad_norm": 0.679591178894043, "learning_rate": 1.7702488250558318e-05, "loss": 3.0703, "step": 86380 }, { "epoch": 0.0004864, "grad_norm": 0.6765504479408264, "learning_rate": 1.7701974664486305e-05, "loss": 2.7876, "step": 86390 }, { "epoch": 0.000512, "grad_norm": 0.7454819679260254, "learning_rate": 1.770146102846903e-05, "loss": 3.2539, "step": 86400 }, { "epoch": 0.0005376, "grad_norm": 0.6764469146728516, "learning_rate": 1.770094734250983e-05, "loss": 3.1067, "step": 86410 }, { "epoch": 0.0005632, "grad_norm": 0.8262908458709717, "learning_rate": 1.770043360661203e-05, "loss": 2.9455, "step": 86420 }, { "epoch": 0.0005888, "grad_norm": 0.7286064624786377, "learning_rate": 1.769991982077897e-05, "loss": 3.187, "step": 86430 }, { "epoch": 0.0006144, "grad_norm": 0.7420425415039062, "learning_rate": 1.7699405985013976e-05, "loss": 3.0315, "step": 86440 }, { "epoch": 0.00064, "grad_norm": 0.6964461207389832, "learning_rate": 1.769889209932038e-05, "loss": 3.0695, "step": 86450 }, { "epoch": 0.0006656, "grad_norm": 0.721649706363678, "learning_rate": 1.769837816370151e-05, "loss": 3.186, "step": 86460 }, { "epoch": 0.0006912, "grad_norm": 0.6793991923332214, "learning_rate": 1.7697864178160706e-05, "loss": 3.2686, "step": 86470 }, { "epoch": 0.0007168, "grad_norm": 0.7635166049003601, "learning_rate": 1.76973501427013e-05, "loss": 2.9932, "step": 86480 }, { "epoch": 0.0007424, "grad_norm": 0.7141431570053101, "learning_rate": 1.7696836057326622e-05, "loss": 3.1916, "step": 86490 }, { "epoch": 0.000768, "grad_norm": 0.730729341506958, "learning_rate": 1.7696321922040008e-05, "loss": 3.0829, "step": 86500 }, { "epoch": 0.0007936, "grad_norm": 0.8929576277732849, "learning_rate": 1.7695807736844796e-05, "loss": 3.0254, "step": 86510 }, { "epoch": 0.0008192, "grad_norm": 0.9702097773551941, "learning_rate": 1.7695293501744314e-05, "loss": 3.02, "step": 86520 }, { "epoch": 0.0008448, "grad_norm": 0.7745619416236877, "learning_rate": 1.7694779216741895e-05, "loss": 3.1769, "step": 86530 }, { "epoch": 0.0008704, "grad_norm": 0.7130779027938843, "learning_rate": 1.769426488184088e-05, "loss": 3.2275, "step": 86540 }, { "epoch": 0.000896, "grad_norm": 0.6692334413528442, "learning_rate": 1.7693750497044605e-05, "loss": 2.9746, "step": 86550 }, { "epoch": 0.0009216, "grad_norm": 1.9841307401657104, "learning_rate": 1.76932360623564e-05, "loss": 3.2847, "step": 86560 }, { "epoch": 0.0009472, "grad_norm": 0.6820372939109802, "learning_rate": 1.7692721577779603e-05, "loss": 3.1234, "step": 86570 }, { "epoch": 0.0009728, "grad_norm": 0.7569597959518433, "learning_rate": 1.7692207043317552e-05, "loss": 3.3564, "step": 86580 }, { "epoch": 0.0009984, "grad_norm": 0.6590626239776611, "learning_rate": 1.7691692458973585e-05, "loss": 3.1027, "step": 86590 }, { "epoch": 0.001024, "grad_norm": 1.7570244073867798, "learning_rate": 1.7691177824751032e-05, "loss": 3.1679, "step": 86600 }, { "epoch": 0.0010496, "grad_norm": 0.9997776746749878, "learning_rate": 1.7690663140653238e-05, "loss": 3.1267, "step": 86610 }, { "epoch": 0.0010752, "grad_norm": 0.8539971113204956, "learning_rate": 1.7690148406683537e-05, "loss": 2.8908, "step": 86620 }, { "epoch": 0.0011008, "grad_norm": 0.7367894649505615, "learning_rate": 1.768963362284527e-05, "loss": 2.9851, "step": 86630 }, { "epoch": 0.0011264, "grad_norm": 0.7162945866584778, "learning_rate": 1.7689118789141768e-05, "loss": 3.1577, "step": 86640 }, { "epoch": 0.001152, "grad_norm": 0.6382483243942261, "learning_rate": 1.7688603905576375e-05, "loss": 3.0633, "step": 86650 }, { "epoch": 0.0011776, "grad_norm": 0.8340044617652893, "learning_rate": 1.768808897215243e-05, "loss": 3.1285, "step": 86660 }, { "epoch": 0.0012032, "grad_norm": 0.7888493537902832, "learning_rate": 1.7687573988873272e-05, "loss": 3.0764, "step": 86670 }, { "epoch": 0.0012288, "grad_norm": 0.6707699298858643, "learning_rate": 1.768705895574224e-05, "loss": 2.9856, "step": 86680 }, { "epoch": 0.0012544, "grad_norm": 0.7802980542182922, "learning_rate": 1.7686543872762673e-05, "loss": 2.9618, "step": 86690 }, { "epoch": 0.00128, "grad_norm": 0.6707441806793213, "learning_rate": 1.768602873993791e-05, "loss": 2.9607, "step": 86700 }, { "epoch": 2.56e-05, "grad_norm": 0.7444815635681152, "learning_rate": 1.7685513557271298e-05, "loss": 3.0765, "step": 86710 }, { "epoch": 5.12e-05, "grad_norm": 0.9139389991760254, "learning_rate": 1.768499832476617e-05, "loss": 3.1008, "step": 86720 }, { "epoch": 7.68e-05, "grad_norm": 0.6718953847885132, "learning_rate": 1.7684483042425868e-05, "loss": 2.9737, "step": 86730 }, { "epoch": 0.0001024, "grad_norm": 0.7320129871368408, "learning_rate": 1.768396771025374e-05, "loss": 3.1753, "step": 86740 }, { "epoch": 0.000128, "grad_norm": 0.8791325688362122, "learning_rate": 1.768345232825312e-05, "loss": 3.1246, "step": 86750 }, { "epoch": 0.0001536, "grad_norm": 1.018784523010254, "learning_rate": 1.7682936896427353e-05, "loss": 3.0081, "step": 86760 }, { "epoch": 0.0001792, "grad_norm": 0.8207268714904785, "learning_rate": 1.7682421414779786e-05, "loss": 3.091, "step": 86770 }, { "epoch": 0.0002048, "grad_norm": 0.7312043309211731, "learning_rate": 1.7681905883313758e-05, "loss": 2.7599, "step": 86780 }, { "epoch": 0.0002304, "grad_norm": 1.0003783702850342, "learning_rate": 1.7681390302032612e-05, "loss": 3.0331, "step": 86790 }, { "epoch": 0.000256, "grad_norm": 0.7418711185455322, "learning_rate": 1.7680874670939688e-05, "loss": 3.0909, "step": 86800 }, { "epoch": 0.0002816, "grad_norm": 0.7960107922554016, "learning_rate": 1.7680358990038336e-05, "loss": 2.9977, "step": 86810 }, { "epoch": 0.0003072, "grad_norm": 0.8185462355613708, "learning_rate": 1.7679843259331894e-05, "loss": 3.0643, "step": 86820 }, { "epoch": 0.0003328, "grad_norm": 1.0016591548919678, "learning_rate": 1.7679327478823713e-05, "loss": 3.2214, "step": 86830 }, { "epoch": 0.0003584, "grad_norm": 0.7273074388504028, "learning_rate": 1.767881164851713e-05, "loss": 3.3193, "step": 86840 }, { "epoch": 0.000384, "grad_norm": 0.7161663770675659, "learning_rate": 1.7678295768415496e-05, "loss": 3.006, "step": 86850 }, { "epoch": 0.0004096, "grad_norm": 0.8086885809898376, "learning_rate": 1.7677779838522158e-05, "loss": 3.0848, "step": 86860 }, { "epoch": 0.0004352, "grad_norm": 0.8213224411010742, "learning_rate": 1.7677263858840454e-05, "loss": 3.3839, "step": 86870 }, { "epoch": 0.0004608, "grad_norm": 0.9905738830566406, "learning_rate": 1.767674782937374e-05, "loss": 3.0895, "step": 86880 }, { "epoch": 0.0004864, "grad_norm": 0.7696829438209534, "learning_rate": 1.767623175012535e-05, "loss": 3.035, "step": 86890 }, { "epoch": 0.000512, "grad_norm": 1.112295389175415, "learning_rate": 1.767571562109864e-05, "loss": 3.0549, "step": 86900 }, { "epoch": 0.0005376, "grad_norm": 0.6580521464347839, "learning_rate": 1.7675199442296953e-05, "loss": 2.9933, "step": 86910 }, { "epoch": 0.0005632, "grad_norm": 0.8415687680244446, "learning_rate": 1.7674683213723637e-05, "loss": 2.945, "step": 86920 }, { "epoch": 0.0005888, "grad_norm": 0.6536645889282227, "learning_rate": 1.7674166935382042e-05, "loss": 2.7294, "step": 86930 }, { "epoch": 0.0006144, "grad_norm": 0.7046947479248047, "learning_rate": 1.7673650607275514e-05, "loss": 2.8596, "step": 86940 }, { "epoch": 0.00064, "grad_norm": 0.7614631652832031, "learning_rate": 1.76731342294074e-05, "loss": 3.2272, "step": 86950 }, { "epoch": 0.0006656, "grad_norm": 0.687192976474762, "learning_rate": 1.767261780178105e-05, "loss": 3.0671, "step": 86960 }, { "epoch": 0.0006912, "grad_norm": 0.7348160743713379, "learning_rate": 1.767210132439981e-05, "loss": 3.2704, "step": 86970 }, { "epoch": 0.0007168, "grad_norm": 0.6841763854026794, "learning_rate": 1.767158479726704e-05, "loss": 3.0114, "step": 86980 }, { "epoch": 0.0007424, "grad_norm": 0.7673133015632629, "learning_rate": 1.7671068220386077e-05, "loss": 3.0167, "step": 86990 }, { "epoch": 0.000768, "grad_norm": 0.7408788204193115, "learning_rate": 1.7670551593760273e-05, "loss": 3.0785, "step": 87000 }, { "epoch": 0.0007936, "grad_norm": 2.162400245666504, "learning_rate": 1.767003491739298e-05, "loss": 2.9211, "step": 87010 }, { "epoch": 0.0008192, "grad_norm": 1.201883316040039, "learning_rate": 1.7669518191287555e-05, "loss": 3.0835, "step": 87020 }, { "epoch": 0.0008448, "grad_norm": 0.7308161854743958, "learning_rate": 1.766900141544734e-05, "loss": 2.7369, "step": 87030 }, { "epoch": 0.0008704, "grad_norm": 0.9128721356391907, "learning_rate": 1.7668484589875684e-05, "loss": 3.0835, "step": 87040 }, { "epoch": 0.000896, "grad_norm": 0.7110142707824707, "learning_rate": 1.766796771457595e-05, "loss": 2.9924, "step": 87050 }, { "epoch": 0.0009216, "grad_norm": 0.7250414490699768, "learning_rate": 1.7667450789551482e-05, "loss": 2.9612, "step": 87060 }, { "epoch": 0.0009472, "grad_norm": 0.6983067989349365, "learning_rate": 1.7666933814805633e-05, "loss": 3.2553, "step": 87070 }, { "epoch": 0.0009728, "grad_norm": 0.6539955735206604, "learning_rate": 1.766641679034175e-05, "loss": 3.0552, "step": 87080 }, { "epoch": 0.0009984, "grad_norm": 0.8998467326164246, "learning_rate": 1.76658997161632e-05, "loss": 2.9389, "step": 87090 }, { "epoch": 0.001024, "grad_norm": 0.7736408114433289, "learning_rate": 1.7665382592273324e-05, "loss": 3.1542, "step": 87100 }, { "epoch": 0.0010496, "grad_norm": 0.8101139664649963, "learning_rate": 1.766486541867548e-05, "loss": 3.0333, "step": 87110 }, { "epoch": 0.0010752, "grad_norm": 0.7541834115982056, "learning_rate": 1.766434819537302e-05, "loss": 3.0849, "step": 87120 }, { "epoch": 0.0011008, "grad_norm": 0.6507870554924011, "learning_rate": 1.76638309223693e-05, "loss": 2.8989, "step": 87130 }, { "epoch": 0.0011264, "grad_norm": 1.0218197107315063, "learning_rate": 1.766331359966767e-05, "loss": 3.1692, "step": 87140 }, { "epoch": 0.001152, "grad_norm": 0.8603965044021606, "learning_rate": 1.766279622727149e-05, "loss": 2.8724, "step": 87150 }, { "epoch": 0.0011776, "grad_norm": 0.7346203923225403, "learning_rate": 1.7662278805184114e-05, "loss": 3.1584, "step": 87160 }, { "epoch": 0.0012032, "grad_norm": 0.9028226137161255, "learning_rate": 1.7661761333408896e-05, "loss": 3.1863, "step": 87170 }, { "epoch": 0.0012288, "grad_norm": 0.9100760221481323, "learning_rate": 1.7661243811949192e-05, "loss": 3.3942, "step": 87180 }, { "epoch": 0.0012544, "grad_norm": 0.7429245114326477, "learning_rate": 1.766072624080836e-05, "loss": 3.0565, "step": 87190 }, { "epoch": 0.00128, "grad_norm": 0.7897778153419495, "learning_rate": 1.7660208619989753e-05, "loss": 3.1733, "step": 87200 }, { "epoch": 0.0013056, "grad_norm": 0.873781144618988, "learning_rate": 1.7659690949496726e-05, "loss": 3.0857, "step": 87210 }, { "epoch": 0.0013312, "grad_norm": 1.2768720388412476, "learning_rate": 1.7659173229332644e-05, "loss": 3.2621, "step": 87220 }, { "epoch": 0.0013568, "grad_norm": 0.7548202872276306, "learning_rate": 1.7658655459500853e-05, "loss": 3.1913, "step": 87230 }, { "epoch": 0.0013824, "grad_norm": 0.6900081038475037, "learning_rate": 1.765813764000472e-05, "loss": 3.1502, "step": 87240 }, { "epoch": 0.001408, "grad_norm": 0.824451208114624, "learning_rate": 1.76576197708476e-05, "loss": 2.9712, "step": 87250 }, { "epoch": 0.0014336, "grad_norm": 0.7047469019889832, "learning_rate": 1.765710185203285e-05, "loss": 2.9142, "step": 87260 }, { "epoch": 0.0014592, "grad_norm": 0.7962784767150879, "learning_rate": 1.7656583883563828e-05, "loss": 3.0949, "step": 87270 }, { "epoch": 0.0014848, "grad_norm": 0.6909545063972473, "learning_rate": 1.76560658654439e-05, "loss": 2.8947, "step": 87280 }, { "epoch": 0.0015104, "grad_norm": 0.7538129687309265, "learning_rate": 1.7655547797676415e-05, "loss": 2.6113, "step": 87290 }, { "epoch": 0.001536, "grad_norm": 1.2368335723876953, "learning_rate": 1.765502968026474e-05, "loss": 3.2475, "step": 87300 }, { "epoch": 0.0015616, "grad_norm": 0.8122758269309998, "learning_rate": 1.765451151321223e-05, "loss": 3.0883, "step": 87310 }, { "epoch": 0.0015872, "grad_norm": 0.789008378982544, "learning_rate": 1.7653993296522245e-05, "loss": 3.2002, "step": 87320 }, { "epoch": 0.0016128, "grad_norm": 0.7885162830352783, "learning_rate": 1.7653475030198152e-05, "loss": 2.9527, "step": 87330 }, { "epoch": 0.0016384, "grad_norm": 0.8078469634056091, "learning_rate": 1.7652956714243302e-05, "loss": 3.1493, "step": 87340 }, { "epoch": 0.001664, "grad_norm": 0.7423650026321411, "learning_rate": 1.7652438348661066e-05, "loss": 3.1142, "step": 87350 }, { "epoch": 0.0016896, "grad_norm": 0.7974101901054382, "learning_rate": 1.76519199334548e-05, "loss": 3.0895, "step": 87360 }, { "epoch": 0.0017152, "grad_norm": 2.4055988788604736, "learning_rate": 1.7651401468627867e-05, "loss": 3.2187, "step": 87370 }, { "epoch": 0.0017408, "grad_norm": 0.8386815786361694, "learning_rate": 1.765088295418363e-05, "loss": 3.0278, "step": 87380 }, { "epoch": 0.0017664, "grad_norm": 0.9239630699157715, "learning_rate": 1.765036439012545e-05, "loss": 3.2361, "step": 87390 }, { "epoch": 0.001792, "grad_norm": 0.8234713673591614, "learning_rate": 1.764984577645669e-05, "loss": 3.0026, "step": 87400 }, { "epoch": 0.0018176, "grad_norm": 0.8334180116653442, "learning_rate": 1.7649327113180713e-05, "loss": 3.1486, "step": 87410 }, { "epoch": 0.0018432, "grad_norm": 0.7999060750007629, "learning_rate": 1.764880840030088e-05, "loss": 2.8991, "step": 87420 }, { "epoch": 0.0018688, "grad_norm": 0.6733655333518982, "learning_rate": 1.764828963782056e-05, "loss": 3.1134, "step": 87430 }, { "epoch": 0.0018944, "grad_norm": 0.6877169609069824, "learning_rate": 1.7647770825743115e-05, "loss": 2.8257, "step": 87440 }, { "epoch": 0.00192, "grad_norm": 0.9742264747619629, "learning_rate": 1.7647251964071907e-05, "loss": 3.1797, "step": 87450 }, { "epoch": 0.0019456, "grad_norm": 0.804026186466217, "learning_rate": 1.7646733052810304e-05, "loss": 3.1544, "step": 87460 }, { "epoch": 0.0019712, "grad_norm": 0.761442244052887, "learning_rate": 1.7646214091961668e-05, "loss": 2.8092, "step": 87470 }, { "epoch": 0.0019968, "grad_norm": 0.7190984487533569, "learning_rate": 1.7645695081529367e-05, "loss": 2.8728, "step": 87480 }, { "epoch": 0.0020224, "grad_norm": 0.7786910533905029, "learning_rate": 1.7645176021516764e-05, "loss": 3.1114, "step": 87490 }, { "epoch": 0.002048, "grad_norm": 0.7299113273620605, "learning_rate": 1.7644656911927228e-05, "loss": 3.1118, "step": 87500 }, { "epoch": 0.0020736, "grad_norm": 0.8520238399505615, "learning_rate": 1.7644137752764124e-05, "loss": 2.8366, "step": 87510 }, { "epoch": 0.0020992, "grad_norm": 0.9170851111412048, "learning_rate": 1.7643618544030815e-05, "loss": 3.001, "step": 87520 }, { "epoch": 0.0021248, "grad_norm": 0.7784470319747925, "learning_rate": 1.7643099285730672e-05, "loss": 3.1332, "step": 87530 }, { "epoch": 0.0021504, "grad_norm": 0.7679725289344788, "learning_rate": 1.764257997786706e-05, "loss": 3.218, "step": 87540 }, { "epoch": 0.002176, "grad_norm": 0.7508382201194763, "learning_rate": 1.7642060620443352e-05, "loss": 3.0636, "step": 87550 }, { "epoch": 0.0022016, "grad_norm": 0.7049858570098877, "learning_rate": 1.7641541213462914e-05, "loss": 3.1301, "step": 87560 }, { "epoch": 0.0022272, "grad_norm": 0.9706542491912842, "learning_rate": 1.764102175692911e-05, "loss": 3.1143, "step": 87570 }, { "epoch": 0.0022528, "grad_norm": 0.8636168241500854, "learning_rate": 1.7640502250845308e-05, "loss": 3.0802, "step": 87580 }, { "epoch": 0.0022784, "grad_norm": 0.7531412243843079, "learning_rate": 1.763998269521488e-05, "loss": 3.3892, "step": 87590 }, { "epoch": 0.002304, "grad_norm": 0.7580934166908264, "learning_rate": 1.7639463090041196e-05, "loss": 3.1534, "step": 87600 }, { "epoch": 0.0023296, "grad_norm": 0.7523968815803528, "learning_rate": 1.7638943435327622e-05, "loss": 3.2656, "step": 87610 }, { "epoch": 0.0023552, "grad_norm": 0.9443408250808716, "learning_rate": 1.763842373107753e-05, "loss": 3.252, "step": 87620 }, { "epoch": 0.0023808, "grad_norm": 0.7492494583129883, "learning_rate": 1.7637903977294293e-05, "loss": 2.9745, "step": 87630 }, { "epoch": 0.0024064, "grad_norm": 0.705418050289154, "learning_rate": 1.7637384173981277e-05, "loss": 3.1543, "step": 87640 }, { "epoch": 0.002432, "grad_norm": 0.7562901973724365, "learning_rate": 1.7636864321141854e-05, "loss": 3.0407, "step": 87650 }, { "epoch": 0.0024576, "grad_norm": 0.786775529384613, "learning_rate": 1.7636344418779397e-05, "loss": 3.0538, "step": 87660 }, { "epoch": 0.0024832, "grad_norm": 0.6653182506561279, "learning_rate": 1.7635824466897277e-05, "loss": 3.2135, "step": 87670 }, { "epoch": 0.0025088, "grad_norm": 0.9013116955757141, "learning_rate": 1.763530446549886e-05, "loss": 3.314, "step": 87680 }, { "epoch": 0.0025344, "grad_norm": 1.0372809171676636, "learning_rate": 1.7634784414587525e-05, "loss": 3.312, "step": 87690 }, { "epoch": 0.00256, "grad_norm": 0.6589071750640869, "learning_rate": 1.7634264314166644e-05, "loss": 3.0661, "step": 87700 }, { "epoch": 0.0025856, "grad_norm": 0.8051645159721375, "learning_rate": 1.7633744164239587e-05, "loss": 3.3287, "step": 87710 }, { "epoch": 0.0026112, "grad_norm": 0.733892023563385, "learning_rate": 1.7633223964809726e-05, "loss": 3.2034, "step": 87720 }, { "epoch": 0.0026368, "grad_norm": 0.8518341183662415, "learning_rate": 1.763270371588044e-05, "loss": 3.1504, "step": 87730 }, { "epoch": 0.0026624, "grad_norm": 0.8664030432701111, "learning_rate": 1.7632183417455096e-05, "loss": 3.1664, "step": 87740 }, { "epoch": 0.002688, "grad_norm": 0.7465702295303345, "learning_rate": 1.7631663069537073e-05, "loss": 3.1125, "step": 87750 }, { "epoch": 0.0027136, "grad_norm": 0.7040249705314636, "learning_rate": 1.763114267212974e-05, "loss": 3.2616, "step": 87760 }, { "epoch": 0.0027392, "grad_norm": 0.7268482446670532, "learning_rate": 1.7630622225236476e-05, "loss": 3.2621, "step": 87770 }, { "epoch": 0.0027648, "grad_norm": 0.6741511821746826, "learning_rate": 1.763010172886066e-05, "loss": 2.9948, "step": 87780 }, { "epoch": 0.0027904, "grad_norm": 0.772881269454956, "learning_rate": 1.762958118300566e-05, "loss": 3.1514, "step": 87790 }, { "epoch": 0.002816, "grad_norm": 1.1140780448913574, "learning_rate": 1.7629060587674848e-05, "loss": 3.2314, "step": 87800 }, { "epoch": 0.0028416, "grad_norm": 0.7167626619338989, "learning_rate": 1.762853994287161e-05, "loss": 3.0655, "step": 87810 }, { "epoch": 0.0028672, "grad_norm": 0.7053471207618713, "learning_rate": 1.7628019248599315e-05, "loss": 3.1552, "step": 87820 }, { "epoch": 0.0028928, "grad_norm": 0.8590969443321228, "learning_rate": 1.7627498504861345e-05, "loss": 3.1947, "step": 87830 }, { "epoch": 0.0029184, "grad_norm": 0.7966821193695068, "learning_rate": 1.7626977711661073e-05, "loss": 3.2029, "step": 87840 }, { "epoch": 0.002944, "grad_norm": 0.7687954306602478, "learning_rate": 1.762645686900188e-05, "loss": 3.2385, "step": 87850 }, { "epoch": 0.0029696, "grad_norm": 0.7787836790084839, "learning_rate": 1.762593597688714e-05, "loss": 3.0012, "step": 87860 }, { "epoch": 0.0029952, "grad_norm": 0.8887073397636414, "learning_rate": 1.762541503532023e-05, "loss": 3.2563, "step": 87870 }, { "epoch": 0.0030208, "grad_norm": 0.7390779852867126, "learning_rate": 1.762489404430453e-05, "loss": 3.0684, "step": 87880 }, { "epoch": 0.0030464, "grad_norm": 0.7191071510314941, "learning_rate": 1.7624373003843422e-05, "loss": 3.3207, "step": 87890 }, { "epoch": 0.003072, "grad_norm": 0.8549396395683289, "learning_rate": 1.7623851913940278e-05, "loss": 3.3067, "step": 87900 }, { "epoch": 0.0030976, "grad_norm": 0.8436385989189148, "learning_rate": 1.7623330774598484e-05, "loss": 3.0218, "step": 87910 }, { "epoch": 0.0031232, "grad_norm": 1.1290194988250732, "learning_rate": 1.7622809585821415e-05, "loss": 3.1688, "step": 87920 }, { "epoch": 0.0031488, "grad_norm": 0.7759792804718018, "learning_rate": 1.7622288347612448e-05, "loss": 3.0911, "step": 87930 }, { "epoch": 0.0031744, "grad_norm": 0.7874084115028381, "learning_rate": 1.7621767059974972e-05, "loss": 3.3388, "step": 87940 }, { "epoch": 0.0032, "grad_norm": 0.6991609334945679, "learning_rate": 1.7621245722912357e-05, "loss": 3.3061, "step": 87950 }, { "epoch": 0.0032256, "grad_norm": 0.8614990711212158, "learning_rate": 1.7620724336427993e-05, "loss": 3.1971, "step": 87960 }, { "epoch": 0.0032512, "grad_norm": 0.7234736084938049, "learning_rate": 1.7620202900525256e-05, "loss": 3.3009, "step": 87970 }, { "epoch": 0.0032768, "grad_norm": 0.6765320301055908, "learning_rate": 1.761968141520753e-05, "loss": 3.089, "step": 87980 }, { "epoch": 0.0033024, "grad_norm": 0.8131476044654846, "learning_rate": 1.7619159880478192e-05, "loss": 3.0617, "step": 87990 }, { "epoch": 0.003328, "grad_norm": 0.650672972202301, "learning_rate": 1.7618638296340628e-05, "loss": 3.0857, "step": 88000 }, { "epoch": 0.0033536, "grad_norm": 0.7758305668830872, "learning_rate": 1.761811666279822e-05, "loss": 3.2737, "step": 88010 }, { "epoch": 0.0033792, "grad_norm": 1.0926427841186523, "learning_rate": 1.761759497985435e-05, "loss": 3.2413, "step": 88020 }, { "epoch": 0.0034048, "grad_norm": 0.8516579866409302, "learning_rate": 1.7617073247512403e-05, "loss": 3.1415, "step": 88030 }, { "epoch": 0.0034304, "grad_norm": 0.7951352596282959, "learning_rate": 1.7616551465775756e-05, "loss": 3.3889, "step": 88040 }, { "epoch": 0.003456, "grad_norm": 0.768547534942627, "learning_rate": 1.76160296346478e-05, "loss": 3.0168, "step": 88050 }, { "epoch": 0.0034816, "grad_norm": 0.6645597219467163, "learning_rate": 1.7615507754131912e-05, "loss": 3.0412, "step": 88060 }, { "epoch": 0.0035072, "grad_norm": 1.0580507516860962, "learning_rate": 1.7614985824231487e-05, "loss": 3.2412, "step": 88070 }, { "epoch": 0.0035328, "grad_norm": 0.7077932953834534, "learning_rate": 1.7614463844949898e-05, "loss": 3.0591, "step": 88080 }, { "epoch": 0.0035584, "grad_norm": 0.9296275973320007, "learning_rate": 1.7613941816290537e-05, "loss": 3.1292, "step": 88090 }, { "epoch": 0.003584, "grad_norm": 0.6703597903251648, "learning_rate": 1.761341973825678e-05, "loss": 3.1703, "step": 88100 }, { "epoch": 0.0036096, "grad_norm": 1.3995335102081299, "learning_rate": 1.761289761085203e-05, "loss": 3.354, "step": 88110 }, { "epoch": 0.0036352, "grad_norm": 0.7054426074028015, "learning_rate": 1.7612375434079658e-05, "loss": 3.1879, "step": 88120 }, { "epoch": 0.0036608, "grad_norm": 0.8570812940597534, "learning_rate": 1.7611853207943053e-05, "loss": 3.2453, "step": 88130 }, { "epoch": 0.0036864, "grad_norm": 0.7780048847198486, "learning_rate": 1.7611330932445602e-05, "loss": 3.2284, "step": 88140 }, { "epoch": 0.003712, "grad_norm": 0.695452868938446, "learning_rate": 1.7610808607590695e-05, "loss": 3.3091, "step": 88150 }, { "epoch": 0.0037376, "grad_norm": 0.6840187907218933, "learning_rate": 1.7610286233381716e-05, "loss": 3.4657, "step": 88160 }, { "epoch": 0.0037632, "grad_norm": 0.849882185459137, "learning_rate": 1.7609763809822052e-05, "loss": 3.0736, "step": 88170 }, { "epoch": 0.0037888, "grad_norm": 1.215506672859192, "learning_rate": 1.7609241336915096e-05, "loss": 3.2095, "step": 88180 }, { "epoch": 0.0038144, "grad_norm": 0.7102605700492859, "learning_rate": 1.760871881466423e-05, "loss": 3.2019, "step": 88190 }, { "epoch": 0.00384, "grad_norm": 0.70079106092453, "learning_rate": 1.7608196243072843e-05, "loss": 3.4404, "step": 88200 }, { "epoch": 0.0038656, "grad_norm": 0.8182222247123718, "learning_rate": 1.7607673622144328e-05, "loss": 3.0325, "step": 88210 }, { "epoch": 0.0038912, "grad_norm": 0.7183645963668823, "learning_rate": 1.760715095188207e-05, "loss": 3.0803, "step": 88220 }, { "epoch": 0.0039168, "grad_norm": 0.9107880592346191, "learning_rate": 1.760662823228946e-05, "loss": 3.1862, "step": 88230 }, { "epoch": 0.0039424, "grad_norm": 0.7350866198539734, "learning_rate": 1.760610546336989e-05, "loss": 3.1794, "step": 88240 }, { "epoch": 0.003968, "grad_norm": 0.7670534253120422, "learning_rate": 1.7605582645126748e-05, "loss": 3.2031, "step": 88250 }, { "epoch": 0.0039936, "grad_norm": 0.7294168472290039, "learning_rate": 1.7605059777563418e-05, "loss": 3.2059, "step": 88260 }, { "epoch": 0.0040192, "grad_norm": 0.7259302735328674, "learning_rate": 1.76045368606833e-05, "loss": 3.307, "step": 88270 }, { "epoch": 0.0040448, "grad_norm": 1.3169666528701782, "learning_rate": 1.760401389448978e-05, "loss": 3.579, "step": 88280 }, { "epoch": 0.0040704, "grad_norm": 0.7938028573989868, "learning_rate": 1.7603490878986253e-05, "loss": 3.2295, "step": 88290 }, { "epoch": 0.004096, "grad_norm": 0.7019129991531372, "learning_rate": 1.760296781417611e-05, "loss": 2.9521, "step": 88300 }, { "epoch": 0.0041216, "grad_norm": 0.7243034839630127, "learning_rate": 1.760244470006274e-05, "loss": 3.2362, "step": 88310 }, { "epoch": 0.0041472, "grad_norm": 0.7358291745185852, "learning_rate": 1.7601921536649534e-05, "loss": 3.1984, "step": 88320 }, { "epoch": 0.0041728, "grad_norm": 0.6858118772506714, "learning_rate": 1.7601398323939885e-05, "loss": 3.2482, "step": 88330 }, { "epoch": 0.0041984, "grad_norm": 0.7524966597557068, "learning_rate": 1.760087506193719e-05, "loss": 3.1375, "step": 88340 }, { "epoch": 0.004224, "grad_norm": 0.7581201195716858, "learning_rate": 1.7600351750644848e-05, "loss": 3.1147, "step": 88350 }, { "epoch": 0.0042496, "grad_norm": 0.8409121036529541, "learning_rate": 1.7599828390066235e-05, "loss": 3.3059, "step": 88360 }, { "epoch": 0.0042752, "grad_norm": 1.0179693698883057, "learning_rate": 1.759930498020476e-05, "loss": 3.1857, "step": 88370 }, { "epoch": 0.0043008, "grad_norm": 0.915175199508667, "learning_rate": 1.759878152106381e-05, "loss": 3.1056, "step": 88380 }, { "epoch": 0.0043264, "grad_norm": 0.698762834072113, "learning_rate": 1.759825801264678e-05, "loss": 3.0976, "step": 88390 }, { "epoch": 0.004352, "grad_norm": 0.9137099981307983, "learning_rate": 1.759773445495707e-05, "loss": 3.3885, "step": 88400 }, { "epoch": 0.0043776, "grad_norm": 0.7014741897583008, "learning_rate": 1.759721084799807e-05, "loss": 3.2381, "step": 88410 }, { "epoch": 0.0044032, "grad_norm": 0.8695681691169739, "learning_rate": 1.7596687191773174e-05, "loss": 3.3059, "step": 88420 }, { "epoch": 0.0044288, "grad_norm": 0.7447514533996582, "learning_rate": 1.759616348628578e-05, "loss": 3.425, "step": 88430 }, { "epoch": 0.0044544, "grad_norm": 0.8558635711669922, "learning_rate": 1.759563973153929e-05, "loss": 3.1722, "step": 88440 }, { "epoch": 0.00448, "grad_norm": 0.8304480314254761, "learning_rate": 1.759511592753709e-05, "loss": 3.1418, "step": 88450 }, { "epoch": 0.0045056, "grad_norm": 0.6610480546951294, "learning_rate": 1.7594592074282584e-05, "loss": 3.1215, "step": 88460 }, { "epoch": 0.0045312, "grad_norm": 0.8956723809242249, "learning_rate": 1.759406817177917e-05, "loss": 2.9639, "step": 88470 }, { "epoch": 0.0045568, "grad_norm": 0.7527256608009338, "learning_rate": 1.7593544220030236e-05, "loss": 3.3282, "step": 88480 }, { "epoch": 0.0045824, "grad_norm": 0.7314112782478333, "learning_rate": 1.7593020219039187e-05, "loss": 3.3213, "step": 88490 }, { "epoch": 0.004608, "grad_norm": 0.72891765832901, "learning_rate": 1.7592496168809425e-05, "loss": 3.155, "step": 88500 }, { "epoch": 0.0046336, "grad_norm": 1.5641640424728394, "learning_rate": 1.7591972069344338e-05, "loss": 3.2719, "step": 88510 }, { "epoch": 0.0046592, "grad_norm": 1.0603209733963013, "learning_rate": 1.759144792064733e-05, "loss": 3.1051, "step": 88520 }, { "epoch": 0.0046848, "grad_norm": 0.8593615293502808, "learning_rate": 1.7590923722721803e-05, "loss": 3.488, "step": 88530 }, { "epoch": 0.0047104, "grad_norm": 0.8182860016822815, "learning_rate": 1.7590399475571152e-05, "loss": 3.1996, "step": 88540 }, { "epoch": 0.004736, "grad_norm": 0.7317187190055847, "learning_rate": 1.7589875179198776e-05, "loss": 3.0509, "step": 88550 }, { "epoch": 0.0047616, "grad_norm": 0.808373749256134, "learning_rate": 1.758935083360808e-05, "loss": 3.4385, "step": 88560 }, { "epoch": 0.0047872, "grad_norm": 0.7178140878677368, "learning_rate": 1.7588826438802458e-05, "loss": 2.9696, "step": 88570 }, { "epoch": 0.0048128, "grad_norm": 0.8384464979171753, "learning_rate": 1.7588301994785314e-05, "loss": 3.068, "step": 88580 }, { "epoch": 0.0048384, "grad_norm": 0.8469061255455017, "learning_rate": 1.7587777501560048e-05, "loss": 3.1297, "step": 88590 }, { "epoch": 0.004864, "grad_norm": 0.7602402567863464, "learning_rate": 1.7587252959130062e-05, "loss": 3.1573, "step": 88600 }, { "epoch": 0.0048896, "grad_norm": 0.8026084899902344, "learning_rate": 1.7586728367498755e-05, "loss": 3.2329, "step": 88610 }, { "epoch": 0.0049152, "grad_norm": 0.8356468677520752, "learning_rate": 1.7586203726669534e-05, "loss": 3.1992, "step": 88620 }, { "epoch": 0.0049408, "grad_norm": 0.7809629440307617, "learning_rate": 1.7585679036645796e-05, "loss": 3.0925, "step": 88630 }, { "epoch": 0.0049664, "grad_norm": 0.7676171660423279, "learning_rate": 1.7585154297430947e-05, "loss": 2.9839, "step": 88640 }, { "epoch": 0.004992, "grad_norm": 0.7719355821609497, "learning_rate": 1.7584629509028387e-05, "loss": 3.2178, "step": 88650 }, { "epoch": 0.0050176, "grad_norm": 0.9209770560264587, "learning_rate": 1.758410467144152e-05, "loss": 3.2549, "step": 88660 }, { "epoch": 0.0050432, "grad_norm": 0.871971070766449, "learning_rate": 1.758357978467375e-05, "loss": 3.194, "step": 88670 }, { "epoch": 0.0050688, "grad_norm": 0.9049177169799805, "learning_rate": 1.7583054848728482e-05, "loss": 3.119, "step": 88680 }, { "epoch": 0.0050944, "grad_norm": 0.7723633050918579, "learning_rate": 1.758252986360912e-05, "loss": 3.1496, "step": 88690 }, { "epoch": 0.00512, "grad_norm": 0.9181063175201416, "learning_rate": 1.7582004829319062e-05, "loss": 3.2586, "step": 88700 }, { "epoch": 0.0051456, "grad_norm": 0.7588092088699341, "learning_rate": 1.758147974586172e-05, "loss": 3.0895, "step": 88710 }, { "epoch": 0.0051712, "grad_norm": 0.7601069808006287, "learning_rate": 1.7580954613240498e-05, "loss": 3.1057, "step": 88720 }, { "epoch": 0.0051968, "grad_norm": 0.8564867973327637, "learning_rate": 1.7580429431458798e-05, "loss": 3.1216, "step": 88730 }, { "epoch": 0.0052224, "grad_norm": 0.8838126063346863, "learning_rate": 1.757990420052003e-05, "loss": 3.1503, "step": 88740 }, { "epoch": 0.005248, "grad_norm": 0.7211958169937134, "learning_rate": 1.75793789204276e-05, "loss": 3.0505, "step": 88750 }, { "epoch": 0.0052736, "grad_norm": 0.8717153072357178, "learning_rate": 1.7578853591184904e-05, "loss": 3.2303, "step": 88760 }, { "epoch": 0.0052992, "grad_norm": 0.657476007938385, "learning_rate": 1.7578328212795365e-05, "loss": 3.1527, "step": 88770 }, { "epoch": 0.0053248, "grad_norm": 0.8361315727233887, "learning_rate": 1.7577802785262377e-05, "loss": 3.3209, "step": 88780 }, { "epoch": 0.0053504, "grad_norm": 0.7904840707778931, "learning_rate": 1.757727730858935e-05, "loss": 3.254, "step": 88790 }, { "epoch": 0.005376, "grad_norm": 1.077832579612732, "learning_rate": 1.75767517827797e-05, "loss": 3.3278, "step": 88800 }, { "epoch": 0.0054016, "grad_norm": 0.7500284910202026, "learning_rate": 1.7576226207836824e-05, "loss": 3.2669, "step": 88810 }, { "epoch": 0.0054272, "grad_norm": 1.0778820514678955, "learning_rate": 1.7575700583764138e-05, "loss": 3.3434, "step": 88820 }, { "epoch": 0.0054528, "grad_norm": 0.7602941989898682, "learning_rate": 1.7575174910565046e-05, "loss": 3.1668, "step": 88830 }, { "epoch": 0.0054784, "grad_norm": 0.7835508584976196, "learning_rate": 1.7574649188242952e-05, "loss": 2.985, "step": 88840 }, { "epoch": 0.005504, "grad_norm": 0.737718939781189, "learning_rate": 1.757412341680128e-05, "loss": 3.122, "step": 88850 }, { "epoch": 0.0055296, "grad_norm": 0.8896776437759399, "learning_rate": 1.7573597596243424e-05, "loss": 3.1805, "step": 88860 }, { "epoch": 0.0055552, "grad_norm": 0.7483769059181213, "learning_rate": 1.75730717265728e-05, "loss": 3.0951, "step": 88870 }, { "epoch": 0.0055808, "grad_norm": 0.8013837337493896, "learning_rate": 1.7572545807792823e-05, "loss": 3.2963, "step": 88880 }, { "epoch": 0.0056064, "grad_norm": 0.893354058265686, "learning_rate": 1.75720198399069e-05, "loss": 3.1239, "step": 88890 }, { "epoch": 0.005632, "grad_norm": 0.6922040581703186, "learning_rate": 1.7571493822918438e-05, "loss": 3.1814, "step": 88900 }, { "epoch": 0.0056576, "grad_norm": 0.7805924415588379, "learning_rate": 1.7570967756830848e-05, "loss": 3.1101, "step": 88910 }, { "epoch": 0.0056832, "grad_norm": 2.4406471252441406, "learning_rate": 1.757044164164755e-05, "loss": 3.0792, "step": 88920 }, { "epoch": 0.0057088, "grad_norm": 0.747158944606781, "learning_rate": 1.7569915477371947e-05, "loss": 3.0109, "step": 88930 }, { "epoch": 0.0057344, "grad_norm": 0.7859488725662231, "learning_rate": 1.756938926400745e-05, "loss": 3.3108, "step": 88940 }, { "epoch": 0.00576, "grad_norm": 0.701296865940094, "learning_rate": 1.7568863001557483e-05, "loss": 3.1112, "step": 88950 }, { "epoch": 0.0057856, "grad_norm": 1.040105938911438, "learning_rate": 1.7568336690025444e-05, "loss": 3.1454, "step": 88960 }, { "epoch": 0.0058112, "grad_norm": 1.114288330078125, "learning_rate": 1.7567810329414758e-05, "loss": 3.1545, "step": 88970 }, { "epoch": 0.0058368, "grad_norm": 0.6856187582015991, "learning_rate": 1.756728391972883e-05, "loss": 3.2917, "step": 88980 }, { "epoch": 0.0058624, "grad_norm": 0.9632576107978821, "learning_rate": 1.7566757460971076e-05, "loss": 3.1057, "step": 88990 }, { "epoch": 0.005888, "grad_norm": 0.8087217807769775, "learning_rate": 1.756623095314492e-05, "loss": 2.9651, "step": 89000 }, { "epoch": 0.0059136, "grad_norm": 0.9658265113830566, "learning_rate": 1.7565704396253757e-05, "loss": 2.9738, "step": 89010 }, { "epoch": 0.0059392, "grad_norm": 0.7338629364967346, "learning_rate": 1.7565177790301014e-05, "loss": 3.0816, "step": 89020 }, { "epoch": 0.0059648, "grad_norm": 0.9709406495094299, "learning_rate": 1.756465113529011e-05, "loss": 3.1717, "step": 89030 }, { "epoch": 0.0059904, "grad_norm": 0.6604009866714478, "learning_rate": 1.7564124431224442e-05, "loss": 3.1237, "step": 89040 }, { "epoch": 0.006016, "grad_norm": 0.8735620975494385, "learning_rate": 1.7563597678107446e-05, "loss": 3.0506, "step": 89050 }, { "epoch": 0.0060416, "grad_norm": 0.698471188545227, "learning_rate": 1.7563070875942527e-05, "loss": 3.2302, "step": 89060 }, { "epoch": 0.0060672, "grad_norm": 0.751316249370575, "learning_rate": 1.7562544024733105e-05, "loss": 3.1151, "step": 89070 }, { "epoch": 0.0060928, "grad_norm": 0.664348840713501, "learning_rate": 1.756201712448259e-05, "loss": 2.9294, "step": 89080 }, { "epoch": 0.0061184, "grad_norm": 0.8915156722068787, "learning_rate": 1.756149017519441e-05, "loss": 3.1915, "step": 89090 }, { "epoch": 0.006144, "grad_norm": 1.1301357746124268, "learning_rate": 1.756096317687197e-05, "loss": 3.1114, "step": 89100 }, { "epoch": 0.0061696, "grad_norm": 0.7779869437217712, "learning_rate": 1.7560436129518697e-05, "loss": 3.039, "step": 89110 }, { "epoch": 0.0061952, "grad_norm": 0.7350479960441589, "learning_rate": 1.7559909033138005e-05, "loss": 3.2287, "step": 89120 }, { "epoch": 0.0062208, "grad_norm": 0.7603198885917664, "learning_rate": 1.7559381887733312e-05, "loss": 3.2815, "step": 89130 }, { "epoch": 0.0062464, "grad_norm": 0.6795520782470703, "learning_rate": 1.7558854693308038e-05, "loss": 2.977, "step": 89140 }, { "epoch": 0.006272, "grad_norm": 1.443578839302063, "learning_rate": 1.7558327449865597e-05, "loss": 3.2222, "step": 89150 }, { "epoch": 0.0062976, "grad_norm": 0.8738045692443848, "learning_rate": 1.755780015740941e-05, "loss": 3.1749, "step": 89160 }, { "epoch": 0.0063232, "grad_norm": 0.8088482618331909, "learning_rate": 1.75572728159429e-05, "loss": 3.1818, "step": 89170 }, { "epoch": 0.0063488, "grad_norm": 0.7571561336517334, "learning_rate": 1.7556745425469485e-05, "loss": 3.1068, "step": 89180 }, { "epoch": 0.0063744, "grad_norm": 0.7987313270568848, "learning_rate": 1.7556217985992585e-05, "loss": 3.1188, "step": 89190 }, { "epoch": 0.0064, "grad_norm": 0.7411290407180786, "learning_rate": 1.7555690497515618e-05, "loss": 3.2806, "step": 89200 }, { "epoch": 2.56e-05, "grad_norm": 0.7093892097473145, "learning_rate": 1.7555162960042006e-05, "loss": 2.9771, "step": 89210 }, { "epoch": 5.12e-05, "grad_norm": 0.7889713048934937, "learning_rate": 1.755463537357517e-05, "loss": 3.1162, "step": 89220 }, { "epoch": 7.68e-05, "grad_norm": 0.7743122577667236, "learning_rate": 1.755410773811853e-05, "loss": 3.1998, "step": 89230 }, { "epoch": 0.0001024, "grad_norm": 0.8012115359306335, "learning_rate": 1.755358005367551e-05, "loss": 2.9729, "step": 89240 }, { "epoch": 0.000128, "grad_norm": 0.9378045797348022, "learning_rate": 1.7553052320249532e-05, "loss": 3.0025, "step": 89250 }, { "epoch": 0.0001536, "grad_norm": 0.7841284275054932, "learning_rate": 1.7552524537844018e-05, "loss": 3.1081, "step": 89260 }, { "epoch": 0.0001792, "grad_norm": 0.7976873517036438, "learning_rate": 1.7551996706462388e-05, "loss": 2.9424, "step": 89270 }, { "epoch": 0.0002048, "grad_norm": 0.7388816475868225, "learning_rate": 1.7551468826108068e-05, "loss": 3.2574, "step": 89280 }, { "epoch": 0.0002304, "grad_norm": 0.7454960346221924, "learning_rate": 1.7550940896784475e-05, "loss": 2.8673, "step": 89290 }, { "epoch": 0.000256, "grad_norm": 0.7914450168609619, "learning_rate": 1.755041291849504e-05, "loss": 2.987, "step": 89300 }, { "epoch": 0.0002816, "grad_norm": 0.8134008049964905, "learning_rate": 1.7549884891243187e-05, "loss": 3.1719, "step": 89310 }, { "epoch": 0.0003072, "grad_norm": 0.7980820536613464, "learning_rate": 1.7549356815032332e-05, "loss": 3.0912, "step": 89320 }, { "epoch": 0.0003328, "grad_norm": 0.6912443041801453, "learning_rate": 1.7548828689865906e-05, "loss": 2.8454, "step": 89330 }, { "epoch": 0.0003584, "grad_norm": 0.7540885210037231, "learning_rate": 1.7548300515747336e-05, "loss": 3.2436, "step": 89340 }, { "epoch": 0.000384, "grad_norm": 0.6523033380508423, "learning_rate": 1.754777229268004e-05, "loss": 2.9389, "step": 89350 }, { "epoch": 0.0004096, "grad_norm": 0.7461825013160706, "learning_rate": 1.7547244020667445e-05, "loss": 3.0045, "step": 89360 }, { "epoch": 0.0004352, "grad_norm": 0.7527627348899841, "learning_rate": 1.754671569971298e-05, "loss": 3.0023, "step": 89370 }, { "epoch": 0.0004608, "grad_norm": 0.8670065999031067, "learning_rate": 1.754618732982007e-05, "loss": 3.1391, "step": 89380 }, { "epoch": 0.0004864, "grad_norm": 0.9666017293930054, "learning_rate": 1.754565891099214e-05, "loss": 2.9578, "step": 89390 }, { "epoch": 0.000512, "grad_norm": 0.7884359955787659, "learning_rate": 1.754513044323262e-05, "loss": 3.2873, "step": 89400 }, { "epoch": 0.0005376, "grad_norm": 0.7589279413223267, "learning_rate": 1.7544601926544927e-05, "loss": 2.9869, "step": 89410 }, { "epoch": 0.0005632, "grad_norm": 0.6518863439559937, "learning_rate": 1.7544073360932502e-05, "loss": 2.9962, "step": 89420 }, { "epoch": 0.0005888, "grad_norm": 0.7294121980667114, "learning_rate": 1.7543544746398765e-05, "loss": 2.9753, "step": 89430 }, { "epoch": 0.0006144, "grad_norm": 0.8188264966011047, "learning_rate": 1.7543016082947146e-05, "loss": 3.0911, "step": 89440 }, { "epoch": 0.00064, "grad_norm": 0.7382212281227112, "learning_rate": 1.7542487370581073e-05, "loss": 3.265, "step": 89450 }, { "epoch": 0.0006656, "grad_norm": 0.7277348637580872, "learning_rate": 1.754195860930397e-05, "loss": 3.058, "step": 89460 }, { "epoch": 0.0006912, "grad_norm": 0.8625161647796631, "learning_rate": 1.7541429799119272e-05, "loss": 3.0442, "step": 89470 }, { "epoch": 0.0007168, "grad_norm": 0.6847667098045349, "learning_rate": 1.7540900940030406e-05, "loss": 2.9189, "step": 89480 }, { "epoch": 0.0007424, "grad_norm": 0.783582866191864, "learning_rate": 1.7540372032040802e-05, "loss": 3.1945, "step": 89490 }, { "epoch": 0.000768, "grad_norm": 0.6653429865837097, "learning_rate": 1.7539843075153888e-05, "loss": 2.934, "step": 89500 }, { "epoch": 0.0007936, "grad_norm": 0.6832624673843384, "learning_rate": 1.7539314069373098e-05, "loss": 3.0921, "step": 89510 }, { "epoch": 0.0008192, "grad_norm": 0.687330961227417, "learning_rate": 1.7538785014701857e-05, "loss": 2.8492, "step": 89520 }, { "epoch": 0.0008448, "grad_norm": 0.7036597728729248, "learning_rate": 1.75382559111436e-05, "loss": 3.0926, "step": 89530 }, { "epoch": 0.0008704, "grad_norm": 0.8440065383911133, "learning_rate": 1.7537726758701757e-05, "loss": 2.9489, "step": 89540 }, { "epoch": 0.000896, "grad_norm": 0.7178053855895996, "learning_rate": 1.753719755737976e-05, "loss": 2.852, "step": 89550 }, { "epoch": 0.0009216, "grad_norm": 0.7717252373695374, "learning_rate": 1.753666830718104e-05, "loss": 2.8795, "step": 89560 }, { "epoch": 0.0009472, "grad_norm": 0.7249424457550049, "learning_rate": 1.753613900810903e-05, "loss": 3.1587, "step": 89570 }, { "epoch": 0.0009728, "grad_norm": 0.6840929388999939, "learning_rate": 1.7535609660167155e-05, "loss": 3.1563, "step": 89580 }, { "epoch": 0.0009984, "grad_norm": 1.0006047487258911, "learning_rate": 1.753508026335886e-05, "loss": 2.9637, "step": 89590 }, { "epoch": 0.001024, "grad_norm": 0.7195070385932922, "learning_rate": 1.753455081768757e-05, "loss": 3.0015, "step": 89600 }, { "epoch": 0.0010496, "grad_norm": 0.6921126842498779, "learning_rate": 1.7534021323156723e-05, "loss": 3.0405, "step": 89610 }, { "epoch": 0.0010752, "grad_norm": 0.7469034790992737, "learning_rate": 1.7533491779769745e-05, "loss": 3.0879, "step": 89620 }, { "epoch": 0.0011008, "grad_norm": 0.7228841185569763, "learning_rate": 1.753296218753008e-05, "loss": 3.1094, "step": 89630 }, { "epoch": 0.0011264, "grad_norm": 0.7325500249862671, "learning_rate": 1.7532432546441157e-05, "loss": 2.8992, "step": 89640 }, { "epoch": 0.001152, "grad_norm": 0.8351637125015259, "learning_rate": 1.7531902856506407e-05, "loss": 3.0885, "step": 89650 }, { "epoch": 0.0011776, "grad_norm": 0.7057843208312988, "learning_rate": 1.753137311772927e-05, "loss": 2.8179, "step": 89660 }, { "epoch": 0.0012032, "grad_norm": 2.5137178897857666, "learning_rate": 1.7530843330113184e-05, "loss": 2.8788, "step": 89670 }, { "epoch": 0.0012288, "grad_norm": 0.7960616946220398, "learning_rate": 1.7530313493661575e-05, "loss": 3.0231, "step": 89680 }, { "epoch": 0.0012544, "grad_norm": 0.8106331825256348, "learning_rate": 1.7529783608377887e-05, "loss": 3.1237, "step": 89690 }, { "epoch": 0.00128, "grad_norm": 0.7076644897460938, "learning_rate": 1.7529253674265554e-05, "loss": 2.8989, "step": 89700 }, { "epoch": 0.0013056, "grad_norm": 0.7345626354217529, "learning_rate": 1.752872369132801e-05, "loss": 3.0532, "step": 89710 }, { "epoch": 0.0013312, "grad_norm": 0.8813334703445435, "learning_rate": 1.7528193659568696e-05, "loss": 3.0682, "step": 89720 }, { "epoch": 0.0013568, "grad_norm": 0.7222887277603149, "learning_rate": 1.7527663578991045e-05, "loss": 2.8762, "step": 89730 }, { "epoch": 0.0013824, "grad_norm": 0.7761253118515015, "learning_rate": 1.7527133449598496e-05, "loss": 2.8782, "step": 89740 }, { "epoch": 0.001408, "grad_norm": 1.0396891832351685, "learning_rate": 1.752660327139449e-05, "loss": 3.1349, "step": 89750 }, { "epoch": 0.0014336, "grad_norm": 0.8463903665542603, "learning_rate": 1.752607304438246e-05, "loss": 3.0637, "step": 89760 }, { "epoch": 0.0014592, "grad_norm": 0.9198597073554993, "learning_rate": 1.7525542768565846e-05, "loss": 2.8315, "step": 89770 }, { "epoch": 0.0014848, "grad_norm": 0.763375461101532, "learning_rate": 1.752501244394809e-05, "loss": 2.9782, "step": 89780 }, { "epoch": 0.0015104, "grad_norm": 0.8139882683753967, "learning_rate": 1.7524482070532625e-05, "loss": 2.7695, "step": 89790 }, { "epoch": 0.001536, "grad_norm": 0.7348301410675049, "learning_rate": 1.7523951648322893e-05, "loss": 2.867, "step": 89800 }, { "epoch": 2.56e-05, "grad_norm": 0.7718484997749329, "learning_rate": 1.7523421177322336e-05, "loss": 2.7988, "step": 89810 }, { "epoch": 5.12e-05, "grad_norm": 0.9060959815979004, "learning_rate": 1.752289065753439e-05, "loss": 3.0999, "step": 89820 }, { "epoch": 7.68e-05, "grad_norm": 1.9403480291366577, "learning_rate": 1.7522360088962502e-05, "loss": 3.366, "step": 89830 }, { "epoch": 0.0001024, "grad_norm": 0.8428324460983276, "learning_rate": 1.7521829471610105e-05, "loss": 3.2382, "step": 89840 }, { "epoch": 0.000128, "grad_norm": 0.7075491547584534, "learning_rate": 1.7521298805480644e-05, "loss": 3.0608, "step": 89850 }, { "epoch": 0.0001536, "grad_norm": 0.7066842913627625, "learning_rate": 1.752076809057756e-05, "loss": 2.9591, "step": 89860 }, { "epoch": 0.0001792, "grad_norm": 0.7594413757324219, "learning_rate": 1.7520237326904292e-05, "loss": 3.0669, "step": 89870 }, { "epoch": 0.0002048, "grad_norm": 0.8676992058753967, "learning_rate": 1.7519706514464282e-05, "loss": 3.0759, "step": 89880 }, { "epoch": 0.0002304, "grad_norm": 0.7539830803871155, "learning_rate": 1.7519175653260976e-05, "loss": 2.834, "step": 89890 }, { "epoch": 0.000256, "grad_norm": 0.7047944664955139, "learning_rate": 1.7518644743297814e-05, "loss": 3.1247, "step": 89900 }, { "epoch": 0.0002816, "grad_norm": 0.79027259349823, "learning_rate": 1.751811378457824e-05, "loss": 2.9679, "step": 89910 }, { "epoch": 0.0003072, "grad_norm": 0.8781192898750305, "learning_rate": 1.7517582777105696e-05, "loss": 3.1624, "step": 89920 }, { "epoch": 0.0003328, "grad_norm": 0.7727410793304443, "learning_rate": 1.7517051720883624e-05, "loss": 3.2024, "step": 89930 }, { "epoch": 0.0003584, "grad_norm": 1.5103130340576172, "learning_rate": 1.751652061591547e-05, "loss": 3.1415, "step": 89940 }, { "epoch": 0.000384, "grad_norm": 0.7131243944168091, "learning_rate": 1.7515989462204676e-05, "loss": 3.1275, "step": 89950 }, { "epoch": 0.0004096, "grad_norm": 0.7786942720413208, "learning_rate": 1.7515458259754688e-05, "loss": 3.1813, "step": 89960 }, { "epoch": 0.0004352, "grad_norm": 0.9216504693031311, "learning_rate": 1.7514927008568952e-05, "loss": 3.0282, "step": 89970 }, { "epoch": 0.0004608, "grad_norm": 0.8506332635879517, "learning_rate": 1.7514395708650914e-05, "loss": 3.0241, "step": 89980 }, { "epoch": 0.0004864, "grad_norm": 0.6840225458145142, "learning_rate": 1.7513864360004012e-05, "loss": 3.2173, "step": 89990 }, { "epoch": 0.000512, "grad_norm": 0.8919110894203186, "learning_rate": 1.75133329626317e-05, "loss": 2.8775, "step": 90000 }, { "epoch": 0.0005376, "grad_norm": 0.8016553521156311, "learning_rate": 1.751280151653742e-05, "loss": 2.9201, "step": 90010 }, { "epoch": 0.0005632, "grad_norm": 0.6987894177436829, "learning_rate": 1.751227002172462e-05, "loss": 2.9401, "step": 90020 }, { "epoch": 0.0005888, "grad_norm": 0.7115820050239563, "learning_rate": 1.7511738478196744e-05, "loss": 2.8916, "step": 90030 }, { "epoch": 0.0006144, "grad_norm": 0.6690685749053955, "learning_rate": 1.751120688595724e-05, "loss": 2.7839, "step": 90040 }, { "epoch": 0.00064, "grad_norm": 0.712440013885498, "learning_rate": 1.7510675245009555e-05, "loss": 2.7496, "step": 90050 }, { "epoch": 0.0006656, "grad_norm": 0.9650976061820984, "learning_rate": 1.7510143555357136e-05, "loss": 3.3407, "step": 90060 }, { "epoch": 0.0006912, "grad_norm": 0.8107706308364868, "learning_rate": 1.7509611817003436e-05, "loss": 3.2469, "step": 90070 }, { "epoch": 0.0007168, "grad_norm": 0.7619014978408813, "learning_rate": 1.7509080029951895e-05, "loss": 3.1244, "step": 90080 }, { "epoch": 0.0007424, "grad_norm": 0.7349917888641357, "learning_rate": 1.7508548194205967e-05, "loss": 3.4607, "step": 90090 }, { "epoch": 0.000768, "grad_norm": 0.6720074415206909, "learning_rate": 1.75080163097691e-05, "loss": 2.9028, "step": 90100 }, { "epoch": 0.0007936, "grad_norm": 0.7435144186019897, "learning_rate": 1.7507484376644744e-05, "loss": 2.9279, "step": 90110 }, { "epoch": 0.0008192, "grad_norm": 0.8566271066665649, "learning_rate": 1.7506952394836347e-05, "loss": 2.9867, "step": 90120 }, { "epoch": 0.0008448, "grad_norm": 0.6499602794647217, "learning_rate": 1.750642036434736e-05, "loss": 2.8542, "step": 90130 }, { "epoch": 0.0008704, "grad_norm": 1.2568153142929077, "learning_rate": 1.7505888285181226e-05, "loss": 3.0456, "step": 90140 }, { "epoch": 0.000896, "grad_norm": 0.8312628865242004, "learning_rate": 1.7505356157341404e-05, "loss": 3.0275, "step": 90150 }, { "epoch": 0.0009216, "grad_norm": 0.6900414228439331, "learning_rate": 1.7504823980831344e-05, "loss": 2.8822, "step": 90160 }, { "epoch": 0.0009472, "grad_norm": 0.6811083555221558, "learning_rate": 1.7504291755654494e-05, "loss": 2.9406, "step": 90170 }, { "epoch": 0.0009728, "grad_norm": 0.7437789440155029, "learning_rate": 1.750375948181431e-05, "loss": 3.1085, "step": 90180 }, { "epoch": 0.0009984, "grad_norm": 0.7283947467803955, "learning_rate": 1.7503227159314232e-05, "loss": 3.095, "step": 90190 }, { "epoch": 0.001024, "grad_norm": 0.7303115129470825, "learning_rate": 1.7502694788157725e-05, "loss": 2.9391, "step": 90200 }, { "epoch": 0.0010496, "grad_norm": 0.71155846118927, "learning_rate": 1.7502162368348238e-05, "loss": 2.8997, "step": 90210 }, { "epoch": 0.0010752, "grad_norm": 0.7193708419799805, "learning_rate": 1.7501629899889217e-05, "loss": 3.0211, "step": 90220 }, { "epoch": 0.0011008, "grad_norm": 0.7508090734481812, "learning_rate": 1.7501097382784126e-05, "loss": 3.0707, "step": 90230 }, { "epoch": 0.0011264, "grad_norm": 0.7282299995422363, "learning_rate": 1.7500564817036405e-05, "loss": 3.0152, "step": 90240 }, { "epoch": 0.001152, "grad_norm": 0.733551025390625, "learning_rate": 1.7500032202649518e-05, "loss": 3.1133, "step": 90250 }, { "epoch": 0.0011776, "grad_norm": 0.7402957081794739, "learning_rate": 1.7499499539626918e-05, "loss": 3.1056, "step": 90260 }, { "epoch": 0.0012032, "grad_norm": 0.7675060033798218, "learning_rate": 1.7498966827972053e-05, "loss": 3.0762, "step": 90270 }, { "epoch": 0.0012288, "grad_norm": 0.7558584213256836, "learning_rate": 1.7498434067688383e-05, "loss": 3.0182, "step": 90280 }, { "epoch": 0.0012544, "grad_norm": 0.8381984233856201, "learning_rate": 1.7497901258779357e-05, "loss": 2.9132, "step": 90290 }, { "epoch": 0.00128, "grad_norm": 0.7572138905525208, "learning_rate": 1.7497368401248437e-05, "loss": 2.9509, "step": 90300 }, { "epoch": 0.0013056, "grad_norm": 0.7392259836196899, "learning_rate": 1.7496835495099076e-05, "loss": 2.9477, "step": 90310 }, { "epoch": 0.0013312, "grad_norm": 0.7842286229133606, "learning_rate": 1.7496302540334727e-05, "loss": 2.9789, "step": 90320 }, { "epoch": 0.0013568, "grad_norm": 0.7211163640022278, "learning_rate": 1.7495769536958852e-05, "loss": 2.9707, "step": 90330 }, { "epoch": 0.0013824, "grad_norm": 0.6936587691307068, "learning_rate": 1.7495236484974904e-05, "loss": 3.0202, "step": 90340 }, { "epoch": 0.001408, "grad_norm": 0.7038611769676208, "learning_rate": 1.7494703384386335e-05, "loss": 3.0479, "step": 90350 }, { "epoch": 0.0014336, "grad_norm": 0.6784055233001709, "learning_rate": 1.7494170235196608e-05, "loss": 2.975, "step": 90360 }, { "epoch": 0.0014592, "grad_norm": 0.8503214120864868, "learning_rate": 1.749363703740918e-05, "loss": 3.0019, "step": 90370 }, { "epoch": 0.0014848, "grad_norm": 0.7453076839447021, "learning_rate": 1.7493103791027507e-05, "loss": 2.8176, "step": 90380 }, { "epoch": 0.0015104, "grad_norm": 0.7765692472457886, "learning_rate": 1.7492570496055048e-05, "loss": 2.8136, "step": 90390 }, { "epoch": 0.001536, "grad_norm": 0.7833457589149475, "learning_rate": 1.7492037152495257e-05, "loss": 3.034, "step": 90400 }, { "epoch": 0.0015616, "grad_norm": 1.0136784315109253, "learning_rate": 1.74915037603516e-05, "loss": 3.0728, "step": 90410 }, { "epoch": 0.0015872, "grad_norm": 0.9107474684715271, "learning_rate": 1.749097031962753e-05, "loss": 2.9785, "step": 90420 }, { "epoch": 0.0016128, "grad_norm": 0.8288192749023438, "learning_rate": 1.749043683032651e-05, "loss": 3.0922, "step": 90430 }, { "epoch": 0.0016384, "grad_norm": 0.8462816476821899, "learning_rate": 1.7489903292451998e-05, "loss": 2.9894, "step": 90440 }, { "epoch": 0.001664, "grad_norm": 0.6652730703353882, "learning_rate": 1.7489369706007455e-05, "loss": 3.0366, "step": 90450 }, { "epoch": 0.0016896, "grad_norm": 0.7612448930740356, "learning_rate": 1.7488836070996336e-05, "loss": 3.1787, "step": 90460 }, { "epoch": 0.0017152, "grad_norm": 0.822656512260437, "learning_rate": 1.7488302387422105e-05, "loss": 3.1564, "step": 90470 }, { "epoch": 0.0017408, "grad_norm": 0.6702625155448914, "learning_rate": 1.7487768655288227e-05, "loss": 3.0032, "step": 90480 }, { "epoch": 0.0017664, "grad_norm": 0.8353177905082703, "learning_rate": 1.7487234874598158e-05, "loss": 2.9089, "step": 90490 }, { "epoch": 0.001792, "grad_norm": 0.7807754874229431, "learning_rate": 1.7486701045355358e-05, "loss": 3.0024, "step": 90500 }, { "epoch": 0.0018176, "grad_norm": 0.7582554221153259, "learning_rate": 1.7486167167563295e-05, "loss": 2.8745, "step": 90510 }, { "epoch": 0.0018432, "grad_norm": 0.8272963762283325, "learning_rate": 1.7485633241225427e-05, "loss": 3.1668, "step": 90520 }, { "epoch": 0.0018688, "grad_norm": 0.7961472272872925, "learning_rate": 1.7485099266345214e-05, "loss": 3.0797, "step": 90530 }, { "epoch": 0.0018944, "grad_norm": 0.7676002979278564, "learning_rate": 1.7484565242926125e-05, "loss": 2.9866, "step": 90540 }, { "epoch": 0.00192, "grad_norm": 0.7800533175468445, "learning_rate": 1.7484031170971617e-05, "loss": 2.9523, "step": 90550 }, { "epoch": 0.0019456, "grad_norm": 0.7310987710952759, "learning_rate": 1.7483497050485157e-05, "loss": 2.871, "step": 90560 }, { "epoch": 0.0019712, "grad_norm": 0.9817436933517456, "learning_rate": 1.7482962881470208e-05, "loss": 3.276, "step": 90570 }, { "epoch": 0.0019968, "grad_norm": 0.8435827493667603, "learning_rate": 1.7482428663930232e-05, "loss": 3.0662, "step": 90580 }, { "epoch": 0.0020224, "grad_norm": 0.6880326271057129, "learning_rate": 1.74818943978687e-05, "loss": 2.8153, "step": 90590 }, { "epoch": 0.002048, "grad_norm": 0.7066754698753357, "learning_rate": 1.7481360083289064e-05, "loss": 2.8893, "step": 90600 }, { "epoch": 0.0020736, "grad_norm": 0.873711109161377, "learning_rate": 1.74808257201948e-05, "loss": 3.0152, "step": 90610 }, { "epoch": 0.0020992, "grad_norm": 0.8112943172454834, "learning_rate": 1.7480291308589368e-05, "loss": 3.1508, "step": 90620 }, { "epoch": 0.0021248, "grad_norm": 1.0301700830459595, "learning_rate": 1.7479756848476235e-05, "loss": 2.8555, "step": 90630 }, { "epoch": 0.0021504, "grad_norm": 0.9400896430015564, "learning_rate": 1.747922233985887e-05, "loss": 3.2699, "step": 90640 }, { "epoch": 0.002176, "grad_norm": 0.8347300291061401, "learning_rate": 1.747868778274073e-05, "loss": 3.1725, "step": 90650 }, { "epoch": 0.0022016, "grad_norm": 0.7333155870437622, "learning_rate": 1.747815317712529e-05, "loss": 3.09, "step": 90660 }, { "epoch": 0.0022272, "grad_norm": 0.866084635257721, "learning_rate": 1.7477618523016017e-05, "loss": 3.1157, "step": 90670 }, { "epoch": 0.0022528, "grad_norm": 0.98607337474823, "learning_rate": 1.747708382041637e-05, "loss": 3.096, "step": 90680 }, { "epoch": 0.0022784, "grad_norm": 2.4757535457611084, "learning_rate": 1.7476549069329825e-05, "loss": 2.9703, "step": 90690 }, { "epoch": 0.002304, "grad_norm": 0.7932751178741455, "learning_rate": 1.7476014269759846e-05, "loss": 2.7875, "step": 90700 }, { "epoch": 0.0023296, "grad_norm": 0.8256769776344299, "learning_rate": 1.7475479421709903e-05, "loss": 3.03, "step": 90710 }, { "epoch": 0.0023552, "grad_norm": 0.9062606692314148, "learning_rate": 1.7474944525183458e-05, "loss": 2.9916, "step": 90720 }, { "epoch": 0.0023808, "grad_norm": 0.7732576131820679, "learning_rate": 1.7474409580183987e-05, "loss": 2.9108, "step": 90730 }, { "epoch": 0.0024064, "grad_norm": 0.6762285828590393, "learning_rate": 1.747387458671496e-05, "loss": 2.983, "step": 90740 }, { "epoch": 0.002432, "grad_norm": 0.9567148089408875, "learning_rate": 1.7473339544779836e-05, "loss": 3.1752, "step": 90750 }, { "epoch": 0.0024576, "grad_norm": 0.6988472938537598, "learning_rate": 1.7472804454382094e-05, "loss": 2.9393, "step": 90760 }, { "epoch": 0.0024832, "grad_norm": 0.7003776431083679, "learning_rate": 1.74722693155252e-05, "loss": 3.113, "step": 90770 }, { "epoch": 0.0025088, "grad_norm": 0.9425477981567383, "learning_rate": 1.7471734128212627e-05, "loss": 3.1197, "step": 90780 }, { "epoch": 0.0025344, "grad_norm": 0.832767128944397, "learning_rate": 1.7471198892447843e-05, "loss": 3.1434, "step": 90790 }, { "epoch": 0.00256, "grad_norm": 0.781799852848053, "learning_rate": 1.7470663608234323e-05, "loss": 3.1867, "step": 90800 }, { "epoch": 0.0025856, "grad_norm": 0.7262840270996094, "learning_rate": 1.747012827557553e-05, "loss": 3.0579, "step": 90810 }, { "epoch": 0.0026112, "grad_norm": 0.7911447286605835, "learning_rate": 1.7469592894474945e-05, "loss": 3.2271, "step": 90820 }, { "epoch": 0.0026368, "grad_norm": 0.7968208193778992, "learning_rate": 1.7469057464936033e-05, "loss": 3.3513, "step": 90830 }, { "epoch": 0.0026624, "grad_norm": 0.7731850147247314, "learning_rate": 1.7468521986962266e-05, "loss": 3.2767, "step": 90840 }, { "epoch": 0.002688, "grad_norm": 0.8343953490257263, "learning_rate": 1.7467986460557124e-05, "loss": 2.9894, "step": 90850 }, { "epoch": 0.0027136, "grad_norm": 0.8116205930709839, "learning_rate": 1.7467450885724072e-05, "loss": 3.1487, "step": 90860 }, { "epoch": 0.0027392, "grad_norm": 0.7225197553634644, "learning_rate": 1.7466915262466587e-05, "loss": 3.2251, "step": 90870 }, { "epoch": 0.0027648, "grad_norm": 2.0973126888275146, "learning_rate": 1.7466379590788135e-05, "loss": 3.154, "step": 90880 }, { "epoch": 0.0027904, "grad_norm": 0.6941642761230469, "learning_rate": 1.7465843870692203e-05, "loss": 2.8966, "step": 90890 }, { "epoch": 0.002816, "grad_norm": 0.8053696155548096, "learning_rate": 1.7465308102182255e-05, "loss": 3.213, "step": 90900 }, { "epoch": 0.0028416, "grad_norm": 0.7687053680419922, "learning_rate": 1.7464772285261766e-05, "loss": 3.1796, "step": 90910 }, { "epoch": 0.0028672, "grad_norm": 0.762666642665863, "learning_rate": 1.7464236419934212e-05, "loss": 3.1966, "step": 90920 }, { "epoch": 0.0028928, "grad_norm": 0.848492443561554, "learning_rate": 1.7463700506203075e-05, "loss": 3.1134, "step": 90930 }, { "epoch": 0.0029184, "grad_norm": 0.7031280994415283, "learning_rate": 1.746316454407182e-05, "loss": 3.0685, "step": 90940 }, { "epoch": 0.002944, "grad_norm": 0.7736741304397583, "learning_rate": 1.7462628533543923e-05, "loss": 3.2568, "step": 90950 }, { "epoch": 0.0029696, "grad_norm": 0.7479258179664612, "learning_rate": 1.746209247462287e-05, "loss": 3.1958, "step": 90960 }, { "epoch": 0.0029952, "grad_norm": 1.788123369216919, "learning_rate": 1.7461556367312126e-05, "loss": 3.0693, "step": 90970 }, { "epoch": 0.0030208, "grad_norm": 0.8807259798049927, "learning_rate": 1.7461020211615172e-05, "loss": 3.3345, "step": 90980 }, { "epoch": 0.0030464, "grad_norm": 0.7682965993881226, "learning_rate": 1.746048400753549e-05, "loss": 2.9247, "step": 90990 }, { "epoch": 0.003072, "grad_norm": 0.6717550754547119, "learning_rate": 1.7459947755076544e-05, "loss": 3.0941, "step": 91000 }, { "epoch": 0.0030976, "grad_norm": 0.8227336406707764, "learning_rate": 1.7459411454241822e-05, "loss": 3.2383, "step": 91010 }, { "epoch": 0.0031232, "grad_norm": 0.7586768269538879, "learning_rate": 1.7458875105034804e-05, "loss": 3.3644, "step": 91020 }, { "epoch": 0.0031488, "grad_norm": 0.7172437906265259, "learning_rate": 1.7458338707458956e-05, "loss": 3.1408, "step": 91030 }, { "epoch": 0.0031744, "grad_norm": 0.9249356985092163, "learning_rate": 1.745780226151777e-05, "loss": 3.2602, "step": 91040 }, { "epoch": 0.0032, "grad_norm": 0.773296594619751, "learning_rate": 1.7457265767214716e-05, "loss": 3.4307, "step": 91050 }, { "epoch": 0.0032256, "grad_norm": 0.774019181728363, "learning_rate": 1.7456729224553274e-05, "loss": 2.9078, "step": 91060 }, { "epoch": 0.0032512, "grad_norm": 0.6738027930259705, "learning_rate": 1.7456192633536926e-05, "loss": 3.1635, "step": 91070 }, { "epoch": 0.0032768, "grad_norm": 0.721119225025177, "learning_rate": 1.745565599416915e-05, "loss": 3.0566, "step": 91080 }, { "epoch": 0.0033024, "grad_norm": 0.6866554617881775, "learning_rate": 1.7455119306453427e-05, "loss": 3.1256, "step": 91090 }, { "epoch": 0.003328, "grad_norm": 0.9160412549972534, "learning_rate": 1.7454582570393235e-05, "loss": 3.3352, "step": 91100 }, { "epoch": 0.0033536, "grad_norm": 0.6568057537078857, "learning_rate": 1.745404578599206e-05, "loss": 3.0965, "step": 91110 }, { "epoch": 0.0033792, "grad_norm": 0.7443190813064575, "learning_rate": 1.7453508953253373e-05, "loss": 3.1622, "step": 91120 }, { "epoch": 0.0034048, "grad_norm": 0.70013827085495, "learning_rate": 1.745297207218067e-05, "loss": 2.8856, "step": 91130 }, { "epoch": 0.0034304, "grad_norm": 0.8264867067337036, "learning_rate": 1.7452435142777417e-05, "loss": 3.0066, "step": 91140 }, { "epoch": 0.003456, "grad_norm": 0.8297504186630249, "learning_rate": 1.7451898165047103e-05, "loss": 3.0908, "step": 91150 }, { "epoch": 0.0034816, "grad_norm": 0.7220191359519958, "learning_rate": 1.745136113899321e-05, "loss": 3.1078, "step": 91160 }, { "epoch": 0.0035072, "grad_norm": 0.7998504638671875, "learning_rate": 1.7450824064619223e-05, "loss": 3.179, "step": 91170 }, { "epoch": 0.0035328, "grad_norm": 0.6634871959686279, "learning_rate": 1.745028694192862e-05, "loss": 3.1981, "step": 91180 }, { "epoch": 0.0035584, "grad_norm": 0.7342963218688965, "learning_rate": 1.744974977092489e-05, "loss": 2.9612, "step": 91190 }, { "epoch": 0.003584, "grad_norm": 0.8778988122940063, "learning_rate": 1.7449212551611506e-05, "loss": 3.1321, "step": 91200 }, { "epoch": 0.0036096, "grad_norm": 1.1073567867279053, "learning_rate": 1.7448675283991963e-05, "loss": 3.146, "step": 91210 }, { "epoch": 0.0036352, "grad_norm": 0.7245727777481079, "learning_rate": 1.744813796806974e-05, "loss": 3.1326, "step": 91220 }, { "epoch": 0.0036608, "grad_norm": 0.8516531586647034, "learning_rate": 1.7447600603848323e-05, "loss": 3.0948, "step": 91230 }, { "epoch": 0.0036864, "grad_norm": 0.8588503003120422, "learning_rate": 1.7447063191331193e-05, "loss": 3.1657, "step": 91240 }, { "epoch": 0.003712, "grad_norm": 0.812628984451294, "learning_rate": 1.744652573052184e-05, "loss": 3.2136, "step": 91250 }, { "epoch": 0.0037376, "grad_norm": 0.7137179374694824, "learning_rate": 1.7445988221423742e-05, "loss": 3.1585, "step": 91260 }, { "epoch": 0.0037632, "grad_norm": 0.8923192620277405, "learning_rate": 1.7445450664040394e-05, "loss": 3.1435, "step": 91270 }, { "epoch": 0.0037888, "grad_norm": 0.7483110427856445, "learning_rate": 1.7444913058375273e-05, "loss": 3.145, "step": 91280 }, { "epoch": 0.0038144, "grad_norm": 0.7921314835548401, "learning_rate": 1.7444375404431873e-05, "loss": 2.997, "step": 91290 }, { "epoch": 0.00384, "grad_norm": 0.6997482776641846, "learning_rate": 1.7443837702213676e-05, "loss": 2.9695, "step": 91300 }, { "epoch": 0.0038656, "grad_norm": 0.7382249236106873, "learning_rate": 1.744329995172417e-05, "loss": 3.2551, "step": 91310 }, { "epoch": 0.0038912, "grad_norm": 0.7782419919967651, "learning_rate": 1.744276215296684e-05, "loss": 3.2366, "step": 91320 }, { "epoch": 0.0039168, "grad_norm": 0.7943083047866821, "learning_rate": 1.7442224305945176e-05, "loss": 3.0716, "step": 91330 }, { "epoch": 0.0039424, "grad_norm": 0.8703163266181946, "learning_rate": 1.744168641066267e-05, "loss": 3.0723, "step": 91340 }, { "epoch": 0.003968, "grad_norm": 0.7419992089271545, "learning_rate": 1.7441148467122803e-05, "loss": 3.2258, "step": 91350 }, { "epoch": 0.0039936, "grad_norm": 0.7720925211906433, "learning_rate": 1.744061047532906e-05, "loss": 3.1715, "step": 91360 }, { "epoch": 0.0040192, "grad_norm": 0.7961704134941101, "learning_rate": 1.7440072435284942e-05, "loss": 3.2321, "step": 91370 }, { "epoch": 0.0040448, "grad_norm": 0.7623945474624634, "learning_rate": 1.743953434699393e-05, "loss": 3.3438, "step": 91380 }, { "epoch": 0.0040704, "grad_norm": 0.7706907391548157, "learning_rate": 1.7438996210459514e-05, "loss": 3.4429, "step": 91390 }, { "epoch": 0.004096, "grad_norm": 0.7136208415031433, "learning_rate": 1.7438458025685184e-05, "loss": 3.113, "step": 91400 }, { "epoch": 0.0041216, "grad_norm": 0.8031986355781555, "learning_rate": 1.743791979267443e-05, "loss": 3.1646, "step": 91410 }, { "epoch": 0.0041472, "grad_norm": 0.6776062846183777, "learning_rate": 1.7437381511430746e-05, "loss": 3.2142, "step": 91420 }, { "epoch": 0.0041728, "grad_norm": 0.6930829882621765, "learning_rate": 1.743684318195762e-05, "loss": 2.9735, "step": 91430 }, { "epoch": 0.0041984, "grad_norm": 0.7447504997253418, "learning_rate": 1.7436304804258538e-05, "loss": 3.0408, "step": 91440 }, { "epoch": 0.004224, "grad_norm": 1.0311707258224487, "learning_rate": 1.7435766378337e-05, "loss": 3.0815, "step": 91450 }, { "epoch": 0.0042496, "grad_norm": 0.7861741185188293, "learning_rate": 1.7435227904196493e-05, "loss": 2.8742, "step": 91460 }, { "epoch": 0.0042752, "grad_norm": 0.7351830005645752, "learning_rate": 1.7434689381840505e-05, "loss": 3.1852, "step": 91470 }, { "epoch": 0.0043008, "grad_norm": 0.8988085985183716, "learning_rate": 1.7434150811272537e-05, "loss": 3.1415, "step": 91480 }, { "epoch": 0.0043264, "grad_norm": 0.8472475409507751, "learning_rate": 1.7433612192496075e-05, "loss": 3.1905, "step": 91490 }, { "epoch": 0.004352, "grad_norm": 0.8265113234519958, "learning_rate": 1.7433073525514613e-05, "loss": 3.3287, "step": 91500 }, { "epoch": 0.0043776, "grad_norm": 0.7128486037254333, "learning_rate": 1.7432534810331646e-05, "loss": 3.2697, "step": 91510 }, { "epoch": 0.0044032, "grad_norm": 0.7429105639457703, "learning_rate": 1.7431996046950664e-05, "loss": 3.1803, "step": 91520 }, { "epoch": 0.0044288, "grad_norm": 0.7527937293052673, "learning_rate": 1.7431457235375165e-05, "loss": 3.1736, "step": 91530 }, { "epoch": 0.0044544, "grad_norm": 2.6276357173919678, "learning_rate": 1.7430918375608638e-05, "loss": 3.1017, "step": 91540 }, { "epoch": 0.00448, "grad_norm": 0.7654339671134949, "learning_rate": 1.7430379467654582e-05, "loss": 3.1461, "step": 91550 }, { "epoch": 0.0045056, "grad_norm": 0.692213237285614, "learning_rate": 1.742984051151649e-05, "loss": 3.2386, "step": 91560 }, { "epoch": 0.0045312, "grad_norm": 0.7173845767974854, "learning_rate": 1.7429301507197856e-05, "loss": 3.1349, "step": 91570 }, { "epoch": 0.0045568, "grad_norm": 0.7972906231880188, "learning_rate": 1.7428762454702178e-05, "loss": 3.0736, "step": 91580 }, { "epoch": 0.0045824, "grad_norm": 0.9747942686080933, "learning_rate": 1.7428223354032948e-05, "loss": 3.2331, "step": 91590 }, { "epoch": 0.004608, "grad_norm": 0.6871944665908813, "learning_rate": 1.7427684205193666e-05, "loss": 3.155, "step": 91600 }, { "epoch": 0.0046336, "grad_norm": 1.0161186456680298, "learning_rate": 1.7427145008187825e-05, "loss": 3.1941, "step": 91610 }, { "epoch": 0.0046592, "grad_norm": 0.7653314471244812, "learning_rate": 1.742660576301892e-05, "loss": 3.1593, "step": 91620 }, { "epoch": 0.0046848, "grad_norm": 0.6620191335678101, "learning_rate": 1.742606646969045e-05, "loss": 3.1479, "step": 91630 }, { "epoch": 0.0047104, "grad_norm": 0.6912354826927185, "learning_rate": 1.742552712820592e-05, "loss": 3.2626, "step": 91640 }, { "epoch": 0.004736, "grad_norm": 0.8192541003227234, "learning_rate": 1.7424987738568814e-05, "loss": 3.3293, "step": 91650 }, { "epoch": 0.0047616, "grad_norm": 0.7880113124847412, "learning_rate": 1.7424448300782636e-05, "loss": 3.1094, "step": 91660 }, { "epoch": 0.0047872, "grad_norm": 0.7142509818077087, "learning_rate": 1.7423908814850885e-05, "loss": 3.2793, "step": 91670 }, { "epoch": 0.0048128, "grad_norm": 0.7509519457817078, "learning_rate": 1.7423369280777058e-05, "loss": 3.1719, "step": 91680 }, { "epoch": 0.0048384, "grad_norm": 0.7908020615577698, "learning_rate": 1.7422829698564654e-05, "loss": 3.0932, "step": 91690 }, { "epoch": 0.004864, "grad_norm": 0.8602398633956909, "learning_rate": 1.742229006821717e-05, "loss": 3.0871, "step": 91700 }, { "epoch": 0.0048896, "grad_norm": 0.716533899307251, "learning_rate": 1.7421750389738114e-05, "loss": 3.0795, "step": 91710 }, { "epoch": 0.0049152, "grad_norm": 0.746271550655365, "learning_rate": 1.7421210663130974e-05, "loss": 3.1506, "step": 91720 }, { "epoch": 0.0049408, "grad_norm": 0.72944575548172, "learning_rate": 1.7420670888399257e-05, "loss": 3.135, "step": 91730 }, { "epoch": 0.0049664, "grad_norm": 0.7526459693908691, "learning_rate": 1.742013106554646e-05, "loss": 3.1253, "step": 91740 }, { "epoch": 0.004992, "grad_norm": 0.6823228001594543, "learning_rate": 1.7419591194576086e-05, "loss": 3.2537, "step": 91750 }, { "epoch": 0.0050176, "grad_norm": 1.0156364440917969, "learning_rate": 1.7419051275491635e-05, "loss": 3.0343, "step": 91760 }, { "epoch": 0.0050432, "grad_norm": 0.8244926929473877, "learning_rate": 1.7418511308296607e-05, "loss": 3.823, "step": 91770 }, { "epoch": 0.0050688, "grad_norm": 0.849818229675293, "learning_rate": 1.7417971292994506e-05, "loss": 3.2539, "step": 91780 }, { "epoch": 0.0050944, "grad_norm": 1.087821125984192, "learning_rate": 1.741743122958883e-05, "loss": 3.1912, "step": 91790 }, { "epoch": 0.00512, "grad_norm": 0.7561004161834717, "learning_rate": 1.741689111808309e-05, "loss": 3.1781, "step": 91800 }, { "epoch": 0.0051456, "grad_norm": 0.7867112159729004, "learning_rate": 1.7416350958480777e-05, "loss": 3.2705, "step": 91810 }, { "epoch": 0.0051712, "grad_norm": 0.9362084865570068, "learning_rate": 1.74158107507854e-05, "loss": 3.2843, "step": 91820 }, { "epoch": 0.0051968, "grad_norm": 0.7933134436607361, "learning_rate": 1.741527049500046e-05, "loss": 3.183, "step": 91830 }, { "epoch": 0.0052224, "grad_norm": 0.9999598264694214, "learning_rate": 1.7414730191129464e-05, "loss": 3.2896, "step": 91840 }, { "epoch": 0.005248, "grad_norm": 0.7709701657295227, "learning_rate": 1.7414189839175914e-05, "loss": 3.195, "step": 91850 }, { "epoch": 0.0052736, "grad_norm": 0.8359023332595825, "learning_rate": 1.7413649439143313e-05, "loss": 3.004, "step": 91860 }, { "epoch": 0.0052992, "grad_norm": 0.7461755275726318, "learning_rate": 1.7413108991035165e-05, "loss": 3.1319, "step": 91870 }, { "epoch": 0.0053248, "grad_norm": 0.7722505927085876, "learning_rate": 1.7412568494854973e-05, "loss": 3.0645, "step": 91880 }, { "epoch": 0.0053504, "grad_norm": 0.7960137724876404, "learning_rate": 1.7412027950606248e-05, "loss": 2.949, "step": 91890 }, { "epoch": 0.005376, "grad_norm": 0.8225464224815369, "learning_rate": 1.741148735829249e-05, "loss": 3.1742, "step": 91900 }, { "epoch": 0.0054016, "grad_norm": 0.7210431098937988, "learning_rate": 1.7410946717917207e-05, "loss": 2.9413, "step": 91910 }, { "epoch": 0.0054272, "grad_norm": 1.098550796508789, "learning_rate": 1.74104060294839e-05, "loss": 3.3347, "step": 91920 }, { "epoch": 0.0054528, "grad_norm": 0.687559187412262, "learning_rate": 1.7409865292996084e-05, "loss": 3.1315, "step": 91930 }, { "epoch": 0.0054784, "grad_norm": 0.7143179774284363, "learning_rate": 1.740932450845726e-05, "loss": 3.0936, "step": 91940 }, { "epoch": 0.005504, "grad_norm": 0.8618747591972351, "learning_rate": 1.7408783675870935e-05, "loss": 2.9973, "step": 91950 }, { "epoch": 0.0055296, "grad_norm": 0.8529782295227051, "learning_rate": 1.7408242795240618e-05, "loss": 3.2008, "step": 91960 }, { "epoch": 0.0055552, "grad_norm": 0.8739913702011108, "learning_rate": 1.7407701866569815e-05, "loss": 3.3667, "step": 91970 }, { "epoch": 0.0055808, "grad_norm": 0.6860520243644714, "learning_rate": 1.7407160889862032e-05, "loss": 3.3855, "step": 91980 }, { "epoch": 0.0056064, "grad_norm": 0.7557374238967896, "learning_rate": 1.7406619865120783e-05, "loss": 3.2866, "step": 91990 }, { "epoch": 0.005632, "grad_norm": 0.7382397651672363, "learning_rate": 1.7406078792349567e-05, "loss": 3.0293, "step": 92000 }, { "epoch": 0.0056576, "grad_norm": 0.6712401509284973, "learning_rate": 1.7405537671551904e-05, "loss": 3.1142, "step": 92010 }, { "epoch": 0.0056832, "grad_norm": 1.3186911344528198, "learning_rate": 1.7404996502731296e-05, "loss": 3.27, "step": 92020 }, { "epoch": 0.0057088, "grad_norm": 0.7002826929092407, "learning_rate": 1.7404455285891254e-05, "loss": 3.1471, "step": 92030 }, { "epoch": 0.0057344, "grad_norm": 0.7312697768211365, "learning_rate": 1.7403914021035283e-05, "loss": 3.1774, "step": 92040 }, { "epoch": 0.00576, "grad_norm": 0.8082467913627625, "learning_rate": 1.74033727081669e-05, "loss": 2.9651, "step": 92050 }, { "epoch": 0.0057856, "grad_norm": 0.8173248171806335, "learning_rate": 1.7402831347289614e-05, "loss": 3.0642, "step": 92060 }, { "epoch": 0.0058112, "grad_norm": 0.7791054248809814, "learning_rate": 1.7402289938406936e-05, "loss": 3.1939, "step": 92070 }, { "epoch": 0.0058368, "grad_norm": 0.7219802737236023, "learning_rate": 1.7401748481522372e-05, "loss": 3.0213, "step": 92080 }, { "epoch": 0.0058624, "grad_norm": 0.689076840877533, "learning_rate": 1.7401206976639436e-05, "loss": 3.1238, "step": 92090 }, { "epoch": 0.005888, "grad_norm": 0.7289261817932129, "learning_rate": 1.7400665423761637e-05, "loss": 3.103, "step": 92100 }, { "epoch": 0.0059136, "grad_norm": 0.716365396976471, "learning_rate": 1.7400123822892494e-05, "loss": 3.1776, "step": 92110 }, { "epoch": 0.0059392, "grad_norm": 0.9042869210243225, "learning_rate": 1.7399582174035515e-05, "loss": 3.1763, "step": 92120 }, { "epoch": 0.0059648, "grad_norm": 0.6944425702095032, "learning_rate": 1.7399040477194212e-05, "loss": 3.1726, "step": 92130 }, { "epoch": 0.0059904, "grad_norm": 0.8041067719459534, "learning_rate": 1.7398498732372094e-05, "loss": 3.3451, "step": 92140 }, { "epoch": 0.006016, "grad_norm": 0.8209594488143921, "learning_rate": 1.739795693957268e-05, "loss": 3.0568, "step": 92150 }, { "epoch": 0.0060416, "grad_norm": 0.7250490784645081, "learning_rate": 1.7397415098799483e-05, "loss": 3.0967, "step": 92160 }, { "epoch": 0.0060672, "grad_norm": 0.7611508965492249, "learning_rate": 1.7396873210056015e-05, "loss": 3.0563, "step": 92170 }, { "epoch": 0.0060928, "grad_norm": 0.764194130897522, "learning_rate": 1.739633127334579e-05, "loss": 3.0658, "step": 92180 }, { "epoch": 0.0061184, "grad_norm": 0.7337713837623596, "learning_rate": 1.739578928867232e-05, "loss": 3.0193, "step": 92190 }, { "epoch": 0.006144, "grad_norm": 2.019536256790161, "learning_rate": 1.7395247256039125e-05, "loss": 3.203, "step": 92200 }, { "epoch": 0.0061696, "grad_norm": 0.7754158973693848, "learning_rate": 1.7394705175449714e-05, "loss": 3.1432, "step": 92210 }, { "epoch": 0.0061952, "grad_norm": 0.6975699663162231, "learning_rate": 1.739416304690761e-05, "loss": 3.0967, "step": 92220 }, { "epoch": 0.0062208, "grad_norm": 0.7114226818084717, "learning_rate": 1.7393620870416315e-05, "loss": 3.0798, "step": 92230 }, { "epoch": 0.0062464, "grad_norm": 0.7550817728042603, "learning_rate": 1.739307864597936e-05, "loss": 3.128, "step": 92240 }, { "epoch": 0.006272, "grad_norm": 0.7911688089370728, "learning_rate": 1.7392536373600258e-05, "loss": 3.1802, "step": 92250 }, { "epoch": 0.0062976, "grad_norm": 1.1717935800552368, "learning_rate": 1.739199405328252e-05, "loss": 3.1728, "step": 92260 }, { "epoch": 0.0063232, "grad_norm": 0.8083217144012451, "learning_rate": 1.739145168502966e-05, "loss": 3.0059, "step": 92270 }, { "epoch": 0.0063488, "grad_norm": 0.7820820212364197, "learning_rate": 1.7390909268845202e-05, "loss": 3.0085, "step": 92280 }, { "epoch": 0.0063744, "grad_norm": 0.8039422631263733, "learning_rate": 1.7390366804732666e-05, "loss": 3.2522, "step": 92290 }, { "epoch": 0.0064, "grad_norm": 1.0710763931274414, "learning_rate": 1.7389824292695562e-05, "loss": 3.2134, "step": 92300 }, { "epoch": 0.0064256, "grad_norm": 0.9651048183441162, "learning_rate": 1.7389281732737412e-05, "loss": 3.1931, "step": 92310 }, { "epoch": 0.0064512, "grad_norm": 0.7226186394691467, "learning_rate": 1.7388739124861735e-05, "loss": 3.0463, "step": 92320 }, { "epoch": 0.0064768, "grad_norm": 0.8019182085990906, "learning_rate": 1.7388196469072047e-05, "loss": 3.1348, "step": 92330 }, { "epoch": 0.0065024, "grad_norm": 0.742724597454071, "learning_rate": 1.738765376537187e-05, "loss": 3.2047, "step": 92340 }, { "epoch": 0.006528, "grad_norm": 0.6770052313804626, "learning_rate": 1.738711101376472e-05, "loss": 2.865, "step": 92350 }, { "epoch": 0.0065536, "grad_norm": 0.7216173410415649, "learning_rate": 1.738656821425412e-05, "loss": 3.1868, "step": 92360 }, { "epoch": 0.0065792, "grad_norm": 0.8104188442230225, "learning_rate": 1.7386025366843587e-05, "loss": 2.9998, "step": 92370 }, { "epoch": 0.0066048, "grad_norm": 0.8173940181732178, "learning_rate": 1.738548247153664e-05, "loss": 3.065, "step": 92380 }, { "epoch": 0.0066304, "grad_norm": 1.1840015649795532, "learning_rate": 1.7384939528336804e-05, "loss": 3.1846, "step": 92390 }, { "epoch": 0.006656, "grad_norm": 0.980670690536499, "learning_rate": 1.7384396537247603e-05, "loss": 2.9736, "step": 92400 }, { "epoch": 0.0066816, "grad_norm": 0.7280896306037903, "learning_rate": 1.7383853498272548e-05, "loss": 3.1923, "step": 92410 }, { "epoch": 0.0067072, "grad_norm": 0.8048612475395203, "learning_rate": 1.7383310411415165e-05, "loss": 3.192, "step": 92420 }, { "epoch": 0.0067328, "grad_norm": 0.733686089515686, "learning_rate": 1.7382767276678978e-05, "loss": 3.0784, "step": 92430 }, { "epoch": 0.0067584, "grad_norm": 0.7528539896011353, "learning_rate": 1.7382224094067508e-05, "loss": 3.088, "step": 92440 }, { "epoch": 0.006784, "grad_norm": 0.8432319760322571, "learning_rate": 1.7381680863584275e-05, "loss": 3.0966, "step": 92450 }, { "epoch": 0.0068096, "grad_norm": 0.9045695066452026, "learning_rate": 1.73811375852328e-05, "loss": 3.0983, "step": 92460 }, { "epoch": 0.0068352, "grad_norm": 0.8038491606712341, "learning_rate": 1.7380594259016615e-05, "loss": 3.0653, "step": 92470 }, { "epoch": 0.0068608, "grad_norm": 0.7609168291091919, "learning_rate": 1.7380050884939237e-05, "loss": 3.0706, "step": 92480 }, { "epoch": 0.0068864, "grad_norm": 0.6778897047042847, "learning_rate": 1.737950746300419e-05, "loss": 3.1283, "step": 92490 }, { "epoch": 0.006912, "grad_norm": 0.7916220426559448, "learning_rate": 1.7378963993214997e-05, "loss": 3.1667, "step": 92500 }, { "epoch": 0.0069376, "grad_norm": 0.8420612215995789, "learning_rate": 1.7378420475575184e-05, "loss": 3.1153, "step": 92510 }, { "epoch": 0.0069632, "grad_norm": 0.9036852717399597, "learning_rate": 1.7377876910088275e-05, "loss": 2.9088, "step": 92520 }, { "epoch": 0.0069888, "grad_norm": 2.181473970413208, "learning_rate": 1.7377333296757797e-05, "loss": 3.0191, "step": 92530 }, { "epoch": 0.0070144, "grad_norm": 0.7101929783821106, "learning_rate": 1.737678963558727e-05, "loss": 3.3292, "step": 92540 }, { "epoch": 0.00704, "grad_norm": 0.6841883063316345, "learning_rate": 1.7376245926580225e-05, "loss": 2.9987, "step": 92550 }, { "epoch": 0.0070656, "grad_norm": 1.9503679275512695, "learning_rate": 1.7375702169740184e-05, "loss": 3.1089, "step": 92560 }, { "epoch": 0.0070912, "grad_norm": 0.9566271305084229, "learning_rate": 1.7375158365070676e-05, "loss": 3.3551, "step": 92570 }, { "epoch": 0.0071168, "grad_norm": 0.8574581742286682, "learning_rate": 1.7374614512575228e-05, "loss": 3.1613, "step": 92580 }, { "epoch": 0.0071424, "grad_norm": 0.7515103816986084, "learning_rate": 1.7374070612257357e-05, "loss": 3.2726, "step": 92590 }, { "epoch": 0.007168, "grad_norm": 1.1530166864395142, "learning_rate": 1.7373526664120607e-05, "loss": 3.1794, "step": 92600 }, { "epoch": 0.0071936, "grad_norm": 0.7535209655761719, "learning_rate": 1.737298266816849e-05, "loss": 3.298, "step": 92610 }, { "epoch": 0.0072192, "grad_norm": 0.7131289839744568, "learning_rate": 1.7372438624404547e-05, "loss": 3.3452, "step": 92620 }, { "epoch": 0.0072448, "grad_norm": 0.7967248558998108, "learning_rate": 1.7371894532832292e-05, "loss": 3.1582, "step": 92630 }, { "epoch": 0.0072704, "grad_norm": 0.8524859547615051, "learning_rate": 1.7371350393455265e-05, "loss": 3.2834, "step": 92640 }, { "epoch": 0.007296, "grad_norm": 0.7754468321800232, "learning_rate": 1.7370806206276984e-05, "loss": 3.0196, "step": 92650 }, { "epoch": 0.0073216, "grad_norm": 0.8672564029693604, "learning_rate": 1.7370261971300987e-05, "loss": 3.1751, "step": 92660 }, { "epoch": 0.0073472, "grad_norm": 0.7807691693305969, "learning_rate": 1.73697176885308e-05, "loss": 3.0958, "step": 92670 }, { "epoch": 0.0073728, "grad_norm": 0.8383861780166626, "learning_rate": 1.7369173357969953e-05, "loss": 3.1118, "step": 92680 }, { "epoch": 0.0073984, "grad_norm": 0.7664291858673096, "learning_rate": 1.7368628979621974e-05, "loss": 3.1843, "step": 92690 }, { "epoch": 0.007424, "grad_norm": 0.8200198411941528, "learning_rate": 1.7368084553490396e-05, "loss": 3.1643, "step": 92700 }, { "epoch": 0.0074496, "grad_norm": 0.8132080435752869, "learning_rate": 1.7367540079578748e-05, "loss": 3.3123, "step": 92710 }, { "epoch": 0.0074752, "grad_norm": 0.9171515703201294, "learning_rate": 1.7366995557890558e-05, "loss": 3.1817, "step": 92720 }, { "epoch": 0.0075008, "grad_norm": 0.9017375707626343, "learning_rate": 1.736645098842936e-05, "loss": 3.1318, "step": 92730 }, { "epoch": 0.0075264, "grad_norm": 0.7797361612319946, "learning_rate": 1.736590637119869e-05, "loss": 3.1443, "step": 92740 }, { "epoch": 0.007552, "grad_norm": 0.7496025562286377, "learning_rate": 1.736536170620207e-05, "loss": 3.1409, "step": 92750 }, { "epoch": 0.0075776, "grad_norm": 0.893040120601654, "learning_rate": 1.736481699344304e-05, "loss": 3.0513, "step": 92760 }, { "epoch": 0.0076032, "grad_norm": 0.8398213386535645, "learning_rate": 1.7364272232925125e-05, "loss": 3.5409, "step": 92770 }, { "epoch": 0.0076288, "grad_norm": 0.8217018246650696, "learning_rate": 1.7363727424651865e-05, "loss": 3.2297, "step": 92780 }, { "epoch": 0.0076544, "grad_norm": 0.8869420289993286, "learning_rate": 1.7363182568626788e-05, "loss": 3.1942, "step": 92790 }, { "epoch": 0.00768, "grad_norm": 0.7959392666816711, "learning_rate": 1.7362637664853426e-05, "loss": 3.338, "step": 92800 }, { "epoch": 0.0077056, "grad_norm": 0.7442058324813843, "learning_rate": 1.7362092713335325e-05, "loss": 3.0597, "step": 92810 }, { "epoch": 0.0077312, "grad_norm": 0.7618200182914734, "learning_rate": 1.7361547714076e-05, "loss": 3.2445, "step": 92820 }, { "epoch": 0.0077568, "grad_norm": 0.7352526783943176, "learning_rate": 1.7361002667079e-05, "loss": 3.0717, "step": 92830 }, { "epoch": 0.0077824, "grad_norm": 0.7970811128616333, "learning_rate": 1.736045757234785e-05, "loss": 3.237, "step": 92840 }, { "epoch": 0.007808, "grad_norm": 0.785243570804596, "learning_rate": 1.735991242988609e-05, "loss": 3.3357, "step": 92850 }, { "epoch": 0.0078336, "grad_norm": 0.8003327250480652, "learning_rate": 1.7359367239697254e-05, "loss": 3.2689, "step": 92860 }, { "epoch": 0.0078592, "grad_norm": 0.6988140344619751, "learning_rate": 1.735882200178488e-05, "loss": 3.0203, "step": 92870 }, { "epoch": 0.0078848, "grad_norm": 0.717951774597168, "learning_rate": 1.7358276716152497e-05, "loss": 3.2059, "step": 92880 }, { "epoch": 0.0079104, "grad_norm": 0.7257916927337646, "learning_rate": 1.735773138280365e-05, "loss": 3.1474, "step": 92890 }, { "epoch": 0.007936, "grad_norm": 0.6822704672813416, "learning_rate": 1.7357186001741864e-05, "loss": 3.1295, "step": 92900 }, { "epoch": 0.0079616, "grad_norm": 0.7649284601211548, "learning_rate": 1.735664057297069e-05, "loss": 3.2545, "step": 92910 }, { "epoch": 0.0079872, "grad_norm": 0.8345293402671814, "learning_rate": 1.735609509649365e-05, "loss": 3.1789, "step": 92920 }, { "epoch": 0.0080128, "grad_norm": 0.7174111008644104, "learning_rate": 1.7355549572314288e-05, "loss": 3.4863, "step": 92930 }, { "epoch": 0.0080384, "grad_norm": 0.7602567076683044, "learning_rate": 1.7355004000436147e-05, "loss": 3.1124, "step": 92940 }, { "epoch": 0.008064, "grad_norm": 0.7431374192237854, "learning_rate": 1.735445838086276e-05, "loss": 3.1028, "step": 92950 }, { "epoch": 0.0080896, "grad_norm": 0.8948839902877808, "learning_rate": 1.735391271359766e-05, "loss": 3.2522, "step": 92960 }, { "epoch": 0.0081152, "grad_norm": 0.8598973155021667, "learning_rate": 1.7353366998644394e-05, "loss": 3.1485, "step": 92970 }, { "epoch": 0.0081408, "grad_norm": 0.7540343403816223, "learning_rate": 1.7352821236006495e-05, "loss": 3.1825, "step": 92980 }, { "epoch": 0.0081664, "grad_norm": 0.771597683429718, "learning_rate": 1.735227542568751e-05, "loss": 3.2257, "step": 92990 }, { "epoch": 0.008192, "grad_norm": 0.8877357840538025, "learning_rate": 1.7351729567690966e-05, "loss": 3.2105, "step": 93000 }, { "epoch": 0.0082176, "grad_norm": 0.7740198969841003, "learning_rate": 1.7351183662020414e-05, "loss": 3.2495, "step": 93010 }, { "epoch": 0.0082432, "grad_norm": 0.9717979431152344, "learning_rate": 1.7350637708679386e-05, "loss": 3.2416, "step": 93020 }, { "epoch": 0.0082688, "grad_norm": 1.2776093482971191, "learning_rate": 1.735009170767143e-05, "loss": 3.1865, "step": 93030 }, { "epoch": 0.0082944, "grad_norm": 0.7794378995895386, "learning_rate": 1.734954565900008e-05, "loss": 3.108, "step": 93040 }, { "epoch": 0.00832, "grad_norm": 0.7164286375045776, "learning_rate": 1.734899956266888e-05, "loss": 3.1744, "step": 93050 }, { "epoch": 0.0083456, "grad_norm": 0.9095838069915771, "learning_rate": 1.7348453418681372e-05, "loss": 3.2121, "step": 93060 }, { "epoch": 0.0083712, "grad_norm": 0.8065826892852783, "learning_rate": 1.7347907227041096e-05, "loss": 3.1948, "step": 93070 }, { "epoch": 0.0083968, "grad_norm": 1.1951854228973389, "learning_rate": 1.7347360987751595e-05, "loss": 3.0187, "step": 93080 }, { "epoch": 0.0084224, "grad_norm": 0.7165895700454712, "learning_rate": 1.734681470081641e-05, "loss": 3.1534, "step": 93090 }, { "epoch": 0.008448, "grad_norm": 0.7144908308982849, "learning_rate": 1.7346268366239085e-05, "loss": 3.0224, "step": 93100 }, { "epoch": 0.0084736, "grad_norm": 0.8375822305679321, "learning_rate": 1.7345721984023162e-05, "loss": 3.1901, "step": 93110 }, { "epoch": 0.0084992, "grad_norm": 0.8203615546226501, "learning_rate": 1.7345175554172183e-05, "loss": 3.236, "step": 93120 }, { "epoch": 0.0085248, "grad_norm": 0.8112195134162903, "learning_rate": 1.734462907668969e-05, "loss": 3.1912, "step": 93130 }, { "epoch": 0.0085504, "grad_norm": 0.7107431888580322, "learning_rate": 1.7344082551579235e-05, "loss": 3.2679, "step": 93140 }, { "epoch": 0.008576, "grad_norm": 0.6806276440620422, "learning_rate": 1.7343535978844355e-05, "loss": 3.0426, "step": 93150 }, { "epoch": 0.0086016, "grad_norm": 0.8527563214302063, "learning_rate": 1.7342989358488592e-05, "loss": 3.2275, "step": 93160 }, { "epoch": 0.0086272, "grad_norm": 0.7674190998077393, "learning_rate": 1.73424426905155e-05, "loss": 3.1577, "step": 93170 }, { "epoch": 0.0086528, "grad_norm": 0.885173499584198, "learning_rate": 1.7341895974928617e-05, "loss": 3.0986, "step": 93180 }, { "epoch": 0.0086784, "grad_norm": 0.867941677570343, "learning_rate": 1.7341349211731488e-05, "loss": 3.0995, "step": 93190 }, { "epoch": 0.008704, "grad_norm": 0.6846035122871399, "learning_rate": 1.734080240092766e-05, "loss": 3.0932, "step": 93200 }, { "epoch": 0.0087296, "grad_norm": 0.7893560528755188, "learning_rate": 1.7340255542520684e-05, "loss": 3.1144, "step": 93210 }, { "epoch": 0.0087552, "grad_norm": 0.6956750750541687, "learning_rate": 1.7339708636514097e-05, "loss": 3.12, "step": 93220 }, { "epoch": 0.0087808, "grad_norm": 0.6460295915603638, "learning_rate": 1.733916168291145e-05, "loss": 3.1442, "step": 93230 }, { "epoch": 0.0088064, "grad_norm": 0.7726292610168457, "learning_rate": 1.7338614681716292e-05, "loss": 3.2166, "step": 93240 }, { "epoch": 0.008832, "grad_norm": 0.6763836145401001, "learning_rate": 1.733806763293217e-05, "loss": 3.2377, "step": 93250 }, { "epoch": 0.0088576, "grad_norm": 0.7244909405708313, "learning_rate": 1.7337520536562625e-05, "loss": 3.1921, "step": 93260 }, { "epoch": 0.0088832, "grad_norm": 0.7945707440376282, "learning_rate": 1.7336973392611212e-05, "loss": 3.1913, "step": 93270 }, { "epoch": 0.0089088, "grad_norm": 0.8890196681022644, "learning_rate": 1.7336426201081475e-05, "loss": 3.1293, "step": 93280 }, { "epoch": 0.0089344, "grad_norm": 1.1872272491455078, "learning_rate": 1.7335878961976966e-05, "loss": 3.1328, "step": 93290 }, { "epoch": 0.00896, "grad_norm": 0.67961186170578, "learning_rate": 1.7335331675301233e-05, "loss": 3.1117, "step": 93300 }, { "epoch": 0.0089856, "grad_norm": 0.8002622723579407, "learning_rate": 1.733478434105782e-05, "loss": 3.1312, "step": 93310 }, { "epoch": 0.0090112, "grad_norm": 0.8074969053268433, "learning_rate": 1.733423695925028e-05, "loss": 3.2366, "step": 93320 }, { "epoch": 0.0090368, "grad_norm": 0.8137736320495605, "learning_rate": 1.7333689529882165e-05, "loss": 3.0634, "step": 93330 }, { "epoch": 0.0090624, "grad_norm": 0.9006907939910889, "learning_rate": 1.7333142052957018e-05, "loss": 3.2952, "step": 93340 }, { "epoch": 0.009088, "grad_norm": 0.884063184261322, "learning_rate": 1.73325945284784e-05, "loss": 3.2694, "step": 93350 }, { "epoch": 0.0091136, "grad_norm": 0.7249205708503723, "learning_rate": 1.733204695644985e-05, "loss": 2.98, "step": 93360 }, { "epoch": 0.0091392, "grad_norm": 0.8475188612937927, "learning_rate": 1.733149933687493e-05, "loss": 3.0917, "step": 93370 }, { "epoch": 0.0091648, "grad_norm": 0.7747344970703125, "learning_rate": 1.7330951669757178e-05, "loss": 3.098, "step": 93380 }, { "epoch": 0.0091904, "grad_norm": 0.7888266444206238, "learning_rate": 1.7330403955100156e-05, "loss": 3.1838, "step": 93390 }, { "epoch": 0.009216, "grad_norm": 0.80126953125, "learning_rate": 1.732985619290741e-05, "loss": 3.4083, "step": 93400 }, { "epoch": 0.0092416, "grad_norm": 0.7715027332305908, "learning_rate": 1.73293083831825e-05, "loss": 3.2554, "step": 93410 }, { "epoch": 0.0092672, "grad_norm": 1.097644329071045, "learning_rate": 1.732876052592897e-05, "loss": 3.0226, "step": 93420 }, { "epoch": 0.0092928, "grad_norm": 0.7623871564865112, "learning_rate": 1.732821262115038e-05, "loss": 3.3243, "step": 93430 }, { "epoch": 0.0093184, "grad_norm": 2.8710997104644775, "learning_rate": 1.732766466885027e-05, "loss": 3.087, "step": 93440 }, { "epoch": 0.009344, "grad_norm": 1.3552544116973877, "learning_rate": 1.732711666903221e-05, "loss": 3.0867, "step": 93450 }, { "epoch": 0.0093696, "grad_norm": 0.705146849155426, "learning_rate": 1.732656862169974e-05, "loss": 3.1208, "step": 93460 }, { "epoch": 0.0093952, "grad_norm": 0.744170606136322, "learning_rate": 1.7326020526856426e-05, "loss": 3.1738, "step": 93470 }, { "epoch": 0.0094208, "grad_norm": 0.7981209754943848, "learning_rate": 1.732547238450581e-05, "loss": 3.0763, "step": 93480 }, { "epoch": 0.0094464, "grad_norm": 0.9818738698959351, "learning_rate": 1.7324924194651454e-05, "loss": 3.1964, "step": 93490 }, { "epoch": 0.009472, "grad_norm": 0.919427752494812, "learning_rate": 1.7324375957296916e-05, "loss": 3.1328, "step": 93500 }, { "epoch": 0.0094976, "grad_norm": 0.683251678943634, "learning_rate": 1.732382767244574e-05, "loss": 3.333, "step": 93510 }, { "epoch": 0.0095232, "grad_norm": 0.9416906237602234, "learning_rate": 1.7323279340101492e-05, "loss": 3.1174, "step": 93520 }, { "epoch": 0.0095488, "grad_norm": 0.7151750326156616, "learning_rate": 1.732273096026772e-05, "loss": 3.3099, "step": 93530 }, { "epoch": 0.0095744, "grad_norm": 0.7754715085029602, "learning_rate": 1.7322182532947986e-05, "loss": 3.1465, "step": 93540 }, { "epoch": 0.0096, "grad_norm": 0.9460291862487793, "learning_rate": 1.7321634058145847e-05, "loss": 3.2242, "step": 93550 }, { "epoch": 0.0096256, "grad_norm": 0.7209237217903137, "learning_rate": 1.7321085535864855e-05, "loss": 3.2814, "step": 93560 }, { "epoch": 0.0096512, "grad_norm": 0.9088671207427979, "learning_rate": 1.7320536966108568e-05, "loss": 3.1263, "step": 93570 }, { "epoch": 0.0096768, "grad_norm": 0.66240394115448, "learning_rate": 1.7319988348880542e-05, "loss": 3.1023, "step": 93580 }, { "epoch": 0.0097024, "grad_norm": 0.7080121636390686, "learning_rate": 1.7319439684184343e-05, "loss": 3.0927, "step": 93590 }, { "epoch": 0.009728, "grad_norm": 0.7551180124282837, "learning_rate": 1.7318890972023518e-05, "loss": 3.4195, "step": 93600 }, { "epoch": 0.0097536, "grad_norm": 0.919628918170929, "learning_rate": 1.7318342212401634e-05, "loss": 3.2903, "step": 93610 }, { "epoch": 0.0097792, "grad_norm": 0.8806402683258057, "learning_rate": 1.731779340532224e-05, "loss": 3.0479, "step": 93620 }, { "epoch": 0.0098048, "grad_norm": 0.8966565132141113, "learning_rate": 1.7317244550788904e-05, "loss": 3.1626, "step": 93630 }, { "epoch": 0.0098304, "grad_norm": 0.8659905195236206, "learning_rate": 1.7316695648805184e-05, "loss": 3.1814, "step": 93640 }, { "epoch": 0.009856, "grad_norm": 0.8857724070549011, "learning_rate": 1.7316146699374634e-05, "loss": 3.1983, "step": 93650 }, { "epoch": 0.0098816, "grad_norm": 0.8006410002708435, "learning_rate": 1.7315597702500818e-05, "loss": 3.1586, "step": 93660 }, { "epoch": 0.0099072, "grad_norm": 0.7953347563743591, "learning_rate": 1.7315048658187298e-05, "loss": 3.0983, "step": 93670 }, { "epoch": 0.0099328, "grad_norm": 0.8419412970542908, "learning_rate": 1.731449956643763e-05, "loss": 3.1793, "step": 93680 }, { "epoch": 0.0099584, "grad_norm": 0.7533960938453674, "learning_rate": 1.7313950427255372e-05, "loss": 3.3705, "step": 93690 }, { "epoch": 0.009984, "grad_norm": 1.1296100616455078, "learning_rate": 1.7313401240644095e-05, "loss": 3.3164, "step": 93700 }, { "epoch": 0.0100096, "grad_norm": 0.8908794522285461, "learning_rate": 1.7312852006607352e-05, "loss": 3.1478, "step": 93710 }, { "epoch": 0.0100352, "grad_norm": 0.8611295819282532, "learning_rate": 1.7312302725148707e-05, "loss": 3.2108, "step": 93720 }, { "epoch": 0.0100608, "grad_norm": 0.8126685619354248, "learning_rate": 1.731175339627172e-05, "loss": 3.2878, "step": 93730 }, { "epoch": 0.0100864, "grad_norm": 0.7090275287628174, "learning_rate": 1.731120401997996e-05, "loss": 3.1617, "step": 93740 }, { "epoch": 0.010112, "grad_norm": 0.6979836821556091, "learning_rate": 1.7310654596276982e-05, "loss": 3.0593, "step": 93750 }, { "epoch": 0.0101376, "grad_norm": 0.7021238207817078, "learning_rate": 1.7310105125166356e-05, "loss": 3.2005, "step": 93760 }, { "epoch": 0.0101632, "grad_norm": 0.7262365818023682, "learning_rate": 1.7309555606651636e-05, "loss": 3.1604, "step": 93770 }, { "epoch": 0.0101888, "grad_norm": 0.7142983675003052, "learning_rate": 1.7309006040736392e-05, "loss": 3.1776, "step": 93780 }, { "epoch": 0.0102144, "grad_norm": 0.7475740909576416, "learning_rate": 1.730845642742419e-05, "loss": 3.0897, "step": 93790 }, { "epoch": 0.01024, "grad_norm": 0.6917153596878052, "learning_rate": 1.7307906766718585e-05, "loss": 3.1276, "step": 93800 }, { "epoch": 0.0102656, "grad_norm": 0.6625188589096069, "learning_rate": 1.730735705862315e-05, "loss": 3.2564, "step": 93810 }, { "epoch": 0.0102912, "grad_norm": 0.7691434621810913, "learning_rate": 1.7306807303141444e-05, "loss": 3.1661, "step": 93820 }, { "epoch": 0.0103168, "grad_norm": 0.8007916808128357, "learning_rate": 1.730625750027704e-05, "loss": 3.3511, "step": 93830 }, { "epoch": 0.0103424, "grad_norm": 0.9312620162963867, "learning_rate": 1.730570765003349e-05, "loss": 3.2003, "step": 93840 }, { "epoch": 0.010368, "grad_norm": 0.727609395980835, "learning_rate": 1.7305157752414374e-05, "loss": 3.1871, "step": 93850 }, { "epoch": 0.0103936, "grad_norm": 1.2050755023956299, "learning_rate": 1.7304607807423247e-05, "loss": 3.5498, "step": 93860 }, { "epoch": 0.0104192, "grad_norm": 0.7992560267448425, "learning_rate": 1.7304057815063683e-05, "loss": 3.1626, "step": 93870 }, { "epoch": 0.0104448, "grad_norm": 0.7758188843727112, "learning_rate": 1.7303507775339246e-05, "loss": 3.1373, "step": 93880 }, { "epoch": 0.0104704, "grad_norm": 0.7799906730651855, "learning_rate": 1.73029576882535e-05, "loss": 3.1531, "step": 93890 }, { "epoch": 0.010496, "grad_norm": 0.7000567317008972, "learning_rate": 1.7302407553810014e-05, "loss": 2.9481, "step": 93900 }, { "epoch": 0.0105216, "grad_norm": 0.7432994246482849, "learning_rate": 1.7301857372012354e-05, "loss": 3.1049, "step": 93910 }, { "epoch": 0.0105472, "grad_norm": 0.7789033651351929, "learning_rate": 1.7301307142864093e-05, "loss": 3.1342, "step": 93920 }, { "epoch": 0.0105728, "grad_norm": 0.8582574129104614, "learning_rate": 1.7300756866368797e-05, "loss": 3.049, "step": 93930 }, { "epoch": 0.0105984, "grad_norm": 0.8011514544487, "learning_rate": 1.7300206542530028e-05, "loss": 2.9421, "step": 93940 }, { "epoch": 0.010624, "grad_norm": 1.3163092136383057, "learning_rate": 1.7299656171351363e-05, "loss": 3.0566, "step": 93950 }, { "epoch": 0.0106496, "grad_norm": 0.7608312368392944, "learning_rate": 1.7299105752836367e-05, "loss": 3.2002, "step": 93960 }, { "epoch": 0.0106752, "grad_norm": 0.763051450252533, "learning_rate": 1.729855528698861e-05, "loss": 3.1419, "step": 93970 }, { "epoch": 0.0107008, "grad_norm": 0.8267378807067871, "learning_rate": 1.7298004773811662e-05, "loss": 3.1075, "step": 93980 }, { "epoch": 0.0107264, "grad_norm": 0.703325092792511, "learning_rate": 1.7297454213309092e-05, "loss": 3.1717, "step": 93990 }, { "epoch": 0.010752, "grad_norm": 0.7563923001289368, "learning_rate": 1.729690360548447e-05, "loss": 3.1283, "step": 94000 }, { "epoch": 0.0107776, "grad_norm": 0.7755301594734192, "learning_rate": 1.729635295034137e-05, "loss": 2.9136, "step": 94010 }, { "epoch": 0.0108032, "grad_norm": 0.8489538431167603, "learning_rate": 1.729580224788336e-05, "loss": 3.1391, "step": 94020 }, { "epoch": 0.0108288, "grad_norm": 0.6773113012313843, "learning_rate": 1.729525149811401e-05, "loss": 3.1122, "step": 94030 }, { "epoch": 0.0108544, "grad_norm": 0.7332168817520142, "learning_rate": 1.7294700701036894e-05, "loss": 3.1806, "step": 94040 }, { "epoch": 0.01088, "grad_norm": 0.7801334261894226, "learning_rate": 1.729414985665558e-05, "loss": 3.1937, "step": 94050 }, { "epoch": 0.0109056, "grad_norm": 0.9884483814239502, "learning_rate": 1.7293598964973643e-05, "loss": 3.3174, "step": 94060 }, { "epoch": 0.0109312, "grad_norm": 0.784538745880127, "learning_rate": 1.7293048025994655e-05, "loss": 3.3117, "step": 94070 }, { "epoch": 0.0109568, "grad_norm": 0.746959388256073, "learning_rate": 1.7292497039722192e-05, "loss": 3.0645, "step": 94080 }, { "epoch": 0.0109824, "grad_norm": 0.8018479943275452, "learning_rate": 1.7291946006159822e-05, "loss": 3.3817, "step": 94090 }, { "epoch": 0.011008, "grad_norm": 0.8068310022354126, "learning_rate": 1.729139492531112e-05, "loss": 3.1299, "step": 94100 }, { "epoch": 0.0110336, "grad_norm": 0.8095101714134216, "learning_rate": 1.7290843797179663e-05, "loss": 2.9942, "step": 94110 }, { "epoch": 0.0110592, "grad_norm": 0.9009932279586792, "learning_rate": 1.729029262176902e-05, "loss": 3.3105, "step": 94120 }, { "epoch": 0.0110848, "grad_norm": 0.7384021282196045, "learning_rate": 1.728974139908276e-05, "loss": 3.1769, "step": 94130 }, { "epoch": 0.0111104, "grad_norm": 0.8918314576148987, "learning_rate": 1.7289190129124474e-05, "loss": 3.0917, "step": 94140 }, { "epoch": 0.011136, "grad_norm": 0.742692232131958, "learning_rate": 1.7288638811897722e-05, "loss": 3.0515, "step": 94150 }, { "epoch": 0.0111616, "grad_norm": 1.4048854112625122, "learning_rate": 1.728808744740609e-05, "loss": 3.3152, "step": 94160 }, { "epoch": 0.0111872, "grad_norm": 0.7688637971878052, "learning_rate": 1.728753603565314e-05, "loss": 3.3512, "step": 94170 }, { "epoch": 0.0112128, "grad_norm": 0.8839700818061829, "learning_rate": 1.7286984576642458e-05, "loss": 3.0937, "step": 94180 }, { "epoch": 0.0112384, "grad_norm": 0.6918076276779175, "learning_rate": 1.728643307037762e-05, "loss": 3.0926, "step": 94190 }, { "epoch": 0.011264, "grad_norm": 0.7294672131538391, "learning_rate": 1.72858815168622e-05, "loss": 3.3627, "step": 94200 }, { "epoch": 0.0112896, "grad_norm": 1.0227668285369873, "learning_rate": 1.7285329916099775e-05, "loss": 3.0936, "step": 94210 }, { "epoch": 0.0113152, "grad_norm": 0.6827560067176819, "learning_rate": 1.7284778268093914e-05, "loss": 3.4017, "step": 94220 }, { "epoch": 0.0113408, "grad_norm": 0.7517516613006592, "learning_rate": 1.728422657284821e-05, "loss": 3.0621, "step": 94230 }, { "epoch": 0.0113664, "grad_norm": 0.8729749321937561, "learning_rate": 1.7283674830366233e-05, "loss": 3.0645, "step": 94240 }, { "epoch": 0.011392, "grad_norm": 0.7074085474014282, "learning_rate": 1.7283123040651558e-05, "loss": 3.1249, "step": 94250 }, { "epoch": 0.0114176, "grad_norm": 0.764123797416687, "learning_rate": 1.7282571203707768e-05, "loss": 3.0699, "step": 94260 }, { "epoch": 0.0114432, "grad_norm": 0.8186538219451904, "learning_rate": 1.7282019319538436e-05, "loss": 3.0819, "step": 94270 }, { "epoch": 0.0114688, "grad_norm": 1.1537970304489136, "learning_rate": 1.7281467388147143e-05, "loss": 3.1018, "step": 94280 }, { "epoch": 0.0114944, "grad_norm": 0.8413915038108826, "learning_rate": 1.7280915409537472e-05, "loss": 3.2987, "step": 94290 }, { "epoch": 0.01152, "grad_norm": 0.7783328890800476, "learning_rate": 1.7280363383713e-05, "loss": 3.1807, "step": 94300 }, { "epoch": 0.0115456, "grad_norm": 0.7369843125343323, "learning_rate": 1.7279811310677306e-05, "loss": 3.2489, "step": 94310 }, { "epoch": 0.0115712, "grad_norm": 1.4441920518875122, "learning_rate": 1.727925919043397e-05, "loss": 3.0007, "step": 94320 }, { "epoch": 0.0115968, "grad_norm": 1.004643201828003, "learning_rate": 1.727870702298657e-05, "loss": 2.9914, "step": 94330 }, { "epoch": 0.0116224, "grad_norm": 0.8303868174552917, "learning_rate": 1.7278154808338694e-05, "loss": 3.2822, "step": 94340 }, { "epoch": 0.011648, "grad_norm": 0.7033947110176086, "learning_rate": 1.7277602546493915e-05, "loss": 3.4576, "step": 94350 }, { "epoch": 0.0116736, "grad_norm": 0.870834469795227, "learning_rate": 1.7277050237455818e-05, "loss": 3.6575, "step": 94360 }, { "epoch": 0.0116992, "grad_norm": 0.7163054347038269, "learning_rate": 1.7276497881227987e-05, "loss": 3.0425, "step": 94370 }, { "epoch": 0.0117248, "grad_norm": 0.7888188362121582, "learning_rate": 1.7275945477814e-05, "loss": 3.0723, "step": 94380 }, { "epoch": 0.0117504, "grad_norm": 0.6865720748901367, "learning_rate": 1.727539302721744e-05, "loss": 3.255, "step": 94390 }, { "epoch": 0.011776, "grad_norm": 0.7896755337715149, "learning_rate": 1.7274840529441886e-05, "loss": 3.1573, "step": 94400 }, { "epoch": 0.0118016, "grad_norm": 0.6789391040802002, "learning_rate": 1.727428798449093e-05, "loss": 3.3425, "step": 94410 }, { "epoch": 0.0118272, "grad_norm": 1.1562267541885376, "learning_rate": 1.7273735392368148e-05, "loss": 3.2849, "step": 94420 }, { "epoch": 0.0118528, "grad_norm": 0.8086495399475098, "learning_rate": 1.7273182753077125e-05, "loss": 3.2543, "step": 94430 }, { "epoch": 0.0118784, "grad_norm": 0.8051975965499878, "learning_rate": 1.7272630066621445e-05, "loss": 3.1535, "step": 94440 }, { "epoch": 0.011904, "grad_norm": 0.7712538838386536, "learning_rate": 1.7272077333004692e-05, "loss": 3.2017, "step": 94450 }, { "epoch": 0.0119296, "grad_norm": 0.7712885141372681, "learning_rate": 1.727152455223045e-05, "loss": 3.388, "step": 94460 }, { "epoch": 0.0119552, "grad_norm": 0.7092480659484863, "learning_rate": 1.72709717243023e-05, "loss": 3.0799, "step": 94470 }, { "epoch": 0.0119808, "grad_norm": 0.748698353767395, "learning_rate": 1.7270418849223838e-05, "loss": 3.1867, "step": 94480 }, { "epoch": 0.0120064, "grad_norm": 0.9249881505966187, "learning_rate": 1.7269865926998637e-05, "loss": 3.0601, "step": 94490 }, { "epoch": 0.012032, "grad_norm": 0.8731756806373596, "learning_rate": 1.726931295763029e-05, "loss": 3.0622, "step": 94500 }, { "epoch": 0.0120576, "grad_norm": 0.8335596919059753, "learning_rate": 1.7268759941122378e-05, "loss": 3.2145, "step": 94510 }, { "epoch": 0.0120832, "grad_norm": 0.8385361433029175, "learning_rate": 1.726820687747849e-05, "loss": 3.2416, "step": 94520 }, { "epoch": 0.0121088, "grad_norm": 1.45125150680542, "learning_rate": 1.726765376670221e-05, "loss": 3.1985, "step": 94530 }, { "epoch": 0.0121344, "grad_norm": 1.4356745481491089, "learning_rate": 1.726710060879713e-05, "loss": 3.2922, "step": 94540 }, { "epoch": 0.01216, "grad_norm": 0.7856946587562561, "learning_rate": 1.7266547403766837e-05, "loss": 3.2255, "step": 94550 }, { "epoch": 0.0121856, "grad_norm": 0.798582136631012, "learning_rate": 1.726599415161491e-05, "loss": 3.0125, "step": 94560 }, { "epoch": 0.0122112, "grad_norm": 0.7539711594581604, "learning_rate": 1.7265440852344947e-05, "loss": 3.2849, "step": 94570 }, { "epoch": 0.0122368, "grad_norm": 0.8404974341392517, "learning_rate": 1.7264887505960527e-05, "loss": 2.9842, "step": 94580 }, { "epoch": 0.0122624, "grad_norm": 0.6901054382324219, "learning_rate": 1.7264334112465243e-05, "loss": 2.9672, "step": 94590 }, { "epoch": 0.012288, "grad_norm": 0.931810736656189, "learning_rate": 1.7263780671862682e-05, "loss": 3.1914, "step": 94600 }, { "epoch": 0.0123136, "grad_norm": 0.6762802600860596, "learning_rate": 1.7263227184156436e-05, "loss": 3.3506, "step": 94610 }, { "epoch": 0.0123392, "grad_norm": 0.7990363836288452, "learning_rate": 1.726267364935009e-05, "loss": 3.2557, "step": 94620 }, { "epoch": 0.0123648, "grad_norm": 0.9243021607398987, "learning_rate": 1.7262120067447238e-05, "loss": 3.467, "step": 94630 }, { "epoch": 0.0123904, "grad_norm": 0.8405270576477051, "learning_rate": 1.7261566438451465e-05, "loss": 3.4657, "step": 94640 }, { "epoch": 0.012416, "grad_norm": 0.8549935817718506, "learning_rate": 1.7261012762366362e-05, "loss": 3.2607, "step": 94650 }, { "epoch": 0.0124416, "grad_norm": 0.8375645875930786, "learning_rate": 1.7260459039195528e-05, "loss": 3.3199, "step": 94660 }, { "epoch": 0.0124672, "grad_norm": 0.8611116409301758, "learning_rate": 1.725990526894254e-05, "loss": 3.4323, "step": 94670 }, { "epoch": 0.0124928, "grad_norm": 0.7212010025978088, "learning_rate": 1.7259351451610997e-05, "loss": 3.1282, "step": 94680 }, { "epoch": 0.0125184, "grad_norm": 0.7147647738456726, "learning_rate": 1.7258797587204493e-05, "loss": 3.0403, "step": 94690 }, { "epoch": 0.012544, "grad_norm": 0.732096791267395, "learning_rate": 1.725824367572661e-05, "loss": 3.2764, "step": 94700 }, { "epoch": 0.0125696, "grad_norm": 0.7465362548828125, "learning_rate": 1.725768971718095e-05, "loss": 3.0791, "step": 94710 }, { "epoch": 0.0125952, "grad_norm": 0.8495108485221863, "learning_rate": 1.7257135711571102e-05, "loss": 3.1745, "step": 94720 }, { "epoch": 0.0126208, "grad_norm": 0.7798982858657837, "learning_rate": 1.725658165890065e-05, "loss": 3.2168, "step": 94730 }, { "epoch": 0.0126464, "grad_norm": 0.9057630300521851, "learning_rate": 1.7256027559173204e-05, "loss": 3.1551, "step": 94740 }, { "epoch": 0.012672, "grad_norm": 0.8518266081809998, "learning_rate": 1.725547341239234e-05, "loss": 3.2097, "step": 94750 }, { "epoch": 0.0126976, "grad_norm": 0.7248418927192688, "learning_rate": 1.7254919218561665e-05, "loss": 3.2676, "step": 94760 }, { "epoch": 0.0127232, "grad_norm": 0.7301065921783447, "learning_rate": 1.7254364977684764e-05, "loss": 3.1749, "step": 94770 }, { "epoch": 0.0127488, "grad_norm": 0.7217004895210266, "learning_rate": 1.7253810689765236e-05, "loss": 3.0865, "step": 94780 }, { "epoch": 0.0127744, "grad_norm": 0.7888465523719788, "learning_rate": 1.725325635480667e-05, "loss": 3.1824, "step": 94790 }, { "epoch": 0.0128, "grad_norm": 1.193306803703308, "learning_rate": 1.725270197281267e-05, "loss": 3.121, "step": 94800 }, { "epoch": 0.0128256, "grad_norm": 2.187653064727783, "learning_rate": 1.725214754378682e-05, "loss": 2.9556, "step": 94810 }, { "epoch": 0.0128512, "grad_norm": 0.6434693336486816, "learning_rate": 1.7251593067732722e-05, "loss": 3.3033, "step": 94820 }, { "epoch": 0.0128768, "grad_norm": 0.6497882604598999, "learning_rate": 1.725103854465397e-05, "loss": 3.2964, "step": 94830 }, { "epoch": 0.0129024, "grad_norm": 0.7064864635467529, "learning_rate": 1.725048397455416e-05, "loss": 3.1089, "step": 94840 }, { "epoch": 0.012928, "grad_norm": 0.954188346862793, "learning_rate": 1.724992935743689e-05, "loss": 3.4, "step": 94850 }, { "epoch": 0.0129536, "grad_norm": 0.7807631492614746, "learning_rate": 1.7249374693305754e-05, "loss": 3.072, "step": 94860 }, { "epoch": 0.0129792, "grad_norm": 0.8065080642700195, "learning_rate": 1.724881998216435e-05, "loss": 3.3216, "step": 94870 }, { "epoch": 0.0130048, "grad_norm": 2.1420183181762695, "learning_rate": 1.7248265224016277e-05, "loss": 3.4038, "step": 94880 }, { "epoch": 0.0130304, "grad_norm": 0.7357505559921265, "learning_rate": 1.7247710418865126e-05, "loss": 3.131, "step": 94890 }, { "epoch": 0.013056, "grad_norm": 0.8572623133659363, "learning_rate": 1.7247155566714502e-05, "loss": 3.2915, "step": 94900 }, { "epoch": 0.0130816, "grad_norm": 0.7568173408508301, "learning_rate": 1.7246600667568e-05, "loss": 3.0157, "step": 94910 }, { "epoch": 0.0131072, "grad_norm": 1.2116155624389648, "learning_rate": 1.724604572142922e-05, "loss": 3.3571, "step": 94920 }, { "epoch": 0.0131328, "grad_norm": 0.8930266499519348, "learning_rate": 1.7245490728301756e-05, "loss": 3.1962, "step": 94930 }, { "epoch": 0.0131584, "grad_norm": 0.8193693161010742, "learning_rate": 1.724493568818921e-05, "loss": 3.621, "step": 94940 }, { "epoch": 0.013184, "grad_norm": 0.6591323614120483, "learning_rate": 1.7244380601095185e-05, "loss": 3.2917, "step": 94950 }, { "epoch": 0.0132096, "grad_norm": 0.7299767732620239, "learning_rate": 1.7243825467023275e-05, "loss": 2.8242, "step": 94960 }, { "epoch": 0.0132352, "grad_norm": 0.8286871910095215, "learning_rate": 1.724327028597708e-05, "loss": 3.1983, "step": 94970 }, { "epoch": 0.0132608, "grad_norm": 0.7015586495399475, "learning_rate": 1.7242715057960206e-05, "loss": 2.9901, "step": 94980 }, { "epoch": 0.0132864, "grad_norm": 0.7407956719398499, "learning_rate": 1.724215978297625e-05, "loss": 3.1484, "step": 94990 }, { "epoch": 0.013312, "grad_norm": 0.6844242811203003, "learning_rate": 1.724160446102881e-05, "loss": 3.268, "step": 95000 }, { "epoch": 0.0133376, "grad_norm": 0.9506523013114929, "learning_rate": 1.7241049092121488e-05, "loss": 3.156, "step": 95010 }, { "epoch": 0.0133632, "grad_norm": 0.7385103702545166, "learning_rate": 1.724049367625789e-05, "loss": 3.3043, "step": 95020 }, { "epoch": 0.0133888, "grad_norm": 0.9426880478858948, "learning_rate": 1.7239938213441615e-05, "loss": 3.238, "step": 95030 }, { "epoch": 0.0134144, "grad_norm": 0.9169508814811707, "learning_rate": 1.7239382703676264e-05, "loss": 3.2177, "step": 95040 }, { "epoch": 0.01344, "grad_norm": 0.9289965629577637, "learning_rate": 1.723882714696544e-05, "loss": 3.5506, "step": 95050 }, { "epoch": 0.0134656, "grad_norm": 0.7538127899169922, "learning_rate": 1.7238271543312744e-05, "loss": 3.1615, "step": 95060 }, { "epoch": 0.0134912, "grad_norm": 0.9248297214508057, "learning_rate": 1.7237715892721786e-05, "loss": 3.1253, "step": 95070 }, { "epoch": 0.0135168, "grad_norm": 0.8497112989425659, "learning_rate": 1.723716019519616e-05, "loss": 3.2309, "step": 95080 }, { "epoch": 0.0135424, "grad_norm": 0.8411637544631958, "learning_rate": 1.7236604450739473e-05, "loss": 3.1661, "step": 95090 }, { "epoch": 0.013568, "grad_norm": 0.8627228140830994, "learning_rate": 1.7236048659355332e-05, "loss": 3.1818, "step": 95100 }, { "epoch": 0.0135936, "grad_norm": 1.131909728050232, "learning_rate": 1.7235492821047338e-05, "loss": 3.1995, "step": 95110 }, { "epoch": 0.0136192, "grad_norm": 1.7580440044403076, "learning_rate": 1.723493693581909e-05, "loss": 3.206, "step": 95120 }, { "epoch": 0.0136448, "grad_norm": 0.6883326768875122, "learning_rate": 1.723438100367421e-05, "loss": 3.4409, "step": 95130 }, { "epoch": 0.0136704, "grad_norm": 0.6860763430595398, "learning_rate": 1.7233825024616284e-05, "loss": 3.2765, "step": 95140 }, { "epoch": 0.013696, "grad_norm": 0.6568614840507507, "learning_rate": 1.7233268998648927e-05, "loss": 3.2891, "step": 95150 }, { "epoch": 0.0137216, "grad_norm": 0.7644044756889343, "learning_rate": 1.7232712925775744e-05, "loss": 3.1795, "step": 95160 }, { "epoch": 0.0137472, "grad_norm": 0.9449668526649475, "learning_rate": 1.723215680600034e-05, "loss": 3.305, "step": 95170 }, { "epoch": 0.0137728, "grad_norm": 0.7216495275497437, "learning_rate": 1.723160063932632e-05, "loss": 3.2518, "step": 95180 }, { "epoch": 0.0137984, "grad_norm": 0.7427451610565186, "learning_rate": 1.723104442575729e-05, "loss": 2.9467, "step": 95190 }, { "epoch": 0.013824, "grad_norm": 0.708303689956665, "learning_rate": 1.723048816529686e-05, "loss": 3.1763, "step": 95200 }, { "epoch": 0.0138496, "grad_norm": 0.7622811794281006, "learning_rate": 1.7229931857948638e-05, "loss": 3.2139, "step": 95210 }, { "epoch": 0.0138752, "grad_norm": 0.7457553744316101, "learning_rate": 1.722937550371623e-05, "loss": 3.1712, "step": 95220 }, { "epoch": 0.0139008, "grad_norm": 1.6310850381851196, "learning_rate": 1.7228819102603238e-05, "loss": 3.3653, "step": 95230 }, { "epoch": 0.0139264, "grad_norm": 0.7987040281295776, "learning_rate": 1.722826265461328e-05, "loss": 3.1367, "step": 95240 }, { "epoch": 0.013952, "grad_norm": 0.7777169942855835, "learning_rate": 1.7227706159749957e-05, "loss": 3.266, "step": 95250 }, { "epoch": 0.0139776, "grad_norm": 0.9913284778594971, "learning_rate": 1.7227149618016882e-05, "loss": 3.2301, "step": 95260 }, { "epoch": 0.0140032, "grad_norm": 0.7866868376731873, "learning_rate": 1.722659302941766e-05, "loss": 3.1979, "step": 95270 }, { "epoch": 0.0140288, "grad_norm": 0.7397075891494751, "learning_rate": 1.7226036393955907e-05, "loss": 3.2116, "step": 95280 }, { "epoch": 0.0140544, "grad_norm": 0.8114287853240967, "learning_rate": 1.7225479711635228e-05, "loss": 3.0231, "step": 95290 }, { "epoch": 0.01408, "grad_norm": 0.691363513469696, "learning_rate": 1.722492298245923e-05, "loss": 3.0929, "step": 95300 }, { "epoch": 0.0141056, "grad_norm": 0.8196120262145996, "learning_rate": 1.722436620643153e-05, "loss": 3.0124, "step": 95310 }, { "epoch": 0.0141312, "grad_norm": 0.6820077896118164, "learning_rate": 1.7223809383555733e-05, "loss": 3.211, "step": 95320 }, { "epoch": 0.0141568, "grad_norm": 0.6715770363807678, "learning_rate": 1.722325251383545e-05, "loss": 3.3109, "step": 95330 }, { "epoch": 0.0141824, "grad_norm": 0.8329690098762512, "learning_rate": 1.72226955972743e-05, "loss": 3.0953, "step": 95340 }, { "epoch": 0.014208, "grad_norm": 0.6477508544921875, "learning_rate": 1.7222138633875883e-05, "loss": 3.0794, "step": 95350 }, { "epoch": 0.0142336, "grad_norm": 0.7623692154884338, "learning_rate": 1.722158162364382e-05, "loss": 3.4624, "step": 95360 }, { "epoch": 0.0142592, "grad_norm": 0.7367314100265503, "learning_rate": 1.7221024566581716e-05, "loss": 3.1476, "step": 95370 }, { "epoch": 0.0142848, "grad_norm": 0.8677640557289124, "learning_rate": 1.7220467462693193e-05, "loss": 3.0744, "step": 95380 }, { "epoch": 0.0143104, "grad_norm": 0.7093092203140259, "learning_rate": 1.721991031198185e-05, "loss": 3.233, "step": 95390 }, { "epoch": 0.014336, "grad_norm": 0.625702440738678, "learning_rate": 1.721935311445131e-05, "loss": 3.0317, "step": 95400 }, { "epoch": 0.0143616, "grad_norm": 1.3565305471420288, "learning_rate": 1.7218795870105184e-05, "loss": 3.1294, "step": 95410 }, { "epoch": 0.0143872, "grad_norm": 0.74025958776474, "learning_rate": 1.721823857894709e-05, "loss": 3.172, "step": 95420 }, { "epoch": 0.0144128, "grad_norm": 0.7332914471626282, "learning_rate": 1.721768124098063e-05, "loss": 3.1637, "step": 95430 }, { "epoch": 0.0144384, "grad_norm": 0.8856309056282043, "learning_rate": 1.721712385620943e-05, "loss": 3.079, "step": 95440 }, { "epoch": 0.014464, "grad_norm": 0.6896685361862183, "learning_rate": 1.72165664246371e-05, "loss": 2.9873, "step": 95450 }, { "epoch": 0.0144896, "grad_norm": 0.693795382976532, "learning_rate": 1.7216008946267253e-05, "loss": 3.1398, "step": 95460 }, { "epoch": 0.0145152, "grad_norm": 0.7156041860580444, "learning_rate": 1.72154514211035e-05, "loss": 3.1012, "step": 95470 }, { "epoch": 0.0145408, "grad_norm": 0.9229040145874023, "learning_rate": 1.7214893849149474e-05, "loss": 3.0119, "step": 95480 }, { "epoch": 0.0145664, "grad_norm": 0.8305131793022156, "learning_rate": 1.721433623040877e-05, "loss": 3.1177, "step": 95490 }, { "epoch": 0.014592, "grad_norm": 0.7713622450828552, "learning_rate": 1.7213778564885016e-05, "loss": 3.0226, "step": 95500 }, { "epoch": 0.0146176, "grad_norm": 0.680813729763031, "learning_rate": 1.7213220852581824e-05, "loss": 3.2765, "step": 95510 }, { "epoch": 0.0146432, "grad_norm": 0.693114697933197, "learning_rate": 1.7212663093502813e-05, "loss": 2.9412, "step": 95520 }, { "epoch": 0.0146688, "grad_norm": 0.6892583966255188, "learning_rate": 1.72121052876516e-05, "loss": 3.1554, "step": 95530 }, { "epoch": 0.0146944, "grad_norm": 0.9698768258094788, "learning_rate": 1.72115474350318e-05, "loss": 3.1546, "step": 95540 }, { "epoch": 0.01472, "grad_norm": 0.6590858697891235, "learning_rate": 1.721098953564703e-05, "loss": 3.0345, "step": 95550 }, { "epoch": 0.0147456, "grad_norm": 0.7860508561134338, "learning_rate": 1.721043158950091e-05, "loss": 3.2625, "step": 95560 }, { "epoch": 0.0147712, "grad_norm": 0.900047779083252, "learning_rate": 1.7209873596597057e-05, "loss": 3.1499, "step": 95570 }, { "epoch": 0.0147968, "grad_norm": 0.8233383297920227, "learning_rate": 1.720931555693909e-05, "loss": 3.2753, "step": 95580 }, { "epoch": 0.0148224, "grad_norm": 0.6702061295509338, "learning_rate": 1.7208757470530627e-05, "loss": 3.2103, "step": 95590 }, { "epoch": 0.014848, "grad_norm": 0.6887328624725342, "learning_rate": 1.720819933737529e-05, "loss": 3.1003, "step": 95600 }, { "epoch": 0.0148736, "grad_norm": 0.7687342762947083, "learning_rate": 1.7207641157476693e-05, "loss": 3.2013, "step": 95610 }, { "epoch": 0.0148992, "grad_norm": 0.8285666704177856, "learning_rate": 1.7207082930838462e-05, "loss": 3.3461, "step": 95620 }, { "epoch": 0.0149248, "grad_norm": 0.6997067928314209, "learning_rate": 1.7206524657464213e-05, "loss": 3.202, "step": 95630 }, { "epoch": 0.0149504, "grad_norm": 1.2128957509994507, "learning_rate": 1.7205966337357564e-05, "loss": 3.1327, "step": 95640 }, { "epoch": 0.014976, "grad_norm": 0.9527180790901184, "learning_rate": 1.720540797052214e-05, "loss": 2.9896, "step": 95650 }, { "epoch": 0.0150016, "grad_norm": 0.8125967979431152, "learning_rate": 1.720484955696156e-05, "loss": 3.3119, "step": 95660 }, { "epoch": 0.0150272, "grad_norm": 0.838928759098053, "learning_rate": 1.720429109667945e-05, "loss": 3.3057, "step": 95670 }, { "epoch": 0.0150528, "grad_norm": 0.7665916681289673, "learning_rate": 1.7203732589679422e-05, "loss": 3.1819, "step": 95680 }, { "epoch": 0.0150784, "grad_norm": 0.8938239812850952, "learning_rate": 1.72031740359651e-05, "loss": 3.1378, "step": 95690 }, { "epoch": 0.015104, "grad_norm": 0.7104379534721375, "learning_rate": 1.7202615435540114e-05, "loss": 3.0957, "step": 95700 }, { "epoch": 0.0151296, "grad_norm": 0.7665375471115112, "learning_rate": 1.720205678840808e-05, "loss": 3.0798, "step": 95710 }, { "epoch": 0.0151552, "grad_norm": 0.7328991889953613, "learning_rate": 1.720149809457262e-05, "loss": 3.3837, "step": 95720 }, { "epoch": 0.0151808, "grad_norm": 0.8374317288398743, "learning_rate": 1.7200939354037363e-05, "loss": 3.1833, "step": 95730 }, { "epoch": 0.0152064, "grad_norm": 0.9430124163627625, "learning_rate": 1.7200380566805924e-05, "loss": 3.0472, "step": 95740 }, { "epoch": 0.015232, "grad_norm": 0.8248913288116455, "learning_rate": 1.719982173288193e-05, "loss": 2.8231, "step": 95750 }, { "epoch": 0.0152576, "grad_norm": 0.7155736088752747, "learning_rate": 1.719926285226901e-05, "loss": 3.2868, "step": 95760 }, { "epoch": 0.0152832, "grad_norm": 0.745661199092865, "learning_rate": 1.719870392497078e-05, "loss": 3.1228, "step": 95770 }, { "epoch": 0.0153088, "grad_norm": 0.792977511882782, "learning_rate": 1.719814495099087e-05, "loss": 3.0929, "step": 95780 }, { "epoch": 0.0153344, "grad_norm": 0.7526711821556091, "learning_rate": 1.7197585930332903e-05, "loss": 3.2967, "step": 95790 }, { "epoch": 0.01536, "grad_norm": 0.9004913568496704, "learning_rate": 1.71970268630005e-05, "loss": 3.4438, "step": 95800 }, { "epoch": 0.0153856, "grad_norm": 0.8162134289741516, "learning_rate": 1.7196467748997297e-05, "loss": 3.201, "step": 95810 }, { "epoch": 0.0154112, "grad_norm": 0.7247445583343506, "learning_rate": 1.719590858832691e-05, "loss": 3.0998, "step": 95820 }, { "epoch": 0.0154368, "grad_norm": 0.7030186057090759, "learning_rate": 1.719534938099297e-05, "loss": 3.3029, "step": 95830 }, { "epoch": 0.0154624, "grad_norm": 0.7946361303329468, "learning_rate": 1.7194790126999098e-05, "loss": 3.0407, "step": 95840 }, { "epoch": 0.015488, "grad_norm": 1.419547200202942, "learning_rate": 1.7194230826348925e-05, "loss": 2.8757, "step": 95850 }, { "epoch": 0.0155136, "grad_norm": 0.8519739508628845, "learning_rate": 1.719367147904608e-05, "loss": 3.2913, "step": 95860 }, { "epoch": 0.0155392, "grad_norm": 1.2930725812911987, "learning_rate": 1.7193112085094186e-05, "loss": 3.1581, "step": 95870 }, { "epoch": 0.0155648, "grad_norm": 0.8246579766273499, "learning_rate": 1.719255264449687e-05, "loss": 3.3913, "step": 95880 }, { "epoch": 0.0155904, "grad_norm": 0.6826273798942566, "learning_rate": 1.7191993157257764e-05, "loss": 3.0102, "step": 95890 }, { "epoch": 0.015616, "grad_norm": 0.6844046115875244, "learning_rate": 1.7191433623380494e-05, "loss": 3.1135, "step": 95900 }, { "epoch": 2.56e-05, "grad_norm": 0.8188198804855347, "learning_rate": 1.7190874042868687e-05, "loss": 2.695, "step": 95910 }, { "epoch": 5.12e-05, "grad_norm": 0.7320538759231567, "learning_rate": 1.719031441572597e-05, "loss": 3.1462, "step": 95920 }, { "epoch": 7.68e-05, "grad_norm": 3.2638702392578125, "learning_rate": 1.718975474195598e-05, "loss": 3.2035, "step": 95930 }, { "epoch": 0.0001024, "grad_norm": 0.6552111506462097, "learning_rate": 1.718919502156234e-05, "loss": 2.9323, "step": 95940 }, { "epoch": 0.000128, "grad_norm": 0.6845597624778748, "learning_rate": 1.718863525454868e-05, "loss": 3.069, "step": 95950 }, { "epoch": 0.0001536, "grad_norm": 0.7363821864128113, "learning_rate": 1.718807544091863e-05, "loss": 2.8673, "step": 95960 }, { "epoch": 0.0001792, "grad_norm": 0.8317474126815796, "learning_rate": 1.7187515580675823e-05, "loss": 3.1348, "step": 95970 }, { "epoch": 0.0002048, "grad_norm": 0.851207435131073, "learning_rate": 1.7186955673823885e-05, "loss": 3.0746, "step": 95980 }, { "epoch": 0.0002304, "grad_norm": 0.768208920955658, "learning_rate": 1.7186395720366453e-05, "loss": 3.1672, "step": 95990 }, { "epoch": 0.000256, "grad_norm": 0.7145197987556458, "learning_rate": 1.7185835720307155e-05, "loss": 3.0088, "step": 96000 }, { "epoch": 0.0002816, "grad_norm": 0.6304397583007812, "learning_rate": 1.7185275673649617e-05, "loss": 2.6682, "step": 96010 }, { "epoch": 0.0003072, "grad_norm": 0.7340236902236938, "learning_rate": 1.7184715580397478e-05, "loss": 3.295, "step": 96020 }, { "epoch": 0.0003328, "grad_norm": 0.8057198524475098, "learning_rate": 1.718415544055437e-05, "loss": 3.2135, "step": 96030 }, { "epoch": 0.0003584, "grad_norm": 0.7491086721420288, "learning_rate": 1.7183595254123924e-05, "loss": 3.0469, "step": 96040 }, { "epoch": 0.000384, "grad_norm": 0.7271429300308228, "learning_rate": 1.718303502110977e-05, "loss": 2.9882, "step": 96050 }, { "epoch": 0.0004096, "grad_norm": 0.7983660101890564, "learning_rate": 1.7182474741515545e-05, "loss": 3.1166, "step": 96060 }, { "epoch": 0.0004352, "grad_norm": 0.7246465682983398, "learning_rate": 1.718191441534488e-05, "loss": 2.9824, "step": 96070 }, { "epoch": 0.0004608, "grad_norm": 0.7306522130966187, "learning_rate": 1.7181354042601402e-05, "loss": 3.3121, "step": 96080 }, { "epoch": 0.0004864, "grad_norm": 0.6292621493339539, "learning_rate": 1.7180793623288758e-05, "loss": 2.8381, "step": 96090 }, { "epoch": 0.000512, "grad_norm": 0.6964051127433777, "learning_rate": 1.7180233157410576e-05, "loss": 3.0345, "step": 96100 }, { "epoch": 0.0005376, "grad_norm": 0.925684928894043, "learning_rate": 1.7179672644970487e-05, "loss": 3.1178, "step": 96110 }, { "epoch": 0.0005632, "grad_norm": 0.6349377036094666, "learning_rate": 1.7179112085972132e-05, "loss": 2.8899, "step": 96120 }, { "epoch": 0.0005888, "grad_norm": 0.6984460949897766, "learning_rate": 1.7178551480419142e-05, "loss": 3.0125, "step": 96130 }, { "epoch": 0.0006144, "grad_norm": 0.8231548070907593, "learning_rate": 1.7177990828315155e-05, "loss": 2.9901, "step": 96140 }, { "epoch": 0.00064, "grad_norm": 0.7652457356452942, "learning_rate": 1.71774301296638e-05, "loss": 3.2726, "step": 96150 }, { "epoch": 0.0006656, "grad_norm": 0.7963167428970337, "learning_rate": 1.7176869384468722e-05, "loss": 3.0136, "step": 96160 }, { "epoch": 0.0006912, "grad_norm": 0.6841186285018921, "learning_rate": 1.7176308592733553e-05, "loss": 2.9323, "step": 96170 }, { "epoch": 0.0007168, "grad_norm": 0.7008190751075745, "learning_rate": 1.717574775446193e-05, "loss": 3.1124, "step": 96180 }, { "epoch": 0.0007424, "grad_norm": 0.7475179433822632, "learning_rate": 1.7175186869657488e-05, "loss": 2.8483, "step": 96190 }, { "epoch": 0.000768, "grad_norm": 0.802000880241394, "learning_rate": 1.7174625938323873e-05, "loss": 2.9799, "step": 96200 }, { "epoch": 0.0007936, "grad_norm": 0.6639800071716309, "learning_rate": 1.717406496046471e-05, "loss": 3.0429, "step": 96210 }, { "epoch": 0.0008192, "grad_norm": 0.7509283423423767, "learning_rate": 1.717350393608364e-05, "loss": 2.8854, "step": 96220 }, { "epoch": 0.0008448, "grad_norm": 0.8900246620178223, "learning_rate": 1.717294286518431e-05, "loss": 2.962, "step": 96230 }, { "epoch": 0.0008704, "grad_norm": 0.696662962436676, "learning_rate": 1.717238174777035e-05, "loss": 2.8931, "step": 96240 }, { "epoch": 0.000896, "grad_norm": 1.046254277229309, "learning_rate": 1.71718205838454e-05, "loss": 3.1726, "step": 96250 }, { "epoch": 0.0009216, "grad_norm": 0.7567388415336609, "learning_rate": 1.7171259373413093e-05, "loss": 3.0229, "step": 96260 }, { "epoch": 0.0009472, "grad_norm": 0.6836420297622681, "learning_rate": 1.7170698116477082e-05, "loss": 3.0772, "step": 96270 }, { "epoch": 0.0009728, "grad_norm": 0.7773184180259705, "learning_rate": 1.7170136813040995e-05, "loss": 3.0145, "step": 96280 }, { "epoch": 0.0009984, "grad_norm": 0.9224005937576294, "learning_rate": 1.716957546310848e-05, "loss": 2.995, "step": 96290 }, { "epoch": 0.001024, "grad_norm": 0.7159631252288818, "learning_rate": 1.7169014066683173e-05, "loss": 3.1597, "step": 96300 }, { "epoch": 0.0010496, "grad_norm": 0.6975980401039124, "learning_rate": 1.7168452623768715e-05, "loss": 2.9552, "step": 96310 }, { "epoch": 0.0010752, "grad_norm": 0.8468323945999146, "learning_rate": 1.7167891134368747e-05, "loss": 3.2562, "step": 96320 }, { "epoch": 0.0011008, "grad_norm": 0.8323689103126526, "learning_rate": 1.716732959848691e-05, "loss": 3.4021, "step": 96330 }, { "epoch": 0.0011264, "grad_norm": 0.7749467492103577, "learning_rate": 1.7166768016126845e-05, "loss": 3.0326, "step": 96340 }, { "epoch": 0.001152, "grad_norm": 0.6793791055679321, "learning_rate": 1.7166206387292196e-05, "loss": 3.0373, "step": 96350 }, { "epoch": 0.0011776, "grad_norm": 0.7704700827598572, "learning_rate": 1.71656447119866e-05, "loss": 3.095, "step": 96360 }, { "epoch": 0.0012032, "grad_norm": 0.7893907427787781, "learning_rate": 1.7165082990213703e-05, "loss": 3.0193, "step": 96370 }, { "epoch": 0.0012288, "grad_norm": 0.6381842494010925, "learning_rate": 1.716452122197715e-05, "loss": 2.9244, "step": 96380 }, { "epoch": 0.0012544, "grad_norm": 3.3445897102355957, "learning_rate": 1.7163959407280578e-05, "loss": 3.1094, "step": 96390 }, { "epoch": 0.00128, "grad_norm": 0.7760923504829407, "learning_rate": 1.7163397546127634e-05, "loss": 3.0622, "step": 96400 }, { "epoch": 0.0013056, "grad_norm": 0.6342892646789551, "learning_rate": 1.716283563852196e-05, "loss": 2.913, "step": 96410 }, { "epoch": 0.0013312, "grad_norm": 1.001412034034729, "learning_rate": 1.7162273684467202e-05, "loss": 3.4304, "step": 96420 }, { "epoch": 0.0013568, "grad_norm": 0.6946651339530945, "learning_rate": 1.7161711683967e-05, "loss": 2.9586, "step": 96430 }, { "epoch": 0.0013824, "grad_norm": 1.1065317392349243, "learning_rate": 1.7161149637025004e-05, "loss": 3.0224, "step": 96440 }, { "epoch": 0.001408, "grad_norm": 0.7614277005195618, "learning_rate": 1.7160587543644855e-05, "loss": 3.158, "step": 96450 }, { "epoch": 0.0014336, "grad_norm": 0.6259116530418396, "learning_rate": 1.7160025403830198e-05, "loss": 2.772, "step": 96460 }, { "epoch": 0.0014592, "grad_norm": 0.6977232098579407, "learning_rate": 1.715946321758468e-05, "loss": 2.961, "step": 96470 }, { "epoch": 0.0014848, "grad_norm": 0.7467309236526489, "learning_rate": 1.7158900984911947e-05, "loss": 3.0948, "step": 96480 }, { "epoch": 0.0015104, "grad_norm": 0.6925081014633179, "learning_rate": 1.715833870581564e-05, "loss": 2.9612, "step": 96490 }, { "epoch": 0.001536, "grad_norm": 0.8527274131774902, "learning_rate": 1.715777638029941e-05, "loss": 3.1334, "step": 96500 }, { "epoch": 0.0015616, "grad_norm": 0.6565311551094055, "learning_rate": 1.7157214008366904e-05, "loss": 2.9722, "step": 96510 }, { "epoch": 0.0015872, "grad_norm": 0.7433024644851685, "learning_rate": 1.715665159002177e-05, "loss": 3.0604, "step": 96520 }, { "epoch": 0.0016128, "grad_norm": 0.9354746341705322, "learning_rate": 1.7156089125267648e-05, "loss": 3.2048, "step": 96530 }, { "epoch": 0.0016384, "grad_norm": 0.6410428285598755, "learning_rate": 1.715552661410819e-05, "loss": 3.0466, "step": 96540 }, { "epoch": 0.001664, "grad_norm": 0.7768917083740234, "learning_rate": 1.7154964056547045e-05, "loss": 3.1609, "step": 96550 }, { "epoch": 0.0016896, "grad_norm": 0.7340030670166016, "learning_rate": 1.715440145258786e-05, "loss": 2.909, "step": 96560 }, { "epoch": 0.0017152, "grad_norm": 0.6646959781646729, "learning_rate": 1.7153838802234285e-05, "loss": 3.0479, "step": 96570 }, { "epoch": 0.0017408, "grad_norm": 0.6853362321853638, "learning_rate": 1.7153276105489963e-05, "loss": 2.9112, "step": 96580 }, { "epoch": 0.0017664, "grad_norm": 1.015451192855835, "learning_rate": 1.7152713362358548e-05, "loss": 3.0165, "step": 96590 }, { "epoch": 0.001792, "grad_norm": 0.6569464802742004, "learning_rate": 1.7152150572843688e-05, "loss": 2.7443, "step": 96600 }, { "epoch": 0.0018176, "grad_norm": 0.9318958520889282, "learning_rate": 1.715158773694903e-05, "loss": 3.0127, "step": 96610 }, { "epoch": 0.0018432, "grad_norm": 0.6724623441696167, "learning_rate": 1.7151024854678227e-05, "loss": 3.0552, "step": 96620 }, { "epoch": 0.0018688, "grad_norm": 0.7244054079055786, "learning_rate": 1.7150461926034932e-05, "loss": 3.1658, "step": 96630 }, { "epoch": 0.0018944, "grad_norm": 0.7374611496925354, "learning_rate": 1.714989895102279e-05, "loss": 3.2756, "step": 96640 }, { "epoch": 0.00192, "grad_norm": 0.7125588059425354, "learning_rate": 1.7149335929645454e-05, "loss": 2.7276, "step": 96650 }, { "epoch": 0.0019456, "grad_norm": 0.6114538311958313, "learning_rate": 1.7148772861906572e-05, "loss": 2.7231, "step": 96660 }, { "epoch": 0.0019712, "grad_norm": 0.8618456125259399, "learning_rate": 1.7148209747809798e-05, "loss": 3.1284, "step": 96670 }, { "epoch": 0.0019968, "grad_norm": 0.7974724769592285, "learning_rate": 1.7147646587358786e-05, "loss": 3.0275, "step": 96680 }, { "epoch": 0.0020224, "grad_norm": 0.7616164088249207, "learning_rate": 1.7147083380557186e-05, "loss": 3.202, "step": 96690 }, { "epoch": 0.002048, "grad_norm": 0.7844250798225403, "learning_rate": 1.7146520127408647e-05, "loss": 2.9004, "step": 96700 }, { "epoch": 0.0020736, "grad_norm": 0.6866908073425293, "learning_rate": 1.7145956827916824e-05, "loss": 2.8918, "step": 96710 }, { "epoch": 0.0020992, "grad_norm": 0.7209980487823486, "learning_rate": 1.714539348208537e-05, "loss": 2.8972, "step": 96720 }, { "epoch": 0.0021248, "grad_norm": 0.6859113574028015, "learning_rate": 1.7144830089917942e-05, "loss": 3.2407, "step": 96730 }, { "epoch": 0.0021504, "grad_norm": 0.6538119912147522, "learning_rate": 1.7144266651418188e-05, "loss": 3.1704, "step": 96740 }, { "epoch": 0.002176, "grad_norm": 0.6831783652305603, "learning_rate": 1.7143703166589762e-05, "loss": 3.0708, "step": 96750 }, { "epoch": 0.0022016, "grad_norm": 3.860102891921997, "learning_rate": 1.714313963543632e-05, "loss": 3.0218, "step": 96760 }, { "epoch": 0.0022272, "grad_norm": 0.7863857746124268, "learning_rate": 1.7142576057961518e-05, "loss": 2.9745, "step": 96770 }, { "epoch": 0.0022528, "grad_norm": 0.7908477783203125, "learning_rate": 1.7142012434169006e-05, "loss": 3.1081, "step": 96780 }, { "epoch": 0.0022784, "grad_norm": 0.6577082872390747, "learning_rate": 1.7141448764062444e-05, "loss": 3.0002, "step": 96790 }, { "epoch": 0.002304, "grad_norm": 0.8465343117713928, "learning_rate": 1.7140885047645484e-05, "loss": 3.0432, "step": 96800 }, { "epoch": 0.0023296, "grad_norm": 0.6908205151557922, "learning_rate": 1.7140321284921778e-05, "loss": 3.0118, "step": 96810 }, { "epoch": 0.0023552, "grad_norm": 0.7248814105987549, "learning_rate": 1.7139757475894993e-05, "loss": 3.0588, "step": 96820 }, { "epoch": 0.0023808, "grad_norm": 1.0022969245910645, "learning_rate": 1.7139193620568773e-05, "loss": 3.2712, "step": 96830 }, { "epoch": 0.0024064, "grad_norm": 0.884573757648468, "learning_rate": 1.713862971894678e-05, "loss": 2.8839, "step": 96840 }, { "epoch": 0.002432, "grad_norm": 0.6804940700531006, "learning_rate": 1.7138065771032674e-05, "loss": 3.3074, "step": 96850 }, { "epoch": 0.0024576, "grad_norm": 0.7074170112609863, "learning_rate": 1.7137501776830107e-05, "loss": 3.2006, "step": 96860 }, { "epoch": 0.0024832, "grad_norm": 0.6900574564933777, "learning_rate": 1.713693773634274e-05, "loss": 3.0128, "step": 96870 }, { "epoch": 0.0025088, "grad_norm": 0.7866234183311462, "learning_rate": 1.7136373649574223e-05, "loss": 2.81, "step": 96880 }, { "epoch": 0.0025344, "grad_norm": 0.7835409045219421, "learning_rate": 1.7135809516528224e-05, "loss": 3.1655, "step": 96890 }, { "epoch": 0.00256, "grad_norm": 0.9432882070541382, "learning_rate": 1.713524533720839e-05, "loss": 3.2205, "step": 96900 }, { "epoch": 0.0025856, "grad_norm": 0.8068422675132751, "learning_rate": 1.7134681111618395e-05, "loss": 3.3845, "step": 96910 }, { "epoch": 0.0026112, "grad_norm": 0.6778401136398315, "learning_rate": 1.7134116839761887e-05, "loss": 3.1436, "step": 96920 }, { "epoch": 0.0026368, "grad_norm": 0.7891331315040588, "learning_rate": 1.7133552521642527e-05, "loss": 3.1669, "step": 96930 }, { "epoch": 0.0026624, "grad_norm": 0.7061665654182434, "learning_rate": 1.7132988157263973e-05, "loss": 3.1853, "step": 96940 }, { "epoch": 0.002688, "grad_norm": 0.7072274088859558, "learning_rate": 1.713242374662989e-05, "loss": 3.1893, "step": 96950 }, { "epoch": 0.0027136, "grad_norm": 0.733315110206604, "learning_rate": 1.713185928974393e-05, "loss": 2.9903, "step": 96960 }, { "epoch": 0.0027392, "grad_norm": 0.9199950098991394, "learning_rate": 1.7131294786609757e-05, "loss": 3.0674, "step": 96970 }, { "epoch": 0.0027648, "grad_norm": 1.0431832075119019, "learning_rate": 1.7130730237231035e-05, "loss": 3.1714, "step": 96980 }, { "epoch": 0.0027904, "grad_norm": 0.8184259533882141, "learning_rate": 1.7130165641611423e-05, "loss": 3.1905, "step": 96990 }, { "epoch": 0.002816, "grad_norm": 0.8008385896682739, "learning_rate": 1.7129600999754586e-05, "loss": 3.1442, "step": 97000 }, { "epoch": 0.0028416, "grad_norm": 0.7744402885437012, "learning_rate": 1.7129036311664177e-05, "loss": 3.2586, "step": 97010 }, { "epoch": 0.0028672, "grad_norm": 0.6462035179138184, "learning_rate": 1.712847157734386e-05, "loss": 3.1218, "step": 97020 }, { "epoch": 0.0028928, "grad_norm": 0.6570467352867126, "learning_rate": 1.7127906796797306e-05, "loss": 3.2324, "step": 97030 }, { "epoch": 0.0029184, "grad_norm": 1.050955057144165, "learning_rate": 1.7127341970028165e-05, "loss": 2.9652, "step": 97040 }, { "epoch": 0.002944, "grad_norm": 0.8258428573608398, "learning_rate": 1.7126777097040106e-05, "loss": 3.1121, "step": 97050 }, { "epoch": 0.0029696, "grad_norm": 0.6641935706138611, "learning_rate": 1.712621217783679e-05, "loss": 3.143, "step": 97060 }, { "epoch": 0.0029952, "grad_norm": 0.688342273235321, "learning_rate": 1.7125647212421887e-05, "loss": 3.2016, "step": 97070 }, { "epoch": 0.0030208, "grad_norm": 0.7617918848991394, "learning_rate": 1.7125082200799054e-05, "loss": 3.0597, "step": 97080 }, { "epoch": 0.0030464, "grad_norm": 0.8468484282493591, "learning_rate": 1.7124517142971953e-05, "loss": 3.2532, "step": 97090 }, { "epoch": 0.003072, "grad_norm": 0.8203907012939453, "learning_rate": 1.7123952038944257e-05, "loss": 3.1928, "step": 97100 }, { "epoch": 0.0030976, "grad_norm": 0.819573700428009, "learning_rate": 1.712338688871962e-05, "loss": 3.0114, "step": 97110 }, { "epoch": 0.0031232, "grad_norm": 0.7346654534339905, "learning_rate": 1.7122821692301717e-05, "loss": 3.2176, "step": 97120 }, { "epoch": 0.0031488, "grad_norm": 0.7140784859657288, "learning_rate": 1.7122256449694206e-05, "loss": 2.9847, "step": 97130 }, { "epoch": 0.0031744, "grad_norm": 0.7214703559875488, "learning_rate": 1.7121691160900757e-05, "loss": 3.2573, "step": 97140 }, { "epoch": 0.0032, "grad_norm": 1.1778326034545898, "learning_rate": 1.712112582592503e-05, "loss": 3.1835, "step": 97150 }, { "epoch": 0.0032256, "grad_norm": 0.7172421813011169, "learning_rate": 1.7120560444770696e-05, "loss": 3.1273, "step": 97160 }, { "epoch": 0.0032512, "grad_norm": 1.368874430656433, "learning_rate": 1.7119995017441423e-05, "loss": 3.2569, "step": 97170 }, { "epoch": 0.0032768, "grad_norm": 1.0402195453643799, "learning_rate": 1.7119429543940872e-05, "loss": 3.2243, "step": 97180 }, { "epoch": 0.0033024, "grad_norm": 1.0227025747299194, "learning_rate": 1.711886402427271e-05, "loss": 3.2325, "step": 97190 }, { "epoch": 0.003328, "grad_norm": 1.0622048377990723, "learning_rate": 1.711829845844061e-05, "loss": 3.2297, "step": 97200 }, { "epoch": 0.0033536, "grad_norm": 3.5997378826141357, "learning_rate": 1.7117732846448235e-05, "loss": 3.0355, "step": 97210 }, { "epoch": 0.0033792, "grad_norm": 1.2983818054199219, "learning_rate": 1.7117167188299255e-05, "loss": 3.2791, "step": 97220 }, { "epoch": 0.0034048, "grad_norm": 0.7023141384124756, "learning_rate": 1.7116601483997337e-05, "loss": 3.0358, "step": 97230 }, { "epoch": 0.0034304, "grad_norm": 0.77740877866745, "learning_rate": 1.7116035733546147e-05, "loss": 3.0986, "step": 97240 }, { "epoch": 0.003456, "grad_norm": 0.712219774723053, "learning_rate": 1.711546993694936e-05, "loss": 3.0467, "step": 97250 }, { "epoch": 0.0034816, "grad_norm": 0.6838272213935852, "learning_rate": 1.7114904094210642e-05, "loss": 3.0867, "step": 97260 }, { "epoch": 0.0035072, "grad_norm": 0.7217069268226624, "learning_rate": 1.7114338205333662e-05, "loss": 3.0474, "step": 97270 }, { "epoch": 0.0035328, "grad_norm": 0.8486436009407043, "learning_rate": 1.7113772270322087e-05, "loss": 3.1581, "step": 97280 }, { "epoch": 0.0035584, "grad_norm": 0.7667872309684753, "learning_rate": 1.711320628917959e-05, "loss": 3.0802, "step": 97290 }, { "epoch": 0.003584, "grad_norm": 0.7250868082046509, "learning_rate": 1.711264026190984e-05, "loss": 3.0374, "step": 97300 }, { "epoch": 0.0036096, "grad_norm": 0.7208192944526672, "learning_rate": 1.711207418851651e-05, "loss": 3.2736, "step": 97310 }, { "epoch": 0.0036352, "grad_norm": 0.8173514008522034, "learning_rate": 1.711150806900327e-05, "loss": 3.0952, "step": 97320 }, { "epoch": 0.0036608, "grad_norm": 0.7242875695228577, "learning_rate": 1.711094190337379e-05, "loss": 3.2794, "step": 97330 }, { "epoch": 0.0036864, "grad_norm": 0.859281599521637, "learning_rate": 1.711037569163174e-05, "loss": 3.2752, "step": 97340 }, { "epoch": 0.003712, "grad_norm": 0.7251374125480652, "learning_rate": 1.710980943378079e-05, "loss": 2.9604, "step": 97350 }, { "epoch": 0.0037376, "grad_norm": 0.8332594633102417, "learning_rate": 1.710924312982462e-05, "loss": 3.1392, "step": 97360 }, { "epoch": 0.0037632, "grad_norm": 0.722610592842102, "learning_rate": 1.71086767797669e-05, "loss": 3.1015, "step": 97370 }, { "epoch": 0.0037888, "grad_norm": 0.722746729850769, "learning_rate": 1.7108110383611295e-05, "loss": 3.0878, "step": 97380 }, { "epoch": 0.0038144, "grad_norm": 0.718352198600769, "learning_rate": 1.7107543941361483e-05, "loss": 3.0059, "step": 97390 }, { "epoch": 0.00384, "grad_norm": 0.7379565238952637, "learning_rate": 1.710697745302114e-05, "loss": 3.2967, "step": 97400 }, { "epoch": 0.0038656, "grad_norm": 0.928436279296875, "learning_rate": 1.710641091859394e-05, "loss": 3.157, "step": 97410 }, { "epoch": 0.0038912, "grad_norm": 0.9791051149368286, "learning_rate": 1.710584433808355e-05, "loss": 2.9599, "step": 97420 }, { "epoch": 0.0039168, "grad_norm": 0.8815745711326599, "learning_rate": 1.710527771149365e-05, "loss": 3.0584, "step": 97430 }, { "epoch": 0.0039424, "grad_norm": 0.7015538811683655, "learning_rate": 1.710471103882791e-05, "loss": 3.1437, "step": 97440 }, { "epoch": 0.003968, "grad_norm": 0.7621333003044128, "learning_rate": 1.710414432009001e-05, "loss": 3.2898, "step": 97450 }, { "epoch": 0.0039936, "grad_norm": 0.7681944966316223, "learning_rate": 1.7103577555283616e-05, "loss": 3.2065, "step": 97460 }, { "epoch": 0.0040192, "grad_norm": 0.8747596144676208, "learning_rate": 1.7103010744412415e-05, "loss": 3.1039, "step": 97470 }, { "epoch": 0.0040448, "grad_norm": 0.744846761226654, "learning_rate": 1.7102443887480077e-05, "loss": 2.979, "step": 97480 }, { "epoch": 0.0040704, "grad_norm": 0.9087570309638977, "learning_rate": 1.7101876984490275e-05, "loss": 3.3211, "step": 97490 }, { "epoch": 0.004096, "grad_norm": 1.2041507959365845, "learning_rate": 1.710131003544669e-05, "loss": 3.2059, "step": 97500 }, { "epoch": 0.0041216, "grad_norm": 0.7326367497444153, "learning_rate": 1.7100743040353e-05, "loss": 3.0931, "step": 97510 }, { "epoch": 0.0041472, "grad_norm": 0.7381828427314758, "learning_rate": 1.7100175999212873e-05, "loss": 3.1312, "step": 97520 }, { "epoch": 0.0041728, "grad_norm": 0.7399926781654358, "learning_rate": 1.7099608912029993e-05, "loss": 3.1478, "step": 97530 }, { "epoch": 0.0041984, "grad_norm": 0.7021884918212891, "learning_rate": 1.7099041778808037e-05, "loss": 3.2469, "step": 97540 }, { "epoch": 0.004224, "grad_norm": 0.9348466396331787, "learning_rate": 1.7098474599550684e-05, "loss": 2.9562, "step": 97550 }, { "epoch": 0.0042496, "grad_norm": 0.7084819674491882, "learning_rate": 1.7097907374261606e-05, "loss": 3.0822, "step": 97560 }, { "epoch": 0.0042752, "grad_norm": 0.8118166327476501, "learning_rate": 1.7097340102944487e-05, "loss": 3.1676, "step": 97570 }, { "epoch": 0.0043008, "grad_norm": 0.8539758324623108, "learning_rate": 1.7096772785603003e-05, "loss": 3.3653, "step": 97580 }, { "epoch": 0.0043264, "grad_norm": 0.7977530360221863, "learning_rate": 1.7096205422240833e-05, "loss": 3.2001, "step": 97590 }, { "epoch": 0.004352, "grad_norm": 0.7985868453979492, "learning_rate": 1.7095638012861657e-05, "loss": 3.2221, "step": 97600 }, { "epoch": 0.0043776, "grad_norm": 0.7666782736778259, "learning_rate": 1.7095070557469157e-05, "loss": 3.1984, "step": 97610 }, { "epoch": 0.0044032, "grad_norm": 0.6783486604690552, "learning_rate": 1.709450305606701e-05, "loss": 3.0557, "step": 97620 }, { "epoch": 0.0044288, "grad_norm": 0.7670270800590515, "learning_rate": 1.7093935508658892e-05, "loss": 3.2561, "step": 97630 }, { "epoch": 0.0044544, "grad_norm": 0.9470139145851135, "learning_rate": 1.709336791524849e-05, "loss": 3.1454, "step": 97640 }, { "epoch": 0.00448, "grad_norm": 0.8091501593589783, "learning_rate": 1.7092800275839485e-05, "loss": 3.1534, "step": 97650 }, { "epoch": 0.0045056, "grad_norm": 1.0588068962097168, "learning_rate": 1.709223259043555e-05, "loss": 3.1914, "step": 97660 }, { "epoch": 0.0045312, "grad_norm": 0.6812363266944885, "learning_rate": 1.7091664859040374e-05, "loss": 3.1542, "step": 97670 }, { "epoch": 0.0045568, "grad_norm": 0.7464112639427185, "learning_rate": 1.7091097081657633e-05, "loss": 3.007, "step": 97680 }, { "epoch": 0.0045824, "grad_norm": 0.8858690857887268, "learning_rate": 1.7090529258291016e-05, "loss": 3.2355, "step": 97690 }, { "epoch": 0.004608, "grad_norm": 0.6572077870368958, "learning_rate": 1.70899613889442e-05, "loss": 3.1589, "step": 97700 }, { "epoch": 2.56e-05, "grad_norm": 0.8692609071731567, "learning_rate": 1.708939347362087e-05, "loss": 2.7203, "step": 97710 }, { "epoch": 5.12e-05, "grad_norm": 0.9136074781417847, "learning_rate": 1.7088825512324703e-05, "loss": 2.9086, "step": 97720 }, { "epoch": 7.68e-05, "grad_norm": 0.6518642902374268, "learning_rate": 1.7088257505059393e-05, "loss": 3.1402, "step": 97730 }, { "epoch": 0.0001024, "grad_norm": 0.7756896018981934, "learning_rate": 1.7087689451828613e-05, "loss": 3.1614, "step": 97740 }, { "epoch": 0.000128, "grad_norm": 0.7330282926559448, "learning_rate": 1.708712135263605e-05, "loss": 3.1216, "step": 97750 }, { "epoch": 0.0001536, "grad_norm": 0.7023009657859802, "learning_rate": 1.708655320748539e-05, "loss": 2.852, "step": 97760 }, { "epoch": 0.0001792, "grad_norm": 0.8300282955169678, "learning_rate": 1.7085985016380317e-05, "loss": 2.9754, "step": 97770 }, { "epoch": 0.0002048, "grad_norm": 0.7680101990699768, "learning_rate": 1.7085416779324516e-05, "loss": 2.996, "step": 97780 }, { "epoch": 0.0002304, "grad_norm": 0.7340623736381531, "learning_rate": 1.7084848496321663e-05, "loss": 2.8844, "step": 97790 }, { "epoch": 0.000256, "grad_norm": 0.7002426385879517, "learning_rate": 1.708428016737546e-05, "loss": 2.9243, "step": 97800 }, { "epoch": 0.0002816, "grad_norm": 0.6892679333686829, "learning_rate": 1.7083711792489576e-05, "loss": 2.8902, "step": 97810 }, { "epoch": 0.0003072, "grad_norm": 0.8227244019508362, "learning_rate": 1.7083143371667703e-05, "loss": 3.1538, "step": 97820 }, { "epoch": 0.0003328, "grad_norm": 0.8150776624679565, "learning_rate": 1.708257490491353e-05, "loss": 3.3217, "step": 97830 }, { "epoch": 0.0003584, "grad_norm": 0.7157617211341858, "learning_rate": 1.7082006392230742e-05, "loss": 3.0122, "step": 97840 }, { "epoch": 0.000384, "grad_norm": 0.6442682147026062, "learning_rate": 1.7081437833623023e-05, "loss": 3.008, "step": 97850 }, { "epoch": 0.0004096, "grad_norm": 0.9846924543380737, "learning_rate": 1.7080869229094063e-05, "loss": 3.4284, "step": 97860 }, { "epoch": 0.0004352, "grad_norm": 0.8325411081314087, "learning_rate": 1.708030057864755e-05, "loss": 3.0536, "step": 97870 }, { "epoch": 0.0004608, "grad_norm": 0.7939269542694092, "learning_rate": 1.7079731882287163e-05, "loss": 3.0929, "step": 97880 }, { "epoch": 0.0004864, "grad_norm": 0.6816971898078918, "learning_rate": 1.7079163140016602e-05, "loss": 2.9676, "step": 97890 }, { "epoch": 0.000512, "grad_norm": 0.7334587574005127, "learning_rate": 1.707859435183955e-05, "loss": 2.8108, "step": 97900 }, { "epoch": 0.0005376, "grad_norm": 1.1693485975265503, "learning_rate": 1.707802551775969e-05, "loss": 2.9323, "step": 97910 }, { "epoch": 0.0005632, "grad_norm": 0.7416588068008423, "learning_rate": 1.707745663778072e-05, "loss": 2.8735, "step": 97920 }, { "epoch": 0.0005888, "grad_norm": 0.8728213310241699, "learning_rate": 1.707688771190632e-05, "loss": 2.9571, "step": 97930 }, { "epoch": 0.0006144, "grad_norm": 0.664774477481842, "learning_rate": 1.7076318740140188e-05, "loss": 3.1054, "step": 97940 }, { "epoch": 0.00064, "grad_norm": 0.8891397714614868, "learning_rate": 1.707574972248601e-05, "loss": 2.9399, "step": 97950 }, { "epoch": 0.0006656, "grad_norm": 0.8371104001998901, "learning_rate": 1.7075180658947473e-05, "loss": 3.4159, "step": 97960 }, { "epoch": 0.0006912, "grad_norm": 0.6779100894927979, "learning_rate": 1.7074611549528272e-05, "loss": 3.1985, "step": 97970 }, { "epoch": 0.0007168, "grad_norm": 1.12696373462677, "learning_rate": 1.7074042394232093e-05, "loss": 3.4741, "step": 97980 }, { "epoch": 0.0007424, "grad_norm": 0.7355062365531921, "learning_rate": 1.707347319306263e-05, "loss": 2.92, "step": 97990 }, { "epoch": 0.000768, "grad_norm": 0.7271464467048645, "learning_rate": 1.7072903946023572e-05, "loss": 2.95, "step": 98000 }, { "epoch": 0.0007936, "grad_norm": 0.7832618951797485, "learning_rate": 1.7072334653118614e-05, "loss": 2.9077, "step": 98010 }, { "epoch": 0.0008192, "grad_norm": 0.814172089099884, "learning_rate": 1.7071765314351446e-05, "loss": 3.0783, "step": 98020 }, { "epoch": 0.0008448, "grad_norm": 0.7021558284759521, "learning_rate": 1.7071195929725758e-05, "loss": 2.9006, "step": 98030 }, { "epoch": 0.0008704, "grad_norm": 0.7962903380393982, "learning_rate": 1.707062649924524e-05, "loss": 3.0491, "step": 98040 }, { "epoch": 0.000896, "grad_norm": 1.7178128957748413, "learning_rate": 1.7070057022913597e-05, "loss": 3.2446, "step": 98050 }, { "epoch": 0.0009216, "grad_norm": 0.5906525254249573, "learning_rate": 1.706948750073451e-05, "loss": 2.8883, "step": 98060 }, { "epoch": 0.0009472, "grad_norm": 0.671209454536438, "learning_rate": 1.706891793271167e-05, "loss": 3.0384, "step": 98070 }, { "epoch": 0.0009728, "grad_norm": 0.6758866906166077, "learning_rate": 1.7068348318848777e-05, "loss": 2.9713, "step": 98080 }, { "epoch": 0.0009984, "grad_norm": 0.9335502982139587, "learning_rate": 1.7067778659149526e-05, "loss": 3.1277, "step": 98090 }, { "epoch": 0.001024, "grad_norm": 0.9246643781661987, "learning_rate": 1.706720895361761e-05, "loss": 3.2653, "step": 98100 }, { "epoch": 0.0010496, "grad_norm": 0.7772267460823059, "learning_rate": 1.706663920225672e-05, "loss": 3.0627, "step": 98110 }, { "epoch": 0.0010752, "grad_norm": 0.6936805248260498, "learning_rate": 1.7066069405070553e-05, "loss": 3.0419, "step": 98120 }, { "epoch": 0.0011008, "grad_norm": 0.7694855332374573, "learning_rate": 1.706549956206281e-05, "loss": 3.0894, "step": 98130 }, { "epoch": 0.0011264, "grad_norm": 0.7861539721488953, "learning_rate": 1.706492967323717e-05, "loss": 3.0535, "step": 98140 }, { "epoch": 0.001152, "grad_norm": 0.7768762707710266, "learning_rate": 1.7064359738597348e-05, "loss": 3.1376, "step": 98150 }, { "epoch": 0.0011776, "grad_norm": 0.664388120174408, "learning_rate": 1.7063789758147025e-05, "loss": 3.0519, "step": 98160 }, { "epoch": 0.0012032, "grad_norm": 0.6832517385482788, "learning_rate": 1.7063219731889902e-05, "loss": 3.2001, "step": 98170 }, { "epoch": 0.0012288, "grad_norm": 0.708406388759613, "learning_rate": 1.7062649659829676e-05, "loss": 2.7527, "step": 98180 }, { "epoch": 0.0012544, "grad_norm": 0.7165277004241943, "learning_rate": 1.706207954197005e-05, "loss": 3.1406, "step": 98190 }, { "epoch": 0.00128, "grad_norm": 0.7033838033676147, "learning_rate": 1.706150937831471e-05, "loss": 3.0758, "step": 98200 }, { "epoch": 0.0013056, "grad_norm": 0.7147420048713684, "learning_rate": 1.706093916886736e-05, "loss": 2.8667, "step": 98210 }, { "epoch": 0.0013312, "grad_norm": 0.7696802616119385, "learning_rate": 1.7060368913631695e-05, "loss": 3.1126, "step": 98220 }, { "epoch": 0.0013568, "grad_norm": 0.6700315475463867, "learning_rate": 1.7059798612611416e-05, "loss": 2.9965, "step": 98230 }, { "epoch": 0.0013824, "grad_norm": 0.7572527527809143, "learning_rate": 1.7059228265810215e-05, "loss": 2.9992, "step": 98240 }, { "epoch": 0.001408, "grad_norm": 0.6967811584472656, "learning_rate": 1.7058657873231797e-05, "loss": 3.0946, "step": 98250 }, { "epoch": 0.0014336, "grad_norm": 0.7392900586128235, "learning_rate": 1.705808743487986e-05, "loss": 3.1392, "step": 98260 }, { "epoch": 0.0014592, "grad_norm": 0.7687879204750061, "learning_rate": 1.7057516950758103e-05, "loss": 3.0486, "step": 98270 }, { "epoch": 0.0014848, "grad_norm": 0.7249293923377991, "learning_rate": 1.7056946420870223e-05, "loss": 2.7915, "step": 98280 }, { "epoch": 0.0015104, "grad_norm": 0.764680802822113, "learning_rate": 1.705637584521992e-05, "loss": 2.7134, "step": 98290 }, { "epoch": 0.001536, "grad_norm": 0.6802111864089966, "learning_rate": 1.7055805223810896e-05, "loss": 3.0429, "step": 98300 }, { "epoch": 0.0015616, "grad_norm": 0.6371496915817261, "learning_rate": 1.705523455664685e-05, "loss": 3.1131, "step": 98310 }, { "epoch": 0.0015872, "grad_norm": 0.8037528395652771, "learning_rate": 1.705466384373148e-05, "loss": 3.196, "step": 98320 }, { "epoch": 0.0016128, "grad_norm": 0.8527507781982422, "learning_rate": 1.7054093085068495e-05, "loss": 3.0809, "step": 98330 }, { "epoch": 0.0016384, "grad_norm": 0.8895035982131958, "learning_rate": 1.7053522280661588e-05, "loss": 2.9801, "step": 98340 }, { "epoch": 0.001664, "grad_norm": 0.8378057479858398, "learning_rate": 1.7052951430514463e-05, "loss": 3.1657, "step": 98350 }, { "epoch": 0.0016896, "grad_norm": 1.2708407640457153, "learning_rate": 1.7052380534630826e-05, "loss": 2.8781, "step": 98360 }, { "epoch": 0.0017152, "grad_norm": 0.791692316532135, "learning_rate": 1.7051809593014373e-05, "loss": 3.1882, "step": 98370 }, { "epoch": 0.0017408, "grad_norm": 0.7346146106719971, "learning_rate": 1.705123860566881e-05, "loss": 2.9387, "step": 98380 }, { "epoch": 0.0017664, "grad_norm": 0.6799153685569763, "learning_rate": 1.7050667572597836e-05, "loss": 2.8396, "step": 98390 }, { "epoch": 0.001792, "grad_norm": 0.7023864388465881, "learning_rate": 1.705009649380516e-05, "loss": 2.9685, "step": 98400 }, { "epoch": 0.0018176, "grad_norm": 0.8098397254943848, "learning_rate": 1.704952536929448e-05, "loss": 2.7982, "step": 98410 }, { "epoch": 0.0018432, "grad_norm": 0.8378589749336243, "learning_rate": 1.7048954199069503e-05, "loss": 3.0573, "step": 98420 }, { "epoch": 0.0018688, "grad_norm": 0.7434520721435547, "learning_rate": 1.704838298313393e-05, "loss": 2.9024, "step": 98430 }, { "epoch": 0.0018944, "grad_norm": 0.7049421072006226, "learning_rate": 1.7047811721491466e-05, "loss": 3.1865, "step": 98440 }, { "epoch": 0.00192, "grad_norm": 0.8253280520439148, "learning_rate": 1.7047240414145817e-05, "loss": 2.9902, "step": 98450 }, { "epoch": 0.0019456, "grad_norm": 0.6105465292930603, "learning_rate": 1.7046669061100684e-05, "loss": 2.78, "step": 98460 }, { "epoch": 0.0019712, "grad_norm": 0.8689247369766235, "learning_rate": 1.704609766235978e-05, "loss": 3.0042, "step": 98470 }, { "epoch": 0.0019968, "grad_norm": 0.7979324460029602, "learning_rate": 1.70455262179268e-05, "loss": 2.858, "step": 98480 }, { "epoch": 0.0020224, "grad_norm": 0.796120285987854, "learning_rate": 1.704495472780546e-05, "loss": 3.1133, "step": 98490 }, { "epoch": 0.002048, "grad_norm": 0.7757472395896912, "learning_rate": 1.7044383191999458e-05, "loss": 2.7717, "step": 98500 }, { "epoch": 0.0020736, "grad_norm": 0.6427645087242126, "learning_rate": 1.7043811610512506e-05, "loss": 2.958, "step": 98510 }, { "epoch": 0.0020992, "grad_norm": 0.6959121227264404, "learning_rate": 1.7043239983348304e-05, "loss": 3.0306, "step": 98520 }, { "epoch": 0.0021248, "grad_norm": 0.7283143997192383, "learning_rate": 1.7042668310510567e-05, "loss": 3.0778, "step": 98530 }, { "epoch": 0.0021504, "grad_norm": 1.477466344833374, "learning_rate": 1.7042096592002995e-05, "loss": 3.1364, "step": 98540 }, { "epoch": 0.002176, "grad_norm": 0.6806459426879883, "learning_rate": 1.7041524827829296e-05, "loss": 3.1542, "step": 98550 }, { "epoch": 0.0022016, "grad_norm": 0.7244477868080139, "learning_rate": 1.7040953017993182e-05, "loss": 2.9889, "step": 98560 }, { "epoch": 0.0022272, "grad_norm": 0.6496001482009888, "learning_rate": 1.704038116249836e-05, "loss": 2.8729, "step": 98570 }, { "epoch": 0.0022528, "grad_norm": 0.7153236865997314, "learning_rate": 1.7039809261348537e-05, "loss": 2.7527, "step": 98580 }, { "epoch": 0.0022784, "grad_norm": 0.7318360805511475, "learning_rate": 1.7039237314547422e-05, "loss": 2.7654, "step": 98590 }, { "epoch": 0.002304, "grad_norm": 0.7330282330513, "learning_rate": 1.7038665322098723e-05, "loss": 3.0169, "step": 98600 }, { "epoch": 0.0023296, "grad_norm": 0.7978213429450989, "learning_rate": 1.703809328400615e-05, "loss": 2.9324, "step": 98610 }, { "epoch": 0.0023552, "grad_norm": 0.7422904372215271, "learning_rate": 1.7037521200273414e-05, "loss": 2.7856, "step": 98620 }, { "epoch": 0.0023808, "grad_norm": 0.7875343561172485, "learning_rate": 1.703694907090422e-05, "loss": 3.1026, "step": 98630 }, { "epoch": 0.0024064, "grad_norm": 0.8493356704711914, "learning_rate": 1.7036376895902282e-05, "loss": 2.8281, "step": 98640 }, { "epoch": 0.002432, "grad_norm": 0.7566995024681091, "learning_rate": 1.7035804675271313e-05, "loss": 3.0279, "step": 98650 }, { "epoch": 0.0024576, "grad_norm": 0.7099981904029846, "learning_rate": 1.703523240901502e-05, "loss": 3.138, "step": 98660 }, { "epoch": 0.0024832, "grad_norm": 1.08323073387146, "learning_rate": 1.7034660097137114e-05, "loss": 3.1306, "step": 98670 }, { "epoch": 0.0025088, "grad_norm": 0.760612964630127, "learning_rate": 1.7034087739641304e-05, "loss": 3.0074, "step": 98680 }, { "epoch": 0.0025344, "grad_norm": 0.7684496641159058, "learning_rate": 1.703351533653131e-05, "loss": 3.2894, "step": 98690 }, { "epoch": 0.00256, "grad_norm": 0.701318085193634, "learning_rate": 1.7032942887810834e-05, "loss": 2.8531, "step": 98700 }, { "epoch": 0.0025856, "grad_norm": 0.7060912251472473, "learning_rate": 1.7032370393483593e-05, "loss": 3.1319, "step": 98710 }, { "epoch": 0.0026112, "grad_norm": 0.8223101496696472, "learning_rate": 1.70317978535533e-05, "loss": 3.1034, "step": 98720 }, { "epoch": 0.0026368, "grad_norm": 0.7103957533836365, "learning_rate": 1.7031225268023666e-05, "loss": 2.9922, "step": 98730 }, { "epoch": 0.0026624, "grad_norm": 0.7143649458885193, "learning_rate": 1.7030652636898403e-05, "loss": 3.0883, "step": 98740 }, { "epoch": 0.002688, "grad_norm": 0.782746434211731, "learning_rate": 1.7030079960181232e-05, "loss": 2.934, "step": 98750 }, { "epoch": 0.0027136, "grad_norm": 0.7597983479499817, "learning_rate": 1.7029507237875857e-05, "loss": 3.1023, "step": 98760 }, { "epoch": 0.0027392, "grad_norm": 0.7074618339538574, "learning_rate": 1.7028934469985997e-05, "loss": 3.1236, "step": 98770 }, { "epoch": 0.0027648, "grad_norm": 0.9984597563743591, "learning_rate": 1.702836165651536e-05, "loss": 3.0698, "step": 98780 }, { "epoch": 0.0027904, "grad_norm": 0.9990344643592834, "learning_rate": 1.702778879746767e-05, "loss": 3.1208, "step": 98790 }, { "epoch": 0.002816, "grad_norm": 0.8498750925064087, "learning_rate": 1.7027215892846638e-05, "loss": 3.1577, "step": 98800 }, { "epoch": 2.56e-05, "grad_norm": 0.8591064810752869, "learning_rate": 1.7026642942655976e-05, "loss": 2.4987, "step": 98810 }, { "epoch": 5.12e-05, "grad_norm": 0.8151722550392151, "learning_rate": 1.7026069946899407e-05, "loss": 3.0862, "step": 98820 }, { "epoch": 7.68e-05, "grad_norm": 0.7923853397369385, "learning_rate": 1.7025496905580638e-05, "loss": 2.9671, "step": 98830 }, { "epoch": 0.0001024, "grad_norm": 0.8148379921913147, "learning_rate": 1.7024923818703388e-05, "loss": 3.0408, "step": 98840 }, { "epoch": 0.000128, "grad_norm": 0.7343773245811462, "learning_rate": 1.7024350686271375e-05, "loss": 2.9302, "step": 98850 }, { "epoch": 0.0001536, "grad_norm": 0.7971540689468384, "learning_rate": 1.7023777508288315e-05, "loss": 3.1297, "step": 98860 }, { "epoch": 0.0001792, "grad_norm": 0.6836139559745789, "learning_rate": 1.7023204284757922e-05, "loss": 3.194, "step": 98870 }, { "epoch": 0.0002048, "grad_norm": 0.7558373212814331, "learning_rate": 1.702263101568392e-05, "loss": 3.1135, "step": 98880 }, { "epoch": 0.0002304, "grad_norm": 0.6618276834487915, "learning_rate": 1.702205770107002e-05, "loss": 2.9868, "step": 98890 }, { "epoch": 0.000256, "grad_norm": 0.8136909008026123, "learning_rate": 1.7021484340919942e-05, "loss": 2.9792, "step": 98900 }, { "epoch": 0.0002816, "grad_norm": 0.8031215071678162, "learning_rate": 1.70209109352374e-05, "loss": 2.8926, "step": 98910 }, { "epoch": 0.0003072, "grad_norm": 0.7315375804901123, "learning_rate": 1.7020337484026127e-05, "loss": 2.7917, "step": 98920 }, { "epoch": 0.0003328, "grad_norm": 0.8372888565063477, "learning_rate": 1.7019763987289824e-05, "loss": 2.9595, "step": 98930 }, { "epoch": 0.0003584, "grad_norm": 0.7167302370071411, "learning_rate": 1.701919044503222e-05, "loss": 3.0744, "step": 98940 }, { "epoch": 0.000384, "grad_norm": 0.7171369194984436, "learning_rate": 1.7018616857257025e-05, "loss": 2.995, "step": 98950 }, { "epoch": 0.0004096, "grad_norm": 0.8134163022041321, "learning_rate": 1.701804322396797e-05, "loss": 2.9886, "step": 98960 }, { "epoch": 0.0004352, "grad_norm": 0.6673274040222168, "learning_rate": 1.701746954516877e-05, "loss": 2.8598, "step": 98970 }, { "epoch": 0.0004608, "grad_norm": 0.7331478595733643, "learning_rate": 1.7016895820863145e-05, "loss": 2.9682, "step": 98980 }, { "epoch": 0.0004864, "grad_norm": 0.79412841796875, "learning_rate": 1.7016322051054818e-05, "loss": 3.1143, "step": 98990 }, { "epoch": 0.000512, "grad_norm": 0.6740530133247375, "learning_rate": 1.7015748235747502e-05, "loss": 2.8479, "step": 99000 }, { "epoch": 0.0005376, "grad_norm": 0.7010242342948914, "learning_rate": 1.7015174374944926e-05, "loss": 2.8386, "step": 99010 }, { "epoch": 0.0005632, "grad_norm": 0.7979180216789246, "learning_rate": 1.701460046865081e-05, "loss": 2.8742, "step": 99020 }, { "epoch": 0.0005888, "grad_norm": 0.7159961462020874, "learning_rate": 1.701402651686887e-05, "loss": 2.854, "step": 99030 }, { "epoch": 0.0006144, "grad_norm": 0.7030190229415894, "learning_rate": 1.7013452519602832e-05, "loss": 2.7488, "step": 99040 }, { "epoch": 0.00064, "grad_norm": 0.6732873916625977, "learning_rate": 1.7012878476856426e-05, "loss": 2.6831, "step": 99050 }, { "epoch": 0.0006656, "grad_norm": 0.7996186017990112, "learning_rate": 1.7012304388633357e-05, "loss": 3.4589, "step": 99060 }, { "epoch": 0.0006912, "grad_norm": 0.7521770000457764, "learning_rate": 1.7011730254937365e-05, "loss": 3.0417, "step": 99070 }, { "epoch": 0.0007168, "grad_norm": 0.7190354466438293, "learning_rate": 1.7011156075772163e-05, "loss": 3.0598, "step": 99080 }, { "epoch": 0.0007424, "grad_norm": 0.7428975105285645, "learning_rate": 1.7010581851141478e-05, "loss": 3.2788, "step": 99090 }, { "epoch": 0.000768, "grad_norm": 0.6963837742805481, "learning_rate": 1.701000758104903e-05, "loss": 2.9198, "step": 99100 }, { "epoch": 0.0007936, "grad_norm": 0.680555522441864, "learning_rate": 1.700943326549855e-05, "loss": 2.9555, "step": 99110 }, { "epoch": 0.0008192, "grad_norm": 0.6815059781074524, "learning_rate": 1.7008858904493756e-05, "loss": 2.9185, "step": 99120 }, { "epoch": 0.0008448, "grad_norm": 0.6783283948898315, "learning_rate": 1.7008284498038375e-05, "loss": 2.9231, "step": 99130 }, { "epoch": 0.0008704, "grad_norm": 0.824239194393158, "learning_rate": 1.7007710046136133e-05, "loss": 2.9882, "step": 99140 }, { "epoch": 0.000896, "grad_norm": 0.7355965971946716, "learning_rate": 1.7007135548790754e-05, "loss": 3.0345, "step": 99150 }, { "epoch": 0.0009216, "grad_norm": 0.6772774457931519, "learning_rate": 1.700656100600596e-05, "loss": 2.6834, "step": 99160 }, { "epoch": 0.0009472, "grad_norm": 0.645607054233551, "learning_rate": 1.700598641778548e-05, "loss": 2.8749, "step": 99170 }, { "epoch": 0.0009728, "grad_norm": 0.6598203778266907, "learning_rate": 1.700541178413304e-05, "loss": 3.0349, "step": 99180 }, { "epoch": 0.0009984, "grad_norm": 0.7174056172370911, "learning_rate": 1.700483710505237e-05, "loss": 2.6909, "step": 99190 }, { "epoch": 0.001024, "grad_norm": 6.175410747528076, "learning_rate": 1.7004262380547192e-05, "loss": 3.1101, "step": 99200 }, { "epoch": 0.0010496, "grad_norm": 0.6946680545806885, "learning_rate": 1.7003687610621233e-05, "loss": 3.1305, "step": 99210 }, { "epoch": 0.0010752, "grad_norm": 0.9114315509796143, "learning_rate": 1.700311279527822e-05, "loss": 3.1906, "step": 99220 }, { "epoch": 0.0011008, "grad_norm": 0.7993954420089722, "learning_rate": 1.7002537934521886e-05, "loss": 3.0242, "step": 99230 }, { "epoch": 0.0011264, "grad_norm": 0.6683521866798401, "learning_rate": 1.700196302835595e-05, "loss": 2.7689, "step": 99240 }, { "epoch": 0.001152, "grad_norm": 0.7496097087860107, "learning_rate": 1.7001388076784143e-05, "loss": 3.2947, "step": 99250 }, { "epoch": 0.0011776, "grad_norm": 0.6319785118103027, "learning_rate": 1.7000813079810203e-05, "loss": 2.9182, "step": 99260 }, { "epoch": 0.0012032, "grad_norm": 0.8220082521438599, "learning_rate": 1.7000238037437844e-05, "loss": 3.1997, "step": 99270 }, { "epoch": 0.0012288, "grad_norm": 0.6946061253547668, "learning_rate": 1.6999662949670805e-05, "loss": 3.1118, "step": 99280 }, { "epoch": 0.0012544, "grad_norm": 0.7651723027229309, "learning_rate": 1.6999087816512808e-05, "loss": 3.434, "step": 99290 }, { "epoch": 0.00128, "grad_norm": 0.7008416056632996, "learning_rate": 1.699851263796759e-05, "loss": 2.8614, "step": 99300 }, { "epoch": 0.0013056, "grad_norm": 0.73825603723526, "learning_rate": 1.699793741403888e-05, "loss": 3.1953, "step": 99310 }, { "epoch": 0.0013312, "grad_norm": 0.7797605991363525, "learning_rate": 1.69973621447304e-05, "loss": 2.9271, "step": 99320 }, { "epoch": 0.0013568, "grad_norm": 0.8647593855857849, "learning_rate": 1.699678683004589e-05, "loss": 3.0892, "step": 99330 }, { "epoch": 0.0013824, "grad_norm": 0.7012277841567993, "learning_rate": 1.6996211469989074e-05, "loss": 2.9734, "step": 99340 }, { "epoch": 0.001408, "grad_norm": 0.8058099746704102, "learning_rate": 1.6995636064563693e-05, "loss": 3.101, "step": 99350 }, { "epoch": 0.0014336, "grad_norm": 0.6861569881439209, "learning_rate": 1.6995060613773463e-05, "loss": 3.1331, "step": 99360 }, { "epoch": 0.0014592, "grad_norm": 0.7084606289863586, "learning_rate": 1.699448511762213e-05, "loss": 2.8704, "step": 99370 }, { "epoch": 0.0014848, "grad_norm": 0.660554051399231, "learning_rate": 1.6993909576113418e-05, "loss": 2.978, "step": 99380 }, { "epoch": 0.0015104, "grad_norm": 0.6603445410728455, "learning_rate": 1.699333398925106e-05, "loss": 2.8577, "step": 99390 }, { "epoch": 0.001536, "grad_norm": 0.7922876477241516, "learning_rate": 1.6992758357038786e-05, "loss": 3.0379, "step": 99400 }, { "epoch": 0.0015616, "grad_norm": 0.8044219017028809, "learning_rate": 1.6992182679480337e-05, "loss": 3.1722, "step": 99410 }, { "epoch": 0.0015872, "grad_norm": 0.7143486142158508, "learning_rate": 1.6991606956579442e-05, "loss": 3.0706, "step": 99420 }, { "epoch": 0.0016128, "grad_norm": 0.8201897740364075, "learning_rate": 1.699103118833983e-05, "loss": 2.9941, "step": 99430 }, { "epoch": 0.0016384, "grad_norm": 0.7141014933586121, "learning_rate": 1.6990455374765245e-05, "loss": 2.946, "step": 99440 }, { "epoch": 0.001664, "grad_norm": 0.7858115434646606, "learning_rate": 1.698987951585941e-05, "loss": 3.1677, "step": 99450 }, { "epoch": 0.0016896, "grad_norm": 0.7375379800796509, "learning_rate": 1.698930361162607e-05, "loss": 2.9385, "step": 99460 }, { "epoch": 0.0017152, "grad_norm": 0.7280722856521606, "learning_rate": 1.6988727662068947e-05, "loss": 2.9582, "step": 99470 }, { "epoch": 0.0017408, "grad_norm": 0.6288456916809082, "learning_rate": 1.6988151667191783e-05, "loss": 2.7612, "step": 99480 }, { "epoch": 0.0017664, "grad_norm": 0.6676846146583557, "learning_rate": 1.6987575626998312e-05, "loss": 2.8078, "step": 99490 }, { "epoch": 0.001792, "grad_norm": 0.6953171491622925, "learning_rate": 1.6986999541492275e-05, "loss": 2.846, "step": 99500 }, { "epoch": 0.0018176, "grad_norm": 0.6714804768562317, "learning_rate": 1.6986423410677402e-05, "loss": 2.6951, "step": 99510 }, { "epoch": 0.0018432, "grad_norm": 0.7032709121704102, "learning_rate": 1.6985847234557427e-05, "loss": 3.0527, "step": 99520 }, { "epoch": 0.0018688, "grad_norm": 0.7702165842056274, "learning_rate": 1.698527101313609e-05, "loss": 3.3235, "step": 99530 }, { "epoch": 0.0018944, "grad_norm": 0.6723122596740723, "learning_rate": 1.698469474641713e-05, "loss": 2.9761, "step": 99540 }, { "epoch": 0.00192, "grad_norm": 1.4248906373977661, "learning_rate": 1.6984118434404277e-05, "loss": 2.922, "step": 99550 }, { "epoch": 0.0019456, "grad_norm": 0.7374867796897888, "learning_rate": 1.6983542077101275e-05, "loss": 3.0966, "step": 99560 }, { "epoch": 0.0019712, "grad_norm": 0.7950145602226257, "learning_rate": 1.698296567451186e-05, "loss": 2.5985, "step": 99570 }, { "epoch": 0.0019968, "grad_norm": 0.7574740052223206, "learning_rate": 1.698238922663976e-05, "loss": 3.2491, "step": 99580 }, { "epoch": 0.0020224, "grad_norm": 0.6814941763877869, "learning_rate": 1.698181273348873e-05, "loss": 3.2653, "step": 99590 }, { "epoch": 0.002048, "grad_norm": 0.6633294820785522, "learning_rate": 1.6981236195062497e-05, "loss": 2.8156, "step": 99600 }, { "epoch": 0.0020736, "grad_norm": 0.7842681407928467, "learning_rate": 1.6980659611364803e-05, "loss": 3.0203, "step": 99610 }, { "epoch": 0.0020992, "grad_norm": 0.8784698843955994, "learning_rate": 1.6980082982399388e-05, "loss": 3.1598, "step": 99620 }, { "epoch": 0.0021248, "grad_norm": 0.9005814790725708, "learning_rate": 1.697950630816999e-05, "loss": 2.9053, "step": 99630 }, { "epoch": 0.0021504, "grad_norm": 0.6768035888671875, "learning_rate": 1.6978929588680343e-05, "loss": 3.1547, "step": 99640 }, { "epoch": 0.002176, "grad_norm": 0.8591440916061401, "learning_rate": 1.6978352823934195e-05, "loss": 3.2101, "step": 99650 }, { "epoch": 0.0022016, "grad_norm": 0.7729890942573547, "learning_rate": 1.6977776013935287e-05, "loss": 3.2813, "step": 99660 }, { "epoch": 0.0022272, "grad_norm": 0.701104462146759, "learning_rate": 1.6977199158687355e-05, "loss": 2.7602, "step": 99670 }, { "epoch": 0.0022528, "grad_norm": 1.6272387504577637, "learning_rate": 1.6976622258194138e-05, "loss": 3.3864, "step": 99680 }, { "epoch": 0.0022784, "grad_norm": 0.6693437099456787, "learning_rate": 1.6976045312459383e-05, "loss": 3.0136, "step": 99690 }, { "epoch": 0.002304, "grad_norm": 0.8664726614952087, "learning_rate": 1.6975468321486825e-05, "loss": 3.0464, "step": 99700 }, { "epoch": 0.0023296, "grad_norm": 0.6648334264755249, "learning_rate": 1.6974891285280213e-05, "loss": 3.1512, "step": 99710 }, { "epoch": 0.0023552, "grad_norm": 0.7426807284355164, "learning_rate": 1.697431420384328e-05, "loss": 2.5888, "step": 99720 }, { "epoch": 0.0023808, "grad_norm": 0.7697530388832092, "learning_rate": 1.6973737077179776e-05, "loss": 2.8486, "step": 99730 }, { "epoch": 0.0024064, "grad_norm": 0.6489160060882568, "learning_rate": 1.697315990529344e-05, "loss": 3.2382, "step": 99740 }, { "epoch": 0.002432, "grad_norm": 0.7625892758369446, "learning_rate": 1.6972582688188013e-05, "loss": 3.0299, "step": 99750 }, { "epoch": 0.0024576, "grad_norm": 0.7192431688308716, "learning_rate": 1.6972005425867244e-05, "loss": 3.0903, "step": 99760 }, { "epoch": 0.0024832, "grad_norm": 0.9549127817153931, "learning_rate": 1.697142811833487e-05, "loss": 3.2176, "step": 99770 }, { "epoch": 0.0025088, "grad_norm": 0.7176723480224609, "learning_rate": 1.6970850765594636e-05, "loss": 3.1541, "step": 99780 }, { "epoch": 0.0025344, "grad_norm": 1.0880390405654907, "learning_rate": 1.6970273367650296e-05, "loss": 2.9219, "step": 99790 }, { "epoch": 0.00256, "grad_norm": 0.6877667903900146, "learning_rate": 1.6969695924505577e-05, "loss": 2.8864, "step": 99800 }, { "epoch": 0.0025856, "grad_norm": 0.6721354722976685, "learning_rate": 1.6969118436164237e-05, "loss": 2.9661, "step": 99810 }, { "epoch": 0.0026112, "grad_norm": 0.966087818145752, "learning_rate": 1.696854090263001e-05, "loss": 3.3978, "step": 99820 }, { "epoch": 0.0026368, "grad_norm": 0.7637101411819458, "learning_rate": 1.6967963323906652e-05, "loss": 3.0741, "step": 99830 }, { "epoch": 0.0026624, "grad_norm": 0.7768253087997437, "learning_rate": 1.6967385699997903e-05, "loss": 3.0077, "step": 99840 }, { "epoch": 0.002688, "grad_norm": 0.7368304133415222, "learning_rate": 1.696680803090751e-05, "loss": 3.3268, "step": 99850 }, { "epoch": 0.0027136, "grad_norm": 0.8768184185028076, "learning_rate": 1.6966230316639217e-05, "loss": 3.1955, "step": 99860 }, { "epoch": 0.0027392, "grad_norm": 0.7054524421691895, "learning_rate": 1.6965652557196773e-05, "loss": 3.1507, "step": 99870 }, { "epoch": 0.0027648, "grad_norm": 0.7056463360786438, "learning_rate": 1.6965074752583922e-05, "loss": 3.1288, "step": 99880 }, { "epoch": 0.0027904, "grad_norm": 1.1445828676223755, "learning_rate": 1.6964496902804414e-05, "loss": 3.1137, "step": 99890 }, { "epoch": 0.002816, "grad_norm": 0.7144861817359924, "learning_rate": 1.6963919007861994e-05, "loss": 3.1519, "step": 99900 }, { "epoch": 0.0028416, "grad_norm": 0.6871825456619263, "learning_rate": 1.696334106776041e-05, "loss": 3.1614, "step": 99910 }, { "epoch": 0.0028672, "grad_norm": 0.7334980368614197, "learning_rate": 1.696276308250341e-05, "loss": 3.0333, "step": 99920 }, { "epoch": 0.0028928, "grad_norm": 0.7073168158531189, "learning_rate": 1.696218505209474e-05, "loss": 3.1213, "step": 99930 }, { "epoch": 0.0029184, "grad_norm": 0.8232581615447998, "learning_rate": 1.6961606976538152e-05, "loss": 3.0756, "step": 99940 }, { "epoch": 0.002944, "grad_norm": 0.7553401589393616, "learning_rate": 1.6961028855837396e-05, "loss": 3.0322, "step": 99950 }, { "epoch": 0.0029696, "grad_norm": 0.7228344082832336, "learning_rate": 1.696045068999621e-05, "loss": 3.2244, "step": 99960 }, { "epoch": 0.0029952, "grad_norm": 0.7688562870025635, "learning_rate": 1.695987247901836e-05, "loss": 3.3296, "step": 99970 }, { "epoch": 0.0030208, "grad_norm": 0.8430259823799133, "learning_rate": 1.695929422290758e-05, "loss": 3.0772, "step": 99980 }, { "epoch": 0.0030464, "grad_norm": 0.6958354115486145, "learning_rate": 1.6958715921667628e-05, "loss": 3.0623, "step": 99990 }, { "epoch": 0.003072, "grad_norm": 1.1448267698287964, "learning_rate": 1.6958137575302254e-05, "loss": 3.1928, "step": 100000 }, { "epoch": 0.0030976, "grad_norm": 1.0324716567993164, "learning_rate": 1.695755918381521e-05, "loss": 3.0673, "step": 100010 }, { "epoch": 0.0031232, "grad_norm": 0.8100910186767578, "learning_rate": 1.695698074721024e-05, "loss": 3.0938, "step": 100020 }, { "epoch": 0.0031488, "grad_norm": 0.663480818271637, "learning_rate": 1.6956402265491098e-05, "loss": 3.4074, "step": 100030 }, { "epoch": 0.0031744, "grad_norm": 0.7617153525352478, "learning_rate": 1.6955823738661537e-05, "loss": 3.0215, "step": 100040 }, { "epoch": 0.0032, "grad_norm": 0.9109271764755249, "learning_rate": 1.695524516672531e-05, "loss": 3.0757, "step": 100050 }, { "epoch": 0.0032256, "grad_norm": 0.84029221534729, "learning_rate": 1.6954666549686164e-05, "loss": 3.1944, "step": 100060 }, { "epoch": 0.0032512, "grad_norm": 2.1981678009033203, "learning_rate": 1.6954087887547856e-05, "loss": 3.1273, "step": 100070 }, { "epoch": 0.0032768, "grad_norm": 0.7153274416923523, "learning_rate": 1.6953509180314132e-05, "loss": 3.0191, "step": 100080 }, { "epoch": 0.0033024, "grad_norm": 0.688018262386322, "learning_rate": 1.695293042798875e-05, "loss": 3.0262, "step": 100090 }, { "epoch": 0.003328, "grad_norm": 1.308377981185913, "learning_rate": 1.6952351630575466e-05, "loss": 3.2111, "step": 100100 }, { "epoch": 0.0033536, "grad_norm": 0.6384870409965515, "learning_rate": 1.6951772788078027e-05, "loss": 3.0112, "step": 100110 }, { "epoch": 0.0033792, "grad_norm": 0.7070585489273071, "learning_rate": 1.695119390050019e-05, "loss": 2.889, "step": 100120 }, { "epoch": 0.0034048, "grad_norm": 0.7044341564178467, "learning_rate": 1.6950614967845705e-05, "loss": 3.2095, "step": 100130 }, { "epoch": 0.0034304, "grad_norm": 0.671585202217102, "learning_rate": 1.6950035990118335e-05, "loss": 3.1585, "step": 100140 }, { "epoch": 0.003456, "grad_norm": 0.6967851519584656, "learning_rate": 1.6949456967321825e-05, "loss": 2.932, "step": 100150 }, { "epoch": 0.0034816, "grad_norm": 0.9354735612869263, "learning_rate": 1.6948877899459933e-05, "loss": 3.0171, "step": 100160 }, { "epoch": 0.0035072, "grad_norm": 0.6882855296134949, "learning_rate": 1.6948298786536414e-05, "loss": 2.9751, "step": 100170 }, { "epoch": 0.0035328, "grad_norm": 0.88937908411026, "learning_rate": 1.6947719628555024e-05, "loss": 3.0718, "step": 100180 }, { "epoch": 0.0035584, "grad_norm": 0.6748154163360596, "learning_rate": 1.694714042551952e-05, "loss": 3.1232, "step": 100190 }, { "epoch": 0.003584, "grad_norm": 0.8080770969390869, "learning_rate": 1.6946561177433654e-05, "loss": 3.279, "step": 100200 }, { "epoch": 0.0036096, "grad_norm": 0.7948331832885742, "learning_rate": 1.694598188430119e-05, "loss": 3.0254, "step": 100210 }, { "epoch": 0.0036352, "grad_norm": 0.806746244430542, "learning_rate": 1.6945402546125876e-05, "loss": 3.1486, "step": 100220 }, { "epoch": 0.0036608, "grad_norm": 0.672913670539856, "learning_rate": 1.6944823162911473e-05, "loss": 3.1757, "step": 100230 }, { "epoch": 0.0036864, "grad_norm": 0.8498513698577881, "learning_rate": 1.694424373466174e-05, "loss": 3.0875, "step": 100240 }, { "epoch": 0.003712, "grad_norm": 0.9512403607368469, "learning_rate": 1.694366426138043e-05, "loss": 3.1515, "step": 100250 }, { "epoch": 0.0037376, "grad_norm": 0.7967090606689453, "learning_rate": 1.69430847430713e-05, "loss": 3.1348, "step": 100260 }, { "epoch": 0.0037632, "grad_norm": 0.7842018008232117, "learning_rate": 1.6942505179738114e-05, "loss": 3.2369, "step": 100270 }, { "epoch": 0.0037888, "grad_norm": 0.7401222586631775, "learning_rate": 1.6941925571384628e-05, "loss": 3.1409, "step": 100280 }, { "epoch": 0.0038144, "grad_norm": 0.7301095128059387, "learning_rate": 1.6941345918014596e-05, "loss": 3.1239, "step": 100290 }, { "epoch": 0.00384, "grad_norm": 0.6834807395935059, "learning_rate": 1.6940766219631783e-05, "loss": 2.915, "step": 100300 }, { "epoch": 0.0038656, "grad_norm": 0.7180691957473755, "learning_rate": 1.6940186476239944e-05, "loss": 3.1395, "step": 100310 }, { "epoch": 0.0038912, "grad_norm": 0.8326249718666077, "learning_rate": 1.693960668784284e-05, "loss": 3.0704, "step": 100320 }, { "epoch": 0.0039168, "grad_norm": 0.9473055005073547, "learning_rate": 1.6939026854444234e-05, "loss": 2.8581, "step": 100330 }, { "epoch": 0.0039424, "grad_norm": 0.8426802754402161, "learning_rate": 1.693844697604788e-05, "loss": 3.0958, "step": 100340 }, { "epoch": 0.003968, "grad_norm": 0.7271156311035156, "learning_rate": 1.693786705265754e-05, "loss": 3.0989, "step": 100350 }, { "epoch": 0.0039936, "grad_norm": 0.7108110785484314, "learning_rate": 1.6937287084276982e-05, "loss": 3.1257, "step": 100360 }, { "epoch": 0.0040192, "grad_norm": 1.3530642986297607, "learning_rate": 1.6936707070909955e-05, "loss": 3.2879, "step": 100370 }, { "epoch": 0.0040448, "grad_norm": 0.8053974509239197, "learning_rate": 1.693612701256023e-05, "loss": 3.1226, "step": 100380 }, { "epoch": 0.0040704, "grad_norm": 0.6412279605865479, "learning_rate": 1.6935546909231562e-05, "loss": 3.1574, "step": 100390 }, { "epoch": 0.004096, "grad_norm": 1.0126521587371826, "learning_rate": 1.6934966760927718e-05, "loss": 3.091, "step": 100400 }, { "epoch": 0.0041216, "grad_norm": 0.8357797861099243, "learning_rate": 1.693438656765246e-05, "loss": 3.1747, "step": 100410 }, { "epoch": 0.0041472, "grad_norm": 0.6363036632537842, "learning_rate": 1.6933806329409545e-05, "loss": 2.9688, "step": 100420 }, { "epoch": 0.0041728, "grad_norm": 0.8797812461853027, "learning_rate": 1.6933226046202734e-05, "loss": 3.1311, "step": 100430 }, { "epoch": 0.0041984, "grad_norm": 0.642212450504303, "learning_rate": 1.6932645718035797e-05, "loss": 3.0674, "step": 100440 }, { "epoch": 0.004224, "grad_norm": 0.823810338973999, "learning_rate": 1.69320653449125e-05, "loss": 3.1583, "step": 100450 }, { "epoch": 0.0042496, "grad_norm": 0.8749567270278931, "learning_rate": 1.69314849268366e-05, "loss": 3.0678, "step": 100460 }, { "epoch": 0.0042752, "grad_norm": 2.786731719970703, "learning_rate": 1.693090446381186e-05, "loss": 3.2291, "step": 100470 }, { "epoch": 0.0043008, "grad_norm": 0.7038099765777588, "learning_rate": 1.6930323955842047e-05, "loss": 3.2278, "step": 100480 }, { "epoch": 0.0043264, "grad_norm": 0.804652750492096, "learning_rate": 1.6929743402930926e-05, "loss": 3.0748, "step": 100490 }, { "epoch": 0.004352, "grad_norm": 0.7959933280944824, "learning_rate": 1.692916280508226e-05, "loss": 3.398, "step": 100500 }, { "epoch": 2.56e-05, "grad_norm": 0.803730309009552, "learning_rate": 1.6928582162299817e-05, "loss": 2.3709, "step": 100510 }, { "epoch": 5.12e-05, "grad_norm": 0.9472833275794983, "learning_rate": 1.6928001474587357e-05, "loss": 3.1207, "step": 100520 }, { "epoch": 7.68e-05, "grad_norm": 0.833762526512146, "learning_rate": 1.692742074194865e-05, "loss": 3.1025, "step": 100530 }, { "epoch": 0.0001024, "grad_norm": 0.7162473797798157, "learning_rate": 1.692683996438746e-05, "loss": 2.911, "step": 100540 }, { "epoch": 0.000128, "grad_norm": 0.710305392742157, "learning_rate": 1.6926259141907552e-05, "loss": 2.927, "step": 100550 }, { "epoch": 0.0001536, "grad_norm": 0.6812201738357544, "learning_rate": 1.6925678274512698e-05, "loss": 3.0246, "step": 100560 }, { "epoch": 0.0001792, "grad_norm": 0.6857778429985046, "learning_rate": 1.692509736220666e-05, "loss": 2.9907, "step": 100570 }, { "epoch": 0.0002048, "grad_norm": 0.9966495633125305, "learning_rate": 1.6924516404993206e-05, "loss": 3.083, "step": 100580 }, { "epoch": 0.0002304, "grad_norm": 0.7355807423591614, "learning_rate": 1.6923935402876105e-05, "loss": 3.0953, "step": 100590 }, { "epoch": 0.000256, "grad_norm": 1.0155274868011475, "learning_rate": 1.692335435585912e-05, "loss": 2.8122, "step": 100600 }, { "epoch": 0.0002816, "grad_norm": 0.7951746582984924, "learning_rate": 1.692277326394602e-05, "loss": 2.9187, "step": 100610 }, { "epoch": 0.0003072, "grad_norm": 0.7869424819946289, "learning_rate": 1.6922192127140577e-05, "loss": 2.8377, "step": 100620 }, { "epoch": 0.0003328, "grad_norm": 0.8516090512275696, "learning_rate": 1.6921610945446563e-05, "loss": 3.2098, "step": 100630 }, { "epoch": 0.0003584, "grad_norm": 0.8560510277748108, "learning_rate": 1.6921029718867737e-05, "loss": 2.9904, "step": 100640 }, { "epoch": 0.000384, "grad_norm": 0.7775896191596985, "learning_rate": 1.6920448447407874e-05, "loss": 2.81, "step": 100650 }, { "epoch": 0.0004096, "grad_norm": 0.7276194095611572, "learning_rate": 1.691986713107074e-05, "loss": 2.9153, "step": 100660 }, { "epoch": 0.0004352, "grad_norm": 0.7641254663467407, "learning_rate": 1.691928576986011e-05, "loss": 2.8449, "step": 100670 }, { "epoch": 0.0004608, "grad_norm": 0.8887467980384827, "learning_rate": 1.6918704363779748e-05, "loss": 3.0711, "step": 100680 }, { "epoch": 0.0004864, "grad_norm": 0.7844114303588867, "learning_rate": 1.6918122912833424e-05, "loss": 3.1154, "step": 100690 }, { "epoch": 0.000512, "grad_norm": 1.0836045742034912, "learning_rate": 1.6917541417024915e-05, "loss": 2.9739, "step": 100700 }, { "epoch": 0.0005376, "grad_norm": 0.7436509728431702, "learning_rate": 1.6916959876357992e-05, "loss": 2.9279, "step": 100710 }, { "epoch": 0.0005632, "grad_norm": 0.6669607162475586, "learning_rate": 1.691637829083642e-05, "loss": 3.0491, "step": 100720 }, { "epoch": 0.0005888, "grad_norm": 0.6831357479095459, "learning_rate": 1.691579666046397e-05, "loss": 3.0009, "step": 100730 }, { "epoch": 0.0006144, "grad_norm": 0.6615148186683655, "learning_rate": 1.691521498524442e-05, "loss": 2.8758, "step": 100740 }, { "epoch": 0.00064, "grad_norm": 0.7930908799171448, "learning_rate": 1.6914633265181536e-05, "loss": 3.0785, "step": 100750 }, { "epoch": 0.0006656, "grad_norm": 0.8062642216682434, "learning_rate": 1.6914051500279096e-05, "loss": 2.944, "step": 100760 }, { "epoch": 0.0006912, "grad_norm": 0.7986634969711304, "learning_rate": 1.6913469690540868e-05, "loss": 3.2469, "step": 100770 }, { "epoch": 0.0007168, "grad_norm": 0.7041900753974915, "learning_rate": 1.6912887835970626e-05, "loss": 2.8597, "step": 100780 }, { "epoch": 0.0007424, "grad_norm": 1.1185377836227417, "learning_rate": 1.691230593657214e-05, "loss": 3.1008, "step": 100790 }, { "epoch": 0.000768, "grad_norm": 0.8337806463241577, "learning_rate": 1.6911723992349194e-05, "loss": 3.1711, "step": 100800 }, { "epoch": 0.0007936, "grad_norm": 0.7002528309822083, "learning_rate": 1.6911142003305552e-05, "loss": 2.7636, "step": 100810 }, { "epoch": 0.0008192, "grad_norm": 0.6557912826538086, "learning_rate": 1.691055996944499e-05, "loss": 3.0242, "step": 100820 }, { "epoch": 0.0008448, "grad_norm": 0.6133113503456116, "learning_rate": 1.6909977890771285e-05, "loss": 2.6722, "step": 100830 }, { "epoch": 0.0008704, "grad_norm": 0.7123347520828247, "learning_rate": 1.690939576728821e-05, "loss": 3.1703, "step": 100840 }, { "epoch": 0.000896, "grad_norm": 1.0122665166854858, "learning_rate": 1.6908813598999537e-05, "loss": 2.6159, "step": 100850 }, { "epoch": 0.0009216, "grad_norm": 1.4565989971160889, "learning_rate": 1.6908231385909048e-05, "loss": 3.6233, "step": 100860 }, { "epoch": 0.0009472, "grad_norm": 0.6558087468147278, "learning_rate": 1.690764912802051e-05, "loss": 2.778, "step": 100870 }, { "epoch": 0.0009728, "grad_norm": 0.7530174851417542, "learning_rate": 1.6907066825337706e-05, "loss": 3.0971, "step": 100880 }, { "epoch": 0.0009984, "grad_norm": 0.7676523327827454, "learning_rate": 1.6906484477864406e-05, "loss": 3.0877, "step": 100890 }, { "epoch": 0.001024, "grad_norm": 0.7503030300140381, "learning_rate": 1.6905902085604394e-05, "loss": 3.0468, "step": 100900 }, { "epoch": 2.56e-05, "grad_norm": 0.7873113751411438, "learning_rate": 1.690531964856144e-05, "loss": 2.5043, "step": 100910 }, { "epoch": 5.12e-05, "grad_norm": 0.8242897391319275, "learning_rate": 1.6904737166739322e-05, "loss": 3.0881, "step": 100920 }, { "epoch": 7.68e-05, "grad_norm": 1.755650281906128, "learning_rate": 1.690415464014182e-05, "loss": 3.1055, "step": 100930 }, { "epoch": 0.0001024, "grad_norm": 0.7058800458908081, "learning_rate": 1.6903572068772714e-05, "loss": 2.9748, "step": 100940 }, { "epoch": 0.000128, "grad_norm": 0.9411375522613525, "learning_rate": 1.6902989452635773e-05, "loss": 2.9576, "step": 100950 }, { "epoch": 0.0001536, "grad_norm": 0.7953975200653076, "learning_rate": 1.690240679173478e-05, "loss": 2.7997, "step": 100960 }, { "epoch": 0.0001792, "grad_norm": 0.8042258620262146, "learning_rate": 1.6901824086073517e-05, "loss": 3.0069, "step": 100970 }, { "epoch": 0.0002048, "grad_norm": 0.7461644411087036, "learning_rate": 1.6901241335655757e-05, "loss": 2.9646, "step": 100980 }, { "epoch": 0.0002304, "grad_norm": 0.7888522744178772, "learning_rate": 1.6900658540485277e-05, "loss": 2.9126, "step": 100990 }, { "epoch": 0.000256, "grad_norm": 0.7954078912734985, "learning_rate": 1.6900075700565866e-05, "loss": 2.7986, "step": 101000 }, { "epoch": 0.0002816, "grad_norm": 0.752295970916748, "learning_rate": 1.6899492815901296e-05, "loss": 2.8669, "step": 101010 }, { "epoch": 0.0003072, "grad_norm": 0.7481247782707214, "learning_rate": 1.6898909886495348e-05, "loss": 2.8283, "step": 101020 }, { "epoch": 0.0003328, "grad_norm": 0.725655198097229, "learning_rate": 1.6898326912351807e-05, "loss": 3.277, "step": 101030 }, { "epoch": 0.0003584, "grad_norm": 0.8134909272193909, "learning_rate": 1.6897743893474443e-05, "loss": 3.0277, "step": 101040 }, { "epoch": 0.000384, "grad_norm": 1.0427266359329224, "learning_rate": 1.6897160829867044e-05, "loss": 3.033, "step": 101050 }, { "epoch": 0.0004096, "grad_norm": 0.8399773240089417, "learning_rate": 1.6896577721533395e-05, "loss": 3.1402, "step": 101060 }, { "epoch": 0.0004352, "grad_norm": 0.6697103381156921, "learning_rate": 1.689599456847727e-05, "loss": 3.0385, "step": 101070 }, { "epoch": 0.0004608, "grad_norm": 0.7436553239822388, "learning_rate": 1.6895411370702452e-05, "loss": 2.8752, "step": 101080 }, { "epoch": 0.0004864, "grad_norm": 0.8440961241722107, "learning_rate": 1.6894828128212727e-05, "loss": 3.1351, "step": 101090 }, { "epoch": 0.000512, "grad_norm": 0.944381594657898, "learning_rate": 1.689424484101187e-05, "loss": 2.9246, "step": 101100 }, { "epoch": 0.0005376, "grad_norm": 0.7085240483283997, "learning_rate": 1.6893661509103668e-05, "loss": 3.1206, "step": 101110 }, { "epoch": 0.0005632, "grad_norm": 0.7742406129837036, "learning_rate": 1.6893078132491903e-05, "loss": 2.9771, "step": 101120 }, { "epoch": 0.0005888, "grad_norm": 0.6425790786743164, "learning_rate": 1.689249471118036e-05, "loss": 2.7108, "step": 101130 }, { "epoch": 0.0006144, "grad_norm": 0.8095271587371826, "learning_rate": 1.689191124517282e-05, "loss": 3.0592, "step": 101140 }, { "epoch": 0.00064, "grad_norm": 1.0190662145614624, "learning_rate": 1.6891327734473066e-05, "loss": 3.1619, "step": 101150 }, { "epoch": 0.0006656, "grad_norm": 0.734601616859436, "learning_rate": 1.6890744179084887e-05, "loss": 3.0531, "step": 101160 }, { "epoch": 0.0006912, "grad_norm": 0.6471226215362549, "learning_rate": 1.6890160579012058e-05, "loss": 2.9598, "step": 101170 }, { "epoch": 0.0007168, "grad_norm": 0.8228419423103333, "learning_rate": 1.688957693425837e-05, "loss": 3.1207, "step": 101180 }, { "epoch": 0.0007424, "grad_norm": 0.6496638059616089, "learning_rate": 1.688899324482761e-05, "loss": 2.9471, "step": 101190 }, { "epoch": 0.000768, "grad_norm": 0.6597681641578674, "learning_rate": 1.6888409510723557e-05, "loss": 2.944, "step": 101200 }, { "epoch": 0.0007936, "grad_norm": 0.703763484954834, "learning_rate": 1.688782573195e-05, "loss": 2.9616, "step": 101210 }, { "epoch": 0.0008192, "grad_norm": 0.7097216248512268, "learning_rate": 1.6887241908510726e-05, "loss": 2.9523, "step": 101220 }, { "epoch": 0.0008448, "grad_norm": 0.6875361800193787, "learning_rate": 1.6886658040409513e-05, "loss": 3.0149, "step": 101230 }, { "epoch": 0.0008704, "grad_norm": 0.8261755108833313, "learning_rate": 1.688607412765016e-05, "loss": 2.8125, "step": 101240 }, { "epoch": 0.000896, "grad_norm": 0.7170066833496094, "learning_rate": 1.688549017023644e-05, "loss": 2.9982, "step": 101250 }, { "epoch": 0.0009216, "grad_norm": 0.9126941561698914, "learning_rate": 1.688490616817215e-05, "loss": 2.8578, "step": 101260 }, { "epoch": 0.0009472, "grad_norm": 0.7933796644210815, "learning_rate": 1.6884322121461073e-05, "loss": 3.0333, "step": 101270 }, { "epoch": 0.0009728, "grad_norm": 0.9739693999290466, "learning_rate": 1.6883738030106993e-05, "loss": 3.0795, "step": 101280 }, { "epoch": 0.0009984, "grad_norm": 0.8853882551193237, "learning_rate": 1.6883153894113704e-05, "loss": 3.151, "step": 101290 }, { "epoch": 0.001024, "grad_norm": 0.7789326310157776, "learning_rate": 1.688256971348499e-05, "loss": 3.0175, "step": 101300 }, { "epoch": 0.0010496, "grad_norm": 0.9743580222129822, "learning_rate": 1.6881985488224642e-05, "loss": 3.0155, "step": 101310 }, { "epoch": 0.0010752, "grad_norm": 0.7213042974472046, "learning_rate": 1.6881401218336447e-05, "loss": 2.9305, "step": 101320 }, { "epoch": 0.0011008, "grad_norm": 0.7572692632675171, "learning_rate": 1.6880816903824192e-05, "loss": 3.2296, "step": 101330 }, { "epoch": 0.0011264, "grad_norm": 0.6423254013061523, "learning_rate": 1.688023254469167e-05, "loss": 2.8151, "step": 101340 }, { "epoch": 0.001152, "grad_norm": 0.9206522107124329, "learning_rate": 1.6879648140942672e-05, "loss": 3.0663, "step": 101350 }, { "epoch": 0.0011776, "grad_norm": 0.9990029335021973, "learning_rate": 1.687906369258098e-05, "loss": 3.0915, "step": 101360 }, { "epoch": 0.0012032, "grad_norm": 1.0270309448242188, "learning_rate": 1.6878479199610385e-05, "loss": 3.2199, "step": 101370 }, { "epoch": 0.0012288, "grad_norm": 0.6818568110466003, "learning_rate": 1.6877894662034687e-05, "loss": 3.309, "step": 101380 }, { "epoch": 0.0012544, "grad_norm": 0.7350576519966125, "learning_rate": 1.6877310079857666e-05, "loss": 3.0768, "step": 101390 }, { "epoch": 0.00128, "grad_norm": 0.9463582038879395, "learning_rate": 1.687672545308312e-05, "loss": 2.997, "step": 101400 }, { "epoch": 0.0013056, "grad_norm": 0.7771509885787964, "learning_rate": 1.6876140781714834e-05, "loss": 2.8546, "step": 101410 }, { "epoch": 0.0013312, "grad_norm": 0.798899233341217, "learning_rate": 1.6875556065756605e-05, "loss": 2.9481, "step": 101420 }, { "epoch": 0.0013568, "grad_norm": 0.7708727121353149, "learning_rate": 1.687497130521222e-05, "loss": 3.0141, "step": 101430 }, { "epoch": 0.0013824, "grad_norm": 1.0625392198562622, "learning_rate": 1.6874386500085474e-05, "loss": 3.3294, "step": 101440 }, { "epoch": 0.001408, "grad_norm": 0.6692903637886047, "learning_rate": 1.687380165038016e-05, "loss": 2.9569, "step": 101450 }, { "epoch": 0.0014336, "grad_norm": 0.8496440052986145, "learning_rate": 1.6873216756100068e-05, "loss": 3.1044, "step": 101460 }, { "epoch": 0.0014592, "grad_norm": 1.3515148162841797, "learning_rate": 1.687263181724899e-05, "loss": 3.0971, "step": 101470 }, { "epoch": 0.0014848, "grad_norm": 0.6938154101371765, "learning_rate": 1.6872046833830725e-05, "loss": 2.8572, "step": 101480 }, { "epoch": 0.0015104, "grad_norm": 0.7095115780830383, "learning_rate": 1.687146180584906e-05, "loss": 2.6113, "step": 101490 }, { "epoch": 0.001536, "grad_norm": 0.759998619556427, "learning_rate": 1.6870876733307792e-05, "loss": 2.9561, "step": 101500 }, { "epoch": 0.0015616, "grad_norm": 0.7021715044975281, "learning_rate": 1.6870291616210715e-05, "loss": 2.8245, "step": 101510 }, { "epoch": 0.0015872, "grad_norm": 0.667808473110199, "learning_rate": 1.6869706454561622e-05, "loss": 3.0956, "step": 101520 }, { "epoch": 0.0016128, "grad_norm": 0.8087410926818848, "learning_rate": 1.686912124836431e-05, "loss": 2.9389, "step": 101530 }, { "epoch": 0.0016384, "grad_norm": 0.7640408873558044, "learning_rate": 1.686853599762257e-05, "loss": 3.2601, "step": 101540 }, { "epoch": 0.001664, "grad_norm": 0.6558118462562561, "learning_rate": 1.6867950702340203e-05, "loss": 2.9359, "step": 101550 }, { "epoch": 0.0016896, "grad_norm": 0.6702255010604858, "learning_rate": 1.6867365362520996e-05, "loss": 3.0167, "step": 101560 }, { "epoch": 0.0017152, "grad_norm": 0.8247698545455933, "learning_rate": 1.686677997816875e-05, "loss": 3.0312, "step": 101570 }, { "epoch": 0.0017408, "grad_norm": 0.6442550420761108, "learning_rate": 1.6866194549287264e-05, "loss": 3.0549, "step": 101580 }, { "epoch": 0.0017664, "grad_norm": 0.7168639898300171, "learning_rate": 1.6865609075880333e-05, "loss": 2.9539, "step": 101590 }, { "epoch": 0.001792, "grad_norm": 0.8133969306945801, "learning_rate": 1.6865023557951748e-05, "loss": 3.0058, "step": 101600 }, { "epoch": 0.0018176, "grad_norm": 0.6751573085784912, "learning_rate": 1.686443799550531e-05, "loss": 2.8594, "step": 101610 }, { "epoch": 0.0018432, "grad_norm": 0.7391154766082764, "learning_rate": 1.6863852388544816e-05, "loss": 2.7902, "step": 101620 }, { "epoch": 0.0018688, "grad_norm": 0.6517195105552673, "learning_rate": 1.6863266737074066e-05, "loss": 2.6508, "step": 101630 }, { "epoch": 0.0018944, "grad_norm": 0.7127683162689209, "learning_rate": 1.6862681041096854e-05, "loss": 2.9132, "step": 101640 }, { "epoch": 0.00192, "grad_norm": 0.6138467192649841, "learning_rate": 1.6862095300616976e-05, "loss": 3.0224, "step": 101650 }, { "epoch": 0.0019456, "grad_norm": 0.765409529209137, "learning_rate": 1.6861509515638238e-05, "loss": 3.1735, "step": 101660 }, { "epoch": 0.0019712, "grad_norm": 0.7361674308776855, "learning_rate": 1.6860923686164432e-05, "loss": 3.0025, "step": 101670 }, { "epoch": 0.0019968, "grad_norm": 0.6694597005844116, "learning_rate": 1.686033781219936e-05, "loss": 2.7898, "step": 101680 }, { "epoch": 0.0020224, "grad_norm": 0.689700186252594, "learning_rate": 1.685975189374682e-05, "loss": 3.0368, "step": 101690 }, { "epoch": 0.002048, "grad_norm": 0.7262183427810669, "learning_rate": 1.6859165930810615e-05, "loss": 3.021, "step": 101700 }, { "epoch": 0.0020736, "grad_norm": 0.8602458834648132, "learning_rate": 1.6858579923394537e-05, "loss": 3.0921, "step": 101710 }, { "epoch": 0.0020992, "grad_norm": 0.788913905620575, "learning_rate": 1.6857993871502395e-05, "loss": 2.5869, "step": 101720 }, { "epoch": 0.0021248, "grad_norm": 0.762214720249176, "learning_rate": 1.6857407775137985e-05, "loss": 2.9589, "step": 101730 }, { "epoch": 0.0021504, "grad_norm": 0.8238015174865723, "learning_rate": 1.6856821634305103e-05, "loss": 2.8732, "step": 101740 }, { "epoch": 0.002176, "grad_norm": 0.8244814872741699, "learning_rate": 1.6856235449007563e-05, "loss": 3.0844, "step": 101750 }, { "epoch": 0.0022016, "grad_norm": 0.849177360534668, "learning_rate": 1.6855649219249155e-05, "loss": 2.8994, "step": 101760 }, { "epoch": 0.0022272, "grad_norm": 0.7375616431236267, "learning_rate": 1.6855062945033683e-05, "loss": 3.0532, "step": 101770 }, { "epoch": 0.0022528, "grad_norm": 0.9684115052223206, "learning_rate": 1.685447662636495e-05, "loss": 3.1169, "step": 101780 }, { "epoch": 0.0022784, "grad_norm": 0.6942684650421143, "learning_rate": 1.6853890263246758e-05, "loss": 2.8914, "step": 101790 }, { "epoch": 0.002304, "grad_norm": 0.8138571381568909, "learning_rate": 1.6853303855682907e-05, "loss": 2.9586, "step": 101800 }, { "epoch": 0.0023296, "grad_norm": 0.8319132328033447, "learning_rate": 1.6852717403677202e-05, "loss": 2.7924, "step": 101810 }, { "epoch": 0.0023552, "grad_norm": 0.7084259390830994, "learning_rate": 1.6852130907233452e-05, "loss": 2.9013, "step": 101820 }, { "epoch": 0.0023808, "grad_norm": 0.7911053895950317, "learning_rate": 1.6851544366355444e-05, "loss": 3.0036, "step": 101830 }, { "epoch": 0.0024064, "grad_norm": 1.2115453481674194, "learning_rate": 1.6850957781046998e-05, "loss": 2.7099, "step": 101840 }, { "epoch": 0.002432, "grad_norm": 0.7776544690132141, "learning_rate": 1.6850371151311914e-05, "loss": 3.0476, "step": 101850 }, { "epoch": 0.0024576, "grad_norm": 0.7156400680541992, "learning_rate": 1.6849784477153993e-05, "loss": 3.1797, "step": 101860 }, { "epoch": 0.0024832, "grad_norm": 0.9164809584617615, "learning_rate": 1.6849197758577035e-05, "loss": 2.9565, "step": 101870 }, { "epoch": 0.0025088, "grad_norm": 0.8372583389282227, "learning_rate": 1.6848610995584854e-05, "loss": 2.9179, "step": 101880 }, { "epoch": 0.0025344, "grad_norm": 0.9544363617897034, "learning_rate": 1.684802418818125e-05, "loss": 3.092, "step": 101890 }, { "epoch": 0.00256, "grad_norm": 0.7393370270729065, "learning_rate": 1.684743733637003e-05, "loss": 3.0524, "step": 101900 }, { "epoch": 0.0025856, "grad_norm": 0.7984206080436707, "learning_rate": 1.6846850440154995e-05, "loss": 2.9332, "step": 101910 }, { "epoch": 0.0026112, "grad_norm": 0.7256454825401306, "learning_rate": 1.6846263499539955e-05, "loss": 2.9781, "step": 101920 }, { "epoch": 0.0026368, "grad_norm": 0.9179227948188782, "learning_rate": 1.684567651452872e-05, "loss": 2.8089, "step": 101930 }, { "epoch": 0.0026624, "grad_norm": 0.7283985614776611, "learning_rate": 1.684508948512509e-05, "loss": 3.3035, "step": 101940 }, { "epoch": 0.002688, "grad_norm": 0.8347048759460449, "learning_rate": 1.6844502411332874e-05, "loss": 2.9158, "step": 101950 }, { "epoch": 0.0027136, "grad_norm": 0.9507935047149658, "learning_rate": 1.6843915293155878e-05, "loss": 2.9391, "step": 101960 }, { "epoch": 0.0027392, "grad_norm": 0.9919911623001099, "learning_rate": 1.684332813059791e-05, "loss": 3.07, "step": 101970 }, { "epoch": 0.0027648, "grad_norm": 0.7494273781776428, "learning_rate": 1.6842740923662778e-05, "loss": 2.9807, "step": 101980 }, { "epoch": 0.0027904, "grad_norm": 0.7005533576011658, "learning_rate": 1.684215367235429e-05, "loss": 2.9947, "step": 101990 }, { "epoch": 0.002816, "grad_norm": 0.7118828892707825, "learning_rate": 1.684156637667625e-05, "loss": 3.3009, "step": 102000 }, { "epoch": 0.0028416, "grad_norm": 0.6853107213973999, "learning_rate": 1.684097903663248e-05, "loss": 3.2008, "step": 102010 }, { "epoch": 0.0028672, "grad_norm": 0.9027234315872192, "learning_rate": 1.684039165222677e-05, "loss": 3.2235, "step": 102020 }, { "epoch": 0.0028928, "grad_norm": 0.7291406989097595, "learning_rate": 1.6839804223462947e-05, "loss": 3.2304, "step": 102030 }, { "epoch": 0.0029184, "grad_norm": 0.6749722957611084, "learning_rate": 1.6839216750344804e-05, "loss": 3.0484, "step": 102040 }, { "epoch": 0.002944, "grad_norm": 0.6636192202568054, "learning_rate": 1.6838629232876162e-05, "loss": 3.1506, "step": 102050 }, { "epoch": 0.0029696, "grad_norm": 0.6856165528297424, "learning_rate": 1.683804167106082e-05, "loss": 2.8007, "step": 102060 }, { "epoch": 0.0029952, "grad_norm": 0.8029345870018005, "learning_rate": 1.6837454064902602e-05, "loss": 3.1207, "step": 102070 }, { "epoch": 0.0030208, "grad_norm": 0.7205519676208496, "learning_rate": 1.6836866414405308e-05, "loss": 2.947, "step": 102080 }, { "epoch": 0.0030464, "grad_norm": 0.7678897380828857, "learning_rate": 1.6836278719572757e-05, "loss": 3.1322, "step": 102090 }, { "epoch": 0.003072, "grad_norm": 0.893763542175293, "learning_rate": 1.683569098040875e-05, "loss": 2.942, "step": 102100 }, { "epoch": 0.0030976, "grad_norm": 0.7744585275650024, "learning_rate": 1.6835103196917107e-05, "loss": 3.2307, "step": 102110 }, { "epoch": 0.0031232, "grad_norm": 0.7871280908584595, "learning_rate": 1.6834515369101638e-05, "loss": 3.1267, "step": 102120 }, { "epoch": 0.0031488, "grad_norm": 0.8477934002876282, "learning_rate": 1.683392749696615e-05, "loss": 3.2407, "step": 102130 }, { "epoch": 0.0031744, "grad_norm": 0.6836605072021484, "learning_rate": 1.6833339580514458e-05, "loss": 2.9993, "step": 102140 }, { "epoch": 0.0032, "grad_norm": 0.7234534025192261, "learning_rate": 1.6832751619750377e-05, "loss": 3.039, "step": 102150 }, { "epoch": 0.0032256, "grad_norm": 0.9768186211585999, "learning_rate": 1.6832163614677715e-05, "loss": 3.1733, "step": 102160 }, { "epoch": 0.0032512, "grad_norm": 0.8178269863128662, "learning_rate": 1.683157556530029e-05, "loss": 3.145, "step": 102170 }, { "epoch": 0.0032768, "grad_norm": 0.7219456434249878, "learning_rate": 1.683098747162191e-05, "loss": 2.9446, "step": 102180 }, { "epoch": 0.0033024, "grad_norm": 0.7662034034729004, "learning_rate": 1.6830399333646398e-05, "loss": 3.1805, "step": 102190 }, { "epoch": 0.003328, "grad_norm": 0.7610232830047607, "learning_rate": 1.6829811151377554e-05, "loss": 3.1069, "step": 102200 }, { "epoch": 0.0033536, "grad_norm": 0.8846200108528137, "learning_rate": 1.6829222924819207e-05, "loss": 2.9981, "step": 102210 }, { "epoch": 0.0033792, "grad_norm": 0.7709002494812012, "learning_rate": 1.682863465397516e-05, "loss": 2.971, "step": 102220 }, { "epoch": 0.0034048, "grad_norm": 0.8078203201293945, "learning_rate": 1.6828046338849233e-05, "loss": 3.031, "step": 102230 }, { "epoch": 0.0034304, "grad_norm": 0.8732191920280457, "learning_rate": 1.682745797944524e-05, "loss": 2.9918, "step": 102240 }, { "epoch": 0.003456, "grad_norm": 0.8497092723846436, "learning_rate": 1.6826869575767e-05, "loss": 3.2547, "step": 102250 }, { "epoch": 0.0034816, "grad_norm": 0.6874967217445374, "learning_rate": 1.682628112781832e-05, "loss": 3.1667, "step": 102260 }, { "epoch": 0.0035072, "grad_norm": 0.6998644471168518, "learning_rate": 1.6825692635603022e-05, "loss": 3.0394, "step": 102270 }, { "epoch": 0.0035328, "grad_norm": 0.7745400071144104, "learning_rate": 1.6825104099124922e-05, "loss": 3.1779, "step": 102280 }, { "epoch": 0.0035584, "grad_norm": 0.6872131824493408, "learning_rate": 1.6824515518387835e-05, "loss": 2.8684, "step": 102290 }, { "epoch": 0.003584, "grad_norm": 1.1507188081741333, "learning_rate": 1.682392689339558e-05, "loss": 2.888, "step": 102300 }, { "epoch": 0.0036096, "grad_norm": 0.7270156145095825, "learning_rate": 1.682333822415197e-05, "loss": 3.467, "step": 102310 }, { "epoch": 0.0036352, "grad_norm": 0.7379317283630371, "learning_rate": 1.682274951066083e-05, "loss": 3.2504, "step": 102320 }, { "epoch": 0.0036608, "grad_norm": 0.8546040654182434, "learning_rate": 1.682216075292597e-05, "loss": 3.1296, "step": 102330 }, { "epoch": 0.0036864, "grad_norm": 1.0141510963439941, "learning_rate": 1.682157195095121e-05, "loss": 3.2148, "step": 102340 }, { "epoch": 0.003712, "grad_norm": 0.7357850074768066, "learning_rate": 1.682098310474037e-05, "loss": 3.1378, "step": 102350 }, { "epoch": 0.0037376, "grad_norm": 0.7332456111907959, "learning_rate": 1.6820394214297265e-05, "loss": 3.0709, "step": 102360 }, { "epoch": 0.0037632, "grad_norm": 0.7305366396903992, "learning_rate": 1.681980527962572e-05, "loss": 3.2271, "step": 102370 }, { "epoch": 0.0037888, "grad_norm": 0.7140345573425293, "learning_rate": 1.6819216300729548e-05, "loss": 3.1933, "step": 102380 }, { "epoch": 0.0038144, "grad_norm": 0.7559223175048828, "learning_rate": 1.681862727761257e-05, "loss": 3.1051, "step": 102390 }, { "epoch": 0.00384, "grad_norm": 0.8653836250305176, "learning_rate": 1.681803821027861e-05, "loss": 2.9792, "step": 102400 }, { "epoch": 0.0038656, "grad_norm": 0.6992835402488708, "learning_rate": 1.6817449098731482e-05, "loss": 3.1215, "step": 102410 }, { "epoch": 0.0038912, "grad_norm": 0.7681947350502014, "learning_rate": 1.6816859942975007e-05, "loss": 2.8528, "step": 102420 }, { "epoch": 0.0039168, "grad_norm": 0.7107990384101868, "learning_rate": 1.6816270743013007e-05, "loss": 3.0012, "step": 102430 }, { "epoch": 0.0039424, "grad_norm": 0.789132833480835, "learning_rate": 1.6815681498849304e-05, "loss": 3.3124, "step": 102440 }, { "epoch": 0.003968, "grad_norm": 0.8084405064582825, "learning_rate": 1.6815092210487717e-05, "loss": 3.1634, "step": 102450 }, { "epoch": 0.0039936, "grad_norm": 0.7896605134010315, "learning_rate": 1.6814502877932067e-05, "loss": 3.0315, "step": 102460 }, { "epoch": 0.0040192, "grad_norm": 0.7915164232254028, "learning_rate": 1.681391350118618e-05, "loss": 3.0176, "step": 102470 }, { "epoch": 0.0040448, "grad_norm": 0.6713359951972961, "learning_rate": 1.6813324080253874e-05, "loss": 3.0781, "step": 102480 }, { "epoch": 0.0040704, "grad_norm": 0.7046420574188232, "learning_rate": 1.681273461513897e-05, "loss": 3.1087, "step": 102490 }, { "epoch": 0.004096, "grad_norm": 0.8192359209060669, "learning_rate": 1.6812145105845296e-05, "loss": 3.1445, "step": 102500 }, { "epoch": 0.0041216, "grad_norm": 0.7569969296455383, "learning_rate": 1.681155555237667e-05, "loss": 3.1446, "step": 102510 }, { "epoch": 0.0041472, "grad_norm": 1.1535406112670898, "learning_rate": 1.6810965954736917e-05, "loss": 3.1302, "step": 102520 }, { "epoch": 0.0041728, "grad_norm": 0.8250330090522766, "learning_rate": 1.681037631292986e-05, "loss": 3.1348, "step": 102530 }, { "epoch": 0.0041984, "grad_norm": 0.6731181144714355, "learning_rate": 1.680978662695932e-05, "loss": 3.1355, "step": 102540 }, { "epoch": 0.004224, "grad_norm": 0.8172406554222107, "learning_rate": 1.6809196896829124e-05, "loss": 3.0963, "step": 102550 }, { "epoch": 0.0042496, "grad_norm": 0.7477667927742004, "learning_rate": 1.68086071225431e-05, "loss": 3.3106, "step": 102560 }, { "epoch": 0.0042752, "grad_norm": 0.9956827759742737, "learning_rate": 1.6808017304105064e-05, "loss": 3.0991, "step": 102570 }, { "epoch": 0.0043008, "grad_norm": 0.6664271950721741, "learning_rate": 1.6807427441518846e-05, "loss": 3.0364, "step": 102580 }, { "epoch": 0.0043264, "grad_norm": 0.7560567259788513, "learning_rate": 1.6806837534788272e-05, "loss": 2.9667, "step": 102590 }, { "epoch": 0.004352, "grad_norm": 0.7713939547538757, "learning_rate": 1.6806247583917162e-05, "loss": 3.2805, "step": 102600 }, { "epoch": 0.0043776, "grad_norm": 0.7418460249900818, "learning_rate": 1.6805657588909346e-05, "loss": 3.2541, "step": 102610 }, { "epoch": 0.0044032, "grad_norm": 0.730593204498291, "learning_rate": 1.6805067549768655e-05, "loss": 2.9459, "step": 102620 }, { "epoch": 0.0044288, "grad_norm": 0.7119333744049072, "learning_rate": 1.6804477466498905e-05, "loss": 3.038, "step": 102630 }, { "epoch": 0.0044544, "grad_norm": 0.6930179595947266, "learning_rate": 1.680388733910393e-05, "loss": 3.0968, "step": 102640 }, { "epoch": 0.00448, "grad_norm": 0.7177897691726685, "learning_rate": 1.680329716758755e-05, "loss": 3.11, "step": 102650 }, { "epoch": 0.0045056, "grad_norm": 0.8098289966583252, "learning_rate": 1.6802706951953598e-05, "loss": 3.0782, "step": 102660 }, { "epoch": 0.0045312, "grad_norm": 0.7060684561729431, "learning_rate": 1.68021166922059e-05, "loss": 3.2602, "step": 102670 }, { "epoch": 0.0045568, "grad_norm": 0.7373763918876648, "learning_rate": 1.680152638834828e-05, "loss": 3.1129, "step": 102680 }, { "epoch": 0.0045824, "grad_norm": 1.0581979751586914, "learning_rate": 1.680093604038457e-05, "loss": 3.2222, "step": 102690 }, { "epoch": 0.004608, "grad_norm": 0.7401909232139587, "learning_rate": 1.6800345648318597e-05, "loss": 2.8993, "step": 102700 }, { "epoch": 0.0046336, "grad_norm": 0.74149489402771, "learning_rate": 1.6799755212154192e-05, "loss": 3.0097, "step": 102710 }, { "epoch": 0.0046592, "grad_norm": 0.789126992225647, "learning_rate": 1.6799164731895178e-05, "loss": 3.1035, "step": 102720 }, { "epoch": 0.0046848, "grad_norm": 0.778826117515564, "learning_rate": 1.6798574207545394e-05, "loss": 3.0044, "step": 102730 }, { "epoch": 0.0047104, "grad_norm": 0.6856095790863037, "learning_rate": 1.679798363910866e-05, "loss": 3.0313, "step": 102740 }, { "epoch": 0.004736, "grad_norm": 0.7959555983543396, "learning_rate": 1.6797393026588805e-05, "loss": 3.2074, "step": 102750 }, { "epoch": 0.0047616, "grad_norm": 0.8224661946296692, "learning_rate": 1.6796802369989667e-05, "loss": 3.3189, "step": 102760 }, { "epoch": 0.0047872, "grad_norm": 0.7859742045402527, "learning_rate": 1.679621166931507e-05, "loss": 3.1435, "step": 102770 }, { "epoch": 0.0048128, "grad_norm": 0.7146328687667847, "learning_rate": 1.6795620924568846e-05, "loss": 3.1968, "step": 102780 }, { "epoch": 0.0048384, "grad_norm": 0.7444461584091187, "learning_rate": 1.679503013575483e-05, "loss": 3.0731, "step": 102790 }, { "epoch": 0.004864, "grad_norm": 1.465753436088562, "learning_rate": 1.6794439302876847e-05, "loss": 3.0326, "step": 102800 }, { "epoch": 0.0048896, "grad_norm": 0.855754554271698, "learning_rate": 1.679384842593873e-05, "loss": 3.0249, "step": 102810 }, { "epoch": 0.0049152, "grad_norm": 0.8902108073234558, "learning_rate": 1.6793257504944312e-05, "loss": 2.91, "step": 102820 }, { "epoch": 0.0049408, "grad_norm": 0.7141739130020142, "learning_rate": 1.6792666539897428e-05, "loss": 2.9093, "step": 102830 }, { "epoch": 0.0049664, "grad_norm": 0.7125675678253174, "learning_rate": 1.6792075530801904e-05, "loss": 3.0031, "step": 102840 }, { "epoch": 0.004992, "grad_norm": 0.6690157055854797, "learning_rate": 1.6791484477661574e-05, "loss": 2.9664, "step": 102850 }, { "epoch": 0.0050176, "grad_norm": 0.6908385157585144, "learning_rate": 1.6790893380480274e-05, "loss": 3.1132, "step": 102860 }, { "epoch": 0.0050432, "grad_norm": 0.679703414440155, "learning_rate": 1.6790302239261833e-05, "loss": 3.1049, "step": 102870 }, { "epoch": 0.0050688, "grad_norm": 0.8050076365470886, "learning_rate": 1.6789711054010086e-05, "loss": 3.2767, "step": 102880 }, { "epoch": 0.0050944, "grad_norm": 0.8206665515899658, "learning_rate": 1.6789119824728873e-05, "loss": 3.1511, "step": 102890 }, { "epoch": 0.00512, "grad_norm": 0.7182747721672058, "learning_rate": 1.6788528551422017e-05, "loss": 3.233, "step": 102900 }, { "epoch": 0.0051456, "grad_norm": 0.7968989014625549, "learning_rate": 1.6787937234093357e-05, "loss": 3.2691, "step": 102910 }, { "epoch": 0.0051712, "grad_norm": 0.6615909337997437, "learning_rate": 1.6787345872746734e-05, "loss": 3.1023, "step": 102920 }, { "epoch": 0.0051968, "grad_norm": 0.8011590838432312, "learning_rate": 1.6786754467385974e-05, "loss": 3.1815, "step": 102930 }, { "epoch": 0.0052224, "grad_norm": 0.8775741457939148, "learning_rate": 1.6786163018014914e-05, "loss": 2.8714, "step": 102940 }, { "epoch": 0.005248, "grad_norm": 0.7297837138175964, "learning_rate": 1.6785571524637386e-05, "loss": 3.1205, "step": 102950 }, { "epoch": 0.0052736, "grad_norm": 0.7274687886238098, "learning_rate": 1.6784979987257238e-05, "loss": 3.1853, "step": 102960 }, { "epoch": 0.0052992, "grad_norm": 0.7219583988189697, "learning_rate": 1.6784388405878295e-05, "loss": 3.1667, "step": 102970 }, { "epoch": 0.0053248, "grad_norm": 0.796226441860199, "learning_rate": 1.6783796780504397e-05, "loss": 2.8579, "step": 102980 }, { "epoch": 0.0053504, "grad_norm": 0.7759775519371033, "learning_rate": 1.678320511113938e-05, "loss": 3.1801, "step": 102990 }, { "epoch": 0.005376, "grad_norm": 0.7211409211158752, "learning_rate": 1.6782613397787075e-05, "loss": 3.1048, "step": 103000 }, { "epoch": 0.0054016, "grad_norm": 0.6621485948562622, "learning_rate": 1.678202164045133e-05, "loss": 3.0456, "step": 103010 }, { "epoch": 0.0054272, "grad_norm": 0.7613874673843384, "learning_rate": 1.6781429839135975e-05, "loss": 3.075, "step": 103020 }, { "epoch": 0.0054528, "grad_norm": 0.8392891883850098, "learning_rate": 1.6780837993844853e-05, "loss": 3.0542, "step": 103030 }, { "epoch": 0.0054784, "grad_norm": 0.7583670616149902, "learning_rate": 1.6780246104581794e-05, "loss": 3.0296, "step": 103040 }, { "epoch": 0.005504, "grad_norm": 0.8714288473129272, "learning_rate": 1.6779654171350644e-05, "loss": 3.2619, "step": 103050 }, { "epoch": 0.0055296, "grad_norm": 1.4821523427963257, "learning_rate": 1.677906219415524e-05, "loss": 3.1419, "step": 103060 }, { "epoch": 0.0055552, "grad_norm": 0.7569463849067688, "learning_rate": 1.677847017299942e-05, "loss": 3.0729, "step": 103070 }, { "epoch": 0.0055808, "grad_norm": 0.7758924961090088, "learning_rate": 1.6777878107887017e-05, "loss": 3.0118, "step": 103080 }, { "epoch": 0.0056064, "grad_norm": 0.8627049922943115, "learning_rate": 1.6777285998821877e-05, "loss": 3.1442, "step": 103090 }, { "epoch": 0.005632, "grad_norm": 0.7590058445930481, "learning_rate": 1.6776693845807842e-05, "loss": 3.1562, "step": 103100 }, { "epoch": 0.0056576, "grad_norm": 0.7216605544090271, "learning_rate": 1.677610164884875e-05, "loss": 2.9145, "step": 103110 }, { "epoch": 0.0056832, "grad_norm": 0.7602929472923279, "learning_rate": 1.6775509407948435e-05, "loss": 2.8905, "step": 103120 }, { "epoch": 0.0057088, "grad_norm": 0.7499889731407166, "learning_rate": 1.677491712311074e-05, "loss": 3.0359, "step": 103130 }, { "epoch": 0.0057344, "grad_norm": 0.7096875905990601, "learning_rate": 1.6774324794339514e-05, "loss": 3.115, "step": 103140 }, { "epoch": 0.00576, "grad_norm": 0.7564333081245422, "learning_rate": 1.6773732421638587e-05, "loss": 3.2372, "step": 103150 }, { "epoch": 0.0057856, "grad_norm": 1.4056017398834229, "learning_rate": 1.677314000501181e-05, "loss": 3.0555, "step": 103160 }, { "epoch": 0.0058112, "grad_norm": 0.8401923179626465, "learning_rate": 1.6772547544463018e-05, "loss": 2.9996, "step": 103170 }, { "epoch": 0.0058368, "grad_norm": 0.9470986723899841, "learning_rate": 1.6771955039996054e-05, "loss": 2.9877, "step": 103180 }, { "epoch": 0.0058624, "grad_norm": 0.674449622631073, "learning_rate": 1.6771362491614764e-05, "loss": 2.9921, "step": 103190 }, { "epoch": 0.005888, "grad_norm": 0.739863395690918, "learning_rate": 1.677076989932299e-05, "loss": 3.096, "step": 103200 }, { "epoch": 0.0059136, "grad_norm": 0.8923059701919556, "learning_rate": 1.6770177263124566e-05, "loss": 3.0551, "step": 103210 }, { "epoch": 0.0059392, "grad_norm": 0.7736741304397583, "learning_rate": 1.6769584583023344e-05, "loss": 3.0289, "step": 103220 }, { "epoch": 0.0059648, "grad_norm": 0.7046627998352051, "learning_rate": 1.6768991859023166e-05, "loss": 3.2827, "step": 103230 }, { "epoch": 0.0059904, "grad_norm": 0.900969386100769, "learning_rate": 1.6768399091127874e-05, "loss": 3.0549, "step": 103240 }, { "epoch": 0.006016, "grad_norm": 0.7713605165481567, "learning_rate": 1.6767806279341314e-05, "loss": 3.1449, "step": 103250 }, { "epoch": 0.0060416, "grad_norm": 1.1830971240997314, "learning_rate": 1.676721342366733e-05, "loss": 3.3166, "step": 103260 }, { "epoch": 0.0060672, "grad_norm": 0.7166346311569214, "learning_rate": 1.6766620524109765e-05, "loss": 3.0898, "step": 103270 }, { "epoch": 0.0060928, "grad_norm": 0.7154086232185364, "learning_rate": 1.6766027580672465e-05, "loss": 2.7998, "step": 103280 }, { "epoch": 0.0061184, "grad_norm": 0.7066528797149658, "learning_rate": 1.6765434593359268e-05, "loss": 2.9699, "step": 103290 }, { "epoch": 0.006144, "grad_norm": 0.7618634700775146, "learning_rate": 1.676484156217403e-05, "loss": 2.967, "step": 103300 }, { "epoch": 0.0061696, "grad_norm": 0.8142610788345337, "learning_rate": 1.6764248487120594e-05, "loss": 3.0798, "step": 103310 }, { "epoch": 0.0061952, "grad_norm": 0.6484953165054321, "learning_rate": 1.6763655368202803e-05, "loss": 3.8883, "step": 103320 }, { "epoch": 0.0062208, "grad_norm": 0.8155924677848816, "learning_rate": 1.6763062205424505e-05, "loss": 3.0971, "step": 103330 }, { "epoch": 0.0062464, "grad_norm": 0.7469926476478577, "learning_rate": 1.6762468998789545e-05, "loss": 3.106, "step": 103340 }, { "epoch": 0.006272, "grad_norm": 1.1747523546218872, "learning_rate": 1.676187574830177e-05, "loss": 3.059, "step": 103350 }, { "epoch": 0.0062976, "grad_norm": 0.7000374794006348, "learning_rate": 1.676128245396503e-05, "loss": 3.287, "step": 103360 }, { "epoch": 0.0063232, "grad_norm": 0.6467869281768799, "learning_rate": 1.676068911578317e-05, "loss": 3.0671, "step": 103370 }, { "epoch": 0.0063488, "grad_norm": 0.764281690120697, "learning_rate": 1.6760095733760036e-05, "loss": 3.1527, "step": 103380 }, { "epoch": 0.0063744, "grad_norm": 0.7734543681144714, "learning_rate": 1.675950230789948e-05, "loss": 3.1697, "step": 103390 }, { "epoch": 0.0064, "grad_norm": 0.7486507892608643, "learning_rate": 1.6758908838205344e-05, "loss": 3.0284, "step": 103400 }, { "epoch": 0.0064256, "grad_norm": 0.6921297311782837, "learning_rate": 1.6758315324681484e-05, "loss": 3.0277, "step": 103410 }, { "epoch": 0.0064512, "grad_norm": 0.7393685579299927, "learning_rate": 1.6757721767331746e-05, "loss": 2.9376, "step": 103420 }, { "epoch": 0.0064768, "grad_norm": 0.8429552912712097, "learning_rate": 1.675712816615998e-05, "loss": 3.0129, "step": 103430 }, { "epoch": 0.0065024, "grad_norm": 0.9054605960845947, "learning_rate": 1.675653452117003e-05, "loss": 3.079, "step": 103440 }, { "epoch": 0.006528, "grad_norm": 0.6488808393478394, "learning_rate": 1.6755940832365748e-05, "loss": 3.1282, "step": 103450 }, { "epoch": 0.0065536, "grad_norm": 0.8119668960571289, "learning_rate": 1.6755347099750988e-05, "loss": 3.1089, "step": 103460 }, { "epoch": 0.0065792, "grad_norm": 0.7500519156455994, "learning_rate": 1.6754753323329595e-05, "loss": 3.0657, "step": 103470 }, { "epoch": 0.0066048, "grad_norm": 0.7248701453208923, "learning_rate": 1.6754159503105425e-05, "loss": 3.136, "step": 103480 }, { "epoch": 0.0066304, "grad_norm": 1.001979947090149, "learning_rate": 1.6753565639082326e-05, "loss": 2.9843, "step": 103490 }, { "epoch": 0.006656, "grad_norm": 1.9089066982269287, "learning_rate": 1.6752971731264144e-05, "loss": 3.1652, "step": 103500 }, { "epoch": 0.0066816, "grad_norm": 0.8155856132507324, "learning_rate": 1.675237777965474e-05, "loss": 3.1797, "step": 103510 }, { "epoch": 0.0067072, "grad_norm": 0.8499515652656555, "learning_rate": 1.6751783784257954e-05, "loss": 3.1358, "step": 103520 }, { "epoch": 0.0067328, "grad_norm": 0.8228253126144409, "learning_rate": 1.675118974507765e-05, "loss": 3.058, "step": 103530 }, { "epoch": 0.0067584, "grad_norm": 0.8294529318809509, "learning_rate": 1.6750595662117677e-05, "loss": 3.1681, "step": 103540 }, { "epoch": 0.006784, "grad_norm": 0.7997361421585083, "learning_rate": 1.6750001535381882e-05, "loss": 3.0489, "step": 103550 }, { "epoch": 0.0068096, "grad_norm": 0.6802389621734619, "learning_rate": 1.674940736487412e-05, "loss": 3.1809, "step": 103560 }, { "epoch": 0.0068352, "grad_norm": 0.9100749492645264, "learning_rate": 1.6748813150598247e-05, "loss": 2.8947, "step": 103570 }, { "epoch": 0.0068608, "grad_norm": 0.8264862895011902, "learning_rate": 1.674821889255811e-05, "loss": 2.9789, "step": 103580 }, { "epoch": 0.0068864, "grad_norm": 0.7491734027862549, "learning_rate": 1.6747624590757574e-05, "loss": 3.0402, "step": 103590 }, { "epoch": 0.006912, "grad_norm": 0.9617857933044434, "learning_rate": 1.6747030245200483e-05, "loss": 3.1179, "step": 103600 }, { "epoch": 0.0069376, "grad_norm": 0.7197037935256958, "learning_rate": 1.674643585589069e-05, "loss": 3.1078, "step": 103610 }, { "epoch": 0.0069632, "grad_norm": 0.9310845136642456, "learning_rate": 1.6745841422832058e-05, "loss": 3.0963, "step": 103620 }, { "epoch": 0.0069888, "grad_norm": 0.6748798489570618, "learning_rate": 1.674524694602844e-05, "loss": 2.9456, "step": 103630 }, { "epoch": 0.0070144, "grad_norm": 0.9637258648872375, "learning_rate": 1.6744652425483682e-05, "loss": 3.1261, "step": 103640 }, { "epoch": 0.00704, "grad_norm": 0.6945666074752808, "learning_rate": 1.6744057861201647e-05, "loss": 3.0988, "step": 103650 }, { "epoch": 0.0070656, "grad_norm": 0.9198399186134338, "learning_rate": 1.6743463253186187e-05, "loss": 3.369, "step": 103660 }, { "epoch": 0.0070912, "grad_norm": 0.718863844871521, "learning_rate": 1.6742868601441163e-05, "loss": 3.0062, "step": 103670 }, { "epoch": 0.0071168, "grad_norm": 0.7182088494300842, "learning_rate": 1.6742273905970428e-05, "loss": 2.9476, "step": 103680 }, { "epoch": 0.0071424, "grad_norm": 0.8156671524047852, "learning_rate": 1.674167916677784e-05, "loss": 3.2553, "step": 103690 }, { "epoch": 0.007168, "grad_norm": 0.7837710380554199, "learning_rate": 1.6741084383867252e-05, "loss": 3.0632, "step": 103700 }, { "epoch": 0.0071936, "grad_norm": 0.7757209539413452, "learning_rate": 1.674048955724253e-05, "loss": 3.1818, "step": 103710 }, { "epoch": 0.0072192, "grad_norm": 0.9576120972633362, "learning_rate": 1.6739894686907514e-05, "loss": 3.4911, "step": 103720 }, { "epoch": 0.0072448, "grad_norm": 0.8017457127571106, "learning_rate": 1.6739299772866076e-05, "loss": 3.0716, "step": 103730 }, { "epoch": 0.0072704, "grad_norm": 0.762948751449585, "learning_rate": 1.6738704815122072e-05, "loss": 3.1332, "step": 103740 }, { "epoch": 0.007296, "grad_norm": 0.7024556398391724, "learning_rate": 1.673810981367936e-05, "loss": 3.1875, "step": 103750 }, { "epoch": 0.0073216, "grad_norm": 0.7251921892166138, "learning_rate": 1.673751476854179e-05, "loss": 3.1787, "step": 103760 }, { "epoch": 0.0073472, "grad_norm": 2.1496806144714355, "learning_rate": 1.6736919679713234e-05, "loss": 3.0411, "step": 103770 }, { "epoch": 0.0073728, "grad_norm": 0.7230221033096313, "learning_rate": 1.673632454719754e-05, "loss": 3.1309, "step": 103780 }, { "epoch": 0.0073984, "grad_norm": 0.8280667066574097, "learning_rate": 1.6735729370998573e-05, "loss": 3.051, "step": 103790 }, { "epoch": 0.007424, "grad_norm": 0.8365049958229065, "learning_rate": 1.6735134151120193e-05, "loss": 3.0473, "step": 103800 }, { "epoch": 0.0074496, "grad_norm": 0.7872888445854187, "learning_rate": 1.6734538887566257e-05, "loss": 3.0381, "step": 103810 }, { "epoch": 0.0074752, "grad_norm": 0.8260585069656372, "learning_rate": 1.6733943580340626e-05, "loss": 3.0755, "step": 103820 }, { "epoch": 0.0075008, "grad_norm": 0.8879576921463013, "learning_rate": 1.6733348229447157e-05, "loss": 3.1291, "step": 103830 }, { "epoch": 0.0075264, "grad_norm": 1.8440907001495361, "learning_rate": 1.673275283488972e-05, "loss": 3.3368, "step": 103840 }, { "epoch": 0.007552, "grad_norm": 0.7139707207679749, "learning_rate": 1.6732157396672168e-05, "loss": 3.1852, "step": 103850 }, { "epoch": 0.0075776, "grad_norm": 0.6813387870788574, "learning_rate": 1.6731561914798362e-05, "loss": 2.7467, "step": 103860 }, { "epoch": 0.0076032, "grad_norm": 0.7114912271499634, "learning_rate": 1.6730966389272168e-05, "loss": 3.0338, "step": 103870 }, { "epoch": 0.0076288, "grad_norm": 0.7152712941169739, "learning_rate": 1.6730370820097448e-05, "loss": 3.2842, "step": 103880 }, { "epoch": 0.0076544, "grad_norm": 0.8281227946281433, "learning_rate": 1.6729775207278058e-05, "loss": 3.2116, "step": 103890 }, { "epoch": 0.00768, "grad_norm": 1.2824581861495972, "learning_rate": 1.6729179550817867e-05, "loss": 3.0802, "step": 103900 }, { "epoch": 0.0077056, "grad_norm": 0.7881393432617188, "learning_rate": 1.6728583850720733e-05, "loss": 2.9103, "step": 103910 }, { "epoch": 0.0077312, "grad_norm": 0.7370342016220093, "learning_rate": 1.6727988106990524e-05, "loss": 3.137, "step": 103920 }, { "epoch": 0.0077568, "grad_norm": 0.7545586824417114, "learning_rate": 1.6727392319631098e-05, "loss": 2.9839, "step": 103930 }, { "epoch": 0.0077824, "grad_norm": 0.829042375087738, "learning_rate": 1.672679648864632e-05, "loss": 3.1746, "step": 103940 }, { "epoch": 0.007808, "grad_norm": 0.8095447421073914, "learning_rate": 1.6726200614040058e-05, "loss": 3.1697, "step": 103950 }, { "epoch": 0.0078336, "grad_norm": 0.7730352282524109, "learning_rate": 1.672560469581617e-05, "loss": 3.0902, "step": 103960 }, { "epoch": 0.0078592, "grad_norm": 0.7899196743965149, "learning_rate": 1.6725008733978523e-05, "loss": 3.2022, "step": 103970 }, { "epoch": 0.0078848, "grad_norm": 1.7695155143737793, "learning_rate": 1.672441272853098e-05, "loss": 2.9945, "step": 103980 }, { "epoch": 0.0079104, "grad_norm": 0.6421588659286499, "learning_rate": 1.672381667947741e-05, "loss": 3.1273, "step": 103990 }, { "epoch": 0.007936, "grad_norm": 2.3838648796081543, "learning_rate": 1.6723220586821675e-05, "loss": 3.2653, "step": 104000 }, { "epoch": 0.0079616, "grad_norm": 0.6797937154769897, "learning_rate": 1.672262445056764e-05, "loss": 3.176, "step": 104010 }, { "epoch": 0.0079872, "grad_norm": 0.7004227638244629, "learning_rate": 1.6722028270719177e-05, "loss": 3.1823, "step": 104020 }, { "epoch": 0.0080128, "grad_norm": 0.9490231871604919, "learning_rate": 1.6721432047280146e-05, "loss": 3.3293, "step": 104030 }, { "epoch": 0.0080384, "grad_norm": 0.7854272127151489, "learning_rate": 1.672083578025441e-05, "loss": 3.3121, "step": 104040 }, { "epoch": 0.008064, "grad_norm": 0.7301101088523865, "learning_rate": 1.6720239469645844e-05, "loss": 2.9819, "step": 104050 }, { "epoch": 0.0080896, "grad_norm": 1.1137744188308716, "learning_rate": 1.6719643115458308e-05, "loss": 3.1206, "step": 104060 }, { "epoch": 0.0081152, "grad_norm": 2.696157455444336, "learning_rate": 1.6719046717695674e-05, "loss": 2.9646, "step": 104070 }, { "epoch": 0.0081408, "grad_norm": 0.859769880771637, "learning_rate": 1.6718450276361808e-05, "loss": 3.2009, "step": 104080 }, { "epoch": 0.0081664, "grad_norm": 1.0473946332931519, "learning_rate": 1.6717853791460575e-05, "loss": 3.1091, "step": 104090 }, { "epoch": 0.008192, "grad_norm": 0.7893844842910767, "learning_rate": 1.671725726299585e-05, "loss": 2.9899, "step": 104100 }, { "epoch": 0.0082176, "grad_norm": 0.6977869272232056, "learning_rate": 1.6716660690971493e-05, "loss": 3.1741, "step": 104110 }, { "epoch": 0.0082432, "grad_norm": 0.7026382684707642, "learning_rate": 1.671606407539138e-05, "loss": 3.1823, "step": 104120 }, { "epoch": 0.0082688, "grad_norm": 0.7281402349472046, "learning_rate": 1.6715467416259376e-05, "loss": 3.251, "step": 104130 }, { "epoch": 0.0082944, "grad_norm": 0.7610952258110046, "learning_rate": 1.6714870713579348e-05, "loss": 3.1689, "step": 104140 }, { "epoch": 0.00832, "grad_norm": 0.7215638756752014, "learning_rate": 1.671427396735517e-05, "loss": 3.219, "step": 104150 }, { "epoch": 0.0083456, "grad_norm": 0.6757287979125977, "learning_rate": 1.6713677177590706e-05, "loss": 3.1176, "step": 104160 }, { "epoch": 0.0083712, "grad_norm": 0.6338397264480591, "learning_rate": 1.6713080344289833e-05, "loss": 3.0693, "step": 104170 }, { "epoch": 0.0083968, "grad_norm": 0.7200393080711365, "learning_rate": 1.671248346745642e-05, "loss": 3.1242, "step": 104180 }, { "epoch": 0.0084224, "grad_norm": 0.6622839570045471, "learning_rate": 1.6711886547094332e-05, "loss": 3.2039, "step": 104190 }, { "epoch": 0.008448, "grad_norm": 0.6706569194793701, "learning_rate": 1.6711289583207446e-05, "loss": 2.8521, "step": 104200 }, { "epoch": 0.0084736, "grad_norm": 0.8509972095489502, "learning_rate": 1.671069257579963e-05, "loss": 2.9208, "step": 104210 }, { "epoch": 0.0084992, "grad_norm": 0.7265480160713196, "learning_rate": 1.6710095524874754e-05, "loss": 3.1002, "step": 104220 }, { "epoch": 0.0085248, "grad_norm": 0.8403474688529968, "learning_rate": 1.6709498430436693e-05, "loss": 3.2491, "step": 104230 }, { "epoch": 0.0085504, "grad_norm": 0.6987869143486023, "learning_rate": 1.670890129248932e-05, "loss": 3.3237, "step": 104240 }, { "epoch": 0.008576, "grad_norm": 0.7011886239051819, "learning_rate": 1.6708304111036504e-05, "loss": 3.1044, "step": 104250 }, { "epoch": 0.0086016, "grad_norm": 0.7245693206787109, "learning_rate": 1.6707706886082117e-05, "loss": 3.2255, "step": 104260 }, { "epoch": 0.0086272, "grad_norm": 0.7112624049186707, "learning_rate": 1.6707109617630034e-05, "loss": 3.4219, "step": 104270 }, { "epoch": 0.0086528, "grad_norm": 0.7260971069335938, "learning_rate": 1.6706512305684127e-05, "loss": 3.1373, "step": 104280 }, { "epoch": 0.0086784, "grad_norm": 0.8275826573371887, "learning_rate": 1.6705914950248273e-05, "loss": 3.0709, "step": 104290 }, { "epoch": 0.008704, "grad_norm": 0.8091362714767456, "learning_rate": 1.6705317551326343e-05, "loss": 3.177, "step": 104300 }, { "epoch": 0.0087296, "grad_norm": 0.8331357836723328, "learning_rate": 1.6704720108922207e-05, "loss": 3.1292, "step": 104310 }, { "epoch": 0.0087552, "grad_norm": 0.6280345916748047, "learning_rate": 1.6704122623039748e-05, "loss": 2.9537, "step": 104320 }, { "epoch": 0.0087808, "grad_norm": 0.7342812418937683, "learning_rate": 1.6703525093682832e-05, "loss": 2.9785, "step": 104330 }, { "epoch": 0.0088064, "grad_norm": 0.7073127031326294, "learning_rate": 1.6702927520855338e-05, "loss": 3.18, "step": 104340 }, { "epoch": 0.008832, "grad_norm": 0.7302444577217102, "learning_rate": 1.670232990456114e-05, "loss": 3.1007, "step": 104350 }, { "epoch": 0.0088576, "grad_norm": 0.8175151348114014, "learning_rate": 1.6701732244804113e-05, "loss": 3.0561, "step": 104360 }, { "epoch": 0.0088832, "grad_norm": 0.7492915391921997, "learning_rate": 1.6701134541588135e-05, "loss": 3.1059, "step": 104370 }, { "epoch": 0.0089088, "grad_norm": 0.7376521825790405, "learning_rate": 1.6700536794917082e-05, "loss": 3.0411, "step": 104380 }, { "epoch": 0.0089344, "grad_norm": 0.6950404644012451, "learning_rate": 1.6699939004794826e-05, "loss": 3.152, "step": 104390 }, { "epoch": 0.00896, "grad_norm": 0.8531895279884338, "learning_rate": 1.6699341171225247e-05, "loss": 3.0099, "step": 104400 }, { "epoch": 2.56e-05, "grad_norm": 0.824535608291626, "learning_rate": 1.669874329421222e-05, "loss": 2.987, "step": 104410 }, { "epoch": 5.12e-05, "grad_norm": 0.6980196237564087, "learning_rate": 1.6698145373759626e-05, "loss": 2.8744, "step": 104420 }, { "epoch": 7.68e-05, "grad_norm": 0.9902951717376709, "learning_rate": 1.6697547409871338e-05, "loss": 3.2396, "step": 104430 }, { "epoch": 0.0001024, "grad_norm": 0.6995611786842346, "learning_rate": 1.6696949402551236e-05, "loss": 3.1318, "step": 104440 }, { "epoch": 0.000128, "grad_norm": 0.9268028140068054, "learning_rate": 1.669635135180319e-05, "loss": 2.9703, "step": 104450 }, { "epoch": 0.0001536, "grad_norm": 0.6832133531570435, "learning_rate": 1.6695753257631093e-05, "loss": 2.8435, "step": 104460 }, { "epoch": 0.0001792, "grad_norm": 0.8817864060401917, "learning_rate": 1.669515512003881e-05, "loss": 2.9551, "step": 104470 }, { "epoch": 0.0002048, "grad_norm": 0.7132088541984558, "learning_rate": 1.6694556939030233e-05, "loss": 2.8192, "step": 104480 }, { "epoch": 0.0002304, "grad_norm": 0.6958026885986328, "learning_rate": 1.6693958714609227e-05, "loss": 2.8869, "step": 104490 }, { "epoch": 0.000256, "grad_norm": 0.7643750905990601, "learning_rate": 1.669336044677968e-05, "loss": 3.1696, "step": 104500 }, { "epoch": 0.0002816, "grad_norm": 0.809762716293335, "learning_rate": 1.669276213554547e-05, "loss": 2.9279, "step": 104510 }, { "epoch": 0.0003072, "grad_norm": 0.6765033006668091, "learning_rate": 1.6692163780910477e-05, "loss": 2.8182, "step": 104520 }, { "epoch": 0.0003328, "grad_norm": 0.872916579246521, "learning_rate": 1.6691565382878576e-05, "loss": 2.9039, "step": 104530 }, { "epoch": 0.0003584, "grad_norm": 0.6917176246643066, "learning_rate": 1.669096694145366e-05, "loss": 3.0793, "step": 104540 }, { "epoch": 0.000384, "grad_norm": 0.8285713791847229, "learning_rate": 1.6690368456639593e-05, "loss": 3.0405, "step": 104550 }, { "epoch": 0.0004096, "grad_norm": 0.6566458344459534, "learning_rate": 1.668976992844027e-05, "loss": 2.7903, "step": 104560 }, { "epoch": 0.0004352, "grad_norm": 0.8782397508621216, "learning_rate": 1.6689171356859565e-05, "loss": 3.2483, "step": 104570 }, { "epoch": 0.0004608, "grad_norm": 0.7308146953582764, "learning_rate": 1.668857274190136e-05, "loss": 2.9837, "step": 104580 }, { "epoch": 0.0004864, "grad_norm": 0.7618383169174194, "learning_rate": 1.6687974083569544e-05, "loss": 2.8761, "step": 104590 }, { "epoch": 0.000512, "grad_norm": 0.7926284074783325, "learning_rate": 1.6687375381867986e-05, "loss": 2.9418, "step": 104600 }, { "epoch": 0.0005376, "grad_norm": 0.7982240319252014, "learning_rate": 1.6686776636800578e-05, "loss": 3.11, "step": 104610 }, { "epoch": 0.0005632, "grad_norm": 0.7474143505096436, "learning_rate": 1.6686177848371202e-05, "loss": 2.9039, "step": 104620 }, { "epoch": 0.0005888, "grad_norm": 0.6887418031692505, "learning_rate": 1.668557901658374e-05, "loss": 2.8799, "step": 104630 }, { "epoch": 0.0006144, "grad_norm": 0.6390653848648071, "learning_rate": 1.6684980141442077e-05, "loss": 2.9785, "step": 104640 }, { "epoch": 0.00064, "grad_norm": 0.7898656129837036, "learning_rate": 1.668438122295009e-05, "loss": 3.0679, "step": 104650 }, { "epoch": 0.0006656, "grad_norm": 0.9034022688865662, "learning_rate": 1.668378226111167e-05, "loss": 2.8798, "step": 104660 }, { "epoch": 0.0006912, "grad_norm": 0.6781902313232422, "learning_rate": 1.6683183255930697e-05, "loss": 3.3898, "step": 104670 }, { "epoch": 0.0007168, "grad_norm": 0.7491949200630188, "learning_rate": 1.668258420741106e-05, "loss": 3.0676, "step": 104680 }, { "epoch": 0.0007424, "grad_norm": 0.8076890707015991, "learning_rate": 1.6681985115556635e-05, "loss": 3.2005, "step": 104690 }, { "epoch": 0.000768, "grad_norm": 0.6948144435882568, "learning_rate": 1.6681385980371317e-05, "loss": 3.0172, "step": 104700 }, { "epoch": 0.0007936, "grad_norm": 0.6864469647407532, "learning_rate": 1.6680786801858982e-05, "loss": 3.0621, "step": 104710 }, { "epoch": 0.0008192, "grad_norm": 1.203776240348816, "learning_rate": 1.6680187580023525e-05, "loss": 2.8535, "step": 104720 }, { "epoch": 0.0008448, "grad_norm": 0.6595858931541443, "learning_rate": 1.6679588314868827e-05, "loss": 2.7447, "step": 104730 }, { "epoch": 0.0008704, "grad_norm": 0.771040678024292, "learning_rate": 1.6678989006398772e-05, "loss": 2.9073, "step": 104740 }, { "epoch": 0.000896, "grad_norm": 0.6404945850372314, "learning_rate": 1.6678389654617246e-05, "loss": 2.879, "step": 104750 }, { "epoch": 0.0009216, "grad_norm": 0.8048907518386841, "learning_rate": 1.667779025952814e-05, "loss": 2.7377, "step": 104760 }, { "epoch": 0.0009472, "grad_norm": 1.0195080041885376, "learning_rate": 1.667719082113534e-05, "loss": 3.2088, "step": 104770 }, { "epoch": 0.0009728, "grad_norm": 0.6793295741081238, "learning_rate": 1.6676591339442732e-05, "loss": 2.926, "step": 104780 }, { "epoch": 0.0009984, "grad_norm": 0.6951155662536621, "learning_rate": 1.6675991814454203e-05, "loss": 2.937, "step": 104790 }, { "epoch": 0.001024, "grad_norm": 0.7106765508651733, "learning_rate": 1.667539224617364e-05, "loss": 2.9776, "step": 104800 }, { "epoch": 0.0010496, "grad_norm": 0.8045181035995483, "learning_rate": 1.6674792634604938e-05, "loss": 3.1575, "step": 104810 }, { "epoch": 0.0010752, "grad_norm": 0.7765724658966064, "learning_rate": 1.6674192979751973e-05, "loss": 3.0519, "step": 104820 }, { "epoch": 0.0011008, "grad_norm": 0.6976145505905151, "learning_rate": 1.6673593281618644e-05, "loss": 3.0538, "step": 104830 }, { "epoch": 0.0011264, "grad_norm": 0.7085605263710022, "learning_rate": 1.6672993540208834e-05, "loss": 2.629, "step": 104840 }, { "epoch": 0.001152, "grad_norm": 0.78253573179245, "learning_rate": 1.6672393755526434e-05, "loss": 2.9489, "step": 104850 }, { "epoch": 0.0011776, "grad_norm": 0.724158763885498, "learning_rate": 1.6671793927575338e-05, "loss": 3.2228, "step": 104860 }, { "epoch": 0.0012032, "grad_norm": 0.9008442163467407, "learning_rate": 1.667119405635943e-05, "loss": 2.9363, "step": 104870 }, { "epoch": 0.0012288, "grad_norm": 0.7091097235679626, "learning_rate": 1.6670594141882596e-05, "loss": 3.2321, "step": 104880 }, { "epoch": 0.0012544, "grad_norm": 0.6999587416648865, "learning_rate": 1.6669994184148735e-05, "loss": 3.1085, "step": 104890 }, { "epoch": 0.00128, "grad_norm": 0.838652491569519, "learning_rate": 1.6669394183161737e-05, "loss": 2.7884, "step": 104900 }, { "epoch": 0.0013056, "grad_norm": 0.7008952498435974, "learning_rate": 1.666879413892549e-05, "loss": 2.8697, "step": 104910 }, { "epoch": 0.0013312, "grad_norm": 0.7328675985336304, "learning_rate": 1.6668194051443883e-05, "loss": 2.7832, "step": 104920 }, { "epoch": 0.0013568, "grad_norm": 1.0264819860458374, "learning_rate": 1.666759392072081e-05, "loss": 3.0942, "step": 104930 }, { "epoch": 0.0013824, "grad_norm": 0.7276491522789001, "learning_rate": 1.6666993746760166e-05, "loss": 3.0035, "step": 104940 }, { "epoch": 0.001408, "grad_norm": 0.6541268825531006, "learning_rate": 1.6666393529565833e-05, "loss": 3.0166, "step": 104950 }, { "epoch": 0.0014336, "grad_norm": 1.1801389455795288, "learning_rate": 1.6665793269141713e-05, "loss": 3.0246, "step": 104960 }, { "epoch": 0.0014592, "grad_norm": 0.6762104034423828, "learning_rate": 1.6665192965491697e-05, "loss": 2.9022, "step": 104970 }, { "epoch": 0.0014848, "grad_norm": 0.7340885400772095, "learning_rate": 1.6664592618619676e-05, "loss": 2.7783, "step": 104980 }, { "epoch": 0.0015104, "grad_norm": 0.6688290238380432, "learning_rate": 1.666399222852954e-05, "loss": 2.5728, "step": 104990 }, { "epoch": 0.001536, "grad_norm": 1.1597552299499512, "learning_rate": 1.6663391795225186e-05, "loss": 2.9705, "step": 105000 }, { "epoch": 0.0015616, "grad_norm": 0.6779282093048096, "learning_rate": 1.666279131871051e-05, "loss": 2.9119, "step": 105010 }, { "epoch": 0.0015872, "grad_norm": 0.9765109419822693, "learning_rate": 1.66621907989894e-05, "loss": 2.9026, "step": 105020 }, { "epoch": 0.0016128, "grad_norm": 0.7480417490005493, "learning_rate": 1.6661590236065754e-05, "loss": 3.1438, "step": 105030 }, { "epoch": 0.0016384, "grad_norm": 0.7464301586151123, "learning_rate": 1.6660989629943467e-05, "loss": 3.1586, "step": 105040 }, { "epoch": 0.001664, "grad_norm": 0.7987822890281677, "learning_rate": 1.6660388980626433e-05, "loss": 2.9717, "step": 105050 }, { "epoch": 0.0016896, "grad_norm": 0.8060093522071838, "learning_rate": 1.6659788288118545e-05, "loss": 3.1456, "step": 105060 }, { "epoch": 0.0017152, "grad_norm": 0.8811111450195312, "learning_rate": 1.66591875524237e-05, "loss": 2.8336, "step": 105070 }, { "epoch": 0.0017408, "grad_norm": 0.7757031321525574, "learning_rate": 1.6658586773545794e-05, "loss": 3.0199, "step": 105080 }, { "epoch": 0.0017664, "grad_norm": 0.8914138674736023, "learning_rate": 1.6657985951488723e-05, "loss": 2.8783, "step": 105090 }, { "epoch": 0.001792, "grad_norm": 0.8840019106864929, "learning_rate": 1.665738508625638e-05, "loss": 2.9914, "step": 105100 }, { "epoch": 0.0018176, "grad_norm": 0.7215061187744141, "learning_rate": 1.6656784177852666e-05, "loss": 2.8725, "step": 105110 }, { "epoch": 0.0018432, "grad_norm": 0.7778321504592896, "learning_rate": 1.6656183226281476e-05, "loss": 2.8927, "step": 105120 }, { "epoch": 0.0018688, "grad_norm": 0.6913646459579468, "learning_rate": 1.6655582231546706e-05, "loss": 2.8173, "step": 105130 }, { "epoch": 0.0018944, "grad_norm": 0.7242841720581055, "learning_rate": 1.6654981193652254e-05, "loss": 2.8828, "step": 105140 }, { "epoch": 0.00192, "grad_norm": 0.6884406805038452, "learning_rate": 1.665438011260202e-05, "loss": 2.9783, "step": 105150 }, { "epoch": 0.0019456, "grad_norm": 0.7355433106422424, "learning_rate": 1.6653778988399896e-05, "loss": 3.043, "step": 105160 }, { "epoch": 0.0019712, "grad_norm": 0.7126909494400024, "learning_rate": 1.6653177821049787e-05, "loss": 3.228, "step": 105170 }, { "epoch": 0.0019968, "grad_norm": 0.6503609418869019, "learning_rate": 1.6652576610555583e-05, "loss": 3.1311, "step": 105180 }, { "epoch": 0.0020224, "grad_norm": 0.8928340077400208, "learning_rate": 1.6651975356921192e-05, "loss": 3.1396, "step": 105190 }, { "epoch": 0.002048, "grad_norm": 0.6490326523780823, "learning_rate": 1.665137406015051e-05, "loss": 3.0932, "step": 105200 }, { "epoch": 0.0020736, "grad_norm": 0.8896421790122986, "learning_rate": 1.6650772720247433e-05, "loss": 2.7624, "step": 105210 }, { "epoch": 0.0020992, "grad_norm": 1.0875178575515747, "learning_rate": 1.665017133721586e-05, "loss": 2.758, "step": 105220 }, { "epoch": 0.0021248, "grad_norm": 0.722495436668396, "learning_rate": 1.6649569911059696e-05, "loss": 2.7529, "step": 105230 }, { "epoch": 0.0021504, "grad_norm": 0.6706650257110596, "learning_rate": 1.664896844178284e-05, "loss": 2.858, "step": 105240 }, { "epoch": 0.002176, "grad_norm": 0.7840092778205872, "learning_rate": 1.6648366929389187e-05, "loss": 3.1367, "step": 105250 }, { "epoch": 0.0022016, "grad_norm": 0.6432861089706421, "learning_rate": 1.6647765373882644e-05, "loss": 2.9961, "step": 105260 }, { "epoch": 0.0022272, "grad_norm": 0.8551596999168396, "learning_rate": 1.664716377526711e-05, "loss": 3.026, "step": 105270 }, { "epoch": 0.0022528, "grad_norm": 0.7448242902755737, "learning_rate": 1.664656213354648e-05, "loss": 2.9887, "step": 105280 }, { "epoch": 0.0022784, "grad_norm": 0.7100070714950562, "learning_rate": 1.664596044872467e-05, "loss": 3.0483, "step": 105290 }, { "epoch": 0.002304, "grad_norm": 0.7119173407554626, "learning_rate": 1.6645358720805563e-05, "loss": 2.9098, "step": 105300 }, { "epoch": 0.0023296, "grad_norm": 0.8682659268379211, "learning_rate": 1.6644756949793073e-05, "loss": 2.8376, "step": 105310 }, { "epoch": 0.0023552, "grad_norm": 0.9098602533340454, "learning_rate": 1.6644155135691106e-05, "loss": 3.0022, "step": 105320 }, { "epoch": 0.0023808, "grad_norm": 1.0551820993423462, "learning_rate": 1.6643553278503555e-05, "loss": 2.8646, "step": 105330 }, { "epoch": 0.0024064, "grad_norm": 0.7057828307151794, "learning_rate": 1.6642951378234325e-05, "loss": 2.9705, "step": 105340 }, { "epoch": 0.002432, "grad_norm": 0.6891871094703674, "learning_rate": 1.6642349434887324e-05, "loss": 3.192, "step": 105350 }, { "epoch": 0.0024576, "grad_norm": 0.7136437892913818, "learning_rate": 1.664174744846645e-05, "loss": 3.0691, "step": 105360 }, { "epoch": 0.0024832, "grad_norm": 0.7274551391601562, "learning_rate": 1.664114541897561e-05, "loss": 2.9809, "step": 105370 }, { "epoch": 0.0025088, "grad_norm": 0.8349859118461609, "learning_rate": 1.6640543346418707e-05, "loss": 3.1469, "step": 105380 }, { "epoch": 0.0025344, "grad_norm": 0.865016758441925, "learning_rate": 1.6639941230799644e-05, "loss": 3.0939, "step": 105390 }, { "epoch": 0.00256, "grad_norm": 0.8417463302612305, "learning_rate": 1.6639339072122326e-05, "loss": 3.1505, "step": 105400 }, { "epoch": 0.0025856, "grad_norm": 0.7129809856414795, "learning_rate": 1.6638736870390662e-05, "loss": 3.1749, "step": 105410 }, { "epoch": 0.0026112, "grad_norm": 0.877212643623352, "learning_rate": 1.6638134625608552e-05, "loss": 3.1261, "step": 105420 }, { "epoch": 0.0026368, "grad_norm": 0.7106478810310364, "learning_rate": 1.6637532337779906e-05, "loss": 3.0791, "step": 105430 }, { "epoch": 0.0026624, "grad_norm": 0.7284533381462097, "learning_rate": 1.6636930006908623e-05, "loss": 3.1667, "step": 105440 }, { "epoch": 0.002688, "grad_norm": 0.8561989068984985, "learning_rate": 1.6636327632998613e-05, "loss": 3.25, "step": 105450 }, { "epoch": 0.0027136, "grad_norm": 0.6871907114982605, "learning_rate": 1.6635725216053784e-05, "loss": 3.0426, "step": 105460 }, { "epoch": 0.0027392, "grad_norm": 0.7894798517227173, "learning_rate": 1.663512275607804e-05, "loss": 3.3389, "step": 105470 }, { "epoch": 0.0027648, "grad_norm": 0.6878867149353027, "learning_rate": 1.6634520253075287e-05, "loss": 3.0398, "step": 105480 }, { "epoch": 0.0027904, "grad_norm": 0.7064356803894043, "learning_rate": 1.6633917707049433e-05, "loss": 3.07, "step": 105490 }, { "epoch": 0.002816, "grad_norm": 0.893223226070404, "learning_rate": 1.6633315118004388e-05, "loss": 3.0339, "step": 105500 }, { "epoch": 0.0028416, "grad_norm": 0.8238965272903442, "learning_rate": 1.6632712485944056e-05, "loss": 3.0985, "step": 105510 }, { "epoch": 0.0028672, "grad_norm": 0.8149630427360535, "learning_rate": 1.663210981087235e-05, "loss": 3.0784, "step": 105520 }, { "epoch": 0.0028928, "grad_norm": 0.7437857389450073, "learning_rate": 1.663150709279317e-05, "loss": 3.1308, "step": 105530 }, { "epoch": 0.0029184, "grad_norm": 0.8963940143585205, "learning_rate": 1.6630904331710433e-05, "loss": 3.1714, "step": 105540 }, { "epoch": 0.002944, "grad_norm": 0.7268811464309692, "learning_rate": 1.663030152762804e-05, "loss": 3.0645, "step": 105550 }, { "epoch": 0.0029696, "grad_norm": 0.7602735161781311, "learning_rate": 1.66296986805499e-05, "loss": 3.0859, "step": 105560 }, { "epoch": 0.0029952, "grad_norm": 0.7835559248924255, "learning_rate": 1.662909579047993e-05, "loss": 3.3557, "step": 105570 }, { "epoch": 0.0030208, "grad_norm": 0.6951190233230591, "learning_rate": 1.6628492857422037e-05, "loss": 2.998, "step": 105580 }, { "epoch": 0.0030464, "grad_norm": 0.7343466877937317, "learning_rate": 1.662788988138013e-05, "loss": 2.9427, "step": 105590 }, { "epoch": 0.003072, "grad_norm": 0.6818963289260864, "learning_rate": 1.662728686235812e-05, "loss": 2.9548, "step": 105600 }, { "epoch": 0.0030976, "grad_norm": 0.6802607774734497, "learning_rate": 1.6626683800359912e-05, "loss": 2.9811, "step": 105610 }, { "epoch": 0.0031232, "grad_norm": 1.2037793397903442, "learning_rate": 1.662608069538942e-05, "loss": 3.2622, "step": 105620 }, { "epoch": 0.0031488, "grad_norm": 0.8174192905426025, "learning_rate": 1.662547754745056e-05, "loss": 3.0001, "step": 105630 }, { "epoch": 0.0031744, "grad_norm": 0.7285802364349365, "learning_rate": 1.6624874356547236e-05, "loss": 3.3345, "step": 105640 }, { "epoch": 0.0032, "grad_norm": 0.7440846562385559, "learning_rate": 1.662427112268336e-05, "loss": 3.0634, "step": 105650 }, { "epoch": 0.0032256, "grad_norm": 1.0388884544372559, "learning_rate": 1.662366784586285e-05, "loss": 3.0439, "step": 105660 }, { "epoch": 0.0032512, "grad_norm": 1.0116209983825684, "learning_rate": 1.6623064526089617e-05, "loss": 3.1392, "step": 105670 }, { "epoch": 0.0032768, "grad_norm": 0.7329239845275879, "learning_rate": 1.662246116336757e-05, "loss": 3.1602, "step": 105680 }, { "epoch": 0.0033024, "grad_norm": 0.6709451675415039, "learning_rate": 1.6621857757700617e-05, "loss": 2.9634, "step": 105690 }, { "epoch": 0.003328, "grad_norm": 0.8003660440444946, "learning_rate": 1.662125430909268e-05, "loss": 3.2099, "step": 105700 }, { "epoch": 0.0033536, "grad_norm": 0.6899294257164001, "learning_rate": 1.6620650817547668e-05, "loss": 3.1094, "step": 105710 }, { "epoch": 0.0033792, "grad_norm": 0.7080681920051575, "learning_rate": 1.6620047283069492e-05, "loss": 3.0359, "step": 105720 }, { "epoch": 0.0034048, "grad_norm": 0.682256817817688, "learning_rate": 1.6619443705662074e-05, "loss": 2.8147, "step": 105730 }, { "epoch": 0.0034304, "grad_norm": 1.0735158920288086, "learning_rate": 1.661884008532932e-05, "loss": 3.3381, "step": 105740 }, { "epoch": 0.003456, "grad_norm": 0.6699202656745911, "learning_rate": 1.6618236422075147e-05, "loss": 3.2068, "step": 105750 }, { "epoch": 0.0034816, "grad_norm": 0.6733370423316956, "learning_rate": 1.6617632715903467e-05, "loss": 2.9573, "step": 105760 }, { "epoch": 0.0035072, "grad_norm": 0.7383571863174438, "learning_rate": 1.6617028966818206e-05, "loss": 2.8953, "step": 105770 }, { "epoch": 0.0035328, "grad_norm": 0.8672634363174438, "learning_rate": 1.6616425174823264e-05, "loss": 3.0865, "step": 105780 }, { "epoch": 0.0035584, "grad_norm": 0.9235935211181641, "learning_rate": 1.6615821339922564e-05, "loss": 3.1151, "step": 105790 }, { "epoch": 0.003584, "grad_norm": 0.7231006622314453, "learning_rate": 1.661521746212002e-05, "loss": 3.183, "step": 105800 }, { "epoch": 0.0036096, "grad_norm": 0.7264082431793213, "learning_rate": 1.6614613541419554e-05, "loss": 3.2292, "step": 105810 }, { "epoch": 0.0036352, "grad_norm": 0.8025087714195251, "learning_rate": 1.6614009577825072e-05, "loss": 3.0742, "step": 105820 }, { "epoch": 0.0036608, "grad_norm": 0.7551420331001282, "learning_rate": 1.6613405571340497e-05, "loss": 3.0623, "step": 105830 }, { "epoch": 0.0036864, "grad_norm": 0.6915070414543152, "learning_rate": 1.6612801521969746e-05, "loss": 3.0765, "step": 105840 }, { "epoch": 0.003712, "grad_norm": 0.8756164908409119, "learning_rate": 1.6612197429716733e-05, "loss": 3.1505, "step": 105850 }, { "epoch": 0.0037376, "grad_norm": 0.7067664265632629, "learning_rate": 1.6611593294585377e-05, "loss": 3.0915, "step": 105860 }, { "epoch": 0.0037632, "grad_norm": 0.6879874467849731, "learning_rate": 1.6610989116579595e-05, "loss": 2.8828, "step": 105870 }, { "epoch": 0.0037888, "grad_norm": 0.8857049942016602, "learning_rate": 1.661038489570331e-05, "loss": 3.2222, "step": 105880 }, { "epoch": 0.0038144, "grad_norm": 0.6955036520957947, "learning_rate": 1.660978063196043e-05, "loss": 3.0574, "step": 105890 }, { "epoch": 0.00384, "grad_norm": 0.7382642030715942, "learning_rate": 1.660917632535488e-05, "loss": 3.1612, "step": 105900 }, { "epoch": 0.0038656, "grad_norm": 0.7400384545326233, "learning_rate": 1.660857197589058e-05, "loss": 2.9753, "step": 105910 }, { "epoch": 0.0038912, "grad_norm": 0.8616741895675659, "learning_rate": 1.6607967583571447e-05, "loss": 3.3344, "step": 105920 }, { "epoch": 0.0039168, "grad_norm": 0.9106396436691284, "learning_rate": 1.66073631484014e-05, "loss": 2.9328, "step": 105930 }, { "epoch": 0.0039424, "grad_norm": 0.7840394377708435, "learning_rate": 1.6606758670384362e-05, "loss": 3.0778, "step": 105940 }, { "epoch": 0.003968, "grad_norm": 0.743530809879303, "learning_rate": 1.6606154149524244e-05, "loss": 3.0125, "step": 105950 }, { "epoch": 0.0039936, "grad_norm": 0.9614514112472534, "learning_rate": 1.6605549585824977e-05, "loss": 3.1612, "step": 105960 }, { "epoch": 0.0040192, "grad_norm": 0.6728450655937195, "learning_rate": 1.6604944979290475e-05, "loss": 3.1459, "step": 105970 }, { "epoch": 0.0040448, "grad_norm": 0.7256367206573486, "learning_rate": 1.6604340329924658e-05, "loss": 3.0309, "step": 105980 }, { "epoch": 0.0040704, "grad_norm": 0.7287602424621582, "learning_rate": 1.6603735637731452e-05, "loss": 3.1045, "step": 105990 }, { "epoch": 0.004096, "grad_norm": 0.6856576800346375, "learning_rate": 1.6603130902714775e-05, "loss": 3.0182, "step": 106000 }, { "epoch": 0.0041216, "grad_norm": 0.9857890009880066, "learning_rate": 1.6602526124878545e-05, "loss": 3.0911, "step": 106010 }, { "epoch": 0.0041472, "grad_norm": 0.6664150953292847, "learning_rate": 1.6601921304226694e-05, "loss": 3.1164, "step": 106020 }, { "epoch": 0.0041728, "grad_norm": 0.7787854075431824, "learning_rate": 1.6601316440763135e-05, "loss": 2.9956, "step": 106030 }, { "epoch": 0.0041984, "grad_norm": 0.789355456829071, "learning_rate": 1.6600711534491795e-05, "loss": 3.1422, "step": 106040 }, { "epoch": 0.004224, "grad_norm": 0.918506383895874, "learning_rate": 1.6600106585416593e-05, "loss": 3.26, "step": 106050 }, { "epoch": 0.0042496, "grad_norm": 0.79727703332901, "learning_rate": 1.6599501593541453e-05, "loss": 3.1667, "step": 106060 }, { "epoch": 0.0042752, "grad_norm": 0.7461289167404175, "learning_rate": 1.6598896558870303e-05, "loss": 3.0983, "step": 106070 }, { "epoch": 0.0043008, "grad_norm": 0.6792963743209839, "learning_rate": 1.6598291481407058e-05, "loss": 3.348, "step": 106080 }, { "epoch": 0.0043264, "grad_norm": 0.7335224151611328, "learning_rate": 1.659768636115565e-05, "loss": 3.1176, "step": 106090 }, { "epoch": 0.004352, "grad_norm": 0.7155961990356445, "learning_rate": 1.6597081198119995e-05, "loss": 3.144, "step": 106100 }, { "epoch": 0.0043776, "grad_norm": 0.6991564631462097, "learning_rate": 1.6596475992304025e-05, "loss": 3.1727, "step": 106110 }, { "epoch": 0.0044032, "grad_norm": 0.7728461623191833, "learning_rate": 1.659587074371166e-05, "loss": 3.1848, "step": 106120 }, { "epoch": 0.0044288, "grad_norm": 0.8283504247665405, "learning_rate": 1.659526545234683e-05, "loss": 3.2479, "step": 106130 }, { "epoch": 0.0044544, "grad_norm": 0.740500271320343, "learning_rate": 1.6594660118213452e-05, "loss": 3.1552, "step": 106140 }, { "epoch": 0.00448, "grad_norm": 0.6870490908622742, "learning_rate": 1.6594054741315458e-05, "loss": 2.8741, "step": 106150 }, { "epoch": 0.0045056, "grad_norm": 0.7608031630516052, "learning_rate": 1.659344932165677e-05, "loss": 3.4211, "step": 106160 }, { "epoch": 0.0045312, "grad_norm": 0.7881180644035339, "learning_rate": 1.6592843859241313e-05, "loss": 3.0339, "step": 106170 }, { "epoch": 0.0045568, "grad_norm": 0.7663413882255554, "learning_rate": 1.659223835407302e-05, "loss": 2.9706, "step": 106180 }, { "epoch": 0.0045824, "grad_norm": 0.7226651310920715, "learning_rate": 1.6591632806155815e-05, "loss": 3.1212, "step": 106190 }, { "epoch": 0.004608, "grad_norm": 2.486542224884033, "learning_rate": 1.659102721549362e-05, "loss": 3.1455, "step": 106200 }, { "epoch": 0.0046336, "grad_norm": 0.8106687664985657, "learning_rate": 1.6590421582090366e-05, "loss": 3.2534, "step": 106210 }, { "epoch": 0.0046592, "grad_norm": 0.6924505233764648, "learning_rate": 1.6589815905949977e-05, "loss": 3.2494, "step": 106220 }, { "epoch": 0.0046848, "grad_norm": 0.9119728207588196, "learning_rate": 1.6589210187076386e-05, "loss": 3.0689, "step": 106230 }, { "epoch": 0.0047104, "grad_norm": 0.774565577507019, "learning_rate": 1.658860442547352e-05, "loss": 3.1946, "step": 106240 }, { "epoch": 0.004736, "grad_norm": 0.9912832975387573, "learning_rate": 1.6587998621145302e-05, "loss": 2.9531, "step": 106250 }, { "epoch": 0.0047616, "grad_norm": 0.7806259393692017, "learning_rate": 1.6587392774095663e-05, "loss": 3.1248, "step": 106260 }, { "epoch": 0.0047872, "grad_norm": 0.678598165512085, "learning_rate": 1.6586786884328533e-05, "loss": 3.0266, "step": 106270 }, { "epoch": 0.0048128, "grad_norm": 0.7636155486106873, "learning_rate": 1.6586180951847842e-05, "loss": 3.0683, "step": 106280 }, { "epoch": 0.0048384, "grad_norm": 0.718947172164917, "learning_rate": 1.6585574976657518e-05, "loss": 3.2727, "step": 106290 }, { "epoch": 0.004864, "grad_norm": 0.7834913730621338, "learning_rate": 1.658496895876149e-05, "loss": 3.1806, "step": 106300 }, { "epoch": 0.0048896, "grad_norm": 0.6926252841949463, "learning_rate": 1.6584362898163687e-05, "loss": 3.0899, "step": 106310 }, { "epoch": 0.0049152, "grad_norm": 0.7023921608924866, "learning_rate": 1.6583756794868045e-05, "loss": 3.2859, "step": 106320 }, { "epoch": 0.0049408, "grad_norm": 0.6675668358802795, "learning_rate": 1.6583150648878487e-05, "loss": 2.9016, "step": 106330 }, { "epoch": 0.0049664, "grad_norm": 0.7603338360786438, "learning_rate": 1.658254446019895e-05, "loss": 3.1291, "step": 106340 }, { "epoch": 0.004992, "grad_norm": 0.7220694422721863, "learning_rate": 1.6581938228833356e-05, "loss": 3.0482, "step": 106350 }, { "epoch": 0.0050176, "grad_norm": 0.7948000431060791, "learning_rate": 1.658133195478564e-05, "loss": 3.1749, "step": 106360 }, { "epoch": 0.0050432, "grad_norm": 0.6885883808135986, "learning_rate": 1.658072563805974e-05, "loss": 3.0284, "step": 106370 }, { "epoch": 0.0050688, "grad_norm": 0.8744683861732483, "learning_rate": 1.6580119278659585e-05, "loss": 3.0573, "step": 106380 }, { "epoch": 0.0050944, "grad_norm": 0.9897986054420471, "learning_rate": 1.6579512876589104e-05, "loss": 3.0401, "step": 106390 }, { "epoch": 0.00512, "grad_norm": 0.7917213439941406, "learning_rate": 1.6578906431852228e-05, "loss": 3.0696, "step": 106400 }, { "epoch": 0.0051456, "grad_norm": 0.7305943369865417, "learning_rate": 1.6578299944452894e-05, "loss": 3.118, "step": 106410 }, { "epoch": 0.0051712, "grad_norm": 0.6947545409202576, "learning_rate": 1.6577693414395033e-05, "loss": 3.0223, "step": 106420 }, { "epoch": 0.0051968, "grad_norm": 0.7522108554840088, "learning_rate": 1.6577086841682578e-05, "loss": 3.0414, "step": 106430 }, { "epoch": 0.0052224, "grad_norm": 0.9842570424079895, "learning_rate": 1.6576480226319463e-05, "loss": 3.1082, "step": 106440 }, { "epoch": 0.005248, "grad_norm": 0.8706157803535461, "learning_rate": 1.6575873568309622e-05, "loss": 3.0038, "step": 106450 }, { "epoch": 0.0052736, "grad_norm": 0.7861028909683228, "learning_rate": 1.6575266867656987e-05, "loss": 3.3413, "step": 106460 }, { "epoch": 0.0052992, "grad_norm": 1.4529167413711548, "learning_rate": 1.6574660124365497e-05, "loss": 3.338, "step": 106470 }, { "epoch": 0.0053248, "grad_norm": 0.7638935446739197, "learning_rate": 1.657405333843908e-05, "loss": 3.0325, "step": 106480 }, { "epoch": 0.0053504, "grad_norm": 0.7663407921791077, "learning_rate": 1.6573446509881676e-05, "loss": 3.0688, "step": 106490 }, { "epoch": 0.005376, "grad_norm": 0.6773641705513, "learning_rate": 1.657283963869722e-05, "loss": 3.0469, "step": 106500 }, { "epoch": 0.0054016, "grad_norm": 0.7379757165908813, "learning_rate": 1.6572232724889642e-05, "loss": 2.8947, "step": 106510 }, { "epoch": 0.0054272, "grad_norm": 0.7000238299369812, "learning_rate": 1.6571625768462883e-05, "loss": 3.0771, "step": 106520 }, { "epoch": 0.0054528, "grad_norm": 0.9531989693641663, "learning_rate": 1.657101876942088e-05, "loss": 2.9855, "step": 106530 }, { "epoch": 0.0054784, "grad_norm": 0.9804726839065552, "learning_rate": 1.657041172776756e-05, "loss": 3.0171, "step": 106540 }, { "epoch": 0.005504, "grad_norm": 0.7701367735862732, "learning_rate": 1.6569804643506874e-05, "loss": 3.121, "step": 106550 }, { "epoch": 0.0055296, "grad_norm": 3.955071210861206, "learning_rate": 1.6569197516642748e-05, "loss": 3.1734, "step": 106560 }, { "epoch": 0.0055552, "grad_norm": 0.6946336627006531, "learning_rate": 1.6568590347179122e-05, "loss": 2.9566, "step": 106570 }, { "epoch": 0.0055808, "grad_norm": 1.0179423093795776, "learning_rate": 1.656798313511993e-05, "loss": 3.0836, "step": 106580 }, { "epoch": 0.0056064, "grad_norm": 0.6759966611862183, "learning_rate": 1.6567375880469117e-05, "loss": 3.2163, "step": 106590 }, { "epoch": 0.005632, "grad_norm": 0.9651687741279602, "learning_rate": 1.6566768583230615e-05, "loss": 3.1101, "step": 106600 }, { "epoch": 0.0056576, "grad_norm": 0.6752814650535583, "learning_rate": 1.6566161243408366e-05, "loss": 3.045, "step": 106610 }, { "epoch": 0.0056832, "grad_norm": 0.7231888175010681, "learning_rate": 1.6565553861006305e-05, "loss": 3.0826, "step": 106620 }, { "epoch": 0.0057088, "grad_norm": 0.6968892216682434, "learning_rate": 1.6564946436028374e-05, "loss": 3.0165, "step": 106630 }, { "epoch": 0.0057344, "grad_norm": 0.8100724220275879, "learning_rate": 1.6564338968478506e-05, "loss": 2.9492, "step": 106640 }, { "epoch": 0.00576, "grad_norm": 0.9221453666687012, "learning_rate": 1.6563731458360648e-05, "loss": 2.9927, "step": 106650 }, { "epoch": 0.0057856, "grad_norm": 0.772891104221344, "learning_rate": 1.6563123905678734e-05, "loss": 3.0052, "step": 106660 }, { "epoch": 0.0058112, "grad_norm": 0.7325231432914734, "learning_rate": 1.6562516310436708e-05, "loss": 2.9178, "step": 106670 }, { "epoch": 0.0058368, "grad_norm": 0.7438202500343323, "learning_rate": 1.6561908672638506e-05, "loss": 3.0232, "step": 106680 }, { "epoch": 0.0058624, "grad_norm": 1.2588287591934204, "learning_rate": 1.6561300992288066e-05, "loss": 3.0195, "step": 106690 }, { "epoch": 0.005888, "grad_norm": 0.7279868125915527, "learning_rate": 1.656069326938934e-05, "loss": 2.8995, "step": 106700 }, { "epoch": 0.0059136, "grad_norm": 0.665975034236908, "learning_rate": 1.656008550394626e-05, "loss": 3.0425, "step": 106710 }, { "epoch": 0.0059392, "grad_norm": 0.7604615688323975, "learning_rate": 1.6559477695962768e-05, "loss": 3.109, "step": 106720 }, { "epoch": 0.0059648, "grad_norm": 0.7669298648834229, "learning_rate": 1.6558869845442806e-05, "loss": 2.935, "step": 106730 }, { "epoch": 0.0059904, "grad_norm": 0.8251059651374817, "learning_rate": 1.6558261952390314e-05, "loss": 3.2326, "step": 106740 }, { "epoch": 0.006016, "grad_norm": 0.75180584192276, "learning_rate": 1.655765401680924e-05, "loss": 3.0195, "step": 106750 }, { "epoch": 0.0060416, "grad_norm": 0.7936665415763855, "learning_rate": 1.655704603870352e-05, "loss": 3.0032, "step": 106760 }, { "epoch": 0.0060672, "grad_norm": 0.8127036690711975, "learning_rate": 1.6556438018077103e-05, "loss": 3.0987, "step": 106770 }, { "epoch": 0.0060928, "grad_norm": 0.7751747965812683, "learning_rate": 1.6555829954933927e-05, "loss": 3.0159, "step": 106780 }, { "epoch": 0.0061184, "grad_norm": 0.7762129902839661, "learning_rate": 1.655522184927793e-05, "loss": 3.1807, "step": 106790 }, { "epoch": 0.006144, "grad_norm": 0.700057864189148, "learning_rate": 1.6554613701113068e-05, "loss": 3.0906, "step": 106800 }, { "epoch": 0.0061696, "grad_norm": 0.9681132435798645, "learning_rate": 1.6554005510443275e-05, "loss": 3.0077, "step": 106810 }, { "epoch": 0.0061952, "grad_norm": 0.6977329850196838, "learning_rate": 1.65533972772725e-05, "loss": 3.1629, "step": 106820 }, { "epoch": 0.0062208, "grad_norm": 1.2170743942260742, "learning_rate": 1.6552789001604685e-05, "loss": 3.1128, "step": 106830 }, { "epoch": 0.0062464, "grad_norm": 0.6310963034629822, "learning_rate": 1.6552180683443776e-05, "loss": 2.9425, "step": 106840 }, { "epoch": 0.006272, "grad_norm": 0.7974526286125183, "learning_rate": 1.6551572322793715e-05, "loss": 3.0419, "step": 106850 }, { "epoch": 0.0062976, "grad_norm": 0.7455421090126038, "learning_rate": 1.6550963919658447e-05, "loss": 3.01, "step": 106860 }, { "epoch": 0.0063232, "grad_norm": 10.073808670043945, "learning_rate": 1.655035547404192e-05, "loss": 3.0294, "step": 106870 }, { "epoch": 0.0063488, "grad_norm": 0.749669075012207, "learning_rate": 1.654974698594808e-05, "loss": 3.1692, "step": 106880 }, { "epoch": 0.0063744, "grad_norm": 0.7156291007995605, "learning_rate": 1.654913845538087e-05, "loss": 3.0154, "step": 106890 }, { "epoch": 0.0064, "grad_norm": 0.7452454566955566, "learning_rate": 1.6548529882344238e-05, "loss": 3.071, "step": 106900 }, { "epoch": 0.0064256, "grad_norm": 0.7296224236488342, "learning_rate": 1.654792126684213e-05, "loss": 2.9785, "step": 106910 }, { "epoch": 0.0064512, "grad_norm": 0.8315496444702148, "learning_rate": 1.6547312608878497e-05, "loss": 3.1326, "step": 106920 }, { "epoch": 0.0064768, "grad_norm": 0.6737388372421265, "learning_rate": 1.6546703908457276e-05, "loss": 3.0378, "step": 106930 }, { "epoch": 0.0065024, "grad_norm": 0.7257678508758545, "learning_rate": 1.6546095165582422e-05, "loss": 3.2016, "step": 106940 }, { "epoch": 0.006528, "grad_norm": 0.710868775844574, "learning_rate": 1.6545486380257885e-05, "loss": 3.0911, "step": 106950 }, { "epoch": 0.0065536, "grad_norm": 0.7181663513183594, "learning_rate": 1.6544877552487603e-05, "loss": 2.9758, "step": 106960 }, { "epoch": 0.0065792, "grad_norm": 0.7918558120727539, "learning_rate": 1.654426868227553e-05, "loss": 3.1365, "step": 106970 }, { "epoch": 0.0066048, "grad_norm": 0.7596262693405151, "learning_rate": 1.6543659769625615e-05, "loss": 3.0886, "step": 106980 }, { "epoch": 0.0066304, "grad_norm": 0.7264249324798584, "learning_rate": 1.6543050814541804e-05, "loss": 3.0587, "step": 106990 }, { "epoch": 0.006656, "grad_norm": 0.6788363456726074, "learning_rate": 1.6542441817028047e-05, "loss": 3.0193, "step": 107000 }, { "epoch": 0.0066816, "grad_norm": 0.8411043882369995, "learning_rate": 1.6541832777088296e-05, "loss": 2.9707, "step": 107010 }, { "epoch": 0.0067072, "grad_norm": 0.7917097806930542, "learning_rate": 1.65412236947265e-05, "loss": 3.1227, "step": 107020 }, { "epoch": 0.0067328, "grad_norm": 0.8946467041969299, "learning_rate": 1.6540614569946604e-05, "loss": 3.0262, "step": 107030 }, { "epoch": 0.0067584, "grad_norm": 0.7894487977027893, "learning_rate": 1.6540005402752557e-05, "loss": 3.0881, "step": 107040 }, { "epoch": 0.006784, "grad_norm": 0.8130916953086853, "learning_rate": 1.653939619314832e-05, "loss": 3.2248, "step": 107050 }, { "epoch": 0.0068096, "grad_norm": 0.7382495999336243, "learning_rate": 1.6538786941137833e-05, "loss": 3.0854, "step": 107060 }, { "epoch": 0.0068352, "grad_norm": 0.8527573347091675, "learning_rate": 1.653817764672505e-05, "loss": 3.0761, "step": 107070 }, { "epoch": 0.0068608, "grad_norm": 6.419358730316162, "learning_rate": 1.6537568309913924e-05, "loss": 3.1215, "step": 107080 }, { "epoch": 0.0068864, "grad_norm": 0.7346651554107666, "learning_rate": 1.6536958930708404e-05, "loss": 3.1719, "step": 107090 }, { "epoch": 0.006912, "grad_norm": 0.6525290012359619, "learning_rate": 1.6536349509112443e-05, "loss": 3.1195, "step": 107100 }, { "epoch": 0.0069376, "grad_norm": 0.7017077207565308, "learning_rate": 1.6535740045129994e-05, "loss": 3.274, "step": 107110 }, { "epoch": 0.0069632, "grad_norm": 0.9263157844543457, "learning_rate": 1.6535130538765005e-05, "loss": 3.0849, "step": 107120 }, { "epoch": 0.0069888, "grad_norm": 0.852705717086792, "learning_rate": 1.6534520990021433e-05, "loss": 3.2865, "step": 107130 }, { "epoch": 0.0070144, "grad_norm": 0.79963219165802, "learning_rate": 1.6533911398903232e-05, "loss": 3.0761, "step": 107140 }, { "epoch": 0.00704, "grad_norm": 0.7152642607688904, "learning_rate": 1.6533301765414346e-05, "loss": 3.2027, "step": 107150 }, { "epoch": 0.0070656, "grad_norm": 0.8797867298126221, "learning_rate": 1.6532692089558738e-05, "loss": 3.1455, "step": 107160 }, { "epoch": 0.0070912, "grad_norm": 0.7312014698982239, "learning_rate": 1.653208237134036e-05, "loss": 3.0564, "step": 107170 }, { "epoch": 0.0071168, "grad_norm": 0.7657479047775269, "learning_rate": 1.653147261076316e-05, "loss": 3.0588, "step": 107180 }, { "epoch": 0.0071424, "grad_norm": 0.8432350158691406, "learning_rate": 1.6530862807831098e-05, "loss": 3.14, "step": 107190 }, { "epoch": 0.007168, "grad_norm": 0.6812677383422852, "learning_rate": 1.6530252962548125e-05, "loss": 3.3095, "step": 107200 }, { "epoch": 0.0071936, "grad_norm": 0.8803184628486633, "learning_rate": 1.6529643074918197e-05, "loss": 3.2051, "step": 107210 }, { "epoch": 0.0072192, "grad_norm": 0.8011219501495361, "learning_rate": 1.6529033144945268e-05, "loss": 3.2349, "step": 107220 }, { "epoch": 0.0072448, "grad_norm": 0.6644448041915894, "learning_rate": 1.6528423172633296e-05, "loss": 3.0869, "step": 107230 }, { "epoch": 0.0072704, "grad_norm": 0.6690495610237122, "learning_rate": 1.6527813157986237e-05, "loss": 3.1084, "step": 107240 }, { "epoch": 0.007296, "grad_norm": 0.7003933191299438, "learning_rate": 1.6527203101008043e-05, "loss": 3.2017, "step": 107250 }, { "epoch": 0.0073216, "grad_norm": 0.724258303642273, "learning_rate": 1.6526593001702668e-05, "loss": 3.1595, "step": 107260 }, { "epoch": 0.0073472, "grad_norm": 0.7503255009651184, "learning_rate": 1.6525982860074075e-05, "loss": 3.1061, "step": 107270 }, { "epoch": 0.0073728, "grad_norm": 0.6980808973312378, "learning_rate": 1.6525372676126218e-05, "loss": 3.0562, "step": 107280 }, { "epoch": 0.0073984, "grad_norm": 0.7985548973083496, "learning_rate": 1.652476244986305e-05, "loss": 3.0824, "step": 107290 }, { "epoch": 0.007424, "grad_norm": 0.7454351782798767, "learning_rate": 1.6524152181288536e-05, "loss": 3.1206, "step": 107300 }, { "epoch": 0.0074496, "grad_norm": 0.6649746894836426, "learning_rate": 1.652354187040663e-05, "loss": 3.005, "step": 107310 }, { "epoch": 0.0074752, "grad_norm": 0.7657681703567505, "learning_rate": 1.6522931517221284e-05, "loss": 3.1514, "step": 107320 }, { "epoch": 0.0075008, "grad_norm": 0.7021831274032593, "learning_rate": 1.6522321121736463e-05, "loss": 3.1941, "step": 107330 }, { "epoch": 0.0075264, "grad_norm": 0.6721652150154114, "learning_rate": 1.6521710683956125e-05, "loss": 3.2157, "step": 107340 }, { "epoch": 0.007552, "grad_norm": 0.720301628112793, "learning_rate": 1.6521100203884225e-05, "loss": 3.0896, "step": 107350 }, { "epoch": 0.0075776, "grad_norm": 0.8317900896072388, "learning_rate": 1.6520489681524724e-05, "loss": 2.9924, "step": 107360 }, { "epoch": 0.0076032, "grad_norm": 0.7847937345504761, "learning_rate": 1.6519879116881575e-05, "loss": 3.1682, "step": 107370 }, { "epoch": 0.0076288, "grad_norm": 0.6745406985282898, "learning_rate": 1.651926850995875e-05, "loss": 3.2767, "step": 107380 }, { "epoch": 0.0076544, "grad_norm": 0.8468344807624817, "learning_rate": 1.6518657860760194e-05, "loss": 3.12, "step": 107390 }, { "epoch": 0.00768, "grad_norm": 0.769224226474762, "learning_rate": 1.651804716928988e-05, "loss": 2.9965, "step": 107400 }, { "epoch": 0.0077056, "grad_norm": 0.7060737609863281, "learning_rate": 1.6517436435551762e-05, "loss": 3.234, "step": 107410 }, { "epoch": 0.0077312, "grad_norm": 0.8325627446174622, "learning_rate": 1.6516825659549796e-05, "loss": 3.123, "step": 107420 }, { "epoch": 0.0077568, "grad_norm": 0.7159736156463623, "learning_rate": 1.6516214841287954e-05, "loss": 3.0625, "step": 107430 }, { "epoch": 0.0077824, "grad_norm": 0.7427327036857605, "learning_rate": 1.6515603980770186e-05, "loss": 3.0995, "step": 107440 }, { "epoch": 0.007808, "grad_norm": 0.6499155759811401, "learning_rate": 1.651499307800046e-05, "loss": 2.9865, "step": 107450 }, { "epoch": 0.0078336, "grad_norm": 0.7668103575706482, "learning_rate": 1.6514382132982736e-05, "loss": 3.1969, "step": 107460 }, { "epoch": 0.0078592, "grad_norm": 0.6817129254341125, "learning_rate": 1.6513771145720975e-05, "loss": 3.0896, "step": 107470 }, { "epoch": 0.0078848, "grad_norm": 0.6503598093986511, "learning_rate": 1.6513160116219135e-05, "loss": 2.8827, "step": 107480 }, { "epoch": 0.0079104, "grad_norm": 0.6471995711326599, "learning_rate": 1.6512549044481185e-05, "loss": 2.9484, "step": 107490 }, { "epoch": 0.007936, "grad_norm": 0.7231259346008301, "learning_rate": 1.6511937930511085e-05, "loss": 3.1683, "step": 107500 }, { "epoch": 0.0079616, "grad_norm": 0.8703631162643433, "learning_rate": 1.65113267743128e-05, "loss": 3.0889, "step": 107510 }, { "epoch": 0.0079872, "grad_norm": 1.0855029821395874, "learning_rate": 1.6510715575890293e-05, "loss": 3.1365, "step": 107520 }, { "epoch": 0.0080128, "grad_norm": 0.7954599857330322, "learning_rate": 1.6510104335247523e-05, "loss": 3.1621, "step": 107530 }, { "epoch": 0.0080384, "grad_norm": 0.7658388018608093, "learning_rate": 1.6509493052388455e-05, "loss": 3.3874, "step": 107540 }, { "epoch": 0.008064, "grad_norm": 0.708634614944458, "learning_rate": 1.650888172731706e-05, "loss": 3.0444, "step": 107550 }, { "epoch": 0.0080896, "grad_norm": 1.5844658613204956, "learning_rate": 1.650827036003729e-05, "loss": 3.1317, "step": 107560 }, { "epoch": 0.0081152, "grad_norm": 0.6961430907249451, "learning_rate": 1.650765895055312e-05, "loss": 3.2292, "step": 107570 }, { "epoch": 0.0081408, "grad_norm": 0.9531350135803223, "learning_rate": 1.6507047498868513e-05, "loss": 3.0758, "step": 107580 }, { "epoch": 0.0081664, "grad_norm": 0.6486190557479858, "learning_rate": 1.650643600498743e-05, "loss": 3.0315, "step": 107590 }, { "epoch": 0.008192, "grad_norm": 0.707250714302063, "learning_rate": 1.650582446891384e-05, "loss": 2.9163, "step": 107600 }, { "epoch": 0.0082176, "grad_norm": 0.661608874797821, "learning_rate": 1.6505212890651704e-05, "loss": 3.025, "step": 107610 }, { "epoch": 0.0082432, "grad_norm": 0.7634052038192749, "learning_rate": 1.6504601270204995e-05, "loss": 3.1768, "step": 107620 }, { "epoch": 0.0082688, "grad_norm": 0.727724015712738, "learning_rate": 1.6503989607577675e-05, "loss": 3.1685, "step": 107630 }, { "epoch": 0.0082944, "grad_norm": 0.7072423696517944, "learning_rate": 1.650337790277371e-05, "loss": 3.092, "step": 107640 }, { "epoch": 0.00832, "grad_norm": 0.7788125872612, "learning_rate": 1.6502766155797068e-05, "loss": 3.1714, "step": 107650 }, { "epoch": 0.0083456, "grad_norm": 0.7408372163772583, "learning_rate": 1.6502154366651713e-05, "loss": 3.1836, "step": 107660 }, { "epoch": 0.0083712, "grad_norm": 0.7752769589424133, "learning_rate": 1.650154253534162e-05, "loss": 3.0684, "step": 107670 }, { "epoch": 0.0083968, "grad_norm": 0.7839319109916687, "learning_rate": 1.6500930661870747e-05, "loss": 3.2741, "step": 107680 }, { "epoch": 0.0084224, "grad_norm": 0.8628256320953369, "learning_rate": 1.650031874624307e-05, "loss": 3.2268, "step": 107690 }, { "epoch": 0.008448, "grad_norm": 0.719298243522644, "learning_rate": 1.6499706788462554e-05, "loss": 2.9548, "step": 107700 }, { "epoch": 0.0084736, "grad_norm": 0.7409751415252686, "learning_rate": 1.6499094788533162e-05, "loss": 3.0382, "step": 107710 }, { "epoch": 0.0084992, "grad_norm": 0.7473001480102539, "learning_rate": 1.6498482746458876e-05, "loss": 3.2428, "step": 107720 }, { "epoch": 0.0085248, "grad_norm": 0.7311194539070129, "learning_rate": 1.6497870662243648e-05, "loss": 3.2146, "step": 107730 }, { "epoch": 0.0085504, "grad_norm": 0.7465600371360779, "learning_rate": 1.6497258535891458e-05, "loss": 3.2605, "step": 107740 }, { "epoch": 0.008576, "grad_norm": 0.8677411079406738, "learning_rate": 1.6496646367406273e-05, "loss": 3.0778, "step": 107750 }, { "epoch": 0.0086016, "grad_norm": 0.7801406383514404, "learning_rate": 1.6496034156792066e-05, "loss": 3.0525, "step": 107760 }, { "epoch": 0.0086272, "grad_norm": 0.7192838788032532, "learning_rate": 1.64954219040528e-05, "loss": 3.1191, "step": 107770 }, { "epoch": 0.0086528, "grad_norm": 0.7344465255737305, "learning_rate": 1.6494809609192453e-05, "loss": 3.0061, "step": 107780 }, { "epoch": 0.0086784, "grad_norm": 0.7442473769187927, "learning_rate": 1.649419727221499e-05, "loss": 3.092, "step": 107790 }, { "epoch": 0.008704, "grad_norm": 0.8911067247390747, "learning_rate": 1.649358489312438e-05, "loss": 3.076, "step": 107800 }, { "epoch": 0.0087296, "grad_norm": 0.806263267993927, "learning_rate": 1.6492972471924603e-05, "loss": 3.1634, "step": 107810 }, { "epoch": 0.0087552, "grad_norm": 0.7299284338951111, "learning_rate": 1.6492360008619623e-05, "loss": 3.1079, "step": 107820 }, { "epoch": 0.0087808, "grad_norm": 0.7603796720504761, "learning_rate": 1.6491747503213416e-05, "loss": 3.0568, "step": 107830 }, { "epoch": 0.0088064, "grad_norm": 0.6951054334640503, "learning_rate": 1.649113495570995e-05, "loss": 3.0538, "step": 107840 }, { "epoch": 0.008832, "grad_norm": 1.7680121660232544, "learning_rate": 1.6490522366113197e-05, "loss": 3.2799, "step": 107850 }, { "epoch": 0.0088576, "grad_norm": 0.7019798159599304, "learning_rate": 1.648990973442713e-05, "loss": 3.0035, "step": 107860 }, { "epoch": 0.0088832, "grad_norm": 0.8920366764068604, "learning_rate": 1.648929706065573e-05, "loss": 3.2548, "step": 107870 }, { "epoch": 0.0089088, "grad_norm": 0.7339498400688171, "learning_rate": 1.648868434480296e-05, "loss": 3.0674, "step": 107880 }, { "epoch": 0.0089344, "grad_norm": 0.9002181887626648, "learning_rate": 1.6488071586872794e-05, "loss": 3.0215, "step": 107890 }, { "epoch": 0.00896, "grad_norm": 0.6761047840118408, "learning_rate": 1.648745878686921e-05, "loss": 3.0823, "step": 107900 }, { "epoch": 0.0089856, "grad_norm": 0.6479434967041016, "learning_rate": 1.6486845944796178e-05, "loss": 3.2007, "step": 107910 }, { "epoch": 0.0090112, "grad_norm": 0.7326242923736572, "learning_rate": 1.6486233060657678e-05, "loss": 3.116, "step": 107920 }, { "epoch": 0.0090368, "grad_norm": 0.7754827737808228, "learning_rate": 1.648562013445768e-05, "loss": 3.1033, "step": 107930 }, { "epoch": 0.0090624, "grad_norm": 0.8170773386955261, "learning_rate": 1.6485007166200158e-05, "loss": 3.0488, "step": 107940 }, { "epoch": 0.009088, "grad_norm": 1.611203670501709, "learning_rate": 1.648439415588909e-05, "loss": 3.1528, "step": 107950 }, { "epoch": 0.0091136, "grad_norm": 0.7472312450408936, "learning_rate": 1.6483781103528444e-05, "loss": 3.1455, "step": 107960 }, { "epoch": 0.0091392, "grad_norm": 0.6767090559005737, "learning_rate": 1.6483168009122207e-05, "loss": 3.074, "step": 107970 }, { "epoch": 0.0091648, "grad_norm": 1.1419522762298584, "learning_rate": 1.6482554872674343e-05, "loss": 3.1306, "step": 107980 }, { "epoch": 0.0091904, "grad_norm": 0.8343557715415955, "learning_rate": 1.6481941694188838e-05, "loss": 3.062, "step": 107990 }, { "epoch": 0.009216, "grad_norm": 0.7280368804931641, "learning_rate": 1.648132847366966e-05, "loss": 3.0656, "step": 108000 }, { "epoch": 0.0092416, "grad_norm": 0.9204459190368652, "learning_rate": 1.6480715211120788e-05, "loss": 3.0834, "step": 108010 }, { "epoch": 0.0092672, "grad_norm": 0.8147127032279968, "learning_rate": 1.6480101906546205e-05, "loss": 3.1728, "step": 108020 }, { "epoch": 0.0092928, "grad_norm": 0.7720587253570557, "learning_rate": 1.647948855994988e-05, "loss": 3.4872, "step": 108030 }, { "epoch": 0.0093184, "grad_norm": 0.8501452803611755, "learning_rate": 1.6478875171335796e-05, "loss": 2.98, "step": 108040 }, { "epoch": 0.009344, "grad_norm": 0.7741602659225464, "learning_rate": 1.6478261740707928e-05, "loss": 2.8336, "step": 108050 }, { "epoch": 0.0093696, "grad_norm": 0.7484472990036011, "learning_rate": 1.6477648268070255e-05, "loss": 3.2296, "step": 108060 }, { "epoch": 0.0093952, "grad_norm": 0.7064381837844849, "learning_rate": 1.647703475342675e-05, "loss": 3.0246, "step": 108070 }, { "epoch": 0.0094208, "grad_norm": 0.7840883135795593, "learning_rate": 1.64764211967814e-05, "loss": 3.0024, "step": 108080 }, { "epoch": 0.0094464, "grad_norm": 0.6954635977745056, "learning_rate": 1.6475807598138182e-05, "loss": 3.1556, "step": 108090 }, { "epoch": 0.009472, "grad_norm": 0.7432703375816345, "learning_rate": 1.647519395750107e-05, "loss": 3.0675, "step": 108100 }, { "epoch": 0.0094976, "grad_norm": 1.0900862216949463, "learning_rate": 1.6474580274874045e-05, "loss": 3.1475, "step": 108110 }, { "epoch": 0.0095232, "grad_norm": 0.7438094019889832, "learning_rate": 1.6473966550261085e-05, "loss": 3.0364, "step": 108120 }, { "epoch": 0.0095488, "grad_norm": 0.7224714159965515, "learning_rate": 1.6473352783666178e-05, "loss": 3.0043, "step": 108130 }, { "epoch": 0.0095744, "grad_norm": 0.7131929993629456, "learning_rate": 1.64727389750933e-05, "loss": 3.1871, "step": 108140 }, { "epoch": 0.0096, "grad_norm": 0.7156469821929932, "learning_rate": 1.6472125124546426e-05, "loss": 3.0712, "step": 108150 }, { "epoch": 0.0096256, "grad_norm": 0.8589982986450195, "learning_rate": 1.647151123202954e-05, "loss": 3.2554, "step": 108160 }, { "epoch": 0.0096512, "grad_norm": 0.9362379908561707, "learning_rate": 1.6470897297546625e-05, "loss": 3.1048, "step": 108170 }, { "epoch": 0.0096768, "grad_norm": 0.7637253403663635, "learning_rate": 1.647028332110166e-05, "loss": 3.188, "step": 108180 }, { "epoch": 0.0097024, "grad_norm": 0.7075826525688171, "learning_rate": 1.646966930269863e-05, "loss": 2.9026, "step": 108190 }, { "epoch": 0.009728, "grad_norm": 0.7496874928474426, "learning_rate": 1.646905524234151e-05, "loss": 2.9925, "step": 108200 }, { "epoch": 0.0097536, "grad_norm": 0.8628906011581421, "learning_rate": 1.646844114003429e-05, "loss": 3.1121, "step": 108210 }, { "epoch": 0.0097792, "grad_norm": 0.9313167333602905, "learning_rate": 1.6467826995780948e-05, "loss": 3.1725, "step": 108220 }, { "epoch": 0.0098048, "grad_norm": 0.8579789996147156, "learning_rate": 1.6467212809585465e-05, "loss": 3.2621, "step": 108230 }, { "epoch": 0.0098304, "grad_norm": 0.6816364526748657, "learning_rate": 1.6466598581451825e-05, "loss": 2.9742, "step": 108240 }, { "epoch": 0.009856, "grad_norm": 0.7679238319396973, "learning_rate": 1.6465984311384013e-05, "loss": 3.0013, "step": 108250 }, { "epoch": 0.0098816, "grad_norm": 0.9116413593292236, "learning_rate": 1.646536999938601e-05, "loss": 3.0725, "step": 108260 }, { "epoch": 0.0099072, "grad_norm": 0.7028353810310364, "learning_rate": 1.6464755645461802e-05, "loss": 3.206, "step": 108270 }, { "epoch": 0.0099328, "grad_norm": 0.8027864694595337, "learning_rate": 1.6464141249615374e-05, "loss": 2.9477, "step": 108280 }, { "epoch": 0.0099584, "grad_norm": 0.8073118329048157, "learning_rate": 1.6463526811850702e-05, "loss": 3.3381, "step": 108290 }, { "epoch": 0.009984, "grad_norm": 0.7513050436973572, "learning_rate": 1.6462912332171784e-05, "loss": 3.3718, "step": 108300 }, { "epoch": 2.56e-05, "grad_norm": 0.6897835731506348, "learning_rate": 1.646229781058259e-05, "loss": 2.5341, "step": 108310 }, { "epoch": 5.12e-05, "grad_norm": 0.7686895728111267, "learning_rate": 1.6461683247087116e-05, "loss": 3.0243, "step": 108320 }, { "epoch": 7.68e-05, "grad_norm": 0.7075116038322449, "learning_rate": 1.6461068641689345e-05, "loss": 3.0183, "step": 108330 }, { "epoch": 0.0001024, "grad_norm": 0.9204853773117065, "learning_rate": 1.646045399439326e-05, "loss": 3.0002, "step": 108340 }, { "epoch": 0.000128, "grad_norm": 0.7272182703018188, "learning_rate": 1.645983930520284e-05, "loss": 2.975, "step": 108350 }, { "epoch": 0.0001536, "grad_norm": 0.76381915807724, "learning_rate": 1.6459224574122088e-05, "loss": 2.7903, "step": 108360 }, { "epoch": 0.0001792, "grad_norm": 0.7883957624435425, "learning_rate": 1.645860980115498e-05, "loss": 3.0338, "step": 108370 }, { "epoch": 0.0002048, "grad_norm": 0.8513695597648621, "learning_rate": 1.6457994986305504e-05, "loss": 3.0225, "step": 108380 }, { "epoch": 0.0002304, "grad_norm": 0.676240861415863, "learning_rate": 1.6457380129577645e-05, "loss": 2.7814, "step": 108390 }, { "epoch": 0.000256, "grad_norm": 2.2927932739257812, "learning_rate": 1.645676523097539e-05, "loss": 3.0484, "step": 108400 }, { "epoch": 0.0002816, "grad_norm": 0.6944056749343872, "learning_rate": 1.6456150290502733e-05, "loss": 2.8325, "step": 108410 }, { "epoch": 0.0003072, "grad_norm": 0.7969713807106018, "learning_rate": 1.645553530816366e-05, "loss": 2.7941, "step": 108420 }, { "epoch": 0.0003328, "grad_norm": 0.7224138975143433, "learning_rate": 1.645492028396215e-05, "loss": 2.9119, "step": 108430 }, { "epoch": 0.0003584, "grad_norm": 0.8073437213897705, "learning_rate": 1.6454305217902197e-05, "loss": 3.2784, "step": 108440 }, { "epoch": 0.000384, "grad_norm": 0.7788808941841125, "learning_rate": 1.645369010998779e-05, "loss": 2.9198, "step": 108450 }, { "epoch": 0.0004096, "grad_norm": 0.645248532295227, "learning_rate": 1.645307496022292e-05, "loss": 2.7215, "step": 108460 }, { "epoch": 0.0004352, "grad_norm": 0.7473063468933105, "learning_rate": 1.6452459768611574e-05, "loss": 2.9786, "step": 108470 }, { "epoch": 0.0004608, "grad_norm": 0.7832837104797363, "learning_rate": 1.6451844535157745e-05, "loss": 3.5918, "step": 108480 }, { "epoch": 0.0004864, "grad_norm": 0.9366413354873657, "learning_rate": 1.645122925986541e-05, "loss": 3.0937, "step": 108490 }, { "epoch": 0.000512, "grad_norm": 0.9261021614074707, "learning_rate": 1.6450613942738573e-05, "loss": 2.8421, "step": 108500 }, { "epoch": 0.0005376, "grad_norm": 0.7751553058624268, "learning_rate": 1.644999858378122e-05, "loss": 2.9073, "step": 108510 }, { "epoch": 0.0005632, "grad_norm": 0.6837868690490723, "learning_rate": 1.644938318299734e-05, "loss": 2.7594, "step": 108520 }, { "epoch": 0.0005888, "grad_norm": 1.3542494773864746, "learning_rate": 1.644876774039092e-05, "loss": 2.9886, "step": 108530 }, { "epoch": 0.0006144, "grad_norm": 0.7065299153327942, "learning_rate": 1.6448152255965962e-05, "loss": 2.993, "step": 108540 }, { "epoch": 0.00064, "grad_norm": 0.7602637410163879, "learning_rate": 1.6447536729726446e-05, "loss": 3.2918, "step": 108550 }, { "epoch": 0.0006656, "grad_norm": 0.865060031414032, "learning_rate": 1.6446921161676368e-05, "loss": 2.8955, "step": 108560 }, { "epoch": 0.0006912, "grad_norm": 0.7535492777824402, "learning_rate": 1.6446305551819718e-05, "loss": 2.902, "step": 108570 }, { "epoch": 0.0007168, "grad_norm": 0.7395052909851074, "learning_rate": 1.6445689900160494e-05, "loss": 2.992, "step": 108580 }, { "epoch": 0.0007424, "grad_norm": 0.7658018469810486, "learning_rate": 1.6445074206702682e-05, "loss": 2.8719, "step": 108590 }, { "epoch": 0.000768, "grad_norm": 0.7486628890037537, "learning_rate": 1.6444458471450278e-05, "loss": 2.9107, "step": 108600 }, { "epoch": 0.0007936, "grad_norm": 0.7057726979255676, "learning_rate": 1.6443842694407268e-05, "loss": 2.9256, "step": 108610 }, { "epoch": 0.0008192, "grad_norm": 0.8538981676101685, "learning_rate": 1.6443226875577656e-05, "loss": 2.8665, "step": 108620 }, { "epoch": 0.0008448, "grad_norm": 0.781518816947937, "learning_rate": 1.644261101496543e-05, "loss": 2.8691, "step": 108630 }, { "epoch": 0.0008704, "grad_norm": 0.6652350425720215, "learning_rate": 1.644199511257458e-05, "loss": 2.7752, "step": 108640 }, { "epoch": 0.000896, "grad_norm": 0.6842780113220215, "learning_rate": 1.6441379168409107e-05, "loss": 2.989, "step": 108650 }, { "epoch": 0.0009216, "grad_norm": 1.1062923669815063, "learning_rate": 1.6440763182473003e-05, "loss": 3.2008, "step": 108660 }, { "epoch": 0.0009472, "grad_norm": 0.9181281328201294, "learning_rate": 1.644014715477026e-05, "loss": 2.7973, "step": 108670 }, { "epoch": 0.0009728, "grad_norm": 0.7602585554122925, "learning_rate": 1.6439531085304873e-05, "loss": 3.0874, "step": 108680 }, { "epoch": 0.0009984, "grad_norm": 0.6869996786117554, "learning_rate": 1.6438914974080843e-05, "loss": 3.0865, "step": 108690 }, { "epoch": 0.001024, "grad_norm": 0.899243950843811, "learning_rate": 1.6438298821102154e-05, "loss": 2.9599, "step": 108700 }, { "epoch": 0.0010496, "grad_norm": 0.7898143529891968, "learning_rate": 1.6437682626372812e-05, "loss": 3.0558, "step": 108710 }, { "epoch": 0.0010752, "grad_norm": 0.9058713912963867, "learning_rate": 1.6437066389896807e-05, "loss": 3.0932, "step": 108720 }, { "epoch": 0.0011008, "grad_norm": 0.6965468525886536, "learning_rate": 1.643645011167814e-05, "loss": 3.0131, "step": 108730 }, { "epoch": 0.0011264, "grad_norm": 0.7171657085418701, "learning_rate": 1.6435833791720804e-05, "loss": 2.9084, "step": 108740 }, { "epoch": 0.001152, "grad_norm": 0.783880889415741, "learning_rate": 1.6435217430028796e-05, "loss": 3.0373, "step": 108750 }, { "epoch": 0.0011776, "grad_norm": 0.6884785294532776, "learning_rate": 1.6434601026606114e-05, "loss": 3.1903, "step": 108760 }, { "epoch": 0.0012032, "grad_norm": 0.7748542428016663, "learning_rate": 1.6433984581456753e-05, "loss": 3.0552, "step": 108770 }, { "epoch": 0.0012288, "grad_norm": 0.6694175004959106, "learning_rate": 1.6433368094584713e-05, "loss": 3.0583, "step": 108780 }, { "epoch": 0.0012544, "grad_norm": 0.6583889722824097, "learning_rate": 1.643275156599399e-05, "loss": 2.8041, "step": 108790 }, { "epoch": 0.00128, "grad_norm": 0.607729434967041, "learning_rate": 1.643213499568858e-05, "loss": 2.8557, "step": 108800 }, { "epoch": 0.0013056, "grad_norm": 0.7252740263938904, "learning_rate": 1.6431518383672487e-05, "loss": 2.9773, "step": 108810 }, { "epoch": 0.0013312, "grad_norm": 0.8144744038581848, "learning_rate": 1.643090172994971e-05, "loss": 3.1196, "step": 108820 }, { "epoch": 0.0013568, "grad_norm": 0.8157680630683899, "learning_rate": 1.643028503452424e-05, "loss": 3.1096, "step": 108830 }, { "epoch": 0.0013824, "grad_norm": 0.7053258419036865, "learning_rate": 1.642966829740008e-05, "loss": 2.996, "step": 108840 }, { "epoch": 0.001408, "grad_norm": 0.693078875541687, "learning_rate": 1.642905151858123e-05, "loss": 3.034, "step": 108850 }, { "epoch": 0.0014336, "grad_norm": 0.7506831884384155, "learning_rate": 1.642843469807169e-05, "loss": 3.0226, "step": 108860 }, { "epoch": 0.0014592, "grad_norm": 0.7563691735267639, "learning_rate": 1.642781783587546e-05, "loss": 3.0249, "step": 108870 }, { "epoch": 0.0014848, "grad_norm": 0.6872094869613647, "learning_rate": 1.642720093199654e-05, "loss": 2.7929, "step": 108880 }, { "epoch": 0.0015104, "grad_norm": 0.7158775925636292, "learning_rate": 1.642658398643893e-05, "loss": 2.7193, "step": 108890 }, { "epoch": 0.001536, "grad_norm": 0.7453139424324036, "learning_rate": 1.642596699920663e-05, "loss": 2.9178, "step": 108900 }, { "epoch": 2.56e-05, "grad_norm": 0.7597218751907349, "learning_rate": 1.6425349970303645e-05, "loss": 2.7085, "step": 108910 }, { "epoch": 5.12e-05, "grad_norm": 1.255072832107544, "learning_rate": 1.642473289973397e-05, "loss": 2.7952, "step": 108920 }, { "epoch": 7.68e-05, "grad_norm": 0.9246278405189514, "learning_rate": 1.642411578750161e-05, "loss": 2.8868, "step": 108930 }, { "epoch": 0.0001024, "grad_norm": 0.6460037231445312, "learning_rate": 1.642349863361057e-05, "loss": 2.6957, "step": 108940 }, { "epoch": 0.000128, "grad_norm": 0.7184552550315857, "learning_rate": 1.6422881438064843e-05, "loss": 3.0924, "step": 108950 }, { "epoch": 0.0001536, "grad_norm": 0.7169367074966431, "learning_rate": 1.6422264200868437e-05, "loss": 2.9461, "step": 108960 }, { "epoch": 0.0001792, "grad_norm": 0.6731632947921753, "learning_rate": 1.6421646922025356e-05, "loss": 2.7879, "step": 108970 }, { "epoch": 0.0002048, "grad_norm": 0.7125076651573181, "learning_rate": 1.6421029601539604e-05, "loss": 2.8765, "step": 108980 }, { "epoch": 0.0002304, "grad_norm": 0.6251295804977417, "learning_rate": 1.6420412239415176e-05, "loss": 3.0113, "step": 108990 }, { "epoch": 0.000256, "grad_norm": 0.8508307933807373, "learning_rate": 1.6419794835656087e-05, "loss": 3.042, "step": 109000 }, { "epoch": 0.0002816, "grad_norm": 0.8576762080192566, "learning_rate": 1.641917739026633e-05, "loss": 3.0315, "step": 109010 }, { "epoch": 0.0003072, "grad_norm": 0.6977095603942871, "learning_rate": 1.641855990324992e-05, "loss": 2.75, "step": 109020 }, { "epoch": 0.0003328, "grad_norm": 0.7316831946372986, "learning_rate": 1.641794237461085e-05, "loss": 2.9584, "step": 109030 }, { "epoch": 0.0003584, "grad_norm": 0.6828370094299316, "learning_rate": 1.6417324804353127e-05, "loss": 3.1065, "step": 109040 }, { "epoch": 0.000384, "grad_norm": 0.6860023140907288, "learning_rate": 1.6416707192480764e-05, "loss": 2.9439, "step": 109050 }, { "epoch": 0.0004096, "grad_norm": 0.6908953189849854, "learning_rate": 1.6416089538997758e-05, "loss": 2.8961, "step": 109060 }, { "epoch": 0.0004352, "grad_norm": 0.8281963467597961, "learning_rate": 1.6415471843908114e-05, "loss": 3.1089, "step": 109070 }, { "epoch": 0.0004608, "grad_norm": 0.747322142124176, "learning_rate": 1.6414854107215843e-05, "loss": 2.9846, "step": 109080 }, { "epoch": 0.0004864, "grad_norm": 0.8100401759147644, "learning_rate": 1.6414236328924945e-05, "loss": 3.0193, "step": 109090 }, { "epoch": 0.000512, "grad_norm": 0.7461996078491211, "learning_rate": 1.6413618509039433e-05, "loss": 2.9747, "step": 109100 }, { "epoch": 0.0005376, "grad_norm": 0.841895341873169, "learning_rate": 1.641300064756331e-05, "loss": 3.0423, "step": 109110 }, { "epoch": 0.0005632, "grad_norm": 0.6942349076271057, "learning_rate": 1.6412382744500578e-05, "loss": 2.8355, "step": 109120 }, { "epoch": 0.0005888, "grad_norm": 0.7446191310882568, "learning_rate": 1.641176479985525e-05, "loss": 2.8447, "step": 109130 }, { "epoch": 0.0006144, "grad_norm": 0.7702944874763489, "learning_rate": 1.641114681363133e-05, "loss": 2.7632, "step": 109140 }, { "epoch": 0.00064, "grad_norm": 0.7219663858413696, "learning_rate": 1.641052878583283e-05, "loss": 2.7991, "step": 109150 }, { "epoch": 0.0006656, "grad_norm": 0.8200390338897705, "learning_rate": 1.640991071646375e-05, "loss": 3.2433, "step": 109160 }, { "epoch": 0.0006912, "grad_norm": 0.7175070643424988, "learning_rate": 1.640929260552811e-05, "loss": 3.4157, "step": 109170 }, { "epoch": 0.0007168, "grad_norm": 0.8494502902030945, "learning_rate": 1.6408674453029903e-05, "loss": 2.8942, "step": 109180 }, { "epoch": 0.0007424, "grad_norm": 0.9322285652160645, "learning_rate": 1.6408056258973147e-05, "loss": 3.0378, "step": 109190 }, { "epoch": 0.000768, "grad_norm": 0.8085647225379944, "learning_rate": 1.6407438023361852e-05, "loss": 3.0451, "step": 109200 }, { "epoch": 0.0007936, "grad_norm": 0.778186559677124, "learning_rate": 1.6406819746200023e-05, "loss": 2.7846, "step": 109210 }, { "epoch": 0.0008192, "grad_norm": 0.6736340522766113, "learning_rate": 1.640620142749167e-05, "loss": 2.859, "step": 109220 }, { "epoch": 0.0008448, "grad_norm": 1.1742340326309204, "learning_rate": 1.6405583067240804e-05, "loss": 2.8861, "step": 109230 }, { "epoch": 0.0008704, "grad_norm": 0.8180504441261292, "learning_rate": 1.6404964665451433e-05, "loss": 2.6987, "step": 109240 }, { "epoch": 0.000896, "grad_norm": 0.7496407628059387, "learning_rate": 1.640434622212757e-05, "loss": 2.9762, "step": 109250 }, { "epoch": 0.0009216, "grad_norm": 0.736904501914978, "learning_rate": 1.6403727737273226e-05, "loss": 2.6385, "step": 109260 }, { "epoch": 0.0009472, "grad_norm": 0.7290767431259155, "learning_rate": 1.6403109210892407e-05, "loss": 3.1347, "step": 109270 }, { "epoch": 0.0009728, "grad_norm": 0.7185920476913452, "learning_rate": 1.6402490642989124e-05, "loss": 2.9914, "step": 109280 }, { "epoch": 0.0009984, "grad_norm": 0.8998292684555054, "learning_rate": 1.6401872033567394e-05, "loss": 2.8871, "step": 109290 }, { "epoch": 0.001024, "grad_norm": 0.681910514831543, "learning_rate": 1.6401253382631223e-05, "loss": 2.8216, "step": 109300 }, { "epoch": 0.0010496, "grad_norm": 0.893530011177063, "learning_rate": 1.6400634690184627e-05, "loss": 3.0389, "step": 109310 }, { "epoch": 0.0010752, "grad_norm": 0.7777000069618225, "learning_rate": 1.6400015956231615e-05, "loss": 3.047, "step": 109320 }, { "epoch": 0.0011008, "grad_norm": 1.0641225576400757, "learning_rate": 1.6399397180776204e-05, "loss": 2.8795, "step": 109330 }, { "epoch": 0.0011264, "grad_norm": 0.6738158464431763, "learning_rate": 1.63987783638224e-05, "loss": 2.9094, "step": 109340 }, { "epoch": 0.001152, "grad_norm": 0.7399192452430725, "learning_rate": 1.6398159505374218e-05, "loss": 2.6722, "step": 109350 }, { "epoch": 0.0011776, "grad_norm": 0.864986777305603, "learning_rate": 1.639754060543567e-05, "loss": 3.0524, "step": 109360 }, { "epoch": 0.0012032, "grad_norm": 0.7791028022766113, "learning_rate": 1.6396921664010777e-05, "loss": 3.0866, "step": 109370 }, { "epoch": 0.0012288, "grad_norm": 0.6941142082214355, "learning_rate": 1.6396302681103543e-05, "loss": 3.3288, "step": 109380 }, { "epoch": 0.0012544, "grad_norm": 0.7293569445610046, "learning_rate": 1.639568365671799e-05, "loss": 3.2283, "step": 109390 }, { "epoch": 0.00128, "grad_norm": 0.724479615688324, "learning_rate": 1.6395064590858125e-05, "loss": 2.8166, "step": 109400 }, { "epoch": 0.0013056, "grad_norm": 0.7786470651626587, "learning_rate": 1.639444548352797e-05, "loss": 2.7999, "step": 109410 }, { "epoch": 0.0013312, "grad_norm": 0.8045465350151062, "learning_rate": 1.6393826334731527e-05, "loss": 2.9408, "step": 109420 }, { "epoch": 0.0013568, "grad_norm": 0.8779246807098389, "learning_rate": 1.6393207144472826e-05, "loss": 3.0603, "step": 109430 }, { "epoch": 0.0013824, "grad_norm": 0.7299241423606873, "learning_rate": 1.6392587912755874e-05, "loss": 3.2008, "step": 109440 }, { "epoch": 0.001408, "grad_norm": 0.8397967219352722, "learning_rate": 1.639196863958469e-05, "loss": 2.8264, "step": 109450 }, { "epoch": 0.0014336, "grad_norm": 1.2428473234176636, "learning_rate": 1.6391349324963286e-05, "loss": 3.0207, "step": 109460 }, { "epoch": 0.0014592, "grad_norm": 0.7705429196357727, "learning_rate": 1.639072996889568e-05, "loss": 2.6315, "step": 109470 }, { "epoch": 0.0014848, "grad_norm": 0.6992063522338867, "learning_rate": 1.639011057138589e-05, "loss": 2.772, "step": 109480 }, { "epoch": 0.0015104, "grad_norm": 0.7168248295783997, "learning_rate": 1.638949113243793e-05, "loss": 2.6933, "step": 109490 }, { "epoch": 0.001536, "grad_norm": 0.7225431799888611, "learning_rate": 1.638887165205582e-05, "loss": 2.729, "step": 109500 }, { "epoch": 0.0015616, "grad_norm": 0.6812772750854492, "learning_rate": 1.6388252130243576e-05, "loss": 2.9597, "step": 109510 }, { "epoch": 0.0015872, "grad_norm": 0.8535072207450867, "learning_rate": 1.6387632567005213e-05, "loss": 3.1757, "step": 109520 }, { "epoch": 0.0016128, "grad_norm": 0.6944493055343628, "learning_rate": 1.638701296234475e-05, "loss": 3.3474, "step": 109530 }, { "epoch": 0.0016384, "grad_norm": 0.7052708268165588, "learning_rate": 1.6386393316266206e-05, "loss": 2.9905, "step": 109540 }, { "epoch": 0.001664, "grad_norm": 0.7513744831085205, "learning_rate": 1.6385773628773598e-05, "loss": 2.902, "step": 109550 }, { "epoch": 0.0016896, "grad_norm": 0.7578203082084656, "learning_rate": 1.6385153899870946e-05, "loss": 3.0754, "step": 109560 }, { "epoch": 0.0017152, "grad_norm": 0.740742027759552, "learning_rate": 1.6384534129562266e-05, "loss": 2.9398, "step": 109570 }, { "epoch": 0.0017408, "grad_norm": 0.7161943912506104, "learning_rate": 1.6383914317851582e-05, "loss": 2.9677, "step": 109580 }, { "epoch": 0.0017664, "grad_norm": 0.6748253107070923, "learning_rate": 1.638329446474291e-05, "loss": 2.5844, "step": 109590 }, { "epoch": 0.001792, "grad_norm": 0.7739773988723755, "learning_rate": 1.6382674570240267e-05, "loss": 2.8301, "step": 109600 }, { "epoch": 0.0018176, "grad_norm": 0.7243894934654236, "learning_rate": 1.6382054634347678e-05, "loss": 2.828, "step": 109610 }, { "epoch": 0.0018432, "grad_norm": 1.7109932899475098, "learning_rate": 1.6381434657069158e-05, "loss": 2.9161, "step": 109620 }, { "epoch": 0.0018688, "grad_norm": 0.7186312079429626, "learning_rate": 1.6380814638408735e-05, "loss": 2.8304, "step": 109630 }, { "epoch": 0.0018944, "grad_norm": 0.6777514219284058, "learning_rate": 1.638019457837042e-05, "loss": 2.8207, "step": 109640 }, { "epoch": 0.00192, "grad_norm": 0.7202258706092834, "learning_rate": 1.6379574476958242e-05, "loss": 3.2325, "step": 109650 }, { "epoch": 0.0019456, "grad_norm": 0.6964299082756042, "learning_rate": 1.6378954334176217e-05, "loss": 2.9743, "step": 109660 }, { "epoch": 0.0019712, "grad_norm": 0.6956004500389099, "learning_rate": 1.637833415002837e-05, "loss": 2.7886, "step": 109670 }, { "epoch": 0.0019968, "grad_norm": 1.0623286962509155, "learning_rate": 1.6377713924518725e-05, "loss": 2.9018, "step": 109680 }, { "epoch": 0.0020224, "grad_norm": 0.8570565581321716, "learning_rate": 1.6377093657651292e-05, "loss": 2.888, "step": 109690 }, { "epoch": 0.002048, "grad_norm": 0.8623736500740051, "learning_rate": 1.6376473349430108e-05, "loss": 2.8586, "step": 109700 }, { "epoch": 0.0020736, "grad_norm": 0.7334372997283936, "learning_rate": 1.6375852999859184e-05, "loss": 2.8782, "step": 109710 }, { "epoch": 0.0020992, "grad_norm": 0.7809265851974487, "learning_rate": 1.6375232608942553e-05, "loss": 2.9122, "step": 109720 }, { "epoch": 0.0021248, "grad_norm": 0.7888941168785095, "learning_rate": 1.637461217668423e-05, "loss": 3.0213, "step": 109730 }, { "epoch": 0.0021504, "grad_norm": 0.7460488677024841, "learning_rate": 1.6373991703088237e-05, "loss": 3.0167, "step": 109740 }, { "epoch": 0.002176, "grad_norm": 0.703356146812439, "learning_rate": 1.637337118815861e-05, "loss": 3.0638, "step": 109750 }, { "epoch": 0.0022016, "grad_norm": 0.8764145970344543, "learning_rate": 1.6372750631899362e-05, "loss": 3.0239, "step": 109760 }, { "epoch": 0.0022272, "grad_norm": 0.7432277202606201, "learning_rate": 1.637213003431452e-05, "loss": 3.0428, "step": 109770 }, { "epoch": 0.0022528, "grad_norm": 0.6622296571731567, "learning_rate": 1.6371509395408104e-05, "loss": 2.7247, "step": 109780 }, { "epoch": 0.0022784, "grad_norm": 0.7132869362831116, "learning_rate": 1.6370888715184148e-05, "loss": 3.1646, "step": 109790 }, { "epoch": 0.002304, "grad_norm": 0.8290990591049194, "learning_rate": 1.637026799364667e-05, "loss": 2.7209, "step": 109800 }, { "epoch": 0.0023296, "grad_norm": 1.1123483180999756, "learning_rate": 1.6369647230799695e-05, "loss": 2.3558, "step": 109810 }, { "epoch": 0.0023552, "grad_norm": 0.7938584685325623, "learning_rate": 1.6369026426647253e-05, "loss": 2.816, "step": 109820 }, { "epoch": 0.0023808, "grad_norm": 0.7692499756813049, "learning_rate": 1.6368405581193365e-05, "loss": 3.0149, "step": 109830 }, { "epoch": 0.0024064, "grad_norm": 1.1552646160125732, "learning_rate": 1.6367784694442065e-05, "loss": 3.0775, "step": 109840 }, { "epoch": 0.002432, "grad_norm": 0.7281319499015808, "learning_rate": 1.6367163766397366e-05, "loss": 2.9526, "step": 109850 }, { "epoch": 0.0024576, "grad_norm": 0.6746090054512024, "learning_rate": 1.6366542797063307e-05, "loss": 3.0487, "step": 109860 }, { "epoch": 0.0024832, "grad_norm": 1.0546246767044067, "learning_rate": 1.6365921786443907e-05, "loss": 2.955, "step": 109870 }, { "epoch": 0.0025088, "grad_norm": 0.7297434210777283, "learning_rate": 1.6365300734543198e-05, "loss": 3.1081, "step": 109880 }, { "epoch": 0.0025344, "grad_norm": 0.6769047379493713, "learning_rate": 1.6364679641365208e-05, "loss": 2.9874, "step": 109890 }, { "epoch": 0.00256, "grad_norm": 2.360194444656372, "learning_rate": 1.636405850691396e-05, "loss": 2.893, "step": 109900 }, { "epoch": 0.0025856, "grad_norm": 0.793959379196167, "learning_rate": 1.6363437331193482e-05, "loss": 2.9055, "step": 109910 }, { "epoch": 0.0026112, "grad_norm": 0.8775243163108826, "learning_rate": 1.6362816114207806e-05, "loss": 3.0941, "step": 109920 }, { "epoch": 0.0026368, "grad_norm": 1.0284172296524048, "learning_rate": 1.6362194855960957e-05, "loss": 3.235, "step": 109930 }, { "epoch": 0.0026624, "grad_norm": 0.7894393801689148, "learning_rate": 1.6361573556456965e-05, "loss": 3.0163, "step": 109940 }, { "epoch": 0.002688, "grad_norm": 0.7595880031585693, "learning_rate": 1.636095221569986e-05, "loss": 2.9777, "step": 109950 }, { "epoch": 0.0027136, "grad_norm": 0.7140309810638428, "learning_rate": 1.636033083369367e-05, "loss": 3.0333, "step": 109960 }, { "epoch": 0.0027392, "grad_norm": 0.8013100624084473, "learning_rate": 1.6359709410442425e-05, "loss": 3.1413, "step": 109970 }, { "epoch": 0.0027648, "grad_norm": 0.747697651386261, "learning_rate": 1.6359087945950158e-05, "loss": 2.6264, "step": 109980 }, { "epoch": 0.0027904, "grad_norm": 0.8588478565216064, "learning_rate": 1.635846644022089e-05, "loss": 3.1126, "step": 109990 }, { "epoch": 0.002816, "grad_norm": 0.7043382525444031, "learning_rate": 1.6357844893258664e-05, "loss": 3.005, "step": 110000 }, { "epoch": 0.0028416, "grad_norm": 0.6864445209503174, "learning_rate": 1.6357223305067498e-05, "loss": 2.8924, "step": 110010 }, { "epoch": 0.0028672, "grad_norm": 0.7480781674385071, "learning_rate": 1.635660167565143e-05, "loss": 2.9966, "step": 110020 }, { "epoch": 0.0028928, "grad_norm": 0.811230480670929, "learning_rate": 1.6355980005014492e-05, "loss": 3.112, "step": 110030 }, { "epoch": 0.0029184, "grad_norm": 0.7507380843162537, "learning_rate": 1.635535829316071e-05, "loss": 3.1882, "step": 110040 }, { "epoch": 0.002944, "grad_norm": 0.8145148754119873, "learning_rate": 1.635473654009412e-05, "loss": 2.9757, "step": 110050 }, { "epoch": 0.0029696, "grad_norm": 0.7071405053138733, "learning_rate": 1.635411474581875e-05, "loss": 3.2168, "step": 110060 }, { "epoch": 0.0029952, "grad_norm": 1.0296987295150757, "learning_rate": 1.6353492910338635e-05, "loss": 3.0319, "step": 110070 }, { "epoch": 0.0030208, "grad_norm": 0.6980001330375671, "learning_rate": 1.6352871033657807e-05, "loss": 3.0725, "step": 110080 }, { "epoch": 0.0030464, "grad_norm": 0.6727724671363831, "learning_rate": 1.63522491157803e-05, "loss": 3.079, "step": 110090 }, { "epoch": 0.003072, "grad_norm": 0.7842531800270081, "learning_rate": 1.6351627156710147e-05, "loss": 3.1956, "step": 110100 }, { "epoch": 0.0030976, "grad_norm": 0.8066098690032959, "learning_rate": 1.635100515645138e-05, "loss": 2.971, "step": 110110 }, { "epoch": 0.0031232, "grad_norm": 0.7829585671424866, "learning_rate": 1.6350383115008028e-05, "loss": 3.2698, "step": 110120 }, { "epoch": 0.0031488, "grad_norm": 0.873510479927063, "learning_rate": 1.6349761032384133e-05, "loss": 3.1291, "step": 110130 }, { "epoch": 0.0031744, "grad_norm": 0.7786264419555664, "learning_rate": 1.6349138908583728e-05, "loss": 3.1655, "step": 110140 }, { "epoch": 0.0032, "grad_norm": 0.7077651619911194, "learning_rate": 1.634851674361084e-05, "loss": 3.1524, "step": 110150 }, { "epoch": 0.0032256, "grad_norm": 0.8179801106452942, "learning_rate": 1.634789453746951e-05, "loss": 3.1141, "step": 110160 }, { "epoch": 0.0032512, "grad_norm": 0.7571385502815247, "learning_rate": 1.6347272290163766e-05, "loss": 3.316, "step": 110170 }, { "epoch": 0.0032768, "grad_norm": 0.7350624203681946, "learning_rate": 1.634665000169765e-05, "loss": 2.9669, "step": 110180 }, { "epoch": 0.0033024, "grad_norm": 0.7436214089393616, "learning_rate": 1.6346027672075198e-05, "loss": 2.9467, "step": 110190 }, { "epoch": 0.003328, "grad_norm": 0.732567310333252, "learning_rate": 1.6345405301300446e-05, "loss": 3.1849, "step": 110200 }, { "epoch": 0.0033536, "grad_norm": 0.6523665189743042, "learning_rate": 1.6344782889377422e-05, "loss": 3.0633, "step": 110210 }, { "epoch": 0.0033792, "grad_norm": 0.7312653064727783, "learning_rate": 1.634416043631017e-05, "loss": 2.9546, "step": 110220 }, { "epoch": 0.0034048, "grad_norm": 0.7066973447799683, "learning_rate": 1.6343537942102724e-05, "loss": 3.0314, "step": 110230 }, { "epoch": 0.0034304, "grad_norm": 0.6600110530853271, "learning_rate": 1.634291540675912e-05, "loss": 3.2702, "step": 110240 }, { "epoch": 0.003456, "grad_norm": 1.711883544921875, "learning_rate": 1.634229283028339e-05, "loss": 3.0255, "step": 110250 }, { "epoch": 0.0034816, "grad_norm": 0.700640082359314, "learning_rate": 1.6341670212679583e-05, "loss": 2.9968, "step": 110260 }, { "epoch": 0.0035072, "grad_norm": 0.7509238719940186, "learning_rate": 1.6341047553951728e-05, "loss": 3.0516, "step": 110270 }, { "epoch": 0.0035328, "grad_norm": 0.8118425607681274, "learning_rate": 1.6340424854103866e-05, "loss": 2.9255, "step": 110280 }, { "epoch": 0.0035584, "grad_norm": 0.9941901564598083, "learning_rate": 1.633980211314003e-05, "loss": 3.1428, "step": 110290 }, { "epoch": 0.003584, "grad_norm": 0.7041386961936951, "learning_rate": 1.6339179331064268e-05, "loss": 3.1262, "step": 110300 }, { "epoch": 0.0036096, "grad_norm": 0.7050407528877258, "learning_rate": 1.633855650788061e-05, "loss": 3.1482, "step": 110310 }, { "epoch": 0.0036352, "grad_norm": 0.9290329217910767, "learning_rate": 1.6337933643593098e-05, "loss": 3.3634, "step": 110320 }, { "epoch": 0.0036608, "grad_norm": 0.7191817164421082, "learning_rate": 1.633731073820577e-05, "loss": 3.1024, "step": 110330 }, { "epoch": 0.0036864, "grad_norm": 0.7577074766159058, "learning_rate": 1.6336687791722665e-05, "loss": 3.0352, "step": 110340 }, { "epoch": 0.003712, "grad_norm": 0.7251332402229309, "learning_rate": 1.6336064804147824e-05, "loss": 3.156, "step": 110350 }, { "epoch": 0.0037376, "grad_norm": 0.7202481031417847, "learning_rate": 1.633544177548529e-05, "loss": 2.8909, "step": 110360 }, { "epoch": 0.0037632, "grad_norm": 0.7187169790267944, "learning_rate": 1.6334818705739098e-05, "loss": 2.933, "step": 110370 }, { "epoch": 0.0037888, "grad_norm": 0.7583009600639343, "learning_rate": 1.633419559491329e-05, "loss": 2.9158, "step": 110380 }, { "epoch": 0.0038144, "grad_norm": 0.8327713012695312, "learning_rate": 1.6333572443011905e-05, "loss": 3.1541, "step": 110390 }, { "epoch": 0.00384, "grad_norm": 0.76065593957901, "learning_rate": 1.6332949250038988e-05, "loss": 3.0522, "step": 110400 }, { "epoch": 0.0038656, "grad_norm": 0.8228922486305237, "learning_rate": 1.6332326015998576e-05, "loss": 3.1396, "step": 110410 }, { "epoch": 0.0038912, "grad_norm": 0.7606241703033447, "learning_rate": 1.633170274089472e-05, "loss": 3.2205, "step": 110420 }, { "epoch": 0.0039168, "grad_norm": 1.2904744148254395, "learning_rate": 1.6331079424731447e-05, "loss": 3.035, "step": 110430 }, { "epoch": 0.0039424, "grad_norm": 0.819921612739563, "learning_rate": 1.6330456067512807e-05, "loss": 3.0059, "step": 110440 }, { "epoch": 0.003968, "grad_norm": 1.089281439781189, "learning_rate": 1.6329832669242844e-05, "loss": 3.0499, "step": 110450 }, { "epoch": 0.0039936, "grad_norm": 0.6822454333305359, "learning_rate": 1.6329209229925597e-05, "loss": 3.043, "step": 110460 }, { "epoch": 0.0040192, "grad_norm": 0.9145568609237671, "learning_rate": 1.632858574956511e-05, "loss": 3.061, "step": 110470 }, { "epoch": 0.0040448, "grad_norm": 0.8040980696678162, "learning_rate": 1.632796222816543e-05, "loss": 3.0993, "step": 110480 }, { "epoch": 0.0040704, "grad_norm": 0.7924991250038147, "learning_rate": 1.6327338665730596e-05, "loss": 3.1581, "step": 110490 }, { "epoch": 0.004096, "grad_norm": 0.7832645177841187, "learning_rate": 1.6326715062264652e-05, "loss": 3.0464, "step": 110500 }, { "epoch": 0.0041216, "grad_norm": 0.9385616779327393, "learning_rate": 1.632609141777164e-05, "loss": 2.9522, "step": 110510 }, { "epoch": 0.0041472, "grad_norm": 0.9446231722831726, "learning_rate": 1.6325467732255603e-05, "loss": 2.9506, "step": 110520 }, { "epoch": 0.0041728, "grad_norm": 0.7506377100944519, "learning_rate": 1.6324844005720593e-05, "loss": 3.0902, "step": 110530 }, { "epoch": 0.0041984, "grad_norm": 0.732115626335144, "learning_rate": 1.6324220238170655e-05, "loss": 3.001, "step": 110540 }, { "epoch": 0.004224, "grad_norm": 0.8549696803092957, "learning_rate": 1.6323596429609825e-05, "loss": 3.3517, "step": 110550 }, { "epoch": 0.0042496, "grad_norm": 0.753413200378418, "learning_rate": 1.632297258004215e-05, "loss": 3.0514, "step": 110560 }, { "epoch": 0.0042752, "grad_norm": 0.7886752486228943, "learning_rate": 1.6322348689471682e-05, "loss": 2.8868, "step": 110570 }, { "epoch": 0.0043008, "grad_norm": 0.753720760345459, "learning_rate": 1.6321724757902464e-05, "loss": 3.0394, "step": 110580 }, { "epoch": 0.0043264, "grad_norm": 0.7266347408294678, "learning_rate": 1.632110078533854e-05, "loss": 2.9066, "step": 110590 }, { "epoch": 0.004352, "grad_norm": 0.7671356797218323, "learning_rate": 1.6320476771783958e-05, "loss": 3.0269, "step": 110600 }, { "epoch": 0.0043776, "grad_norm": 0.7487277388572693, "learning_rate": 1.6319852717242758e-05, "loss": 3.0021, "step": 110610 }, { "epoch": 0.0044032, "grad_norm": 0.7593289017677307, "learning_rate": 1.6319228621719e-05, "loss": 3.0971, "step": 110620 }, { "epoch": 0.0044288, "grad_norm": 0.8011700510978699, "learning_rate": 1.631860448521672e-05, "loss": 2.9709, "step": 110630 }, { "epoch": 0.0044544, "grad_norm": 0.7659568786621094, "learning_rate": 1.6317980307739972e-05, "loss": 2.8211, "step": 110640 }, { "epoch": 0.00448, "grad_norm": 0.7425219416618347, "learning_rate": 1.63173560892928e-05, "loss": 3.2373, "step": 110650 }, { "epoch": 0.0045056, "grad_norm": 0.6874550580978394, "learning_rate": 1.631673182987925e-05, "loss": 3.0004, "step": 110660 }, { "epoch": 0.0045312, "grad_norm": 0.764357328414917, "learning_rate": 1.6316107529503376e-05, "loss": 3.04, "step": 110670 }, { "epoch": 0.0045568, "grad_norm": 0.7528886795043945, "learning_rate": 1.6315483188169223e-05, "loss": 3.266, "step": 110680 }, { "epoch": 0.0045824, "grad_norm": 0.7289993762969971, "learning_rate": 1.631485880588084e-05, "loss": 3.0356, "step": 110690 }, { "epoch": 0.004608, "grad_norm": 1.6766692399978638, "learning_rate": 1.6314234382642272e-05, "loss": 3.1649, "step": 110700 }, { "epoch": 0.0046336, "grad_norm": 0.7856569886207581, "learning_rate": 1.6313609918457575e-05, "loss": 3.2161, "step": 110710 }, { "epoch": 0.0046592, "grad_norm": 0.8013682961463928, "learning_rate": 1.6312985413330798e-05, "loss": 2.968, "step": 110720 }, { "epoch": 0.0046848, "grad_norm": 0.7956802248954773, "learning_rate": 1.6312360867265986e-05, "loss": 3.1613, "step": 110730 }, { "epoch": 0.0047104, "grad_norm": 0.9066449403762817, "learning_rate": 1.6311736280267194e-05, "loss": 3.1754, "step": 110740 }, { "epoch": 0.004736, "grad_norm": 0.740561842918396, "learning_rate": 1.6311111652338464e-05, "loss": 2.9844, "step": 110750 }, { "epoch": 0.0047616, "grad_norm": 0.8797049522399902, "learning_rate": 1.6310486983483857e-05, "loss": 3.2434, "step": 110760 }, { "epoch": 0.0047872, "grad_norm": 0.7330320477485657, "learning_rate": 1.630986227370742e-05, "loss": 3.004, "step": 110770 }, { "epoch": 0.0048128, "grad_norm": 1.1252665519714355, "learning_rate": 1.6309237523013196e-05, "loss": 3.1527, "step": 110780 }, { "epoch": 0.0048384, "grad_norm": 0.7754290699958801, "learning_rate": 1.630861273140525e-05, "loss": 3.0308, "step": 110790 }, { "epoch": 0.004864, "grad_norm": 0.7779684066772461, "learning_rate": 1.6307987898887625e-05, "loss": 3.1193, "step": 110800 }, { "epoch": 0.0048896, "grad_norm": 0.7372971773147583, "learning_rate": 1.6307363025464376e-05, "loss": 3.1257, "step": 110810 }, { "epoch": 0.0049152, "grad_norm": 0.7431859374046326, "learning_rate": 1.6306738111139554e-05, "loss": 2.9321, "step": 110820 }, { "epoch": 0.0049408, "grad_norm": 0.8030596971511841, "learning_rate": 1.630611315591721e-05, "loss": 3.1489, "step": 110830 }, { "epoch": 0.0049664, "grad_norm": 0.8115175366401672, "learning_rate": 1.63054881598014e-05, "loss": 3.1209, "step": 110840 }, { "epoch": 0.004992, "grad_norm": 0.6909031271934509, "learning_rate": 1.630486312279617e-05, "loss": 3.0113, "step": 110850 }, { "epoch": 0.0050176, "grad_norm": 0.7172101736068726, "learning_rate": 1.6304238044905583e-05, "loss": 3.0142, "step": 110860 }, { "epoch": 0.0050432, "grad_norm": 0.741071343421936, "learning_rate": 1.630361292613369e-05, "loss": 2.8768, "step": 110870 }, { "epoch": 0.0050688, "grad_norm": 0.6890760064125061, "learning_rate": 1.630298776648454e-05, "loss": 2.8811, "step": 110880 }, { "epoch": 0.0050944, "grad_norm": 0.8436999320983887, "learning_rate": 1.6302362565962185e-05, "loss": 3.0772, "step": 110890 }, { "epoch": 0.00512, "grad_norm": 0.6413560509681702, "learning_rate": 1.630173732457069e-05, "loss": 2.9334, "step": 110900 }, { "epoch": 0.0051456, "grad_norm": 1.425009846687317, "learning_rate": 1.63011120423141e-05, "loss": 3.1019, "step": 110910 }, { "epoch": 0.0051712, "grad_norm": 0.7647849917411804, "learning_rate": 1.6300486719196474e-05, "loss": 3.1119, "step": 110920 }, { "epoch": 0.0051968, "grad_norm": 0.6594929695129395, "learning_rate": 1.6299861355221867e-05, "loss": 3.1461, "step": 110930 }, { "epoch": 0.0052224, "grad_norm": 0.7726595401763916, "learning_rate": 1.6299235950394332e-05, "loss": 3.0581, "step": 110940 }, { "epoch": 0.005248, "grad_norm": 0.7645649313926697, "learning_rate": 1.6298610504717928e-05, "loss": 3.08, "step": 110950 }, { "epoch": 0.0052736, "grad_norm": 0.7546001076698303, "learning_rate": 1.6297985018196705e-05, "loss": 3.2261, "step": 110960 }, { "epoch": 0.0052992, "grad_norm": 0.6845559477806091, "learning_rate": 1.6297359490834724e-05, "loss": 3.044, "step": 110970 }, { "epoch": 0.0053248, "grad_norm": 0.6626899242401123, "learning_rate": 1.6296733922636045e-05, "loss": 3.0822, "step": 110980 }, { "epoch": 0.0053504, "grad_norm": 1.1044411659240723, "learning_rate": 1.6296108313604716e-05, "loss": 3.031, "step": 110990 }, { "epoch": 0.005376, "grad_norm": 0.6742497086524963, "learning_rate": 1.6295482663744796e-05, "loss": 3.2427, "step": 111000 }, { "epoch": 0.0054016, "grad_norm": 0.6161617040634155, "learning_rate": 1.6294856973060352e-05, "loss": 3.1163, "step": 111010 }, { "epoch": 0.0054272, "grad_norm": 1.041998028755188, "learning_rate": 1.6294231241555425e-05, "loss": 3.0487, "step": 111020 }, { "epoch": 0.0054528, "grad_norm": 0.8991832137107849, "learning_rate": 1.6293605469234087e-05, "loss": 3.1458, "step": 111030 }, { "epoch": 0.0054784, "grad_norm": 0.8668157458305359, "learning_rate": 1.6292979656100385e-05, "loss": 3.0704, "step": 111040 }, { "epoch": 0.005504, "grad_norm": 2.3544039726257324, "learning_rate": 1.6292353802158383e-05, "loss": 2.9814, "step": 111050 }, { "epoch": 0.0055296, "grad_norm": 0.7149984240531921, "learning_rate": 1.629172790741214e-05, "loss": 2.8765, "step": 111060 }, { "epoch": 0.0055552, "grad_norm": 1.2874431610107422, "learning_rate": 1.6291101971865715e-05, "loss": 3.3122, "step": 111070 }, { "epoch": 0.0055808, "grad_norm": 0.7279128432273865, "learning_rate": 1.6290475995523167e-05, "loss": 3.0677, "step": 111080 }, { "epoch": 0.0056064, "grad_norm": 0.7571876049041748, "learning_rate": 1.6289849978388552e-05, "loss": 3.0653, "step": 111090 }, { "epoch": 0.005632, "grad_norm": 0.7681862115859985, "learning_rate": 1.6289223920465933e-05, "loss": 2.7337, "step": 111100 }, { "epoch": 0.0056576, "grad_norm": 0.6598387360572815, "learning_rate": 1.6288597821759365e-05, "loss": 2.9511, "step": 111110 }, { "epoch": 0.0056832, "grad_norm": 0.8912038803100586, "learning_rate": 1.6287971682272914e-05, "loss": 3.0398, "step": 111120 }, { "epoch": 0.0057088, "grad_norm": 3.3191442489624023, "learning_rate": 1.6287345502010636e-05, "loss": 2.9268, "step": 111130 }, { "epoch": 0.0057344, "grad_norm": 0.6927063465118408, "learning_rate": 1.6286719280976597e-05, "loss": 3.0927, "step": 111140 }, { "epoch": 0.00576, "grad_norm": 0.7809391021728516, "learning_rate": 1.6286093019174848e-05, "loss": 3.0017, "step": 111150 }, { "epoch": 0.0057856, "grad_norm": 0.6967118978500366, "learning_rate": 1.6285466716609462e-05, "loss": 2.9783, "step": 111160 }, { "epoch": 0.0058112, "grad_norm": 0.9864261746406555, "learning_rate": 1.6284840373284496e-05, "loss": 3.0802, "step": 111170 }, { "epoch": 0.0058368, "grad_norm": 0.809986412525177, "learning_rate": 1.6284213989204007e-05, "loss": 2.9912, "step": 111180 }, { "epoch": 0.0058624, "grad_norm": 0.7287131547927856, "learning_rate": 1.628358756437206e-05, "loss": 3.1137, "step": 111190 }, { "epoch": 0.005888, "grad_norm": 0.7294580936431885, "learning_rate": 1.628296109879272e-05, "loss": 3.1031, "step": 111200 }, { "epoch": 0.0059136, "grad_norm": 0.719447672367096, "learning_rate": 1.6282334592470044e-05, "loss": 2.9147, "step": 111210 }, { "epoch": 0.0059392, "grad_norm": 0.7592419981956482, "learning_rate": 1.62817080454081e-05, "loss": 3.0868, "step": 111220 }, { "epoch": 0.0059648, "grad_norm": 0.7317597270011902, "learning_rate": 1.6281081457610946e-05, "loss": 3.0737, "step": 111230 }, { "epoch": 0.0059904, "grad_norm": 0.7128769159317017, "learning_rate": 1.6280454829082654e-05, "loss": 3.0501, "step": 111240 }, { "epoch": 0.006016, "grad_norm": 0.6858521699905396, "learning_rate": 1.6279828159827276e-05, "loss": 2.8269, "step": 111250 }, { "epoch": 0.0060416, "grad_norm": 0.8424468636512756, "learning_rate": 1.6279201449848882e-05, "loss": 2.9444, "step": 111260 }, { "epoch": 0.0060672, "grad_norm": 0.7523114085197449, "learning_rate": 1.6278574699151538e-05, "loss": 3.1212, "step": 111270 }, { "epoch": 0.0060928, "grad_norm": 0.795647919178009, "learning_rate": 1.6277947907739308e-05, "loss": 2.9774, "step": 111280 }, { "epoch": 0.0061184, "grad_norm": 0.7269482016563416, "learning_rate": 1.627732107561625e-05, "loss": 2.963, "step": 111290 }, { "epoch": 0.006144, "grad_norm": 1.403368592262268, "learning_rate": 1.6276694202786433e-05, "loss": 3.1921, "step": 111300 }, { "epoch": 0.0061696, "grad_norm": 0.7780163884162903, "learning_rate": 1.627606728925392e-05, "loss": 3.6359, "step": 111310 }, { "epoch": 0.0061952, "grad_norm": 0.752802848815918, "learning_rate": 1.6275440335022784e-05, "loss": 3.1998, "step": 111320 }, { "epoch": 0.0062208, "grad_norm": 0.7483195066452026, "learning_rate": 1.6274813340097082e-05, "loss": 3.3091, "step": 111330 }, { "epoch": 0.0062464, "grad_norm": 1.3249715566635132, "learning_rate": 1.6274186304480886e-05, "loss": 3.187, "step": 111340 }, { "epoch": 0.006272, "grad_norm": 3.345616102218628, "learning_rate": 1.6273559228178256e-05, "loss": 3.2798, "step": 111350 }, { "epoch": 0.0062976, "grad_norm": 0.6735289692878723, "learning_rate": 1.6272932111193263e-05, "loss": 3.1207, "step": 111360 }, { "epoch": 0.0063232, "grad_norm": 0.6901282072067261, "learning_rate": 1.6272304953529974e-05, "loss": 2.9786, "step": 111370 }, { "epoch": 0.0063488, "grad_norm": 0.731377124786377, "learning_rate": 1.627167775519245e-05, "loss": 2.8434, "step": 111380 }, { "epoch": 0.0063744, "grad_norm": 0.723349392414093, "learning_rate": 1.6271050516184765e-05, "loss": 2.9681, "step": 111390 }, { "epoch": 0.0064, "grad_norm": 0.7062875032424927, "learning_rate": 1.6270423236510985e-05, "loss": 3.031, "step": 111400 }, { "epoch": 0.0064256, "grad_norm": 0.9856829643249512, "learning_rate": 1.6269795916175177e-05, "loss": 3.1538, "step": 111410 }, { "epoch": 0.0064512, "grad_norm": 0.7602010369300842, "learning_rate": 1.626916855518141e-05, "loss": 3.4175, "step": 111420 }, { "epoch": 0.0064768, "grad_norm": 0.8031951189041138, "learning_rate": 1.626854115353375e-05, "loss": 3.0207, "step": 111430 }, { "epoch": 0.0065024, "grad_norm": 0.7918127775192261, "learning_rate": 1.6267913711236265e-05, "loss": 2.9207, "step": 111440 }, { "epoch": 0.006528, "grad_norm": 0.7476118803024292, "learning_rate": 1.6267286228293026e-05, "loss": 2.9387, "step": 111450 }, { "epoch": 0.0065536, "grad_norm": 0.7345893383026123, "learning_rate": 1.62666587047081e-05, "loss": 3.0641, "step": 111460 }, { "epoch": 0.0065792, "grad_norm": 0.8601425886154175, "learning_rate": 1.626603114048556e-05, "loss": 3.103, "step": 111470 }, { "epoch": 0.0066048, "grad_norm": 0.6943868398666382, "learning_rate": 1.626540353562947e-05, "loss": 2.892, "step": 111480 }, { "epoch": 0.0066304, "grad_norm": 0.7785977721214294, "learning_rate": 1.6264775890143908e-05, "loss": 3.2443, "step": 111490 }, { "epoch": 0.006656, "grad_norm": 0.7382953763008118, "learning_rate": 1.6264148204032938e-05, "loss": 3.1857, "step": 111500 }, { "epoch": 0.0066816, "grad_norm": 0.7939864993095398, "learning_rate": 1.626352047730063e-05, "loss": 3.0455, "step": 111510 }, { "epoch": 0.0067072, "grad_norm": 0.7692867517471313, "learning_rate": 1.626289270995106e-05, "loss": 3.0892, "step": 111520 }, { "epoch": 0.0067328, "grad_norm": 0.858004093170166, "learning_rate": 1.6262264901988292e-05, "loss": 3.1382, "step": 111530 }, { "epoch": 0.0067584, "grad_norm": 0.7764757871627808, "learning_rate": 1.62616370534164e-05, "loss": 3.166, "step": 111540 }, { "epoch": 0.006784, "grad_norm": 0.7277525663375854, "learning_rate": 1.6261009164239455e-05, "loss": 3.0608, "step": 111550 }, { "epoch": 0.0068096, "grad_norm": 0.6572389006614685, "learning_rate": 1.6260381234461534e-05, "loss": 2.9946, "step": 111560 }, { "epoch": 0.0068352, "grad_norm": 0.6979449987411499, "learning_rate": 1.62597532640867e-05, "loss": 3.1324, "step": 111570 }, { "epoch": 0.0068608, "grad_norm": 0.9025924801826477, "learning_rate": 1.6259125253119032e-05, "loss": 3.0425, "step": 111580 }, { "epoch": 0.0068864, "grad_norm": 0.7503573298454285, "learning_rate": 1.6258497201562598e-05, "loss": 2.8611, "step": 111590 }, { "epoch": 0.006912, "grad_norm": 0.6643836498260498, "learning_rate": 1.625786910942148e-05, "loss": 2.9965, "step": 111600 }, { "epoch": 0.0069376, "grad_norm": 0.6999651789665222, "learning_rate": 1.6257240976699736e-05, "loss": 3.0678, "step": 111610 }, { "epoch": 0.0069632, "grad_norm": 0.7310221195220947, "learning_rate": 1.625661280340145e-05, "loss": 3.3221, "step": 111620 }, { "epoch": 0.0069888, "grad_norm": 0.7009432315826416, "learning_rate": 1.625598458953069e-05, "loss": 3.1347, "step": 111630 }, { "epoch": 0.0070144, "grad_norm": 0.7049671411514282, "learning_rate": 1.625535633509154e-05, "loss": 3.0866, "step": 111640 }, { "epoch": 0.00704, "grad_norm": 0.6739674806594849, "learning_rate": 1.625472804008806e-05, "loss": 3.1577, "step": 111650 }, { "epoch": 0.0070656, "grad_norm": 0.7440071105957031, "learning_rate": 1.625409970452433e-05, "loss": 3.0529, "step": 111660 }, { "epoch": 0.0070912, "grad_norm": 0.9626307487487793, "learning_rate": 1.625347132840443e-05, "loss": 2.9366, "step": 111670 }, { "epoch": 0.0071168, "grad_norm": 0.8341855406761169, "learning_rate": 1.625284291173243e-05, "loss": 3.1256, "step": 111680 }, { "epoch": 0.0071424, "grad_norm": 0.7251095771789551, "learning_rate": 1.6252214454512406e-05, "loss": 2.9496, "step": 111690 }, { "epoch": 0.007168, "grad_norm": 0.7838795185089111, "learning_rate": 1.625158595674843e-05, "loss": 3.3505, "step": 111700 }, { "epoch": 0.0071936, "grad_norm": 0.8392860293388367, "learning_rate": 1.6250957418444582e-05, "loss": 3.1937, "step": 111710 }, { "epoch": 0.0072192, "grad_norm": 1.1271684169769287, "learning_rate": 1.625032883960494e-05, "loss": 3.0846, "step": 111720 }, { "epoch": 0.0072448, "grad_norm": 0.9455386400222778, "learning_rate": 1.624970022023357e-05, "loss": 3.1856, "step": 111730 }, { "epoch": 0.0072704, "grad_norm": 0.6645054817199707, "learning_rate": 1.6249071560334555e-05, "loss": 3.2314, "step": 111740 }, { "epoch": 0.007296, "grad_norm": 0.7139641046524048, "learning_rate": 1.6248442859911978e-05, "loss": 3.058, "step": 111750 }, { "epoch": 0.0073216, "grad_norm": 0.8319581747055054, "learning_rate": 1.6247814118969903e-05, "loss": 3.2622, "step": 111760 }, { "epoch": 0.0073472, "grad_norm": 0.8435764908790588, "learning_rate": 1.6247185337512422e-05, "loss": 3.0858, "step": 111770 }, { "epoch": 0.0073728, "grad_norm": 0.9261461496353149, "learning_rate": 1.6246556515543596e-05, "loss": 3.0719, "step": 111780 }, { "epoch": 0.0073984, "grad_norm": 0.845880925655365, "learning_rate": 1.6245927653067515e-05, "loss": 3.272, "step": 111790 }, { "epoch": 0.007424, "grad_norm": 0.6597696542739868, "learning_rate": 1.6245298750088254e-05, "loss": 2.9321, "step": 111800 }, { "epoch": 0.0074496, "grad_norm": 0.7817533016204834, "learning_rate": 1.6244669806609888e-05, "loss": 3.0299, "step": 111810 }, { "epoch": 0.0074752, "grad_norm": 0.7393878102302551, "learning_rate": 1.6244040822636503e-05, "loss": 3.2486, "step": 111820 }, { "epoch": 0.0075008, "grad_norm": 0.6902705430984497, "learning_rate": 1.6243411798172165e-05, "loss": 3.1764, "step": 111830 }, { "epoch": 0.0075264, "grad_norm": 0.7265027165412903, "learning_rate": 1.6242782733220966e-05, "loss": 3.1341, "step": 111840 }, { "epoch": 0.007552, "grad_norm": 0.8595990538597107, "learning_rate": 1.624215362778698e-05, "loss": 3.1059, "step": 111850 }, { "epoch": 0.0075776, "grad_norm": 0.6901662945747375, "learning_rate": 1.624152448187429e-05, "loss": 3.0608, "step": 111860 }, { "epoch": 0.0076032, "grad_norm": 0.677302896976471, "learning_rate": 1.6240895295486965e-05, "loss": 3.2359, "step": 111870 }, { "epoch": 0.0076288, "grad_norm": 0.677112340927124, "learning_rate": 1.6240266068629094e-05, "loss": 3.188, "step": 111880 }, { "epoch": 0.0076544, "grad_norm": 0.7801728248596191, "learning_rate": 1.623963680130476e-05, "loss": 2.9584, "step": 111890 }, { "epoch": 0.00768, "grad_norm": 0.641732931137085, "learning_rate": 1.623900749351804e-05, "loss": 3.0155, "step": 111900 }, { "epoch": 0.0077056, "grad_norm": 0.6362382769584656, "learning_rate": 1.6238378145273008e-05, "loss": 2.8657, "step": 111910 }, { "epoch": 0.0077312, "grad_norm": 0.6903960704803467, "learning_rate": 1.6237748756573755e-05, "loss": 3.125, "step": 111920 }, { "epoch": 0.0077568, "grad_norm": 0.7248846292495728, "learning_rate": 1.623711932742436e-05, "loss": 3.108, "step": 111930 }, { "epoch": 0.0077824, "grad_norm": 0.7400584816932678, "learning_rate": 1.6236489857828903e-05, "loss": 3.0656, "step": 111940 }, { "epoch": 0.007808, "grad_norm": 0.8611815571784973, "learning_rate": 1.623586034779147e-05, "loss": 3.4039, "step": 111950 }, { "epoch": 0.0078336, "grad_norm": 0.7357204556465149, "learning_rate": 1.6235230797316134e-05, "loss": 3.0622, "step": 111960 }, { "epoch": 0.0078592, "grad_norm": 0.772526204586029, "learning_rate": 1.623460120640699e-05, "loss": 3.0217, "step": 111970 }, { "epoch": 0.0078848, "grad_norm": 0.6642471551895142, "learning_rate": 1.623397157506811e-05, "loss": 3.0924, "step": 111980 }, { "epoch": 0.0079104, "grad_norm": 0.6894156336784363, "learning_rate": 1.623334190330358e-05, "loss": 2.9485, "step": 111990 }, { "epoch": 0.007936, "grad_norm": 0.675730288028717, "learning_rate": 1.6232712191117488e-05, "loss": 3.1316, "step": 112000 }, { "epoch": 0.0079616, "grad_norm": 0.7304441332817078, "learning_rate": 1.623208243851391e-05, "loss": 3.1011, "step": 112010 }, { "epoch": 0.0079872, "grad_norm": 0.8035944104194641, "learning_rate": 1.623145264549694e-05, "loss": 3.2263, "step": 112020 }, { "epoch": 0.0080128, "grad_norm": 0.7081320881843567, "learning_rate": 1.623082281207065e-05, "loss": 2.9888, "step": 112030 }, { "epoch": 0.0080384, "grad_norm": 0.7015580534934998, "learning_rate": 1.623019293823913e-05, "loss": 2.9134, "step": 112040 }, { "epoch": 0.008064, "grad_norm": 0.7952503561973572, "learning_rate": 1.6229563024006466e-05, "loss": 3.1384, "step": 112050 }, { "epoch": 0.0080896, "grad_norm": 0.7251965403556824, "learning_rate": 1.6228933069376745e-05, "loss": 3.0336, "step": 112060 }, { "epoch": 0.0081152, "grad_norm": 0.9047871232032776, "learning_rate": 1.6228303074354043e-05, "loss": 3.1216, "step": 112070 }, { "epoch": 0.0081408, "grad_norm": 0.6844422221183777, "learning_rate": 1.6227673038942455e-05, "loss": 3.3393, "step": 112080 }, { "epoch": 0.0081664, "grad_norm": 0.6939225196838379, "learning_rate": 1.6227042963146057e-05, "loss": 2.8394, "step": 112090 }, { "epoch": 0.008192, "grad_norm": 0.7718369960784912, "learning_rate": 1.6226412846968943e-05, "loss": 3.1207, "step": 112100 }, { "epoch": 0.0082176, "grad_norm": 0.7799562811851501, "learning_rate": 1.62257826904152e-05, "loss": 3.1572, "step": 112110 }, { "epoch": 0.0082432, "grad_norm": 0.6947033405303955, "learning_rate": 1.622515249348891e-05, "loss": 3.0719, "step": 112120 }, { "epoch": 0.0082688, "grad_norm": 0.767820417881012, "learning_rate": 1.6224522256194158e-05, "loss": 3.2082, "step": 112130 }, { "epoch": 0.0082944, "grad_norm": 0.9446948170661926, "learning_rate": 1.6223891978535034e-05, "loss": 3.2469, "step": 112140 }, { "epoch": 0.00832, "grad_norm": 0.7296448945999146, "learning_rate": 1.6223261660515626e-05, "loss": 3.0491, "step": 112150 }, { "epoch": 0.0083456, "grad_norm": 0.6890026926994324, "learning_rate": 1.6222631302140018e-05, "loss": 3.0175, "step": 112160 }, { "epoch": 0.0083712, "grad_norm": 0.7152645587921143, "learning_rate": 1.6222000903412302e-05, "loss": 3.2207, "step": 112170 }, { "epoch": 0.0083968, "grad_norm": 0.7209406495094299, "learning_rate": 1.6221370464336558e-05, "loss": 3.1518, "step": 112180 }, { "epoch": 0.0084224, "grad_norm": 1.0018333196640015, "learning_rate": 1.6220739984916886e-05, "loss": 2.9793, "step": 112190 }, { "epoch": 0.008448, "grad_norm": 0.715107798576355, "learning_rate": 1.6220109465157364e-05, "loss": 3.138, "step": 112200 }, { "epoch": 0.0084736, "grad_norm": 0.7562996745109558, "learning_rate": 1.621947890506209e-05, "loss": 3.0865, "step": 112210 }, { "epoch": 0.0084992, "grad_norm": 0.9032649397850037, "learning_rate": 1.6218848304635148e-05, "loss": 3.3054, "step": 112220 }, { "epoch": 0.0085248, "grad_norm": 0.7166388034820557, "learning_rate": 1.6218217663880626e-05, "loss": 3.2043, "step": 112230 }, { "epoch": 0.0085504, "grad_norm": 0.7374144792556763, "learning_rate": 1.6217586982802612e-05, "loss": 3.0943, "step": 112240 }, { "epoch": 0.008576, "grad_norm": 0.912166953086853, "learning_rate": 1.6216956261405203e-05, "loss": 3.151, "step": 112250 }, { "epoch": 0.0086016, "grad_norm": 0.6459321975708008, "learning_rate": 1.6216325499692485e-05, "loss": 3.0858, "step": 112260 }, { "epoch": 0.0086272, "grad_norm": 0.7474772334098816, "learning_rate": 1.6215694697668545e-05, "loss": 3.1198, "step": 112270 }, { "epoch": 0.0086528, "grad_norm": 0.9508581757545471, "learning_rate": 1.6215063855337477e-05, "loss": 2.9536, "step": 112280 }, { "epoch": 0.0086784, "grad_norm": 0.810150146484375, "learning_rate": 1.6214432972703376e-05, "loss": 2.867, "step": 112290 }, { "epoch": 0.008704, "grad_norm": 0.7202020883560181, "learning_rate": 1.6213802049770327e-05, "loss": 3.012, "step": 112300 }, { "epoch": 0.0087296, "grad_norm": 0.747871458530426, "learning_rate": 1.621317108654242e-05, "loss": 3.1786, "step": 112310 }, { "epoch": 0.0087552, "grad_norm": 0.7409653663635254, "learning_rate": 1.6212540083023754e-05, "loss": 2.9068, "step": 112320 }, { "epoch": 0.0087808, "grad_norm": 0.7284985780715942, "learning_rate": 1.6211909039218413e-05, "loss": 3.0926, "step": 112330 }, { "epoch": 0.0088064, "grad_norm": 0.9961779713630676, "learning_rate": 1.6211277955130495e-05, "loss": 2.955, "step": 112340 }, { "epoch": 0.008832, "grad_norm": 0.7834521532058716, "learning_rate": 1.621064683076409e-05, "loss": 3.0582, "step": 112350 }, { "epoch": 0.0088576, "grad_norm": 0.8424550890922546, "learning_rate": 1.6210015666123288e-05, "loss": 3.0903, "step": 112360 }, { "epoch": 0.0088832, "grad_norm": 0.6986085772514343, "learning_rate": 1.6209384461212188e-05, "loss": 3.0595, "step": 112370 }, { "epoch": 0.0089088, "grad_norm": 0.8322094082832336, "learning_rate": 1.620875321603488e-05, "loss": 3.0847, "step": 112380 }, { "epoch": 0.0089344, "grad_norm": 0.7150031328201294, "learning_rate": 1.6208121930595457e-05, "loss": 3.0687, "step": 112390 }, { "epoch": 0.00896, "grad_norm": 0.7058475017547607, "learning_rate": 1.6207490604898015e-05, "loss": 3.2672, "step": 112400 }, { "epoch": 0.0089856, "grad_norm": 0.6713724136352539, "learning_rate": 1.6206859238946643e-05, "loss": 3.0995, "step": 112410 }, { "epoch": 0.0090112, "grad_norm": 0.8005036115646362, "learning_rate": 1.620622783274544e-05, "loss": 3.1968, "step": 112420 }, { "epoch": 0.0090368, "grad_norm": 0.7799215912818909, "learning_rate": 1.6205596386298502e-05, "loss": 3.0525, "step": 112430 }, { "epoch": 0.0090624, "grad_norm": 0.8456237316131592, "learning_rate": 1.6204964899609915e-05, "loss": 3.2619, "step": 112440 }, { "epoch": 0.009088, "grad_norm": 0.6478429436683655, "learning_rate": 1.620433337268378e-05, "loss": 2.9561, "step": 112450 }, { "epoch": 0.0091136, "grad_norm": 0.828481912612915, "learning_rate": 1.6203701805524194e-05, "loss": 3.2131, "step": 112460 }, { "epoch": 0.0091392, "grad_norm": 0.9380221962928772, "learning_rate": 1.6203070198135252e-05, "loss": 3.1136, "step": 112470 }, { "epoch": 0.0091648, "grad_norm": 0.7456597089767456, "learning_rate": 1.6202438550521048e-05, "loss": 3.0629, "step": 112480 }, { "epoch": 0.0091904, "grad_norm": 0.7188569903373718, "learning_rate": 1.6201806862685677e-05, "loss": 3.0695, "step": 112490 }, { "epoch": 0.009216, "grad_norm": 0.7535415887832642, "learning_rate": 1.6201175134633236e-05, "loss": 3.1256, "step": 112500 }, { "epoch": 0.0092416, "grad_norm": 1.0457514524459839, "learning_rate": 1.620054336636782e-05, "loss": 2.9299, "step": 112510 }, { "epoch": 0.0092672, "grad_norm": 0.9862115979194641, "learning_rate": 1.619991155789353e-05, "loss": 3.1511, "step": 112520 }, { "epoch": 0.0092928, "grad_norm": 0.9939778447151184, "learning_rate": 1.6199279709214465e-05, "loss": 3.1024, "step": 112530 }, { "epoch": 0.0093184, "grad_norm": 0.6822584867477417, "learning_rate": 1.619864782033472e-05, "loss": 3.0513, "step": 112540 }, { "epoch": 0.009344, "grad_norm": 0.82772296667099, "learning_rate": 1.6198015891258388e-05, "loss": 3.1386, "step": 112550 }, { "epoch": 0.0093696, "grad_norm": 0.8996903896331787, "learning_rate": 1.6197383921989568e-05, "loss": 2.9911, "step": 112560 }, { "epoch": 0.0093952, "grad_norm": 0.713539183139801, "learning_rate": 1.619675191253236e-05, "loss": 3.1128, "step": 112570 }, { "epoch": 0.0094208, "grad_norm": 0.679800808429718, "learning_rate": 1.619611986289087e-05, "loss": 3.1275, "step": 112580 }, { "epoch": 0.0094464, "grad_norm": 0.9570578336715698, "learning_rate": 1.6195487773069182e-05, "loss": 2.9209, "step": 112590 }, { "epoch": 0.009472, "grad_norm": 0.8542211055755615, "learning_rate": 1.6194855643071405e-05, "loss": 3.3753, "step": 112600 }, { "epoch": 0.0094976, "grad_norm": 0.696427583694458, "learning_rate": 1.6194223472901636e-05, "loss": 3.1607, "step": 112610 }, { "epoch": 0.0095232, "grad_norm": 0.8034717440605164, "learning_rate": 1.6193591262563973e-05, "loss": 3.1947, "step": 112620 }, { "epoch": 0.0095488, "grad_norm": 0.8359224796295166, "learning_rate": 1.619295901206252e-05, "loss": 3.0711, "step": 112630 }, { "epoch": 0.0095744, "grad_norm": 0.8632981777191162, "learning_rate": 1.6192326721401368e-05, "loss": 3.0373, "step": 112640 }, { "epoch": 0.0096, "grad_norm": 0.771894097328186, "learning_rate": 1.6191694390584625e-05, "loss": 2.99, "step": 112650 }, { "epoch": 0.0096256, "grad_norm": 0.7697548866271973, "learning_rate": 1.619106201961639e-05, "loss": 2.8558, "step": 112660 }, { "epoch": 0.0096512, "grad_norm": 0.7103152871131897, "learning_rate": 1.6190429608500764e-05, "loss": 2.839, "step": 112670 }, { "epoch": 0.0096768, "grad_norm": 0.6941413879394531, "learning_rate": 1.6189797157241845e-05, "loss": 3.1849, "step": 112680 }, { "epoch": 0.0097024, "grad_norm": 0.7749711871147156, "learning_rate": 1.618916466584374e-05, "loss": 3.1529, "step": 112690 }, { "epoch": 0.009728, "grad_norm": 0.8400499820709229, "learning_rate": 1.6188532134310545e-05, "loss": 3.1489, "step": 112700 }, { "epoch": 0.0097536, "grad_norm": 0.7524868845939636, "learning_rate": 1.6187899562646364e-05, "loss": 3.1207, "step": 112710 }, { "epoch": 0.0097792, "grad_norm": 0.7584912180900574, "learning_rate": 1.6187266950855297e-05, "loss": 3.1377, "step": 112720 }, { "epoch": 0.0098048, "grad_norm": 0.7215922474861145, "learning_rate": 1.618663429894145e-05, "loss": 2.9955, "step": 112730 }, { "epoch": 0.0098304, "grad_norm": 0.9522319436073303, "learning_rate": 1.6186001606908925e-05, "loss": 2.9354, "step": 112740 }, { "epoch": 0.009856, "grad_norm": 1.0351277589797974, "learning_rate": 1.6185368874761823e-05, "loss": 3.1417, "step": 112750 }, { "epoch": 0.0098816, "grad_norm": 0.714982807636261, "learning_rate": 1.6184736102504247e-05, "loss": 3.1577, "step": 112760 }, { "epoch": 0.0099072, "grad_norm": 0.8558354377746582, "learning_rate": 1.6184103290140302e-05, "loss": 3.1888, "step": 112770 }, { "epoch": 0.0099328, "grad_norm": 0.9245544075965881, "learning_rate": 1.618347043767409e-05, "loss": 3.1547, "step": 112780 }, { "epoch": 0.0099584, "grad_norm": 0.9080178737640381, "learning_rate": 1.618283754510971e-05, "loss": 3.0909, "step": 112790 }, { "epoch": 0.009984, "grad_norm": 0.7387744784355164, "learning_rate": 1.618220461245128e-05, "loss": 3.1026, "step": 112800 }, { "epoch": 0.0100096, "grad_norm": 1.0498777627944946, "learning_rate": 1.6181571639702895e-05, "loss": 3.2026, "step": 112810 }, { "epoch": 0.0100352, "grad_norm": 0.7836905121803284, "learning_rate": 1.6180938626868657e-05, "loss": 3.1211, "step": 112820 }, { "epoch": 0.0100608, "grad_norm": 0.6714757680892944, "learning_rate": 1.6180305573952678e-05, "loss": 3.0686, "step": 112830 }, { "epoch": 0.0100864, "grad_norm": 0.7229371666908264, "learning_rate": 1.617967248095906e-05, "loss": 3.2684, "step": 112840 }, { "epoch": 0.010112, "grad_norm": 1.2354108095169067, "learning_rate": 1.6179039347891907e-05, "loss": 3.091, "step": 112850 }, { "epoch": 0.0101376, "grad_norm": 0.6801518797874451, "learning_rate": 1.6178406174755325e-05, "loss": 3.2111, "step": 112860 }, { "epoch": 0.0101632, "grad_norm": 0.7291268706321716, "learning_rate": 1.617777296155342e-05, "loss": 3.1072, "step": 112870 }, { "epoch": 0.0101888, "grad_norm": 0.7562692761421204, "learning_rate": 1.6177139708290304e-05, "loss": 2.9489, "step": 112880 }, { "epoch": 0.0102144, "grad_norm": 1.0323498249053955, "learning_rate": 1.6176506414970074e-05, "loss": 2.7763, "step": 112890 }, { "epoch": 0.01024, "grad_norm": 0.779506504535675, "learning_rate": 1.6175873081596844e-05, "loss": 2.9026, "step": 112900 }, { "epoch": 0.0102656, "grad_norm": 0.743442952632904, "learning_rate": 1.6175239708174717e-05, "loss": 3.0075, "step": 112910 }, { "epoch": 0.0102912, "grad_norm": 1.0191514492034912, "learning_rate": 1.6174606294707803e-05, "loss": 3.151, "step": 112920 }, { "epoch": 0.0103168, "grad_norm": 0.7835716605186462, "learning_rate": 1.617397284120021e-05, "loss": 3.1682, "step": 112930 }, { "epoch": 0.0103424, "grad_norm": 0.8151763081550598, "learning_rate": 1.617333934765604e-05, "loss": 3.2314, "step": 112940 }, { "epoch": 0.010368, "grad_norm": 0.7371060848236084, "learning_rate": 1.6172705814079407e-05, "loss": 3.1082, "step": 112950 }, { "epoch": 0.0103936, "grad_norm": 0.6805897951126099, "learning_rate": 1.6172072240474416e-05, "loss": 3.1324, "step": 112960 }, { "epoch": 0.0104192, "grad_norm": 1.4639428853988647, "learning_rate": 1.6171438626845177e-05, "loss": 3.3833, "step": 112970 }, { "epoch": 0.0104448, "grad_norm": 0.6724688410758972, "learning_rate": 1.61708049731958e-05, "loss": 3.2623, "step": 112980 }, { "epoch": 0.0104704, "grad_norm": 0.7747681736946106, "learning_rate": 1.617017127953039e-05, "loss": 3.1043, "step": 112990 }, { "epoch": 0.010496, "grad_norm": 0.8124310970306396, "learning_rate": 1.6169537545853066e-05, "loss": 3.1405, "step": 113000 }, { "epoch": 0.0105216, "grad_norm": 0.8678972721099854, "learning_rate": 1.6168903772167922e-05, "loss": 3.1318, "step": 113010 }, { "epoch": 0.0105472, "grad_norm": 0.7980047464370728, "learning_rate": 1.6168269958479082e-05, "loss": 3.3358, "step": 113020 }, { "epoch": 0.0105728, "grad_norm": 0.7921180725097656, "learning_rate": 1.616763610479065e-05, "loss": 2.9671, "step": 113030 }, { "epoch": 0.0105984, "grad_norm": 0.8646658658981323, "learning_rate": 1.6167002211106738e-05, "loss": 3.1145, "step": 113040 }, { "epoch": 0.010624, "grad_norm": 0.7615454196929932, "learning_rate": 1.616636827743145e-05, "loss": 3.0521, "step": 113050 }, { "epoch": 0.0106496, "grad_norm": 0.665052056312561, "learning_rate": 1.6165734303768905e-05, "loss": 3.1071, "step": 113060 }, { "epoch": 0.0106752, "grad_norm": 1.8516868352890015, "learning_rate": 1.6165100290123213e-05, "loss": 2.9297, "step": 113070 }, { "epoch": 0.0107008, "grad_norm": 0.6757867336273193, "learning_rate": 1.6164466236498485e-05, "loss": 2.9293, "step": 113080 }, { "epoch": 0.0107264, "grad_norm": 0.8185513019561768, "learning_rate": 1.6163832142898832e-05, "loss": 3.1108, "step": 113090 }, { "epoch": 0.010752, "grad_norm": 0.7081509232521057, "learning_rate": 1.6163198009328366e-05, "loss": 3.1326, "step": 113100 }, { "epoch": 0.0107776, "grad_norm": 0.748053789138794, "learning_rate": 1.6162563835791196e-05, "loss": 3.2389, "step": 113110 }, { "epoch": 0.0108032, "grad_norm": 0.8025566339492798, "learning_rate": 1.6161929622291438e-05, "loss": 3.0354, "step": 113120 }, { "epoch": 0.0108288, "grad_norm": 0.7983729839324951, "learning_rate": 1.6161295368833206e-05, "loss": 3.192, "step": 113130 }, { "epoch": 0.0108544, "grad_norm": 0.714061439037323, "learning_rate": 1.616066107542061e-05, "loss": 3.0496, "step": 113140 }, { "epoch": 0.01088, "grad_norm": 0.7257843017578125, "learning_rate": 1.6160026742057764e-05, "loss": 3.1338, "step": 113150 }, { "epoch": 0.0109056, "grad_norm": 0.7837279438972473, "learning_rate": 1.615939236874878e-05, "loss": 3.1294, "step": 113160 }, { "epoch": 0.0109312, "grad_norm": 0.8039405345916748, "learning_rate": 1.6158757955497777e-05, "loss": 3.2212, "step": 113170 }, { "epoch": 0.0109568, "grad_norm": 0.8016884326934814, "learning_rate": 1.615812350230886e-05, "loss": 3.142, "step": 113180 }, { "epoch": 0.0109824, "grad_norm": 0.7318016290664673, "learning_rate": 1.615748900918616e-05, "loss": 3.0228, "step": 113190 }, { "epoch": 0.011008, "grad_norm": 0.8254014849662781, "learning_rate": 1.615685447613377e-05, "loss": 3.1728, "step": 113200 }, { "epoch": 0.0110336, "grad_norm": 0.7068798542022705, "learning_rate": 1.6156219903155817e-05, "loss": 2.9864, "step": 113210 }, { "epoch": 0.0110592, "grad_norm": 0.881252110004425, "learning_rate": 1.6155585290256413e-05, "loss": 2.9643, "step": 113220 }, { "epoch": 0.0110848, "grad_norm": 0.7441108822822571, "learning_rate": 1.6154950637439676e-05, "loss": 3.0809, "step": 113230 }, { "epoch": 0.0111104, "grad_norm": 0.6916786432266235, "learning_rate": 1.615431594470972e-05, "loss": 3.1528, "step": 113240 }, { "epoch": 0.011136, "grad_norm": 0.6945155262947083, "learning_rate": 1.6153681212070662e-05, "loss": 3.1136, "step": 113250 }, { "epoch": 0.0111616, "grad_norm": 0.96392422914505, "learning_rate": 1.6153046439526615e-05, "loss": 3.1155, "step": 113260 }, { "epoch": 0.0111872, "grad_norm": 0.7744250893592834, "learning_rate": 1.6152411627081698e-05, "loss": 3.3171, "step": 113270 }, { "epoch": 0.0112128, "grad_norm": 0.682978630065918, "learning_rate": 1.6151776774740028e-05, "loss": 2.8738, "step": 113280 }, { "epoch": 0.0112384, "grad_norm": 0.8403294086456299, "learning_rate": 1.615114188250572e-05, "loss": 3.1297, "step": 113290 }, { "epoch": 0.011264, "grad_norm": 0.7227822542190552, "learning_rate": 1.615050695038289e-05, "loss": 3.1002, "step": 113300 }, { "epoch": 0.0112896, "grad_norm": 0.8630802631378174, "learning_rate": 1.6149871978375658e-05, "loss": 3.4261, "step": 113310 }, { "epoch": 0.0113152, "grad_norm": 0.789590060710907, "learning_rate": 1.6149236966488137e-05, "loss": 3.0977, "step": 113320 }, { "epoch": 0.0113408, "grad_norm": 1.3111943006515503, "learning_rate": 1.6148601914724452e-05, "loss": 3.4635, "step": 113330 }, { "epoch": 0.0113664, "grad_norm": 0.7958963513374329, "learning_rate": 1.6147966823088717e-05, "loss": 3.1086, "step": 113340 }, { "epoch": 0.011392, "grad_norm": 0.7152823209762573, "learning_rate": 1.6147331691585053e-05, "loss": 3.083, "step": 113350 }, { "epoch": 0.0114176, "grad_norm": 1.4462634325027466, "learning_rate": 1.6146696520217578e-05, "loss": 3.2566, "step": 113360 }, { "epoch": 0.0114432, "grad_norm": 0.811548113822937, "learning_rate": 1.6146061308990404e-05, "loss": 3.2188, "step": 113370 }, { "epoch": 0.0114688, "grad_norm": 0.9485940933227539, "learning_rate": 1.614542605790766e-05, "loss": 3.1554, "step": 113380 }, { "epoch": 0.0114944, "grad_norm": 0.708651065826416, "learning_rate": 1.614479076697346e-05, "loss": 2.9318, "step": 113390 }, { "epoch": 0.01152, "grad_norm": 0.7506431937217712, "learning_rate": 1.6144155436191926e-05, "loss": 3.1386, "step": 113400 }, { "epoch": 0.0115456, "grad_norm": 1.585780382156372, "learning_rate": 1.6143520065567174e-05, "loss": 3.2756, "step": 113410 }, { "epoch": 0.0115712, "grad_norm": 0.7947617173194885, "learning_rate": 1.614288465510333e-05, "loss": 3.0926, "step": 113420 }, { "epoch": 0.0115968, "grad_norm": 0.7923992276191711, "learning_rate": 1.6142249204804513e-05, "loss": 3.1547, "step": 113430 }, { "epoch": 0.0116224, "grad_norm": 0.9723368287086487, "learning_rate": 1.6141613714674837e-05, "loss": 3.0337, "step": 113440 }, { "epoch": 0.011648, "grad_norm": 0.7302764654159546, "learning_rate": 1.6140978184718434e-05, "loss": 2.9109, "step": 113450 }, { "epoch": 0.0116736, "grad_norm": 0.7248371839523315, "learning_rate": 1.6140342614939415e-05, "loss": 3.2036, "step": 113460 }, { "epoch": 0.0116992, "grad_norm": 0.8761436343193054, "learning_rate": 1.6139707005341908e-05, "loss": 2.9839, "step": 113470 }, { "epoch": 0.0117248, "grad_norm": 0.743474543094635, "learning_rate": 1.6139071355930034e-05, "loss": 3.0574, "step": 113480 }, { "epoch": 0.0117504, "grad_norm": 1.0588809251785278, "learning_rate": 1.6138435666707914e-05, "loss": 3.0895, "step": 113490 }, { "epoch": 0.011776, "grad_norm": 0.9341185688972473, "learning_rate": 1.6137799937679666e-05, "loss": 3.1532, "step": 113500 }, { "epoch": 0.0118016, "grad_norm": 1.286488652229309, "learning_rate": 1.6137164168849418e-05, "loss": 3.3725, "step": 113510 }, { "epoch": 0.0118272, "grad_norm": 0.8138867020606995, "learning_rate": 1.6136528360221297e-05, "loss": 3.0862, "step": 113520 }, { "epoch": 0.0118528, "grad_norm": 0.7401095628738403, "learning_rate": 1.6135892511799415e-05, "loss": 3.0668, "step": 113530 }, { "epoch": 0.0118784, "grad_norm": 0.725258469581604, "learning_rate": 1.6135256623587903e-05, "loss": 3.1369, "step": 113540 }, { "epoch": 0.011904, "grad_norm": 0.7258860468864441, "learning_rate": 1.613462069559088e-05, "loss": 3.185, "step": 113550 }, { "epoch": 0.0119296, "grad_norm": 0.8221699595451355, "learning_rate": 1.613398472781247e-05, "loss": 3.1294, "step": 113560 }, { "epoch": 0.0119552, "grad_norm": 0.8887892365455627, "learning_rate": 1.6133348720256806e-05, "loss": 3.4155, "step": 113570 }, { "epoch": 0.0119808, "grad_norm": 0.9239745140075684, "learning_rate": 1.6132712672928e-05, "loss": 2.7803, "step": 113580 }, { "epoch": 0.0120064, "grad_norm": 0.6817312240600586, "learning_rate": 1.6132076585830187e-05, "loss": 3.0765, "step": 113590 }, { "epoch": 0.012032, "grad_norm": 0.7876426577568054, "learning_rate": 1.6131440458967485e-05, "loss": 3.1581, "step": 113600 }, { "epoch": 0.0120576, "grad_norm": 0.8045613169670105, "learning_rate": 1.613080429234402e-05, "loss": 3.1305, "step": 113610 }, { "epoch": 0.0120832, "grad_norm": 0.7251316905021667, "learning_rate": 1.613016808596392e-05, "loss": 3.1507, "step": 113620 }, { "epoch": 0.0121088, "grad_norm": 0.7199756503105164, "learning_rate": 1.612953183983131e-05, "loss": 3.1789, "step": 113630 }, { "epoch": 0.0121344, "grad_norm": 0.7462698221206665, "learning_rate": 1.6128895553950312e-05, "loss": 3.1435, "step": 113640 }, { "epoch": 0.01216, "grad_norm": 0.760113537311554, "learning_rate": 1.6128259228325057e-05, "loss": 2.9596, "step": 113650 }, { "epoch": 0.0121856, "grad_norm": 0.7787204384803772, "learning_rate": 1.612762286295967e-05, "loss": 3.2164, "step": 113660 }, { "epoch": 0.0122112, "grad_norm": 0.7747787833213806, "learning_rate": 1.6126986457858274e-05, "loss": 3.0464, "step": 113670 }, { "epoch": 0.0122368, "grad_norm": 0.745343804359436, "learning_rate": 1.6126350013025004e-05, "loss": 2.9943, "step": 113680 }, { "epoch": 0.0122624, "grad_norm": 0.6367192268371582, "learning_rate": 1.612571352846398e-05, "loss": 3.101, "step": 113690 }, { "epoch": 0.012288, "grad_norm": 1.8138659000396729, "learning_rate": 1.612507700417933e-05, "loss": 3.1867, "step": 113700 }, { "epoch": 0.0123136, "grad_norm": 0.8813657164573669, "learning_rate": 1.6124440440175186e-05, "loss": 3.2323, "step": 113710 }, { "epoch": 0.0123392, "grad_norm": 2.5475809574127197, "learning_rate": 1.6123803836455675e-05, "loss": 3.1054, "step": 113720 }, { "epoch": 0.0123648, "grad_norm": 0.665407657623291, "learning_rate": 1.6123167193024922e-05, "loss": 2.9341, "step": 113730 }, { "epoch": 0.0123904, "grad_norm": 0.7164245247840881, "learning_rate": 1.6122530509887053e-05, "loss": 3.4994, "step": 113740 }, { "epoch": 0.012416, "grad_norm": 0.9470813870429993, "learning_rate": 1.6121893787046206e-05, "loss": 3.176, "step": 113750 }, { "epoch": 0.0124416, "grad_norm": 0.6746749877929688, "learning_rate": 1.6121257024506506e-05, "loss": 3.0436, "step": 113760 }, { "epoch": 0.0124672, "grad_norm": 0.8855879306793213, "learning_rate": 1.6120620222272075e-05, "loss": 3.2251, "step": 113770 }, { "epoch": 0.0124928, "grad_norm": 1.1859506368637085, "learning_rate": 1.611998338034705e-05, "loss": 3.2675, "step": 113780 }, { "epoch": 0.0125184, "grad_norm": 0.7586250901222229, "learning_rate": 1.6119346498735563e-05, "loss": 3.0859, "step": 113790 }, { "epoch": 0.012544, "grad_norm": 0.6897703409194946, "learning_rate": 1.611870957744174e-05, "loss": 3.0804, "step": 113800 }, { "epoch": 0.0125696, "grad_norm": 0.6479098796844482, "learning_rate": 1.6118072616469712e-05, "loss": 3.0738, "step": 113810 }, { "epoch": 0.0125952, "grad_norm": 0.7091588973999023, "learning_rate": 1.6117435615823603e-05, "loss": 3.3234, "step": 113820 }, { "epoch": 0.0126208, "grad_norm": 0.6578148007392883, "learning_rate": 1.6116798575507557e-05, "loss": 2.9721, "step": 113830 }, { "epoch": 0.0126464, "grad_norm": 0.7146003842353821, "learning_rate": 1.6116161495525692e-05, "loss": 3.1478, "step": 113840 }, { "epoch": 0.012672, "grad_norm": 0.80430668592453, "learning_rate": 1.6115524375882147e-05, "loss": 3.0807, "step": 113850 }, { "epoch": 0.0126976, "grad_norm": 1.138719081878662, "learning_rate": 1.6114887216581054e-05, "loss": 3.0053, "step": 113860 }, { "epoch": 0.0127232, "grad_norm": 0.6778662800788879, "learning_rate": 1.611425001762654e-05, "loss": 3.1484, "step": 113870 }, { "epoch": 0.0127488, "grad_norm": 1.0336521863937378, "learning_rate": 1.611361277902274e-05, "loss": 3.125, "step": 113880 }, { "epoch": 0.0127744, "grad_norm": 0.9818047285079956, "learning_rate": 1.6112975500773788e-05, "loss": 3.7319, "step": 113890 }, { "epoch": 0.0128, "grad_norm": 0.7059646844863892, "learning_rate": 1.6112338182883814e-05, "loss": 3.2486, "step": 113900 }, { "epoch": 0.0128256, "grad_norm": 0.7615141868591309, "learning_rate": 1.6111700825356954e-05, "loss": 3.0623, "step": 113910 }, { "epoch": 0.0128512, "grad_norm": 0.6994110941886902, "learning_rate": 1.611106342819733e-05, "loss": 2.9735, "step": 113920 }, { "epoch": 0.0128768, "grad_norm": 0.8467179536819458, "learning_rate": 1.611042599140909e-05, "loss": 3.0014, "step": 113930 }, { "epoch": 0.0129024, "grad_norm": 0.7065028548240662, "learning_rate": 1.610978851499636e-05, "loss": 2.8901, "step": 113940 }, { "epoch": 0.012928, "grad_norm": 1.77944016456604, "learning_rate": 1.6109150998963277e-05, "loss": 3.2617, "step": 113950 }, { "epoch": 0.0129536, "grad_norm": 0.681781530380249, "learning_rate": 1.610851344331397e-05, "loss": 3.0881, "step": 113960 }, { "epoch": 0.0129792, "grad_norm": 0.7678276896476746, "learning_rate": 1.610787584805258e-05, "loss": 2.9511, "step": 113970 }, { "epoch": 0.0130048, "grad_norm": 0.7150656580924988, "learning_rate": 1.610723821318324e-05, "loss": 3.0008, "step": 113980 }, { "epoch": 0.0130304, "grad_norm": 1.0296721458435059, "learning_rate": 1.6106600538710077e-05, "loss": 3.0831, "step": 113990 }, { "epoch": 0.013056, "grad_norm": 0.7786634564399719, "learning_rate": 1.6105962824637234e-05, "loss": 3.2049, "step": 114000 }, { "epoch": 0.0130816, "grad_norm": 0.6959714889526367, "learning_rate": 1.6105325070968847e-05, "loss": 3.0761, "step": 114010 }, { "epoch": 0.0131072, "grad_norm": 0.7761219143867493, "learning_rate": 1.6104687277709052e-05, "loss": 3.0453, "step": 114020 }, { "epoch": 0.0131328, "grad_norm": 0.72740238904953, "learning_rate": 1.610404944486198e-05, "loss": 3.1362, "step": 114030 }, { "epoch": 0.0131584, "grad_norm": 1.2260676622390747, "learning_rate": 1.610341157243177e-05, "loss": 3.3542, "step": 114040 }, { "epoch": 0.013184, "grad_norm": 0.8443571329116821, "learning_rate": 1.6102773660422553e-05, "loss": 3.0796, "step": 114050 }, { "epoch": 0.0132096, "grad_norm": 0.6890577673912048, "learning_rate": 1.6102135708838477e-05, "loss": 3.0438, "step": 114060 }, { "epoch": 0.0132352, "grad_norm": 0.7663807272911072, "learning_rate": 1.610149771768367e-05, "loss": 3.0295, "step": 114070 }, { "epoch": 0.0132608, "grad_norm": 0.775509774684906, "learning_rate": 1.6100859686962272e-05, "loss": 2.9556, "step": 114080 }, { "epoch": 0.0132864, "grad_norm": 0.8517870306968689, "learning_rate": 1.6100221616678424e-05, "loss": 3.4359, "step": 114090 }, { "epoch": 0.013312, "grad_norm": 0.6311568021774292, "learning_rate": 1.6099583506836253e-05, "loss": 3.1431, "step": 114100 }, { "epoch": 0.0133376, "grad_norm": 0.7767282724380493, "learning_rate": 1.6098945357439907e-05, "loss": 3.1053, "step": 114110 }, { "epoch": 0.0133632, "grad_norm": 1.718165636062622, "learning_rate": 1.6098307168493526e-05, "loss": 3.168, "step": 114120 }, { "epoch": 0.0133888, "grad_norm": 0.7486051321029663, "learning_rate": 1.609766894000124e-05, "loss": 3.0266, "step": 114130 }, { "epoch": 0.0134144, "grad_norm": 0.8160644769668579, "learning_rate": 1.6097030671967192e-05, "loss": 3.13, "step": 114140 }, { "epoch": 0.01344, "grad_norm": 0.8419447541236877, "learning_rate": 1.6096392364395517e-05, "loss": 3.4726, "step": 114150 }, { "epoch": 0.0134656, "grad_norm": 0.94510817527771, "learning_rate": 1.6095754017290362e-05, "loss": 3.5432, "step": 114160 }, { "epoch": 0.0134912, "grad_norm": 1.022130012512207, "learning_rate": 1.6095115630655862e-05, "loss": 3.0721, "step": 114170 }, { "epoch": 0.0135168, "grad_norm": 0.6541037559509277, "learning_rate": 1.6094477204496156e-05, "loss": 3.0835, "step": 114180 }, { "epoch": 0.0135424, "grad_norm": 0.6817511916160583, "learning_rate": 1.6093838738815383e-05, "loss": 2.9621, "step": 114190 }, { "epoch": 0.013568, "grad_norm": 0.8783578276634216, "learning_rate": 1.6093200233617688e-05, "loss": 3.0464, "step": 114200 }, { "epoch": 0.0135936, "grad_norm": 0.8067980408668518, "learning_rate": 1.609256168890721e-05, "loss": 2.9771, "step": 114210 }, { "epoch": 0.0136192, "grad_norm": 0.7332670092582703, "learning_rate": 1.6091923104688086e-05, "loss": 3.3189, "step": 114220 }, { "epoch": 0.0136448, "grad_norm": 0.7221235036849976, "learning_rate": 1.609128448096446e-05, "loss": 3.1419, "step": 114230 }, { "epoch": 0.0136704, "grad_norm": 1.884511113166809, "learning_rate": 1.6090645817740472e-05, "loss": 3.0659, "step": 114240 }, { "epoch": 0.013696, "grad_norm": 0.7163938283920288, "learning_rate": 1.6090007115020267e-05, "loss": 3.2435, "step": 114250 }, { "epoch": 0.0137216, "grad_norm": 0.7165133357048035, "learning_rate": 1.6089368372807984e-05, "loss": 3.0637, "step": 114260 }, { "epoch": 0.0137472, "grad_norm": 0.7542372941970825, "learning_rate": 1.6088729591107766e-05, "loss": 3.0295, "step": 114270 }, { "epoch": 0.0137728, "grad_norm": 0.7112265825271606, "learning_rate": 1.6088090769923752e-05, "loss": 3.1609, "step": 114280 }, { "epoch": 0.0137984, "grad_norm": 0.7642626166343689, "learning_rate": 1.6087451909260088e-05, "loss": 3.1493, "step": 114290 }, { "epoch": 0.013824, "grad_norm": 0.6825520396232605, "learning_rate": 1.608681300912092e-05, "loss": 3.2171, "step": 114300 }, { "epoch": 0.0138496, "grad_norm": 1.2432829141616821, "learning_rate": 1.6086174069510384e-05, "loss": 3.1607, "step": 114310 }, { "epoch": 0.0138752, "grad_norm": 0.7129283547401428, "learning_rate": 1.608553509043263e-05, "loss": 2.9989, "step": 114320 }, { "epoch": 0.0139008, "grad_norm": 0.7567805051803589, "learning_rate": 1.6084896071891794e-05, "loss": 3.188, "step": 114330 }, { "epoch": 0.0139264, "grad_norm": 0.7049436569213867, "learning_rate": 1.6084257013892025e-05, "loss": 2.8835, "step": 114340 }, { "epoch": 0.013952, "grad_norm": 1.070764422416687, "learning_rate": 1.6083617916437468e-05, "loss": 3.2466, "step": 114350 }, { "epoch": 0.0139776, "grad_norm": 0.7867716550827026, "learning_rate": 1.608297877953226e-05, "loss": 3.342, "step": 114360 }, { "epoch": 0.0140032, "grad_norm": 0.9347223043441772, "learning_rate": 1.6082339603180558e-05, "loss": 3.2872, "step": 114370 }, { "epoch": 0.0140288, "grad_norm": 0.7660688161849976, "learning_rate": 1.6081700387386502e-05, "loss": 3.209, "step": 114380 }, { "epoch": 0.0140544, "grad_norm": 0.7442470788955688, "learning_rate": 1.6081061132154227e-05, "loss": 3.065, "step": 114390 }, { "epoch": 0.01408, "grad_norm": 0.6559311747550964, "learning_rate": 1.6080421837487892e-05, "loss": 3.0901, "step": 114400 }, { "epoch": 0.0141056, "grad_norm": 3.7777016162872314, "learning_rate": 1.607978250339164e-05, "loss": 3.1481, "step": 114410 }, { "epoch": 0.0141312, "grad_norm": 0.7892477512359619, "learning_rate": 1.607914312986961e-05, "loss": 2.9925, "step": 114420 }, { "epoch": 0.0141568, "grad_norm": 0.7230266332626343, "learning_rate": 1.6078503716925953e-05, "loss": 3.3019, "step": 114430 }, { "epoch": 0.0141824, "grad_norm": 1.0549781322479248, "learning_rate": 1.6077864264564814e-05, "loss": 3.121, "step": 114440 }, { "epoch": 0.014208, "grad_norm": 0.9015457630157471, "learning_rate": 1.607722477279034e-05, "loss": 3.0775, "step": 114450 }, { "epoch": 0.0142336, "grad_norm": 0.7773035168647766, "learning_rate": 1.6076585241606682e-05, "loss": 3.0102, "step": 114460 }, { "epoch": 0.0142592, "grad_norm": 0.7121808528900146, "learning_rate": 1.6075945671017982e-05, "loss": 3.2396, "step": 114470 }, { "epoch": 0.0142848, "grad_norm": 1.2138285636901855, "learning_rate": 1.607530606102839e-05, "loss": 3.9159, "step": 114480 }, { "epoch": 0.0143104, "grad_norm": 1.0723053216934204, "learning_rate": 1.607466641164205e-05, "loss": 2.9854, "step": 114490 }, { "epoch": 0.014336, "grad_norm": 0.9952747821807861, "learning_rate": 1.6074026722863118e-05, "loss": 3.2946, "step": 114500 }, { "epoch": 0.0143616, "grad_norm": 0.7136436104774475, "learning_rate": 1.607338699469573e-05, "loss": 3.0172, "step": 114510 }, { "epoch": 0.0143872, "grad_norm": 0.8543176651000977, "learning_rate": 1.6072747227144043e-05, "loss": 3.1384, "step": 114520 }, { "epoch": 0.0144128, "grad_norm": 0.7502086162567139, "learning_rate": 1.607210742021221e-05, "loss": 3.4007, "step": 114530 }, { "epoch": 0.0144384, "grad_norm": 0.633707582950592, "learning_rate": 1.6071467573904368e-05, "loss": 3.0772, "step": 114540 }, { "epoch": 0.014464, "grad_norm": 0.7578427195549011, "learning_rate": 1.6070827688224677e-05, "loss": 2.8954, "step": 114550 }, { "epoch": 0.0144896, "grad_norm": 0.7575653195381165, "learning_rate": 1.6070187763177277e-05, "loss": 3.1142, "step": 114560 }, { "epoch": 0.0145152, "grad_norm": 0.7817381024360657, "learning_rate": 1.6069547798766328e-05, "loss": 3.0888, "step": 114570 }, { "epoch": 0.0145408, "grad_norm": 0.7679798603057861, "learning_rate": 1.6068907794995973e-05, "loss": 3.3231, "step": 114580 }, { "epoch": 0.0145664, "grad_norm": 0.784938633441925, "learning_rate": 1.6068267751870357e-05, "loss": 3.1857, "step": 114590 }, { "epoch": 0.014592, "grad_norm": 0.6803986430168152, "learning_rate": 1.6067627669393647e-05, "loss": 3.254, "step": 114600 }, { "epoch": 0.0146176, "grad_norm": 0.7258595824241638, "learning_rate": 1.606698754756998e-05, "loss": 3.2246, "step": 114610 }, { "epoch": 0.0146432, "grad_norm": 0.6573030352592468, "learning_rate": 1.606634738640351e-05, "loss": 3.0431, "step": 114620 }, { "epoch": 0.0146688, "grad_norm": 0.7723381519317627, "learning_rate": 1.6065707185898392e-05, "loss": 3.013, "step": 114630 }, { "epoch": 0.0146944, "grad_norm": 0.8441811203956604, "learning_rate": 1.6065066946058773e-05, "loss": 3.1759, "step": 114640 }, { "epoch": 0.01472, "grad_norm": 0.6426613330841064, "learning_rate": 1.606442666688881e-05, "loss": 2.8116, "step": 114650 }, { "epoch": 0.0147456, "grad_norm": 0.8760749697685242, "learning_rate": 1.6063786348392646e-05, "loss": 2.9522, "step": 114660 }, { "epoch": 0.0147712, "grad_norm": 0.9049730896949768, "learning_rate": 1.6063145990574446e-05, "loss": 2.9734, "step": 114670 }, { "epoch": 0.0147968, "grad_norm": 1.0298540592193604, "learning_rate": 1.606250559343835e-05, "loss": 3.2185, "step": 114680 }, { "epoch": 0.0148224, "grad_norm": 0.7778957486152649, "learning_rate": 1.6061865156988518e-05, "loss": 3.076, "step": 114690 }, { "epoch": 0.014848, "grad_norm": 1.120097279548645, "learning_rate": 1.60612246812291e-05, "loss": 3.0806, "step": 114700 }, { "epoch": 0.0148736, "grad_norm": 0.7395840287208557, "learning_rate": 1.6060584166164255e-05, "loss": 3.0077, "step": 114710 }, { "epoch": 0.0148992, "grad_norm": 0.7761955857276917, "learning_rate": 1.605994361179813e-05, "loss": 3.0085, "step": 114720 }, { "epoch": 0.0149248, "grad_norm": 0.7284514307975769, "learning_rate": 1.6059303018134883e-05, "loss": 2.9811, "step": 114730 }, { "epoch": 0.0149504, "grad_norm": 1.1094568967819214, "learning_rate": 1.6058662385178663e-05, "loss": 2.9212, "step": 114740 }, { "epoch": 0.014976, "grad_norm": 0.9328688383102417, "learning_rate": 1.605802171293363e-05, "loss": 2.931, "step": 114750 }, { "epoch": 0.0150016, "grad_norm": 0.8567693829536438, "learning_rate": 1.6057381001403935e-05, "loss": 3.2734, "step": 114760 }, { "epoch": 0.0150272, "grad_norm": 0.7321374416351318, "learning_rate": 1.6056740250593733e-05, "loss": 3.0654, "step": 114770 }, { "epoch": 0.0150528, "grad_norm": 0.7501481771469116, "learning_rate": 1.6056099460507186e-05, "loss": 3.168, "step": 114780 }, { "epoch": 0.0150784, "grad_norm": 0.6276878118515015, "learning_rate": 1.6055458631148438e-05, "loss": 3.0573, "step": 114790 }, { "epoch": 0.015104, "grad_norm": 1.078951358795166, "learning_rate": 1.6054817762521652e-05, "loss": 3.1112, "step": 114800 }, { "epoch": 0.0151296, "grad_norm": 0.7857349514961243, "learning_rate": 1.605417685463098e-05, "loss": 3.2271, "step": 114810 }, { "epoch": 0.0151552, "grad_norm": 0.7856634855270386, "learning_rate": 1.6053535907480583e-05, "loss": 3.1766, "step": 114820 }, { "epoch": 0.0151808, "grad_norm": 0.7436631917953491, "learning_rate": 1.605289492107461e-05, "loss": 3.1131, "step": 114830 }, { "epoch": 0.0152064, "grad_norm": 0.7563173174858093, "learning_rate": 1.6052253895417228e-05, "loss": 3.1197, "step": 114840 }, { "epoch": 0.015232, "grad_norm": 0.7868609428405762, "learning_rate": 1.6051612830512583e-05, "loss": 3.0872, "step": 114850 }, { "epoch": 0.0152576, "grad_norm": 0.7017580270767212, "learning_rate": 1.605097172636484e-05, "loss": 2.9545, "step": 114860 }, { "epoch": 0.0152832, "grad_norm": 0.7288659811019897, "learning_rate": 1.6050330582978154e-05, "loss": 3.1899, "step": 114870 }, { "epoch": 0.0153088, "grad_norm": 0.766656756401062, "learning_rate": 1.604968940035668e-05, "loss": 3.2246, "step": 114880 }, { "epoch": 0.0153344, "grad_norm": 0.6785381436347961, "learning_rate": 1.604904817850458e-05, "loss": 2.9728, "step": 114890 }, { "epoch": 0.01536, "grad_norm": 0.7427111864089966, "learning_rate": 1.604840691742601e-05, "loss": 3.1944, "step": 114900 }, { "epoch": 0.0153856, "grad_norm": 0.737044632434845, "learning_rate": 1.6047765617125125e-05, "loss": 3.1083, "step": 114910 }, { "epoch": 0.0154112, "grad_norm": 0.7230579853057861, "learning_rate": 1.604712427760609e-05, "loss": 3.0508, "step": 114920 }, { "epoch": 0.0154368, "grad_norm": 0.7662153244018555, "learning_rate": 1.6046482898873065e-05, "loss": 3.2834, "step": 114930 }, { "epoch": 0.0154624, "grad_norm": 0.7803110480308533, "learning_rate": 1.60458414809302e-05, "loss": 3.013, "step": 114940 }, { "epoch": 0.015488, "grad_norm": 0.7466351985931396, "learning_rate": 1.6045200023781663e-05, "loss": 3.1466, "step": 114950 }, { "epoch": 0.0155136, "grad_norm": 0.8642958402633667, "learning_rate": 1.6044558527431608e-05, "loss": 3.079, "step": 114960 }, { "epoch": 0.0155392, "grad_norm": 0.7939163446426392, "learning_rate": 1.6043916991884195e-05, "loss": 2.7435, "step": 114970 }, { "epoch": 0.0155648, "grad_norm": 0.676986813545227, "learning_rate": 1.6043275417143592e-05, "loss": 3.1706, "step": 114980 }, { "epoch": 0.0155904, "grad_norm": 0.6681327819824219, "learning_rate": 1.6042633803213952e-05, "loss": 3.0289, "step": 114990 }, { "epoch": 0.015616, "grad_norm": 0.8520525097846985, "learning_rate": 1.6041992150099437e-05, "loss": 3.227, "step": 115000 }, { "epoch": 0.0156416, "grad_norm": 0.6685444116592407, "learning_rate": 1.604135045780421e-05, "loss": 2.8593, "step": 115010 }, { "epoch": 0.0156672, "grad_norm": 0.9414058923721313, "learning_rate": 1.6040708726332432e-05, "loss": 3.2803, "step": 115020 }, { "epoch": 0.0156928, "grad_norm": 0.6998686790466309, "learning_rate": 1.604006695568826e-05, "loss": 3.1441, "step": 115030 }, { "epoch": 0.0157184, "grad_norm": 1.14535653591156, "learning_rate": 1.603942514587586e-05, "loss": 3.1703, "step": 115040 }, { "epoch": 0.015744, "grad_norm": 0.8030408024787903, "learning_rate": 1.6038783296899393e-05, "loss": 3.1996, "step": 115050 }, { "epoch": 0.0157696, "grad_norm": 1.064174771308899, "learning_rate": 1.603814140876302e-05, "loss": 2.8874, "step": 115060 }, { "epoch": 0.0157952, "grad_norm": 1.0333279371261597, "learning_rate": 1.603749948147091e-05, "loss": 3.0805, "step": 115070 }, { "epoch": 0.0158208, "grad_norm": 0.7447709441184998, "learning_rate": 1.6036857515027214e-05, "loss": 3.2643, "step": 115080 }, { "epoch": 0.0158464, "grad_norm": 0.7447245717048645, "learning_rate": 1.603621550943611e-05, "loss": 3.1025, "step": 115090 }, { "epoch": 0.015872, "grad_norm": 0.7942084670066833, "learning_rate": 1.6035573464701742e-05, "loss": 3.2008, "step": 115100 }, { "epoch": 0.0158976, "grad_norm": 0.7152696251869202, "learning_rate": 1.6034931380828293e-05, "loss": 2.9674, "step": 115110 }, { "epoch": 0.0159232, "grad_norm": 0.9160066843032837, "learning_rate": 1.6034289257819912e-05, "loss": 3.2149, "step": 115120 }, { "epoch": 0.0159488, "grad_norm": 1.2587887048721313, "learning_rate": 1.6033647095680768e-05, "loss": 3.2401, "step": 115130 }, { "epoch": 0.0159744, "grad_norm": 0.8977571129798889, "learning_rate": 1.6033004894415033e-05, "loss": 3.1365, "step": 115140 }, { "epoch": 0.016, "grad_norm": 0.7025513052940369, "learning_rate": 1.603236265402686e-05, "loss": 3.2163, "step": 115150 }, { "epoch": 0.0160256, "grad_norm": 0.805053174495697, "learning_rate": 1.6031720374520418e-05, "loss": 3.0719, "step": 115160 }, { "epoch": 0.0160512, "grad_norm": 0.9791232943534851, "learning_rate": 1.6031078055899875e-05, "loss": 3.2197, "step": 115170 }, { "epoch": 0.0160768, "grad_norm": 0.8027019500732422, "learning_rate": 1.603043569816939e-05, "loss": 3.1089, "step": 115180 }, { "epoch": 0.0161024, "grad_norm": 0.7308597564697266, "learning_rate": 1.6029793301333138e-05, "loss": 3.0988, "step": 115190 }, { "epoch": 0.016128, "grad_norm": 0.7338427305221558, "learning_rate": 1.6029150865395272e-05, "loss": 3.0682, "step": 115200 }, { "epoch": 0.0161536, "grad_norm": 0.6986845135688782, "learning_rate": 1.6028508390359968e-05, "loss": 2.9406, "step": 115210 }, { "epoch": 0.0161792, "grad_norm": 1.1114041805267334, "learning_rate": 1.6027865876231393e-05, "loss": 3.1562, "step": 115220 }, { "epoch": 0.0162048, "grad_norm": 0.7886353731155396, "learning_rate": 1.6027223323013705e-05, "loss": 3.0096, "step": 115230 }, { "epoch": 0.0162304, "grad_norm": 0.8236733675003052, "learning_rate": 1.6026580730711077e-05, "loss": 3.0609, "step": 115240 }, { "epoch": 0.016256, "grad_norm": 0.7628327012062073, "learning_rate": 1.6025938099327673e-05, "loss": 3.0934, "step": 115250 }, { "epoch": 0.0162816, "grad_norm": 0.7370908856391907, "learning_rate": 1.6025295428867665e-05, "loss": 3.2893, "step": 115260 }, { "epoch": 0.0163072, "grad_norm": 0.802183210849762, "learning_rate": 1.6024652719335217e-05, "loss": 3.0126, "step": 115270 }, { "epoch": 0.0163328, "grad_norm": 1.0384125709533691, "learning_rate": 1.6024009970734495e-05, "loss": 3.1975, "step": 115280 }, { "epoch": 0.0163584, "grad_norm": 0.6827748417854309, "learning_rate": 1.602336718306967e-05, "loss": 3.176, "step": 115290 }, { "epoch": 0.016384, "grad_norm": 0.8084807991981506, "learning_rate": 1.602272435634491e-05, "loss": 3.0274, "step": 115300 }, { "epoch": 0.0164096, "grad_norm": 1.3865342140197754, "learning_rate": 1.6022081490564385e-05, "loss": 2.9203, "step": 115310 }, { "epoch": 0.0164352, "grad_norm": 0.7272325754165649, "learning_rate": 1.602143858573226e-05, "loss": 3.197, "step": 115320 }, { "epoch": 0.0164608, "grad_norm": 0.6476426720619202, "learning_rate": 1.6020795641852708e-05, "loss": 2.9658, "step": 115330 }, { "epoch": 0.0164864, "grad_norm": 0.8031585216522217, "learning_rate": 1.6020152658929893e-05, "loss": 3.2625, "step": 115340 }, { "epoch": 0.016512, "grad_norm": 1.2928467988967896, "learning_rate": 1.6019509636967993e-05, "loss": 2.8862, "step": 115350 }, { "epoch": 0.0165376, "grad_norm": 0.6688327193260193, "learning_rate": 1.601886657597117e-05, "loss": 2.8729, "step": 115360 }, { "epoch": 0.0165632, "grad_norm": 0.7758086323738098, "learning_rate": 1.6018223475943596e-05, "loss": 3.0859, "step": 115370 }, { "epoch": 0.0165888, "grad_norm": 0.8047590255737305, "learning_rate": 1.601758033688944e-05, "loss": 3.078, "step": 115380 }, { "epoch": 0.0166144, "grad_norm": 0.8481021523475647, "learning_rate": 1.6016937158812877e-05, "loss": 2.9188, "step": 115390 }, { "epoch": 0.01664, "grad_norm": 0.8344007730484009, "learning_rate": 1.6016293941718077e-05, "loss": 3.1781, "step": 115400 }, { "epoch": 0.0166656, "grad_norm": 0.7482524514198303, "learning_rate": 1.601565068560921e-05, "loss": 3.0194, "step": 115410 }, { "epoch": 0.0166912, "grad_norm": 0.9898269772529602, "learning_rate": 1.6015007390490446e-05, "loss": 3.3523, "step": 115420 }, { "epoch": 0.0167168, "grad_norm": 0.6875491738319397, "learning_rate": 1.601436405636596e-05, "loss": 3.4024, "step": 115430 }, { "epoch": 0.0167424, "grad_norm": 0.7393812537193298, "learning_rate": 1.601372068323992e-05, "loss": 3.2448, "step": 115440 }, { "epoch": 0.016768, "grad_norm": 0.9212729334831238, "learning_rate": 1.6013077271116498e-05, "loss": 3.4747, "step": 115450 }, { "epoch": 0.0167936, "grad_norm": 0.9322811961174011, "learning_rate": 1.601243381999987e-05, "loss": 3.1511, "step": 115460 }, { "epoch": 0.0168192, "grad_norm": 0.7821263670921326, "learning_rate": 1.6011790329894204e-05, "loss": 2.9308, "step": 115470 }, { "epoch": 0.0168448, "grad_norm": 0.8395355939865112, "learning_rate": 1.6011146800803677e-05, "loss": 3.0643, "step": 115480 }, { "epoch": 0.0168704, "grad_norm": 0.69486403465271, "learning_rate": 1.601050323273246e-05, "loss": 3.1112, "step": 115490 }, { "epoch": 0.016896, "grad_norm": 0.8140734434127808, "learning_rate": 1.6009859625684727e-05, "loss": 3.2031, "step": 115500 }, { "epoch": 0.0169216, "grad_norm": 0.7693402767181396, "learning_rate": 1.6009215979664653e-05, "loss": 3.3642, "step": 115510 }, { "epoch": 0.0169472, "grad_norm": 0.8258143067359924, "learning_rate": 1.600857229467641e-05, "loss": 3.2567, "step": 115520 }, { "epoch": 0.0169728, "grad_norm": 0.7910342216491699, "learning_rate": 1.600792857072417e-05, "loss": 3.2822, "step": 115530 }, { "epoch": 0.0169984, "grad_norm": 0.8211557865142822, "learning_rate": 1.6007284807812115e-05, "loss": 3.3722, "step": 115540 }, { "epoch": 0.017024, "grad_norm": 0.7134290337562561, "learning_rate": 1.600664100594441e-05, "loss": 2.9221, "step": 115550 }, { "epoch": 0.0170496, "grad_norm": 0.940189778804779, "learning_rate": 1.6005997165125235e-05, "loss": 2.9737, "step": 115560 }, { "epoch": 0.0170752, "grad_norm": 0.7043961882591248, "learning_rate": 1.600535328535877e-05, "loss": 3.212, "step": 115570 }, { "epoch": 0.0171008, "grad_norm": 0.763268232345581, "learning_rate": 1.6004709366649176e-05, "loss": 3.1015, "step": 115580 }, { "epoch": 0.0171264, "grad_norm": 0.7650354504585266, "learning_rate": 1.6004065409000642e-05, "loss": 2.9863, "step": 115590 }, { "epoch": 0.017152, "grad_norm": 0.6234548091888428, "learning_rate": 1.600342141241734e-05, "loss": 3.0367, "step": 115600 }, { "epoch": 0.0171776, "grad_norm": 0.6678217053413391, "learning_rate": 1.6002777376903447e-05, "loss": 3.2121, "step": 115610 }, { "epoch": 0.0172032, "grad_norm": 0.7243049740791321, "learning_rate": 1.6002133302463137e-05, "loss": 3.0612, "step": 115620 }, { "epoch": 0.0172288, "grad_norm": 0.7800476551055908, "learning_rate": 1.6001489189100586e-05, "loss": 2.9698, "step": 115630 }, { "epoch": 0.0172544, "grad_norm": 0.6961314082145691, "learning_rate": 1.6000845036819974e-05, "loss": 3.2499, "step": 115640 }, { "epoch": 0.01728, "grad_norm": 0.8300138711929321, "learning_rate": 1.6000200845625476e-05, "loss": 3.1052, "step": 115650 }, { "epoch": 0.0173056, "grad_norm": 0.6719551086425781, "learning_rate": 1.599955661552127e-05, "loss": 3.158, "step": 115660 }, { "epoch": 0.0173312, "grad_norm": 1.118791937828064, "learning_rate": 1.5998912346511532e-05, "loss": 3.1875, "step": 115670 }, { "epoch": 0.0173568, "grad_norm": 0.9025167226791382, "learning_rate": 1.5998268038600445e-05, "loss": 3.038, "step": 115680 }, { "epoch": 0.0173824, "grad_norm": 0.8134221434593201, "learning_rate": 1.5997623691792185e-05, "loss": 3.0753, "step": 115690 }, { "epoch": 0.017408, "grad_norm": 0.7893900275230408, "learning_rate": 1.5996979306090926e-05, "loss": 3.1847, "step": 115700 }, { "epoch": 0.0174336, "grad_norm": 0.6552165746688843, "learning_rate": 1.5996334881500853e-05, "loss": 3.2332, "step": 115710 }, { "epoch": 0.0174592, "grad_norm": 0.7184738516807556, "learning_rate": 1.599569041802614e-05, "loss": 3.2231, "step": 115720 }, { "epoch": 0.0174848, "grad_norm": 0.728987991809845, "learning_rate": 1.599504591567097e-05, "loss": 3.1182, "step": 115730 }, { "epoch": 0.0175104, "grad_norm": 0.8022724986076355, "learning_rate": 1.5994401374439517e-05, "loss": 2.9993, "step": 115740 }, { "epoch": 0.017536, "grad_norm": 0.7433023452758789, "learning_rate": 1.5993756794335967e-05, "loss": 3.3362, "step": 115750 }, { "epoch": 0.0175616, "grad_norm": 0.7817097902297974, "learning_rate": 1.5993112175364494e-05, "loss": 3.3814, "step": 115760 }, { "epoch": 0.0175872, "grad_norm": 1.1861135959625244, "learning_rate": 1.5992467517529283e-05, "loss": 3.4756, "step": 115770 }, { "epoch": 0.0176128, "grad_norm": 0.7478834390640259, "learning_rate": 1.5991822820834518e-05, "loss": 3.3361, "step": 115780 }, { "epoch": 0.0176384, "grad_norm": 0.8732203841209412, "learning_rate": 1.599117808528437e-05, "loss": 3.0954, "step": 115790 }, { "epoch": 0.017664, "grad_norm": 0.7600667476654053, "learning_rate": 1.5990533310883022e-05, "loss": 3.1824, "step": 115800 }, { "epoch": 0.0176896, "grad_norm": 1.0353705883026123, "learning_rate": 1.598988849763466e-05, "loss": 3.1522, "step": 115810 }, { "epoch": 0.0177152, "grad_norm": 0.7971363663673401, "learning_rate": 1.5989243645543464e-05, "loss": 3.1952, "step": 115820 }, { "epoch": 0.0177408, "grad_norm": 0.6732283234596252, "learning_rate": 1.5988598754613612e-05, "loss": 3.1771, "step": 115830 }, { "epoch": 0.0177664, "grad_norm": 0.6766202449798584, "learning_rate": 1.598795382484929e-05, "loss": 2.7872, "step": 115840 }, { "epoch": 0.017792, "grad_norm": 0.7349610328674316, "learning_rate": 1.598730885625468e-05, "loss": 3.1611, "step": 115850 }, { "epoch": 0.0178176, "grad_norm": 0.7828022837638855, "learning_rate": 1.5986663848833962e-05, "loss": 3.3506, "step": 115860 }, { "epoch": 0.0178432, "grad_norm": 0.70601886510849, "learning_rate": 1.5986018802591322e-05, "loss": 3.329, "step": 115870 }, { "epoch": 0.0178688, "grad_norm": 0.7685275673866272, "learning_rate": 1.5985373717530938e-05, "loss": 3.1704, "step": 115880 }, { "epoch": 0.0178944, "grad_norm": 1.1065454483032227, "learning_rate": 1.5984728593656996e-05, "loss": 3.2181, "step": 115890 }, { "epoch": 0.01792, "grad_norm": 0.7052695155143738, "learning_rate": 1.5984083430973686e-05, "loss": 3.1074, "step": 115900 }, { "epoch": 0.0179456, "grad_norm": 0.7356778383255005, "learning_rate": 1.5983438229485177e-05, "loss": 3.0499, "step": 115910 }, { "epoch": 0.0179712, "grad_norm": 0.7430433630943298, "learning_rate": 1.5982792989195664e-05, "loss": 3.1008, "step": 115920 }, { "epoch": 0.0179968, "grad_norm": 0.7432559728622437, "learning_rate": 1.5982147710109332e-05, "loss": 3.1699, "step": 115930 }, { "epoch": 0.0180224, "grad_norm": 0.7198444604873657, "learning_rate": 1.5981502392230358e-05, "loss": 3.1879, "step": 115940 }, { "epoch": 0.018048, "grad_norm": 0.6884022355079651, "learning_rate": 1.598085703556293e-05, "loss": 3.1276, "step": 115950 }, { "epoch": 0.0180736, "grad_norm": 0.8655784726142883, "learning_rate": 1.5980211640111235e-05, "loss": 3.2393, "step": 115960 }, { "epoch": 0.0180992, "grad_norm": 0.7311469912528992, "learning_rate": 1.597956620587946e-05, "loss": 3.2048, "step": 115970 }, { "epoch": 0.0181248, "grad_norm": 0.7367813587188721, "learning_rate": 1.5978920732871786e-05, "loss": 3.1006, "step": 115980 }, { "epoch": 0.0181504, "grad_norm": 0.864578127861023, "learning_rate": 1.5978275221092395e-05, "loss": 3.1892, "step": 115990 }, { "epoch": 0.018176, "grad_norm": 0.7580136060714722, "learning_rate": 1.5977629670545485e-05, "loss": 3.3275, "step": 116000 }, { "epoch": 0.0182016, "grad_norm": 0.7863845825195312, "learning_rate": 1.597698408123523e-05, "loss": 2.9609, "step": 116010 }, { "epoch": 0.0182272, "grad_norm": 0.8380716443061829, "learning_rate": 1.5976338453165824e-05, "loss": 3.091, "step": 116020 }, { "epoch": 0.0182528, "grad_norm": 0.7446456551551819, "learning_rate": 1.5975692786341453e-05, "loss": 3.291, "step": 116030 }, { "epoch": 0.0182784, "grad_norm": 0.7119495868682861, "learning_rate": 1.59750470807663e-05, "loss": 3.1899, "step": 116040 }, { "epoch": 0.018304, "grad_norm": 0.6888519525527954, "learning_rate": 1.5974401336444553e-05, "loss": 3.2883, "step": 116050 }, { "epoch": 0.0183296, "grad_norm": 0.7093097567558289, "learning_rate": 1.5973755553380407e-05, "loss": 3.0621, "step": 116060 }, { "epoch": 0.0183552, "grad_norm": 0.7957096099853516, "learning_rate": 1.597310973157804e-05, "loss": 3.1372, "step": 116070 }, { "epoch": 0.0183808, "grad_norm": 0.7687450051307678, "learning_rate": 1.5972463871041644e-05, "loss": 3.1422, "step": 116080 }, { "epoch": 0.0184064, "grad_norm": 0.8770846128463745, "learning_rate": 1.5971817971775406e-05, "loss": 3.2032, "step": 116090 }, { "epoch": 0.018432, "grad_norm": 0.7433145046234131, "learning_rate": 1.5971172033783517e-05, "loss": 3.0429, "step": 116100 }, { "epoch": 0.0184576, "grad_norm": 0.7795640826225281, "learning_rate": 1.5970526057070163e-05, "loss": 3.0905, "step": 116110 }, { "epoch": 0.0184832, "grad_norm": 0.7321749925613403, "learning_rate": 1.596988004163954e-05, "loss": 3.3232, "step": 116120 }, { "epoch": 0.0185088, "grad_norm": 0.6608676314353943, "learning_rate": 1.5969233987495824e-05, "loss": 3.0072, "step": 116130 }, { "epoch": 0.0185344, "grad_norm": 0.7646887898445129, "learning_rate": 1.5968587894643215e-05, "loss": 3.071, "step": 116140 }, { "epoch": 0.01856, "grad_norm": 0.8070118427276611, "learning_rate": 1.5967941763085903e-05, "loss": 3.1587, "step": 116150 }, { "epoch": 0.0185856, "grad_norm": 0.745459794998169, "learning_rate": 1.596729559282807e-05, "loss": 3.2505, "step": 116160 }, { "epoch": 0.0186112, "grad_norm": 1.0021979808807373, "learning_rate": 1.596664938387391e-05, "loss": 3.0839, "step": 116170 }, { "epoch": 0.0186368, "grad_norm": 0.6781618595123291, "learning_rate": 1.596600313622762e-05, "loss": 3.0539, "step": 116180 }, { "epoch": 0.0186624, "grad_norm": 0.6714193224906921, "learning_rate": 1.5965356849893382e-05, "loss": 3.1147, "step": 116190 }, { "epoch": 0.018688, "grad_norm": 0.6752594113349915, "learning_rate": 1.5964710524875388e-05, "loss": 3.1658, "step": 116200 }, { "epoch": 0.0187136, "grad_norm": 0.7714471817016602, "learning_rate": 1.5964064161177834e-05, "loss": 3.1015, "step": 116210 }, { "epoch": 0.0187392, "grad_norm": 1.2152022123336792, "learning_rate": 1.5963417758804908e-05, "loss": 3.0801, "step": 116220 }, { "epoch": 0.0187648, "grad_norm": 0.6867991089820862, "learning_rate": 1.5962771317760805e-05, "loss": 3.3777, "step": 116230 }, { "epoch": 0.0187904, "grad_norm": 0.6859958171844482, "learning_rate": 1.5962124838049712e-05, "loss": 3.2438, "step": 116240 }, { "epoch": 0.018816, "grad_norm": 0.7151476740837097, "learning_rate": 1.5961478319675827e-05, "loss": 3.1584, "step": 116250 }, { "epoch": 0.0188416, "grad_norm": 0.8205493688583374, "learning_rate": 1.5960831762643335e-05, "loss": 2.9715, "step": 116260 }, { "epoch": 0.0188672, "grad_norm": 0.7340593934059143, "learning_rate": 1.5960185166956437e-05, "loss": 3.1287, "step": 116270 }, { "epoch": 0.0188928, "grad_norm": 0.7125808596611023, "learning_rate": 1.5959538532619324e-05, "loss": 3.1227, "step": 116280 }, { "epoch": 0.0189184, "grad_norm": 0.7796851992607117, "learning_rate": 1.5958891859636184e-05, "loss": 3.2533, "step": 116290 }, { "epoch": 0.018944, "grad_norm": 0.9184260368347168, "learning_rate": 1.5958245148011214e-05, "loss": 3.1414, "step": 116300 }, { "epoch": 0.0189696, "grad_norm": 0.7899242639541626, "learning_rate": 1.595759839774861e-05, "loss": 3.0917, "step": 116310 }, { "epoch": 0.0189952, "grad_norm": 1.7869995832443237, "learning_rate": 1.595695160885256e-05, "loss": 3.1419, "step": 116320 }, { "epoch": 0.0190208, "grad_norm": 0.7580373883247375, "learning_rate": 1.5956304781327266e-05, "loss": 3.0676, "step": 116330 }, { "epoch": 0.0190464, "grad_norm": 0.6885644793510437, "learning_rate": 1.5955657915176917e-05, "loss": 3.1742, "step": 116340 }, { "epoch": 0.019072, "grad_norm": 0.6684139966964722, "learning_rate": 1.5955011010405707e-05, "loss": 3.1537, "step": 116350 }, { "epoch": 0.0190976, "grad_norm": 0.6731612086296082, "learning_rate": 1.5954364067017835e-05, "loss": 3.1781, "step": 116360 }, { "epoch": 0.0191232, "grad_norm": 0.6968110203742981, "learning_rate": 1.5953717085017496e-05, "loss": 3.069, "step": 116370 }, { "epoch": 0.0191488, "grad_norm": 0.765446126461029, "learning_rate": 1.5953070064408883e-05, "loss": 3.1136, "step": 116380 }, { "epoch": 0.0191744, "grad_norm": 0.6649048328399658, "learning_rate": 1.5952423005196194e-05, "loss": 3.0974, "step": 116390 }, { "epoch": 0.0192, "grad_norm": 0.6390575766563416, "learning_rate": 1.595177590738362e-05, "loss": 3.0934, "step": 116400 }, { "epoch": 0.0192256, "grad_norm": 0.7013968229293823, "learning_rate": 1.5951128770975362e-05, "loss": 2.9673, "step": 116410 }, { "epoch": 0.0192512, "grad_norm": 0.8479849100112915, "learning_rate": 1.5950481595975618e-05, "loss": 3.0854, "step": 116420 }, { "epoch": 0.0192768, "grad_norm": 1.0688611268997192, "learning_rate": 1.5949834382388583e-05, "loss": 2.8783, "step": 116430 }, { "epoch": 0.0193024, "grad_norm": 1.1311588287353516, "learning_rate": 1.5949187130218452e-05, "loss": 2.9489, "step": 116440 }, { "epoch": 0.019328, "grad_norm": 0.7387770414352417, "learning_rate": 1.594853983946942e-05, "loss": 2.9976, "step": 116450 }, { "epoch": 0.0193536, "grad_norm": 0.7126542329788208, "learning_rate": 1.5947892510145693e-05, "loss": 2.7328, "step": 116460 }, { "epoch": 0.0193792, "grad_norm": 0.7380394339561462, "learning_rate": 1.594724514225146e-05, "loss": 2.9054, "step": 116470 }, { "epoch": 0.0194048, "grad_norm": 0.7398974895477295, "learning_rate": 1.5946597735790924e-05, "loss": 2.8674, "step": 116480 }, { "epoch": 0.0194304, "grad_norm": 0.7608436346054077, "learning_rate": 1.5945950290768287e-05, "loss": 2.7314, "step": 116490 }, { "epoch": 0.019456, "grad_norm": 0.6998135447502136, "learning_rate": 1.5945302807187735e-05, "loss": 2.6786, "step": 116500 }, { "epoch": 0.0194816, "grad_norm": 0.6951549649238586, "learning_rate": 1.594465528505348e-05, "loss": 3.0026, "step": 116510 }, { "epoch": 0.0195072, "grad_norm": 0.7314658761024475, "learning_rate": 1.5944007724369713e-05, "loss": 2.8312, "step": 116520 }, { "epoch": 0.0195328, "grad_norm": 0.6979000568389893, "learning_rate": 1.5943360125140635e-05, "loss": 2.897, "step": 116530 }, { "epoch": 0.0195584, "grad_norm": 0.7021425366401672, "learning_rate": 1.5942712487370447e-05, "loss": 2.7636, "step": 116540 }, { "epoch": 0.019584, "grad_norm": 0.84238201379776, "learning_rate": 1.5942064811063347e-05, "loss": 2.7829, "step": 116550 }, { "epoch": 0.0196096, "grad_norm": 0.8110218644142151, "learning_rate": 1.594141709622354e-05, "loss": 3.0247, "step": 116560 }, { "epoch": 0.0196352, "grad_norm": 0.7858093976974487, "learning_rate": 1.5940769342855217e-05, "loss": 2.824, "step": 116570 }, { "epoch": 0.0196608, "grad_norm": 0.7253391146659851, "learning_rate": 1.5940121550962583e-05, "loss": 2.8476, "step": 116580 }, { "epoch": 0.0196864, "grad_norm": 0.8394119143486023, "learning_rate": 1.5939473720549842e-05, "loss": 2.8856, "step": 116590 }, { "epoch": 0.019712, "grad_norm": 0.6973821520805359, "learning_rate": 1.5938825851621195e-05, "loss": 2.8814, "step": 116600 }, { "epoch": 0.0197376, "grad_norm": 0.8368275165557861, "learning_rate": 1.5938177944180836e-05, "loss": 2.9566, "step": 116610 }, { "epoch": 0.0197632, "grad_norm": 0.8090068697929382, "learning_rate": 1.5937529998232974e-05, "loss": 3.1355, "step": 116620 }, { "epoch": 0.0197888, "grad_norm": 0.8193808197975159, "learning_rate": 1.5936882013781806e-05, "loss": 2.9179, "step": 116630 }, { "epoch": 0.0198144, "grad_norm": 0.6248961091041565, "learning_rate": 1.5936233990831537e-05, "loss": 2.6742, "step": 116640 }, { "epoch": 0.01984, "grad_norm": 0.6905590891838074, "learning_rate": 1.593558592938637e-05, "loss": 3.0226, "step": 116650 }, { "epoch": 0.0198656, "grad_norm": 0.6988979578018188, "learning_rate": 1.5934937829450505e-05, "loss": 2.9167, "step": 116660 }, { "epoch": 0.0198912, "grad_norm": 0.7304912805557251, "learning_rate": 1.5934289691028144e-05, "loss": 2.926, "step": 116670 }, { "epoch": 0.0199168, "grad_norm": 0.7822540402412415, "learning_rate": 1.593364151412349e-05, "loss": 2.9612, "step": 116680 }, { "epoch": 0.0199424, "grad_norm": 0.6682946085929871, "learning_rate": 1.5932993298740754e-05, "loss": 3.0985, "step": 116690 }, { "epoch": 0.019968, "grad_norm": 0.7156141996383667, "learning_rate": 1.5932345044884126e-05, "loss": 2.893, "step": 116700 }, { "epoch": 0.0199936, "grad_norm": 0.8108616471290588, "learning_rate": 1.593169675255782e-05, "loss": 3.1697, "step": 116710 }, { "epoch": 0.0200192, "grad_norm": 1.3595941066741943, "learning_rate": 1.593104842176604e-05, "loss": 3.2262, "step": 116720 }, { "epoch": 0.0200448, "grad_norm": 0.7529960870742798, "learning_rate": 1.5930400052512986e-05, "loss": 2.8526, "step": 116730 }, { "epoch": 0.0200704, "grad_norm": 0.777850866317749, "learning_rate": 1.592975164480286e-05, "loss": 2.8313, "step": 116740 }, { "epoch": 0.020096, "grad_norm": 0.6641184687614441, "learning_rate": 1.5929103198639873e-05, "loss": 2.9141, "step": 116750 }, { "epoch": 0.0201216, "grad_norm": 0.6927085518836975, "learning_rate": 1.5928454714028228e-05, "loss": 2.9286, "step": 116760 }, { "epoch": 0.0201472, "grad_norm": 0.7146192789077759, "learning_rate": 1.592780619097213e-05, "loss": 2.7088, "step": 116770 }, { "epoch": 0.0201728, "grad_norm": 0.7125889658927917, "learning_rate": 1.5927157629475782e-05, "loss": 2.9951, "step": 116780 }, { "epoch": 0.0201984, "grad_norm": 0.6941321492195129, "learning_rate": 1.5926509029543393e-05, "loss": 2.7029, "step": 116790 }, { "epoch": 0.020224, "grad_norm": 0.5982988476753235, "learning_rate": 1.5925860391179168e-05, "loss": 2.6999, "step": 116800 }, { "epoch": 0.0202496, "grad_norm": 0.6504491567611694, "learning_rate": 1.5925211714387315e-05, "loss": 2.8352, "step": 116810 }, { "epoch": 0.0202752, "grad_norm": 0.7482448220252991, "learning_rate": 1.592456299917204e-05, "loss": 2.8262, "step": 116820 }, { "epoch": 0.0203008, "grad_norm": 0.6756120324134827, "learning_rate": 1.5923914245537545e-05, "loss": 2.7633, "step": 116830 }, { "epoch": 0.0203264, "grad_norm": 0.9408334493637085, "learning_rate": 1.592326545348804e-05, "loss": 3.026, "step": 116840 }, { "epoch": 0.020352, "grad_norm": 0.7091398239135742, "learning_rate": 1.5922616623027735e-05, "loss": 3.1658, "step": 116850 }, { "epoch": 0.0203776, "grad_norm": 0.7065417170524597, "learning_rate": 1.5921967754160834e-05, "loss": 2.7685, "step": 116860 }, { "epoch": 0.0204032, "grad_norm": 0.8153745532035828, "learning_rate": 1.5921318846891546e-05, "loss": 3.1634, "step": 116870 }, { "epoch": 0.0204288, "grad_norm": 1.0865517854690552, "learning_rate": 1.5920669901224082e-05, "loss": 3.03, "step": 116880 }, { "epoch": 0.0204544, "grad_norm": 0.7776234745979309, "learning_rate": 1.5920020917162645e-05, "loss": 2.9049, "step": 116890 }, { "epoch": 0.02048, "grad_norm": 0.8028183579444885, "learning_rate": 1.5919371894711445e-05, "loss": 3.6483, "step": 116900 }, { "epoch": 0.0205056, "grad_norm": 0.7335556745529175, "learning_rate": 1.591872283387469e-05, "loss": 2.982, "step": 116910 }, { "epoch": 0.0205312, "grad_norm": 0.7230268120765686, "learning_rate": 1.5918073734656594e-05, "loss": 2.9267, "step": 116920 }, { "epoch": 0.0205568, "grad_norm": 0.7761354446411133, "learning_rate": 1.5917424597061358e-05, "loss": 2.9235, "step": 116930 }, { "epoch": 0.0205824, "grad_norm": 0.7828847169876099, "learning_rate": 1.59167754210932e-05, "loss": 2.908, "step": 116940 }, { "epoch": 0.020608, "grad_norm": 0.7519556283950806, "learning_rate": 1.5916126206756327e-05, "loss": 2.7601, "step": 116950 }, { "epoch": 0.0206336, "grad_norm": 1.0596320629119873, "learning_rate": 1.5915476954054945e-05, "loss": 2.9226, "step": 116960 }, { "epoch": 0.0206592, "grad_norm": 0.7434589266777039, "learning_rate": 1.591482766299327e-05, "loss": 2.9765, "step": 116970 }, { "epoch": 0.0206848, "grad_norm": 0.9127278923988342, "learning_rate": 1.5914178333575504e-05, "loss": 2.9392, "step": 116980 }, { "epoch": 0.0207104, "grad_norm": 0.7569241523742676, "learning_rate": 1.5913528965805868e-05, "loss": 3.0855, "step": 116990 }, { "epoch": 0.020736, "grad_norm": 0.7935746312141418, "learning_rate": 1.5912879559688567e-05, "loss": 2.9035, "step": 117000 }, { "epoch": 0.0207616, "grad_norm": 0.7069028615951538, "learning_rate": 1.5912230115227813e-05, "loss": 2.9717, "step": 117010 }, { "epoch": 0.0207872, "grad_norm": 0.7055628895759583, "learning_rate": 1.591158063242782e-05, "loss": 2.6301, "step": 117020 }, { "epoch": 0.0208128, "grad_norm": 0.7661616206169128, "learning_rate": 1.5910931111292796e-05, "loss": 2.9074, "step": 117030 }, { "epoch": 0.0208384, "grad_norm": 0.6844205856323242, "learning_rate": 1.5910281551826956e-05, "loss": 2.916, "step": 117040 }, { "epoch": 0.020864, "grad_norm": 1.0415657758712769, "learning_rate": 1.590963195403451e-05, "loss": 3.0402, "step": 117050 }, { "epoch": 0.0208896, "grad_norm": 0.7061111927032471, "learning_rate": 1.5908982317919672e-05, "loss": 3.0217, "step": 117060 }, { "epoch": 0.0209152, "grad_norm": 0.8276165723800659, "learning_rate": 1.5908332643486653e-05, "loss": 2.9813, "step": 117070 }, { "epoch": 0.0209408, "grad_norm": 0.6783965229988098, "learning_rate": 1.5907682930739668e-05, "loss": 3.0045, "step": 117080 }, { "epoch": 0.0209664, "grad_norm": 1.0645784139633179, "learning_rate": 1.5907033179682928e-05, "loss": 3.0256, "step": 117090 }, { "epoch": 0.020992, "grad_norm": 0.755622386932373, "learning_rate": 1.5906383390320647e-05, "loss": 3.0217, "step": 117100 }, { "epoch": 0.0210176, "grad_norm": 0.6685400009155273, "learning_rate": 1.590573356265704e-05, "loss": 2.9515, "step": 117110 }, { "epoch": 0.0210432, "grad_norm": 0.7576996684074402, "learning_rate": 1.5905083696696323e-05, "loss": 2.9101, "step": 117120 }, { "epoch": 0.0210688, "grad_norm": 0.6891135573387146, "learning_rate": 1.5904433792442704e-05, "loss": 2.8402, "step": 117130 }, { "epoch": 0.0210944, "grad_norm": 0.7214924097061157, "learning_rate": 1.5903783849900403e-05, "loss": 2.9598, "step": 117140 }, { "epoch": 0.02112, "grad_norm": 0.6290484070777893, "learning_rate": 1.5903133869073633e-05, "loss": 2.7906, "step": 117150 }, { "epoch": 0.0211456, "grad_norm": 0.6918420195579529, "learning_rate": 1.5902483849966608e-05, "loss": 3.0181, "step": 117160 }, { "epoch": 0.0211712, "grad_norm": 0.6779277324676514, "learning_rate": 1.5901833792583546e-05, "loss": 2.9485, "step": 117170 }, { "epoch": 0.0211968, "grad_norm": 0.7857188582420349, "learning_rate": 1.590118369692866e-05, "loss": 2.9887, "step": 117180 }, { "epoch": 0.0212224, "grad_norm": 0.7297973036766052, "learning_rate": 1.5900533563006163e-05, "loss": 3.2223, "step": 117190 }, { "epoch": 0.021248, "grad_norm": 0.7884229421615601, "learning_rate": 1.5899883390820277e-05, "loss": 3.1827, "step": 117200 }, { "epoch": 0.0212736, "grad_norm": 0.7041546702384949, "learning_rate": 1.5899233180375214e-05, "loss": 2.902, "step": 117210 }, { "epoch": 0.0212992, "grad_norm": 0.7087988257408142, "learning_rate": 1.589858293167519e-05, "loss": 3.0353, "step": 117220 }, { "epoch": 0.0213248, "grad_norm": 0.7229814529418945, "learning_rate": 1.5897932644724427e-05, "loss": 2.8407, "step": 117230 }, { "epoch": 0.0213504, "grad_norm": 0.7250144481658936, "learning_rate": 1.5897282319527135e-05, "loss": 2.9453, "step": 117240 }, { "epoch": 0.021376, "grad_norm": 0.8890979886054993, "learning_rate": 1.5896631956087536e-05, "loss": 2.9426, "step": 117250 }, { "epoch": 0.0214016, "grad_norm": 0.7439370155334473, "learning_rate": 1.5895981554409847e-05, "loss": 2.9767, "step": 117260 }, { "epoch": 0.0214272, "grad_norm": 0.753272533416748, "learning_rate": 1.5895331114498283e-05, "loss": 2.8757, "step": 117270 }, { "epoch": 0.0214528, "grad_norm": 0.717753529548645, "learning_rate": 1.5894680636357063e-05, "loss": 3.125, "step": 117280 }, { "epoch": 0.0214784, "grad_norm": 0.7526137828826904, "learning_rate": 1.5894030119990405e-05, "loss": 2.8955, "step": 117290 }, { "epoch": 0.021504, "grad_norm": 0.9809553623199463, "learning_rate": 1.5893379565402534e-05, "loss": 2.9256, "step": 117300 }, { "epoch": 0.0215296, "grad_norm": 0.8391861319541931, "learning_rate": 1.5892728972597655e-05, "loss": 3.0678, "step": 117310 }, { "epoch": 0.0215552, "grad_norm": 0.7152854800224304, "learning_rate": 1.5892078341579998e-05, "loss": 2.7639, "step": 117320 }, { "epoch": 0.0215808, "grad_norm": 0.7291511297225952, "learning_rate": 1.589142767235378e-05, "loss": 2.404, "step": 117330 }, { "epoch": 0.0216064, "grad_norm": 1.0621691942214966, "learning_rate": 1.5890776964923218e-05, "loss": 2.9887, "step": 117340 }, { "epoch": 0.021632, "grad_norm": 0.7236348986625671, "learning_rate": 1.5890126219292534e-05, "loss": 2.8525, "step": 117350 }, { "epoch": 0.0216576, "grad_norm": 1.1048105955123901, "learning_rate": 1.5889475435465946e-05, "loss": 2.8416, "step": 117360 }, { "epoch": 0.0216832, "grad_norm": 0.7088680863380432, "learning_rate": 1.5888824613447676e-05, "loss": 3.0334, "step": 117370 }, { "epoch": 0.0217088, "grad_norm": 0.6673048138618469, "learning_rate": 1.5888173753241942e-05, "loss": 2.8802, "step": 117380 }, { "epoch": 0.0217344, "grad_norm": 0.6626047492027283, "learning_rate": 1.588752285485297e-05, "loss": 3.0843, "step": 117390 }, { "epoch": 0.02176, "grad_norm": 1.3614211082458496, "learning_rate": 1.588687191828497e-05, "loss": 3.1443, "step": 117400 }, { "epoch": 0.0217856, "grad_norm": 0.7394474744796753, "learning_rate": 1.5886220943542176e-05, "loss": 3.093, "step": 117410 }, { "epoch": 0.0218112, "grad_norm": 0.7453150749206543, "learning_rate": 1.5885569930628798e-05, "loss": 3.1551, "step": 117420 }, { "epoch": 0.0218368, "grad_norm": 0.9721931219100952, "learning_rate": 1.588491887954907e-05, "loss": 3.0541, "step": 117430 }, { "epoch": 0.0218624, "grad_norm": 0.8203706741333008, "learning_rate": 1.58842677903072e-05, "loss": 3.0341, "step": 117440 }, { "epoch": 0.021888, "grad_norm": 0.8538007736206055, "learning_rate": 1.588361666290742e-05, "loss": 3.0624, "step": 117450 }, { "epoch": 0.0219136, "grad_norm": 0.7342122197151184, "learning_rate": 1.5882965497353953e-05, "loss": 3.0559, "step": 117460 }, { "epoch": 0.0219392, "grad_norm": 0.7265629172325134, "learning_rate": 1.5882314293651016e-05, "loss": 3.1869, "step": 117470 }, { "epoch": 0.0219648, "grad_norm": 0.8500844836235046, "learning_rate": 1.588166305180283e-05, "loss": 3.0451, "step": 117480 }, { "epoch": 0.0219904, "grad_norm": 0.7067074179649353, "learning_rate": 1.588101177181363e-05, "loss": 2.9685, "step": 117490 }, { "epoch": 0.022016, "grad_norm": 0.6487573385238647, "learning_rate": 1.5880360453687626e-05, "loss": 2.875, "step": 117500 }, { "epoch": 0.0220416, "grad_norm": 0.7484943866729736, "learning_rate": 1.587970909742905e-05, "loss": 2.9319, "step": 117510 }, { "epoch": 0.0220672, "grad_norm": 0.7059506177902222, "learning_rate": 1.587905770304212e-05, "loss": 3.0459, "step": 117520 }, { "epoch": 0.0220928, "grad_norm": 0.7070468664169312, "learning_rate": 1.5878406270531068e-05, "loss": 2.9929, "step": 117530 }, { "epoch": 0.0221184, "grad_norm": 0.7051637768745422, "learning_rate": 1.587775479990011e-05, "loss": 3.2551, "step": 117540 }, { "epoch": 0.022144, "grad_norm": 0.8438124656677246, "learning_rate": 1.5877103291153477e-05, "loss": 3.015, "step": 117550 }, { "epoch": 0.0221696, "grad_norm": 0.8122435808181763, "learning_rate": 1.587645174429539e-05, "loss": 3.0945, "step": 117560 }, { "epoch": 0.0221952, "grad_norm": 0.6347606778144836, "learning_rate": 1.5875800159330073e-05, "loss": 3.0366, "step": 117570 }, { "epoch": 0.0222208, "grad_norm": 0.6992306709289551, "learning_rate": 1.5875148536261757e-05, "loss": 2.9165, "step": 117580 }, { "epoch": 0.0222464, "grad_norm": 0.790565013885498, "learning_rate": 1.5874496875094662e-05, "loss": 3.0177, "step": 117590 }, { "epoch": 0.022272, "grad_norm": 0.7268771529197693, "learning_rate": 1.5873845175833015e-05, "loss": 2.9652, "step": 117600 }, { "epoch": 0.0222976, "grad_norm": 0.698628842830658, "learning_rate": 1.587319343848105e-05, "loss": 3.1397, "step": 117610 }, { "epoch": 0.0223232, "grad_norm": 0.7300692796707153, "learning_rate": 1.5872541663042978e-05, "loss": 2.9343, "step": 117620 }, { "epoch": 0.0223488, "grad_norm": 0.6411874294281006, "learning_rate": 1.5871889849523037e-05, "loss": 3.035, "step": 117630 }, { "epoch": 0.0223744, "grad_norm": 1.0336116552352905, "learning_rate": 1.587123799792545e-05, "loss": 3.2233, "step": 117640 }, { "epoch": 0.0224, "grad_norm": 0.6581589579582214, "learning_rate": 1.5870586108254448e-05, "loss": 3.1575, "step": 117650 }, { "epoch": 0.0224256, "grad_norm": 0.691175103187561, "learning_rate": 1.5869934180514253e-05, "loss": 2.9139, "step": 117660 }, { "epoch": 0.0224512, "grad_norm": 0.7478624582290649, "learning_rate": 1.5869282214709094e-05, "loss": 2.9459, "step": 117670 }, { "epoch": 0.0224768, "grad_norm": 0.7349675297737122, "learning_rate": 1.58686302108432e-05, "loss": 3.17, "step": 117680 }, { "epoch": 0.0225024, "grad_norm": 0.6618598699569702, "learning_rate": 1.58679781689208e-05, "loss": 3.0726, "step": 117690 }, { "epoch": 0.022528, "grad_norm": 0.736844539642334, "learning_rate": 1.586732608894612e-05, "loss": 2.9629, "step": 117700 }, { "epoch": 0.0225536, "grad_norm": 0.7047948241233826, "learning_rate": 1.5866673970923392e-05, "loss": 3.0889, "step": 117710 }, { "epoch": 0.0225792, "grad_norm": 0.839353621006012, "learning_rate": 1.586602181485684e-05, "loss": 3.0729, "step": 117720 }, { "epoch": 0.0226048, "grad_norm": 0.8346564769744873, "learning_rate": 1.5865369620750698e-05, "loss": 3.0964, "step": 117730 }, { "epoch": 0.0226304, "grad_norm": 0.6684945225715637, "learning_rate": 1.5864717388609186e-05, "loss": 2.99, "step": 117740 }, { "epoch": 0.022656, "grad_norm": 0.8392608761787415, "learning_rate": 1.5864065118436548e-05, "loss": 2.9611, "step": 117750 }, { "epoch": 0.0226816, "grad_norm": 0.8604039549827576, "learning_rate": 1.5863412810237003e-05, "loss": 3.0273, "step": 117760 }, { "epoch": 0.0227072, "grad_norm": 0.7896856069564819, "learning_rate": 1.5862760464014786e-05, "loss": 3.0269, "step": 117770 }, { "epoch": 0.0227328, "grad_norm": 0.7008787989616394, "learning_rate": 1.5862108079774126e-05, "loss": 3.01, "step": 117780 }, { "epoch": 0.0227584, "grad_norm": 0.7908503413200378, "learning_rate": 1.586145565751925e-05, "loss": 3.0925, "step": 117790 }, { "epoch": 0.022784, "grad_norm": 0.6831607222557068, "learning_rate": 1.5860803197254394e-05, "loss": 2.9013, "step": 117800 }, { "epoch": 0.0228096, "grad_norm": 0.7116988301277161, "learning_rate": 1.586015069898379e-05, "loss": 3.0095, "step": 117810 }, { "epoch": 0.0228352, "grad_norm": 0.7066211104393005, "learning_rate": 1.5859498162711662e-05, "loss": 3.0393, "step": 117820 }, { "epoch": 0.0228608, "grad_norm": 0.6900517344474792, "learning_rate": 1.5858845588442243e-05, "loss": 3.1731, "step": 117830 }, { "epoch": 0.0228864, "grad_norm": 0.7332448363304138, "learning_rate": 1.585819297617977e-05, "loss": 3.2633, "step": 117840 }, { "epoch": 0.022912, "grad_norm": 0.7843621373176575, "learning_rate": 1.5857540325928474e-05, "loss": 2.992, "step": 117850 }, { "epoch": 0.0229376, "grad_norm": 0.7764554023742676, "learning_rate": 1.5856887637692585e-05, "loss": 2.9626, "step": 117860 }, { "epoch": 0.0229632, "grad_norm": 0.7365886569023132, "learning_rate": 1.5856234911476337e-05, "loss": 3.0929, "step": 117870 }, { "epoch": 0.0229888, "grad_norm": 0.7499637603759766, "learning_rate": 1.585558214728396e-05, "loss": 2.9238, "step": 117880 }, { "epoch": 0.0230144, "grad_norm": 0.7652825713157654, "learning_rate": 1.585492934511969e-05, "loss": 2.9528, "step": 117890 }, { "epoch": 0.02304, "grad_norm": 0.7552051544189453, "learning_rate": 1.585427650498776e-05, "loss": 3.3507, "step": 117900 }, { "epoch": 0.0230656, "grad_norm": 0.8403623104095459, "learning_rate": 1.5853623626892404e-05, "loss": 2.9193, "step": 117910 }, { "epoch": 0.0230912, "grad_norm": 0.7063391804695129, "learning_rate": 1.585297071083785e-05, "loss": 3.0497, "step": 117920 }, { "epoch": 0.0231168, "grad_norm": 0.7172421216964722, "learning_rate": 1.585231775682834e-05, "loss": 2.936, "step": 117930 }, { "epoch": 0.0231424, "grad_norm": 0.8512545228004456, "learning_rate": 1.5851664764868103e-05, "loss": 3.0508, "step": 117940 }, { "epoch": 0.023168, "grad_norm": 0.7742657661437988, "learning_rate": 1.5851011734961376e-05, "loss": 3.0176, "step": 117950 }, { "epoch": 0.0231936, "grad_norm": 0.8908333778381348, "learning_rate": 1.5850358667112394e-05, "loss": 3.134, "step": 117960 }, { "epoch": 0.0232192, "grad_norm": 0.7248685359954834, "learning_rate": 1.584970556132539e-05, "loss": 3.1193, "step": 117970 }, { "epoch": 0.0232448, "grad_norm": 0.7794482111930847, "learning_rate": 1.58490524176046e-05, "loss": 2.9548, "step": 117980 }, { "epoch": 0.0232704, "grad_norm": 0.6642507314682007, "learning_rate": 1.584839923595426e-05, "loss": 2.9922, "step": 117990 }, { "epoch": 0.023296, "grad_norm": 0.7172739505767822, "learning_rate": 1.5847746016378605e-05, "loss": 3.0433, "step": 118000 }, { "epoch": 0.0233216, "grad_norm": 1.008695363998413, "learning_rate": 1.5847092758881867e-05, "loss": 3.3407, "step": 118010 }, { "epoch": 0.0233472, "grad_norm": 2.82110595703125, "learning_rate": 1.584643946346829e-05, "loss": 3.0548, "step": 118020 }, { "epoch": 0.0233728, "grad_norm": 0.7848962545394897, "learning_rate": 1.5845786130142108e-05, "loss": 3.0145, "step": 118030 }, { "epoch": 0.0233984, "grad_norm": 0.8098962306976318, "learning_rate": 1.5845132758907557e-05, "loss": 3.2311, "step": 118040 }, { "epoch": 0.023424, "grad_norm": 0.7583710551261902, "learning_rate": 1.5844479349768872e-05, "loss": 2.9227, "step": 118050 }, { "epoch": 0.0234496, "grad_norm": 0.6458153128623962, "learning_rate": 1.5843825902730294e-05, "loss": 2.9435, "step": 118060 }, { "epoch": 0.0234752, "grad_norm": 0.7550063729286194, "learning_rate": 1.5843172417796055e-05, "loss": 3.0198, "step": 118070 }, { "epoch": 0.0235008, "grad_norm": 0.6649935841560364, "learning_rate": 1.5842518894970396e-05, "loss": 3.0957, "step": 118080 }, { "epoch": 0.0235264, "grad_norm": 0.9314299821853638, "learning_rate": 1.5841865334257557e-05, "loss": 3.1167, "step": 118090 }, { "epoch": 0.023552, "grad_norm": 0.808692455291748, "learning_rate": 1.5841211735661772e-05, "loss": 2.9208, "step": 118100 }, { "epoch": 0.0235776, "grad_norm": 0.7522433996200562, "learning_rate": 1.584055809918728e-05, "loss": 3.0794, "step": 118110 }, { "epoch": 0.0236032, "grad_norm": 0.6914024949073792, "learning_rate": 1.5839904424838324e-05, "loss": 3.0022, "step": 118120 }, { "epoch": 0.0236288, "grad_norm": 0.7809137105941772, "learning_rate": 1.583925071261914e-05, "loss": 3.182, "step": 118130 }, { "epoch": 0.0236544, "grad_norm": 0.7350773215293884, "learning_rate": 1.5838596962533964e-05, "loss": 3.0624, "step": 118140 }, { "epoch": 0.02368, "grad_norm": 0.6366870999336243, "learning_rate": 1.583794317458704e-05, "loss": 2.8676, "step": 118150 }, { "epoch": 0.0237056, "grad_norm": 0.6933838725090027, "learning_rate": 1.5837289348782606e-05, "loss": 3.1414, "step": 118160 }, { "epoch": 0.0237312, "grad_norm": 0.7307800054550171, "learning_rate": 1.5836635485124906e-05, "loss": 3.0179, "step": 118170 }, { "epoch": 0.0237568, "grad_norm": 0.7215752601623535, "learning_rate": 1.583598158361817e-05, "loss": 3.036, "step": 118180 }, { "epoch": 0.0237824, "grad_norm": 0.7433914542198181, "learning_rate": 1.5835327644266647e-05, "loss": 3.0582, "step": 118190 }, { "epoch": 0.023808, "grad_norm": 0.7081514596939087, "learning_rate": 1.5834673667074572e-05, "loss": 2.8667, "step": 118200 }, { "epoch": 0.0238336, "grad_norm": 1.0517771244049072, "learning_rate": 1.5834019652046193e-05, "loss": 3.1289, "step": 118210 }, { "epoch": 0.0238592, "grad_norm": 0.6863741278648376, "learning_rate": 1.5833365599185748e-05, "loss": 3.0816, "step": 118220 }, { "epoch": 0.0238848, "grad_norm": 0.682758092880249, "learning_rate": 1.5832711508497475e-05, "loss": 3.0629, "step": 118230 }, { "epoch": 0.0239104, "grad_norm": 0.9043822884559631, "learning_rate": 1.5832057379985617e-05, "loss": 2.9909, "step": 118240 }, { "epoch": 0.023936, "grad_norm": 0.7333313226699829, "learning_rate": 1.583140321365442e-05, "loss": 3.073, "step": 118250 }, { "epoch": 0.0239616, "grad_norm": 0.9519574642181396, "learning_rate": 1.5830749009508116e-05, "loss": 3.1182, "step": 118260 }, { "epoch": 0.0239872, "grad_norm": 0.7062398791313171, "learning_rate": 1.583009476755096e-05, "loss": 3.1113, "step": 118270 }, { "epoch": 0.0240128, "grad_norm": 0.8102542757987976, "learning_rate": 1.582944048778719e-05, "loss": 3.1302, "step": 118280 }, { "epoch": 0.0240384, "grad_norm": 0.933796226978302, "learning_rate": 1.5828786170221047e-05, "loss": 3.0025, "step": 118290 }, { "epoch": 0.024064, "grad_norm": 0.8818724751472473, "learning_rate": 1.582813181485677e-05, "loss": 2.9833, "step": 118300 }, { "epoch": 0.0240896, "grad_norm": 0.7083673477172852, "learning_rate": 1.5827477421698613e-05, "loss": 3.0112, "step": 118310 }, { "epoch": 0.0241152, "grad_norm": 0.724678635597229, "learning_rate": 1.5826822990750812e-05, "loss": 3.0964, "step": 118320 }, { "epoch": 0.0241408, "grad_norm": 0.7667543292045593, "learning_rate": 1.582616852201761e-05, "loss": 3.1554, "step": 118330 }, { "epoch": 0.0241664, "grad_norm": 0.6971215605735779, "learning_rate": 1.5825514015503255e-05, "loss": 3.216, "step": 118340 }, { "epoch": 0.024192, "grad_norm": 0.778709352016449, "learning_rate": 1.582485947121199e-05, "loss": 2.9949, "step": 118350 }, { "epoch": 0.0242176, "grad_norm": 0.8469071984291077, "learning_rate": 1.582420488914806e-05, "loss": 2.9446, "step": 118360 }, { "epoch": 0.0242432, "grad_norm": 0.831908643245697, "learning_rate": 1.582355026931571e-05, "loss": 3.1384, "step": 118370 }, { "epoch": 0.0242688, "grad_norm": 0.6684411764144897, "learning_rate": 1.5822895611719182e-05, "loss": 2.978, "step": 118380 }, { "epoch": 0.0242944, "grad_norm": 0.8592739105224609, "learning_rate": 1.5822240916362725e-05, "loss": 3.0629, "step": 118390 }, { "epoch": 0.02432, "grad_norm": 0.7872419953346252, "learning_rate": 1.5821586183250583e-05, "loss": 2.942, "step": 118400 }, { "epoch": 0.0243456, "grad_norm": 0.780857503414154, "learning_rate": 1.5820931412387e-05, "loss": 3.0687, "step": 118410 }, { "epoch": 0.0243712, "grad_norm": 0.8583473563194275, "learning_rate": 1.5820276603776228e-05, "loss": 3.1293, "step": 118420 }, { "epoch": 0.0243968, "grad_norm": 0.8788266777992249, "learning_rate": 1.5819621757422504e-05, "loss": 2.9885, "step": 118430 }, { "epoch": 0.0244224, "grad_norm": 1.0847318172454834, "learning_rate": 1.581896687333008e-05, "loss": 3.0203, "step": 118440 }, { "epoch": 0.024448, "grad_norm": 1.0180000066757202, "learning_rate": 1.5818311951503207e-05, "loss": 3.1082, "step": 118450 }, { "epoch": 0.0244736, "grad_norm": 0.7119346857070923, "learning_rate": 1.5817656991946125e-05, "loss": 3.0822, "step": 118460 }, { "epoch": 0.0244992, "grad_norm": 1.0378392934799194, "learning_rate": 1.5817001994663083e-05, "loss": 3.0671, "step": 118470 }, { "epoch": 0.0245248, "grad_norm": 0.8273143172264099, "learning_rate": 1.581634695965833e-05, "loss": 2.9822, "step": 118480 }, { "epoch": 0.0245504, "grad_norm": 1.1104810237884521, "learning_rate": 1.581569188693611e-05, "loss": 3.1675, "step": 118490 }, { "epoch": 0.024576, "grad_norm": 0.7321491837501526, "learning_rate": 1.5815036776500675e-05, "loss": 2.9622, "step": 118500 }, { "epoch": 0.0246016, "grad_norm": 0.8060581088066101, "learning_rate": 1.581438162835627e-05, "loss": 3.1519, "step": 118510 }, { "epoch": 0.0246272, "grad_norm": 0.7303034067153931, "learning_rate": 1.581372644250715e-05, "loss": 3.1385, "step": 118520 }, { "epoch": 0.0246528, "grad_norm": 0.6300415992736816, "learning_rate": 1.581307121895756e-05, "loss": 2.8491, "step": 118530 }, { "epoch": 0.0246784, "grad_norm": 0.7340370416641235, "learning_rate": 1.5812415957711743e-05, "loss": 3.0024, "step": 118540 }, { "epoch": 0.024704, "grad_norm": 0.6903543472290039, "learning_rate": 1.5811760658773956e-05, "loss": 3.0682, "step": 118550 }, { "epoch": 0.0247296, "grad_norm": 0.8683553338050842, "learning_rate": 1.5811105322148445e-05, "loss": 3.0577, "step": 118560 }, { "epoch": 0.0247552, "grad_norm": 0.7734999060630798, "learning_rate": 1.581044994783946e-05, "loss": 3.1449, "step": 118570 }, { "epoch": 0.0247808, "grad_norm": 0.7359703779220581, "learning_rate": 1.5809794535851253e-05, "loss": 2.9353, "step": 118580 }, { "epoch": 0.0248064, "grad_norm": 0.7244194746017456, "learning_rate": 1.5809139086188073e-05, "loss": 2.9161, "step": 118590 }, { "epoch": 0.024832, "grad_norm": 0.7203825116157532, "learning_rate": 1.5808483598854167e-05, "loss": 3.0601, "step": 118600 }, { "epoch": 0.0248576, "grad_norm": 0.650849461555481, "learning_rate": 1.5807828073853787e-05, "loss": 3.0377, "step": 118610 }, { "epoch": 0.0248832, "grad_norm": 0.814752995967865, "learning_rate": 1.580717251119119e-05, "loss": 3.0367, "step": 118620 }, { "epoch": 0.0249088, "grad_norm": 0.7364596724510193, "learning_rate": 1.580651691087062e-05, "loss": 2.9159, "step": 118630 }, { "epoch": 0.0249344, "grad_norm": 2.1547763347625732, "learning_rate": 1.5805861272896334e-05, "loss": 3.005, "step": 118640 }, { "epoch": 0.02496, "grad_norm": 0.8468695282936096, "learning_rate": 1.580520559727258e-05, "loss": 3.2201, "step": 118650 }, { "epoch": 0.0249856, "grad_norm": 0.7240563035011292, "learning_rate": 1.580454988400361e-05, "loss": 3.2587, "step": 118660 }, { "epoch": 0.0250112, "grad_norm": 1.0960958003997803, "learning_rate": 1.5803894133093674e-05, "loss": 3.2612, "step": 118670 }, { "epoch": 0.0250368, "grad_norm": 0.8625494241714478, "learning_rate": 1.5803238344547028e-05, "loss": 2.9362, "step": 118680 }, { "epoch": 0.0250624, "grad_norm": 0.8126355409622192, "learning_rate": 1.5802582518367925e-05, "loss": 2.8523, "step": 118690 }, { "epoch": 0.025088, "grad_norm": 0.7467857003211975, "learning_rate": 1.5801926654560614e-05, "loss": 2.9083, "step": 118700 }, { "epoch": 0.0251136, "grad_norm": 0.7142470479011536, "learning_rate": 1.5801270753129352e-05, "loss": 2.8857, "step": 118710 }, { "epoch": 0.0251392, "grad_norm": 0.728125810623169, "learning_rate": 1.5800614814078393e-05, "loss": 2.9226, "step": 118720 }, { "epoch": 0.0251648, "grad_norm": 0.7458196878433228, "learning_rate": 1.5799958837411988e-05, "loss": 3.2161, "step": 118730 }, { "epoch": 0.0251904, "grad_norm": 0.7269836068153381, "learning_rate": 1.579930282313439e-05, "loss": 2.9684, "step": 118740 }, { "epoch": 0.025216, "grad_norm": 0.778931200504303, "learning_rate": 1.5798646771249855e-05, "loss": 2.9759, "step": 118750 }, { "epoch": 0.0252416, "grad_norm": 0.7002513408660889, "learning_rate": 1.579799068176264e-05, "loss": 2.9716, "step": 118760 }, { "epoch": 0.0252672, "grad_norm": 0.7370846271514893, "learning_rate": 1.579733455467699e-05, "loss": 3.1653, "step": 118770 }, { "epoch": 0.0252928, "grad_norm": 0.7978376746177673, "learning_rate": 1.579667838999717e-05, "loss": 3.17, "step": 118780 }, { "epoch": 0.0253184, "grad_norm": 0.751237154006958, "learning_rate": 1.579602218772743e-05, "loss": 2.9081, "step": 118790 }, { "epoch": 0.025344, "grad_norm": 0.6962341070175171, "learning_rate": 1.5795365947872025e-05, "loss": 2.939, "step": 118800 }, { "epoch": 0.0253696, "grad_norm": 0.7112219333648682, "learning_rate": 1.5794709670435215e-05, "loss": 2.9069, "step": 118810 }, { "epoch": 0.0253952, "grad_norm": 0.871933102607727, "learning_rate": 1.5794053355421252e-05, "loss": 2.8915, "step": 118820 }, { "epoch": 0.0254208, "grad_norm": 0.8577027320861816, "learning_rate": 1.5793397002834395e-05, "loss": 2.9065, "step": 118830 }, { "epoch": 0.0254464, "grad_norm": 0.8319653868675232, "learning_rate": 1.5792740612678895e-05, "loss": 2.9481, "step": 118840 }, { "epoch": 0.025472, "grad_norm": 1.088891625404358, "learning_rate": 1.5792084184959013e-05, "loss": 3.0405, "step": 118850 }, { "epoch": 0.0254976, "grad_norm": 0.7355894446372986, "learning_rate": 1.5791427719679006e-05, "loss": 2.795, "step": 118860 }, { "epoch": 0.0255232, "grad_norm": 0.7026501893997192, "learning_rate": 1.5790771216843123e-05, "loss": 2.9351, "step": 118870 }, { "epoch": 0.0255488, "grad_norm": 0.709409236907959, "learning_rate": 1.579011467645563e-05, "loss": 3.0, "step": 118880 }, { "epoch": 0.0255744, "grad_norm": 0.7522956728935242, "learning_rate": 1.5789458098520788e-05, "loss": 3.2, "step": 118890 }, { "epoch": 0.0256, "grad_norm": 0.7508769035339355, "learning_rate": 1.5788801483042845e-05, "loss": 3.1647, "step": 118900 }, { "epoch": 0.0256256, "grad_norm": 0.8113420605659485, "learning_rate": 1.5788144830026063e-05, "loss": 2.9324, "step": 118910 }, { "epoch": 0.0256512, "grad_norm": 0.7472780346870422, "learning_rate": 1.57874881394747e-05, "loss": 3.1367, "step": 118920 }, { "epoch": 0.0256768, "grad_norm": 0.8324251174926758, "learning_rate": 1.5786831411393014e-05, "loss": 3.2268, "step": 118930 }, { "epoch": 0.0257024, "grad_norm": 0.8000217080116272, "learning_rate": 1.5786174645785265e-05, "loss": 3.1036, "step": 118940 }, { "epoch": 0.025728, "grad_norm": 0.703023374080658, "learning_rate": 1.578551784265571e-05, "loss": 2.9779, "step": 118950 }, { "epoch": 0.0257536, "grad_norm": 1.1690958738327026, "learning_rate": 1.578486100200861e-05, "loss": 3.2879, "step": 118960 }, { "epoch": 0.0257792, "grad_norm": 0.802724301815033, "learning_rate": 1.5784204123848222e-05, "loss": 3.1472, "step": 118970 }, { "epoch": 0.0258048, "grad_norm": 0.6594672799110413, "learning_rate": 1.5783547208178812e-05, "loss": 3.0303, "step": 118980 }, { "epoch": 0.0258304, "grad_norm": 0.810194730758667, "learning_rate": 1.5782890255004636e-05, "loss": 2.9857, "step": 118990 }, { "epoch": 0.025856, "grad_norm": 0.6820563077926636, "learning_rate": 1.578223326432995e-05, "loss": 3.128, "step": 119000 }, { "epoch": 0.0258816, "grad_norm": 0.77767413854599, "learning_rate": 1.578157623615902e-05, "loss": 3.0094, "step": 119010 }, { "epoch": 0.0259072, "grad_norm": 0.7410066723823547, "learning_rate": 1.57809191704961e-05, "loss": 3.0459, "step": 119020 }, { "epoch": 0.0259328, "grad_norm": 0.8804969191551208, "learning_rate": 1.578026206734546e-05, "loss": 2.9989, "step": 119030 }, { "epoch": 0.0259584, "grad_norm": 0.8290525674819946, "learning_rate": 1.5779604926711356e-05, "loss": 3.1209, "step": 119040 }, { "epoch": 0.025984, "grad_norm": 0.7575038075447083, "learning_rate": 1.5778947748598047e-05, "loss": 3.2928, "step": 119050 }, { "epoch": 0.0260096, "grad_norm": 0.9590944051742554, "learning_rate": 1.57782905330098e-05, "loss": 3.0081, "step": 119060 }, { "epoch": 0.0260352, "grad_norm": 0.9505782127380371, "learning_rate": 1.5777633279950877e-05, "loss": 2.9442, "step": 119070 }, { "epoch": 0.0260608, "grad_norm": 0.8237128257751465, "learning_rate": 1.5776975989425535e-05, "loss": 2.9648, "step": 119080 }, { "epoch": 0.0260864, "grad_norm": 0.6554962396621704, "learning_rate": 1.5776318661438043e-05, "loss": 3.119, "step": 119090 }, { "epoch": 0.026112, "grad_norm": 0.7202283143997192, "learning_rate": 1.5775661295992653e-05, "loss": 2.9579, "step": 119100 }, { "epoch": 0.0261376, "grad_norm": 0.9309374094009399, "learning_rate": 1.5775003893093642e-05, "loss": 3.1665, "step": 119110 }, { "epoch": 0.0261632, "grad_norm": 0.7356905341148376, "learning_rate": 1.5774346452745258e-05, "loss": 2.9535, "step": 119120 }, { "epoch": 0.0261888, "grad_norm": 0.7834895849227905, "learning_rate": 1.5773688974951776e-05, "loss": 3.1978, "step": 119130 }, { "epoch": 0.0262144, "grad_norm": 0.7470413446426392, "learning_rate": 1.5773031459717457e-05, "loss": 2.8559, "step": 119140 }, { "epoch": 0.02624, "grad_norm": 0.6731562614440918, "learning_rate": 1.577237390704656e-05, "loss": 2.9977, "step": 119150 }, { "epoch": 0.0262656, "grad_norm": 0.7466083765029907, "learning_rate": 1.5771716316943355e-05, "loss": 3.1049, "step": 119160 }, { "epoch": 0.0262912, "grad_norm": 0.7306852340698242, "learning_rate": 1.5771058689412104e-05, "loss": 3.069, "step": 119170 }, { "epoch": 0.0263168, "grad_norm": 0.8011466264724731, "learning_rate": 1.5770401024457063e-05, "loss": 3.2525, "step": 119180 }, { "epoch": 0.0263424, "grad_norm": 0.8334375619888306, "learning_rate": 1.5769743322082516e-05, "loss": 3.1063, "step": 119190 }, { "epoch": 0.026368, "grad_norm": 0.7533141374588013, "learning_rate": 1.5769085582292708e-05, "loss": 2.9394, "step": 119200 }, { "epoch": 0.0263936, "grad_norm": 0.6948071718215942, "learning_rate": 1.576842780509192e-05, "loss": 2.9735, "step": 119210 }, { "epoch": 0.0264192, "grad_norm": 0.7106432318687439, "learning_rate": 1.5767769990484408e-05, "loss": 2.949, "step": 119220 }, { "epoch": 0.0264448, "grad_norm": 0.7220146656036377, "learning_rate": 1.576711213847444e-05, "loss": 3.0786, "step": 119230 }, { "epoch": 0.0264704, "grad_norm": 0.867019534111023, "learning_rate": 1.5766454249066284e-05, "loss": 3.3207, "step": 119240 }, { "epoch": 0.026496, "grad_norm": 0.7695000171661377, "learning_rate": 1.5765796322264205e-05, "loss": 3.0754, "step": 119250 }, { "epoch": 0.0265216, "grad_norm": 0.7095667123794556, "learning_rate": 1.576513835807247e-05, "loss": 3.0909, "step": 119260 }, { "epoch": 0.0265472, "grad_norm": 0.7987155318260193, "learning_rate": 1.576448035649534e-05, "loss": 3.1363, "step": 119270 }, { "epoch": 0.0265728, "grad_norm": 0.7850795984268188, "learning_rate": 1.576382231753709e-05, "loss": 2.9918, "step": 119280 }, { "epoch": 0.0265984, "grad_norm": 0.8164717555046082, "learning_rate": 1.5763164241201986e-05, "loss": 2.9945, "step": 119290 }, { "epoch": 0.026624, "grad_norm": 0.7555155754089355, "learning_rate": 1.576250612749429e-05, "loss": 3.2446, "step": 119300 }, { "epoch": 0.0266496, "grad_norm": 0.7788746356964111, "learning_rate": 1.5761847976418274e-05, "loss": 3.2401, "step": 119310 }, { "epoch": 0.0266752, "grad_norm": 0.7024664282798767, "learning_rate": 1.5761189787978207e-05, "loss": 3.0319, "step": 119320 }, { "epoch": 0.0267008, "grad_norm": 0.7125629782676697, "learning_rate": 1.5760531562178353e-05, "loss": 3.0961, "step": 119330 }, { "epoch": 0.0267264, "grad_norm": 0.9700383543968201, "learning_rate": 1.5759873299022984e-05, "loss": 3.1285, "step": 119340 }, { "epoch": 0.026752, "grad_norm": 0.6728136539459229, "learning_rate": 1.575921499851637e-05, "loss": 3.0745, "step": 119350 }, { "epoch": 0.0267776, "grad_norm": 0.8423010110855103, "learning_rate": 1.5758556660662773e-05, "loss": 3.0025, "step": 119360 }, { "epoch": 0.0268032, "grad_norm": 0.764691174030304, "learning_rate": 1.575789828546647e-05, "loss": 3.1628, "step": 119370 }, { "epoch": 0.0268288, "grad_norm": 0.8024287819862366, "learning_rate": 1.5757239872931727e-05, "loss": 2.9945, "step": 119380 }, { "epoch": 0.0268544, "grad_norm": 0.8496307134628296, "learning_rate": 1.5756581423062812e-05, "loss": 3.1067, "step": 119390 }, { "epoch": 0.02688, "grad_norm": 0.736853301525116, "learning_rate": 1.5755922935863996e-05, "loss": 3.0098, "step": 119400 }, { "epoch": 0.0269056, "grad_norm": 0.7099543213844299, "learning_rate": 1.5755264411339548e-05, "loss": 3.0371, "step": 119410 }, { "epoch": 0.0269312, "grad_norm": 0.795405924320221, "learning_rate": 1.575460584949374e-05, "loss": 3.2449, "step": 119420 }, { "epoch": 0.0269568, "grad_norm": 0.8692080974578857, "learning_rate": 1.5753947250330847e-05, "loss": 3.0824, "step": 119430 }, { "epoch": 0.0269824, "grad_norm": 0.7394324541091919, "learning_rate": 1.5753288613855132e-05, "loss": 2.9623, "step": 119440 }, { "epoch": 0.027008, "grad_norm": 0.8925713896751404, "learning_rate": 1.5752629940070868e-05, "loss": 3.182, "step": 119450 }, { "epoch": 0.0270336, "grad_norm": 0.7755177021026611, "learning_rate": 1.575197122898233e-05, "loss": 3.2224, "step": 119460 }, { "epoch": 0.0270592, "grad_norm": 1.6688482761383057, "learning_rate": 1.5751312480593788e-05, "loss": 3.2483, "step": 119470 }, { "epoch": 0.0270848, "grad_norm": 0.6614660024642944, "learning_rate": 1.575065369490951e-05, "loss": 3.0553, "step": 119480 }, { "epoch": 0.0271104, "grad_norm": 0.7114155888557434, "learning_rate": 1.5749994871933775e-05, "loss": 3.1871, "step": 119490 }, { "epoch": 0.027136, "grad_norm": 0.8332304954528809, "learning_rate": 1.5749336011670847e-05, "loss": 3.108, "step": 119500 }, { "epoch": 0.0271616, "grad_norm": 0.6647229194641113, "learning_rate": 1.5748677114125005e-05, "loss": 3.0205, "step": 119510 }, { "epoch": 0.0271872, "grad_norm": 0.7057476043701172, "learning_rate": 1.5748018179300522e-05, "loss": 3.0776, "step": 119520 }, { "epoch": 0.0272128, "grad_norm": 0.6819524168968201, "learning_rate": 1.5747359207201666e-05, "loss": 3.0801, "step": 119530 }, { "epoch": 0.0272384, "grad_norm": 0.8096979260444641, "learning_rate": 1.5746700197832713e-05, "loss": 3.1384, "step": 119540 }, { "epoch": 0.027264, "grad_norm": 0.8233697414398193, "learning_rate": 1.574604115119794e-05, "loss": 3.2794, "step": 119550 }, { "epoch": 0.0272896, "grad_norm": 0.7678320407867432, "learning_rate": 1.5745382067301617e-05, "loss": 3.0471, "step": 119560 }, { "epoch": 0.0273152, "grad_norm": 0.7786206603050232, "learning_rate": 1.574472294614801e-05, "loss": 3.2328, "step": 119570 }, { "epoch": 0.0273408, "grad_norm": 0.9189977645874023, "learning_rate": 1.574406378774141e-05, "loss": 3.0133, "step": 119580 }, { "epoch": 0.0273664, "grad_norm": 0.7199508547782898, "learning_rate": 1.574340459208608e-05, "loss": 3.0569, "step": 119590 }, { "epoch": 0.027392, "grad_norm": 0.7097633481025696, "learning_rate": 1.5742745359186297e-05, "loss": 3.0592, "step": 119600 }, { "epoch": 0.0274176, "grad_norm": 0.6757251024246216, "learning_rate": 1.574208608904634e-05, "loss": 3.1336, "step": 119610 }, { "epoch": 0.0274432, "grad_norm": 0.7451558113098145, "learning_rate": 1.5741426781670477e-05, "loss": 3.0297, "step": 119620 }, { "epoch": 0.0274688, "grad_norm": 0.9542856812477112, "learning_rate": 1.5740767437062987e-05, "loss": 3.19, "step": 119630 }, { "epoch": 0.0274944, "grad_norm": 0.8099550604820251, "learning_rate": 1.5740108055228148e-05, "loss": 3.0426, "step": 119640 }, { "epoch": 0.02752, "grad_norm": 0.8503132462501526, "learning_rate": 1.5739448636170233e-05, "loss": 3.2732, "step": 119650 }, { "epoch": 0.0275456, "grad_norm": 0.7561261653900146, "learning_rate": 1.573878917989352e-05, "loss": 2.9896, "step": 119660 }, { "epoch": 0.0275712, "grad_norm": 0.9380553364753723, "learning_rate": 1.5738129686402284e-05, "loss": 3.0702, "step": 119670 }, { "epoch": 0.0275968, "grad_norm": 0.7707901000976562, "learning_rate": 1.57374701557008e-05, "loss": 3.1081, "step": 119680 }, { "epoch": 0.0276224, "grad_norm": 0.7158870100975037, "learning_rate": 1.573681058779335e-05, "loss": 3.167, "step": 119690 }, { "epoch": 0.027648, "grad_norm": 0.6691503524780273, "learning_rate": 1.5736150982684203e-05, "loss": 3.0102, "step": 119700 }, { "epoch": 0.0276736, "grad_norm": 0.7960184216499329, "learning_rate": 1.5735491340377646e-05, "loss": 3.1417, "step": 119710 }, { "epoch": 0.0276992, "grad_norm": 0.6507644057273865, "learning_rate": 1.573483166087795e-05, "loss": 2.9869, "step": 119720 }, { "epoch": 0.0277248, "grad_norm": 0.7810119986534119, "learning_rate": 1.5734171944189398e-05, "loss": 2.9707, "step": 119730 }, { "epoch": 0.0277504, "grad_norm": 0.6907485127449036, "learning_rate": 1.573351219031626e-05, "loss": 3.1308, "step": 119740 }, { "epoch": 0.027776, "grad_norm": 0.7072343826293945, "learning_rate": 1.5732852399262822e-05, "loss": 3.0502, "step": 119750 }, { "epoch": 0.0278016, "grad_norm": 0.7379907369613647, "learning_rate": 1.573219257103336e-05, "loss": 3.0056, "step": 119760 }, { "epoch": 0.0278272, "grad_norm": 0.8227431774139404, "learning_rate": 1.573153270563215e-05, "loss": 3.021, "step": 119770 }, { "epoch": 0.0278528, "grad_norm": 0.9095590114593506, "learning_rate": 1.5730872803063476e-05, "loss": 3.133, "step": 119780 }, { "epoch": 0.0278784, "grad_norm": 0.7731046080589294, "learning_rate": 1.5730212863331616e-05, "loss": 2.97, "step": 119790 }, { "epoch": 0.027904, "grad_norm": 0.7592473030090332, "learning_rate": 1.5729552886440845e-05, "loss": 3.3136, "step": 119800 }, { "epoch": 0.0279296, "grad_norm": 0.7252969741821289, "learning_rate": 1.572889287239545e-05, "loss": 3.0302, "step": 119810 }, { "epoch": 0.0279552, "grad_norm": 0.7376800775527954, "learning_rate": 1.5728232821199707e-05, "loss": 3.0703, "step": 119820 }, { "epoch": 0.0279808, "grad_norm": 0.8491965532302856, "learning_rate": 1.5727572732857895e-05, "loss": 2.933, "step": 119830 }, { "epoch": 0.0280064, "grad_norm": 0.7272160053253174, "learning_rate": 1.5726912607374295e-05, "loss": 3.0036, "step": 119840 }, { "epoch": 0.028032, "grad_norm": 0.6809101700782776, "learning_rate": 1.572625244475319e-05, "loss": 2.9814, "step": 119850 }, { "epoch": 0.0280576, "grad_norm": 0.7643553018569946, "learning_rate": 1.572559224499886e-05, "loss": 2.929, "step": 119860 }, { "epoch": 0.0280832, "grad_norm": 0.9226939082145691, "learning_rate": 1.5724932008115585e-05, "loss": 2.8754, "step": 119870 }, { "epoch": 0.0281088, "grad_norm": 0.7176644802093506, "learning_rate": 1.5724271734107648e-05, "loss": 2.9759, "step": 119880 }, { "epoch": 0.0281344, "grad_norm": 0.7598268389701843, "learning_rate": 1.572361142297933e-05, "loss": 2.9918, "step": 119890 }, { "epoch": 0.02816, "grad_norm": 0.7353162169456482, "learning_rate": 1.5722951074734914e-05, "loss": 2.9358, "step": 119900 }, { "epoch": 0.0281856, "grad_norm": 0.7537452578544617, "learning_rate": 1.5722290689378682e-05, "loss": 2.905, "step": 119910 }, { "epoch": 0.0282112, "grad_norm": 0.8297394514083862, "learning_rate": 1.5721630266914912e-05, "loss": 3.0688, "step": 119920 }, { "epoch": 0.0282368, "grad_norm": 0.75654536485672, "learning_rate": 1.572096980734789e-05, "loss": 3.0006, "step": 119930 }, { "epoch": 0.0282624, "grad_norm": 0.7963055968284607, "learning_rate": 1.5720309310681904e-05, "loss": 3.1938, "step": 119940 }, { "epoch": 0.028288, "grad_norm": 0.685093104839325, "learning_rate": 1.571964877692123e-05, "loss": 3.0456, "step": 119950 }, { "epoch": 0.0283136, "grad_norm": 0.9300757646560669, "learning_rate": 1.5718988206070157e-05, "loss": 2.8233, "step": 119960 }, { "epoch": 0.0283392, "grad_norm": 0.6441997289657593, "learning_rate": 1.571832759813296e-05, "loss": 3.0594, "step": 119970 }, { "epoch": 0.0283648, "grad_norm": 0.7489911317825317, "learning_rate": 1.571766695311393e-05, "loss": 3.0288, "step": 119980 }, { "epoch": 0.0283904, "grad_norm": 0.6782160997390747, "learning_rate": 1.5717006271017355e-05, "loss": 3.1408, "step": 119990 }, { "epoch": 0.028416, "grad_norm": 0.7576929926872253, "learning_rate": 1.571634555184751e-05, "loss": 2.975, "step": 120000 }, { "epoch": 0.0284416, "grad_norm": 0.7917011976242065, "learning_rate": 1.5715684795608678e-05, "loss": 3.056, "step": 120010 }, { "epoch": 0.0284672, "grad_norm": 1.503775715827942, "learning_rate": 1.5715024002305153e-05, "loss": 3.2748, "step": 120020 }, { "epoch": 0.0284928, "grad_norm": 1.0335760116577148, "learning_rate": 1.5714363171941216e-05, "loss": 3.098, "step": 120030 }, { "epoch": 0.0285184, "grad_norm": 0.6882169842720032, "learning_rate": 1.5713702304521152e-05, "loss": 3.1542, "step": 120040 }, { "epoch": 0.028544, "grad_norm": 0.7402321696281433, "learning_rate": 1.5713041400049252e-05, "loss": 2.717, "step": 120050 }, { "epoch": 0.0285696, "grad_norm": 0.7183836698532104, "learning_rate": 1.571238045852979e-05, "loss": 2.8072, "step": 120060 }, { "epoch": 0.0285952, "grad_norm": 0.9958657026290894, "learning_rate": 1.571171947996706e-05, "loss": 3.126, "step": 120070 }, { "epoch": 0.0286208, "grad_norm": 0.7700120806694031, "learning_rate": 1.5711058464365354e-05, "loss": 3.0728, "step": 120080 }, { "epoch": 0.0286464, "grad_norm": 0.7778180241584778, "learning_rate": 1.5710397411728946e-05, "loss": 3.1726, "step": 120090 }, { "epoch": 0.028672, "grad_norm": 0.7850180268287659, "learning_rate": 1.5709736322062128e-05, "loss": 3.0028, "step": 120100 }, { "epoch": 0.0286976, "grad_norm": 0.758768618106842, "learning_rate": 1.570907519536919e-05, "loss": 3.0798, "step": 120110 }, { "epoch": 0.0287232, "grad_norm": 0.7710169553756714, "learning_rate": 1.5708414031654413e-05, "loss": 2.9576, "step": 120120 }, { "epoch": 0.0287488, "grad_norm": 0.8894284963607788, "learning_rate": 1.570775283092209e-05, "loss": 3.0567, "step": 120130 }, { "epoch": 0.0287744, "grad_norm": 0.7636568546295166, "learning_rate": 1.570709159317651e-05, "loss": 3.137, "step": 120140 }, { "epoch": 0.0288, "grad_norm": 0.6921336054801941, "learning_rate": 1.5706430318421953e-05, "loss": 3.2716, "step": 120150 }, { "epoch": 0.0288256, "grad_norm": 0.7232047319412231, "learning_rate": 1.5705769006662714e-05, "loss": 3.0285, "step": 120160 }, { "epoch": 0.0288512, "grad_norm": 0.7295318841934204, "learning_rate": 1.570510765790308e-05, "loss": 3.0251, "step": 120170 }, { "epoch": 0.0288768, "grad_norm": 1.4612895250320435, "learning_rate": 1.5704446272147336e-05, "loss": 3.1093, "step": 120180 }, { "epoch": 0.0289024, "grad_norm": 0.7044032216072083, "learning_rate": 1.5703784849399777e-05, "loss": 3.1284, "step": 120190 }, { "epoch": 0.028928, "grad_norm": 0.8729687929153442, "learning_rate": 1.570312338966469e-05, "loss": 3.2308, "step": 120200 }, { "epoch": 0.0289536, "grad_norm": 0.7068844437599182, "learning_rate": 1.5702461892946362e-05, "loss": 2.9674, "step": 120210 }, { "epoch": 0.0289792, "grad_norm": 1.1867631673812866, "learning_rate": 1.5701800359249085e-05, "loss": 2.9763, "step": 120220 }, { "epoch": 0.0290048, "grad_norm": 0.7800089120864868, "learning_rate": 1.570113878857715e-05, "loss": 2.9788, "step": 120230 }, { "epoch": 0.0290304, "grad_norm": 0.6890209317207336, "learning_rate": 1.570047718093484e-05, "loss": 2.9235, "step": 120240 }, { "epoch": 0.029056, "grad_norm": 1.0047651529312134, "learning_rate": 1.5699815536326452e-05, "loss": 2.7907, "step": 120250 }, { "epoch": 0.0290816, "grad_norm": 0.7096021771430969, "learning_rate": 1.5699153854756277e-05, "loss": 3.1076, "step": 120260 }, { "epoch": 0.0291072, "grad_norm": 0.7098080515861511, "learning_rate": 1.56984921362286e-05, "loss": 3.0828, "step": 120270 }, { "epoch": 0.0291328, "grad_norm": 0.7800062894821167, "learning_rate": 1.569783038074772e-05, "loss": 3.1142, "step": 120280 }, { "epoch": 0.0291584, "grad_norm": 0.6856850981712341, "learning_rate": 1.5697168588317923e-05, "loss": 2.9283, "step": 120290 }, { "epoch": 0.029184, "grad_norm": 0.6818794012069702, "learning_rate": 1.56965067589435e-05, "loss": 2.9977, "step": 120300 }, { "epoch": 0.0292096, "grad_norm": 0.9953253865242004, "learning_rate": 1.569584489262875e-05, "loss": 3.2651, "step": 120310 }, { "epoch": 0.0292352, "grad_norm": 0.8506261706352234, "learning_rate": 1.5695182989377955e-05, "loss": 3.0796, "step": 120320 }, { "epoch": 0.0292608, "grad_norm": 0.7986269593238831, "learning_rate": 1.5694521049195414e-05, "loss": 3.121, "step": 120330 }, { "epoch": 0.0292864, "grad_norm": 0.6744458675384521, "learning_rate": 1.5693859072085418e-05, "loss": 3.1138, "step": 120340 }, { "epoch": 0.029312, "grad_norm": 1.2960529327392578, "learning_rate": 1.5693197058052253e-05, "loss": 3.0691, "step": 120350 }, { "epoch": 0.0293376, "grad_norm": 0.7934141159057617, "learning_rate": 1.5692535007100225e-05, "loss": 3.029, "step": 120360 }, { "epoch": 0.0293632, "grad_norm": 0.8103048205375671, "learning_rate": 1.569187291923362e-05, "loss": 3.1451, "step": 120370 }, { "epoch": 0.0293888, "grad_norm": 0.6816081404685974, "learning_rate": 1.569121079445673e-05, "loss": 3.0388, "step": 120380 }, { "epoch": 0.0294144, "grad_norm": 0.8914951086044312, "learning_rate": 1.569054863277385e-05, "loss": 3.1299, "step": 120390 }, { "epoch": 0.02944, "grad_norm": 0.7088703513145447, "learning_rate": 1.5689886434189274e-05, "loss": 2.9705, "step": 120400 }, { "epoch": 0.0294656, "grad_norm": 0.6800735592842102, "learning_rate": 1.56892241987073e-05, "loss": 3.0425, "step": 120410 }, { "epoch": 0.0294912, "grad_norm": 0.8587555885314941, "learning_rate": 1.5688561926332217e-05, "loss": 2.9379, "step": 120420 }, { "epoch": 0.0295168, "grad_norm": 0.7380599975585938, "learning_rate": 1.568789961706832e-05, "loss": 3.0975, "step": 120430 }, { "epoch": 0.0295424, "grad_norm": 0.7359968423843384, "learning_rate": 1.5687237270919906e-05, "loss": 3.0718, "step": 120440 }, { "epoch": 0.029568, "grad_norm": 0.7147209048271179, "learning_rate": 1.568657488789127e-05, "loss": 3.1308, "step": 120450 }, { "epoch": 0.0295936, "grad_norm": 0.8162083625793457, "learning_rate": 1.568591246798671e-05, "loss": 3.267, "step": 120460 }, { "epoch": 0.0296192, "grad_norm": 0.6928228139877319, "learning_rate": 1.568525001121052e-05, "loss": 2.9373, "step": 120470 }, { "epoch": 0.0296448, "grad_norm": 0.7892473340034485, "learning_rate": 1.568458751756699e-05, "loss": 3.1128, "step": 120480 }, { "epoch": 0.0296704, "grad_norm": 0.7890087366104126, "learning_rate": 1.5683924987060422e-05, "loss": 3.1163, "step": 120490 }, { "epoch": 0.029696, "grad_norm": 0.6935847401618958, "learning_rate": 1.568326241969511e-05, "loss": 3.1937, "step": 120500 }, { "epoch": 0.0297216, "grad_norm": 0.9821865558624268, "learning_rate": 1.5682599815475354e-05, "loss": 2.933, "step": 120510 }, { "epoch": 0.0297472, "grad_norm": 1.4794238805770874, "learning_rate": 1.5681937174405447e-05, "loss": 3.0671, "step": 120520 }, { "epoch": 0.0297728, "grad_norm": 2.613821268081665, "learning_rate": 1.5681274496489687e-05, "loss": 3.0401, "step": 120530 }, { "epoch": 0.0297984, "grad_norm": 0.722828209400177, "learning_rate": 1.5680611781732372e-05, "loss": 3.0396, "step": 120540 }, { "epoch": 0.029824, "grad_norm": 0.8031870126724243, "learning_rate": 1.56799490301378e-05, "loss": 3.2347, "step": 120550 }, { "epoch": 0.0298496, "grad_norm": 0.7143955230712891, "learning_rate": 1.567928624171027e-05, "loss": 3.0821, "step": 120560 }, { "epoch": 0.0298752, "grad_norm": 0.8066424131393433, "learning_rate": 1.5678623416454074e-05, "loss": 3.1353, "step": 120570 }, { "epoch": 0.0299008, "grad_norm": 0.8464199900627136, "learning_rate": 1.5677960554373518e-05, "loss": 2.9233, "step": 120580 }, { "epoch": 0.0299264, "grad_norm": 0.7704408764839172, "learning_rate": 1.5677297655472897e-05, "loss": 3.0806, "step": 120590 }, { "epoch": 0.029952, "grad_norm": 0.8957446217536926, "learning_rate": 1.5676634719756507e-05, "loss": 3.0898, "step": 120600 }, { "epoch": 0.0299776, "grad_norm": 0.7080380320549011, "learning_rate": 1.567597174722865e-05, "loss": 2.9955, "step": 120610 }, { "epoch": 0.0300032, "grad_norm": 1.0980178117752075, "learning_rate": 1.5675308737893623e-05, "loss": 3.0592, "step": 120620 }, { "epoch": 0.0300288, "grad_norm": 1.1542052030563354, "learning_rate": 1.5674645691755732e-05, "loss": 3.0819, "step": 120630 }, { "epoch": 0.0300544, "grad_norm": 0.7761768102645874, "learning_rate": 1.567398260881927e-05, "loss": 3.0729, "step": 120640 }, { "epoch": 0.03008, "grad_norm": 0.9541090726852417, "learning_rate": 1.5673319489088536e-05, "loss": 2.9589, "step": 120650 }, { "epoch": 0.0301056, "grad_norm": 0.7983007431030273, "learning_rate": 1.5672656332567835e-05, "loss": 3.1289, "step": 120660 }, { "epoch": 0.0301312, "grad_norm": 0.7387617826461792, "learning_rate": 1.5671993139261465e-05, "loss": 3.3328, "step": 120670 }, { "epoch": 0.0301568, "grad_norm": 1.3833945989608765, "learning_rate": 1.5671329909173727e-05, "loss": 3.1504, "step": 120680 }, { "epoch": 0.0301824, "grad_norm": 0.6829736828804016, "learning_rate": 1.5670666642308922e-05, "loss": 3.1438, "step": 120690 }, { "epoch": 0.030208, "grad_norm": 0.8751630783081055, "learning_rate": 1.567000333867135e-05, "loss": 2.9587, "step": 120700 }, { "epoch": 0.0302336, "grad_norm": 1.0376230478286743, "learning_rate": 1.5669339998265316e-05, "loss": 3.1489, "step": 120710 }, { "epoch": 0.0302592, "grad_norm": 0.6936051249504089, "learning_rate": 1.5668676621095117e-05, "loss": 3.1088, "step": 120720 }, { "epoch": 0.0302848, "grad_norm": 0.7793996930122375, "learning_rate": 1.5668013207165057e-05, "loss": 3.0338, "step": 120730 }, { "epoch": 0.0303104, "grad_norm": 0.7988933324813843, "learning_rate": 1.566734975647944e-05, "loss": 3.1025, "step": 120740 }, { "epoch": 0.030336, "grad_norm": 0.9657689929008484, "learning_rate": 1.5666686269042562e-05, "loss": 2.9772, "step": 120750 }, { "epoch": 0.0303616, "grad_norm": 0.7293961048126221, "learning_rate": 1.566602274485873e-05, "loss": 3.2537, "step": 120760 }, { "epoch": 0.0303872, "grad_norm": 0.7867388129234314, "learning_rate": 1.5665359183932248e-05, "loss": 3.3089, "step": 120770 }, { "epoch": 0.0304128, "grad_norm": 0.7968007326126099, "learning_rate": 1.5664695586267416e-05, "loss": 3.0572, "step": 120780 }, { "epoch": 0.0304384, "grad_norm": 0.7025959491729736, "learning_rate": 1.5664031951868537e-05, "loss": 3.0694, "step": 120790 }, { "epoch": 0.030464, "grad_norm": 0.7708142399787903, "learning_rate": 1.5663368280739918e-05, "loss": 3.1698, "step": 120800 }, { "epoch": 0.0304896, "grad_norm": 0.6430638432502747, "learning_rate": 1.566270457288586e-05, "loss": 3.0109, "step": 120810 }, { "epoch": 0.0305152, "grad_norm": 0.9019932150840759, "learning_rate": 1.566204082831067e-05, "loss": 3.1502, "step": 120820 }, { "epoch": 0.0305408, "grad_norm": 0.8146420121192932, "learning_rate": 1.566137704701865e-05, "loss": 3.1423, "step": 120830 }, { "epoch": 0.0305664, "grad_norm": 0.8878852725028992, "learning_rate": 1.5660713229014103e-05, "loss": 3.0141, "step": 120840 }, { "epoch": 0.030592, "grad_norm": 0.7236299514770508, "learning_rate": 1.5660049374301335e-05, "loss": 2.9125, "step": 120850 }, { "epoch": 0.0306176, "grad_norm": 0.7891235947608948, "learning_rate": 1.5659385482884654e-05, "loss": 3.1087, "step": 120860 }, { "epoch": 0.0306432, "grad_norm": 0.9205425381660461, "learning_rate": 1.565872155476836e-05, "loss": 3.0687, "step": 120870 }, { "epoch": 0.0306688, "grad_norm": 0.7840664982795715, "learning_rate": 1.5658057589956758e-05, "loss": 3.0566, "step": 120880 }, { "epoch": 0.0306944, "grad_norm": 1.0424964427947998, "learning_rate": 1.565739358845416e-05, "loss": 3.0824, "step": 120890 }, { "epoch": 0.03072, "grad_norm": 0.9188322424888611, "learning_rate": 1.5656729550264866e-05, "loss": 3.1325, "step": 120900 }, { "epoch": 0.0307456, "grad_norm": 2.4265823364257812, "learning_rate": 1.5656065475393184e-05, "loss": 2.9351, "step": 120910 }, { "epoch": 0.0307712, "grad_norm": 0.7103089094161987, "learning_rate": 1.5655401363843425e-05, "loss": 2.9899, "step": 120920 }, { "epoch": 0.0307968, "grad_norm": 1.599554419517517, "learning_rate": 1.5654737215619886e-05, "loss": 3.1746, "step": 120930 }, { "epoch": 0.0308224, "grad_norm": 0.6983889937400818, "learning_rate": 1.5654073030726882e-05, "loss": 3.1117, "step": 120940 }, { "epoch": 0.030848, "grad_norm": 0.6849226951599121, "learning_rate": 1.5653408809168715e-05, "loss": 2.8984, "step": 120950 }, { "epoch": 0.0308736, "grad_norm": 0.8143565654754639, "learning_rate": 1.5652744550949694e-05, "loss": 3.0155, "step": 120960 }, { "epoch": 0.0308992, "grad_norm": 0.7313569784164429, "learning_rate": 1.565208025607413e-05, "loss": 2.9087, "step": 120970 }, { "epoch": 0.0309248, "grad_norm": 0.6629632711410522, "learning_rate": 1.5651415924546326e-05, "loss": 3.1764, "step": 120980 }, { "epoch": 0.0309504, "grad_norm": 0.7124858498573303, "learning_rate": 1.565075155637059e-05, "loss": 2.9998, "step": 120990 }, { "epoch": 0.030976, "grad_norm": 2.6650474071502686, "learning_rate": 1.5650087151551235e-05, "loss": 3.0853, "step": 121000 }, { "epoch": 0.0310016, "grad_norm": 1.4779739379882812, "learning_rate": 1.5649422710092565e-05, "loss": 2.9448, "step": 121010 }, { "epoch": 0.0310272, "grad_norm": 1.213951587677002, "learning_rate": 1.5648758231998892e-05, "loss": 3.2902, "step": 121020 }, { "epoch": 0.0310528, "grad_norm": 0.8565121293067932, "learning_rate": 1.564809371727452e-05, "loss": 3.3619, "step": 121030 }, { "epoch": 0.0310784, "grad_norm": 0.8722606301307678, "learning_rate": 1.564742916592376e-05, "loss": 3.0085, "step": 121040 }, { "epoch": 0.031104, "grad_norm": 0.8558236956596375, "learning_rate": 1.5646764577950924e-05, "loss": 3.5265, "step": 121050 }, { "epoch": 0.0311296, "grad_norm": 0.7780798673629761, "learning_rate": 1.5646099953360325e-05, "loss": 2.966, "step": 121060 }, { "epoch": 0.0311552, "grad_norm": 0.7107998132705688, "learning_rate": 1.5645435292156266e-05, "loss": 2.9691, "step": 121070 }, { "epoch": 0.0311808, "grad_norm": 0.68675696849823, "learning_rate": 1.5644770594343057e-05, "loss": 3.0378, "step": 121080 }, { "epoch": 0.0312064, "grad_norm": 0.9235862493515015, "learning_rate": 1.5644105859925014e-05, "loss": 2.83, "step": 121090 }, { "epoch": 0.031232, "grad_norm": 1.7402667999267578, "learning_rate": 1.5643441088906444e-05, "loss": 2.9574, "step": 121100 }, { "epoch": 0.0312576, "grad_norm": 0.6801356077194214, "learning_rate": 1.564277628129166e-05, "loss": 3.066, "step": 121110 }, { "epoch": 0.0312832, "grad_norm": 0.8070452809333801, "learning_rate": 1.5642111437084966e-05, "loss": 2.9213, "step": 121120 }, { "epoch": 0.0313088, "grad_norm": 0.932449460029602, "learning_rate": 1.5641446556290682e-05, "loss": 3.0384, "step": 121130 }, { "epoch": 0.0313344, "grad_norm": 0.7111425995826721, "learning_rate": 1.5640781638913116e-05, "loss": 3.0526, "step": 121140 }, { "epoch": 0.03136, "grad_norm": 0.7568868398666382, "learning_rate": 1.5640116684956578e-05, "loss": 3.1771, "step": 121150 }, { "epoch": 0.0313856, "grad_norm": 1.4217910766601562, "learning_rate": 1.5639451694425387e-05, "loss": 3.3795, "step": 121160 }, { "epoch": 0.0314112, "grad_norm": 0.7493147253990173, "learning_rate": 1.5638786667323847e-05, "loss": 2.9125, "step": 121170 }, { "epoch": 0.0314368, "grad_norm": 0.7301627993583679, "learning_rate": 1.5638121603656274e-05, "loss": 3.0426, "step": 121180 }, { "epoch": 0.0314624, "grad_norm": 0.7834779024124146, "learning_rate": 1.5637456503426985e-05, "loss": 3.0002, "step": 121190 }, { "epoch": 0.031488, "grad_norm": 0.7453407645225525, "learning_rate": 1.5636791366640286e-05, "loss": 3.2951, "step": 121200 }, { "epoch": 0.0315136, "grad_norm": 0.6722946166992188, "learning_rate": 1.5636126193300493e-05, "loss": 3.0826, "step": 121210 }, { "epoch": 0.0315392, "grad_norm": 0.7028414011001587, "learning_rate": 1.563546098341192e-05, "loss": 3.1255, "step": 121220 }, { "epoch": 0.0315648, "grad_norm": 2.110771894454956, "learning_rate": 1.563479573697888e-05, "loss": 3.3137, "step": 121230 }, { "epoch": 0.0315904, "grad_norm": 0.7658068537712097, "learning_rate": 1.5634130454005684e-05, "loss": 3.0524, "step": 121240 }, { "epoch": 0.031616, "grad_norm": 0.6983315348625183, "learning_rate": 1.5633465134496655e-05, "loss": 3.2189, "step": 121250 }, { "epoch": 0.0316416, "grad_norm": 0.8118183016777039, "learning_rate": 1.56327997784561e-05, "loss": 3.1421, "step": 121260 }, { "epoch": 0.0316672, "grad_norm": 0.9876090288162231, "learning_rate": 1.5632134385888335e-05, "loss": 3.1263, "step": 121270 }, { "epoch": 0.0316928, "grad_norm": 0.7141126394271851, "learning_rate": 1.5631468956797674e-05, "loss": 3.0572, "step": 121280 }, { "epoch": 0.0317184, "grad_norm": 0.7042573690414429, "learning_rate": 1.5630803491188434e-05, "loss": 3.0207, "step": 121290 }, { "epoch": 0.031744, "grad_norm": 0.8212665915489197, "learning_rate": 1.5630137989064932e-05, "loss": 3.0525, "step": 121300 }, { "epoch": 0.0317696, "grad_norm": 0.6959993839263916, "learning_rate": 1.5629472450431478e-05, "loss": 3.1226, "step": 121310 }, { "epoch": 0.0317952, "grad_norm": 1.7793874740600586, "learning_rate": 1.5628806875292394e-05, "loss": 2.9876, "step": 121320 }, { "epoch": 0.0318208, "grad_norm": 1.165132999420166, "learning_rate": 1.5628141263651994e-05, "loss": 2.9202, "step": 121330 }, { "epoch": 0.0318464, "grad_norm": 0.727220892906189, "learning_rate": 1.5627475615514588e-05, "loss": 2.7873, "step": 121340 }, { "epoch": 0.031872, "grad_norm": 0.8037835955619812, "learning_rate": 1.5626809930884505e-05, "loss": 3.1219, "step": 121350 }, { "epoch": 0.0318976, "grad_norm": 0.7138456106185913, "learning_rate": 1.5626144209766052e-05, "loss": 2.9819, "step": 121360 }, { "epoch": 0.0319232, "grad_norm": 0.6851904392242432, "learning_rate": 1.562547845216355e-05, "loss": 3.0331, "step": 121370 }, { "epoch": 0.0319488, "grad_norm": 0.7037876844406128, "learning_rate": 1.5624812658081316e-05, "loss": 3.1644, "step": 121380 }, { "epoch": 0.0319744, "grad_norm": 1.2723044157028198, "learning_rate": 1.5624146827523662e-05, "loss": 3.2233, "step": 121390 }, { "epoch": 0.032, "grad_norm": 1.1061811447143555, "learning_rate": 1.5623480960494914e-05, "loss": 3.1956, "step": 121400 }, { "epoch": 0.0320256, "grad_norm": 0.7610477805137634, "learning_rate": 1.5622815056999388e-05, "loss": 3.1379, "step": 121410 }, { "epoch": 0.0320512, "grad_norm": 0.9592013359069824, "learning_rate": 1.5622149117041398e-05, "loss": 3.0678, "step": 121420 }, { "epoch": 0.0320768, "grad_norm": 0.7995350956916809, "learning_rate": 1.562148314062527e-05, "loss": 2.9406, "step": 121430 }, { "epoch": 0.0321024, "grad_norm": 0.7309368252754211, "learning_rate": 1.562081712775531e-05, "loss": 3.1134, "step": 121440 }, { "epoch": 0.032128, "grad_norm": 0.9651811718940735, "learning_rate": 1.562015107843585e-05, "loss": 3.0302, "step": 121450 }, { "epoch": 0.0321536, "grad_norm": 0.7574543952941895, "learning_rate": 1.5619484992671204e-05, "loss": 2.9328, "step": 121460 }, { "epoch": 0.0321792, "grad_norm": 0.7596895694732666, "learning_rate": 1.5618818870465693e-05, "loss": 3.2524, "step": 121470 }, { "epoch": 0.0322048, "grad_norm": 0.6982290148735046, "learning_rate": 1.561815271182363e-05, "loss": 3.1928, "step": 121480 }, { "epoch": 0.0322304, "grad_norm": 0.8064969182014465, "learning_rate": 1.5617486516749342e-05, "loss": 3.397, "step": 121490 }, { "epoch": 0.032256, "grad_norm": 0.7868456244468689, "learning_rate": 1.5616820285247145e-05, "loss": 2.9724, "step": 121500 }, { "epoch": 0.0322816, "grad_norm": 0.7240344285964966, "learning_rate": 1.5616154017321365e-05, "loss": 3.087, "step": 121510 }, { "epoch": 0.0323072, "grad_norm": 0.854056179523468, "learning_rate": 1.561548771297632e-05, "loss": 3.0789, "step": 121520 }, { "epoch": 0.0323328, "grad_norm": 0.8341773748397827, "learning_rate": 1.5614821372216325e-05, "loss": 3.1254, "step": 121530 }, { "epoch": 0.0323584, "grad_norm": 0.781055212020874, "learning_rate": 1.5614154995045705e-05, "loss": 3.0859, "step": 121540 }, { "epoch": 0.032384, "grad_norm": 0.9288667440414429, "learning_rate": 1.5613488581468785e-05, "loss": 3.0472, "step": 121550 }, { "epoch": 0.0324096, "grad_norm": 0.7729520201683044, "learning_rate": 1.5612822131489882e-05, "loss": 3.0397, "step": 121560 }, { "epoch": 0.0324352, "grad_norm": 0.7371137142181396, "learning_rate": 1.561215564511332e-05, "loss": 2.9992, "step": 121570 }, { "epoch": 0.0324608, "grad_norm": 0.7622582912445068, "learning_rate": 1.561148912234342e-05, "loss": 3.2936, "step": 121580 }, { "epoch": 0.0324864, "grad_norm": 0.6955766081809998, "learning_rate": 1.5610822563184505e-05, "loss": 3.0597, "step": 121590 }, { "epoch": 0.032512, "grad_norm": 0.6858052611351013, "learning_rate": 1.5610155967640897e-05, "loss": 3.1642, "step": 121600 }, { "epoch": 0.0325376, "grad_norm": 0.6569207310676575, "learning_rate": 1.5609489335716918e-05, "loss": 2.953, "step": 121610 }, { "epoch": 0.0325632, "grad_norm": 0.7530611753463745, "learning_rate": 1.560882266741689e-05, "loss": 3.1455, "step": 121620 }, { "epoch": 0.0325888, "grad_norm": 1.0445985794067383, "learning_rate": 1.5608155962745137e-05, "loss": 3.3808, "step": 121630 }, { "epoch": 0.0326144, "grad_norm": 0.8491761684417725, "learning_rate": 1.5607489221705987e-05, "loss": 3.0127, "step": 121640 }, { "epoch": 0.03264, "grad_norm": 0.8167476654052734, "learning_rate": 1.5606822444303758e-05, "loss": 2.8378, "step": 121650 }, { "epoch": 0.0326656, "grad_norm": 0.7469803094863892, "learning_rate": 1.5606155630542774e-05, "loss": 3.1621, "step": 121660 }, { "epoch": 0.0326912, "grad_norm": 0.7662940621376038, "learning_rate": 1.5605488780427364e-05, "loss": 2.9824, "step": 121670 }, { "epoch": 0.0327168, "grad_norm": 0.6653591990470886, "learning_rate": 1.5604821893961846e-05, "loss": 3.1304, "step": 121680 }, { "epoch": 0.0327424, "grad_norm": 0.7711138129234314, "learning_rate": 1.5604154971150548e-05, "loss": 2.9756, "step": 121690 }, { "epoch": 0.032768, "grad_norm": 1.1353206634521484, "learning_rate": 1.5603488011997793e-05, "loss": 3.1721, "step": 121700 }, { "epoch": 0.0327936, "grad_norm": 0.7419408559799194, "learning_rate": 1.560282101650791e-05, "loss": 2.8714, "step": 121710 }, { "epoch": 0.0328192, "grad_norm": 0.7171127200126648, "learning_rate": 1.5602153984685223e-05, "loss": 3.2107, "step": 121720 }, { "epoch": 0.0328448, "grad_norm": 0.7707818150520325, "learning_rate": 1.5601486916534052e-05, "loss": 2.989, "step": 121730 }, { "epoch": 0.0328704, "grad_norm": 3.713974714279175, "learning_rate": 1.5600819812058733e-05, "loss": 3.0817, "step": 121740 }, { "epoch": 0.032896, "grad_norm": 0.6852193474769592, "learning_rate": 1.560015267126358e-05, "loss": 3.1066, "step": 121750 }, { "epoch": 0.0329216, "grad_norm": 0.6970987915992737, "learning_rate": 1.5599485494152927e-05, "loss": 3.0903, "step": 121760 }, { "epoch": 0.0329472, "grad_norm": 0.8371245861053467, "learning_rate": 1.55988182807311e-05, "loss": 3.1367, "step": 121770 }, { "epoch": 0.0329728, "grad_norm": 0.7963036894798279, "learning_rate": 1.5598151031002424e-05, "loss": 3.3231, "step": 121780 }, { "epoch": 0.0329984, "grad_norm": 0.718147873878479, "learning_rate": 1.5597483744971228e-05, "loss": 3.2964, "step": 121790 }, { "epoch": 0.033024, "grad_norm": 0.6851111054420471, "learning_rate": 1.5596816422641834e-05, "loss": 2.9036, "step": 121800 }, { "epoch": 0.0330496, "grad_norm": 0.8207707405090332, "learning_rate": 1.5596149064018576e-05, "loss": 3.0398, "step": 121810 }, { "epoch": 0.0330752, "grad_norm": 0.8243818283081055, "learning_rate": 1.5595481669105777e-05, "loss": 3.3338, "step": 121820 }, { "epoch": 0.0331008, "grad_norm": 0.6251834630966187, "learning_rate": 1.5594814237907764e-05, "loss": 2.9102, "step": 121830 }, { "epoch": 0.0331264, "grad_norm": 0.6864221096038818, "learning_rate": 1.559414677042887e-05, "loss": 3.1622, "step": 121840 }, { "epoch": 0.033152, "grad_norm": 0.8381693363189697, "learning_rate": 1.5593479266673424e-05, "loss": 3.2397, "step": 121850 }, { "epoch": 0.0331776, "grad_norm": 0.8025251626968384, "learning_rate": 1.5592811726645748e-05, "loss": 2.98, "step": 121860 }, { "epoch": 0.0332032, "grad_norm": 0.8819319009780884, "learning_rate": 1.5592144150350177e-05, "loss": 3.1131, "step": 121870 }, { "epoch": 0.0332288, "grad_norm": 0.7219656705856323, "learning_rate": 1.5591476537791036e-05, "loss": 3.148, "step": 121880 }, { "epoch": 0.0332544, "grad_norm": 0.7484517097473145, "learning_rate": 1.5590808888972654e-05, "loss": 3.093, "step": 121890 }, { "epoch": 0.03328, "grad_norm": 0.8516309261322021, "learning_rate": 1.5590141203899364e-05, "loss": 2.9823, "step": 121900 }, { "epoch": 0.0333056, "grad_norm": 0.783621072769165, "learning_rate": 1.5589473482575493e-05, "loss": 2.8632, "step": 121910 }, { "epoch": 0.0333312, "grad_norm": 0.730112612247467, "learning_rate": 1.5588805725005372e-05, "loss": 3.0583, "step": 121920 }, { "epoch": 0.0333568, "grad_norm": 0.8809746503829956, "learning_rate": 1.5588137931193333e-05, "loss": 3.1165, "step": 121930 }, { "epoch": 0.0333824, "grad_norm": 0.7940579652786255, "learning_rate": 1.5587470101143703e-05, "loss": 3.2167, "step": 121940 }, { "epoch": 0.033408, "grad_norm": 0.8252488970756531, "learning_rate": 1.5586802234860817e-05, "loss": 3.2312, "step": 121950 }, { "epoch": 0.0334336, "grad_norm": 0.7219293713569641, "learning_rate": 1.5586134332349e-05, "loss": 2.8951, "step": 121960 }, { "epoch": 0.0334592, "grad_norm": 0.9044380187988281, "learning_rate": 1.5585466393612584e-05, "loss": 3.1278, "step": 121970 }, { "epoch": 0.0334848, "grad_norm": 0.9480823278427124, "learning_rate": 1.558479841865591e-05, "loss": 3.2434, "step": 121980 }, { "epoch": 0.0335104, "grad_norm": 0.7616328597068787, "learning_rate": 1.5584130407483296e-05, "loss": 3.0682, "step": 121990 }, { "epoch": 0.033536, "grad_norm": 1.0164105892181396, "learning_rate": 1.5583462360099083e-05, "loss": 3.0713, "step": 122000 }, { "epoch": 0.0335616, "grad_norm": 0.741201639175415, "learning_rate": 1.55827942765076e-05, "loss": 3.0201, "step": 122010 }, { "epoch": 0.0335872, "grad_norm": 0.7492563724517822, "learning_rate": 1.5582126156713183e-05, "loss": 3.0907, "step": 122020 }, { "epoch": 0.0336128, "grad_norm": 0.7651773691177368, "learning_rate": 1.558145800072016e-05, "loss": 2.9073, "step": 122030 }, { "epoch": 0.0336384, "grad_norm": 0.6340529918670654, "learning_rate": 1.5580789808532862e-05, "loss": 3.1256, "step": 122040 }, { "epoch": 0.033664, "grad_norm": 0.6350452899932861, "learning_rate": 1.5580121580155628e-05, "loss": 3.0352, "step": 122050 }, { "epoch": 0.0336896, "grad_norm": 0.7123602628707886, "learning_rate": 1.557945331559279e-05, "loss": 3.0589, "step": 122060 }, { "epoch": 0.0337152, "grad_norm": 0.743506669998169, "learning_rate": 1.5578785014848677e-05, "loss": 3.1357, "step": 122070 }, { "epoch": 0.0337408, "grad_norm": 0.7024487257003784, "learning_rate": 1.5578116677927627e-05, "loss": 3.0116, "step": 122080 }, { "epoch": 0.0337664, "grad_norm": 0.9231582283973694, "learning_rate": 1.5577448304833973e-05, "loss": 3.1872, "step": 122090 }, { "epoch": 0.033792, "grad_norm": 0.7546108961105347, "learning_rate": 1.5576779895572047e-05, "loss": 3.0345, "step": 122100 }, { "epoch": 0.0338176, "grad_norm": 1.1506457328796387, "learning_rate": 1.5576111450146188e-05, "loss": 3.0181, "step": 122110 }, { "epoch": 0.0338432, "grad_norm": 0.6461163759231567, "learning_rate": 1.5575442968560726e-05, "loss": 3.1651, "step": 122120 }, { "epoch": 0.0338688, "grad_norm": 1.196387767791748, "learning_rate": 1.5574774450820002e-05, "loss": 3.0407, "step": 122130 }, { "epoch": 0.0338944, "grad_norm": 0.6569114327430725, "learning_rate": 1.5574105896928345e-05, "loss": 2.889, "step": 122140 }, { "epoch": 0.03392, "grad_norm": 0.7581915259361267, "learning_rate": 1.5573437306890093e-05, "loss": 2.9663, "step": 122150 }, { "epoch": 0.0339456, "grad_norm": 1.5849604606628418, "learning_rate": 1.557276868070958e-05, "loss": 3.2685, "step": 122160 }, { "epoch": 0.0339712, "grad_norm": 0.7348315715789795, "learning_rate": 1.5572100018391148e-05, "loss": 3.2476, "step": 122170 }, { "epoch": 0.0339968, "grad_norm": 0.6852918267250061, "learning_rate": 1.5571431319939123e-05, "loss": 3.0315, "step": 122180 }, { "epoch": 0.0340224, "grad_norm": 1.3447078466415405, "learning_rate": 1.557076258535785e-05, "loss": 3.2375, "step": 122190 }, { "epoch": 0.034048, "grad_norm": 0.7020588517189026, "learning_rate": 1.5570093814651656e-05, "loss": 2.8624, "step": 122200 }, { "epoch": 0.0340736, "grad_norm": 0.7573279738426208, "learning_rate": 1.556942500782489e-05, "loss": 2.9589, "step": 122210 }, { "epoch": 0.0340992, "grad_norm": 0.752347469329834, "learning_rate": 1.556875616488188e-05, "loss": 3.1249, "step": 122220 }, { "epoch": 0.0341248, "grad_norm": 0.6889175176620483, "learning_rate": 1.556808728582697e-05, "loss": 3.06, "step": 122230 }, { "epoch": 0.0341504, "grad_norm": 0.7959426045417786, "learning_rate": 1.556741837066449e-05, "loss": 2.9992, "step": 122240 }, { "epoch": 0.034176, "grad_norm": 0.6822579503059387, "learning_rate": 1.5566749419398784e-05, "loss": 3.0798, "step": 122250 }, { "epoch": 0.0342016, "grad_norm": 0.8666003942489624, "learning_rate": 1.5566080432034185e-05, "loss": 2.9742, "step": 122260 }, { "epoch": 0.0342272, "grad_norm": 0.7197795510292053, "learning_rate": 1.5565411408575035e-05, "loss": 3.0869, "step": 122270 }, { "epoch": 0.0342528, "grad_norm": 1.169396162033081, "learning_rate": 1.556474234902567e-05, "loss": 3.0132, "step": 122280 }, { "epoch": 0.0342784, "grad_norm": 1.7296066284179688, "learning_rate": 1.556407325339043e-05, "loss": 2.7684, "step": 122290 }, { "epoch": 0.034304, "grad_norm": 0.7953703999519348, "learning_rate": 1.556340412167366e-05, "loss": 3.0469, "step": 122300 }, { "epoch": 0.0343296, "grad_norm": 0.6764185428619385, "learning_rate": 1.5562734953879685e-05, "loss": 2.862, "step": 122310 }, { "epoch": 0.0343552, "grad_norm": 0.6956014037132263, "learning_rate": 1.5562065750012854e-05, "loss": 3.2216, "step": 122320 }, { "epoch": 0.0343808, "grad_norm": 0.7475494146347046, "learning_rate": 1.5561396510077504e-05, "loss": 3.2031, "step": 122330 }, { "epoch": 0.0344064, "grad_norm": 0.6687173247337341, "learning_rate": 1.556072723407798e-05, "loss": 3.0791, "step": 122340 }, { "epoch": 0.034432, "grad_norm": 0.7300527691841125, "learning_rate": 1.5560057922018615e-05, "loss": 3.2919, "step": 122350 }, { "epoch": 0.0344576, "grad_norm": 0.68265700340271, "learning_rate": 1.555938857390375e-05, "loss": 3.0624, "step": 122360 }, { "epoch": 0.0344832, "grad_norm": 0.7660710215568542, "learning_rate": 1.555871918973773e-05, "loss": 3.1027, "step": 122370 }, { "epoch": 0.0345088, "grad_norm": 0.7560785412788391, "learning_rate": 1.5558049769524896e-05, "loss": 2.9368, "step": 122380 }, { "epoch": 0.0345344, "grad_norm": 0.699838399887085, "learning_rate": 1.555738031326958e-05, "loss": 3.1524, "step": 122390 }, { "epoch": 0.03456, "grad_norm": 0.7468166351318359, "learning_rate": 1.5556710820976134e-05, "loss": 3.2794, "step": 122400 }, { "epoch": 0.0345856, "grad_norm": 0.6776031851768494, "learning_rate": 1.5556041292648894e-05, "loss": 2.9675, "step": 122410 }, { "epoch": 0.0346112, "grad_norm": 0.7396749258041382, "learning_rate": 1.5555371728292204e-05, "loss": 3.182, "step": 122420 }, { "epoch": 0.0346368, "grad_norm": 0.6910127401351929, "learning_rate": 1.5554702127910402e-05, "loss": 3.0064, "step": 122430 }, { "epoch": 0.0346624, "grad_norm": 0.8751016855239868, "learning_rate": 1.5554032491507833e-05, "loss": 2.8728, "step": 122440 }, { "epoch": 0.034688, "grad_norm": 0.7076098918914795, "learning_rate": 1.555336281908884e-05, "loss": 2.9309, "step": 122450 }, { "epoch": 0.0347136, "grad_norm": 0.8158003091812134, "learning_rate": 1.555269311065777e-05, "loss": 3.1272, "step": 122460 }, { "epoch": 0.0347392, "grad_norm": 1.053727149963379, "learning_rate": 1.5552023366218956e-05, "loss": 3.1442, "step": 122470 }, { "epoch": 0.0347648, "grad_norm": 0.681883692741394, "learning_rate": 1.5551353585776747e-05, "loss": 2.9469, "step": 122480 }, { "epoch": 0.0347904, "grad_norm": 0.6708886027336121, "learning_rate": 1.5550683769335483e-05, "loss": 3.1145, "step": 122490 }, { "epoch": 0.034816, "grad_norm": 0.8005593419075012, "learning_rate": 1.5550013916899515e-05, "loss": 3.0016, "step": 122500 }, { "epoch": 0.0348416, "grad_norm": 0.8409804105758667, "learning_rate": 1.554934402847318e-05, "loss": 3.2058, "step": 122510 }, { "epoch": 0.0348672, "grad_norm": 0.7868899703025818, "learning_rate": 1.5548674104060825e-05, "loss": 2.9396, "step": 122520 }, { "epoch": 0.0348928, "grad_norm": 0.8632394075393677, "learning_rate": 1.5548004143666785e-05, "loss": 3.1272, "step": 122530 }, { "epoch": 0.0349184, "grad_norm": 0.6862985491752625, "learning_rate": 1.554733414729542e-05, "loss": 3.1077, "step": 122540 }, { "epoch": 0.034944, "grad_norm": 0.7323859333992004, "learning_rate": 1.5546664114951068e-05, "loss": 3.1713, "step": 122550 }, { "epoch": 0.0349696, "grad_norm": 0.7796506285667419, "learning_rate": 1.554599404663807e-05, "loss": 3.4655, "step": 122560 }, { "epoch": 0.0349952, "grad_norm": 0.7209028601646423, "learning_rate": 1.5545323942360777e-05, "loss": 3.0448, "step": 122570 }, { "epoch": 0.0350208, "grad_norm": 0.8492228984832764, "learning_rate": 1.5544653802123533e-05, "loss": 3.0865, "step": 122580 }, { "epoch": 0.0350464, "grad_norm": 0.9310436844825745, "learning_rate": 1.554398362593068e-05, "loss": 3.0763, "step": 122590 }, { "epoch": 0.035072, "grad_norm": 0.8239201903343201, "learning_rate": 1.5543313413786565e-05, "loss": 3.1463, "step": 122600 }, { "epoch": 0.0350976, "grad_norm": 0.8493404388427734, "learning_rate": 1.554264316569554e-05, "loss": 2.9113, "step": 122610 }, { "epoch": 0.0351232, "grad_norm": 1.064174771308899, "learning_rate": 1.5541972881661942e-05, "loss": 3.1201, "step": 122620 }, { "epoch": 0.0351488, "grad_norm": 0.7497137784957886, "learning_rate": 1.5541302561690124e-05, "loss": 3.0649, "step": 122630 }, { "epoch": 0.0351744, "grad_norm": 0.6913137435913086, "learning_rate": 1.5540632205784434e-05, "loss": 3.2322, "step": 122640 }, { "epoch": 0.0352, "grad_norm": 0.730289101600647, "learning_rate": 1.553996181394922e-05, "loss": 3.1175, "step": 122650 }, { "epoch": 0.0352256, "grad_norm": 1.2249583005905151, "learning_rate": 1.553929138618882e-05, "loss": 3.0722, "step": 122660 }, { "epoch": 0.0352512, "grad_norm": 0.7132927775382996, "learning_rate": 1.553862092250759e-05, "loss": 3.2951, "step": 122670 }, { "epoch": 0.0352768, "grad_norm": 0.6373891234397888, "learning_rate": 1.5537950422909875e-05, "loss": 3.0137, "step": 122680 }, { "epoch": 0.0353024, "grad_norm": 0.7490044832229614, "learning_rate": 1.5537279887400024e-05, "loss": 2.9374, "step": 122690 }, { "epoch": 0.035328, "grad_norm": 0.6667441129684448, "learning_rate": 1.553660931598238e-05, "loss": 3.0516, "step": 122700 }, { "epoch": 0.0353536, "grad_norm": 1.272605299949646, "learning_rate": 1.55359387086613e-05, "loss": 3.0254, "step": 122710 }, { "epoch": 0.0353792, "grad_norm": 0.651851236820221, "learning_rate": 1.553526806544113e-05, "loss": 2.9509, "step": 122720 }, { "epoch": 0.0354048, "grad_norm": 0.8709374666213989, "learning_rate": 1.553459738632622e-05, "loss": 3.0515, "step": 122730 }, { "epoch": 0.0354304, "grad_norm": 1.4390569925308228, "learning_rate": 1.553392667132091e-05, "loss": 3.0374, "step": 122740 }, { "epoch": 0.035456, "grad_norm": 0.7029870748519897, "learning_rate": 1.553325592042956e-05, "loss": 2.9368, "step": 122750 }, { "epoch": 0.0354816, "grad_norm": 0.8052093386650085, "learning_rate": 1.5532585133656514e-05, "loss": 3.05, "step": 122760 }, { "epoch": 0.0355072, "grad_norm": 0.7017725706100464, "learning_rate": 1.5531914311006125e-05, "loss": 2.8155, "step": 122770 }, { "epoch": 0.0355328, "grad_norm": 0.9254910349845886, "learning_rate": 1.5531243452482746e-05, "loss": 3.0997, "step": 122780 }, { "epoch": 0.0355584, "grad_norm": 0.7819593548774719, "learning_rate": 1.5530572558090717e-05, "loss": 3.0071, "step": 122790 }, { "epoch": 0.035584, "grad_norm": 0.8062415719032288, "learning_rate": 1.5529901627834395e-05, "loss": 2.9566, "step": 122800 }, { "epoch": 0.0356096, "grad_norm": 0.8308310508728027, "learning_rate": 1.5529230661718133e-05, "loss": 3.1943, "step": 122810 }, { "epoch": 0.0356352, "grad_norm": 0.8289026618003845, "learning_rate": 1.5528559659746282e-05, "loss": 3.0948, "step": 122820 }, { "epoch": 0.0356608, "grad_norm": 0.9194530248641968, "learning_rate": 1.5527888621923186e-05, "loss": 3.1428, "step": 122830 }, { "epoch": 0.0356864, "grad_norm": 0.9106895327568054, "learning_rate": 1.5527217548253208e-05, "loss": 3.0748, "step": 122840 }, { "epoch": 0.035712, "grad_norm": 0.7053783535957336, "learning_rate": 1.5526546438740685e-05, "loss": 2.9397, "step": 122850 }, { "epoch": 0.0357376, "grad_norm": 0.9635990262031555, "learning_rate": 1.552587529338998e-05, "loss": 3.0546, "step": 122860 }, { "epoch": 0.0357632, "grad_norm": 0.6589085459709167, "learning_rate": 1.5525204112205444e-05, "loss": 3.0727, "step": 122870 }, { "epoch": 0.0357888, "grad_norm": 0.909595787525177, "learning_rate": 1.552453289519143e-05, "loss": 3.0307, "step": 122880 }, { "epoch": 0.0358144, "grad_norm": 0.746749758720398, "learning_rate": 1.5523861642352283e-05, "loss": 2.9664, "step": 122890 }, { "epoch": 0.03584, "grad_norm": 0.8289572596549988, "learning_rate": 1.5523190353692363e-05, "loss": 3.1569, "step": 122900 }, { "epoch": 0.0358656, "grad_norm": 0.6996680498123169, "learning_rate": 1.5522519029216024e-05, "loss": 3.1346, "step": 122910 }, { "epoch": 0.0358912, "grad_norm": 1.1458982229232788, "learning_rate": 1.5521847668927614e-05, "loss": 2.9891, "step": 122920 }, { "epoch": 0.0359168, "grad_norm": 0.6511052846908569, "learning_rate": 1.552117627283149e-05, "loss": 3.1694, "step": 122930 }, { "epoch": 0.0359424, "grad_norm": 0.7850854396820068, "learning_rate": 1.5520504840932005e-05, "loss": 3.0099, "step": 122940 }, { "epoch": 0.035968, "grad_norm": 0.7798933982849121, "learning_rate": 1.5519833373233515e-05, "loss": 3.2022, "step": 122950 }, { "epoch": 0.0359936, "grad_norm": 0.6909790635108948, "learning_rate": 1.551916186974037e-05, "loss": 3.0775, "step": 122960 }, { "epoch": 0.0360192, "grad_norm": 0.7553197145462036, "learning_rate": 1.551849033045693e-05, "loss": 3.2665, "step": 122970 }, { "epoch": 0.0360448, "grad_norm": 0.6853986382484436, "learning_rate": 1.5517818755387545e-05, "loss": 3.0446, "step": 122980 }, { "epoch": 0.0360704, "grad_norm": 0.7640613913536072, "learning_rate": 1.5517147144536573e-05, "loss": 3.0792, "step": 122990 }, { "epoch": 0.036096, "grad_norm": 0.7755226492881775, "learning_rate": 1.5516475497908367e-05, "loss": 3.1204, "step": 123000 }, { "epoch": 0.0361216, "grad_norm": 0.8294265270233154, "learning_rate": 1.5515803815507285e-05, "loss": 3.0698, "step": 123010 }, { "epoch": 0.0361472, "grad_norm": 2.75762939453125, "learning_rate": 1.5515132097337676e-05, "loss": 3.1984, "step": 123020 }, { "epoch": 0.0361728, "grad_norm": 0.8493973016738892, "learning_rate": 1.551446034340391e-05, "loss": 3.0756, "step": 123030 }, { "epoch": 0.0361984, "grad_norm": 0.7462238669395447, "learning_rate": 1.551378855371033e-05, "loss": 3.148, "step": 123040 }, { "epoch": 0.036224, "grad_norm": 1.0224902629852295, "learning_rate": 1.5513116728261292e-05, "loss": 3.0569, "step": 123050 }, { "epoch": 0.0362496, "grad_norm": 0.7820832133293152, "learning_rate": 1.551244486706116e-05, "loss": 3.1659, "step": 123060 }, { "epoch": 0.0362752, "grad_norm": 0.8225603103637695, "learning_rate": 1.551177297011429e-05, "loss": 2.9572, "step": 123070 }, { "epoch": 0.0363008, "grad_norm": 0.7363958358764648, "learning_rate": 1.5511101037425037e-05, "loss": 3.0033, "step": 123080 }, { "epoch": 0.0363264, "grad_norm": 1.0035334825515747, "learning_rate": 1.5510429068997758e-05, "loss": 3.3671, "step": 123090 }, { "epoch": 0.036352, "grad_norm": 0.691100001335144, "learning_rate": 1.5509757064836812e-05, "loss": 3.4545, "step": 123100 }, { "epoch": 0.0363776, "grad_norm": 3.353126287460327, "learning_rate": 1.5509085024946556e-05, "loss": 3.2213, "step": 123110 }, { "epoch": 0.0364032, "grad_norm": 0.726995050907135, "learning_rate": 1.5508412949331347e-05, "loss": 3.0399, "step": 123120 }, { "epoch": 0.0364288, "grad_norm": 1.0185152292251587, "learning_rate": 1.5507740837995543e-05, "loss": 3.2938, "step": 123130 }, { "epoch": 0.0364544, "grad_norm": 0.909061849117279, "learning_rate": 1.55070686909435e-05, "loss": 3.1046, "step": 123140 }, { "epoch": 0.03648, "grad_norm": 0.7187822461128235, "learning_rate": 1.5506396508179585e-05, "loss": 3.0626, "step": 123150 }, { "epoch": 0.0365056, "grad_norm": 1.05543851852417, "learning_rate": 1.5505724289708153e-05, "loss": 3.0036, "step": 123160 }, { "epoch": 0.0365312, "grad_norm": 0.6838300824165344, "learning_rate": 1.5505052035533558e-05, "loss": 3.119, "step": 123170 }, { "epoch": 0.0365568, "grad_norm": 0.9262920022010803, "learning_rate": 1.5504379745660165e-05, "loss": 3.1373, "step": 123180 }, { "epoch": 0.0365824, "grad_norm": 0.7807247042655945, "learning_rate": 1.5503707420092334e-05, "loss": 3.0808, "step": 123190 }, { "epoch": 0.036608, "grad_norm": 0.7394962906837463, "learning_rate": 1.5503035058834424e-05, "loss": 3.2106, "step": 123200 }, { "epoch": 0.0366336, "grad_norm": 0.723793625831604, "learning_rate": 1.550236266189079e-05, "loss": 3.1506, "step": 123210 }, { "epoch": 0.0366592, "grad_norm": 0.8038948178291321, "learning_rate": 1.5501690229265798e-05, "loss": 3.2556, "step": 123220 }, { "epoch": 0.0366848, "grad_norm": 0.7649756669998169, "learning_rate": 1.5501017760963808e-05, "loss": 3.0594, "step": 123230 }, { "epoch": 0.0367104, "grad_norm": 0.8047980666160583, "learning_rate": 1.550034525698918e-05, "loss": 3.2338, "step": 123240 }, { "epoch": 0.036736, "grad_norm": 1.0867441892623901, "learning_rate": 1.5499672717346274e-05, "loss": 3.3182, "step": 123250 }, { "epoch": 0.0367616, "grad_norm": 0.9627471566200256, "learning_rate": 1.549900014203945e-05, "loss": 3.1499, "step": 123260 }, { "epoch": 0.0367872, "grad_norm": 12.31390380859375, "learning_rate": 1.5498327531073077e-05, "loss": 3.4089, "step": 123270 }, { "epoch": 0.0368128, "grad_norm": 0.7336745262145996, "learning_rate": 1.549765488445151e-05, "loss": 3.1624, "step": 123280 }, { "epoch": 0.0368384, "grad_norm": 0.7938362956047058, "learning_rate": 1.549698220217911e-05, "loss": 3.1677, "step": 123290 }, { "epoch": 0.036864, "grad_norm": 0.8301486968994141, "learning_rate": 1.5496309484260243e-05, "loss": 2.9094, "step": 123300 }, { "epoch": 0.0368896, "grad_norm": 0.6552810072898865, "learning_rate": 1.5495636730699267e-05, "loss": 3.1151, "step": 123310 }, { "epoch": 0.0369152, "grad_norm": 0.9254693984985352, "learning_rate": 1.549496394150055e-05, "loss": 3.0803, "step": 123320 }, { "epoch": 0.0369408, "grad_norm": 0.7458796501159668, "learning_rate": 1.549429111666845e-05, "loss": 2.71, "step": 123330 }, { "epoch": 0.0369664, "grad_norm": 0.8172239661216736, "learning_rate": 1.5493618256207334e-05, "loss": 3.162, "step": 123340 }, { "epoch": 0.036992, "grad_norm": 0.730607271194458, "learning_rate": 1.549294536012156e-05, "loss": 3.1319, "step": 123350 }, { "epoch": 0.0370176, "grad_norm": 0.720620334148407, "learning_rate": 1.5492272428415503e-05, "loss": 3.0682, "step": 123360 }, { "epoch": 0.0370432, "grad_norm": 0.7844652533531189, "learning_rate": 1.5491599461093515e-05, "loss": 3.1733, "step": 123370 }, { "epoch": 0.0370688, "grad_norm": 0.6175827383995056, "learning_rate": 1.5490926458159964e-05, "loss": 2.9951, "step": 123380 }, { "epoch": 0.0370944, "grad_norm": 0.7082812190055847, "learning_rate": 1.549025341961921e-05, "loss": 2.9457, "step": 123390 }, { "epoch": 0.03712, "grad_norm": 0.7392808794975281, "learning_rate": 1.5489580345475625e-05, "loss": 2.9737, "step": 123400 }, { "epoch": 0.0371456, "grad_norm": 0.9962665438652039, "learning_rate": 1.548890723573357e-05, "loss": 3.2358, "step": 123410 }, { "epoch": 0.0371712, "grad_norm": 0.7317200303077698, "learning_rate": 1.5488234090397414e-05, "loss": 3.101, "step": 123420 }, { "epoch": 0.0371968, "grad_norm": 0.7693435549736023, "learning_rate": 1.5487560909471515e-05, "loss": 3.1153, "step": 123430 }, { "epoch": 0.0372224, "grad_norm": 0.9417634606361389, "learning_rate": 1.5486887692960243e-05, "loss": 3.0945, "step": 123440 }, { "epoch": 0.037248, "grad_norm": 0.771549642086029, "learning_rate": 1.5486214440867963e-05, "loss": 2.9288, "step": 123450 }, { "epoch": 0.0372736, "grad_norm": 0.697544276714325, "learning_rate": 1.5485541153199036e-05, "loss": 2.9642, "step": 123460 }, { "epoch": 0.0372992, "grad_norm": 0.7107551097869873, "learning_rate": 1.5484867829957838e-05, "loss": 3.0579, "step": 123470 }, { "epoch": 0.0373248, "grad_norm": 0.7287944555282593, "learning_rate": 1.5484194471148728e-05, "loss": 3.232, "step": 123480 }, { "epoch": 0.0373504, "grad_norm": 0.7455116510391235, "learning_rate": 1.5483521076776072e-05, "loss": 3.114, "step": 123490 }, { "epoch": 0.037376, "grad_norm": 0.7141309976577759, "learning_rate": 1.548284764684424e-05, "loss": 3.1613, "step": 123500 }, { "epoch": 0.0374016, "grad_norm": 0.7174506783485413, "learning_rate": 1.5482174181357602e-05, "loss": 2.9953, "step": 123510 }, { "epoch": 0.0374272, "grad_norm": 0.7485747933387756, "learning_rate": 1.548150068032052e-05, "loss": 3.1711, "step": 123520 }, { "epoch": 0.0374528, "grad_norm": 0.9252510070800781, "learning_rate": 1.5480827143737357e-05, "loss": 3.3163, "step": 123530 }, { "epoch": 0.0374784, "grad_norm": 0.8698096871376038, "learning_rate": 1.5480153571612496e-05, "loss": 3.3793, "step": 123540 }, { "epoch": 0.037504, "grad_norm": 0.8459914922714233, "learning_rate": 1.547947996395029e-05, "loss": 3.2481, "step": 123550 }, { "epoch": 0.0375296, "grad_norm": 0.8101368546485901, "learning_rate": 1.5478806320755112e-05, "loss": 2.8085, "step": 123560 }, { "epoch": 0.0375552, "grad_norm": 0.6857287883758545, "learning_rate": 1.5478132642031334e-05, "loss": 3.2651, "step": 123570 }, { "epoch": 0.0375808, "grad_norm": 0.7502570748329163, "learning_rate": 1.547745892778332e-05, "loss": 3.0056, "step": 123580 }, { "epoch": 0.0376064, "grad_norm": 0.9408870339393616, "learning_rate": 1.547678517801544e-05, "loss": 2.9674, "step": 123590 }, { "epoch": 0.037632, "grad_norm": 0.8327311873435974, "learning_rate": 1.5476111392732067e-05, "loss": 3.3063, "step": 123600 }, { "epoch": 0.0376576, "grad_norm": 1.2855703830718994, "learning_rate": 1.5475437571937563e-05, "loss": 3.1385, "step": 123610 }, { "epoch": 0.0376832, "grad_norm": 0.7807132601737976, "learning_rate": 1.5474763715636302e-05, "loss": 3.1979, "step": 123620 }, { "epoch": 0.0377088, "grad_norm": 0.711430013179779, "learning_rate": 1.5474089823832655e-05, "loss": 3.2087, "step": 123630 }, { "epoch": 0.0377344, "grad_norm": 1.1317481994628906, "learning_rate": 1.5473415896530992e-05, "loss": 2.9946, "step": 123640 }, { "epoch": 0.03776, "grad_norm": 0.7653282284736633, "learning_rate": 1.547274193373568e-05, "loss": 3.1232, "step": 123650 }, { "epoch": 0.0377856, "grad_norm": 0.9114817976951599, "learning_rate": 1.5472067935451088e-05, "loss": 3.3062, "step": 123660 }, { "epoch": 0.0378112, "grad_norm": 0.7543562054634094, "learning_rate": 1.5471393901681592e-05, "loss": 3.2439, "step": 123670 }, { "epoch": 0.0378368, "grad_norm": 0.6964967846870422, "learning_rate": 1.5470719832431562e-05, "loss": 3.1518, "step": 123680 }, { "epoch": 0.0378624, "grad_norm": 0.709596574306488, "learning_rate": 1.5470045727705367e-05, "loss": 3.1573, "step": 123690 }, { "epoch": 0.037888, "grad_norm": 0.768584132194519, "learning_rate": 1.5469371587507377e-05, "loss": 3.0688, "step": 123700 }, { "epoch": 0.0379136, "grad_norm": 0.7576825618743896, "learning_rate": 1.546869741184197e-05, "loss": 2.9739, "step": 123710 }, { "epoch": 0.0379392, "grad_norm": 0.7867745757102966, "learning_rate": 1.546802320071351e-05, "loss": 3.1198, "step": 123720 }, { "epoch": 0.0379648, "grad_norm": 0.7945892214775085, "learning_rate": 1.5467348954126377e-05, "loss": 3.2305, "step": 123730 }, { "epoch": 0.0379904, "grad_norm": 0.8250932693481445, "learning_rate": 1.5466674672084937e-05, "loss": 3.1145, "step": 123740 }, { "epoch": 0.038016, "grad_norm": 0.7635249495506287, "learning_rate": 1.5466000354593563e-05, "loss": 3.0895, "step": 123750 }, { "epoch": 0.0380416, "grad_norm": 0.7790599465370178, "learning_rate": 1.5465326001656627e-05, "loss": 3.2101, "step": 123760 }, { "epoch": 0.0380672, "grad_norm": 0.7990930676460266, "learning_rate": 1.546465161327851e-05, "loss": 3.1003, "step": 123770 }, { "epoch": 0.0380928, "grad_norm": 0.7708519101142883, "learning_rate": 1.5463977189463578e-05, "loss": 3.3111, "step": 123780 }, { "epoch": 0.0381184, "grad_norm": 0.7842068076133728, "learning_rate": 1.5463302730216205e-05, "loss": 3.1029, "step": 123790 }, { "epoch": 0.038144, "grad_norm": 0.7202425003051758, "learning_rate": 1.5462628235540767e-05, "loss": 3.2007, "step": 123800 }, { "epoch": 0.0381696, "grad_norm": 0.7281798720359802, "learning_rate": 1.5461953705441636e-05, "loss": 3.1337, "step": 123810 }, { "epoch": 0.0381952, "grad_norm": 0.681258499622345, "learning_rate": 1.5461279139923185e-05, "loss": 2.9284, "step": 123820 }, { "epoch": 0.0382208, "grad_norm": 0.7104662656784058, "learning_rate": 1.5460604538989792e-05, "loss": 3.2269, "step": 123830 }, { "epoch": 0.0382464, "grad_norm": 0.788145899772644, "learning_rate": 1.545992990264583e-05, "loss": 3.1118, "step": 123840 }, { "epoch": 0.038272, "grad_norm": 0.7237950563430786, "learning_rate": 1.5459255230895672e-05, "loss": 3.2149, "step": 123850 }, { "epoch": 0.0382976, "grad_norm": 1.293419599533081, "learning_rate": 1.5458580523743697e-05, "loss": 2.9887, "step": 123860 }, { "epoch": 0.0383232, "grad_norm": 0.7628561854362488, "learning_rate": 1.545790578119428e-05, "loss": 3.1853, "step": 123870 }, { "epoch": 0.0383488, "grad_norm": 0.7313528060913086, "learning_rate": 1.545723100325179e-05, "loss": 3.0721, "step": 123880 }, { "epoch": 0.0383744, "grad_norm": 0.7587596774101257, "learning_rate": 1.545655618992061e-05, "loss": 2.9555, "step": 123890 }, { "epoch": 0.0384, "grad_norm": 0.7111709713935852, "learning_rate": 1.5455881341205112e-05, "loss": 3.0821, "step": 123900 }, { "epoch": 0.0384256, "grad_norm": 0.8568133115768433, "learning_rate": 1.5455206457109672e-05, "loss": 3.1498, "step": 123910 }, { "epoch": 0.0384512, "grad_norm": 0.729474663734436, "learning_rate": 1.545453153763867e-05, "loss": 2.8282, "step": 123920 }, { "epoch": 0.0384768, "grad_norm": 0.7893701791763306, "learning_rate": 1.545385658279648e-05, "loss": 2.9426, "step": 123930 }, { "epoch": 0.0385024, "grad_norm": 0.6940604448318481, "learning_rate": 1.545318159258748e-05, "loss": 2.7682, "step": 123940 }, { "epoch": 0.038528, "grad_norm": 0.7325557470321655, "learning_rate": 1.5452506567016047e-05, "loss": 2.9105, "step": 123950 }, { "epoch": 0.0385536, "grad_norm": 0.7489163875579834, "learning_rate": 1.5451831506086558e-05, "loss": 2.9244, "step": 123960 }, { "epoch": 0.0385792, "grad_norm": 0.748418390750885, "learning_rate": 1.5451156409803388e-05, "loss": 2.8673, "step": 123970 }, { "epoch": 0.0386048, "grad_norm": 0.7258138060569763, "learning_rate": 1.545048127817092e-05, "loss": 2.8878, "step": 123980 }, { "epoch": 0.0386304, "grad_norm": 0.9095538258552551, "learning_rate": 1.544980611119353e-05, "loss": 2.9565, "step": 123990 }, { "epoch": 0.038656, "grad_norm": 0.7266674041748047, "learning_rate": 1.5449130908875595e-05, "loss": 3.0047, "step": 124000 }, { "epoch": 0.0386816, "grad_norm": 0.6939504146575928, "learning_rate": 1.5448455671221494e-05, "loss": 3.011, "step": 124010 }, { "epoch": 0.0387072, "grad_norm": 0.7175726294517517, "learning_rate": 1.5447780398235606e-05, "loss": 2.8947, "step": 124020 }, { "epoch": 0.0387328, "grad_norm": 0.7908490896224976, "learning_rate": 1.544710508992231e-05, "loss": 2.8416, "step": 124030 }, { "epoch": 0.0387584, "grad_norm": 0.749066948890686, "learning_rate": 1.5446429746285986e-05, "loss": 2.7969, "step": 124040 }, { "epoch": 0.038784, "grad_norm": 0.7130111455917358, "learning_rate": 1.544575436733101e-05, "loss": 2.8721, "step": 124050 }, { "epoch": 0.0388096, "grad_norm": 0.7820339798927307, "learning_rate": 1.544507895306177e-05, "loss": 2.8717, "step": 124060 }, { "epoch": 0.0388352, "grad_norm": 0.873902440071106, "learning_rate": 1.5444403503482634e-05, "loss": 3.0036, "step": 124070 }, { "epoch": 0.0388608, "grad_norm": 0.7036550641059875, "learning_rate": 1.544372801859799e-05, "loss": 3.0853, "step": 124080 }, { "epoch": 0.0388864, "grad_norm": 0.6665043234825134, "learning_rate": 1.5443052498412218e-05, "loss": 2.7785, "step": 124090 }, { "epoch": 0.038912, "grad_norm": 0.8155101537704468, "learning_rate": 1.544237694292969e-05, "loss": 3.0351, "step": 124100 }, { "epoch": 0.0389376, "grad_norm": 0.7354287505149841, "learning_rate": 1.5441701352154806e-05, "loss": 2.6511, "step": 124110 }, { "epoch": 0.0389632, "grad_norm": 0.7390442490577698, "learning_rate": 1.5441025726091928e-05, "loss": 2.8626, "step": 124120 }, { "epoch": 0.0389888, "grad_norm": 0.7153881788253784, "learning_rate": 1.5440350064745445e-05, "loss": 2.7875, "step": 124130 }, { "epoch": 0.0390144, "grad_norm": 0.732684850692749, "learning_rate": 1.5439674368119736e-05, "loss": 3.0, "step": 124140 }, { "epoch": 0.03904, "grad_norm": 0.6880097389221191, "learning_rate": 1.5438998636219186e-05, "loss": 2.8394, "step": 124150 }, { "epoch": 0.0390656, "grad_norm": 0.7524678111076355, "learning_rate": 1.5438322869048178e-05, "loss": 2.925, "step": 124160 }, { "epoch": 0.0390912, "grad_norm": 0.7630980610847473, "learning_rate": 1.543764706661109e-05, "loss": 3.2564, "step": 124170 }, { "epoch": 0.0391168, "grad_norm": 0.8576049208641052, "learning_rate": 1.5436971228912303e-05, "loss": 2.8347, "step": 124180 }, { "epoch": 0.0391424, "grad_norm": 0.8067550659179688, "learning_rate": 1.5436295355956207e-05, "loss": 2.8369, "step": 124190 }, { "epoch": 0.039168, "grad_norm": 0.7309563159942627, "learning_rate": 1.5435619447747176e-05, "loss": 2.8866, "step": 124200 }, { "epoch": 0.0391936, "grad_norm": 0.6553823351860046, "learning_rate": 1.54349435042896e-05, "loss": 2.6897, "step": 124210 }, { "epoch": 0.0392192, "grad_norm": 0.767899215221405, "learning_rate": 1.543426752558786e-05, "loss": 3.0017, "step": 124220 }, { "epoch": 0.0392448, "grad_norm": 0.8052501082420349, "learning_rate": 1.543359151164634e-05, "loss": 2.8692, "step": 124230 }, { "epoch": 0.0392704, "grad_norm": 0.6968886852264404, "learning_rate": 1.543291546246942e-05, "loss": 2.8045, "step": 124240 }, { "epoch": 0.039296, "grad_norm": 0.7088351845741272, "learning_rate": 1.543223937806149e-05, "loss": 2.937, "step": 124250 }, { "epoch": 0.0393216, "grad_norm": 0.661364734172821, "learning_rate": 1.5431563258426932e-05, "loss": 2.8313, "step": 124260 }, { "epoch": 0.0393472, "grad_norm": 1.0621705055236816, "learning_rate": 1.5430887103570126e-05, "loss": 2.7435, "step": 124270 }, { "epoch": 0.0393728, "grad_norm": 0.636944055557251, "learning_rate": 1.5430210913495463e-05, "loss": 2.9002, "step": 124280 }, { "epoch": 0.0393984, "grad_norm": 1.1498262882232666, "learning_rate": 1.5429534688207322e-05, "loss": 3.0294, "step": 124290 }, { "epoch": 0.039424, "grad_norm": 0.883499026298523, "learning_rate": 1.5428858427710096e-05, "loss": 2.9446, "step": 124300 }, { "epoch": 0.0394496, "grad_norm": 1.0296586751937866, "learning_rate": 1.5428182132008164e-05, "loss": 2.7292, "step": 124310 }, { "epoch": 0.0394752, "grad_norm": 0.788736879825592, "learning_rate": 1.5427505801105914e-05, "loss": 2.7744, "step": 124320 }, { "epoch": 0.0395008, "grad_norm": 0.7248668074607849, "learning_rate": 1.5426829435007735e-05, "loss": 2.9659, "step": 124330 }, { "epoch": 0.0395264, "grad_norm": 0.8070342540740967, "learning_rate": 1.5426153033718003e-05, "loss": 3.0141, "step": 124340 }, { "epoch": 0.039552, "grad_norm": 0.6520211100578308, "learning_rate": 1.5425476597241113e-05, "loss": 3.0024, "step": 124350 }, { "epoch": 0.0395776, "grad_norm": 0.6496695280075073, "learning_rate": 1.542480012558145e-05, "loss": 2.9386, "step": 124360 }, { "epoch": 0.0396032, "grad_norm": 0.7789679765701294, "learning_rate": 1.5424123618743402e-05, "loss": 2.8696, "step": 124370 }, { "epoch": 0.0396288, "grad_norm": 0.9245637655258179, "learning_rate": 1.5423447076731347e-05, "loss": 2.9456, "step": 124380 }, { "epoch": 0.0396544, "grad_norm": 0.7019448280334473, "learning_rate": 1.5422770499549687e-05, "loss": 3.3708, "step": 124390 }, { "epoch": 0.03968, "grad_norm": 0.7445545196533203, "learning_rate": 1.5422093887202797e-05, "loss": 2.9724, "step": 124400 }, { "epoch": 0.0397056, "grad_norm": 0.770789384841919, "learning_rate": 1.542141723969507e-05, "loss": 3.0666, "step": 124410 }, { "epoch": 0.0397312, "grad_norm": 0.7832414507865906, "learning_rate": 1.5420740557030895e-05, "loss": 3.0018, "step": 124420 }, { "epoch": 0.0397568, "grad_norm": 0.7735645174980164, "learning_rate": 1.5420063839214654e-05, "loss": 2.6834, "step": 124430 }, { "epoch": 0.0397824, "grad_norm": 0.7426646947860718, "learning_rate": 1.541938708625074e-05, "loss": 2.7089, "step": 124440 }, { "epoch": 0.039808, "grad_norm": 0.7368754148483276, "learning_rate": 1.5418710298143545e-05, "loss": 2.9373, "step": 124450 }, { "epoch": 0.0398336, "grad_norm": 0.6814676523208618, "learning_rate": 1.541803347489745e-05, "loss": 2.8843, "step": 124460 }, { "epoch": 0.0398592, "grad_norm": 0.6652403473854065, "learning_rate": 1.5417356616516853e-05, "loss": 2.8331, "step": 124470 }, { "epoch": 0.0398848, "grad_norm": 0.7418667674064636, "learning_rate": 1.5416679723006135e-05, "loss": 2.9719, "step": 124480 }, { "epoch": 0.0399104, "grad_norm": 0.6671649217605591, "learning_rate": 1.5416002794369685e-05, "loss": 2.9547, "step": 124490 }, { "epoch": 0.039936, "grad_norm": 0.673958957195282, "learning_rate": 1.54153258306119e-05, "loss": 2.6326, "step": 124500 }, { "epoch": 0.0399616, "grad_norm": 0.8226944208145142, "learning_rate": 1.5414648831737166e-05, "loss": 2.8311, "step": 124510 }, { "epoch": 0.0399872, "grad_norm": 0.6225755214691162, "learning_rate": 1.5413971797749873e-05, "loss": 2.8152, "step": 124520 }, { "epoch": 0.0400128, "grad_norm": 0.6822346448898315, "learning_rate": 1.541329472865441e-05, "loss": 2.7686, "step": 124530 }, { "epoch": 0.0400384, "grad_norm": 0.8349315524101257, "learning_rate": 1.5412617624455174e-05, "loss": 2.8792, "step": 124540 }, { "epoch": 0.040064, "grad_norm": 0.7282534837722778, "learning_rate": 1.5411940485156545e-05, "loss": 2.9117, "step": 124550 }, { "epoch": 0.0400896, "grad_norm": 0.6761432886123657, "learning_rate": 1.5411263310762925e-05, "loss": 2.9087, "step": 124560 }, { "epoch": 0.0401152, "grad_norm": 0.7817529439926147, "learning_rate": 1.5410586101278702e-05, "loss": 2.9327, "step": 124570 }, { "epoch": 0.0401408, "grad_norm": 0.8048045039176941, "learning_rate": 1.540990885670826e-05, "loss": 2.9795, "step": 124580 }, { "epoch": 0.0401664, "grad_norm": 0.7798486948013306, "learning_rate": 1.5409231577056004e-05, "loss": 3.0195, "step": 124590 }, { "epoch": 0.040192, "grad_norm": 0.7341333627700806, "learning_rate": 1.540855426232631e-05, "loss": 3.3684, "step": 124600 }, { "epoch": 0.0402176, "grad_norm": 0.697565495967865, "learning_rate": 1.540787691252359e-05, "loss": 2.8195, "step": 124610 }, { "epoch": 0.0402432, "grad_norm": 0.641193687915802, "learning_rate": 1.5407199527652216e-05, "loss": 2.8877, "step": 124620 }, { "epoch": 0.0402688, "grad_norm": 0.6853845119476318, "learning_rate": 1.5406522107716596e-05, "loss": 2.858, "step": 124630 }, { "epoch": 0.0402944, "grad_norm": 0.6064239740371704, "learning_rate": 1.540584465272111e-05, "loss": 2.7189, "step": 124640 }, { "epoch": 0.04032, "grad_norm": 1.0112956762313843, "learning_rate": 1.5405167162670166e-05, "loss": 2.9964, "step": 124650 }, { "epoch": 0.0403456, "grad_norm": 0.6209505200386047, "learning_rate": 1.5404489637568145e-05, "loss": 2.6747, "step": 124660 }, { "epoch": 0.0403712, "grad_norm": 0.8017656803131104, "learning_rate": 1.5403812077419448e-05, "loss": 3.1851, "step": 124670 }, { "epoch": 0.0403968, "grad_norm": 0.6238229274749756, "learning_rate": 1.5403134482228462e-05, "loss": 2.7056, "step": 124680 }, { "epoch": 0.0404224, "grad_norm": 0.9077609181404114, "learning_rate": 1.540245685199959e-05, "loss": 3.1473, "step": 124690 }, { "epoch": 0.040448, "grad_norm": 1.0206263065338135, "learning_rate": 1.5401779186737217e-05, "loss": 3.0323, "step": 124700 }, { "epoch": 0.0404736, "grad_norm": 0.8097306489944458, "learning_rate": 1.540110148644574e-05, "loss": 2.8959, "step": 124710 }, { "epoch": 0.0404992, "grad_norm": 0.6413999795913696, "learning_rate": 1.540042375112956e-05, "loss": 2.9382, "step": 124720 }, { "epoch": 0.0405248, "grad_norm": 0.7765621542930603, "learning_rate": 1.5399745980793067e-05, "loss": 3.027, "step": 124730 }, { "epoch": 0.0405504, "grad_norm": 0.7168742418289185, "learning_rate": 1.539906817544065e-05, "loss": 2.7384, "step": 124740 }, { "epoch": 0.040576, "grad_norm": 0.7290087938308716, "learning_rate": 1.5398390335076717e-05, "loss": 2.8358, "step": 124750 }, { "epoch": 0.0406016, "grad_norm": 0.9078118801116943, "learning_rate": 1.5397712459705658e-05, "loss": 3.0559, "step": 124760 }, { "epoch": 0.0406272, "grad_norm": 0.6892088055610657, "learning_rate": 1.5397034549331866e-05, "loss": 3.0691, "step": 124770 }, { "epoch": 0.0406528, "grad_norm": 0.8019964098930359, "learning_rate": 1.539635660395974e-05, "loss": 3.0105, "step": 124780 }, { "epoch": 0.0406784, "grad_norm": 0.7031724452972412, "learning_rate": 1.5395678623593674e-05, "loss": 2.9374, "step": 124790 }, { "epoch": 0.040704, "grad_norm": 0.7721139192581177, "learning_rate": 1.5395000608238067e-05, "loss": 3.0665, "step": 124800 }, { "epoch": 0.0407296, "grad_norm": 0.7917068600654602, "learning_rate": 1.5394322557897315e-05, "loss": 2.9353, "step": 124810 }, { "epoch": 0.0407552, "grad_norm": 0.9489421248435974, "learning_rate": 1.5393644472575814e-05, "loss": 2.4881, "step": 124820 }, { "epoch": 0.0407808, "grad_norm": 0.7828443050384521, "learning_rate": 1.5392966352277964e-05, "loss": 2.7671, "step": 124830 }, { "epoch": 0.0408064, "grad_norm": 0.7538997530937195, "learning_rate": 1.539228819700816e-05, "loss": 2.888, "step": 124840 }, { "epoch": 0.040832, "grad_norm": 0.8522412180900574, "learning_rate": 1.53916100067708e-05, "loss": 2.9723, "step": 124850 }, { "epoch": 0.0408576, "grad_norm": 0.7509545683860779, "learning_rate": 1.539093178157028e-05, "loss": 2.8373, "step": 124860 }, { "epoch": 0.0408832, "grad_norm": 0.8410879373550415, "learning_rate": 1.5390253521411004e-05, "loss": 2.9846, "step": 124870 }, { "epoch": 0.0409088, "grad_norm": 0.8231285810470581, "learning_rate": 1.5389575226297358e-05, "loss": 2.9106, "step": 124880 }, { "epoch": 0.0409344, "grad_norm": 0.728385865688324, "learning_rate": 1.538889689623376e-05, "loss": 2.7572, "step": 124890 }, { "epoch": 0.04096, "grad_norm": 0.9196845889091492, "learning_rate": 1.5388218531224592e-05, "loss": 3.0063, "step": 124900 }, { "epoch": 0.0409856, "grad_norm": 0.6657965183258057, "learning_rate": 1.5387540131274258e-05, "loss": 2.7852, "step": 124910 }, { "epoch": 0.0410112, "grad_norm": 0.7000821828842163, "learning_rate": 1.5386861696387162e-05, "loss": 3.0102, "step": 124920 }, { "epoch": 0.0410368, "grad_norm": 0.6886584162712097, "learning_rate": 1.53861832265677e-05, "loss": 2.8753, "step": 124930 }, { "epoch": 0.0410624, "grad_norm": 0.6882743835449219, "learning_rate": 1.538550472182027e-05, "loss": 2.989, "step": 124940 }, { "epoch": 0.041088, "grad_norm": 0.8871084451675415, "learning_rate": 1.538482618214927e-05, "loss": 2.9267, "step": 124950 }, { "epoch": 0.0411136, "grad_norm": 0.8116686344146729, "learning_rate": 1.5384147607559104e-05, "loss": 3.1313, "step": 124960 }, { "epoch": 0.0411392, "grad_norm": 0.6768342852592468, "learning_rate": 1.5383468998054173e-05, "loss": 3.0922, "step": 124970 }, { "epoch": 0.0411648, "grad_norm": 0.8136584758758545, "learning_rate": 1.5382790353638875e-05, "loss": 3.1715, "step": 124980 }, { "epoch": 0.0411904, "grad_norm": 2.106398344039917, "learning_rate": 1.5382111674317613e-05, "loss": 3.1771, "step": 124990 }, { "epoch": 0.041216, "grad_norm": 0.7026546001434326, "learning_rate": 1.538143296009479e-05, "loss": 3.0913, "step": 125000 }, { "epoch": 0.0412416, "grad_norm": 0.7203381657600403, "learning_rate": 1.53807542109748e-05, "loss": 3.1639, "step": 125010 }, { "epoch": 0.0412672, "grad_norm": 0.7560357451438904, "learning_rate": 1.538007542696205e-05, "loss": 3.0887, "step": 125020 }, { "epoch": 0.0412928, "grad_norm": 0.7005445957183838, "learning_rate": 1.5379396608060944e-05, "loss": 2.8306, "step": 125030 }, { "epoch": 0.0413184, "grad_norm": 1.2968921661376953, "learning_rate": 1.537871775427588e-05, "loss": 3.0857, "step": 125040 }, { "epoch": 0.041344, "grad_norm": 0.7717990875244141, "learning_rate": 1.5378038865611258e-05, "loss": 2.9433, "step": 125050 }, { "epoch": 0.0413696, "grad_norm": 0.6527106165885925, "learning_rate": 1.5377359942071485e-05, "loss": 3.0983, "step": 125060 }, { "epoch": 0.0413952, "grad_norm": 0.8499014973640442, "learning_rate": 1.5376680983660957e-05, "loss": 3.0332, "step": 125070 }, { "epoch": 0.0414208, "grad_norm": 0.689444899559021, "learning_rate": 1.5376001990384088e-05, "loss": 3.185, "step": 125080 }, { "epoch": 0.0414464, "grad_norm": 0.7559958100318909, "learning_rate": 1.5375322962245273e-05, "loss": 2.9386, "step": 125090 }, { "epoch": 0.041472, "grad_norm": 1.206190586090088, "learning_rate": 1.5374643899248918e-05, "loss": 3.0221, "step": 125100 }, { "epoch": 0.0414976, "grad_norm": 0.7551124095916748, "learning_rate": 1.5373964801399424e-05, "loss": 3.1064, "step": 125110 }, { "epoch": 0.0415232, "grad_norm": 0.8365989327430725, "learning_rate": 1.53732856687012e-05, "loss": 2.9846, "step": 125120 }, { "epoch": 0.0415488, "grad_norm": 0.8656526207923889, "learning_rate": 1.5372606501158643e-05, "loss": 2.9227, "step": 125130 }, { "epoch": 0.0415744, "grad_norm": 0.7529104351997375, "learning_rate": 1.537192729877616e-05, "loss": 2.8263, "step": 125140 }, { "epoch": 0.0416, "grad_norm": 0.988813579082489, "learning_rate": 1.537124806155816e-05, "loss": 3.0766, "step": 125150 }, { "epoch": 0.0416256, "grad_norm": 0.8713960647583008, "learning_rate": 1.537056878950904e-05, "loss": 3.0617, "step": 125160 }, { "epoch": 0.0416512, "grad_norm": 0.9243590831756592, "learning_rate": 1.536988948263321e-05, "loss": 3.0654, "step": 125170 }, { "epoch": 0.0416768, "grad_norm": 1.0318536758422852, "learning_rate": 1.5369210140935074e-05, "loss": 3.105, "step": 125180 }, { "epoch": 0.0417024, "grad_norm": 0.6914140582084656, "learning_rate": 1.5368530764419038e-05, "loss": 3.0035, "step": 125190 }, { "epoch": 0.041728, "grad_norm": 0.6732264161109924, "learning_rate": 1.5367851353089508e-05, "loss": 2.9804, "step": 125200 }, { "epoch": 0.0417536, "grad_norm": 0.8318272829055786, "learning_rate": 1.5367171906950887e-05, "loss": 3.0567, "step": 125210 }, { "epoch": 0.0417792, "grad_norm": 0.6849144697189331, "learning_rate": 1.536649242600758e-05, "loss": 3.0571, "step": 125220 }, { "epoch": 0.0418048, "grad_norm": 0.7352623343467712, "learning_rate": 1.5365812910263998e-05, "loss": 3.1726, "step": 125230 }, { "epoch": 0.0418304, "grad_norm": 2.60026478767395, "learning_rate": 1.5365133359724545e-05, "loss": 2.8388, "step": 125240 }, { "epoch": 0.041856, "grad_norm": 0.671463131904602, "learning_rate": 1.536445377439363e-05, "loss": 2.9603, "step": 125250 }, { "epoch": 0.0418816, "grad_norm": 0.7541422247886658, "learning_rate": 1.5363774154275656e-05, "loss": 3.155, "step": 125260 }, { "epoch": 0.0419072, "grad_norm": 0.657305121421814, "learning_rate": 1.5363094499375032e-05, "loss": 3.1319, "step": 125270 }, { "epoch": 0.0419328, "grad_norm": 0.8069433569908142, "learning_rate": 1.5362414809696167e-05, "loss": 2.9822, "step": 125280 }, { "epoch": 0.0419584, "grad_norm": 0.6672627925872803, "learning_rate": 1.5361735085243466e-05, "loss": 3.0044, "step": 125290 }, { "epoch": 0.041984, "grad_norm": 0.7403931617736816, "learning_rate": 1.536105532602134e-05, "loss": 3.0595, "step": 125300 }, { "epoch": 0.0420096, "grad_norm": 0.6924645900726318, "learning_rate": 1.5360375532034192e-05, "loss": 2.9465, "step": 125310 }, { "epoch": 0.0420352, "grad_norm": 0.8255952596664429, "learning_rate": 1.5359695703286436e-05, "loss": 3.0364, "step": 125320 }, { "epoch": 0.0420608, "grad_norm": 1.181806206703186, "learning_rate": 1.5359015839782475e-05, "loss": 3.0806, "step": 125330 }, { "epoch": 0.0420864, "grad_norm": 0.8153981566429138, "learning_rate": 1.535833594152672e-05, "loss": 3.0589, "step": 125340 }, { "epoch": 0.042112, "grad_norm": 0.8400455117225647, "learning_rate": 1.5357656008523584e-05, "loss": 3.0711, "step": 125350 }, { "epoch": 0.0421376, "grad_norm": 0.8192578554153442, "learning_rate": 1.535697604077747e-05, "loss": 3.0691, "step": 125360 }, { "epoch": 0.0421632, "grad_norm": 0.7185347080230713, "learning_rate": 1.5356296038292794e-05, "loss": 3.0936, "step": 125370 }, { "epoch": 0.0421888, "grad_norm": 0.7270528078079224, "learning_rate": 1.535561600107396e-05, "loss": 3.087, "step": 125380 }, { "epoch": 0.0422144, "grad_norm": 0.71868497133255, "learning_rate": 1.5354935929125375e-05, "loss": 2.84, "step": 125390 }, { "epoch": 0.04224, "grad_norm": 0.8263117074966431, "learning_rate": 1.5354255822451455e-05, "loss": 3.0133, "step": 125400 }, { "epoch": 0.0422656, "grad_norm": 0.7498098611831665, "learning_rate": 1.535357568105661e-05, "loss": 3.1827, "step": 125410 }, { "epoch": 0.0422912, "grad_norm": 0.6766838431358337, "learning_rate": 1.5352895504945254e-05, "loss": 3.004, "step": 125420 }, { "epoch": 0.0423168, "grad_norm": 0.7125664949417114, "learning_rate": 1.535221529412179e-05, "loss": 2.9274, "step": 125430 }, { "epoch": 0.0423424, "grad_norm": 0.7119992971420288, "learning_rate": 1.5351535048590634e-05, "loss": 2.9011, "step": 125440 }, { "epoch": 0.042368, "grad_norm": 0.6962350010871887, "learning_rate": 1.5350854768356194e-05, "loss": 2.9664, "step": 125450 }, { "epoch": 0.0423936, "grad_norm": 0.7322153449058533, "learning_rate": 1.5350174453422882e-05, "loss": 3.0761, "step": 125460 }, { "epoch": 0.0424192, "grad_norm": 0.8431721925735474, "learning_rate": 1.5349494103795112e-05, "loss": 3.0209, "step": 125470 }, { "epoch": 0.0424448, "grad_norm": 0.7521161437034607, "learning_rate": 1.534881371947729e-05, "loss": 3.135, "step": 125480 }, { "epoch": 0.0424704, "grad_norm": 1.045583963394165, "learning_rate": 1.534813330047384e-05, "loss": 3.1925, "step": 125490 }, { "epoch": 0.042496, "grad_norm": 0.696338951587677, "learning_rate": 1.534745284678916e-05, "loss": 2.9788, "step": 125500 }, { "epoch": 0.0425216, "grad_norm": 0.7014554142951965, "learning_rate": 1.5346772358427674e-05, "loss": 3.0545, "step": 125510 }, { "epoch": 0.0425472, "grad_norm": 0.78289794921875, "learning_rate": 1.534609183539379e-05, "loss": 2.8763, "step": 125520 }, { "epoch": 0.0425728, "grad_norm": 0.946419358253479, "learning_rate": 1.534541127769192e-05, "loss": 3.3214, "step": 125530 }, { "epoch": 0.0425984, "grad_norm": 0.7349060773849487, "learning_rate": 1.5344730685326478e-05, "loss": 2.8631, "step": 125540 }, { "epoch": 0.042624, "grad_norm": 0.6443593502044678, "learning_rate": 1.5344050058301882e-05, "loss": 2.9162, "step": 125550 }, { "epoch": 0.0426496, "grad_norm": 0.8416638970375061, "learning_rate": 1.5343369396622535e-05, "loss": 2.946, "step": 125560 }, { "epoch": 0.0426752, "grad_norm": 0.8087143898010254, "learning_rate": 1.534268870029286e-05, "loss": 3.3059, "step": 125570 }, { "epoch": 0.0427008, "grad_norm": 0.7332867980003357, "learning_rate": 1.534200796931727e-05, "loss": 2.9912, "step": 125580 }, { "epoch": 0.0427264, "grad_norm": 0.8304922580718994, "learning_rate": 1.5341327203700177e-05, "loss": 2.9838, "step": 125590 }, { "epoch": 0.042752, "grad_norm": 0.6998001337051392, "learning_rate": 1.5340646403445997e-05, "loss": 3.4129, "step": 125600 }, { "epoch": 0.0427776, "grad_norm": 0.9514519572257996, "learning_rate": 1.5339965568559146e-05, "loss": 3.0944, "step": 125610 }, { "epoch": 0.0428032, "grad_norm": 0.7652586102485657, "learning_rate": 1.5339284699044033e-05, "loss": 2.9559, "step": 125620 }, { "epoch": 0.0428288, "grad_norm": 0.7058196663856506, "learning_rate": 1.5338603794905083e-05, "loss": 2.9329, "step": 125630 }, { "epoch": 0.0428544, "grad_norm": 1.0515921115875244, "learning_rate": 1.5337922856146704e-05, "loss": 3.075, "step": 125640 }, { "epoch": 0.04288, "grad_norm": 0.9387261271476746, "learning_rate": 1.5337241882773314e-05, "loss": 3.0574, "step": 125650 }, { "epoch": 0.0429056, "grad_norm": 0.9013000726699829, "learning_rate": 1.5336560874789328e-05, "loss": 3.0451, "step": 125660 }, { "epoch": 0.0429312, "grad_norm": 0.7126544117927551, "learning_rate": 1.5335879832199163e-05, "loss": 3.0598, "step": 125670 }, { "epoch": 0.0429568, "grad_norm": 0.677494466304779, "learning_rate": 1.5335198755007235e-05, "loss": 2.9117, "step": 125680 }, { "epoch": 0.0429824, "grad_norm": 0.7186419367790222, "learning_rate": 1.5334517643217965e-05, "loss": 2.8401, "step": 125690 }, { "epoch": 0.043008, "grad_norm": 0.7708032131195068, "learning_rate": 1.5333836496835765e-05, "loss": 3.0648, "step": 125700 }, { "epoch": 0.0430336, "grad_norm": 1.4477092027664185, "learning_rate": 1.5333155315865048e-05, "loss": 2.9932, "step": 125710 }, { "epoch": 0.0430592, "grad_norm": 0.7534489631652832, "learning_rate": 1.533247410031024e-05, "loss": 2.9623, "step": 125720 }, { "epoch": 0.0430848, "grad_norm": 0.6906997561454773, "learning_rate": 1.5331792850175754e-05, "loss": 3.1515, "step": 125730 }, { "epoch": 0.0431104, "grad_norm": 1.4843958616256714, "learning_rate": 1.533111156546601e-05, "loss": 3.2824, "step": 125740 }, { "epoch": 0.043136, "grad_norm": 0.90101158618927, "learning_rate": 1.533043024618542e-05, "loss": 3.3757, "step": 125750 }, { "epoch": 0.0431616, "grad_norm": 0.701739490032196, "learning_rate": 1.532974889233841e-05, "loss": 3.0528, "step": 125760 }, { "epoch": 0.0431872, "grad_norm": 0.7503359317779541, "learning_rate": 1.5329067503929392e-05, "loss": 2.9471, "step": 125770 }, { "epoch": 0.0432128, "grad_norm": 0.8074245452880859, "learning_rate": 1.5328386080962793e-05, "loss": 3.0461, "step": 125780 }, { "epoch": 0.0432384, "grad_norm": 0.7993535995483398, "learning_rate": 1.5327704623443022e-05, "loss": 2.8979, "step": 125790 }, { "epoch": 0.043264, "grad_norm": 0.7226826548576355, "learning_rate": 1.5327023131374502e-05, "loss": 2.9887, "step": 125800 }, { "epoch": 0.0432896, "grad_norm": 0.7933639287948608, "learning_rate": 1.5326341604761654e-05, "loss": 2.8532, "step": 125810 }, { "epoch": 0.0433152, "grad_norm": 0.673751711845398, "learning_rate": 1.5325660043608897e-05, "loss": 3.1191, "step": 125820 }, { "epoch": 0.0433408, "grad_norm": 0.7077876329421997, "learning_rate": 1.5324978447920648e-05, "loss": 2.9532, "step": 125830 }, { "epoch": 0.0433664, "grad_norm": 0.8206611275672913, "learning_rate": 1.5324296817701332e-05, "loss": 3.0134, "step": 125840 }, { "epoch": 0.043392, "grad_norm": 0.7509341239929199, "learning_rate": 1.5323615152955364e-05, "loss": 2.8497, "step": 125850 }, { "epoch": 0.0434176, "grad_norm": 0.6646078824996948, "learning_rate": 1.5322933453687166e-05, "loss": 3.1314, "step": 125860 }, { "epoch": 0.0434432, "grad_norm": 0.6429510712623596, "learning_rate": 1.532225171990116e-05, "loss": 2.9667, "step": 125870 }, { "epoch": 0.0434688, "grad_norm": 0.6930200457572937, "learning_rate": 1.5321569951601768e-05, "loss": 3.1092, "step": 125880 }, { "epoch": 0.0434944, "grad_norm": 0.6785938739776611, "learning_rate": 1.532088814879341e-05, "loss": 2.9118, "step": 125890 }, { "epoch": 0.04352, "grad_norm": 0.8100569844245911, "learning_rate": 1.5320206311480502e-05, "loss": 3.1212, "step": 125900 }, { "epoch": 0.0435456, "grad_norm": 0.7093473076820374, "learning_rate": 1.531952443966747e-05, "loss": 3.162, "step": 125910 }, { "epoch": 0.0435712, "grad_norm": 0.7499752640724182, "learning_rate": 1.531884253335874e-05, "loss": 3.0234, "step": 125920 }, { "epoch": 0.0435968, "grad_norm": 0.7450871467590332, "learning_rate": 1.5318160592558723e-05, "loss": 3.1019, "step": 125930 }, { "epoch": 0.0436224, "grad_norm": 0.7780405282974243, "learning_rate": 1.5317478617271855e-05, "loss": 2.9624, "step": 125940 }, { "epoch": 0.043648, "grad_norm": 0.723106861114502, "learning_rate": 1.531679660750255e-05, "loss": 3.1003, "step": 125950 }, { "epoch": 0.0436736, "grad_norm": 0.6687521934509277, "learning_rate": 1.531611456325523e-05, "loss": 3.0202, "step": 125960 }, { "epoch": 0.0436992, "grad_norm": 0.7023062109947205, "learning_rate": 1.531543248453432e-05, "loss": 3.1673, "step": 125970 }, { "epoch": 0.0437248, "grad_norm": 0.6869547367095947, "learning_rate": 1.5314750371344246e-05, "loss": 3.0763, "step": 125980 }, { "epoch": 0.0437504, "grad_norm": 0.6847842335700989, "learning_rate": 1.5314068223689423e-05, "loss": 2.9953, "step": 125990 }, { "epoch": 0.043776, "grad_norm": 0.69320148229599, "learning_rate": 1.5313386041574284e-05, "loss": 3.0375, "step": 126000 }, { "epoch": 0.0438016, "grad_norm": 0.7708196043968201, "learning_rate": 1.531270382500325e-05, "loss": 3.0413, "step": 126010 }, { "epoch": 0.0438272, "grad_norm": 0.9779266119003296, "learning_rate": 1.531202157398074e-05, "loss": 2.9587, "step": 126020 }, { "epoch": 0.0438528, "grad_norm": 0.6883912086486816, "learning_rate": 1.5311339288511186e-05, "loss": 2.9824, "step": 126030 }, { "epoch": 0.0438784, "grad_norm": 0.7274159789085388, "learning_rate": 1.5310656968599006e-05, "loss": 2.8511, "step": 126040 }, { "epoch": 0.043904, "grad_norm": 0.8286908864974976, "learning_rate": 1.5309974614248625e-05, "loss": 3.0717, "step": 126050 }, { "epoch": 0.0439296, "grad_norm": 0.6765512824058533, "learning_rate": 1.530929222546447e-05, "loss": 2.9942, "step": 126060 }, { "epoch": 0.0439552, "grad_norm": 1.9447752237319946, "learning_rate": 1.5308609802250968e-05, "loss": 3.0273, "step": 126070 }, { "epoch": 0.0439808, "grad_norm": 0.7226161360740662, "learning_rate": 1.5307927344612543e-05, "loss": 3.0098, "step": 126080 }, { "epoch": 0.0440064, "grad_norm": 0.8449322581291199, "learning_rate": 1.5307244852553615e-05, "loss": 3.0178, "step": 126090 }, { "epoch": 0.044032, "grad_norm": 0.8341689705848694, "learning_rate": 1.5306562326078617e-05, "loss": 3.0241, "step": 126100 }, { "epoch": 0.0440576, "grad_norm": 0.7988303303718567, "learning_rate": 1.5305879765191975e-05, "loss": 3.0552, "step": 126110 }, { "epoch": 0.0440832, "grad_norm": 0.8232439160346985, "learning_rate": 1.530519716989811e-05, "loss": 3.0448, "step": 126120 }, { "epoch": 0.0441088, "grad_norm": 0.955187201499939, "learning_rate": 1.5304514540201453e-05, "loss": 3.0017, "step": 126130 }, { "epoch": 0.0441344, "grad_norm": 0.6857151985168457, "learning_rate": 1.530383187610643e-05, "loss": 3.1534, "step": 126140 }, { "epoch": 0.04416, "grad_norm": 0.708837628364563, "learning_rate": 1.5303149177617466e-05, "loss": 2.9976, "step": 126150 }, { "epoch": 0.0441856, "grad_norm": 0.737864077091217, "learning_rate": 1.5302466444738987e-05, "loss": 3.0933, "step": 126160 }, { "epoch": 0.0442112, "grad_norm": 0.6731730699539185, "learning_rate": 1.5301783677475427e-05, "loss": 2.9266, "step": 126170 }, { "epoch": 0.0442368, "grad_norm": 0.6535726189613342, "learning_rate": 1.5301100875831205e-05, "loss": 2.8716, "step": 126180 }, { "epoch": 0.0442624, "grad_norm": 0.8065158724784851, "learning_rate": 1.5300418039810755e-05, "loss": 3.0899, "step": 126190 }, { "epoch": 0.044288, "grad_norm": 0.8179424405097961, "learning_rate": 1.52997351694185e-05, "loss": 3.0358, "step": 126200 }, { "epoch": 0.0443136, "grad_norm": 0.778612494468689, "learning_rate": 1.5299052264658872e-05, "loss": 3.0495, "step": 126210 }, { "epoch": 0.0443392, "grad_norm": 0.6769242286682129, "learning_rate": 1.52983693255363e-05, "loss": 2.9048, "step": 126220 }, { "epoch": 0.0443648, "grad_norm": 0.7757065892219543, "learning_rate": 1.5297686352055207e-05, "loss": 3.053, "step": 126230 }, { "epoch": 0.0443904, "grad_norm": 0.6954228281974792, "learning_rate": 1.5297003344220032e-05, "loss": 2.9007, "step": 126240 }, { "epoch": 0.044416, "grad_norm": 0.7991501688957214, "learning_rate": 1.5296320302035194e-05, "loss": 3.1343, "step": 126250 }, { "epoch": 0.0444416, "grad_norm": 0.733173668384552, "learning_rate": 1.5295637225505125e-05, "loss": 3.1492, "step": 126260 }, { "epoch": 0.0444672, "grad_norm": 0.8903587460517883, "learning_rate": 1.529495411463426e-05, "loss": 2.9815, "step": 126270 }, { "epoch": 0.0444928, "grad_norm": 0.7633318901062012, "learning_rate": 1.529427096942702e-05, "loss": 3.0352, "step": 126280 }, { "epoch": 0.0445184, "grad_norm": 1.294397234916687, "learning_rate": 1.5293587789887846e-05, "loss": 3.195, "step": 126290 }, { "epoch": 0.044544, "grad_norm": 0.7189125418663025, "learning_rate": 1.5292904576021156e-05, "loss": 3.0132, "step": 126300 }, { "epoch": 0.0445696, "grad_norm": 0.7020672559738159, "learning_rate": 1.529222132783139e-05, "loss": 3.0023, "step": 126310 }, { "epoch": 0.0445952, "grad_norm": 1.2875311374664307, "learning_rate": 1.5291538045322976e-05, "loss": 3.0177, "step": 126320 }, { "epoch": 0.0446208, "grad_norm": 0.7091734409332275, "learning_rate": 1.5290854728500343e-05, "loss": 3.1092, "step": 126330 }, { "epoch": 0.0446464, "grad_norm": 0.8970181941986084, "learning_rate": 1.5290171377367925e-05, "loss": 3.1481, "step": 126340 }, { "epoch": 0.044672, "grad_norm": 0.7006480097770691, "learning_rate": 1.5289487991930148e-05, "loss": 3.0182, "step": 126350 }, { "epoch": 0.0446976, "grad_norm": 0.7166397571563721, "learning_rate": 1.5288804572191443e-05, "loss": 2.9271, "step": 126360 }, { "epoch": 0.0447232, "grad_norm": 0.8303005695343018, "learning_rate": 1.5288121118156256e-05, "loss": 2.9972, "step": 126370 }, { "epoch": 0.0447488, "grad_norm": 0.9908134937286377, "learning_rate": 1.5287437629829003e-05, "loss": 3.0541, "step": 126380 }, { "epoch": 0.0447744, "grad_norm": 0.7499796152114868, "learning_rate": 1.5286754107214122e-05, "loss": 2.9993, "step": 126390 }, { "epoch": 0.0448, "grad_norm": 0.7012671828269958, "learning_rate": 1.528607055031605e-05, "loss": 3.0418, "step": 126400 }, { "epoch": 0.0448256, "grad_norm": 0.679189145565033, "learning_rate": 1.528538695913921e-05, "loss": 2.9081, "step": 126410 }, { "epoch": 0.0448512, "grad_norm": 0.7880390286445618, "learning_rate": 1.5284703333688045e-05, "loss": 3.0388, "step": 126420 }, { "epoch": 0.0448768, "grad_norm": 0.6902658343315125, "learning_rate": 1.5284019673966982e-05, "loss": 2.9869, "step": 126430 }, { "epoch": 0.0449024, "grad_norm": 0.8798694610595703, "learning_rate": 1.5283335979980453e-05, "loss": 2.9767, "step": 126440 }, { "epoch": 0.044928, "grad_norm": 0.8573668599128723, "learning_rate": 1.5282652251732897e-05, "loss": 3.0463, "step": 126450 }, { "epoch": 0.0449536, "grad_norm": 1.1906356811523438, "learning_rate": 1.5281968489228744e-05, "loss": 3.107, "step": 126460 }, { "epoch": 0.0449792, "grad_norm": 0.7133072018623352, "learning_rate": 1.528128469247243e-05, "loss": 3.0853, "step": 126470 }, { "epoch": 0.0450048, "grad_norm": 0.6835469603538513, "learning_rate": 1.528060086146839e-05, "loss": 2.9681, "step": 126480 }, { "epoch": 0.0450304, "grad_norm": 0.7152994871139526, "learning_rate": 1.5279916996221053e-05, "loss": 2.9704, "step": 126490 }, { "epoch": 0.045056, "grad_norm": 0.7304689288139343, "learning_rate": 1.527923309673486e-05, "loss": 2.9537, "step": 126500 }, { "epoch": 0.0450816, "grad_norm": 0.9734928607940674, "learning_rate": 1.5278549163014242e-05, "loss": 2.9848, "step": 126510 }, { "epoch": 0.0451072, "grad_norm": 2.1973137855529785, "learning_rate": 1.5277865195063637e-05, "loss": 3.0778, "step": 126520 }, { "epoch": 0.0451328, "grad_norm": 0.6838570237159729, "learning_rate": 1.527718119288748e-05, "loss": 2.9493, "step": 126530 }, { "epoch": 0.0451584, "grad_norm": 0.7851141095161438, "learning_rate": 1.52764971564902e-05, "loss": 3.0283, "step": 126540 }, { "epoch": 0.045184, "grad_norm": 1.0141959190368652, "learning_rate": 1.5275813085876245e-05, "loss": 2.9834, "step": 126550 }, { "epoch": 0.0452096, "grad_norm": 0.7551377415657043, "learning_rate": 1.5275128981050038e-05, "loss": 3.1192, "step": 126560 }, { "epoch": 0.0452352, "grad_norm": 0.6856139898300171, "learning_rate": 1.5274444842016026e-05, "loss": 3.1473, "step": 126570 }, { "epoch": 0.0452608, "grad_norm": 0.7533906102180481, "learning_rate": 1.527376066877864e-05, "loss": 2.8959, "step": 126580 }, { "epoch": 0.0452864, "grad_norm": 0.6976428031921387, "learning_rate": 1.5273076461342318e-05, "loss": 3.1245, "step": 126590 }, { "epoch": 0.045312, "grad_norm": 0.7994243502616882, "learning_rate": 1.5272392219711494e-05, "loss": 3.0739, "step": 126600 }, { "epoch": 0.0453376, "grad_norm": 0.7318904995918274, "learning_rate": 1.5271707943890613e-05, "loss": 2.9692, "step": 126610 }, { "epoch": 0.0453632, "grad_norm": 1.3310216665267944, "learning_rate": 1.52710236338841e-05, "loss": 3.0111, "step": 126620 }, { "epoch": 0.0453888, "grad_norm": 0.7077253460884094, "learning_rate": 1.5270339289696407e-05, "loss": 3.1414, "step": 126630 }, { "epoch": 0.0454144, "grad_norm": 0.7721645832061768, "learning_rate": 1.5269654911331958e-05, "loss": 2.9968, "step": 126640 }, { "epoch": 0.04544, "grad_norm": 0.7569018006324768, "learning_rate": 1.5268970498795202e-05, "loss": 2.9377, "step": 126650 }, { "epoch": 0.0454656, "grad_norm": 1.3979566097259521, "learning_rate": 1.5268286052090567e-05, "loss": 3.2102, "step": 126660 }, { "epoch": 0.0454912, "grad_norm": 1.8147257566452026, "learning_rate": 1.5267601571222505e-05, "loss": 2.8848, "step": 126670 }, { "epoch": 0.0455168, "grad_norm": 0.6784709692001343, "learning_rate": 1.526691705619544e-05, "loss": 2.9445, "step": 126680 }, { "epoch": 0.0455424, "grad_norm": 0.8137476444244385, "learning_rate": 1.526623250701382e-05, "loss": 3.1362, "step": 126690 }, { "epoch": 0.045568, "grad_norm": 0.7463153600692749, "learning_rate": 1.5265547923682084e-05, "loss": 3.0619, "step": 126700 }, { "epoch": 0.0455936, "grad_norm": 0.7690771222114563, "learning_rate": 1.5264863306204666e-05, "loss": 2.922, "step": 126710 }, { "epoch": 0.0456192, "grad_norm": 0.8212154507637024, "learning_rate": 1.526417865458601e-05, "loss": 3.0538, "step": 126720 }, { "epoch": 0.0456448, "grad_norm": 0.7406108379364014, "learning_rate": 1.5263493968830554e-05, "loss": 2.9616, "step": 126730 }, { "epoch": 0.0456704, "grad_norm": 0.6921138763427734, "learning_rate": 1.526280924894274e-05, "loss": 3.0113, "step": 126740 }, { "epoch": 0.045696, "grad_norm": 0.761978268623352, "learning_rate": 1.5262124494927006e-05, "loss": 3.1148, "step": 126750 }, { "epoch": 0.0457216, "grad_norm": 0.7175716161727905, "learning_rate": 1.526143970678779e-05, "loss": 3.0937, "step": 126760 }, { "epoch": 0.0457472, "grad_norm": 0.8615398406982422, "learning_rate": 1.526075488452954e-05, "loss": 2.9519, "step": 126770 }, { "epoch": 0.0457728, "grad_norm": 0.788783609867096, "learning_rate": 1.526007002815669e-05, "loss": 3.0432, "step": 126780 }, { "epoch": 0.0457984, "grad_norm": 1.3424615859985352, "learning_rate": 1.5259385137673683e-05, "loss": 3.0833, "step": 126790 }, { "epoch": 0.045824, "grad_norm": 0.7237810492515564, "learning_rate": 1.5258700213084967e-05, "loss": 3.0005, "step": 126800 }, { "epoch": 0.0458496, "grad_norm": 0.8030499815940857, "learning_rate": 1.5258015254394973e-05, "loss": 3.1273, "step": 126810 }, { "epoch": 0.0458752, "grad_norm": 0.7432007789611816, "learning_rate": 1.5257330261608148e-05, "loss": 3.0763, "step": 126820 }, { "epoch": 0.0459008, "grad_norm": 0.6810356974601746, "learning_rate": 1.5256645234728936e-05, "loss": 3.2008, "step": 126830 }, { "epoch": 0.0459264, "grad_norm": 0.7108353972434998, "learning_rate": 1.5255960173761772e-05, "loss": 2.9235, "step": 126840 }, { "epoch": 0.045952, "grad_norm": 0.6772131323814392, "learning_rate": 1.5255275078711105e-05, "loss": 2.9487, "step": 126850 }, { "epoch": 0.0459776, "grad_norm": 0.6688607335090637, "learning_rate": 1.5254589949581377e-05, "loss": 3.1755, "step": 126860 }, { "epoch": 0.0460032, "grad_norm": 0.7804098725318909, "learning_rate": 1.5253904786377032e-05, "loss": 3.3094, "step": 126870 }, { "epoch": 0.0460288, "grad_norm": 0.7294216752052307, "learning_rate": 1.5253219589102504e-05, "loss": 3.1555, "step": 126880 }, { "epoch": 0.0460544, "grad_norm": 0.7984243631362915, "learning_rate": 1.5252534357762248e-05, "loss": 3.166, "step": 126890 }, { "epoch": 0.04608, "grad_norm": 0.6678454279899597, "learning_rate": 1.5251849092360702e-05, "loss": 3.022, "step": 126900 }, { "epoch": 0.0461056, "grad_norm": 0.6687915325164795, "learning_rate": 1.525116379290231e-05, "loss": 2.9554, "step": 126910 }, { "epoch": 0.0461312, "grad_norm": 0.7690653800964355, "learning_rate": 1.5250478459391516e-05, "loss": 3.0414, "step": 126920 }, { "epoch": 0.0461568, "grad_norm": 0.6588954925537109, "learning_rate": 1.5249793091832768e-05, "loss": 3.0775, "step": 126930 }, { "epoch": 0.0461824, "grad_norm": 0.735726535320282, "learning_rate": 1.5249107690230503e-05, "loss": 3.0505, "step": 126940 }, { "epoch": 0.046208, "grad_norm": 0.7158023118972778, "learning_rate": 1.524842225458917e-05, "loss": 3.0489, "step": 126950 }, { "epoch": 0.0462336, "grad_norm": 0.676580548286438, "learning_rate": 1.5247736784913214e-05, "loss": 3.371, "step": 126960 }, { "epoch": 0.0462592, "grad_norm": 0.7427816987037659, "learning_rate": 1.5247051281207079e-05, "loss": 2.9915, "step": 126970 }, { "epoch": 0.0462848, "grad_norm": 0.7611289024353027, "learning_rate": 1.5246365743475214e-05, "loss": 2.9849, "step": 126980 }, { "epoch": 0.0463104, "grad_norm": 0.7401059865951538, "learning_rate": 1.5245680171722056e-05, "loss": 3.0058, "step": 126990 }, { "epoch": 0.046336, "grad_norm": 0.7631123661994934, "learning_rate": 1.5244994565952063e-05, "loss": 3.0803, "step": 127000 }, { "epoch": 0.0463616, "grad_norm": 1.0205602645874023, "learning_rate": 1.524430892616967e-05, "loss": 3.0142, "step": 127010 }, { "epoch": 0.0463872, "grad_norm": 1.3621865510940552, "learning_rate": 1.5243623252379327e-05, "loss": 3.138, "step": 127020 }, { "epoch": 0.0464128, "grad_norm": 0.7352436780929565, "learning_rate": 1.5242937544585484e-05, "loss": 3.133, "step": 127030 }, { "epoch": 0.0464384, "grad_norm": 0.7455176711082458, "learning_rate": 1.5242251802792583e-05, "loss": 3.3093, "step": 127040 }, { "epoch": 0.046464, "grad_norm": 0.7908969521522522, "learning_rate": 1.524156602700507e-05, "loss": 3.0613, "step": 127050 }, { "epoch": 0.0464896, "grad_norm": 0.7441186308860779, "learning_rate": 1.5240880217227397e-05, "loss": 2.9423, "step": 127060 }, { "epoch": 0.0465152, "grad_norm": 0.7027636766433716, "learning_rate": 1.5240194373464006e-05, "loss": 3.0926, "step": 127070 }, { "epoch": 0.0465408, "grad_norm": 0.8556361794471741, "learning_rate": 1.5239508495719348e-05, "loss": 3.2428, "step": 127080 }, { "epoch": 0.0465664, "grad_norm": 0.7600985169410706, "learning_rate": 1.5238822583997873e-05, "loss": 3.0378, "step": 127090 }, { "epoch": 0.046592, "grad_norm": 0.6589226126670837, "learning_rate": 1.5238136638304025e-05, "loss": 3.1439, "step": 127100 }, { "epoch": 0.0466176, "grad_norm": 0.7852801084518433, "learning_rate": 1.523745065864225e-05, "loss": 2.9964, "step": 127110 }, { "epoch": 0.0466432, "grad_norm": 0.7858287692070007, "learning_rate": 1.5236764645017002e-05, "loss": 3.1203, "step": 127120 }, { "epoch": 0.0466688, "grad_norm": 2.719320774078369, "learning_rate": 1.5236078597432727e-05, "loss": 3.2071, "step": 127130 }, { "epoch": 0.0466944, "grad_norm": 0.7757872343063354, "learning_rate": 1.5235392515893871e-05, "loss": 3.1241, "step": 127140 }, { "epoch": 0.04672, "grad_norm": 0.7130602598190308, "learning_rate": 1.5234706400404888e-05, "loss": 3.2707, "step": 127150 }, { "epoch": 0.0467456, "grad_norm": 0.7772128582000732, "learning_rate": 1.5234020250970228e-05, "loss": 3.0637, "step": 127160 }, { "epoch": 0.0467712, "grad_norm": 0.6833972334861755, "learning_rate": 1.5233334067594332e-05, "loss": 2.9377, "step": 127170 }, { "epoch": 0.0467968, "grad_norm": 1.6080231666564941, "learning_rate": 1.5232647850281658e-05, "loss": 3.1899, "step": 127180 }, { "epoch": 0.0468224, "grad_norm": 0.7476280331611633, "learning_rate": 1.5231961599036653e-05, "loss": 3.0788, "step": 127190 }, { "epoch": 0.046848, "grad_norm": 0.858113706111908, "learning_rate": 1.5231275313863769e-05, "loss": 2.9019, "step": 127200 }, { "epoch": 0.0468736, "grad_norm": 0.7925682067871094, "learning_rate": 1.5230588994767455e-05, "loss": 2.873, "step": 127210 }, { "epoch": 0.0468992, "grad_norm": 0.6834678053855896, "learning_rate": 1.5229902641752161e-05, "loss": 3.208, "step": 127220 }, { "epoch": 0.0469248, "grad_norm": 0.6879700422286987, "learning_rate": 1.5229216254822337e-05, "loss": 3.1322, "step": 127230 }, { "epoch": 0.0469504, "grad_norm": 0.7800734043121338, "learning_rate": 1.5228529833982435e-05, "loss": 3.0831, "step": 127240 }, { "epoch": 0.046976, "grad_norm": 0.7160904407501221, "learning_rate": 1.5227843379236906e-05, "loss": 3.2033, "step": 127250 }, { "epoch": 0.0470016, "grad_norm": 0.7920892238616943, "learning_rate": 1.5227156890590207e-05, "loss": 3.0523, "step": 127260 }, { "epoch": 0.0470272, "grad_norm": 0.7569106817245483, "learning_rate": 1.5226470368046782e-05, "loss": 3.1138, "step": 127270 }, { "epoch": 0.0470528, "grad_norm": 1.02555251121521, "learning_rate": 1.5225783811611084e-05, "loss": 3.2013, "step": 127280 }, { "epoch": 0.0470784, "grad_norm": 0.7683366537094116, "learning_rate": 1.5225097221287569e-05, "loss": 2.9312, "step": 127290 }, { "epoch": 0.047104, "grad_norm": 0.7921035885810852, "learning_rate": 1.5224410597080687e-05, "loss": 2.9499, "step": 127300 }, { "epoch": 0.0471296, "grad_norm": 0.7568624019622803, "learning_rate": 1.522372393899489e-05, "loss": 3.0285, "step": 127310 }, { "epoch": 0.0471552, "grad_norm": 0.8537634611129761, "learning_rate": 1.522303724703463e-05, "loss": 2.9577, "step": 127320 }, { "epoch": 0.0471808, "grad_norm": 1.0042994022369385, "learning_rate": 1.5222350521204363e-05, "loss": 2.8897, "step": 127330 }, { "epoch": 0.0472064, "grad_norm": 0.7637802958488464, "learning_rate": 1.5221663761508541e-05, "loss": 3.0394, "step": 127340 }, { "epoch": 0.047232, "grad_norm": 0.8095718026161194, "learning_rate": 1.5220976967951617e-05, "loss": 3.1094, "step": 127350 }, { "epoch": 0.0472576, "grad_norm": 0.7067936658859253, "learning_rate": 1.5220290140538048e-05, "loss": 3.0381, "step": 127360 }, { "epoch": 0.0472832, "grad_norm": 0.7408726215362549, "learning_rate": 1.5219603279272281e-05, "loss": 2.9949, "step": 127370 }, { "epoch": 0.0473088, "grad_norm": 0.7538449764251709, "learning_rate": 1.5218916384158776e-05, "loss": 2.9384, "step": 127380 }, { "epoch": 0.0473344, "grad_norm": 0.640947699546814, "learning_rate": 1.5218229455201983e-05, "loss": 3.1759, "step": 127390 }, { "epoch": 0.04736, "grad_norm": 0.8078674077987671, "learning_rate": 1.5217542492406359e-05, "loss": 3.0327, "step": 127400 }, { "epoch": 0.0473856, "grad_norm": 0.7003697752952576, "learning_rate": 1.5216855495776362e-05, "loss": 2.9852, "step": 127410 }, { "epoch": 0.0474112, "grad_norm": 1.6295132637023926, "learning_rate": 1.521616846531644e-05, "loss": 3.176, "step": 127420 }, { "epoch": 0.0474368, "grad_norm": 0.6702320575714111, "learning_rate": 1.5215481401031054e-05, "loss": 2.9854, "step": 127430 }, { "epoch": 0.0474624, "grad_norm": 0.7479055523872375, "learning_rate": 1.5214794302924656e-05, "loss": 3.0113, "step": 127440 }, { "epoch": 0.047488, "grad_norm": 0.7225912809371948, "learning_rate": 1.5214107171001705e-05, "loss": 2.9819, "step": 127450 }, { "epoch": 0.0475136, "grad_norm": 0.670321524143219, "learning_rate": 1.5213420005266653e-05, "loss": 3.0645, "step": 127460 }, { "epoch": 0.0475392, "grad_norm": 0.7371720671653748, "learning_rate": 1.5212732805723955e-05, "loss": 2.8708, "step": 127470 }, { "epoch": 0.0475648, "grad_norm": 0.7806991338729858, "learning_rate": 1.5212045572378073e-05, "loss": 3.1308, "step": 127480 }, { "epoch": 0.0475904, "grad_norm": 0.8121463656425476, "learning_rate": 1.521135830523346e-05, "loss": 3.0682, "step": 127490 }, { "epoch": 0.047616, "grad_norm": 0.6830133199691772, "learning_rate": 1.5210671004294575e-05, "loss": 3.1453, "step": 127500 }, { "epoch": 0.0476416, "grad_norm": 1.4397999048233032, "learning_rate": 1.520998366956587e-05, "loss": 3.1449, "step": 127510 }, { "epoch": 0.0476672, "grad_norm": 0.7909718155860901, "learning_rate": 1.5209296301051812e-05, "loss": 3.0788, "step": 127520 }, { "epoch": 0.0476928, "grad_norm": 0.8183258771896362, "learning_rate": 1.5208608898756847e-05, "loss": 2.9511, "step": 127530 }, { "epoch": 0.0477184, "grad_norm": 0.7448943853378296, "learning_rate": 1.5207921462685438e-05, "loss": 2.9157, "step": 127540 }, { "epoch": 0.047744, "grad_norm": 0.891798198223114, "learning_rate": 1.5207233992842043e-05, "loss": 3.0511, "step": 127550 }, { "epoch": 0.0477696, "grad_norm": 0.7127847075462341, "learning_rate": 1.520654648923112e-05, "loss": 3.1141, "step": 127560 }, { "epoch": 0.0477952, "grad_norm": 0.730983316898346, "learning_rate": 1.5205858951857125e-05, "loss": 2.9704, "step": 127570 }, { "epoch": 0.0478208, "grad_norm": 0.8121686577796936, "learning_rate": 1.5205171380724517e-05, "loss": 2.8356, "step": 127580 }, { "epoch": 0.0478464, "grad_norm": 0.8947477340698242, "learning_rate": 1.520448377583776e-05, "loss": 3.0941, "step": 127590 }, { "epoch": 0.047872, "grad_norm": 0.723112940788269, "learning_rate": 1.5203796137201308e-05, "loss": 3.018, "step": 127600 }, { "epoch": 0.0478976, "grad_norm": 0.7306171655654907, "learning_rate": 1.5203108464819618e-05, "loss": 3.0225, "step": 127610 }, { "epoch": 0.0479232, "grad_norm": 0.7024168372154236, "learning_rate": 1.5202420758697157e-05, "loss": 3.136, "step": 127620 }, { "epoch": 0.0479488, "grad_norm": 0.7281426787376404, "learning_rate": 1.5201733018838378e-05, "loss": 3.0327, "step": 127630 }, { "epoch": 0.0479744, "grad_norm": 0.7646895051002502, "learning_rate": 1.5201045245247743e-05, "loss": 2.9901, "step": 127640 }, { "epoch": 0.048, "grad_norm": 0.8577759265899658, "learning_rate": 1.520035743792971e-05, "loss": 3.1661, "step": 127650 }, { "epoch": 0.0480256, "grad_norm": 0.7077302932739258, "learning_rate": 1.5199669596888742e-05, "loss": 3.2488, "step": 127660 }, { "epoch": 0.0480512, "grad_norm": 0.7942055463790894, "learning_rate": 1.5198981722129297e-05, "loss": 3.2167, "step": 127670 }, { "epoch": 0.0480768, "grad_norm": 0.7985394597053528, "learning_rate": 1.519829381365584e-05, "loss": 3.2922, "step": 127680 }, { "epoch": 0.0481024, "grad_norm": 0.8875015377998352, "learning_rate": 1.5197605871472831e-05, "loss": 3.1467, "step": 127690 }, { "epoch": 0.048128, "grad_norm": 0.7644611597061157, "learning_rate": 1.5196917895584723e-05, "loss": 2.9542, "step": 127700 }, { "epoch": 0.0481536, "grad_norm": 0.7872587442398071, "learning_rate": 1.5196229885995988e-05, "loss": 3.0623, "step": 127710 }, { "epoch": 0.0481792, "grad_norm": 0.6969557404518127, "learning_rate": 1.5195541842711086e-05, "loss": 2.9833, "step": 127720 }, { "epoch": 0.0482048, "grad_norm": 0.7362402677536011, "learning_rate": 1.5194853765734471e-05, "loss": 2.9227, "step": 127730 }, { "epoch": 0.0482304, "grad_norm": 0.779259979724884, "learning_rate": 1.5194165655070611e-05, "loss": 3.11, "step": 127740 }, { "epoch": 0.048256, "grad_norm": 0.7339951992034912, "learning_rate": 1.5193477510723967e-05, "loss": 2.832, "step": 127750 }, { "epoch": 0.0482816, "grad_norm": 0.8008157014846802, "learning_rate": 1.5192789332699003e-05, "loss": 2.9296, "step": 127760 }, { "epoch": 0.0483072, "grad_norm": 0.7646434903144836, "learning_rate": 1.519210112100018e-05, "loss": 3.0076, "step": 127770 }, { "epoch": 0.0483328, "grad_norm": 1.2721853256225586, "learning_rate": 1.5191412875631959e-05, "loss": 2.8181, "step": 127780 }, { "epoch": 0.0483584, "grad_norm": 0.8667572736740112, "learning_rate": 1.5190724596598809e-05, "loss": 3.4354, "step": 127790 }, { "epoch": 0.048384, "grad_norm": 0.8460628390312195, "learning_rate": 1.5190036283905185e-05, "loss": 3.017, "step": 127800 }, { "epoch": 0.0484096, "grad_norm": 0.6287282705307007, "learning_rate": 1.5189347937555559e-05, "loss": 2.7553, "step": 127810 }, { "epoch": 0.0484352, "grad_norm": 0.6499795317649841, "learning_rate": 1.5188659557554388e-05, "loss": 2.9109, "step": 127820 }, { "epoch": 0.0484608, "grad_norm": 0.7675598859786987, "learning_rate": 1.5187971143906142e-05, "loss": 3.0853, "step": 127830 }, { "epoch": 0.0484864, "grad_norm": 0.6801338195800781, "learning_rate": 1.5187282696615278e-05, "loss": 3.1433, "step": 127840 }, { "epoch": 0.048512, "grad_norm": 0.6656932830810547, "learning_rate": 1.5186594215686267e-05, "loss": 3.1172, "step": 127850 }, { "epoch": 0.0485376, "grad_norm": 0.7867270708084106, "learning_rate": 1.5185905701123569e-05, "loss": 3.2302, "step": 127860 }, { "epoch": 0.0485632, "grad_norm": 0.6665933132171631, "learning_rate": 1.5185217152931651e-05, "loss": 3.0852, "step": 127870 }, { "epoch": 0.0485888, "grad_norm": 0.8122178912162781, "learning_rate": 1.5184528571114978e-05, "loss": 3.1666, "step": 127880 }, { "epoch": 0.0486144, "grad_norm": 0.7259637117385864, "learning_rate": 1.5183839955678016e-05, "loss": 3.0416, "step": 127890 }, { "epoch": 0.04864, "grad_norm": 0.7795698642730713, "learning_rate": 1.518315130662523e-05, "loss": 3.0861, "step": 127900 }, { "epoch": 0.0486656, "grad_norm": 0.9764823913574219, "learning_rate": 1.5182462623961085e-05, "loss": 2.9029, "step": 127910 }, { "epoch": 0.0486912, "grad_norm": 0.7324492335319519, "learning_rate": 1.5181773907690048e-05, "loss": 3.1117, "step": 127920 }, { "epoch": 0.0487168, "grad_norm": 0.8411350846290588, "learning_rate": 1.5181085157816582e-05, "loss": 3.0304, "step": 127930 }, { "epoch": 0.0487424, "grad_norm": 0.7054486274719238, "learning_rate": 1.5180396374345156e-05, "loss": 3.2109, "step": 127940 }, { "epoch": 0.048768, "grad_norm": 0.6799852252006531, "learning_rate": 1.5179707557280239e-05, "loss": 2.8932, "step": 127950 }, { "epoch": 0.0487936, "grad_norm": 0.978777289390564, "learning_rate": 1.5179018706626293e-05, "loss": 3.1352, "step": 127960 }, { "epoch": 0.0488192, "grad_norm": 0.8712519407272339, "learning_rate": 1.5178329822387788e-05, "loss": 3.021, "step": 127970 }, { "epoch": 0.0488448, "grad_norm": 0.7654715180397034, "learning_rate": 1.5177640904569188e-05, "loss": 3.0382, "step": 127980 }, { "epoch": 0.0488704, "grad_norm": 0.7391760349273682, "learning_rate": 1.5176951953174968e-05, "loss": 2.9833, "step": 127990 }, { "epoch": 0.048896, "grad_norm": 0.6940836906433105, "learning_rate": 1.5176262968209586e-05, "loss": 2.9883, "step": 128000 }, { "epoch": 0.0489216, "grad_norm": 0.6823523044586182, "learning_rate": 1.5175573949677513e-05, "loss": 3.2611, "step": 128010 }, { "epoch": 0.0489472, "grad_norm": 0.7398691773414612, "learning_rate": 1.5174884897583222e-05, "loss": 2.845, "step": 128020 }, { "epoch": 0.0489728, "grad_norm": 0.7626541256904602, "learning_rate": 1.5174195811931176e-05, "loss": 3.0305, "step": 128030 }, { "epoch": 0.0489984, "grad_norm": 0.6772891879081726, "learning_rate": 1.5173506692725846e-05, "loss": 2.9183, "step": 128040 }, { "epoch": 0.049024, "grad_norm": 0.7081247568130493, "learning_rate": 1.5172817539971703e-05, "loss": 3.1977, "step": 128050 }, { "epoch": 0.0490496, "grad_norm": 0.7892815470695496, "learning_rate": 1.5172128353673208e-05, "loss": 3.0727, "step": 128060 }, { "epoch": 0.0490752, "grad_norm": 0.9121196269989014, "learning_rate": 1.5171439133834836e-05, "loss": 3.0664, "step": 128070 }, { "epoch": 0.0491008, "grad_norm": 0.7488983273506165, "learning_rate": 1.5170749880461058e-05, "loss": 3.0876, "step": 128080 }, { "epoch": 0.0491264, "grad_norm": 0.9819818735122681, "learning_rate": 1.5170060593556342e-05, "loss": 3.1617, "step": 128090 }, { "epoch": 0.049152, "grad_norm": 0.9635878205299377, "learning_rate": 1.5169371273125154e-05, "loss": 3.0644, "step": 128100 }, { "epoch": 0.0491776, "grad_norm": 0.7029449343681335, "learning_rate": 1.5168681919171966e-05, "loss": 3.2018, "step": 128110 }, { "epoch": 0.0492032, "grad_norm": 0.7992034554481506, "learning_rate": 1.5167992531701252e-05, "loss": 2.971, "step": 128120 }, { "epoch": 0.0492288, "grad_norm": 0.733568012714386, "learning_rate": 1.516730311071748e-05, "loss": 3.0231, "step": 128130 }, { "epoch": 0.0492544, "grad_norm": 0.8665330410003662, "learning_rate": 1.516661365622512e-05, "loss": 3.0957, "step": 128140 }, { "epoch": 0.04928, "grad_norm": 0.7298395037651062, "learning_rate": 1.5165924168228645e-05, "loss": 3.1861, "step": 128150 }, { "epoch": 0.0493056, "grad_norm": 0.6652483344078064, "learning_rate": 1.5165234646732523e-05, "loss": 3.0617, "step": 128160 }, { "epoch": 0.0493312, "grad_norm": 2.010697841644287, "learning_rate": 1.5164545091741227e-05, "loss": 3.0233, "step": 128170 }, { "epoch": 0.0493568, "grad_norm": 0.792352020740509, "learning_rate": 1.5163855503259227e-05, "loss": 3.2783, "step": 128180 }, { "epoch": 0.0493824, "grad_norm": 0.9165970087051392, "learning_rate": 1.5163165881291e-05, "loss": 2.9745, "step": 128190 }, { "epoch": 0.049408, "grad_norm": 1.0061707496643066, "learning_rate": 1.5162476225841014e-05, "loss": 2.9923, "step": 128200 }, { "epoch": 0.0494336, "grad_norm": 0.7132822275161743, "learning_rate": 1.516178653691374e-05, "loss": 2.9453, "step": 128210 }, { "epoch": 0.0494592, "grad_norm": 0.716261625289917, "learning_rate": 1.5161096814513654e-05, "loss": 3.0708, "step": 128220 }, { "epoch": 0.0494848, "grad_norm": 0.7522410750389099, "learning_rate": 1.5160407058645226e-05, "loss": 2.8242, "step": 128230 }, { "epoch": 0.0495104, "grad_norm": 0.7710448503494263, "learning_rate": 1.5159717269312931e-05, "loss": 2.9328, "step": 128240 }, { "epoch": 0.049536, "grad_norm": 0.7386015057563782, "learning_rate": 1.515902744652124e-05, "loss": 2.9829, "step": 128250 }, { "epoch": 0.0495616, "grad_norm": 0.8158722519874573, "learning_rate": 1.5158337590274626e-05, "loss": 2.9657, "step": 128260 }, { "epoch": 0.0495872, "grad_norm": 0.703781008720398, "learning_rate": 1.5157647700577563e-05, "loss": 2.9874, "step": 128270 }, { "epoch": 0.0496128, "grad_norm": 0.9768350124359131, "learning_rate": 1.5156957777434525e-05, "loss": 2.8946, "step": 128280 }, { "epoch": 0.0496384, "grad_norm": 0.7950112223625183, "learning_rate": 1.5156267820849989e-05, "loss": 2.9293, "step": 128290 }, { "epoch": 0.049664, "grad_norm": 0.7314438223838806, "learning_rate": 1.5155577830828426e-05, "loss": 2.9911, "step": 128300 }, { "epoch": 0.0496896, "grad_norm": 0.8890763521194458, "learning_rate": 1.5154887807374308e-05, "loss": 2.9396, "step": 128310 }, { "epoch": 0.0497152, "grad_norm": 0.744499921798706, "learning_rate": 1.5154197750492116e-05, "loss": 3.1318, "step": 128320 }, { "epoch": 0.0497408, "grad_norm": 0.7395046949386597, "learning_rate": 1.515350766018632e-05, "loss": 2.9546, "step": 128330 }, { "epoch": 0.0497664, "grad_norm": 0.6724340319633484, "learning_rate": 1.5152817536461397e-05, "loss": 2.8242, "step": 128340 }, { "epoch": 0.049792, "grad_norm": 0.783562958240509, "learning_rate": 1.5152127379321824e-05, "loss": 3.1074, "step": 128350 }, { "epoch": 0.0498176, "grad_norm": 0.8551502823829651, "learning_rate": 1.515143718877207e-05, "loss": 2.9853, "step": 128360 }, { "epoch": 0.0498432, "grad_norm": 0.7654721140861511, "learning_rate": 1.5150746964816615e-05, "loss": 3.0449, "step": 128370 }, { "epoch": 0.0498688, "grad_norm": 0.7044587731361389, "learning_rate": 1.515005670745994e-05, "loss": 3.0724, "step": 128380 }, { "epoch": 0.0498944, "grad_norm": 0.8207073211669922, "learning_rate": 1.514936641670651e-05, "loss": 3.0771, "step": 128390 }, { "epoch": 0.04992, "grad_norm": 2.0474438667297363, "learning_rate": 1.514867609256081e-05, "loss": 3.3396, "step": 128400 }, { "epoch": 0.0499456, "grad_norm": 0.9293689131736755, "learning_rate": 1.5147985735027313e-05, "loss": 3.2654, "step": 128410 }, { "epoch": 0.0499712, "grad_norm": 0.7881788015365601, "learning_rate": 1.5147295344110497e-05, "loss": 3.0344, "step": 128420 }, { "epoch": 0.0499968, "grad_norm": 0.7620229125022888, "learning_rate": 1.514660491981484e-05, "loss": 3.2507, "step": 128430 }, { "epoch": 0.0500224, "grad_norm": 0.8260372877120972, "learning_rate": 1.5145914462144817e-05, "loss": 2.8472, "step": 128440 }, { "epoch": 0.050048, "grad_norm": 0.856809139251709, "learning_rate": 1.5145223971104905e-05, "loss": 3.1961, "step": 128450 }, { "epoch": 0.0500736, "grad_norm": 0.7051054239273071, "learning_rate": 1.5144533446699585e-05, "loss": 3.1366, "step": 128460 }, { "epoch": 0.0500992, "grad_norm": 0.800512969493866, "learning_rate": 1.5143842888933332e-05, "loss": 3.1011, "step": 128470 }, { "epoch": 0.0501248, "grad_norm": 0.7831434011459351, "learning_rate": 1.5143152297810626e-05, "loss": 2.9452, "step": 128480 }, { "epoch": 0.0501504, "grad_norm": 0.7191097736358643, "learning_rate": 1.5142461673335942e-05, "loss": 3.1512, "step": 128490 }, { "epoch": 0.050176, "grad_norm": 0.7817112803459167, "learning_rate": 1.5141771015513764e-05, "loss": 2.995, "step": 128500 }, { "epoch": 0.0502016, "grad_norm": 0.7218928337097168, "learning_rate": 1.5141080324348565e-05, "loss": 3.21, "step": 128510 }, { "epoch": 0.0502272, "grad_norm": 0.785407543182373, "learning_rate": 1.5140389599844826e-05, "loss": 3.1121, "step": 128520 }, { "epoch": 0.0502528, "grad_norm": 0.6457892060279846, "learning_rate": 1.5139698842007026e-05, "loss": 2.9934, "step": 128530 }, { "epoch": 0.0502784, "grad_norm": 0.7265588641166687, "learning_rate": 1.513900805083965e-05, "loss": 3.0245, "step": 128540 }, { "epoch": 0.050304, "grad_norm": 0.7800309658050537, "learning_rate": 1.5138317226347168e-05, "loss": 3.0401, "step": 128550 }, { "epoch": 0.0503296, "grad_norm": 0.6953880786895752, "learning_rate": 1.5137626368534064e-05, "loss": 3.0457, "step": 128560 }, { "epoch": 0.0503552, "grad_norm": 1.247959852218628, "learning_rate": 1.513693547740482e-05, "loss": 3.3198, "step": 128570 }, { "epoch": 0.0503808, "grad_norm": 0.756116509437561, "learning_rate": 1.5136244552963914e-05, "loss": 2.8511, "step": 128580 }, { "epoch": 0.0504064, "grad_norm": 0.897268533706665, "learning_rate": 1.513555359521583e-05, "loss": 2.8455, "step": 128590 }, { "epoch": 0.050432, "grad_norm": 0.9057093262672424, "learning_rate": 1.5134862604165044e-05, "loss": 3.2022, "step": 128600 }, { "epoch": 0.0504576, "grad_norm": 0.7116617560386658, "learning_rate": 1.5134171579816039e-05, "loss": 3.1726, "step": 128610 }, { "epoch": 0.0504832, "grad_norm": 0.9510066509246826, "learning_rate": 1.5133480522173291e-05, "loss": 3.0146, "step": 128620 }, { "epoch": 0.0505088, "grad_norm": 0.774947464466095, "learning_rate": 1.513278943124129e-05, "loss": 2.9604, "step": 128630 }, { "epoch": 0.0505344, "grad_norm": 0.6251006722450256, "learning_rate": 1.5132098307024517e-05, "loss": 2.9424, "step": 128640 }, { "epoch": 0.05056, "grad_norm": 0.79659104347229, "learning_rate": 1.5131407149527446e-05, "loss": 3.1831, "step": 128650 }, { "epoch": 0.0505856, "grad_norm": 0.9589307904243469, "learning_rate": 1.5130715958754562e-05, "loss": 3.248, "step": 128660 }, { "epoch": 0.0506112, "grad_norm": 0.7904630899429321, "learning_rate": 1.5130024734710354e-05, "loss": 3.2034, "step": 128670 }, { "epoch": 0.0506368, "grad_norm": 0.7406290769577026, "learning_rate": 1.5129333477399297e-05, "loss": 3.0468, "step": 128680 }, { "epoch": 0.0506624, "grad_norm": 0.8129403591156006, "learning_rate": 1.5128642186825872e-05, "loss": 3.0618, "step": 128690 }, { "epoch": 0.050688, "grad_norm": 0.7797231078147888, "learning_rate": 1.5127950862994571e-05, "loss": 3.1189, "step": 128700 }, { "epoch": 0.0507136, "grad_norm": 0.7310222387313843, "learning_rate": 1.5127259505909867e-05, "loss": 3.1311, "step": 128710 }, { "epoch": 0.0507392, "grad_norm": 0.6855611205101013, "learning_rate": 1.512656811557625e-05, "loss": 3.2409, "step": 128720 }, { "epoch": 0.0507648, "grad_norm": 0.9627186059951782, "learning_rate": 1.5125876691998201e-05, "loss": 3.1426, "step": 128730 }, { "epoch": 0.0507904, "grad_norm": 0.6419841051101685, "learning_rate": 1.5125185235180205e-05, "loss": 3.1267, "step": 128740 }, { "epoch": 0.050816, "grad_norm": 1.2006494998931885, "learning_rate": 1.5124493745126742e-05, "loss": 3.032, "step": 128750 }, { "epoch": 0.0508416, "grad_norm": 0.9162566065788269, "learning_rate": 1.51238022218423e-05, "loss": 3.0344, "step": 128760 }, { "epoch": 0.0508672, "grad_norm": 0.9487365484237671, "learning_rate": 1.5123110665331365e-05, "loss": 2.9894, "step": 128770 }, { "epoch": 0.0508928, "grad_norm": 1.2830326557159424, "learning_rate": 1.5122419075598415e-05, "loss": 3.1203, "step": 128780 }, { "epoch": 0.0509184, "grad_norm": 0.7672938704490662, "learning_rate": 1.5121727452647941e-05, "loss": 3.1062, "step": 128790 }, { "epoch": 0.050944, "grad_norm": 0.6960640549659729, "learning_rate": 1.5121035796484425e-05, "loss": 3.0218, "step": 128800 }, { "epoch": 0.0509696, "grad_norm": 0.7437350749969482, "learning_rate": 1.5120344107112353e-05, "loss": 3.0491, "step": 128810 }, { "epoch": 0.0509952, "grad_norm": 0.6831632852554321, "learning_rate": 1.511965238453621e-05, "loss": 3.2571, "step": 128820 }, { "epoch": 0.0510208, "grad_norm": 0.7932926416397095, "learning_rate": 1.5118960628760483e-05, "loss": 3.1049, "step": 128830 }, { "epoch": 0.0510464, "grad_norm": 0.6770755648612976, "learning_rate": 1.5118268839789657e-05, "loss": 3.1821, "step": 128840 }, { "epoch": 0.051072, "grad_norm": 0.8825058341026306, "learning_rate": 1.5117577017628215e-05, "loss": 3.2655, "step": 128850 }, { "epoch": 0.0510976, "grad_norm": 0.7884690165519714, "learning_rate": 1.5116885162280649e-05, "loss": 2.9243, "step": 128860 }, { "epoch": 0.0511232, "grad_norm": 0.6377168297767639, "learning_rate": 1.511619327375144e-05, "loss": 3.0474, "step": 128870 }, { "epoch": 0.0511488, "grad_norm": 0.7430493235588074, "learning_rate": 1.511550135204508e-05, "loss": 3.2069, "step": 128880 }, { "epoch": 0.0511744, "grad_norm": 0.8099373579025269, "learning_rate": 1.5114809397166052e-05, "loss": 3.6338, "step": 128890 }, { "epoch": 0.0512, "grad_norm": 0.7103743553161621, "learning_rate": 1.5114117409118843e-05, "loss": 3.7729, "step": 128900 }, { "epoch": 0.0512256, "grad_norm": 0.7138640880584717, "learning_rate": 1.5113425387907946e-05, "loss": 3.1167, "step": 128910 }, { "epoch": 0.0512512, "grad_norm": 0.6577547788619995, "learning_rate": 1.511273333353784e-05, "loss": 3.0288, "step": 128920 }, { "epoch": 0.0512768, "grad_norm": 0.924614429473877, "learning_rate": 1.511204124601302e-05, "loss": 3.1462, "step": 128930 }, { "epoch": 0.0513024, "grad_norm": 0.8604391813278198, "learning_rate": 1.511134912533797e-05, "loss": 2.9167, "step": 128940 }, { "epoch": 0.051328, "grad_norm": 0.8777470588684082, "learning_rate": 1.5110656971517179e-05, "loss": 3.422, "step": 128950 }, { "epoch": 0.0513536, "grad_norm": 0.7308458089828491, "learning_rate": 1.5109964784555136e-05, "loss": 3.0344, "step": 128960 }, { "epoch": 0.0513792, "grad_norm": 0.8037799596786499, "learning_rate": 1.510927256445633e-05, "loss": 3.3041, "step": 128970 }, { "epoch": 0.0514048, "grad_norm": 0.7866966128349304, "learning_rate": 1.510858031122525e-05, "loss": 3.2811, "step": 128980 }, { "epoch": 0.0514304, "grad_norm": 0.7857362627983093, "learning_rate": 1.5107888024866379e-05, "loss": 3.1242, "step": 128990 }, { "epoch": 0.051456, "grad_norm": 0.7566019892692566, "learning_rate": 1.5107195705384218e-05, "loss": 2.9643, "step": 129000 }, { "epoch": 0.0514816, "grad_norm": 0.6638773083686829, "learning_rate": 1.5106503352783248e-05, "loss": 3.2251, "step": 129010 }, { "epoch": 0.0515072, "grad_norm": 0.8633416891098022, "learning_rate": 1.5105810967067962e-05, "loss": 3.0435, "step": 129020 }, { "epoch": 0.0515328, "grad_norm": 0.6731061935424805, "learning_rate": 1.5105118548242847e-05, "loss": 3.0437, "step": 129030 }, { "epoch": 0.0515584, "grad_norm": 0.7834264636039734, "learning_rate": 1.5104426096312396e-05, "loss": 3.0218, "step": 129040 }, { "epoch": 0.051584, "grad_norm": 0.776831328868866, "learning_rate": 1.5103733611281097e-05, "loss": 3.0418, "step": 129050 }, { "epoch": 0.0516096, "grad_norm": 0.7637120485305786, "learning_rate": 1.5103041093153443e-05, "loss": 3.0435, "step": 129060 }, { "epoch": 0.0516352, "grad_norm": 0.6933563351631165, "learning_rate": 1.5102348541933927e-05, "loss": 2.8258, "step": 129070 }, { "epoch": 0.0516608, "grad_norm": 0.6624906063079834, "learning_rate": 1.5101655957627032e-05, "loss": 3.049, "step": 129080 }, { "epoch": 0.0516864, "grad_norm": 1.474596381187439, "learning_rate": 1.5100963340237255e-05, "loss": 3.9101, "step": 129090 }, { "epoch": 0.051712, "grad_norm": 0.7767882347106934, "learning_rate": 1.5100270689769088e-05, "loss": 3.1562, "step": 129100 }, { "epoch": 0.0517376, "grad_norm": 0.6175568103790283, "learning_rate": 1.509957800622702e-05, "loss": 2.8161, "step": 129110 }, { "epoch": 0.0517632, "grad_norm": 0.7842872142791748, "learning_rate": 1.5098885289615543e-05, "loss": 3.0986, "step": 129120 }, { "epoch": 0.0517888, "grad_norm": 0.7342828512191772, "learning_rate": 1.5098192539939152e-05, "loss": 2.9779, "step": 129130 }, { "epoch": 0.0518144, "grad_norm": 0.8051470518112183, "learning_rate": 1.5097499757202336e-05, "loss": 3.0118, "step": 129140 }, { "epoch": 0.05184, "grad_norm": 1.0362166166305542, "learning_rate": 1.509680694140959e-05, "loss": 3.0635, "step": 129150 }, { "epoch": 0.0518656, "grad_norm": 1.5559526681900024, "learning_rate": 1.5096114092565403e-05, "loss": 3.0176, "step": 129160 }, { "epoch": 0.0518912, "grad_norm": 2.9464516639709473, "learning_rate": 1.5095421210674275e-05, "loss": 3.3633, "step": 129170 }, { "epoch": 0.0519168, "grad_norm": 1.4319639205932617, "learning_rate": 1.509472829574069e-05, "loss": 3.3583, "step": 129180 }, { "epoch": 0.0519424, "grad_norm": 0.7180718779563904, "learning_rate": 1.5094035347769147e-05, "loss": 3.1077, "step": 129190 }, { "epoch": 0.051968, "grad_norm": 0.8133378028869629, "learning_rate": 1.5093342366764141e-05, "loss": 3.2921, "step": 129200 }, { "epoch": 0.0519936, "grad_norm": 0.8013813495635986, "learning_rate": 1.5092649352730165e-05, "loss": 3.5971, "step": 129210 }, { "epoch": 0.0520192, "grad_norm": 4.2014665603637695, "learning_rate": 1.5091956305671707e-05, "loss": 3.2594, "step": 129220 }, { "epoch": 0.0520448, "grad_norm": 0.7029136419296265, "learning_rate": 1.5091263225593268e-05, "loss": 3.2199, "step": 129230 }, { "epoch": 0.0520704, "grad_norm": 0.6732560992240906, "learning_rate": 1.5090570112499337e-05, "loss": 3.1879, "step": 129240 }, { "epoch": 0.052096, "grad_norm": 1.0688587427139282, "learning_rate": 1.5089876966394413e-05, "loss": 3.1839, "step": 129250 }, { "epoch": 0.0521216, "grad_norm": 0.7955114841461182, "learning_rate": 1.508918378728299e-05, "loss": 3.0866, "step": 129260 }, { "epoch": 0.0521472, "grad_norm": 0.664753794670105, "learning_rate": 1.5088490575169563e-05, "loss": 3.1667, "step": 129270 }, { "epoch": 0.0521728, "grad_norm": 0.7473957538604736, "learning_rate": 1.5087797330058629e-05, "loss": 3.14, "step": 129280 }, { "epoch": 0.0521984, "grad_norm": 0.685917854309082, "learning_rate": 1.5087104051954678e-05, "loss": 3.1167, "step": 129290 }, { "epoch": 0.052224, "grad_norm": 0.8260008096694946, "learning_rate": 1.5086410740862211e-05, "loss": 3.0741, "step": 129300 }, { "epoch": 0.0522496, "grad_norm": 0.7859379053115845, "learning_rate": 1.5085717396785722e-05, "loss": 3.0484, "step": 129310 }, { "epoch": 0.0522752, "grad_norm": 0.7225761413574219, "learning_rate": 1.5085024019729706e-05, "loss": 3.2154, "step": 129320 }, { "epoch": 0.0523008, "grad_norm": 0.7035468816757202, "learning_rate": 1.5084330609698663e-05, "loss": 2.7904, "step": 129330 }, { "epoch": 0.0523264, "grad_norm": 0.7655560970306396, "learning_rate": 1.5083637166697087e-05, "loss": 3.0042, "step": 129340 }, { "epoch": 0.052352, "grad_norm": 0.7679634094238281, "learning_rate": 1.5082943690729473e-05, "loss": 3.1304, "step": 129350 }, { "epoch": 0.0523776, "grad_norm": 1.3242560625076294, "learning_rate": 1.5082250181800322e-05, "loss": 3.3858, "step": 129360 }, { "epoch": 0.0524032, "grad_norm": 0.9689163565635681, "learning_rate": 1.508155663991413e-05, "loss": 2.9128, "step": 129370 }, { "epoch": 0.0524288, "grad_norm": 0.7470305562019348, "learning_rate": 1.508086306507539e-05, "loss": 3.0509, "step": 129380 }, { "epoch": 0.0524544, "grad_norm": 0.7461903095245361, "learning_rate": 1.5080169457288608e-05, "loss": 3.0808, "step": 129390 }, { "epoch": 0.05248, "grad_norm": 0.798265278339386, "learning_rate": 1.5079475816558274e-05, "loss": 3.0801, "step": 129400 }, { "epoch": 0.0525056, "grad_norm": 0.6873827576637268, "learning_rate": 1.5078782142888888e-05, "loss": 2.9451, "step": 129410 }, { "epoch": 0.0525312, "grad_norm": 0.7073207497596741, "learning_rate": 1.507808843628495e-05, "loss": 3.057, "step": 129420 }, { "epoch": 0.0525568, "grad_norm": 0.8449299335479736, "learning_rate": 1.5077394696750962e-05, "loss": 3.2687, "step": 129430 }, { "epoch": 0.0525824, "grad_norm": 0.9305628538131714, "learning_rate": 1.5076700924291416e-05, "loss": 3.2575, "step": 129440 }, { "epoch": 0.052608, "grad_norm": 0.9028809070587158, "learning_rate": 1.5076007118910814e-05, "loss": 3.1297, "step": 129450 }, { "epoch": 0.0526336, "grad_norm": 0.6640967130661011, "learning_rate": 1.5075313280613655e-05, "loss": 2.9969, "step": 129460 }, { "epoch": 0.0526592, "grad_norm": 0.9836876392364502, "learning_rate": 1.507461940940444e-05, "loss": 2.9954, "step": 129470 }, { "epoch": 0.0526848, "grad_norm": 0.9016789197921753, "learning_rate": 1.5073925505287665e-05, "loss": 3.0203, "step": 129480 }, { "epoch": 0.0527104, "grad_norm": 0.9482855200767517, "learning_rate": 1.5073231568267828e-05, "loss": 3.1116, "step": 129490 }, { "epoch": 0.052736, "grad_norm": 0.7206461429595947, "learning_rate": 1.5072537598349439e-05, "loss": 3.0224, "step": 129500 }, { "epoch": 0.0527616, "grad_norm": 1.3102136850357056, "learning_rate": 1.507184359553699e-05, "loss": 2.9722, "step": 129510 }, { "epoch": 0.0527872, "grad_norm": 1.1730670928955078, "learning_rate": 1.5071149559834979e-05, "loss": 3.0126, "step": 129520 }, { "epoch": 0.0528128, "grad_norm": 0.7554535269737244, "learning_rate": 1.5070455491247918e-05, "loss": 3.0642, "step": 129530 }, { "epoch": 0.0528384, "grad_norm": 0.8501593470573425, "learning_rate": 1.5069761389780296e-05, "loss": 3.1463, "step": 129540 }, { "epoch": 0.052864, "grad_norm": 0.7129285335540771, "learning_rate": 1.506906725543662e-05, "loss": 2.9635, "step": 129550 }, { "epoch": 0.0528896, "grad_norm": 0.6954265236854553, "learning_rate": 1.506837308822139e-05, "loss": 2.8193, "step": 129560 }, { "epoch": 0.0529152, "grad_norm": 0.7219118475914001, "learning_rate": 1.5067678888139105e-05, "loss": 3.424, "step": 129570 }, { "epoch": 0.0529408, "grad_norm": 0.6846635937690735, "learning_rate": 1.5066984655194272e-05, "loss": 2.8245, "step": 129580 }, { "epoch": 0.0529664, "grad_norm": 0.7000362873077393, "learning_rate": 1.506629038939139e-05, "loss": 3.007, "step": 129590 }, { "epoch": 0.052992, "grad_norm": 0.6389757394790649, "learning_rate": 1.506559609073496e-05, "loss": 3.0677, "step": 129600 }, { "epoch": 0.0530176, "grad_norm": 0.8232146501541138, "learning_rate": 1.5064901759229486e-05, "loss": 3.1489, "step": 129610 }, { "epoch": 0.0530432, "grad_norm": 0.7589119672775269, "learning_rate": 1.5064207394879474e-05, "loss": 3.1383, "step": 129620 }, { "epoch": 0.0530688, "grad_norm": 0.7328688502311707, "learning_rate": 1.5063512997689417e-05, "loss": 3.0277, "step": 129630 }, { "epoch": 0.0530944, "grad_norm": 0.6414045095443726, "learning_rate": 1.506281856766383e-05, "loss": 3.0124, "step": 129640 }, { "epoch": 0.05312, "grad_norm": 2.6041784286499023, "learning_rate": 1.5062124104807203e-05, "loss": 2.9211, "step": 129650 }, { "epoch": 0.0531456, "grad_norm": 0.796134889125824, "learning_rate": 1.506142960912405e-05, "loss": 2.9496, "step": 129660 }, { "epoch": 0.0531712, "grad_norm": 1.3694161176681519, "learning_rate": 1.5060735080618872e-05, "loss": 2.8457, "step": 129670 }, { "epoch": 0.0531968, "grad_norm": 0.6744800806045532, "learning_rate": 1.506004051929617e-05, "loss": 3.1324, "step": 129680 }, { "epoch": 0.0532224, "grad_norm": 0.7204030156135559, "learning_rate": 1.505934592516045e-05, "loss": 3.0472, "step": 129690 }, { "epoch": 0.053248, "grad_norm": 0.8034929633140564, "learning_rate": 1.5058651298216221e-05, "loss": 2.9407, "step": 129700 }, { "epoch": 0.0532736, "grad_norm": 0.8426151871681213, "learning_rate": 1.505795663846798e-05, "loss": 2.9722, "step": 129710 }, { "epoch": 0.0532992, "grad_norm": 0.7573427557945251, "learning_rate": 1.5057261945920231e-05, "loss": 3.0786, "step": 129720 }, { "epoch": 0.0533248, "grad_norm": 0.7868319153785706, "learning_rate": 1.5056567220577489e-05, "loss": 2.9146, "step": 129730 }, { "epoch": 0.0533504, "grad_norm": 1.3486852645874023, "learning_rate": 1.5055872462444247e-05, "loss": 3.2334, "step": 129740 }, { "epoch": 0.053376, "grad_norm": 1.5440126657485962, "learning_rate": 1.5055177671525017e-05, "loss": 3.2533, "step": 129750 }, { "epoch": 0.0534016, "grad_norm": 0.6509987115859985, "learning_rate": 1.5054482847824305e-05, "loss": 2.8632, "step": 129760 }, { "epoch": 0.0534272, "grad_norm": 0.866673469543457, "learning_rate": 1.5053787991346613e-05, "loss": 3.1001, "step": 129770 }, { "epoch": 0.0534528, "grad_norm": 0.7645587921142578, "learning_rate": 1.505309310209645e-05, "loss": 3.0176, "step": 129780 }, { "epoch": 0.0534784, "grad_norm": 0.7641358971595764, "learning_rate": 1.505239818007832e-05, "loss": 3.1249, "step": 129790 }, { "epoch": 0.053504, "grad_norm": 0.7745864391326904, "learning_rate": 1.5051703225296734e-05, "loss": 3.0472, "step": 129800 }, { "epoch": 0.0535296, "grad_norm": 0.6948561072349548, "learning_rate": 1.5051008237756191e-05, "loss": 2.9219, "step": 129810 }, { "epoch": 0.0535552, "grad_norm": 0.8573179841041565, "learning_rate": 1.5050313217461204e-05, "loss": 3.2224, "step": 129820 }, { "epoch": 0.0535808, "grad_norm": 0.833448052406311, "learning_rate": 1.504961816441628e-05, "loss": 2.8814, "step": 129830 }, { "epoch": 0.0536064, "grad_norm": 0.7740306854248047, "learning_rate": 1.5048923078625918e-05, "loss": 3.1845, "step": 129840 }, { "epoch": 0.053632, "grad_norm": 0.7965713143348694, "learning_rate": 1.5048227960094637e-05, "loss": 3.0823, "step": 129850 }, { "epoch": 0.0536576, "grad_norm": 0.725566565990448, "learning_rate": 1.5047532808826939e-05, "loss": 3.305, "step": 129860 }, { "epoch": 0.0536832, "grad_norm": 0.7057721018791199, "learning_rate": 1.504683762482733e-05, "loss": 3.4345, "step": 129870 }, { "epoch": 0.0537088, "grad_norm": 0.7975460886955261, "learning_rate": 1.5046142408100321e-05, "loss": 3.1805, "step": 129880 }, { "epoch": 0.0537344, "grad_norm": 1.8185150623321533, "learning_rate": 1.5045447158650422e-05, "loss": 3.1335, "step": 129890 }, { "epoch": 0.05376, "grad_norm": 0.7134320139884949, "learning_rate": 1.5044751876482138e-05, "loss": 3.1544, "step": 129900 }, { "epoch": 0.0537856, "grad_norm": 0.8278115391731262, "learning_rate": 1.5044056561599978e-05, "loss": 3.0339, "step": 129910 }, { "epoch": 0.0538112, "grad_norm": 0.7359062433242798, "learning_rate": 1.504336121400845e-05, "loss": 3.0449, "step": 129920 }, { "epoch": 0.0538368, "grad_norm": 0.6984592080116272, "learning_rate": 1.5042665833712067e-05, "loss": 3.254, "step": 129930 }, { "epoch": 0.0538624, "grad_norm": 0.7980424761772156, "learning_rate": 1.5041970420715335e-05, "loss": 3.1701, "step": 129940 }, { "epoch": 0.053888, "grad_norm": 0.8120068311691284, "learning_rate": 1.5041274975022766e-05, "loss": 2.9724, "step": 129950 }, { "epoch": 0.0539136, "grad_norm": 0.8429736495018005, "learning_rate": 1.504057949663887e-05, "loss": 3.1148, "step": 129960 }, { "epoch": 0.0539392, "grad_norm": 0.7133706212043762, "learning_rate": 1.5039883985568154e-05, "loss": 3.1968, "step": 129970 }, { "epoch": 0.0539648, "grad_norm": 0.7499210834503174, "learning_rate": 1.5039188441815128e-05, "loss": 3.0147, "step": 129980 }, { "epoch": 0.0539904, "grad_norm": 0.776050865650177, "learning_rate": 1.5038492865384307e-05, "loss": 3.0348, "step": 129990 }, { "epoch": 0.054016, "grad_norm": 0.6892354488372803, "learning_rate": 1.5037797256280198e-05, "loss": 3.0078, "step": 130000 }, { "epoch": 0.0540416, "grad_norm": 1.0534143447875977, "learning_rate": 1.503710161450731e-05, "loss": 3.0903, "step": 130010 }, { "epoch": 0.0540672, "grad_norm": 0.6434007287025452, "learning_rate": 1.503640594007016e-05, "loss": 2.9126, "step": 130020 }, { "epoch": 0.0540928, "grad_norm": 0.6481518745422363, "learning_rate": 1.5035710232973256e-05, "loss": 2.9556, "step": 130030 }, { "epoch": 0.0541184, "grad_norm": 1.1304155588150024, "learning_rate": 1.5035014493221108e-05, "loss": 3.1618, "step": 130040 }, { "epoch": 0.054144, "grad_norm": 0.9566487669944763, "learning_rate": 1.5034318720818228e-05, "loss": 3.1863, "step": 130050 }, { "epoch": 0.0541696, "grad_norm": 1.2103030681610107, "learning_rate": 1.5033622915769135e-05, "loss": 2.8358, "step": 130060 }, { "epoch": 0.0541952, "grad_norm": 0.7791178226470947, "learning_rate": 1.503292707807833e-05, "loss": 3.2492, "step": 130070 }, { "epoch": 0.0542208, "grad_norm": 0.7289243340492249, "learning_rate": 1.5032231207750329e-05, "loss": 3.0799, "step": 130080 }, { "epoch": 0.0542464, "grad_norm": 0.7225679159164429, "learning_rate": 1.503153530478965e-05, "loss": 3.0529, "step": 130090 }, { "epoch": 0.054272, "grad_norm": 0.7138689160346985, "learning_rate": 1.5030839369200801e-05, "loss": 3.2147, "step": 130100 }, { "epoch": 0.0542976, "grad_norm": 0.8768152594566345, "learning_rate": 1.5030143400988294e-05, "loss": 2.9678, "step": 130110 }, { "epoch": 0.0543232, "grad_norm": 0.8648607730865479, "learning_rate": 1.5029447400156646e-05, "loss": 3.1867, "step": 130120 }, { "epoch": 0.0543488, "grad_norm": 0.675653338432312, "learning_rate": 1.5028751366710365e-05, "loss": 2.9451, "step": 130130 }, { "epoch": 0.0543744, "grad_norm": 0.746480405330658, "learning_rate": 1.502805530065397e-05, "loss": 3.0581, "step": 130140 }, { "epoch": 0.0544, "grad_norm": 0.7031347751617432, "learning_rate": 1.5027359201991973e-05, "loss": 3.0, "step": 130150 }, { "epoch": 0.0544256, "grad_norm": 0.7260912656784058, "learning_rate": 1.502666307072889e-05, "loss": 3.0047, "step": 130160 }, { "epoch": 0.0544512, "grad_norm": 0.6516491770744324, "learning_rate": 1.502596690686923e-05, "loss": 2.976, "step": 130170 }, { "epoch": 0.0544768, "grad_norm": 0.7188768982887268, "learning_rate": 1.502527071041751e-05, "loss": 2.9517, "step": 130180 }, { "epoch": 0.0545024, "grad_norm": 0.6430644392967224, "learning_rate": 1.5024574481378245e-05, "loss": 3.1105, "step": 130190 }, { "epoch": 0.054528, "grad_norm": 0.8532121181488037, "learning_rate": 1.5023878219755949e-05, "loss": 3.0861, "step": 130200 }, { "epoch": 0.0545536, "grad_norm": 0.6999014616012573, "learning_rate": 1.5023181925555141e-05, "loss": 2.9986, "step": 130210 }, { "epoch": 0.0545792, "grad_norm": 1.0161375999450684, "learning_rate": 1.5022485598780333e-05, "loss": 3.0795, "step": 130220 }, { "epoch": 0.0546048, "grad_norm": 0.8264204263687134, "learning_rate": 1.5021789239436037e-05, "loss": 3.0417, "step": 130230 }, { "epoch": 0.0546304, "grad_norm": 0.6843150854110718, "learning_rate": 1.5021092847526775e-05, "loss": 3.2274, "step": 130240 }, { "epoch": 0.054656, "grad_norm": 0.756102979183197, "learning_rate": 1.502039642305706e-05, "loss": 3.0552, "step": 130250 }, { "epoch": 0.0546816, "grad_norm": 0.7853014469146729, "learning_rate": 1.5019699966031408e-05, "loss": 2.9002, "step": 130260 }, { "epoch": 0.0547072, "grad_norm": 0.7664072513580322, "learning_rate": 1.5019003476454337e-05, "loss": 3.0909, "step": 130270 }, { "epoch": 0.0547328, "grad_norm": 0.7685438990592957, "learning_rate": 1.5018306954330361e-05, "loss": 2.9791, "step": 130280 }, { "epoch": 0.0547584, "grad_norm": 0.7721203565597534, "learning_rate": 1.5017610399664e-05, "loss": 3.222, "step": 130290 }, { "epoch": 0.054784, "grad_norm": 0.7681277394294739, "learning_rate": 1.5016913812459769e-05, "loss": 3.0703, "step": 130300 }, { "epoch": 0.0548096, "grad_norm": 0.6596646904945374, "learning_rate": 1.5016217192722183e-05, "loss": 2.8587, "step": 130310 }, { "epoch": 0.0548352, "grad_norm": 0.7461696267127991, "learning_rate": 1.5015520540455764e-05, "loss": 3.0051, "step": 130320 }, { "epoch": 0.0548608, "grad_norm": 0.6345539689064026, "learning_rate": 1.5014823855665026e-05, "loss": 3.185, "step": 130330 }, { "epoch": 0.0548864, "grad_norm": 0.9569090008735657, "learning_rate": 1.5014127138354488e-05, "loss": 3.1214, "step": 130340 }, { "epoch": 0.054912, "grad_norm": 0.7855456471443176, "learning_rate": 1.5013430388528668e-05, "loss": 3.0301, "step": 130350 }, { "epoch": 0.0549376, "grad_norm": 0.9449374079704285, "learning_rate": 1.5012733606192083e-05, "loss": 3.2731, "step": 130360 }, { "epoch": 0.0549632, "grad_norm": 0.7699745297431946, "learning_rate": 1.5012036791349255e-05, "loss": 3.0081, "step": 130370 }, { "epoch": 0.0549888, "grad_norm": 0.7821857333183289, "learning_rate": 1.5011339944004699e-05, "loss": 3.0491, "step": 130380 }, { "epoch": 0.0550144, "grad_norm": 0.6783038377761841, "learning_rate": 1.5010643064162938e-05, "loss": 3.1357, "step": 130390 }, { "epoch": 0.05504, "grad_norm": 1.1997902393341064, "learning_rate": 1.5009946151828487e-05, "loss": 3.232, "step": 130400 }, { "epoch": 0.0550656, "grad_norm": 0.6424487233161926, "learning_rate": 1.5009249207005865e-05, "loss": 3.1951, "step": 130410 }, { "epoch": 0.0550912, "grad_norm": 0.8906088471412659, "learning_rate": 1.5008552229699595e-05, "loss": 3.1753, "step": 130420 }, { "epoch": 0.0551168, "grad_norm": 0.8752138018608093, "learning_rate": 1.5007855219914196e-05, "loss": 3.2005, "step": 130430 }, { "epoch": 0.0551424, "grad_norm": 0.8260735869407654, "learning_rate": 1.5007158177654182e-05, "loss": 3.0685, "step": 130440 }, { "epoch": 0.055168, "grad_norm": 0.77822345495224, "learning_rate": 1.5006461102924084e-05, "loss": 2.7765, "step": 130450 }, { "epoch": 0.0551936, "grad_norm": 0.6701822876930237, "learning_rate": 1.500576399572841e-05, "loss": 3.201, "step": 130460 }, { "epoch": 0.0552192, "grad_norm": 1.1603749990463257, "learning_rate": 1.500506685607169e-05, "loss": 2.993, "step": 130470 }, { "epoch": 0.0552448, "grad_norm": 1.0254108905792236, "learning_rate": 1.5004369683958442e-05, "loss": 3.0814, "step": 130480 }, { "epoch": 0.0552704, "grad_norm": 0.8207913637161255, "learning_rate": 1.5003672479393187e-05, "loss": 2.9941, "step": 130490 }, { "epoch": 0.055296, "grad_norm": 0.7298461198806763, "learning_rate": 1.5002975242380446e-05, "loss": 3.0661, "step": 130500 }, { "epoch": 0.0553216, "grad_norm": 0.8204057812690735, "learning_rate": 1.5002277972924741e-05, "loss": 3.0164, "step": 130510 }, { "epoch": 0.0553472, "grad_norm": 0.7005780935287476, "learning_rate": 1.5001580671030591e-05, "loss": 2.9401, "step": 130520 }, { "epoch": 0.0553728, "grad_norm": 4.756045818328857, "learning_rate": 1.500088333670252e-05, "loss": 3.1301, "step": 130530 }, { "epoch": 0.0553984, "grad_norm": 0.756591796875, "learning_rate": 1.5000185969945048e-05, "loss": 3.0936, "step": 130540 }, { "epoch": 0.055424, "grad_norm": 0.762712299823761, "learning_rate": 1.49994885707627e-05, "loss": 3.0103, "step": 130550 }, { "epoch": 0.0554496, "grad_norm": 0.835871696472168, "learning_rate": 1.4998791139159998e-05, "loss": 2.9506, "step": 130560 }, { "epoch": 0.0554752, "grad_norm": 0.8058941960334778, "learning_rate": 1.4998093675141463e-05, "loss": 3.0951, "step": 130570 }, { "epoch": 0.0555008, "grad_norm": 0.8485520482063293, "learning_rate": 1.4997396178711621e-05, "loss": 2.862, "step": 130580 }, { "epoch": 0.0555264, "grad_norm": 0.7963038086891174, "learning_rate": 1.4996698649874993e-05, "loss": 3.0529, "step": 130590 }, { "epoch": 0.055552, "grad_norm": 0.7419357895851135, "learning_rate": 1.4996001088636099e-05, "loss": 3.0166, "step": 130600 }, { "epoch": 0.0555776, "grad_norm": 0.981001615524292, "learning_rate": 1.4995303494999466e-05, "loss": 2.9771, "step": 130610 }, { "epoch": 0.0556032, "grad_norm": 0.8553689122200012, "learning_rate": 1.499460586896962e-05, "loss": 3.2079, "step": 130620 }, { "epoch": 0.0556288, "grad_norm": 0.7636007070541382, "learning_rate": 1.499390821055108e-05, "loss": 2.7295, "step": 130630 }, { "epoch": 0.0556544, "grad_norm": 0.6737553477287292, "learning_rate": 1.4993210519748373e-05, "loss": 2.9657, "step": 130640 }, { "epoch": 0.05568, "grad_norm": 0.747266948223114, "learning_rate": 1.4992512796566025e-05, "loss": 3.1833, "step": 130650 }, { "epoch": 0.0557056, "grad_norm": 0.6866562962532043, "learning_rate": 1.4991815041008557e-05, "loss": 3.1161, "step": 130660 }, { "epoch": 0.0557312, "grad_norm": 1.0681499242782593, "learning_rate": 1.4991117253080496e-05, "loss": 3.2306, "step": 130670 }, { "epoch": 0.0557568, "grad_norm": 0.6894387602806091, "learning_rate": 1.4990419432786363e-05, "loss": 3.0594, "step": 130680 }, { "epoch": 0.0557824, "grad_norm": 0.7181791663169861, "learning_rate": 1.4989721580130691e-05, "loss": 3.0589, "step": 130690 }, { "epoch": 0.055808, "grad_norm": 0.8472244143486023, "learning_rate": 1.4989023695117996e-05, "loss": 3.052, "step": 130700 }, { "epoch": 0.0558336, "grad_norm": 0.7137201428413391, "learning_rate": 1.4988325777752813e-05, "loss": 3.0872, "step": 130710 }, { "epoch": 0.0558592, "grad_norm": 0.7334112524986267, "learning_rate": 1.4987627828039659e-05, "loss": 3.223, "step": 130720 }, { "epoch": 0.0558848, "grad_norm": 0.7473485469818115, "learning_rate": 1.4986929845983065e-05, "loss": 2.9965, "step": 130730 }, { "epoch": 0.0559104, "grad_norm": 0.7069958448410034, "learning_rate": 1.4986231831587556e-05, "loss": 3.2008, "step": 130740 }, { "epoch": 0.055936, "grad_norm": 0.7914525270462036, "learning_rate": 1.4985533784857662e-05, "loss": 3.1269, "step": 130750 }, { "epoch": 0.0559616, "grad_norm": 0.8506519198417664, "learning_rate": 1.4984835705797903e-05, "loss": 3.0108, "step": 130760 }, { "epoch": 0.0559872, "grad_norm": 0.7246622443199158, "learning_rate": 1.4984137594412808e-05, "loss": 3.0652, "step": 130770 }, { "epoch": 0.0560128, "grad_norm": 0.7273083329200745, "learning_rate": 1.4983439450706908e-05, "loss": 3.2032, "step": 130780 }, { "epoch": 0.0560384, "grad_norm": 0.7130270004272461, "learning_rate": 1.4982741274684726e-05, "loss": 3.1485, "step": 130790 }, { "epoch": 0.056064, "grad_norm": 0.7942357659339905, "learning_rate": 1.4982043066350794e-05, "loss": 3.4052, "step": 130800 }, { "epoch": 0.0560896, "grad_norm": 0.9805663228034973, "learning_rate": 1.4981344825709637e-05, "loss": 3.1456, "step": 130810 }, { "epoch": 0.0561152, "grad_norm": 1.060542106628418, "learning_rate": 1.498064655276578e-05, "loss": 2.9764, "step": 130820 }, { "epoch": 0.0561408, "grad_norm": 0.7286435961723328, "learning_rate": 1.4979948247523752e-05, "loss": 3.2111, "step": 130830 }, { "epoch": 0.0561664, "grad_norm": 0.8789868950843811, "learning_rate": 1.4979249909988088e-05, "loss": 2.9958, "step": 130840 }, { "epoch": 0.056192, "grad_norm": 1.1693934202194214, "learning_rate": 1.4978551540163309e-05, "loss": 2.9158, "step": 130850 }, { "epoch": 0.0562176, "grad_norm": 0.7538801431655884, "learning_rate": 1.4977853138053945e-05, "loss": 2.9878, "step": 130860 }, { "epoch": 0.0562432, "grad_norm": 0.7302262187004089, "learning_rate": 1.4977154703664529e-05, "loss": 2.9878, "step": 130870 }, { "epoch": 0.0562688, "grad_norm": 1.0514395236968994, "learning_rate": 1.4976456236999586e-05, "loss": 3.02, "step": 130880 }, { "epoch": 0.0562944, "grad_norm": 0.7546784281730652, "learning_rate": 1.4975757738063648e-05, "loss": 3.1448, "step": 130890 }, { "epoch": 0.05632, "grad_norm": 0.84095698595047, "learning_rate": 1.4975059206861242e-05, "loss": 3.0256, "step": 130900 }, { "epoch": 0.0563456, "grad_norm": 0.7356117963790894, "learning_rate": 1.4974360643396902e-05, "loss": 2.8891, "step": 130910 }, { "epoch": 0.0563712, "grad_norm": 0.8007916808128357, "learning_rate": 1.4973662047675153e-05, "loss": 2.9364, "step": 130920 }, { "epoch": 0.0563968, "grad_norm": 0.7212464213371277, "learning_rate": 1.4972963419700525e-05, "loss": 3.0964, "step": 130930 }, { "epoch": 0.0564224, "grad_norm": 0.7940086722373962, "learning_rate": 1.4972264759477554e-05, "loss": 3.0483, "step": 130940 }, { "epoch": 0.056448, "grad_norm": 0.7219845652580261, "learning_rate": 1.497156606701077e-05, "loss": 3.1626, "step": 130950 }, { "epoch": 0.0564736, "grad_norm": 0.711316704750061, "learning_rate": 1.4970867342304699e-05, "loss": 3.0327, "step": 130960 }, { "epoch": 0.0564992, "grad_norm": 0.737640917301178, "learning_rate": 1.4970168585363872e-05, "loss": 3.0795, "step": 130970 }, { "epoch": 0.0565248, "grad_norm": 0.693602979183197, "learning_rate": 1.4969469796192827e-05, "loss": 2.9378, "step": 130980 }, { "epoch": 0.0565504, "grad_norm": 0.7205409407615662, "learning_rate": 1.4968770974796084e-05, "loss": 3.1503, "step": 130990 }, { "epoch": 0.056576, "grad_norm": 0.8726969957351685, "learning_rate": 1.4968072121178187e-05, "loss": 3.1046, "step": 131000 }, { "epoch": 0.0566016, "grad_norm": 0.9253693222999573, "learning_rate": 1.4967373235343665e-05, "loss": 3.1522, "step": 131010 }, { "epoch": 0.0566272, "grad_norm": 0.6918565630912781, "learning_rate": 1.4966674317297045e-05, "loss": 3.175, "step": 131020 }, { "epoch": 0.0566528, "grad_norm": 0.802909791469574, "learning_rate": 1.496597536704286e-05, "loss": 3.1217, "step": 131030 }, { "epoch": 0.0566784, "grad_norm": 0.7039731740951538, "learning_rate": 1.4965276384585648e-05, "loss": 3.0753, "step": 131040 }, { "epoch": 0.056704, "grad_norm": 0.6826228499412537, "learning_rate": 1.4964577369929938e-05, "loss": 2.9785, "step": 131050 }, { "epoch": 0.0567296, "grad_norm": 0.939628005027771, "learning_rate": 1.496387832308026e-05, "loss": 2.8853, "step": 131060 }, { "epoch": 0.0567552, "grad_norm": 0.795819103717804, "learning_rate": 1.4963179244041151e-05, "loss": 3.4324, "step": 131070 }, { "epoch": 0.0567808, "grad_norm": 0.7768266797065735, "learning_rate": 1.4962480132817146e-05, "loss": 2.9504, "step": 131080 }, { "epoch": 0.0568064, "grad_norm": 0.7180092334747314, "learning_rate": 1.4961780989412775e-05, "loss": 3.1216, "step": 131090 }, { "epoch": 0.056832, "grad_norm": 0.6957339644432068, "learning_rate": 1.4961081813832572e-05, "loss": 3.0812, "step": 131100 }, { "epoch": 0.0568576, "grad_norm": 0.7937475442886353, "learning_rate": 1.4960382606081074e-05, "loss": 3.0136, "step": 131110 }, { "epoch": 0.0568832, "grad_norm": 0.8074996471405029, "learning_rate": 1.4959683366162811e-05, "loss": 2.9815, "step": 131120 }, { "epoch": 0.0569088, "grad_norm": 1.1271239519119263, "learning_rate": 1.4958984094082321e-05, "loss": 2.959, "step": 131130 }, { "epoch": 0.0569344, "grad_norm": 0.917710542678833, "learning_rate": 1.4958284789844135e-05, "loss": 3.0791, "step": 131140 }, { "epoch": 0.05696, "grad_norm": 0.7990673184394836, "learning_rate": 1.4957585453452793e-05, "loss": 2.9727, "step": 131150 }, { "epoch": 0.0569856, "grad_norm": 0.75277179479599, "learning_rate": 1.4956886084912824e-05, "loss": 2.8791, "step": 131160 }, { "epoch": 0.0570112, "grad_norm": 0.6754884719848633, "learning_rate": 1.4956186684228769e-05, "loss": 3.1416, "step": 131170 }, { "epoch": 0.0570368, "grad_norm": 0.8260498046875, "learning_rate": 1.4955487251405161e-05, "loss": 3.1696, "step": 131180 }, { "epoch": 0.0570624, "grad_norm": 0.7776967287063599, "learning_rate": 1.4954787786446531e-05, "loss": 3.0964, "step": 131190 }, { "epoch": 0.057088, "grad_norm": 0.6952288150787354, "learning_rate": 1.4954088289357422e-05, "loss": 3.0877, "step": 131200 }, { "epoch": 0.0571136, "grad_norm": 0.7553994655609131, "learning_rate": 1.4953388760142366e-05, "loss": 2.9923, "step": 131210 }, { "epoch": 0.0571392, "grad_norm": 0.673367440700531, "learning_rate": 1.49526891988059e-05, "loss": 3.112, "step": 131220 }, { "epoch": 0.0571648, "grad_norm": 0.7003836035728455, "learning_rate": 1.4951989605352562e-05, "loss": 3.0564, "step": 131230 }, { "epoch": 0.0571904, "grad_norm": 0.7016487121582031, "learning_rate": 1.4951289979786886e-05, "loss": 3.192, "step": 131240 }, { "epoch": 0.057216, "grad_norm": 1.1925008296966553, "learning_rate": 1.4950590322113413e-05, "loss": 3.0653, "step": 131250 }, { "epoch": 0.0572416, "grad_norm": 0.9730677604675293, "learning_rate": 1.4949890632336675e-05, "loss": 2.9475, "step": 131260 }, { "epoch": 0.0572672, "grad_norm": 1.0280755758285522, "learning_rate": 1.4949190910461214e-05, "loss": 3.0859, "step": 131270 }, { "epoch": 0.0572928, "grad_norm": 0.7610648274421692, "learning_rate": 1.4948491156491564e-05, "loss": 2.9956, "step": 131280 }, { "epoch": 0.0573184, "grad_norm": 0.7131748199462891, "learning_rate": 1.4947791370432263e-05, "loss": 2.9646, "step": 131290 }, { "epoch": 0.057344, "grad_norm": 1.0701295137405396, "learning_rate": 1.4947091552287852e-05, "loss": 2.9182, "step": 131300 }, { "epoch": 0.0573696, "grad_norm": 1.0303311347961426, "learning_rate": 1.4946391702062865e-05, "loss": 3.107, "step": 131310 }, { "epoch": 0.0573952, "grad_norm": 0.8966115117073059, "learning_rate": 1.4945691819761843e-05, "loss": 3.115, "step": 131320 }, { "epoch": 0.0574208, "grad_norm": 0.7818470597267151, "learning_rate": 1.4944991905389323e-05, "loss": 3.1157, "step": 131330 }, { "epoch": 0.0574464, "grad_norm": 0.7733495831489563, "learning_rate": 1.4944291958949848e-05, "loss": 3.0339, "step": 131340 }, { "epoch": 0.057472, "grad_norm": 0.6704601645469666, "learning_rate": 1.494359198044795e-05, "loss": 2.9925, "step": 131350 }, { "epoch": 0.0574976, "grad_norm": 0.7258123755455017, "learning_rate": 1.4942891969888176e-05, "loss": 3.2578, "step": 131360 }, { "epoch": 0.0575232, "grad_norm": 0.9644456505775452, "learning_rate": 1.4942191927275058e-05, "loss": 3.1405, "step": 131370 }, { "epoch": 0.0575488, "grad_norm": 0.7241207957267761, "learning_rate": 1.494149185261314e-05, "loss": 3.0538, "step": 131380 }, { "epoch": 0.0575744, "grad_norm": 0.72626793384552, "learning_rate": 1.4940791745906959e-05, "loss": 2.8844, "step": 131390 }, { "epoch": 0.0576, "grad_norm": 0.7385469079017639, "learning_rate": 1.4940091607161057e-05, "loss": 3.2782, "step": 131400 }, { "epoch": 0.0576256, "grad_norm": 0.7582634687423706, "learning_rate": 1.4939391436379975e-05, "loss": 3.1409, "step": 131410 }, { "epoch": 0.0576512, "grad_norm": 0.7741273045539856, "learning_rate": 1.4938691233568253e-05, "loss": 2.9107, "step": 131420 }, { "epoch": 0.0576768, "grad_norm": 0.754563570022583, "learning_rate": 1.493799099873043e-05, "loss": 3.2115, "step": 131430 }, { "epoch": 0.0577024, "grad_norm": 0.7736526727676392, "learning_rate": 1.4937290731871049e-05, "loss": 3.1525, "step": 131440 }, { "epoch": 0.057728, "grad_norm": 0.7329264879226685, "learning_rate": 1.4936590432994647e-05, "loss": 3.1519, "step": 131450 }, { "epoch": 0.0577536, "grad_norm": 0.6972744464874268, "learning_rate": 1.493589010210577e-05, "loss": 2.8222, "step": 131460 }, { "epoch": 0.0577792, "grad_norm": 0.8002253770828247, "learning_rate": 1.4935189739208955e-05, "loss": 2.8343, "step": 131470 }, { "epoch": 0.0578048, "grad_norm": 0.7566887736320496, "learning_rate": 1.493448934430875e-05, "loss": 2.953, "step": 131480 }, { "epoch": 0.0578304, "grad_norm": 0.7695633172988892, "learning_rate": 1.4933788917409689e-05, "loss": 2.7687, "step": 131490 }, { "epoch": 0.057856, "grad_norm": 0.8457813262939453, "learning_rate": 1.493308845851632e-05, "loss": 3.0846, "step": 131500 }, { "epoch": 0.0578816, "grad_norm": 0.8623905181884766, "learning_rate": 1.4932387967633184e-05, "loss": 2.8869, "step": 131510 }, { "epoch": 0.0579072, "grad_norm": 0.7533519864082336, "learning_rate": 1.4931687444764824e-05, "loss": 2.9345, "step": 131520 }, { "epoch": 0.0579328, "grad_norm": 0.7147701978683472, "learning_rate": 1.4930986889915779e-05, "loss": 2.9221, "step": 131530 }, { "epoch": 0.0579584, "grad_norm": 0.7213287353515625, "learning_rate": 1.4930286303090598e-05, "loss": 2.9926, "step": 131540 }, { "epoch": 0.057984, "grad_norm": 0.7120655179023743, "learning_rate": 1.4929585684293817e-05, "loss": 2.7822, "step": 131550 }, { "epoch": 0.0580096, "grad_norm": 0.7597536444664001, "learning_rate": 1.4928885033529983e-05, "loss": 2.8367, "step": 131560 }, { "epoch": 0.0580352, "grad_norm": 0.7059605717658997, "learning_rate": 1.492818435080364e-05, "loss": 2.8983, "step": 131570 }, { "epoch": 0.0580608, "grad_norm": 1.1854777336120605, "learning_rate": 1.4927483636119334e-05, "loss": 2.9591, "step": 131580 }, { "epoch": 0.0580864, "grad_norm": 0.6548255085945129, "learning_rate": 1.4926782889481603e-05, "loss": 2.9225, "step": 131590 }, { "epoch": 0.058112, "grad_norm": 0.7102105617523193, "learning_rate": 1.4926082110894995e-05, "loss": 3.034, "step": 131600 }, { "epoch": 0.0581376, "grad_norm": 0.8172444105148315, "learning_rate": 1.4925381300364055e-05, "loss": 2.9818, "step": 131610 }, { "epoch": 0.0581632, "grad_norm": 0.7296550273895264, "learning_rate": 1.4924680457893326e-05, "loss": 2.972, "step": 131620 }, { "epoch": 0.0581888, "grad_norm": 0.7944116592407227, "learning_rate": 1.492397958348735e-05, "loss": 2.8896, "step": 131630 }, { "epoch": 0.0582144, "grad_norm": 0.6693174242973328, "learning_rate": 1.492327867715068e-05, "loss": 2.8661, "step": 131640 }, { "epoch": 0.05824, "grad_norm": 1.1727960109710693, "learning_rate": 1.4922577738887855e-05, "loss": 2.8119, "step": 131650 }, { "epoch": 0.0582656, "grad_norm": 0.6955945491790771, "learning_rate": 1.4921876768703417e-05, "loss": 2.9109, "step": 131660 }, { "epoch": 0.0582912, "grad_norm": 0.6976997256278992, "learning_rate": 1.4921175766601923e-05, "loss": 2.8984, "step": 131670 }, { "epoch": 0.0583168, "grad_norm": 0.6494850516319275, "learning_rate": 1.4920474732587908e-05, "loss": 2.6857, "step": 131680 }, { "epoch": 0.0583424, "grad_norm": 0.7041134834289551, "learning_rate": 1.4919773666665923e-05, "loss": 2.8021, "step": 131690 }, { "epoch": 0.058368, "grad_norm": 0.7844234704971313, "learning_rate": 1.4919072568840513e-05, "loss": 3.3785, "step": 131700 }, { "epoch": 0.0583936, "grad_norm": 0.6618282198905945, "learning_rate": 1.4918371439116225e-05, "loss": 2.8464, "step": 131710 }, { "epoch": 0.0584192, "grad_norm": 0.6792876124382019, "learning_rate": 1.4917670277497605e-05, "loss": 2.7715, "step": 131720 }, { "epoch": 0.0584448, "grad_norm": 0.8626121878623962, "learning_rate": 1.4916969083989198e-05, "loss": 3.028, "step": 131730 }, { "epoch": 0.0584704, "grad_norm": 0.750151515007019, "learning_rate": 1.491626785859556e-05, "loss": 3.4318, "step": 131740 }, { "epoch": 0.058496, "grad_norm": 0.6740269064903259, "learning_rate": 1.4915566601321226e-05, "loss": 2.6868, "step": 131750 }, { "epoch": 0.0585216, "grad_norm": 1.024188756942749, "learning_rate": 1.4914865312170748e-05, "loss": 2.8784, "step": 131760 }, { "epoch": 0.0585472, "grad_norm": 0.6890820860862732, "learning_rate": 1.4914163991148679e-05, "loss": 2.8405, "step": 131770 }, { "epoch": 0.0585728, "grad_norm": 0.6585286259651184, "learning_rate": 1.491346263825956e-05, "loss": 2.8185, "step": 131780 }, { "epoch": 0.0585984, "grad_norm": 0.7301946878433228, "learning_rate": 1.491276125350794e-05, "loss": 2.5982, "step": 131790 }, { "epoch": 0.058624, "grad_norm": 0.9098458886146545, "learning_rate": 1.4912059836898374e-05, "loss": 2.7545, "step": 131800 }, { "epoch": 0.0586496, "grad_norm": 0.7369422912597656, "learning_rate": 1.4911358388435402e-05, "loss": 2.9638, "step": 131810 }, { "epoch": 0.0586752, "grad_norm": 0.6996477246284485, "learning_rate": 1.4910656908123575e-05, "loss": 2.8557, "step": 131820 }, { "epoch": 0.0587008, "grad_norm": 0.6504570841789246, "learning_rate": 1.4909955395967445e-05, "loss": 2.7125, "step": 131830 }, { "epoch": 0.0587264, "grad_norm": 0.6975361704826355, "learning_rate": 1.4909253851971559e-05, "loss": 2.9705, "step": 131840 }, { "epoch": 0.058752, "grad_norm": 0.8049175143241882, "learning_rate": 1.4908552276140464e-05, "loss": 3.0362, "step": 131850 }, { "epoch": 0.0587776, "grad_norm": 0.7103980183601379, "learning_rate": 1.4907850668478715e-05, "loss": 2.8386, "step": 131860 }, { "epoch": 0.0588032, "grad_norm": 0.7202678322792053, "learning_rate": 1.4907149028990859e-05, "loss": 2.6308, "step": 131870 }, { "epoch": 0.0588288, "grad_norm": 0.707983672618866, "learning_rate": 1.4906447357681445e-05, "loss": 2.6328, "step": 131880 }, { "epoch": 0.0588544, "grad_norm": 0.7479917407035828, "learning_rate": 1.4905745654555022e-05, "loss": 2.7309, "step": 131890 }, { "epoch": 0.05888, "grad_norm": 0.8326422572135925, "learning_rate": 1.4905043919616147e-05, "loss": 2.8965, "step": 131900 }, { "epoch": 0.0589056, "grad_norm": 0.6807600259780884, "learning_rate": 1.4904342152869363e-05, "loss": 3.0131, "step": 131910 }, { "epoch": 0.0589312, "grad_norm": 0.6344304084777832, "learning_rate": 1.490364035431922e-05, "loss": 2.8944, "step": 131920 }, { "epoch": 0.0589568, "grad_norm": 0.7417724132537842, "learning_rate": 1.4902938523970276e-05, "loss": 2.9175, "step": 131930 }, { "epoch": 0.0589824, "grad_norm": 0.7182006239891052, "learning_rate": 1.4902236661827079e-05, "loss": 2.8168, "step": 131940 }, { "epoch": 0.059008, "grad_norm": 0.8666926026344299, "learning_rate": 1.4901534767894179e-05, "loss": 2.6995, "step": 131950 }, { "epoch": 0.0590336, "grad_norm": 0.8409590125083923, "learning_rate": 1.4900832842176129e-05, "loss": 2.9638, "step": 131960 }, { "epoch": 0.0590592, "grad_norm": 0.7175100445747375, "learning_rate": 1.4900130884677482e-05, "loss": 2.8136, "step": 131970 }, { "epoch": 0.0590848, "grad_norm": 0.6767575144767761, "learning_rate": 1.4899428895402786e-05, "loss": 2.9144, "step": 131980 }, { "epoch": 0.0591104, "grad_norm": 0.8892197608947754, "learning_rate": 1.4898726874356597e-05, "loss": 2.9582, "step": 131990 }, { "epoch": 0.059136, "grad_norm": 0.689699113368988, "learning_rate": 1.4898024821543469e-05, "loss": 3.1073, "step": 132000 }, { "epoch": 0.0591616, "grad_norm": 0.7110569477081299, "learning_rate": 1.4897322736967945e-05, "loss": 2.7756, "step": 132010 }, { "epoch": 0.0591872, "grad_norm": 0.6935116648674011, "learning_rate": 1.489662062063459e-05, "loss": 2.6894, "step": 132020 }, { "epoch": 0.0592128, "grad_norm": 0.8210614919662476, "learning_rate": 1.489591847254795e-05, "loss": 3.0045, "step": 132030 }, { "epoch": 0.0592384, "grad_norm": 0.6637828946113586, "learning_rate": 1.489521629271258e-05, "loss": 2.805, "step": 132040 }, { "epoch": 0.059264, "grad_norm": 0.7030468583106995, "learning_rate": 1.4894514081133032e-05, "loss": 2.8908, "step": 132050 }, { "epoch": 0.0592896, "grad_norm": 0.6786969304084778, "learning_rate": 1.4893811837813862e-05, "loss": 2.9136, "step": 132060 }, { "epoch": 0.0593152, "grad_norm": 0.8401268720626831, "learning_rate": 1.4893109562759625e-05, "loss": 2.9086, "step": 132070 }, { "epoch": 0.0593408, "grad_norm": 0.8322815299034119, "learning_rate": 1.4892407255974872e-05, "loss": 3.058, "step": 132080 }, { "epoch": 0.0593664, "grad_norm": 0.7078414559364319, "learning_rate": 1.4891704917464156e-05, "loss": 2.994, "step": 132090 }, { "epoch": 0.059392, "grad_norm": 0.6911024451255798, "learning_rate": 1.4891002547232035e-05, "loss": 3.0697, "step": 132100 }, { "epoch": 0.0594176, "grad_norm": 0.7068758606910706, "learning_rate": 1.4890300145283061e-05, "loss": 2.7322, "step": 132110 }, { "epoch": 0.0594432, "grad_norm": 0.6935430765151978, "learning_rate": 1.488959771162179e-05, "loss": 2.961, "step": 132120 }, { "epoch": 0.0594688, "grad_norm": 0.6727972626686096, "learning_rate": 1.4888895246252782e-05, "loss": 2.795, "step": 132130 }, { "epoch": 0.0594944, "grad_norm": 0.7622265219688416, "learning_rate": 1.4888192749180583e-05, "loss": 2.8409, "step": 132140 }, { "epoch": 0.05952, "grad_norm": 0.6928433775901794, "learning_rate": 1.4887490220409755e-05, "loss": 2.6591, "step": 132150 }, { "epoch": 0.0595456, "grad_norm": 0.7029484510421753, "learning_rate": 1.4886787659944852e-05, "loss": 3.0705, "step": 132160 }, { "epoch": 0.0595712, "grad_norm": 0.6760627031326294, "learning_rate": 1.488608506779043e-05, "loss": 2.961, "step": 132170 }, { "epoch": 0.0595968, "grad_norm": 0.6815301775932312, "learning_rate": 1.4885382443951044e-05, "loss": 3.0543, "step": 132180 }, { "epoch": 0.0596224, "grad_norm": 0.7592172622680664, "learning_rate": 1.4884679788431248e-05, "loss": 2.9913, "step": 132190 }, { "epoch": 0.059648, "grad_norm": 0.7058264017105103, "learning_rate": 1.488397710123561e-05, "loss": 3.044, "step": 132200 }, { "epoch": 0.0596736, "grad_norm": 0.7330719232559204, "learning_rate": 1.4883274382368671e-05, "loss": 3.0578, "step": 132210 }, { "epoch": 0.0596992, "grad_norm": 1.0799566507339478, "learning_rate": 1.4882571631834999e-05, "loss": 2.7797, "step": 132220 }, { "epoch": 0.0597248, "grad_norm": 0.6361766457557678, "learning_rate": 1.4881868849639147e-05, "loss": 2.7067, "step": 132230 }, { "epoch": 0.0597504, "grad_norm": 0.708354115486145, "learning_rate": 1.4881166035785673e-05, "loss": 2.8557, "step": 132240 }, { "epoch": 0.059776, "grad_norm": 0.8243893980979919, "learning_rate": 1.4880463190279133e-05, "loss": 2.7545, "step": 132250 }, { "epoch": 0.0598016, "grad_norm": 0.7750579714775085, "learning_rate": 1.4879760313124088e-05, "loss": 3.0556, "step": 132260 }, { "epoch": 0.0598272, "grad_norm": 1.0035629272460938, "learning_rate": 1.4879057404325093e-05, "loss": 2.8859, "step": 132270 }, { "epoch": 0.0598528, "grad_norm": 0.8861965537071228, "learning_rate": 1.4878354463886708e-05, "loss": 3.0545, "step": 132280 }, { "epoch": 0.0598784, "grad_norm": 1.4118049144744873, "learning_rate": 1.4877651491813491e-05, "loss": 2.9513, "step": 132290 }, { "epoch": 0.059904, "grad_norm": 0.753663182258606, "learning_rate": 1.4876948488110001e-05, "loss": 2.8179, "step": 132300 }, { "epoch": 0.0599296, "grad_norm": 0.8093264102935791, "learning_rate": 1.4876245452780796e-05, "loss": 2.7128, "step": 132310 }, { "epoch": 0.0599552, "grad_norm": 0.6844539642333984, "learning_rate": 1.4875542385830435e-05, "loss": 2.9967, "step": 132320 }, { "epoch": 0.0599808, "grad_norm": 0.7089518308639526, "learning_rate": 1.4874839287263479e-05, "loss": 2.7204, "step": 132330 }, { "epoch": 0.0600064, "grad_norm": 0.8336825966835022, "learning_rate": 1.4874136157084482e-05, "loss": 2.5127, "step": 132340 }, { "epoch": 0.060032, "grad_norm": 0.6652316451072693, "learning_rate": 1.487343299529801e-05, "loss": 2.8101, "step": 132350 }, { "epoch": 0.0600576, "grad_norm": 0.7260022163391113, "learning_rate": 1.4872729801908619e-05, "loss": 2.9599, "step": 132360 }, { "epoch": 0.0600832, "grad_norm": 0.7759114503860474, "learning_rate": 1.4872026576920871e-05, "loss": 2.9464, "step": 132370 }, { "epoch": 0.0601088, "grad_norm": 0.9642140865325928, "learning_rate": 1.4871323320339323e-05, "loss": 3.1317, "step": 132380 }, { "epoch": 0.0601344, "grad_norm": 0.7089341878890991, "learning_rate": 1.4870620032168542e-05, "loss": 2.8456, "step": 132390 }, { "epoch": 0.06016, "grad_norm": 0.65104740858078, "learning_rate": 1.4869916712413085e-05, "loss": 3.1278, "step": 132400 }, { "epoch": 0.0601856, "grad_norm": 0.7462838888168335, "learning_rate": 1.4869213361077508e-05, "loss": 3.1128, "step": 132410 }, { "epoch": 0.0602112, "grad_norm": 0.7081199884414673, "learning_rate": 1.4868509978166378e-05, "loss": 2.9168, "step": 132420 }, { "epoch": 0.0602368, "grad_norm": 0.6629088521003723, "learning_rate": 1.4867806563684256e-05, "loss": 2.6785, "step": 132430 }, { "epoch": 0.0602624, "grad_norm": 0.7953763008117676, "learning_rate": 1.4867103117635702e-05, "loss": 2.8877, "step": 132440 }, { "epoch": 0.060288, "grad_norm": 0.7674140930175781, "learning_rate": 1.4866399640025276e-05, "loss": 2.8219, "step": 132450 }, { "epoch": 0.0603136, "grad_norm": 0.6984959840774536, "learning_rate": 1.4865696130857545e-05, "loss": 2.9889, "step": 132460 }, { "epoch": 0.0603392, "grad_norm": 0.7230883240699768, "learning_rate": 1.4864992590137064e-05, "loss": 3.0454, "step": 132470 }, { "epoch": 0.0603648, "grad_norm": 1.0262242555618286, "learning_rate": 1.4864289017868401e-05, "loss": 3.0776, "step": 132480 }, { "epoch": 0.0603904, "grad_norm": 0.7291548252105713, "learning_rate": 1.4863585414056117e-05, "loss": 3.0319, "step": 132490 }, { "epoch": 0.060416, "grad_norm": 0.7288016080856323, "learning_rate": 1.4862881778704774e-05, "loss": 2.9176, "step": 132500 }, { "epoch": 0.0604416, "grad_norm": 0.7032866477966309, "learning_rate": 1.4862178111818933e-05, "loss": 2.9907, "step": 132510 }, { "epoch": 0.0604672, "grad_norm": 0.7408750057220459, "learning_rate": 1.4861474413403162e-05, "loss": 2.9476, "step": 132520 }, { "epoch": 0.0604928, "grad_norm": 0.6869224905967712, "learning_rate": 1.4860770683462018e-05, "loss": 3.0926, "step": 132530 }, { "epoch": 0.0605184, "grad_norm": 1.209273099899292, "learning_rate": 1.486006692200007e-05, "loss": 3.0799, "step": 132540 }, { "epoch": 0.060544, "grad_norm": 0.7423692941665649, "learning_rate": 1.485936312902188e-05, "loss": 3.1414, "step": 132550 }, { "epoch": 0.0605696, "grad_norm": 0.8087826371192932, "learning_rate": 1.4858659304532012e-05, "loss": 2.9791, "step": 132560 }, { "epoch": 0.0605952, "grad_norm": 0.6972132325172424, "learning_rate": 1.485795544853503e-05, "loss": 2.9092, "step": 132570 }, { "epoch": 0.0606208, "grad_norm": 0.6527661085128784, "learning_rate": 1.4857251561035495e-05, "loss": 3.0321, "step": 132580 }, { "epoch": 0.0606464, "grad_norm": 0.6931836605072021, "learning_rate": 1.4856547642037975e-05, "loss": 3.0462, "step": 132590 }, { "epoch": 0.060672, "grad_norm": 0.8422311544418335, "learning_rate": 1.4855843691547038e-05, "loss": 2.7401, "step": 132600 }, { "epoch": 0.0606976, "grad_norm": 0.7816349267959595, "learning_rate": 1.4855139709567243e-05, "loss": 2.9804, "step": 132610 }, { "epoch": 0.0607232, "grad_norm": 0.7290034890174866, "learning_rate": 1.4854435696103155e-05, "loss": 3.1022, "step": 132620 }, { "epoch": 0.0607488, "grad_norm": 0.8548576831817627, "learning_rate": 1.4853731651159344e-05, "loss": 3.0509, "step": 132630 }, { "epoch": 0.0607744, "grad_norm": 0.8164876699447632, "learning_rate": 1.4853027574740373e-05, "loss": 3.0154, "step": 132640 }, { "epoch": 0.0608, "grad_norm": 0.9293161034584045, "learning_rate": 1.485232346685081e-05, "loss": 2.9343, "step": 132650 }, { "epoch": 0.0608256, "grad_norm": 0.7867525815963745, "learning_rate": 1.4851619327495217e-05, "loss": 3.1048, "step": 132660 }, { "epoch": 0.0608512, "grad_norm": 0.6877582669258118, "learning_rate": 1.4850915156678162e-05, "loss": 3.1153, "step": 132670 }, { "epoch": 0.0608768, "grad_norm": 0.874610960483551, "learning_rate": 1.4850210954404212e-05, "loss": 2.9891, "step": 132680 }, { "epoch": 0.0609024, "grad_norm": 0.7730860710144043, "learning_rate": 1.4849506720677933e-05, "loss": 3.2089, "step": 132690 }, { "epoch": 0.060928, "grad_norm": 0.9387247562408447, "learning_rate": 1.484880245550389e-05, "loss": 2.9951, "step": 132700 }, { "epoch": 0.0609536, "grad_norm": 0.7027854919433594, "learning_rate": 1.4848098158886651e-05, "loss": 2.8471, "step": 132710 }, { "epoch": 0.0609792, "grad_norm": 0.8520905375480652, "learning_rate": 1.4847393830830788e-05, "loss": 2.8805, "step": 132720 }, { "epoch": 0.0610048, "grad_norm": 0.6901149153709412, "learning_rate": 1.484668947134086e-05, "loss": 2.9922, "step": 132730 }, { "epoch": 0.0610304, "grad_norm": 0.7182703614234924, "learning_rate": 1.4845985080421442e-05, "loss": 3.0608, "step": 132740 }, { "epoch": 0.061056, "grad_norm": 0.8761138916015625, "learning_rate": 1.4845280658077097e-05, "loss": 2.9102, "step": 132750 }, { "epoch": 0.0610816, "grad_norm": 0.7472317218780518, "learning_rate": 1.4844576204312398e-05, "loss": 2.9949, "step": 132760 }, { "epoch": 0.0611072, "grad_norm": 0.7476905584335327, "learning_rate": 1.4843871719131907e-05, "loss": 3.0261, "step": 132770 }, { "epoch": 0.0611328, "grad_norm": 0.9707621335983276, "learning_rate": 1.4843167202540194e-05, "loss": 3.043, "step": 132780 }, { "epoch": 0.0611584, "grad_norm": 0.675992488861084, "learning_rate": 1.4842462654541832e-05, "loss": 2.8792, "step": 132790 }, { "epoch": 0.061184, "grad_norm": 0.761099100112915, "learning_rate": 1.4841758075141383e-05, "loss": 2.9024, "step": 132800 }, { "epoch": 0.0612096, "grad_norm": 0.7585344910621643, "learning_rate": 1.484105346434342e-05, "loss": 2.8777, "step": 132810 }, { "epoch": 0.0612352, "grad_norm": 0.7041283249855042, "learning_rate": 1.4840348822152515e-05, "loss": 2.9188, "step": 132820 }, { "epoch": 0.0612608, "grad_norm": 0.7281336784362793, "learning_rate": 1.4839644148573233e-05, "loss": 3.1606, "step": 132830 }, { "epoch": 0.0612864, "grad_norm": 1.3739534616470337, "learning_rate": 1.4838939443610144e-05, "loss": 3.1484, "step": 132840 }, { "epoch": 0.061312, "grad_norm": 1.2150304317474365, "learning_rate": 1.483823470726782e-05, "loss": 3.2528, "step": 132850 }, { "epoch": 0.0613376, "grad_norm": 0.6566058397293091, "learning_rate": 1.483752993955083e-05, "loss": 2.791, "step": 132860 }, { "epoch": 0.0613632, "grad_norm": 0.7561262845993042, "learning_rate": 1.4836825140463745e-05, "loss": 2.9262, "step": 132870 }, { "epoch": 0.0613888, "grad_norm": 0.9105932116508484, "learning_rate": 1.4836120310011128e-05, "loss": 3.1364, "step": 132880 }, { "epoch": 0.0614144, "grad_norm": 0.8037055730819702, "learning_rate": 1.4835415448197563e-05, "loss": 3.0928, "step": 132890 }, { "epoch": 0.06144, "grad_norm": 1.0136909484863281, "learning_rate": 1.4834710555027611e-05, "loss": 2.8263, "step": 132900 }, { "epoch": 0.0614656, "grad_norm": 1.15106999874115, "learning_rate": 1.4834005630505845e-05, "loss": 2.9825, "step": 132910 }, { "epoch": 0.0614912, "grad_norm": 1.3472710847854614, "learning_rate": 1.483330067463684e-05, "loss": 3.0792, "step": 132920 }, { "epoch": 0.0615168, "grad_norm": 0.8870859742164612, "learning_rate": 1.4832595687425164e-05, "loss": 2.9352, "step": 132930 }, { "epoch": 0.0615424, "grad_norm": 0.7607040405273438, "learning_rate": 1.4831890668875388e-05, "loss": 3.0345, "step": 132940 }, { "epoch": 0.061568, "grad_norm": 0.957359254360199, "learning_rate": 1.4831185618992088e-05, "loss": 3.1997, "step": 132950 }, { "epoch": 0.0615936, "grad_norm": 0.7181737422943115, "learning_rate": 1.4830480537779832e-05, "loss": 3.1232, "step": 132960 }, { "epoch": 0.0616192, "grad_norm": 0.7543672323226929, "learning_rate": 1.4829775425243189e-05, "loss": 2.9934, "step": 132970 }, { "epoch": 0.0616448, "grad_norm": 0.987846314907074, "learning_rate": 1.4829070281386742e-05, "loss": 2.956, "step": 132980 }, { "epoch": 0.0616704, "grad_norm": 0.7163127660751343, "learning_rate": 1.4828365106215055e-05, "loss": 2.9237, "step": 132990 }, { "epoch": 0.061696, "grad_norm": 0.870521068572998, "learning_rate": 1.4827659899732703e-05, "loss": 3.1329, "step": 133000 }, { "epoch": 0.0617216, "grad_norm": 0.7685457468032837, "learning_rate": 1.4826954661944259e-05, "loss": 3.0333, "step": 133010 }, { "epoch": 0.0617472, "grad_norm": 0.6766968965530396, "learning_rate": 1.4826249392854302e-05, "loss": 3.1111, "step": 133020 }, { "epoch": 0.0617728, "grad_norm": 0.7209174036979675, "learning_rate": 1.4825544092467396e-05, "loss": 2.8098, "step": 133030 }, { "epoch": 0.0617984, "grad_norm": 0.7701072096824646, "learning_rate": 1.4824838760788118e-05, "loss": 3.152, "step": 133040 }, { "epoch": 0.061824, "grad_norm": 0.733027994632721, "learning_rate": 1.4824133397821046e-05, "loss": 2.9556, "step": 133050 }, { "epoch": 0.0618496, "grad_norm": 0.9232259392738342, "learning_rate": 1.4823428003570748e-05, "loss": 3.08, "step": 133060 }, { "epoch": 0.0618752, "grad_norm": 0.7048068642616272, "learning_rate": 1.4822722578041804e-05, "loss": 2.9525, "step": 133070 }, { "epoch": 0.0619008, "grad_norm": 0.8122382760047913, "learning_rate": 1.4822017121238783e-05, "loss": 3.1601, "step": 133080 }, { "epoch": 0.0619264, "grad_norm": 0.7275751829147339, "learning_rate": 1.4821311633166266e-05, "loss": 2.9177, "step": 133090 }, { "epoch": 0.061952, "grad_norm": 0.8073979616165161, "learning_rate": 1.4820606113828824e-05, "loss": 2.9142, "step": 133100 }, { "epoch": 0.0619776, "grad_norm": 0.7730475068092346, "learning_rate": 1.4819900563231028e-05, "loss": 3.1987, "step": 133110 }, { "epoch": 0.0620032, "grad_norm": 0.7613698244094849, "learning_rate": 1.4819194981377462e-05, "loss": 3.1573, "step": 133120 }, { "epoch": 0.0620288, "grad_norm": 0.7165111899375916, "learning_rate": 1.4818489368272697e-05, "loss": 3.2435, "step": 133130 }, { "epoch": 0.0620544, "grad_norm": 0.7093604207038879, "learning_rate": 1.4817783723921307e-05, "loss": 2.9488, "step": 133140 }, { "epoch": 0.06208, "grad_norm": 0.7597931027412415, "learning_rate": 1.4817078048327874e-05, "loss": 2.9901, "step": 133150 }, { "epoch": 0.0621056, "grad_norm": 0.7014804482460022, "learning_rate": 1.4816372341496965e-05, "loss": 2.9073, "step": 133160 }, { "epoch": 0.0621312, "grad_norm": 0.7360178828239441, "learning_rate": 1.4815666603433165e-05, "loss": 2.9977, "step": 133170 }, { "epoch": 0.0621568, "grad_norm": 0.8145546913146973, "learning_rate": 1.4814960834141045e-05, "loss": 2.969, "step": 133180 }, { "epoch": 0.0621824, "grad_norm": 0.7501958608627319, "learning_rate": 1.4814255033625187e-05, "loss": 3.3858, "step": 133190 }, { "epoch": 0.062208, "grad_norm": 0.7852097153663635, "learning_rate": 1.4813549201890161e-05, "loss": 2.7949, "step": 133200 }, { "epoch": 0.0622336, "grad_norm": 0.7471276521682739, "learning_rate": 1.4812843338940548e-05, "loss": 3.1259, "step": 133210 }, { "epoch": 0.0622592, "grad_norm": 0.93097323179245, "learning_rate": 1.4812137444780927e-05, "loss": 3.1077, "step": 133220 }, { "epoch": 0.0622848, "grad_norm": 0.9746472835540771, "learning_rate": 1.4811431519415868e-05, "loss": 3.2067, "step": 133230 }, { "epoch": 0.0623104, "grad_norm": 0.8243350386619568, "learning_rate": 1.4810725562849962e-05, "loss": 3.0336, "step": 133240 }, { "epoch": 0.062336, "grad_norm": 1.1196345090866089, "learning_rate": 1.4810019575087776e-05, "loss": 2.9813, "step": 133250 }, { "epoch": 0.0623616, "grad_norm": 0.9417933821678162, "learning_rate": 1.480931355613389e-05, "loss": 3.2394, "step": 133260 }, { "epoch": 0.0623872, "grad_norm": 0.940726101398468, "learning_rate": 1.4808607505992885e-05, "loss": 3.1415, "step": 133270 }, { "epoch": 0.0624128, "grad_norm": 1.0014992952346802, "learning_rate": 1.4807901424669338e-05, "loss": 3.1036, "step": 133280 }, { "epoch": 0.0624384, "grad_norm": 0.8174948692321777, "learning_rate": 1.480719531216783e-05, "loss": 3.0269, "step": 133290 }, { "epoch": 0.062464, "grad_norm": 0.8236557841300964, "learning_rate": 1.4806489168492937e-05, "loss": 2.9625, "step": 133300 }, { "epoch": 0.0624896, "grad_norm": 0.691001296043396, "learning_rate": 1.4805782993649238e-05, "loss": 3.0595, "step": 133310 }, { "epoch": 0.0625152, "grad_norm": 0.7791374921798706, "learning_rate": 1.4805076787641312e-05, "loss": 3.1126, "step": 133320 }, { "epoch": 0.0625408, "grad_norm": 0.7117149233818054, "learning_rate": 1.4804370550473744e-05, "loss": 2.8907, "step": 133330 }, { "epoch": 0.0625664, "grad_norm": 0.6754146814346313, "learning_rate": 1.4803664282151107e-05, "loss": 3.0431, "step": 133340 }, { "epoch": 0.062592, "grad_norm": 0.6535292863845825, "learning_rate": 1.4802957982677985e-05, "loss": 2.8677, "step": 133350 }, { "epoch": 0.0626176, "grad_norm": 0.6847477555274963, "learning_rate": 1.4802251652058956e-05, "loss": 2.786, "step": 133360 }, { "epoch": 0.0626432, "grad_norm": 0.772142767906189, "learning_rate": 1.4801545290298603e-05, "loss": 2.9475, "step": 133370 }, { "epoch": 0.0626688, "grad_norm": 0.7100564241409302, "learning_rate": 1.4800838897401504e-05, "loss": 2.9555, "step": 133380 }, { "epoch": 0.0626944, "grad_norm": 0.9312489032745361, "learning_rate": 1.4800132473372242e-05, "loss": 3.1252, "step": 133390 }, { "epoch": 0.06272, "grad_norm": 0.7537598013877869, "learning_rate": 1.4799426018215393e-05, "loss": 3.1637, "step": 133400 }, { "epoch": 0.0627456, "grad_norm": 0.8515644073486328, "learning_rate": 1.4798719531935545e-05, "loss": 3.0338, "step": 133410 }, { "epoch": 0.0627712, "grad_norm": 0.8710972666740417, "learning_rate": 1.4798013014537276e-05, "loss": 2.9726, "step": 133420 }, { "epoch": 0.0627968, "grad_norm": 0.7897095084190369, "learning_rate": 1.4797306466025165e-05, "loss": 3.0224, "step": 133430 }, { "epoch": 0.0628224, "grad_norm": 1.058178186416626, "learning_rate": 1.4796599886403798e-05, "loss": 3.1479, "step": 133440 }, { "epoch": 0.062848, "grad_norm": 0.6701655387878418, "learning_rate": 1.479589327567776e-05, "loss": 2.9021, "step": 133450 }, { "epoch": 0.0628736, "grad_norm": 0.7160298824310303, "learning_rate": 1.479518663385162e-05, "loss": 3.1379, "step": 133460 }, { "epoch": 0.0628992, "grad_norm": 0.8125600218772888, "learning_rate": 1.4794479960929974e-05, "loss": 3.0297, "step": 133470 }, { "epoch": 0.0629248, "grad_norm": 0.7834421396255493, "learning_rate": 1.47937732569174e-05, "loss": 2.951, "step": 133480 }, { "epoch": 0.0629504, "grad_norm": 0.8024649024009705, "learning_rate": 1.479306652181848e-05, "loss": 3.0519, "step": 133490 }, { "epoch": 0.062976, "grad_norm": 0.9604840874671936, "learning_rate": 1.4792359755637798e-05, "loss": 3.1956, "step": 133500 }, { "epoch": 0.0630016, "grad_norm": 0.6764593124389648, "learning_rate": 1.479165295837994e-05, "loss": 3.1194, "step": 133510 }, { "epoch": 0.0630272, "grad_norm": 0.6651324033737183, "learning_rate": 1.479094613004948e-05, "loss": 2.8666, "step": 133520 }, { "epoch": 0.0630528, "grad_norm": 0.8400362730026245, "learning_rate": 1.4790239270651007e-05, "loss": 2.7415, "step": 133530 }, { "epoch": 0.0630784, "grad_norm": 0.7361498475074768, "learning_rate": 1.478953238018911e-05, "loss": 3.0158, "step": 133540 }, { "epoch": 0.063104, "grad_norm": 0.673339307308197, "learning_rate": 1.4788825458668368e-05, "loss": 2.7995, "step": 133550 }, { "epoch": 0.0631296, "grad_norm": 0.88953697681427, "learning_rate": 1.4788118506093362e-05, "loss": 3.0006, "step": 133560 }, { "epoch": 0.0631552, "grad_norm": 0.6979849338531494, "learning_rate": 1.4787411522468684e-05, "loss": 2.9481, "step": 133570 }, { "epoch": 0.0631808, "grad_norm": 0.8939825892448425, "learning_rate": 1.478670450779891e-05, "loss": 2.9554, "step": 133580 }, { "epoch": 0.0632064, "grad_norm": 0.802797257900238, "learning_rate": 1.4785997462088633e-05, "loss": 2.9318, "step": 133590 }, { "epoch": 0.063232, "grad_norm": 0.7127203345298767, "learning_rate": 1.4785290385342435e-05, "loss": 3.1133, "step": 133600 }, { "epoch": 0.0632576, "grad_norm": 0.7575217485427856, "learning_rate": 1.4784583277564898e-05, "loss": 3.0799, "step": 133610 }, { "epoch": 0.0632832, "grad_norm": 0.7738772630691528, "learning_rate": 1.4783876138760614e-05, "loss": 2.9203, "step": 133620 }, { "epoch": 0.0633088, "grad_norm": 0.7189076542854309, "learning_rate": 1.478316896893416e-05, "loss": 2.8893, "step": 133630 }, { "epoch": 0.0633344, "grad_norm": 0.7848626375198364, "learning_rate": 1.4782461768090128e-05, "loss": 3.044, "step": 133640 }, { "epoch": 0.06336, "grad_norm": 0.7608844041824341, "learning_rate": 1.4781754536233103e-05, "loss": 3.0258, "step": 133650 }, { "epoch": 0.0633856, "grad_norm": 0.8520258665084839, "learning_rate": 1.4781047273367671e-05, "loss": 3.1076, "step": 133660 }, { "epoch": 0.0634112, "grad_norm": 0.6946230530738831, "learning_rate": 1.4780339979498415e-05, "loss": 3.0116, "step": 133670 }, { "epoch": 0.0634368, "grad_norm": 0.714830219745636, "learning_rate": 1.4779632654629929e-05, "loss": 2.9837, "step": 133680 }, { "epoch": 0.0634624, "grad_norm": 0.7749878764152527, "learning_rate": 1.4778925298766793e-05, "loss": 2.9256, "step": 133690 }, { "epoch": 0.063488, "grad_norm": 0.6833897233009338, "learning_rate": 1.4778217911913598e-05, "loss": 2.9142, "step": 133700 }, { "epoch": 0.0635136, "grad_norm": 0.8587101697921753, "learning_rate": 1.477751049407493e-05, "loss": 2.8729, "step": 133710 }, { "epoch": 0.0635392, "grad_norm": 0.8076624870300293, "learning_rate": 1.4776803045255377e-05, "loss": 2.978, "step": 133720 }, { "epoch": 0.0635648, "grad_norm": 0.7015475034713745, "learning_rate": 1.4776095565459523e-05, "loss": 3.1822, "step": 133730 }, { "epoch": 0.0635904, "grad_norm": 0.6339576840400696, "learning_rate": 1.477538805469196e-05, "loss": 3.0479, "step": 133740 }, { "epoch": 0.063616, "grad_norm": 1.1323131322860718, "learning_rate": 1.4774680512957274e-05, "loss": 2.9388, "step": 133750 }, { "epoch": 0.0636416, "grad_norm": 0.6813247799873352, "learning_rate": 1.4773972940260053e-05, "loss": 2.9171, "step": 133760 }, { "epoch": 0.0636672, "grad_norm": 0.672659695148468, "learning_rate": 1.4773265336604885e-05, "loss": 2.8993, "step": 133770 }, { "epoch": 0.0636928, "grad_norm": 0.7105011343955994, "learning_rate": 1.4772557701996367e-05, "loss": 2.9906, "step": 133780 }, { "epoch": 0.0637184, "grad_norm": 0.7128755450248718, "learning_rate": 1.4771850036439073e-05, "loss": 2.9229, "step": 133790 }, { "epoch": 0.063744, "grad_norm": 0.7292500734329224, "learning_rate": 1.4771142339937603e-05, "loss": 2.9002, "step": 133800 }, { "epoch": 0.0637696, "grad_norm": 0.9710416197776794, "learning_rate": 1.4770434612496542e-05, "loss": 2.8796, "step": 133810 }, { "epoch": 0.0637952, "grad_norm": 0.7205061912536621, "learning_rate": 1.4769726854120482e-05, "loss": 3.0229, "step": 133820 }, { "epoch": 0.0638208, "grad_norm": 0.7147035598754883, "learning_rate": 1.476901906481401e-05, "loss": 3.1365, "step": 133830 }, { "epoch": 0.0638464, "grad_norm": 0.6557835936546326, "learning_rate": 1.4768311244581716e-05, "loss": 3.0761, "step": 133840 }, { "epoch": 0.063872, "grad_norm": 0.6478001475334167, "learning_rate": 1.4767603393428192e-05, "loss": 2.945, "step": 133850 }, { "epoch": 0.0638976, "grad_norm": 0.7852048873901367, "learning_rate": 1.4766895511358026e-05, "loss": 3.027, "step": 133860 }, { "epoch": 0.0639232, "grad_norm": 0.7370321750640869, "learning_rate": 1.476618759837581e-05, "loss": 2.9004, "step": 133870 }, { "epoch": 0.0639488, "grad_norm": 0.7242926359176636, "learning_rate": 1.4765479654486135e-05, "loss": 3.0189, "step": 133880 }, { "epoch": 0.0639744, "grad_norm": 1.288878321647644, "learning_rate": 1.476477167969359e-05, "loss": 3.0524, "step": 133890 }, { "epoch": 0.064, "grad_norm": 0.7232392430305481, "learning_rate": 1.4764063674002766e-05, "loss": 2.8843, "step": 133900 }, { "epoch": 0.0640256, "grad_norm": 0.7605088949203491, "learning_rate": 1.476335563741826e-05, "loss": 2.8837, "step": 133910 }, { "epoch": 0.0640512, "grad_norm": 0.7157727479934692, "learning_rate": 1.4762647569944655e-05, "loss": 3.0865, "step": 133920 }, { "epoch": 0.0640768, "grad_norm": 0.6765753030776978, "learning_rate": 1.4761939471586542e-05, "loss": 2.9424, "step": 133930 }, { "epoch": 0.0641024, "grad_norm": 0.6650036573410034, "learning_rate": 1.4761231342348523e-05, "loss": 3.0902, "step": 133940 }, { "epoch": 0.064128, "grad_norm": 0.7741455435752869, "learning_rate": 1.4760523182235183e-05, "loss": 2.8878, "step": 133950 }, { "epoch": 0.0641536, "grad_norm": 0.8386836647987366, "learning_rate": 1.4759814991251113e-05, "loss": 2.9675, "step": 133960 }, { "epoch": 0.0641792, "grad_norm": 0.7216668128967285, "learning_rate": 1.4759106769400911e-05, "loss": 3.0798, "step": 133970 }, { "epoch": 0.0642048, "grad_norm": 0.743595540523529, "learning_rate": 1.4758398516689165e-05, "loss": 2.8118, "step": 133980 }, { "epoch": 0.0642304, "grad_norm": 0.7542601823806763, "learning_rate": 1.4757690233120469e-05, "loss": 3.1065, "step": 133990 }, { "epoch": 0.064256, "grad_norm": 0.7791390419006348, "learning_rate": 1.4756981918699418e-05, "loss": 3.0407, "step": 134000 }, { "epoch": 0.0642816, "grad_norm": 0.717379629611969, "learning_rate": 1.4756273573430601e-05, "loss": 2.6169, "step": 134010 }, { "epoch": 0.0643072, "grad_norm": 0.7238845825195312, "learning_rate": 1.4755565197318616e-05, "loss": 3.4784, "step": 134020 }, { "epoch": 0.0643328, "grad_norm": 0.6823868155479431, "learning_rate": 1.4754856790368052e-05, "loss": 3.0079, "step": 134030 }, { "epoch": 0.0643584, "grad_norm": 0.9546977877616882, "learning_rate": 1.4754148352583508e-05, "loss": 3.1343, "step": 134040 }, { "epoch": 0.064384, "grad_norm": 0.708818256855011, "learning_rate": 1.4753439883969574e-05, "loss": 2.8562, "step": 134050 }, { "epoch": 0.0644096, "grad_norm": 1.3768192529678345, "learning_rate": 1.4752731384530847e-05, "loss": 2.9285, "step": 134060 }, { "epoch": 0.0644352, "grad_norm": 0.7427321076393127, "learning_rate": 1.4752022854271918e-05, "loss": 2.99, "step": 134070 }, { "epoch": 0.0644608, "grad_norm": 0.6789559721946716, "learning_rate": 1.4751314293197387e-05, "loss": 3.0335, "step": 134080 }, { "epoch": 0.0644864, "grad_norm": 0.6940761208534241, "learning_rate": 1.4750605701311841e-05, "loss": 2.9871, "step": 134090 }, { "epoch": 0.064512, "grad_norm": 0.728184163570404, "learning_rate": 1.4749897078619883e-05, "loss": 2.8895, "step": 134100 }, { "epoch": 0.0645376, "grad_norm": 0.7564554810523987, "learning_rate": 1.4749188425126104e-05, "loss": 2.9853, "step": 134110 }, { "epoch": 0.0645632, "grad_norm": 0.7780801653862, "learning_rate": 1.47484797408351e-05, "loss": 3.0469, "step": 134120 }, { "epoch": 0.0645888, "grad_norm": 0.6825897097587585, "learning_rate": 1.4747771025751467e-05, "loss": 3.0675, "step": 134130 }, { "epoch": 0.0646144, "grad_norm": 0.7986465096473694, "learning_rate": 1.4747062279879802e-05, "loss": 2.9824, "step": 134140 }, { "epoch": 0.06464, "grad_norm": 0.6555761098861694, "learning_rate": 1.47463535032247e-05, "loss": 2.9422, "step": 134150 }, { "epoch": 0.0646656, "grad_norm": 0.7755833864212036, "learning_rate": 1.4745644695790753e-05, "loss": 2.8908, "step": 134160 }, { "epoch": 0.0646912, "grad_norm": 0.7244433760643005, "learning_rate": 1.4744935857582566e-05, "loss": 2.9082, "step": 134170 }, { "epoch": 0.0647168, "grad_norm": 0.7577725648880005, "learning_rate": 1.474422698860473e-05, "loss": 3.1707, "step": 134180 }, { "epoch": 0.0647424, "grad_norm": 0.723246157169342, "learning_rate": 1.4743518088861838e-05, "loss": 3.0857, "step": 134190 }, { "epoch": 0.064768, "grad_norm": 0.9770336151123047, "learning_rate": 1.4742809158358498e-05, "loss": 2.9201, "step": 134200 }, { "epoch": 0.0647936, "grad_norm": 0.7840180993080139, "learning_rate": 1.4742100197099296e-05, "loss": 3.1924, "step": 134210 }, { "epoch": 0.0648192, "grad_norm": 0.7101062536239624, "learning_rate": 1.4741391205088838e-05, "loss": 3.0202, "step": 134220 }, { "epoch": 0.0648448, "grad_norm": 0.7179352045059204, "learning_rate": 1.4740682182331716e-05, "loss": 2.8979, "step": 134230 }, { "epoch": 0.0648704, "grad_norm": 0.7400736212730408, "learning_rate": 1.4739973128832535e-05, "loss": 3.0427, "step": 134240 }, { "epoch": 0.064896, "grad_norm": 1.6405770778656006, "learning_rate": 1.4739264044595883e-05, "loss": 3.0625, "step": 134250 }, { "epoch": 0.0649216, "grad_norm": 0.7823337912559509, "learning_rate": 1.4738554929626365e-05, "loss": 3.1297, "step": 134260 }, { "epoch": 0.0649472, "grad_norm": 0.91357421875, "learning_rate": 1.4737845783928575e-05, "loss": 2.956, "step": 134270 }, { "epoch": 0.0649728, "grad_norm": 0.6793233156204224, "learning_rate": 1.4737136607507117e-05, "loss": 2.9935, "step": 134280 }, { "epoch": 0.0649984, "grad_norm": 1.0360534191131592, "learning_rate": 1.4736427400366586e-05, "loss": 2.8974, "step": 134290 }, { "epoch": 0.065024, "grad_norm": 0.8488708734512329, "learning_rate": 1.4735718162511585e-05, "loss": 2.9552, "step": 134300 }, { "epoch": 0.0650496, "grad_norm": 0.7201096415519714, "learning_rate": 1.4735008893946708e-05, "loss": 3.0999, "step": 134310 }, { "epoch": 0.0650752, "grad_norm": 0.7825139760971069, "learning_rate": 1.4734299594676556e-05, "loss": 2.9229, "step": 134320 }, { "epoch": 0.0651008, "grad_norm": 0.6690295934677124, "learning_rate": 1.4733590264705728e-05, "loss": 3.0483, "step": 134330 }, { "epoch": 0.0651264, "grad_norm": 0.653903067111969, "learning_rate": 1.4732880904038828e-05, "loss": 3.0566, "step": 134340 }, { "epoch": 0.065152, "grad_norm": 0.6795715093612671, "learning_rate": 1.4732171512680454e-05, "loss": 3.0667, "step": 134350 }, { "epoch": 0.0651776, "grad_norm": 0.6710231304168701, "learning_rate": 1.4731462090635203e-05, "loss": 2.9945, "step": 134360 }, { "epoch": 0.0652032, "grad_norm": 0.8035829663276672, "learning_rate": 1.4730752637907677e-05, "loss": 3.1179, "step": 134370 }, { "epoch": 0.0652288, "grad_norm": 0.7893949151039124, "learning_rate": 1.473004315450248e-05, "loss": 2.9384, "step": 134380 }, { "epoch": 0.0652544, "grad_norm": 0.7842835187911987, "learning_rate": 1.4729333640424208e-05, "loss": 3.0861, "step": 134390 }, { "epoch": 0.06528, "grad_norm": 0.6813676357269287, "learning_rate": 1.4728624095677468e-05, "loss": 3.1456, "step": 134400 }, { "epoch": 0.0653056, "grad_norm": 0.6566843390464783, "learning_rate": 1.4727914520266855e-05, "loss": 3.0524, "step": 134410 }, { "epoch": 0.0653312, "grad_norm": 0.6711104512214661, "learning_rate": 1.4727204914196972e-05, "loss": 3.2819, "step": 134420 }, { "epoch": 0.0653568, "grad_norm": 0.7713280916213989, "learning_rate": 1.4726495277472422e-05, "loss": 2.9734, "step": 134430 }, { "epoch": 0.0653824, "grad_norm": 0.744461178779602, "learning_rate": 1.472578561009781e-05, "loss": 3.0721, "step": 134440 }, { "epoch": 0.065408, "grad_norm": 0.7314940690994263, "learning_rate": 1.4725075912077726e-05, "loss": 3.0232, "step": 134450 }, { "epoch": 0.0654336, "grad_norm": 0.8827231526374817, "learning_rate": 1.4724366183416785e-05, "loss": 3.0922, "step": 134460 }, { "epoch": 0.0654592, "grad_norm": 0.7309171557426453, "learning_rate": 1.4723656424119587e-05, "loss": 3.0134, "step": 134470 }, { "epoch": 0.0654848, "grad_norm": 0.7749043107032776, "learning_rate": 1.4722946634190732e-05, "loss": 3.2699, "step": 134480 }, { "epoch": 0.0655104, "grad_norm": 0.7093462944030762, "learning_rate": 1.472223681363482e-05, "loss": 3.0116, "step": 134490 }, { "epoch": 0.065536, "grad_norm": 1.1261405944824219, "learning_rate": 1.472152696245646e-05, "loss": 3.0155, "step": 134500 }, { "epoch": 0.0655616, "grad_norm": 0.7665766477584839, "learning_rate": 1.4720817080660252e-05, "loss": 2.8289, "step": 134510 }, { "epoch": 0.0655872, "grad_norm": 0.7285558581352234, "learning_rate": 1.4720107168250798e-05, "loss": 2.9907, "step": 134520 }, { "epoch": 0.0656128, "grad_norm": 0.6719523072242737, "learning_rate": 1.4719397225232705e-05, "loss": 3.1216, "step": 134530 }, { "epoch": 0.0656384, "grad_norm": 0.7315959930419922, "learning_rate": 1.4718687251610578e-05, "loss": 3.171, "step": 134540 }, { "epoch": 0.065664, "grad_norm": 0.7388892769813538, "learning_rate": 1.4717977247389014e-05, "loss": 3.0828, "step": 134550 }, { "epoch": 0.0656896, "grad_norm": 0.8698478937149048, "learning_rate": 1.4717267212572624e-05, "loss": 3.1566, "step": 134560 }, { "epoch": 0.0657152, "grad_norm": 0.6874498128890991, "learning_rate": 1.4716557147166008e-05, "loss": 2.9843, "step": 134570 }, { "epoch": 0.0657408, "grad_norm": 0.8068462610244751, "learning_rate": 1.4715847051173774e-05, "loss": 3.2189, "step": 134580 }, { "epoch": 0.0657664, "grad_norm": 0.6738618016242981, "learning_rate": 1.4715136924600524e-05, "loss": 3.0856, "step": 134590 }, { "epoch": 0.065792, "grad_norm": 0.7022135257720947, "learning_rate": 1.4714426767450867e-05, "loss": 3.0771, "step": 134600 }, { "epoch": 0.0658176, "grad_norm": 0.7546777129173279, "learning_rate": 1.4713716579729402e-05, "loss": 2.9867, "step": 134610 }, { "epoch": 0.0658432, "grad_norm": 0.7427181005477905, "learning_rate": 1.4713006361440737e-05, "loss": 3.252, "step": 134620 }, { "epoch": 0.0658688, "grad_norm": 0.7373681664466858, "learning_rate": 1.4712296112589482e-05, "loss": 3.0017, "step": 134630 }, { "epoch": 0.0658944, "grad_norm": 0.7202866077423096, "learning_rate": 1.4711585833180236e-05, "loss": 3.2728, "step": 134640 }, { "epoch": 0.06592, "grad_norm": 0.7535549998283386, "learning_rate": 1.471087552321761e-05, "loss": 3.0997, "step": 134650 }, { "epoch": 0.0659456, "grad_norm": 0.7201194763183594, "learning_rate": 1.4710165182706204e-05, "loss": 3.0226, "step": 134660 }, { "epoch": 0.0659712, "grad_norm": 0.8133584856987, "learning_rate": 1.4709454811650634e-05, "loss": 2.9322, "step": 134670 }, { "epoch": 0.0659968, "grad_norm": 0.7047604918479919, "learning_rate": 1.47087444100555e-05, "loss": 2.9643, "step": 134680 }, { "epoch": 0.0660224, "grad_norm": 0.6654903292655945, "learning_rate": 1.4708033977925408e-05, "loss": 2.9951, "step": 134690 }, { "epoch": 0.066048, "grad_norm": 0.6725111603736877, "learning_rate": 1.470732351526497e-05, "loss": 3.1101, "step": 134700 }, { "epoch": 0.0660736, "grad_norm": 0.7403898239135742, "learning_rate": 1.4706613022078786e-05, "loss": 2.9528, "step": 134710 }, { "epoch": 0.0660992, "grad_norm": 0.6664618253707886, "learning_rate": 1.470590249837147e-05, "loss": 2.9277, "step": 134720 }, { "epoch": 0.0661248, "grad_norm": 0.8116287589073181, "learning_rate": 1.4705191944147628e-05, "loss": 3.0749, "step": 134730 }, { "epoch": 0.0661504, "grad_norm": 0.7658675312995911, "learning_rate": 1.4704481359411864e-05, "loss": 3.0632, "step": 134740 }, { "epoch": 0.066176, "grad_norm": 0.6406140923500061, "learning_rate": 1.4703770744168791e-05, "loss": 2.9983, "step": 134750 }, { "epoch": 0.0662016, "grad_norm": 0.8248341679573059, "learning_rate": 1.4703060098423011e-05, "loss": 3.0886, "step": 134760 }, { "epoch": 0.0662272, "grad_norm": 0.6400077939033508, "learning_rate": 1.4702349422179141e-05, "loss": 2.908, "step": 134770 }, { "epoch": 0.0662528, "grad_norm": 0.8523517847061157, "learning_rate": 1.4701638715441784e-05, "loss": 2.8721, "step": 134780 }, { "epoch": 0.0662784, "grad_norm": 0.7148156762123108, "learning_rate": 1.4700927978215546e-05, "loss": 3.0009, "step": 134790 }, { "epoch": 0.066304, "grad_norm": 0.7658489346504211, "learning_rate": 1.4700217210505043e-05, "loss": 3.0939, "step": 134800 }, { "epoch": 0.0663296, "grad_norm": 0.8088801503181458, "learning_rate": 1.4699506412314878e-05, "loss": 3.1079, "step": 134810 }, { "epoch": 0.0663552, "grad_norm": 0.7493599653244019, "learning_rate": 1.4698795583649664e-05, "loss": 2.8994, "step": 134820 }, { "epoch": 0.0663808, "grad_norm": 0.788036584854126, "learning_rate": 1.4698084724514011e-05, "loss": 2.9184, "step": 134830 }, { "epoch": 0.0664064, "grad_norm": 0.8815810084342957, "learning_rate": 1.4697373834912525e-05, "loss": 3.0156, "step": 134840 }, { "epoch": 0.066432, "grad_norm": 0.7123252749443054, "learning_rate": 1.4696662914849817e-05, "loss": 2.9484, "step": 134850 }, { "epoch": 0.0664576, "grad_norm": 0.6480118036270142, "learning_rate": 1.4695951964330502e-05, "loss": 3.0145, "step": 134860 }, { "epoch": 0.0664832, "grad_norm": 0.7682172060012817, "learning_rate": 1.4695240983359184e-05, "loss": 3.0206, "step": 134870 }, { "epoch": 0.0665088, "grad_norm": 0.8711473941802979, "learning_rate": 1.4694529971940475e-05, "loss": 3.0183, "step": 134880 }, { "epoch": 0.0665344, "grad_norm": 0.7062976360321045, "learning_rate": 1.469381893007899e-05, "loss": 2.8368, "step": 134890 }, { "epoch": 0.06656, "grad_norm": 0.7035473585128784, "learning_rate": 1.4693107857779335e-05, "loss": 3.0141, "step": 134900 }, { "epoch": 0.0665856, "grad_norm": 0.903220534324646, "learning_rate": 1.4692396755046124e-05, "loss": 3.0085, "step": 134910 }, { "epoch": 0.0666112, "grad_norm": 0.7235351204872131, "learning_rate": 1.4691685621883964e-05, "loss": 3.1368, "step": 134920 }, { "epoch": 0.0666368, "grad_norm": 0.7518353462219238, "learning_rate": 1.4690974458297474e-05, "loss": 2.9073, "step": 134930 }, { "epoch": 0.0666624, "grad_norm": 0.8241447806358337, "learning_rate": 1.4690263264291259e-05, "loss": 3.1355, "step": 134940 }, { "epoch": 0.066688, "grad_norm": 0.6858693957328796, "learning_rate": 1.4689552039869932e-05, "loss": 2.9343, "step": 134950 }, { "epoch": 0.0667136, "grad_norm": 0.7605384588241577, "learning_rate": 1.4688840785038108e-05, "loss": 3.0058, "step": 134960 }, { "epoch": 0.0667392, "grad_norm": 1.0361164808273315, "learning_rate": 1.4688129499800397e-05, "loss": 3.2213, "step": 134970 }, { "epoch": 0.0667648, "grad_norm": 0.7885130643844604, "learning_rate": 1.468741818416141e-05, "loss": 2.8651, "step": 134980 }, { "epoch": 0.0667904, "grad_norm": 0.7515611052513123, "learning_rate": 1.4686706838125767e-05, "loss": 2.9916, "step": 134990 }, { "epoch": 0.066816, "grad_norm": 0.6674558520317078, "learning_rate": 1.4685995461698072e-05, "loss": 3.1197, "step": 135000 }, { "epoch": 0.0668416, "grad_norm": 0.7138276100158691, "learning_rate": 1.4685284054882942e-05, "loss": 3.0301, "step": 135010 }, { "epoch": 0.0668672, "grad_norm": 1.0501049757003784, "learning_rate": 1.468457261768499e-05, "loss": 2.9121, "step": 135020 }, { "epoch": 0.0668928, "grad_norm": 0.6458745002746582, "learning_rate": 1.4683861150108833e-05, "loss": 3.2045, "step": 135030 }, { "epoch": 0.0669184, "grad_norm": 0.7991660833358765, "learning_rate": 1.4683149652159078e-05, "loss": 2.9907, "step": 135040 }, { "epoch": 0.066944, "grad_norm": 0.8845164775848389, "learning_rate": 1.4682438123840343e-05, "loss": 3.0142, "step": 135050 }, { "epoch": 0.0669696, "grad_norm": 0.7422481179237366, "learning_rate": 1.4681726565157242e-05, "loss": 2.8833, "step": 135060 }, { "epoch": 0.0669952, "grad_norm": 1.644012689590454, "learning_rate": 1.4681014976114387e-05, "loss": 3.1752, "step": 135070 }, { "epoch": 0.0670208, "grad_norm": 0.7408275604248047, "learning_rate": 1.4680303356716391e-05, "loss": 3.0288, "step": 135080 }, { "epoch": 0.0670464, "grad_norm": 0.7968072891235352, "learning_rate": 1.467959170696788e-05, "loss": 3.2361, "step": 135090 }, { "epoch": 0.067072, "grad_norm": 0.8094964027404785, "learning_rate": 1.4678880026873455e-05, "loss": 2.9195, "step": 135100 }, { "epoch": 0.0670976, "grad_norm": 0.7062538266181946, "learning_rate": 1.4678168316437735e-05, "loss": 3.0056, "step": 135110 }, { "epoch": 0.0671232, "grad_norm": 0.7071722149848938, "learning_rate": 1.4677456575665339e-05, "loss": 2.9986, "step": 135120 }, { "epoch": 0.0671488, "grad_norm": 0.7343522906303406, "learning_rate": 1.4676744804560882e-05, "loss": 2.9057, "step": 135130 }, { "epoch": 0.0671744, "grad_norm": 0.7222900390625, "learning_rate": 1.4676033003128974e-05, "loss": 2.9793, "step": 135140 }, { "epoch": 0.0672, "grad_norm": 0.7462278604507446, "learning_rate": 1.4675321171374238e-05, "loss": 3.2894, "step": 135150 }, { "epoch": 0.0672256, "grad_norm": 0.7906010746955872, "learning_rate": 1.4674609309301287e-05, "loss": 2.9948, "step": 135160 }, { "epoch": 0.0672512, "grad_norm": 0.7413575053215027, "learning_rate": 1.4673897416914733e-05, "loss": 3.0065, "step": 135170 }, { "epoch": 0.0672768, "grad_norm": 0.7257281541824341, "learning_rate": 1.4673185494219197e-05, "loss": 3.0386, "step": 135180 }, { "epoch": 0.0673024, "grad_norm": 0.7051306962966919, "learning_rate": 1.4672473541219301e-05, "loss": 3.2106, "step": 135190 }, { "epoch": 0.067328, "grad_norm": 0.7788839936256409, "learning_rate": 1.467176155791965e-05, "loss": 2.9369, "step": 135200 }, { "epoch": 0.0673536, "grad_norm": 0.6944273710250854, "learning_rate": 1.467104954432487e-05, "loss": 3.135, "step": 135210 }, { "epoch": 0.0673792, "grad_norm": 0.9293381571769714, "learning_rate": 1.4670337500439573e-05, "loss": 2.9604, "step": 135220 }, { "epoch": 0.0674048, "grad_norm": 0.7294458746910095, "learning_rate": 1.466962542626838e-05, "loss": 2.8384, "step": 135230 }, { "epoch": 0.0674304, "grad_norm": 0.7127357721328735, "learning_rate": 1.4668913321815906e-05, "loss": 2.944, "step": 135240 }, { "epoch": 0.067456, "grad_norm": 0.7694419622421265, "learning_rate": 1.4668201187086768e-05, "loss": 2.9757, "step": 135250 }, { "epoch": 0.0674816, "grad_norm": 0.8403909802436829, "learning_rate": 1.4667489022085591e-05, "loss": 3.1157, "step": 135260 }, { "epoch": 0.0675072, "grad_norm": 0.9239026308059692, "learning_rate": 1.4666776826816986e-05, "loss": 3.3328, "step": 135270 }, { "epoch": 0.0675328, "grad_norm": 0.7476266622543335, "learning_rate": 1.4666064601285572e-05, "loss": 3.1529, "step": 135280 }, { "epoch": 0.0675584, "grad_norm": 0.6865497827529907, "learning_rate": 1.4665352345495971e-05, "loss": 3.1701, "step": 135290 }, { "epoch": 0.067584, "grad_norm": 0.7383613586425781, "learning_rate": 1.46646400594528e-05, "loss": 3.072, "step": 135300 }, { "epoch": 0.0676096, "grad_norm": 0.6761548519134521, "learning_rate": 1.4663927743160678e-05, "loss": 3.1603, "step": 135310 }, { "epoch": 0.0676352, "grad_norm": 1.0561968088150024, "learning_rate": 1.4663215396624223e-05, "loss": 3.1109, "step": 135320 }, { "epoch": 0.0676608, "grad_norm": 0.738257646560669, "learning_rate": 1.4662503019848053e-05, "loss": 3.4456, "step": 135330 }, { "epoch": 0.0676864, "grad_norm": 1.1164264678955078, "learning_rate": 1.4661790612836793e-05, "loss": 3.0783, "step": 135340 }, { "epoch": 0.067712, "grad_norm": 1.0643606185913086, "learning_rate": 1.466107817559506e-05, "loss": 3.2466, "step": 135350 }, { "epoch": 0.0677376, "grad_norm": 0.7300990223884583, "learning_rate": 1.4660365708127476e-05, "loss": 3.122, "step": 135360 }, { "epoch": 0.0677632, "grad_norm": 0.7539510726928711, "learning_rate": 1.4659653210438655e-05, "loss": 3.38, "step": 135370 }, { "epoch": 0.0677888, "grad_norm": 0.7477542757987976, "learning_rate": 1.4658940682533222e-05, "loss": 2.9146, "step": 135380 }, { "epoch": 0.0678144, "grad_norm": 0.941024661064148, "learning_rate": 1.4658228124415799e-05, "loss": 2.9514, "step": 135390 }, { "epoch": 0.06784, "grad_norm": 1.1033174991607666, "learning_rate": 1.4657515536091002e-05, "loss": 2.9268, "step": 135400 }, { "epoch": 0.0678656, "grad_norm": 0.7359266877174377, "learning_rate": 1.4656802917563455e-05, "loss": 3.0815, "step": 135410 }, { "epoch": 0.0678912, "grad_norm": 0.7432069182395935, "learning_rate": 1.465609026883778e-05, "loss": 2.8521, "step": 135420 }, { "epoch": 0.0679168, "grad_norm": 0.7610225677490234, "learning_rate": 1.4655377589918596e-05, "loss": 3.1195, "step": 135430 }, { "epoch": 0.0679424, "grad_norm": 0.6928536295890808, "learning_rate": 1.4654664880810528e-05, "loss": 3.0276, "step": 135440 }, { "epoch": 0.067968, "grad_norm": 0.7004099488258362, "learning_rate": 1.465395214151819e-05, "loss": 3.2049, "step": 135450 }, { "epoch": 0.0679936, "grad_norm": 0.8403373956680298, "learning_rate": 1.4653239372046216e-05, "loss": 3.234, "step": 135460 }, { "epoch": 0.0680192, "grad_norm": 0.7086397409439087, "learning_rate": 1.4652526572399218e-05, "loss": 3.0452, "step": 135470 }, { "epoch": 0.0680448, "grad_norm": 0.795499861240387, "learning_rate": 1.4651813742581821e-05, "loss": 3.2952, "step": 135480 }, { "epoch": 0.0680704, "grad_norm": 0.8146211504936218, "learning_rate": 1.4651100882598647e-05, "loss": 2.9671, "step": 135490 }, { "epoch": 0.068096, "grad_norm": 0.8020851016044617, "learning_rate": 1.4650387992454322e-05, "loss": 3.0896, "step": 135500 }, { "epoch": 0.0681216, "grad_norm": 0.8771374821662903, "learning_rate": 1.4649675072153466e-05, "loss": 3.3154, "step": 135510 }, { "epoch": 0.0681472, "grad_norm": 0.9956561326980591, "learning_rate": 1.4648962121700705e-05, "loss": 3.09, "step": 135520 }, { "epoch": 0.0681728, "grad_norm": 0.712408185005188, "learning_rate": 1.4648249141100658e-05, "loss": 2.8741, "step": 135530 }, { "epoch": 0.0681984, "grad_norm": 0.8021070957183838, "learning_rate": 1.464753613035795e-05, "loss": 3.0294, "step": 135540 }, { "epoch": 0.068224, "grad_norm": 0.8889954090118408, "learning_rate": 1.464682308947721e-05, "loss": 3.0922, "step": 135550 }, { "epoch": 0.0682496, "grad_norm": 0.912135124206543, "learning_rate": 1.4646110018463056e-05, "loss": 3.0273, "step": 135560 }, { "epoch": 0.0682752, "grad_norm": 0.976922333240509, "learning_rate": 1.464539691732011e-05, "loss": 3.2113, "step": 135570 }, { "epoch": 0.0683008, "grad_norm": 0.6957268118858337, "learning_rate": 1.4644683786052998e-05, "loss": 3.0279, "step": 135580 }, { "epoch": 0.0683264, "grad_norm": 0.8099013566970825, "learning_rate": 1.464397062466635e-05, "loss": 2.9052, "step": 135590 }, { "epoch": 0.068352, "grad_norm": 0.7581726312637329, "learning_rate": 1.4643257433164785e-05, "loss": 3.1193, "step": 135600 }, { "epoch": 0.0683776, "grad_norm": 0.8791795372962952, "learning_rate": 1.4642544211552929e-05, "loss": 3.3583, "step": 135610 }, { "epoch": 0.0684032, "grad_norm": 0.7707836031913757, "learning_rate": 1.464183095983541e-05, "loss": 3.0319, "step": 135620 }, { "epoch": 0.0684288, "grad_norm": 0.6908943057060242, "learning_rate": 1.464111767801685e-05, "loss": 3.0401, "step": 135630 }, { "epoch": 0.0684544, "grad_norm": 0.8104336857795715, "learning_rate": 1.4640404366101874e-05, "loss": 2.9904, "step": 135640 }, { "epoch": 0.06848, "grad_norm": 0.8527071475982666, "learning_rate": 1.4639691024095111e-05, "loss": 3.0406, "step": 135650 }, { "epoch": 0.0685056, "grad_norm": 0.7191662788391113, "learning_rate": 1.4638977652001186e-05, "loss": 3.1157, "step": 135660 }, { "epoch": 0.0685312, "grad_norm": 0.7312744855880737, "learning_rate": 1.4638264249824719e-05, "loss": 2.7126, "step": 135670 }, { "epoch": 0.0685568, "grad_norm": 0.6863601207733154, "learning_rate": 1.4637550817570343e-05, "loss": 3.1229, "step": 135680 }, { "epoch": 0.0685824, "grad_norm": 0.6706814765930176, "learning_rate": 1.4636837355242684e-05, "loss": 3.0443, "step": 135690 }, { "epoch": 0.068608, "grad_norm": 0.8780615329742432, "learning_rate": 1.4636123862846368e-05, "loss": 2.9692, "step": 135700 }, { "epoch": 0.0686336, "grad_norm": 0.7349931001663208, "learning_rate": 1.4635410340386015e-05, "loss": 2.9995, "step": 135710 }, { "epoch": 0.0686592, "grad_norm": 0.8461744785308838, "learning_rate": 1.4634696787866264e-05, "loss": 3.2267, "step": 135720 }, { "epoch": 0.0686848, "grad_norm": 0.6807737946510315, "learning_rate": 1.4633983205291736e-05, "loss": 3.1574, "step": 135730 }, { "epoch": 0.0687104, "grad_norm": 0.7845762968063354, "learning_rate": 1.4633269592667057e-05, "loss": 2.8581, "step": 135740 }, { "epoch": 0.068736, "grad_norm": 0.7410246729850769, "learning_rate": 1.4632555949996854e-05, "loss": 3.047, "step": 135750 }, { "epoch": 0.0687616, "grad_norm": 2.2177748680114746, "learning_rate": 1.4631842277285757e-05, "loss": 3.7138, "step": 135760 }, { "epoch": 0.0687872, "grad_norm": 0.7559844255447388, "learning_rate": 1.4631128574538395e-05, "loss": 3.0229, "step": 135770 }, { "epoch": 0.0688128, "grad_norm": 0.7426838278770447, "learning_rate": 1.4630414841759394e-05, "loss": 2.8438, "step": 135780 }, { "epoch": 0.0688384, "grad_norm": 0.7113649249076843, "learning_rate": 1.4629701078953389e-05, "loss": 2.9463, "step": 135790 }, { "epoch": 0.068864, "grad_norm": 0.7289031744003296, "learning_rate": 1.4628987286124998e-05, "loss": 3.2352, "step": 135800 }, { "epoch": 0.0688896, "grad_norm": 0.810825765132904, "learning_rate": 1.4628273463278854e-05, "loss": 3.0243, "step": 135810 }, { "epoch": 0.0689152, "grad_norm": 0.6755541563034058, "learning_rate": 1.462755961041959e-05, "loss": 2.9535, "step": 135820 }, { "epoch": 0.0689408, "grad_norm": 0.8595032095909119, "learning_rate": 1.4626845727551828e-05, "loss": 3.022, "step": 135830 }, { "epoch": 0.0689664, "grad_norm": 0.6961877942085266, "learning_rate": 1.46261318146802e-05, "loss": 2.9866, "step": 135840 }, { "epoch": 0.068992, "grad_norm": 0.9493185877799988, "learning_rate": 1.4625417871809342e-05, "loss": 2.9919, "step": 135850 }, { "epoch": 0.0690176, "grad_norm": 0.7153233885765076, "learning_rate": 1.4624703898943875e-05, "loss": 3.1072, "step": 135860 }, { "epoch": 0.0690432, "grad_norm": 0.7814285159111023, "learning_rate": 1.4623989896088433e-05, "loss": 3.0443, "step": 135870 }, { "epoch": 0.0690688, "grad_norm": 0.7158147692680359, "learning_rate": 1.4623275863247647e-05, "loss": 3.0451, "step": 135880 }, { "epoch": 0.0690944, "grad_norm": 0.9062305092811584, "learning_rate": 1.4622561800426145e-05, "loss": 2.9858, "step": 135890 }, { "epoch": 0.06912, "grad_norm": 0.7931305766105652, "learning_rate": 1.4621847707628556e-05, "loss": 3.1955, "step": 135900 }, { "epoch": 0.0691456, "grad_norm": 0.7335234880447388, "learning_rate": 1.4621133584859514e-05, "loss": 2.9513, "step": 135910 }, { "epoch": 0.0691712, "grad_norm": 0.8481730222702026, "learning_rate": 1.4620419432123652e-05, "loss": 3.0121, "step": 135920 }, { "epoch": 0.0691968, "grad_norm": 0.7107898592948914, "learning_rate": 1.4619705249425592e-05, "loss": 2.8557, "step": 135930 }, { "epoch": 0.0692224, "grad_norm": 0.7127259969711304, "learning_rate": 1.4618991036769977e-05, "loss": 2.8467, "step": 135940 }, { "epoch": 0.069248, "grad_norm": 0.7246230840682983, "learning_rate": 1.461827679416143e-05, "loss": 3.0062, "step": 135950 }, { "epoch": 0.0692736, "grad_norm": 0.832417905330658, "learning_rate": 1.4617562521604585e-05, "loss": 3.1187, "step": 135960 }, { "epoch": 0.0692992, "grad_norm": 0.8029674887657166, "learning_rate": 1.4616848219104074e-05, "loss": 2.9957, "step": 135970 }, { "epoch": 0.0693248, "grad_norm": 1.0517452955245972, "learning_rate": 1.4616133886664532e-05, "loss": 3.0036, "step": 135980 }, { "epoch": 0.0693504, "grad_norm": 0.858705997467041, "learning_rate": 1.4615419524290586e-05, "loss": 2.9344, "step": 135990 }, { "epoch": 0.069376, "grad_norm": 0.7615057229995728, "learning_rate": 1.4614705131986872e-05, "loss": 2.9223, "step": 136000 }, { "epoch": 0.0694016, "grad_norm": 0.6910858154296875, "learning_rate": 1.4613990709758021e-05, "loss": 3.0781, "step": 136010 }, { "epoch": 0.0694272, "grad_norm": 0.7932119369506836, "learning_rate": 1.4613276257608666e-05, "loss": 2.9695, "step": 136020 }, { "epoch": 0.0694528, "grad_norm": 0.7083530426025391, "learning_rate": 1.461256177554344e-05, "loss": 2.9265, "step": 136030 }, { "epoch": 0.0694784, "grad_norm": 1.1978042125701904, "learning_rate": 1.461184726356698e-05, "loss": 3.1891, "step": 136040 }, { "epoch": 0.069504, "grad_norm": 1.2924920320510864, "learning_rate": 1.4611132721683917e-05, "loss": 3.1546, "step": 136050 }, { "epoch": 0.0695296, "grad_norm": 0.7197518944740295, "learning_rate": 1.461041814989888e-05, "loss": 3.0943, "step": 136060 }, { "epoch": 0.0695552, "grad_norm": 1.1275070905685425, "learning_rate": 1.4609703548216507e-05, "loss": 3.1886, "step": 136070 }, { "epoch": 0.0695808, "grad_norm": 0.9882824420928955, "learning_rate": 1.4608988916641433e-05, "loss": 3.1223, "step": 136080 }, { "epoch": 0.0696064, "grad_norm": 0.7189898490905762, "learning_rate": 1.4608274255178292e-05, "loss": 3.1304, "step": 136090 }, { "epoch": 0.069632, "grad_norm": 1.5694218873977661, "learning_rate": 1.4607559563831713e-05, "loss": 3.0898, "step": 136100 }, { "epoch": 0.0696576, "grad_norm": 0.7654491662979126, "learning_rate": 1.4606844842606339e-05, "loss": 3.2202, "step": 136110 }, { "epoch": 0.0696832, "grad_norm": 0.7657212615013123, "learning_rate": 1.4606130091506798e-05, "loss": 2.9105, "step": 136120 }, { "epoch": 0.0697088, "grad_norm": 0.8711572885513306, "learning_rate": 1.4605415310537727e-05, "loss": 2.8097, "step": 136130 }, { "epoch": 0.0697344, "grad_norm": 0.8172405958175659, "learning_rate": 1.4604700499703763e-05, "loss": 3.0992, "step": 136140 }, { "epoch": 0.06976, "grad_norm": 0.726107120513916, "learning_rate": 1.4603985659009541e-05, "loss": 3.0576, "step": 136150 }, { "epoch": 0.0697856, "grad_norm": 0.6617157459259033, "learning_rate": 1.4603270788459693e-05, "loss": 3.0463, "step": 136160 }, { "epoch": 0.0698112, "grad_norm": 0.7863369584083557, "learning_rate": 1.4602555888058861e-05, "loss": 3.0411, "step": 136170 }, { "epoch": 0.0698368, "grad_norm": 0.6727924346923828, "learning_rate": 1.4601840957811674e-05, "loss": 2.941, "step": 136180 }, { "epoch": 0.0698624, "grad_norm": 0.8126001954078674, "learning_rate": 1.4601125997722772e-05, "loss": 2.8494, "step": 136190 }, { "epoch": 0.069888, "grad_norm": 0.6851103901863098, "learning_rate": 1.4600411007796791e-05, "loss": 2.9119, "step": 136200 }, { "epoch": 0.0699136, "grad_norm": 0.8325366973876953, "learning_rate": 1.459969598803837e-05, "loss": 3.04, "step": 136210 }, { "epoch": 0.0699392, "grad_norm": 0.7774674296379089, "learning_rate": 1.4598980938452138e-05, "loss": 3.0874, "step": 136220 }, { "epoch": 0.0699648, "grad_norm": 0.6695281863212585, "learning_rate": 1.4598265859042739e-05, "loss": 3.1389, "step": 136230 }, { "epoch": 0.0699904, "grad_norm": 0.8054704070091248, "learning_rate": 1.4597550749814809e-05, "loss": 3.3243, "step": 136240 }, { "epoch": 0.070016, "grad_norm": 0.6938163042068481, "learning_rate": 1.4596835610772986e-05, "loss": 2.909, "step": 136250 }, { "epoch": 0.0700416, "grad_norm": 0.742913007736206, "learning_rate": 1.4596120441921904e-05, "loss": 3.2654, "step": 136260 }, { "epoch": 0.0700672, "grad_norm": 0.80588299036026, "learning_rate": 1.45954052432662e-05, "loss": 3.1034, "step": 136270 }, { "epoch": 0.0700928, "grad_norm": 0.7063839435577393, "learning_rate": 1.4594690014810515e-05, "loss": 3.0144, "step": 136280 }, { "epoch": 0.0701184, "grad_norm": 0.7091802358627319, "learning_rate": 1.4593974756559488e-05, "loss": 2.9583, "step": 136290 }, { "epoch": 0.070144, "grad_norm": 0.6684953570365906, "learning_rate": 1.4593259468517756e-05, "loss": 3.2227, "step": 136300 }, { "epoch": 0.0701696, "grad_norm": 0.6920188069343567, "learning_rate": 1.4592544150689958e-05, "loss": 2.9604, "step": 136310 }, { "epoch": 0.0701952, "grad_norm": 0.6852650046348572, "learning_rate": 1.4591828803080728e-05, "loss": 2.9377, "step": 136320 }, { "epoch": 0.0702208, "grad_norm": 0.6603186130523682, "learning_rate": 1.4591113425694713e-05, "loss": 3.019, "step": 136330 }, { "epoch": 0.0702464, "grad_norm": 0.7943592071533203, "learning_rate": 1.4590398018536543e-05, "loss": 3.2238, "step": 136340 }, { "epoch": 0.070272, "grad_norm": 0.6731729507446289, "learning_rate": 1.4589682581610865e-05, "loss": 2.883, "step": 136350 }, { "epoch": 0.0702976, "grad_norm": 0.8234896659851074, "learning_rate": 1.4588967114922312e-05, "loss": 3.2149, "step": 136360 }, { "epoch": 0.0703232, "grad_norm": 0.8117375373840332, "learning_rate": 1.4588251618475528e-05, "loss": 2.9592, "step": 136370 }, { "epoch": 0.0703488, "grad_norm": 0.8761759400367737, "learning_rate": 1.4587536092275153e-05, "loss": 2.9612, "step": 136380 }, { "epoch": 0.0703744, "grad_norm": 0.6725341081619263, "learning_rate": 1.4586820536325828e-05, "loss": 3.0151, "step": 136390 }, { "epoch": 0.0704, "grad_norm": 0.6597015857696533, "learning_rate": 1.4586104950632187e-05, "loss": 3.0553, "step": 136400 }, { "epoch": 0.0704256, "grad_norm": 0.7479566335678101, "learning_rate": 1.4585389335198877e-05, "loss": 2.8841, "step": 136410 }, { "epoch": 0.0704512, "grad_norm": 0.6920589208602905, "learning_rate": 1.4584673690030533e-05, "loss": 3.1077, "step": 136420 }, { "epoch": 0.0704768, "grad_norm": 1.2202144861221313, "learning_rate": 1.4583958015131801e-05, "loss": 3.0968, "step": 136430 }, { "epoch": 0.0705024, "grad_norm": 0.8387539386749268, "learning_rate": 1.4583242310507318e-05, "loss": 3.1767, "step": 136440 }, { "epoch": 0.070528, "grad_norm": 0.7826353907585144, "learning_rate": 1.4582526576161727e-05, "loss": 3.0151, "step": 136450 }, { "epoch": 0.0705536, "grad_norm": 0.7320989966392517, "learning_rate": 1.4581810812099668e-05, "loss": 2.9463, "step": 136460 }, { "epoch": 0.0705792, "grad_norm": 0.7257803678512573, "learning_rate": 1.4581095018325786e-05, "loss": 3.0257, "step": 136470 }, { "epoch": 0.0706048, "grad_norm": 0.7184851765632629, "learning_rate": 1.458037919484472e-05, "loss": 3.1324, "step": 136480 }, { "epoch": 0.0706304, "grad_norm": 0.7452049255371094, "learning_rate": 1.457966334166111e-05, "loss": 3.0261, "step": 136490 }, { "epoch": 0.070656, "grad_norm": 0.7200428247451782, "learning_rate": 1.4578947458779603e-05, "loss": 2.9507, "step": 136500 }, { "epoch": 0.0706816, "grad_norm": 0.794300377368927, "learning_rate": 1.4578231546204837e-05, "loss": 3.0973, "step": 136510 }, { "epoch": 0.0707072, "grad_norm": 0.7633105516433716, "learning_rate": 1.4577515603941457e-05, "loss": 3.5563, "step": 136520 }, { "epoch": 0.0707328, "grad_norm": 0.7345170378684998, "learning_rate": 1.4576799631994102e-05, "loss": 2.99, "step": 136530 }, { "epoch": 0.0707584, "grad_norm": 0.7308512330055237, "learning_rate": 1.4576083630367421e-05, "loss": 3.0003, "step": 136540 }, { "epoch": 0.070784, "grad_norm": 0.7518207430839539, "learning_rate": 1.4575367599066055e-05, "loss": 3.1581, "step": 136550 }, { "epoch": 0.0708096, "grad_norm": 2.3077304363250732, "learning_rate": 1.4574651538094641e-05, "loss": 3.3374, "step": 136560 }, { "epoch": 0.0708352, "grad_norm": 0.7332225441932678, "learning_rate": 1.4573935447457831e-05, "loss": 3.1833, "step": 136570 }, { "epoch": 0.0708608, "grad_norm": 0.9505400657653809, "learning_rate": 1.4573219327160265e-05, "loss": 2.9843, "step": 136580 }, { "epoch": 0.0708864, "grad_norm": 0.671355664730072, "learning_rate": 1.457250317720659e-05, "loss": 2.9814, "step": 136590 }, { "epoch": 0.070912, "grad_norm": 0.7220054864883423, "learning_rate": 1.4571786997601443e-05, "loss": 3.0538, "step": 136600 }, { "epoch": 0.0709376, "grad_norm": 0.7178427577018738, "learning_rate": 1.4571070788349474e-05, "loss": 2.9903, "step": 136610 }, { "epoch": 0.0709632, "grad_norm": 0.7705311179161072, "learning_rate": 1.4570354549455324e-05, "loss": 2.9425, "step": 136620 }, { "epoch": 0.0709888, "grad_norm": 0.8860849142074585, "learning_rate": 1.456963828092364e-05, "loss": 3.2925, "step": 136630 }, { "epoch": 0.0710144, "grad_norm": 0.8103097081184387, "learning_rate": 1.4568921982759068e-05, "loss": 3.0409, "step": 136640 }, { "epoch": 0.07104, "grad_norm": 0.6721123456954956, "learning_rate": 1.456820565496625e-05, "loss": 2.9949, "step": 136650 }, { "epoch": 0.0710656, "grad_norm": 0.690711259841919, "learning_rate": 1.4567489297549832e-05, "loss": 3.0874, "step": 136660 }, { "epoch": 0.0710912, "grad_norm": 0.7091537714004517, "learning_rate": 1.456677291051446e-05, "loss": 3.0132, "step": 136670 }, { "epoch": 0.0711168, "grad_norm": 0.7083098888397217, "learning_rate": 1.4566056493864781e-05, "loss": 3.02, "step": 136680 }, { "epoch": 0.0711424, "grad_norm": 0.6727330684661865, "learning_rate": 1.4565340047605439e-05, "loss": 3.073, "step": 136690 }, { "epoch": 0.071168, "grad_norm": 0.7178286910057068, "learning_rate": 1.4564623571741077e-05, "loss": 3.025, "step": 136700 }, { "epoch": 0.0711936, "grad_norm": 0.8229320049285889, "learning_rate": 1.4563907066276346e-05, "loss": 3.1641, "step": 136710 }, { "epoch": 0.0712192, "grad_norm": 0.6700177192687988, "learning_rate": 1.4563190531215892e-05, "loss": 2.9634, "step": 136720 }, { "epoch": 0.0712448, "grad_norm": 0.7569386959075928, "learning_rate": 1.4562473966564357e-05, "loss": 2.9166, "step": 136730 }, { "epoch": 0.0712704, "grad_norm": 0.7231523990631104, "learning_rate": 1.4561757372326396e-05, "loss": 2.9087, "step": 136740 }, { "epoch": 0.071296, "grad_norm": 0.6756207942962646, "learning_rate": 1.4561040748506647e-05, "loss": 3.1754, "step": 136750 }, { "epoch": 0.0713216, "grad_norm": 0.7271532416343689, "learning_rate": 1.456032409510976e-05, "loss": 3.1163, "step": 136760 }, { "epoch": 0.0713472, "grad_norm": 4.679971694946289, "learning_rate": 1.4559607412140384e-05, "loss": 3.0337, "step": 136770 }, { "epoch": 0.0713728, "grad_norm": 0.796703040599823, "learning_rate": 1.455889069960317e-05, "loss": 3.046, "step": 136780 }, { "epoch": 0.0713984, "grad_norm": 0.7892919182777405, "learning_rate": 1.4558173957502757e-05, "loss": 3.0664, "step": 136790 }, { "epoch": 0.071424, "grad_norm": 0.7594782114028931, "learning_rate": 1.4557457185843797e-05, "loss": 2.8997, "step": 136800 }, { "epoch": 0.0714496, "grad_norm": 0.7365861535072327, "learning_rate": 1.4556740384630938e-05, "loss": 3.1277, "step": 136810 }, { "epoch": 0.0714752, "grad_norm": 0.817409873008728, "learning_rate": 1.4556023553868832e-05, "loss": 3.2417, "step": 136820 }, { "epoch": 0.0715008, "grad_norm": 0.7495335936546326, "learning_rate": 1.4555306693562121e-05, "loss": 3.0567, "step": 136830 }, { "epoch": 0.0715264, "grad_norm": 0.8667694926261902, "learning_rate": 1.455458980371546e-05, "loss": 2.92, "step": 136840 }, { "epoch": 0.071552, "grad_norm": 0.7451313734054565, "learning_rate": 1.4553872884333493e-05, "loss": 3.2893, "step": 136850 }, { "epoch": 0.0715776, "grad_norm": 0.8598904013633728, "learning_rate": 1.4553155935420871e-05, "loss": 3.0601, "step": 136860 }, { "epoch": 0.0716032, "grad_norm": 0.781446099281311, "learning_rate": 1.4552438956982242e-05, "loss": 2.649, "step": 136870 }, { "epoch": 0.0716288, "grad_norm": 0.7364176511764526, "learning_rate": 1.4551721949022256e-05, "loss": 3.0195, "step": 136880 }, { "epoch": 0.0716544, "grad_norm": 0.8615832924842834, "learning_rate": 1.4551004911545561e-05, "loss": 3.2096, "step": 136890 }, { "epoch": 0.07168, "grad_norm": 0.7772188186645508, "learning_rate": 1.4550287844556814e-05, "loss": 3.2552, "step": 136900 }, { "epoch": 0.0717056, "grad_norm": 0.6578202247619629, "learning_rate": 1.4549570748060656e-05, "loss": 3.1143, "step": 136910 }, { "epoch": 0.0717312, "grad_norm": 0.7451308965682983, "learning_rate": 1.454885362206174e-05, "loss": 2.9636, "step": 136920 }, { "epoch": 0.0717568, "grad_norm": 0.6663442850112915, "learning_rate": 1.454813646656472e-05, "loss": 3.1529, "step": 136930 }, { "epoch": 0.0717824, "grad_norm": 0.9186345934867859, "learning_rate": 1.4547419281574242e-05, "loss": 2.8899, "step": 136940 }, { "epoch": 0.071808, "grad_norm": 0.8541898131370544, "learning_rate": 1.454670206709496e-05, "loss": 2.8985, "step": 136950 }, { "epoch": 0.0718336, "grad_norm": 2.0808534622192383, "learning_rate": 1.4545984823131521e-05, "loss": 2.8555, "step": 136960 }, { "epoch": 0.0718592, "grad_norm": 0.7059140205383301, "learning_rate": 1.454526754968858e-05, "loss": 3.0242, "step": 136970 }, { "epoch": 0.0718848, "grad_norm": 0.7462812066078186, "learning_rate": 1.4544550246770787e-05, "loss": 3.2115, "step": 136980 }, { "epoch": 0.0719104, "grad_norm": 1.0339919328689575, "learning_rate": 1.4543832914382794e-05, "loss": 3.0073, "step": 136990 }, { "epoch": 0.071936, "grad_norm": 0.7728705406188965, "learning_rate": 1.4543115552529252e-05, "loss": 3.1229, "step": 137000 }, { "epoch": 0.0719616, "grad_norm": 1.0479724407196045, "learning_rate": 1.4542398161214813e-05, "loss": 3.2346, "step": 137010 }, { "epoch": 0.0719872, "grad_norm": 0.8457967042922974, "learning_rate": 1.4541680740444129e-05, "loss": 2.9295, "step": 137020 }, { "epoch": 0.0720128, "grad_norm": 0.8092893362045288, "learning_rate": 1.454096329022185e-05, "loss": 3.0829, "step": 137030 }, { "epoch": 0.0720384, "grad_norm": 0.7511181235313416, "learning_rate": 1.4540245810552636e-05, "loss": 3.0447, "step": 137040 }, { "epoch": 0.072064, "grad_norm": 0.6900845170021057, "learning_rate": 1.453952830144113e-05, "loss": 2.9987, "step": 137050 }, { "epoch": 0.0720896, "grad_norm": 0.8024572730064392, "learning_rate": 1.453881076289199e-05, "loss": 3.342, "step": 137060 }, { "epoch": 0.0721152, "grad_norm": 0.6882420182228088, "learning_rate": 1.453809319490987e-05, "loss": 2.9761, "step": 137070 }, { "epoch": 0.0721408, "grad_norm": 0.7870179414749146, "learning_rate": 1.4537375597499422e-05, "loss": 3.0562, "step": 137080 }, { "epoch": 0.0721664, "grad_norm": 0.7349975109100342, "learning_rate": 1.4536657970665299e-05, "loss": 3.0526, "step": 137090 }, { "epoch": 0.072192, "grad_norm": 0.7900339961051941, "learning_rate": 1.4535940314412154e-05, "loss": 3.0528, "step": 137100 }, { "epoch": 0.0722176, "grad_norm": 0.802727222442627, "learning_rate": 1.453522262874464e-05, "loss": 3.1156, "step": 137110 }, { "epoch": 0.0722432, "grad_norm": 0.7171453237533569, "learning_rate": 1.4534504913667415e-05, "loss": 2.9544, "step": 137120 }, { "epoch": 0.0722688, "grad_norm": 0.8662691116333008, "learning_rate": 1.453378716918513e-05, "loss": 3.0821, "step": 137130 }, { "epoch": 0.0722944, "grad_norm": 0.8415297269821167, "learning_rate": 1.4533069395302441e-05, "loss": 2.9267, "step": 137140 }, { "epoch": 0.07232, "grad_norm": 0.8517220616340637, "learning_rate": 1.4532351592024e-05, "loss": 3.0348, "step": 137150 }, { "epoch": 0.0723456, "grad_norm": 0.7533292174339294, "learning_rate": 1.4531633759354465e-05, "loss": 3.034, "step": 137160 }, { "epoch": 0.0723712, "grad_norm": 0.7762165069580078, "learning_rate": 1.453091589729849e-05, "loss": 2.9914, "step": 137170 }, { "epoch": 0.0723968, "grad_norm": 0.6718074679374695, "learning_rate": 1.4530198005860727e-05, "loss": 2.7541, "step": 137180 }, { "epoch": 0.0724224, "grad_norm": 0.8428837656974792, "learning_rate": 1.4529480085045836e-05, "loss": 3.1817, "step": 137190 }, { "epoch": 0.072448, "grad_norm": 0.8194850087165833, "learning_rate": 1.4528762134858472e-05, "loss": 2.981, "step": 137200 }, { "epoch": 0.0724736, "grad_norm": 0.6710900664329529, "learning_rate": 1.4528044155303285e-05, "loss": 3.0578, "step": 137210 }, { "epoch": 0.0724992, "grad_norm": 0.7167710661888123, "learning_rate": 1.4527326146384937e-05, "loss": 3.0044, "step": 137220 }, { "epoch": 0.0725248, "grad_norm": 0.7445276975631714, "learning_rate": 1.4526608108108082e-05, "loss": 2.9829, "step": 137230 }, { "epoch": 0.0725504, "grad_norm": 0.6980224847793579, "learning_rate": 1.4525890040477375e-05, "loss": 2.8354, "step": 137240 }, { "epoch": 0.072576, "grad_norm": 0.7280999422073364, "learning_rate": 1.4525171943497476e-05, "loss": 3.1165, "step": 137250 }, { "epoch": 0.0726016, "grad_norm": 0.8958001732826233, "learning_rate": 1.4524453817173038e-05, "loss": 2.9976, "step": 137260 }, { "epoch": 0.0726272, "grad_norm": 1.0242611169815063, "learning_rate": 1.4523735661508722e-05, "loss": 2.7108, "step": 137270 }, { "epoch": 0.0726528, "grad_norm": 0.757958710193634, "learning_rate": 1.452301747650918e-05, "loss": 2.8742, "step": 137280 }, { "epoch": 0.0726784, "grad_norm": 0.8168308138847351, "learning_rate": 1.4522299262179075e-05, "loss": 3.2985, "step": 137290 }, { "epoch": 0.072704, "grad_norm": 0.6954008936882019, "learning_rate": 1.4521581018523058e-05, "loss": 3.0661, "step": 137300 }, { "epoch": 0.0727296, "grad_norm": 0.7110769748687744, "learning_rate": 1.4520862745545792e-05, "loss": 3.0335, "step": 137310 }, { "epoch": 0.0727552, "grad_norm": 0.9128572940826416, "learning_rate": 1.4520144443251928e-05, "loss": 3.3582, "step": 137320 }, { "epoch": 0.0727808, "grad_norm": 0.8875327110290527, "learning_rate": 1.4519426111646135e-05, "loss": 2.9557, "step": 137330 }, { "epoch": 0.0728064, "grad_norm": 0.6966755986213684, "learning_rate": 1.4518707750733061e-05, "loss": 2.5908, "step": 137340 }, { "epoch": 0.072832, "grad_norm": 0.7736689448356628, "learning_rate": 1.4517989360517369e-05, "loss": 3.0794, "step": 137350 }, { "epoch": 0.0728576, "grad_norm": 0.6608801484107971, "learning_rate": 1.4517270941003716e-05, "loss": 3.1401, "step": 137360 }, { "epoch": 0.0728832, "grad_norm": 0.721177875995636, "learning_rate": 1.4516552492196762e-05, "loss": 3.1418, "step": 137370 }, { "epoch": 0.0729088, "grad_norm": 0.8290756940841675, "learning_rate": 1.4515834014101166e-05, "loss": 3.1211, "step": 137380 }, { "epoch": 0.0729344, "grad_norm": 0.8866058588027954, "learning_rate": 1.4515115506721586e-05, "loss": 3.0393, "step": 137390 }, { "epoch": 0.07296, "grad_norm": 0.8126180171966553, "learning_rate": 1.4514396970062686e-05, "loss": 3.1846, "step": 137400 }, { "epoch": 0.0729856, "grad_norm": 0.7362178564071655, "learning_rate": 1.4513678404129112e-05, "loss": 2.9835, "step": 137410 }, { "epoch": 0.0730112, "grad_norm": 0.7596219778060913, "learning_rate": 1.4512959808925539e-05, "loss": 3.0708, "step": 137420 }, { "epoch": 0.0730368, "grad_norm": 0.6902622580528259, "learning_rate": 1.4512241184456624e-05, "loss": 3.1424, "step": 137430 }, { "epoch": 0.0730624, "grad_norm": 0.7339823842048645, "learning_rate": 1.451152253072702e-05, "loss": 3.0188, "step": 137440 }, { "epoch": 0.073088, "grad_norm": 0.8842318058013916, "learning_rate": 1.451080384774139e-05, "loss": 3.018, "step": 137450 }, { "epoch": 0.0731136, "grad_norm": 0.8704869747161865, "learning_rate": 1.45100851355044e-05, "loss": 2.8662, "step": 137460 }, { "epoch": 0.0731392, "grad_norm": 0.834307074546814, "learning_rate": 1.4509366394020705e-05, "loss": 3.1324, "step": 137470 }, { "epoch": 0.0731648, "grad_norm": 0.7092186212539673, "learning_rate": 1.4508647623294966e-05, "loss": 3.0788, "step": 137480 }, { "epoch": 0.0731904, "grad_norm": 0.6985361576080322, "learning_rate": 1.4507928823331848e-05, "loss": 3.0642, "step": 137490 }, { "epoch": 0.073216, "grad_norm": 0.6850705742835999, "learning_rate": 1.4507209994136007e-05, "loss": 3.2391, "step": 137500 }, { "epoch": 0.0732416, "grad_norm": 1.023937463760376, "learning_rate": 1.4506491135712106e-05, "loss": 3.1025, "step": 137510 }, { "epoch": 0.0732672, "grad_norm": 0.6785716414451599, "learning_rate": 1.4505772248064811e-05, "loss": 3.0475, "step": 137520 }, { "epoch": 0.0732928, "grad_norm": 0.6675170063972473, "learning_rate": 1.4505053331198782e-05, "loss": 2.9927, "step": 137530 }, { "epoch": 0.0733184, "grad_norm": 0.9702374339103699, "learning_rate": 1.4504334385118675e-05, "loss": 3.0666, "step": 137540 }, { "epoch": 0.073344, "grad_norm": 0.8205657005310059, "learning_rate": 1.4503615409829159e-05, "loss": 3.1291, "step": 137550 }, { "epoch": 0.0733696, "grad_norm": 0.8379870057106018, "learning_rate": 1.4502896405334891e-05, "loss": 3.3067, "step": 137560 }, { "epoch": 0.0733952, "grad_norm": 0.8355761170387268, "learning_rate": 1.4502177371640542e-05, "loss": 3.0345, "step": 137570 }, { "epoch": 0.0734208, "grad_norm": 0.6341465711593628, "learning_rate": 1.4501458308750765e-05, "loss": 2.991, "step": 137580 }, { "epoch": 0.0734464, "grad_norm": 0.8349927663803101, "learning_rate": 1.4500739216670228e-05, "loss": 3.0334, "step": 137590 }, { "epoch": 0.073472, "grad_norm": 0.8230007886886597, "learning_rate": 1.4500020095403592e-05, "loss": 3.1296, "step": 137600 }, { "epoch": 0.0734976, "grad_norm": 0.727642834186554, "learning_rate": 1.4499300944955526e-05, "loss": 3.1161, "step": 137610 }, { "epoch": 0.0735232, "grad_norm": 0.7251062989234924, "learning_rate": 1.4498581765330685e-05, "loss": 3.0753, "step": 137620 }, { "epoch": 0.0735488, "grad_norm": 0.8725531101226807, "learning_rate": 1.449786255653374e-05, "loss": 3.0, "step": 137630 }, { "epoch": 0.0735744, "grad_norm": 0.7186927199363708, "learning_rate": 1.4497143318569349e-05, "loss": 3.0037, "step": 137640 }, { "epoch": 0.0736, "grad_norm": 0.7644299864768982, "learning_rate": 1.4496424051442179e-05, "loss": 3.021, "step": 137650 }, { "epoch": 0.0736256, "grad_norm": 1.0410631895065308, "learning_rate": 1.4495704755156896e-05, "loss": 2.9406, "step": 137660 }, { "epoch": 0.0736512, "grad_norm": 0.7669177651405334, "learning_rate": 1.449498542971816e-05, "loss": 3.0864, "step": 137670 }, { "epoch": 0.0736768, "grad_norm": 0.7696532607078552, "learning_rate": 1.4494266075130638e-05, "loss": 3.0152, "step": 137680 }, { "epoch": 0.0737024, "grad_norm": 1.5777392387390137, "learning_rate": 1.4493546691398999e-05, "loss": 3.0313, "step": 137690 }, { "epoch": 0.073728, "grad_norm": 0.8912967443466187, "learning_rate": 1.4492827278527901e-05, "loss": 2.9627, "step": 137700 }, { "epoch": 0.0737536, "grad_norm": 1.3888940811157227, "learning_rate": 1.4492107836522014e-05, "loss": 3.3879, "step": 137710 }, { "epoch": 0.0737792, "grad_norm": 0.7611891031265259, "learning_rate": 1.4491388365385999e-05, "loss": 3.0498, "step": 137720 }, { "epoch": 0.0738048, "grad_norm": 1.0498892068862915, "learning_rate": 1.4490668865124527e-05, "loss": 3.1703, "step": 137730 }, { "epoch": 0.0738304, "grad_norm": 0.7138067483901978, "learning_rate": 1.4489949335742257e-05, "loss": 3.0445, "step": 137740 }, { "epoch": 0.073856, "grad_norm": 0.6330154538154602, "learning_rate": 1.4489229777243859e-05, "loss": 2.9617, "step": 137750 }, { "epoch": 0.0738816, "grad_norm": 1.1296104192733765, "learning_rate": 1.4488510189634001e-05, "loss": 2.9097, "step": 137760 }, { "epoch": 0.0739072, "grad_norm": 0.8188772797584534, "learning_rate": 1.4487790572917349e-05, "loss": 2.8599, "step": 137770 }, { "epoch": 0.0739328, "grad_norm": 0.7354639172554016, "learning_rate": 1.4487070927098566e-05, "loss": 2.9106, "step": 137780 }, { "epoch": 0.0739584, "grad_norm": 0.7524297833442688, "learning_rate": 1.448635125218232e-05, "loss": 3.202, "step": 137790 }, { "epoch": 0.073984, "grad_norm": 0.6772817373275757, "learning_rate": 1.448563154817328e-05, "loss": 2.9664, "step": 137800 }, { "epoch": 0.0740096, "grad_norm": 0.7843740582466125, "learning_rate": 1.448491181507611e-05, "loss": 2.9917, "step": 137810 }, { "epoch": 0.0740352, "grad_norm": 0.7613982558250427, "learning_rate": 1.4484192052895479e-05, "loss": 3.1655, "step": 137820 }, { "epoch": 0.0740608, "grad_norm": 0.7354992628097534, "learning_rate": 1.4483472261636056e-05, "loss": 2.9981, "step": 137830 }, { "epoch": 0.0740864, "grad_norm": 0.9349342584609985, "learning_rate": 1.4482752441302504e-05, "loss": 3.0092, "step": 137840 }, { "epoch": 0.074112, "grad_norm": 0.9101986885070801, "learning_rate": 1.4482032591899495e-05, "loss": 3.209, "step": 137850 }, { "epoch": 0.0741376, "grad_norm": 0.754279375076294, "learning_rate": 1.4481312713431697e-05, "loss": 3.0646, "step": 137860 }, { "epoch": 0.0741632, "grad_norm": 0.7642857432365417, "learning_rate": 1.4480592805903775e-05, "loss": 2.9952, "step": 137870 }, { "epoch": 0.0741888, "grad_norm": 0.8015186786651611, "learning_rate": 1.4479872869320402e-05, "loss": 3.0298, "step": 137880 }, { "epoch": 0.0742144, "grad_norm": 0.8291166424751282, "learning_rate": 1.4479152903686245e-05, "loss": 3.0057, "step": 137890 }, { "epoch": 0.07424, "grad_norm": 0.7834596633911133, "learning_rate": 1.4478432909005968e-05, "loss": 3.0217, "step": 137900 }, { "epoch": 0.0742656, "grad_norm": 1.1116372346878052, "learning_rate": 1.4477712885284245e-05, "loss": 3.1173, "step": 137910 }, { "epoch": 0.0742912, "grad_norm": 0.7066279649734497, "learning_rate": 1.4476992832525745e-05, "loss": 3.1156, "step": 137920 }, { "epoch": 0.0743168, "grad_norm": 0.7047009468078613, "learning_rate": 1.4476272750735134e-05, "loss": 3.1255, "step": 137930 }, { "epoch": 0.0743424, "grad_norm": 0.7120074033737183, "learning_rate": 1.4475552639917086e-05, "loss": 3.0145, "step": 137940 }, { "epoch": 0.074368, "grad_norm": 0.8270100951194763, "learning_rate": 1.4474832500076268e-05, "loss": 3.0449, "step": 137950 }, { "epoch": 0.0743936, "grad_norm": 0.8100025653839111, "learning_rate": 1.4474112331217352e-05, "loss": 3.1572, "step": 137960 }, { "epoch": 0.0744192, "grad_norm": 0.8197717070579529, "learning_rate": 1.4473392133345003e-05, "loss": 2.9366, "step": 137970 }, { "epoch": 0.0744448, "grad_norm": 0.9186192750930786, "learning_rate": 1.4472671906463898e-05, "loss": 3.5794, "step": 137980 }, { "epoch": 0.0744704, "grad_norm": 0.7823315262794495, "learning_rate": 1.4471951650578706e-05, "loss": 3.1149, "step": 137990 }, { "epoch": 0.074496, "grad_norm": 0.749606728553772, "learning_rate": 1.4471231365694092e-05, "loss": 3.2057, "step": 138000 }, { "epoch": 0.0745216, "grad_norm": 0.7962782382965088, "learning_rate": 1.4470511051814733e-05, "loss": 3.163, "step": 138010 }, { "epoch": 0.0745472, "grad_norm": 0.808138906955719, "learning_rate": 1.4469790708945297e-05, "loss": 3.0277, "step": 138020 }, { "epoch": 0.0745728, "grad_norm": 0.6851063370704651, "learning_rate": 1.4469070337090456e-05, "loss": 2.98, "step": 138030 }, { "epoch": 0.0745984, "grad_norm": 0.802139163017273, "learning_rate": 1.4468349936254882e-05, "loss": 3.1858, "step": 138040 }, { "epoch": 0.074624, "grad_norm": 1.108983039855957, "learning_rate": 1.4467629506443247e-05, "loss": 3.1452, "step": 138050 }, { "epoch": 0.0746496, "grad_norm": 1.5647454261779785, "learning_rate": 1.4466909047660223e-05, "loss": 3.105, "step": 138060 }, { "epoch": 0.0746752, "grad_norm": 0.7922728657722473, "learning_rate": 1.4466188559910478e-05, "loss": 3.1125, "step": 138070 }, { "epoch": 0.0747008, "grad_norm": 0.8113752007484436, "learning_rate": 1.4465468043198689e-05, "loss": 3.1856, "step": 138080 }, { "epoch": 0.0747264, "grad_norm": 0.6648405194282532, "learning_rate": 1.4464747497529527e-05, "loss": 2.8669, "step": 138090 }, { "epoch": 0.074752, "grad_norm": 0.7107937335968018, "learning_rate": 1.4464026922907662e-05, "loss": 2.9754, "step": 138100 }, { "epoch": 0.0747776, "grad_norm": 0.7964227795600891, "learning_rate": 1.446330631933777e-05, "loss": 3.0926, "step": 138110 }, { "epoch": 0.0748032, "grad_norm": 0.6566756963729858, "learning_rate": 1.4462585686824524e-05, "loss": 3.1243, "step": 138120 }, { "epoch": 0.0748288, "grad_norm": 0.962415874004364, "learning_rate": 1.4461865025372595e-05, "loss": 3.1034, "step": 138130 }, { "epoch": 0.0748544, "grad_norm": 0.7541049122810364, "learning_rate": 1.4461144334986655e-05, "loss": 3.0271, "step": 138140 }, { "epoch": 0.07488, "grad_norm": 0.7503150701522827, "learning_rate": 1.4460423615671379e-05, "loss": 3.1858, "step": 138150 }, { "epoch": 0.0749056, "grad_norm": 0.957188606262207, "learning_rate": 1.4459702867431447e-05, "loss": 2.9979, "step": 138160 }, { "epoch": 0.0749312, "grad_norm": 0.7386629581451416, "learning_rate": 1.4458982090271522e-05, "loss": 2.9285, "step": 138170 }, { "epoch": 0.0749568, "grad_norm": 1.1278350353240967, "learning_rate": 1.4458261284196286e-05, "loss": 3.2712, "step": 138180 }, { "epoch": 0.0749824, "grad_norm": 0.7686112523078918, "learning_rate": 1.445754044921041e-05, "loss": 3.1653, "step": 138190 }, { "epoch": 0.075008, "grad_norm": 0.643552303314209, "learning_rate": 1.4456819585318566e-05, "loss": 3.0379, "step": 138200 }, { "epoch": 0.0750336, "grad_norm": 0.7699463963508606, "learning_rate": 1.4456098692525435e-05, "loss": 3.0175, "step": 138210 }, { "epoch": 0.0750592, "grad_norm": 0.750405490398407, "learning_rate": 1.4455377770835687e-05, "loss": 3.0087, "step": 138220 }, { "epoch": 0.0750848, "grad_norm": 0.7788735628128052, "learning_rate": 1.4454656820253998e-05, "loss": 3.0547, "step": 138230 }, { "epoch": 0.0751104, "grad_norm": 0.7736037969589233, "learning_rate": 1.4453935840785046e-05, "loss": 2.9785, "step": 138240 }, { "epoch": 0.075136, "grad_norm": 0.6957564353942871, "learning_rate": 1.44532148324335e-05, "loss": 3.1133, "step": 138250 }, { "epoch": 0.0751616, "grad_norm": 0.8241851925849915, "learning_rate": 1.4452493795204042e-05, "loss": 2.957, "step": 138260 }, { "epoch": 0.0751872, "grad_norm": 0.6821261048316956, "learning_rate": 1.4451772729101344e-05, "loss": 3.1853, "step": 138270 }, { "epoch": 0.0752128, "grad_norm": 0.7421987056732178, "learning_rate": 1.4451051634130082e-05, "loss": 2.9835, "step": 138280 }, { "epoch": 0.0752384, "grad_norm": 0.7376146912574768, "learning_rate": 1.4450330510294931e-05, "loss": 3.162, "step": 138290 }, { "epoch": 0.075264, "grad_norm": 0.6899235844612122, "learning_rate": 1.4449609357600574e-05, "loss": 3.1094, "step": 138300 }, { "epoch": 0.0752896, "grad_norm": 0.6963797211647034, "learning_rate": 1.444888817605168e-05, "loss": 3.0546, "step": 138310 }, { "epoch": 0.0753152, "grad_norm": 0.6680565476417542, "learning_rate": 1.4448166965652932e-05, "loss": 3.0965, "step": 138320 }, { "epoch": 0.0753408, "grad_norm": 0.9038199186325073, "learning_rate": 1.4447445726409003e-05, "loss": 3.0761, "step": 138330 }, { "epoch": 0.0753664, "grad_norm": 0.698241651058197, "learning_rate": 1.4446724458324567e-05, "loss": 3.033, "step": 138340 }, { "epoch": 0.075392, "grad_norm": 0.8831669688224792, "learning_rate": 1.4446003161404305e-05, "loss": 2.9361, "step": 138350 }, { "epoch": 0.0754176, "grad_norm": 0.7599766254425049, "learning_rate": 1.4445281835652895e-05, "loss": 3.0849, "step": 138360 }, { "epoch": 0.0754432, "grad_norm": 0.7153052687644958, "learning_rate": 1.4444560481075015e-05, "loss": 3.1207, "step": 138370 }, { "epoch": 0.0754688, "grad_norm": 0.6982991099357605, "learning_rate": 1.444383909767534e-05, "loss": 3.0182, "step": 138380 }, { "epoch": 0.0754944, "grad_norm": 0.7309590578079224, "learning_rate": 1.4443117685458552e-05, "loss": 3.0368, "step": 138390 }, { "epoch": 0.07552, "grad_norm": 0.7320541739463806, "learning_rate": 1.4442396244429324e-05, "loss": 3.1687, "step": 138400 }, { "epoch": 0.0755456, "grad_norm": 0.8057261109352112, "learning_rate": 1.4441674774592338e-05, "loss": 3.0448, "step": 138410 }, { "epoch": 0.0755712, "grad_norm": 0.7231920957565308, "learning_rate": 1.4440953275952271e-05, "loss": 3.0358, "step": 138420 }, { "epoch": 0.0755968, "grad_norm": 0.6923068165779114, "learning_rate": 1.4440231748513802e-05, "loss": 3.0625, "step": 138430 }, { "epoch": 0.0756224, "grad_norm": 0.7173610329627991, "learning_rate": 1.4439510192281612e-05, "loss": 2.9705, "step": 138440 }, { "epoch": 0.075648, "grad_norm": 0.7549614310264587, "learning_rate": 1.4438788607260377e-05, "loss": 2.943, "step": 138450 }, { "epoch": 0.0756736, "grad_norm": 0.8070223927497864, "learning_rate": 1.4438066993454776e-05, "loss": 3.0531, "step": 138460 }, { "epoch": 0.0756992, "grad_norm": 0.7777757048606873, "learning_rate": 1.4437345350869491e-05, "loss": 2.9316, "step": 138470 }, { "epoch": 0.0757248, "grad_norm": 1.2367521524429321, "learning_rate": 1.4436623679509203e-05, "loss": 3.2627, "step": 138480 }, { "epoch": 0.0757504, "grad_norm": 0.7301132082939148, "learning_rate": 1.4435901979378589e-05, "loss": 3.0051, "step": 138490 }, { "epoch": 0.075776, "grad_norm": 0.9524611830711365, "learning_rate": 1.4435180250482327e-05, "loss": 3.138, "step": 138500 }, { "epoch": 0.0758016, "grad_norm": 0.7026020884513855, "learning_rate": 1.4434458492825101e-05, "loss": 2.9287, "step": 138510 }, { "epoch": 0.0758272, "grad_norm": 0.8655324578285217, "learning_rate": 1.4433736706411592e-05, "loss": 3.213, "step": 138520 }, { "epoch": 0.0758528, "grad_norm": 0.7943993210792542, "learning_rate": 1.4433014891246477e-05, "loss": 3.0061, "step": 138530 }, { "epoch": 0.0758784, "grad_norm": 0.6420594453811646, "learning_rate": 1.4432293047334435e-05, "loss": 2.9723, "step": 138540 }, { "epoch": 0.075904, "grad_norm": 0.7787966132164001, "learning_rate": 1.4431571174680159e-05, "loss": 3.1315, "step": 138550 }, { "epoch": 0.0759296, "grad_norm": 0.8190917372703552, "learning_rate": 1.4430849273288316e-05, "loss": 2.9036, "step": 138560 }, { "epoch": 0.0759552, "grad_norm": 0.7556844353675842, "learning_rate": 1.4430127343163594e-05, "loss": 3.1459, "step": 138570 }, { "epoch": 0.0759808, "grad_norm": 0.6768133640289307, "learning_rate": 1.4429405384310676e-05, "loss": 3.0728, "step": 138580 }, { "epoch": 0.0760064, "grad_norm": 0.6316946744918823, "learning_rate": 1.4428683396734237e-05, "loss": 3.1457, "step": 138590 }, { "epoch": 0.076032, "grad_norm": 0.7041962742805481, "learning_rate": 1.4427961380438966e-05, "loss": 3.0973, "step": 138600 }, { "epoch": 0.0760576, "grad_norm": 0.7727822661399841, "learning_rate": 1.4427239335429541e-05, "loss": 2.8662, "step": 138610 }, { "epoch": 0.0760832, "grad_norm": 1.588766098022461, "learning_rate": 1.4426517261710646e-05, "loss": 2.967, "step": 138620 }, { "epoch": 0.0761088, "grad_norm": 0.7743009328842163, "learning_rate": 1.4425795159286961e-05, "loss": 2.9041, "step": 138630 }, { "epoch": 0.0761344, "grad_norm": 0.7215845584869385, "learning_rate": 1.4425073028163174e-05, "loss": 2.9589, "step": 138640 }, { "epoch": 0.07616, "grad_norm": 0.9125052690505981, "learning_rate": 1.4424350868343964e-05, "loss": 3.1144, "step": 138650 }, { "epoch": 0.0761856, "grad_norm": 0.8456500768661499, "learning_rate": 1.4423628679834014e-05, "loss": 3.0317, "step": 138660 }, { "epoch": 0.0762112, "grad_norm": 0.9866217374801636, "learning_rate": 1.4422906462638007e-05, "loss": 3.151, "step": 138670 }, { "epoch": 0.0762368, "grad_norm": 0.7849451899528503, "learning_rate": 1.4422184216760629e-05, "loss": 3.2206, "step": 138680 }, { "epoch": 0.0762624, "grad_norm": 0.7690137028694153, "learning_rate": 1.4421461942206558e-05, "loss": 2.9972, "step": 138690 }, { "epoch": 0.076288, "grad_norm": 0.7193928360939026, "learning_rate": 1.4420739638980484e-05, "loss": 3.2017, "step": 138700 }, { "epoch": 0.0763136, "grad_norm": 1.0529361963272095, "learning_rate": 1.4420017307087086e-05, "loss": 3.0701, "step": 138710 }, { "epoch": 0.0763392, "grad_norm": 0.8642722964286804, "learning_rate": 1.4419294946531053e-05, "loss": 3.0982, "step": 138720 }, { "epoch": 0.0763648, "grad_norm": 0.8115140795707703, "learning_rate": 1.4418572557317065e-05, "loss": 3.1014, "step": 138730 }, { "epoch": 0.0763904, "grad_norm": 0.9112223386764526, "learning_rate": 1.441785013944981e-05, "loss": 3.2358, "step": 138740 }, { "epoch": 0.076416, "grad_norm": 0.6692140698432922, "learning_rate": 1.441712769293397e-05, "loss": 2.9684, "step": 138750 }, { "epoch": 0.0764416, "grad_norm": 0.7568619847297668, "learning_rate": 1.4416405217774233e-05, "loss": 2.9796, "step": 138760 }, { "epoch": 0.0764672, "grad_norm": 0.8219455480575562, "learning_rate": 1.4415682713975278e-05, "loss": 3.1255, "step": 138770 }, { "epoch": 0.0764928, "grad_norm": 0.7371528744697571, "learning_rate": 1.4414960181541797e-05, "loss": 3.1875, "step": 138780 }, { "epoch": 0.0765184, "grad_norm": 1.148529052734375, "learning_rate": 1.441423762047847e-05, "loss": 3.2164, "step": 138790 }, { "epoch": 0.076544, "grad_norm": 0.7334292531013489, "learning_rate": 1.4413515030789986e-05, "loss": 3.033, "step": 138800 }, { "epoch": 0.0765696, "grad_norm": 0.7572510242462158, "learning_rate": 1.4412792412481032e-05, "loss": 2.9825, "step": 138810 }, { "epoch": 0.0765952, "grad_norm": 0.7164313197135925, "learning_rate": 1.4412069765556291e-05, "loss": 3.0339, "step": 138820 }, { "epoch": 0.0766208, "grad_norm": 0.7262660264968872, "learning_rate": 1.4411347090020448e-05, "loss": 3.1343, "step": 138830 }, { "epoch": 0.0766464, "grad_norm": 0.7423118352890015, "learning_rate": 1.4410624385878197e-05, "loss": 3.13, "step": 138840 }, { "epoch": 0.076672, "grad_norm": 0.7050949335098267, "learning_rate": 1.4409901653134216e-05, "loss": 2.9898, "step": 138850 }, { "epoch": 0.0766976, "grad_norm": 0.6607176661491394, "learning_rate": 1.4409178891793191e-05, "loss": 3.0072, "step": 138860 }, { "epoch": 0.0767232, "grad_norm": 0.7169002294540405, "learning_rate": 1.4408456101859817e-05, "loss": 2.8458, "step": 138870 }, { "epoch": 0.0767488, "grad_norm": 0.678074836730957, "learning_rate": 1.4407733283338778e-05, "loss": 2.9355, "step": 138880 }, { "epoch": 0.0767744, "grad_norm": 0.8085635304450989, "learning_rate": 1.4407010436234755e-05, "loss": 3.0123, "step": 138890 }, { "epoch": 0.0768, "grad_norm": 0.9203239679336548, "learning_rate": 1.4406287560552443e-05, "loss": 2.9857, "step": 138900 }, { "epoch": 0.0768256, "grad_norm": 0.7725054621696472, "learning_rate": 1.440556465629653e-05, "loss": 3.1159, "step": 138910 }, { "epoch": 0.0768512, "grad_norm": 0.7905682325363159, "learning_rate": 1.4404841723471698e-05, "loss": 2.9325, "step": 138920 }, { "epoch": 0.0768768, "grad_norm": 0.7518634796142578, "learning_rate": 1.440411876208264e-05, "loss": 3.0865, "step": 138930 }, { "epoch": 0.0769024, "grad_norm": 2.183689594268799, "learning_rate": 1.4403395772134038e-05, "loss": 3.0962, "step": 138940 }, { "epoch": 0.076928, "grad_norm": 0.8626227378845215, "learning_rate": 1.440267275363059e-05, "loss": 2.957, "step": 138950 }, { "epoch": 0.0769536, "grad_norm": 0.7624696493148804, "learning_rate": 1.440194970657698e-05, "loss": 3.0806, "step": 138960 }, { "epoch": 0.0769792, "grad_norm": 0.753250777721405, "learning_rate": 1.440122663097789e-05, "loss": 3.0243, "step": 138970 }, { "epoch": 0.0770048, "grad_norm": 0.9578465819358826, "learning_rate": 1.4400503526838019e-05, "loss": 3.2937, "step": 138980 }, { "epoch": 0.0770304, "grad_norm": 0.7539263963699341, "learning_rate": 1.4399780394162051e-05, "loss": 2.8023, "step": 138990 }, { "epoch": 0.077056, "grad_norm": 0.6889841556549072, "learning_rate": 1.4399057232954677e-05, "loss": 2.7671, "step": 139000 }, { "epoch": 0.0770816, "grad_norm": 0.7464808821678162, "learning_rate": 1.4398334043220585e-05, "loss": 3.0352, "step": 139010 }, { "epoch": 0.0771072, "grad_norm": 0.695182204246521, "learning_rate": 1.4397610824964468e-05, "loss": 2.7451, "step": 139020 }, { "epoch": 0.0771328, "grad_norm": 0.6897189617156982, "learning_rate": 1.439688757819101e-05, "loss": 3.0059, "step": 139030 }, { "epoch": 0.0771584, "grad_norm": 0.7222037315368652, "learning_rate": 1.4396164302904907e-05, "loss": 2.8934, "step": 139040 }, { "epoch": 0.077184, "grad_norm": 0.6923055052757263, "learning_rate": 1.4395440999110847e-05, "loss": 2.5494, "step": 139050 }, { "epoch": 0.0772096, "grad_norm": 0.6629042029380798, "learning_rate": 1.4394717666813517e-05, "loss": 2.7928, "step": 139060 }, { "epoch": 0.0772352, "grad_norm": 0.7673067450523376, "learning_rate": 1.4393994306017618e-05, "loss": 2.7927, "step": 139070 }, { "epoch": 0.0772608, "grad_norm": 0.7310132384300232, "learning_rate": 1.4393270916727829e-05, "loss": 2.7803, "step": 139080 }, { "epoch": 0.0772864, "grad_norm": 0.7768663763999939, "learning_rate": 1.4392547498948845e-05, "loss": 2.9289, "step": 139090 }, { "epoch": 0.077312, "grad_norm": 0.7405317425727844, "learning_rate": 1.4391824052685357e-05, "loss": 3.2006, "step": 139100 }, { "epoch": 0.0773376, "grad_norm": 0.7025265097618103, "learning_rate": 1.439110057794206e-05, "loss": 2.9421, "step": 139110 }, { "epoch": 0.0773632, "grad_norm": 0.7766634821891785, "learning_rate": 1.4390377074723643e-05, "loss": 2.944, "step": 139120 }, { "epoch": 0.0773888, "grad_norm": 0.7145302295684814, "learning_rate": 1.4389653543034796e-05, "loss": 2.812, "step": 139130 }, { "epoch": 0.0774144, "grad_norm": 0.6851516962051392, "learning_rate": 1.4388929982880213e-05, "loss": 2.9106, "step": 139140 }, { "epoch": 0.07744, "grad_norm": 0.7834727168083191, "learning_rate": 1.4388206394264584e-05, "loss": 2.7639, "step": 139150 }, { "epoch": 0.0774656, "grad_norm": 0.6969383358955383, "learning_rate": 1.4387482777192606e-05, "loss": 2.8232, "step": 139160 }, { "epoch": 0.0774912, "grad_norm": 0.716741681098938, "learning_rate": 1.4386759131668966e-05, "loss": 2.83, "step": 139170 }, { "epoch": 0.0775168, "grad_norm": 0.7477161884307861, "learning_rate": 1.438603545769836e-05, "loss": 2.7659, "step": 139180 }, { "epoch": 0.0775424, "grad_norm": 0.6585568189620972, "learning_rate": 1.4385311755285479e-05, "loss": 2.5816, "step": 139190 }, { "epoch": 0.077568, "grad_norm": 0.6521705985069275, "learning_rate": 1.4384588024435015e-05, "loss": 2.7274, "step": 139200 }, { "epoch": 0.0775936, "grad_norm": 0.7176759839057922, "learning_rate": 1.4383864265151666e-05, "loss": 2.782, "step": 139210 }, { "epoch": 0.0776192, "grad_norm": 0.8143592476844788, "learning_rate": 1.4383140477440121e-05, "loss": 2.9617, "step": 139220 }, { "epoch": 0.0776448, "grad_norm": 1.1823961734771729, "learning_rate": 1.4382416661305072e-05, "loss": 2.9168, "step": 139230 }, { "epoch": 0.0776704, "grad_norm": 0.7198460698127747, "learning_rate": 1.4381692816751221e-05, "loss": 2.7393, "step": 139240 }, { "epoch": 0.077696, "grad_norm": 0.7007437944412231, "learning_rate": 1.4380968943783254e-05, "loss": 2.8126, "step": 139250 }, { "epoch": 0.0777216, "grad_norm": 0.6603631973266602, "learning_rate": 1.4380245042405867e-05, "loss": 2.9536, "step": 139260 }, { "epoch": 0.0777472, "grad_norm": 0.6512635946273804, "learning_rate": 1.4379521112623758e-05, "loss": 3.0991, "step": 139270 }, { "epoch": 0.0777728, "grad_norm": 0.6459760665893555, "learning_rate": 1.4378797154441615e-05, "loss": 2.905, "step": 139280 }, { "epoch": 0.0777984, "grad_norm": 0.6624528765678406, "learning_rate": 1.4378073167864137e-05, "loss": 2.8115, "step": 139290 }, { "epoch": 0.077824, "grad_norm": 0.7321378588676453, "learning_rate": 1.4377349152896018e-05, "loss": 2.9519, "step": 139300 }, { "epoch": 0.0778496, "grad_norm": 0.6939398646354675, "learning_rate": 1.4376625109541955e-05, "loss": 2.7373, "step": 139310 }, { "epoch": 0.0778752, "grad_norm": 0.7653660178184509, "learning_rate": 1.4375901037806638e-05, "loss": 2.7632, "step": 139320 }, { "epoch": 0.0779008, "grad_norm": 0.7561643123626709, "learning_rate": 1.4375176937694768e-05, "loss": 2.706, "step": 139330 }, { "epoch": 0.0779264, "grad_norm": 0.6859797835350037, "learning_rate": 1.4374452809211042e-05, "loss": 3.0245, "step": 139340 }, { "epoch": 0.077952, "grad_norm": 0.6620303392410278, "learning_rate": 1.4373728652360147e-05, "loss": 2.7446, "step": 139350 }, { "epoch": 0.0779776, "grad_norm": 0.6637797355651855, "learning_rate": 1.4373004467146782e-05, "loss": 2.8701, "step": 139360 }, { "epoch": 0.0780032, "grad_norm": 0.6951032876968384, "learning_rate": 1.4372280253575652e-05, "loss": 2.9018, "step": 139370 }, { "epoch": 0.0780288, "grad_norm": 0.8707420825958252, "learning_rate": 1.4371556011651442e-05, "loss": 2.8754, "step": 139380 }, { "epoch": 0.0780544, "grad_norm": 0.7051072120666504, "learning_rate": 1.4370831741378855e-05, "loss": 2.7751, "step": 139390 }, { "epoch": 0.07808, "grad_norm": 0.8312610983848572, "learning_rate": 1.4370107442762584e-05, "loss": 2.8267, "step": 139400 }, { "epoch": 0.0781056, "grad_norm": 0.8103346228599548, "learning_rate": 1.4369383115807329e-05, "loss": 2.7366, "step": 139410 }, { "epoch": 0.0781312, "grad_norm": 0.6554751396179199, "learning_rate": 1.4368658760517783e-05, "loss": 2.917, "step": 139420 }, { "epoch": 0.0781568, "grad_norm": 0.7177314162254333, "learning_rate": 1.4367934376898651e-05, "loss": 2.9749, "step": 139430 }, { "epoch": 0.0781824, "grad_norm": 0.7164790630340576, "learning_rate": 1.4367209964954624e-05, "loss": 2.758, "step": 139440 }, { "epoch": 0.078208, "grad_norm": 0.7499439120292664, "learning_rate": 1.43664855246904e-05, "loss": 3.2061, "step": 139450 }, { "epoch": 0.0782336, "grad_norm": 0.7847282886505127, "learning_rate": 1.4365761056110677e-05, "loss": 2.8708, "step": 139460 }, { "epoch": 0.0782592, "grad_norm": 0.7497707009315491, "learning_rate": 1.4365036559220156e-05, "loss": 2.7421, "step": 139470 }, { "epoch": 0.0782848, "grad_norm": 0.6984232068061829, "learning_rate": 1.4364312034023532e-05, "loss": 3.1156, "step": 139480 }, { "epoch": 0.0783104, "grad_norm": 0.7422033548355103, "learning_rate": 1.4363587480525501e-05, "loss": 2.7733, "step": 139490 }, { "epoch": 0.078336, "grad_norm": 0.7130302786827087, "learning_rate": 1.436286289873077e-05, "loss": 2.8804, "step": 139500 }, { "epoch": 0.0783616, "grad_norm": 0.639137327671051, "learning_rate": 1.4362138288644029e-05, "loss": 3.0624, "step": 139510 }, { "epoch": 0.0783872, "grad_norm": 0.6658255457878113, "learning_rate": 1.4361413650269982e-05, "loss": 2.8652, "step": 139520 }, { "epoch": 0.0784128, "grad_norm": 0.748159646987915, "learning_rate": 1.4360688983613327e-05, "loss": 2.9668, "step": 139530 }, { "epoch": 0.0784384, "grad_norm": 0.6991344094276428, "learning_rate": 1.4359964288678762e-05, "loss": 2.8492, "step": 139540 }, { "epoch": 0.078464, "grad_norm": 0.6878490447998047, "learning_rate": 1.4359239565470988e-05, "loss": 2.7324, "step": 139550 }, { "epoch": 0.0784896, "grad_norm": 0.7392411231994629, "learning_rate": 1.4358514813994705e-05, "loss": 2.6567, "step": 139560 }, { "epoch": 0.0785152, "grad_norm": 0.769707202911377, "learning_rate": 1.4357790034254611e-05, "loss": 2.9771, "step": 139570 }, { "epoch": 0.0785408, "grad_norm": 0.712395966053009, "learning_rate": 1.4357065226255405e-05, "loss": 2.8061, "step": 139580 }, { "epoch": 0.0785664, "grad_norm": 0.9411882162094116, "learning_rate": 1.4356340390001788e-05, "loss": 2.9568, "step": 139590 }, { "epoch": 0.078592, "grad_norm": 0.8816880583763123, "learning_rate": 1.4355615525498468e-05, "loss": 2.9228, "step": 139600 }, { "epoch": 2.56e-05, "grad_norm": 1.5469576120376587, "learning_rate": 1.4354890632750134e-05, "loss": 2.2078, "step": 139610 }, { "epoch": 5.12e-05, "grad_norm": 0.6530632376670837, "learning_rate": 1.4354165711761489e-05, "loss": 2.8608, "step": 139620 }, { "epoch": 7.68e-05, "grad_norm": 0.8113317489624023, "learning_rate": 1.435344076253724e-05, "loss": 2.9601, "step": 139630 }, { "epoch": 0.0001024, "grad_norm": 0.6691372990608215, "learning_rate": 1.4352715785082087e-05, "loss": 2.7865, "step": 139640 }, { "epoch": 0.000128, "grad_norm": 0.6358795762062073, "learning_rate": 1.4351990779400726e-05, "loss": 2.9589, "step": 139650 }, { "epoch": 0.0001536, "grad_norm": 0.6808091998100281, "learning_rate": 1.435126574549786e-05, "loss": 3.053, "step": 139660 }, { "epoch": 0.0001792, "grad_norm": 0.6544643640518188, "learning_rate": 1.4350540683378193e-05, "loss": 2.8961, "step": 139670 }, { "epoch": 0.0002048, "grad_norm": 0.8147395253181458, "learning_rate": 1.4349815593046427e-05, "loss": 2.9438, "step": 139680 }, { "epoch": 0.0002304, "grad_norm": 0.6641374826431274, "learning_rate": 1.4349090474507263e-05, "loss": 2.921, "step": 139690 }, { "epoch": 0.000256, "grad_norm": 0.7807210087776184, "learning_rate": 1.4348365327765405e-05, "loss": 2.8544, "step": 139700 }, { "epoch": 0.0002816, "grad_norm": 0.8245289325714111, "learning_rate": 1.4347640152825551e-05, "loss": 2.8899, "step": 139710 }, { "epoch": 0.0003072, "grad_norm": 0.8759667277336121, "learning_rate": 1.4346914949692405e-05, "loss": 2.8791, "step": 139720 }, { "epoch": 0.0003328, "grad_norm": 0.718227207660675, "learning_rate": 1.4346189718370674e-05, "loss": 2.7443, "step": 139730 }, { "epoch": 0.0003584, "grad_norm": 0.7530933022499084, "learning_rate": 1.4345464458865054e-05, "loss": 3.2133, "step": 139740 }, { "epoch": 0.000384, "grad_norm": 0.840266227722168, "learning_rate": 1.4344739171180255e-05, "loss": 3.0068, "step": 139750 }, { "epoch": 0.0004096, "grad_norm": 1.0586893558502197, "learning_rate": 1.4344013855320972e-05, "loss": 2.91, "step": 139760 }, { "epoch": 0.0004352, "grad_norm": 0.6862390637397766, "learning_rate": 1.4343288511291919e-05, "loss": 2.9657, "step": 139770 }, { "epoch": 0.0004608, "grad_norm": 0.7049717903137207, "learning_rate": 1.4342563139097793e-05, "loss": 3.1924, "step": 139780 }, { "epoch": 0.0004864, "grad_norm": 0.6881759762763977, "learning_rate": 1.43418377387433e-05, "loss": 2.9525, "step": 139790 }, { "epoch": 0.000512, "grad_norm": 0.8303090929985046, "learning_rate": 1.4341112310233146e-05, "loss": 2.9462, "step": 139800 }, { "epoch": 0.0005376, "grad_norm": 0.6895320415496826, "learning_rate": 1.4340386853572026e-05, "loss": 2.723, "step": 139810 }, { "epoch": 0.0005632, "grad_norm": 0.7098149061203003, "learning_rate": 1.4339661368764653e-05, "loss": 2.8896, "step": 139820 }, { "epoch": 0.0005888, "grad_norm": 0.7512398362159729, "learning_rate": 1.433893585581573e-05, "loss": 2.9316, "step": 139830 }, { "epoch": 0.0006144, "grad_norm": 0.7612740993499756, "learning_rate": 1.433821031472996e-05, "loss": 2.6664, "step": 139840 }, { "epoch": 0.00064, "grad_norm": 0.7834928631782532, "learning_rate": 1.4337484745512051e-05, "loss": 2.7276, "step": 139850 }, { "epoch": 0.0006656, "grad_norm": 0.9114246368408203, "learning_rate": 1.4336759148166704e-05, "loss": 3.0341, "step": 139860 }, { "epoch": 0.0006912, "grad_norm": 0.7320348620414734, "learning_rate": 1.433603352269863e-05, "loss": 3.1986, "step": 139870 }, { "epoch": 0.0007168, "grad_norm": 0.7168043851852417, "learning_rate": 1.4335307869112527e-05, "loss": 2.93, "step": 139880 }, { "epoch": 0.0007424, "grad_norm": 0.6940163373947144, "learning_rate": 1.4334582187413106e-05, "loss": 2.955, "step": 139890 }, { "epoch": 0.000768, "grad_norm": 0.7256541848182678, "learning_rate": 1.4333856477605075e-05, "loss": 3.1199, "step": 139900 }, { "epoch": 0.0007936, "grad_norm": 0.7619777917861938, "learning_rate": 1.4333130739693133e-05, "loss": 2.6637, "step": 139910 }, { "epoch": 0.0008192, "grad_norm": 0.8142132759094238, "learning_rate": 1.4332404973681989e-05, "loss": 3.0713, "step": 139920 }, { "epoch": 0.0008448, "grad_norm": 0.6898582577705383, "learning_rate": 1.433167917957635e-05, "loss": 2.9303, "step": 139930 }, { "epoch": 0.0008704, "grad_norm": 0.6669427752494812, "learning_rate": 1.4330953357380924e-05, "loss": 2.8972, "step": 139940 }, { "epoch": 0.000896, "grad_norm": 0.6577410101890564, "learning_rate": 1.4330227507100415e-05, "loss": 2.661, "step": 139950 }, { "epoch": 0.0009216, "grad_norm": 0.7526630163192749, "learning_rate": 1.4329501628739538e-05, "loss": 2.6107, "step": 139960 }, { "epoch": 0.0009472, "grad_norm": 0.7988859415054321, "learning_rate": 1.4328775722302988e-05, "loss": 3.042, "step": 139970 }, { "epoch": 0.0009728, "grad_norm": 0.7289075255393982, "learning_rate": 1.4328049787795478e-05, "loss": 2.8938, "step": 139980 }, { "epoch": 0.0009984, "grad_norm": 0.6908458471298218, "learning_rate": 1.4327323825221715e-05, "loss": 2.9697, "step": 139990 }, { "epoch": 0.001024, "grad_norm": 0.647413432598114, "learning_rate": 1.4326597834586409e-05, "loss": 3.0418, "step": 140000 }, { "epoch": 0.0010496, "grad_norm": 0.715656042098999, "learning_rate": 1.4325871815894266e-05, "loss": 2.8735, "step": 140010 }, { "epoch": 0.0010752, "grad_norm": 0.7791308760643005, "learning_rate": 1.4325145769149988e-05, "loss": 3.165, "step": 140020 }, { "epoch": 0.0011008, "grad_norm": 0.7976940870285034, "learning_rate": 1.4324419694358296e-05, "loss": 3.0021, "step": 140030 }, { "epoch": 0.0011264, "grad_norm": 0.8324291110038757, "learning_rate": 1.4323693591523889e-05, "loss": 3.0944, "step": 140040 }, { "epoch": 0.001152, "grad_norm": 0.7644986510276794, "learning_rate": 1.432296746065148e-05, "loss": 2.9309, "step": 140050 }, { "epoch": 0.0011776, "grad_norm": 0.7627471089363098, "learning_rate": 1.4322241301745776e-05, "loss": 2.7728, "step": 140060 }, { "epoch": 0.0012032, "grad_norm": 0.8019642233848572, "learning_rate": 1.4321515114811483e-05, "loss": 2.9253, "step": 140070 }, { "epoch": 0.0012288, "grad_norm": 0.7931530475616455, "learning_rate": 1.4320788899853314e-05, "loss": 2.8388, "step": 140080 }, { "epoch": 0.0012544, "grad_norm": 0.7418688535690308, "learning_rate": 1.4320062656875976e-05, "loss": 3.1778, "step": 140090 }, { "epoch": 0.00128, "grad_norm": 0.7452643513679504, "learning_rate": 1.4319336385884184e-05, "loss": 2.9003, "step": 140100 }, { "epoch": 0.0013056, "grad_norm": 0.9938323497772217, "learning_rate": 1.431861008688264e-05, "loss": 2.8008, "step": 140110 }, { "epoch": 0.0013312, "grad_norm": 0.7293104529380798, "learning_rate": 1.431788375987606e-05, "loss": 2.8738, "step": 140120 }, { "epoch": 0.0013568, "grad_norm": 0.8781518936157227, "learning_rate": 1.4317157404869151e-05, "loss": 2.9861, "step": 140130 }, { "epoch": 0.0013824, "grad_norm": 1.835423469543457, "learning_rate": 1.4316431021866622e-05, "loss": 3.1324, "step": 140140 }, { "epoch": 0.001408, "grad_norm": 0.7112947106361389, "learning_rate": 1.4315704610873184e-05, "loss": 3.1153, "step": 140150 }, { "epoch": 0.0014336, "grad_norm": 0.7811148166656494, "learning_rate": 1.4314978171893552e-05, "loss": 2.9591, "step": 140160 }, { "epoch": 0.0014592, "grad_norm": 0.6754451394081116, "learning_rate": 1.4314251704932432e-05, "loss": 2.8506, "step": 140170 }, { "epoch": 0.0014848, "grad_norm": 0.668342113494873, "learning_rate": 1.4313525209994534e-05, "loss": 2.7877, "step": 140180 }, { "epoch": 0.0015104, "grad_norm": 0.696399450302124, "learning_rate": 1.4312798687084572e-05, "loss": 2.587, "step": 140190 }, { "epoch": 0.001536, "grad_norm": 0.6798853874206543, "learning_rate": 1.4312072136207258e-05, "loss": 2.722, "step": 140200 }, { "epoch": 0.0015616, "grad_norm": 0.7802225351333618, "learning_rate": 1.4311345557367303e-05, "loss": 3.1273, "step": 140210 }, { "epoch": 0.0015872, "grad_norm": 0.7858554124832153, "learning_rate": 1.4310618950569416e-05, "loss": 2.828, "step": 140220 }, { "epoch": 0.0016128, "grad_norm": 0.8512685894966125, "learning_rate": 1.4309892315818313e-05, "loss": 3.0246, "step": 140230 }, { "epoch": 0.0016384, "grad_norm": 1.1425442695617676, "learning_rate": 1.4309165653118702e-05, "loss": 2.9366, "step": 140240 }, { "epoch": 0.001664, "grad_norm": 0.7401779890060425, "learning_rate": 1.4308438962475295e-05, "loss": 2.8755, "step": 140250 }, { "epoch": 0.0016896, "grad_norm": 0.8281723856925964, "learning_rate": 1.430771224389281e-05, "loss": 2.9305, "step": 140260 }, { "epoch": 0.0017152, "grad_norm": 0.7880387902259827, "learning_rate": 1.4306985497375955e-05, "loss": 3.0336, "step": 140270 }, { "epoch": 0.0017408, "grad_norm": 0.7003595232963562, "learning_rate": 1.4306258722929443e-05, "loss": 2.9406, "step": 140280 }, { "epoch": 0.0017664, "grad_norm": 0.7513831257820129, "learning_rate": 1.4305531920557987e-05, "loss": 2.9556, "step": 140290 }, { "epoch": 0.001792, "grad_norm": 0.67125403881073, "learning_rate": 1.4304805090266303e-05, "loss": 2.7942, "step": 140300 }, { "epoch": 0.0018176, "grad_norm": 0.7080345153808594, "learning_rate": 1.43040782320591e-05, "loss": 2.8042, "step": 140310 }, { "epoch": 0.0018432, "grad_norm": 0.6364331245422363, "learning_rate": 1.4303351345941092e-05, "loss": 2.6162, "step": 140320 }, { "epoch": 0.0018688, "grad_norm": 0.7309866547584534, "learning_rate": 1.4302624431917e-05, "loss": 2.9422, "step": 140330 }, { "epoch": 0.0018944, "grad_norm": 0.8426299095153809, "learning_rate": 1.4301897489991527e-05, "loss": 2.9912, "step": 140340 }, { "epoch": 0.00192, "grad_norm": 0.6882357001304626, "learning_rate": 1.4301170520169393e-05, "loss": 2.9106, "step": 140350 }, { "epoch": 0.0019456, "grad_norm": 0.7445717453956604, "learning_rate": 1.4300443522455311e-05, "loss": 3.0079, "step": 140360 }, { "epoch": 0.0019712, "grad_norm": 1.1641249656677246, "learning_rate": 1.4299716496853999e-05, "loss": 2.7508, "step": 140370 }, { "epoch": 0.0019968, "grad_norm": 0.7888004183769226, "learning_rate": 1.4298989443370163e-05, "loss": 2.8455, "step": 140380 }, { "epoch": 0.0020224, "grad_norm": 0.7049464583396912, "learning_rate": 1.429826236200853e-05, "loss": 2.7383, "step": 140390 }, { "epoch": 0.002048, "grad_norm": 0.7783179879188538, "learning_rate": 1.4297535252773802e-05, "loss": 2.7533, "step": 140400 }, { "epoch": 0.0020736, "grad_norm": 0.7300049066543579, "learning_rate": 1.4296808115670704e-05, "loss": 2.8066, "step": 140410 }, { "epoch": 0.0020992, "grad_norm": 0.6877595782279968, "learning_rate": 1.4296080950703943e-05, "loss": 2.8297, "step": 140420 }, { "epoch": 0.0021248, "grad_norm": 0.7738488912582397, "learning_rate": 1.4295353757878244e-05, "loss": 3.011, "step": 140430 }, { "epoch": 0.0021504, "grad_norm": 0.8864270448684692, "learning_rate": 1.4294626537198314e-05, "loss": 3.0675, "step": 140440 }, { "epoch": 0.002176, "grad_norm": 0.8755119442939758, "learning_rate": 1.4293899288668868e-05, "loss": 3.049, "step": 140450 }, { "epoch": 0.0022016, "grad_norm": 0.8151592016220093, "learning_rate": 1.4293172012294633e-05, "loss": 3.0137, "step": 140460 }, { "epoch": 0.0022272, "grad_norm": 1.2710386514663696, "learning_rate": 1.4292444708080317e-05, "loss": 3.2424, "step": 140470 }, { "epoch": 0.0022528, "grad_norm": 0.942988395690918, "learning_rate": 1.4291717376030638e-05, "loss": 2.7353, "step": 140480 }, { "epoch": 0.0022784, "grad_norm": 0.7175097465515137, "learning_rate": 1.4290990016150313e-05, "loss": 2.9599, "step": 140490 }, { "epoch": 0.002304, "grad_norm": 0.7916256785392761, "learning_rate": 1.4290262628444056e-05, "loss": 2.1474, "step": 140500 }, { "epoch": 2.56e-05, "grad_norm": 0.7095018625259399, "learning_rate": 1.4289535212916588e-05, "loss": 2.534, "step": 140510 }, { "epoch": 5.12e-05, "grad_norm": 0.9010604619979858, "learning_rate": 1.4288807769572621e-05, "loss": 2.8068, "step": 140520 }, { "epoch": 7.68e-05, "grad_norm": 0.6775065660476685, "learning_rate": 1.428808029841688e-05, "loss": 2.9637, "step": 140530 }, { "epoch": 0.0001024, "grad_norm": 0.713778555393219, "learning_rate": 1.4287352799454073e-05, "loss": 2.8571, "step": 140540 }, { "epoch": 0.000128, "grad_norm": 0.6945703625679016, "learning_rate": 1.4286625272688925e-05, "loss": 2.9158, "step": 140550 }, { "epoch": 0.0001536, "grad_norm": 0.7376394271850586, "learning_rate": 1.4285897718126154e-05, "loss": 2.8467, "step": 140560 }, { "epoch": 0.0001792, "grad_norm": 0.8483387231826782, "learning_rate": 1.4285170135770472e-05, "loss": 2.8852, "step": 140570 }, { "epoch": 0.0002048, "grad_norm": 0.9243755340576172, "learning_rate": 1.42844425256266e-05, "loss": 3.0008, "step": 140580 }, { "epoch": 0.0002304, "grad_norm": 0.6764582991600037, "learning_rate": 1.428371488769926e-05, "loss": 2.74, "step": 140590 }, { "epoch": 0.000256, "grad_norm": 0.8330690860748291, "learning_rate": 1.4282987221993166e-05, "loss": 2.7559, "step": 140600 }, { "epoch": 0.0002816, "grad_norm": 0.7889697551727295, "learning_rate": 1.4282259528513036e-05, "loss": 2.9588, "step": 140610 }, { "epoch": 0.0003072, "grad_norm": 0.6762669682502747, "learning_rate": 1.4281531807263594e-05, "loss": 2.7179, "step": 140620 }, { "epoch": 0.0003328, "grad_norm": 0.7114733457565308, "learning_rate": 1.4280804058249553e-05, "loss": 2.8869, "step": 140630 }, { "epoch": 0.0003584, "grad_norm": 0.6755879521369934, "learning_rate": 1.4280076281475638e-05, "loss": 2.9608, "step": 140640 }, { "epoch": 0.000384, "grad_norm": 0.8179463148117065, "learning_rate": 1.4279348476946563e-05, "loss": 2.8524, "step": 140650 }, { "epoch": 0.0004096, "grad_norm": 0.7624916434288025, "learning_rate": 1.4278620644667056e-05, "loss": 2.8653, "step": 140660 }, { "epoch": 0.0004352, "grad_norm": 0.7046999931335449, "learning_rate": 1.4277892784641828e-05, "loss": 2.7522, "step": 140670 }, { "epoch": 0.0004608, "grad_norm": 0.844559907913208, "learning_rate": 1.42771648968756e-05, "loss": 2.8959, "step": 140680 }, { "epoch": 0.0004864, "grad_norm": 0.7457081079483032, "learning_rate": 1.4276436981373099e-05, "loss": 2.8592, "step": 140690 }, { "epoch": 0.000512, "grad_norm": 0.9983905553817749, "learning_rate": 1.4275709038139038e-05, "loss": 2.8406, "step": 140700 }, { "epoch": 0.0005376, "grad_norm": 0.763706624507904, "learning_rate": 1.4274981067178137e-05, "loss": 2.8909, "step": 140710 }, { "epoch": 0.0005632, "grad_norm": 0.7056226134300232, "learning_rate": 1.4274253068495126e-05, "loss": 2.8589, "step": 140720 }, { "epoch": 0.0005888, "grad_norm": 1.034420132637024, "learning_rate": 1.4273525042094715e-05, "loss": 2.8354, "step": 140730 }, { "epoch": 0.0006144, "grad_norm": 0.6867685317993164, "learning_rate": 1.4272796987981634e-05, "loss": 2.7696, "step": 140740 }, { "epoch": 0.00064, "grad_norm": 0.7071799635887146, "learning_rate": 1.4272068906160596e-05, "loss": 3.0011, "step": 140750 }, { "epoch": 0.0006656, "grad_norm": 0.6944435238838196, "learning_rate": 1.4271340796636333e-05, "loss": 2.9534, "step": 140760 }, { "epoch": 0.0006912, "grad_norm": 1.2982946634292603, "learning_rate": 1.4270612659413553e-05, "loss": 2.8343, "step": 140770 }, { "epoch": 0.0007168, "grad_norm": 0.7532864212989807, "learning_rate": 1.4269884494496988e-05, "loss": 2.937, "step": 140780 }, { "epoch": 0.0007424, "grad_norm": 0.7178881764411926, "learning_rate": 1.4269156301891358e-05, "loss": 3.0677, "step": 140790 }, { "epoch": 0.000768, "grad_norm": 0.7064319252967834, "learning_rate": 1.426842808160138e-05, "loss": 2.8406, "step": 140800 }, { "epoch": 0.0007936, "grad_norm": 0.7754916548728943, "learning_rate": 1.4267699833631783e-05, "loss": 3.0009, "step": 140810 }, { "epoch": 0.0008192, "grad_norm": 0.6724854111671448, "learning_rate": 1.4266971557987289e-05, "loss": 2.8885, "step": 140820 }, { "epoch": 0.0008448, "grad_norm": 1.5865955352783203, "learning_rate": 1.4266243254672616e-05, "loss": 3.0451, "step": 140830 }, { "epoch": 0.0008704, "grad_norm": 0.7778950333595276, "learning_rate": 1.4265514923692488e-05, "loss": 2.7485, "step": 140840 }, { "epoch": 0.000896, "grad_norm": 0.7769004702568054, "learning_rate": 1.4264786565051631e-05, "loss": 2.6501, "step": 140850 }, { "epoch": 0.0009216, "grad_norm": 0.7677860260009766, "learning_rate": 1.4264058178754767e-05, "loss": 2.9718, "step": 140860 }, { "epoch": 0.0009472, "grad_norm": 0.633688747882843, "learning_rate": 1.4263329764806618e-05, "loss": 2.8016, "step": 140870 }, { "epoch": 0.0009728, "grad_norm": 0.7301807403564453, "learning_rate": 1.4262601323211908e-05, "loss": 2.8012, "step": 140880 }, { "epoch": 0.0009984, "grad_norm": 0.689495861530304, "learning_rate": 1.4261872853975363e-05, "loss": 2.7325, "step": 140890 }, { "epoch": 0.001024, "grad_norm": 0.8477718234062195, "learning_rate": 1.4261144357101705e-05, "loss": 2.6775, "step": 140900 }, { "epoch": 0.0010496, "grad_norm": 0.7333324551582336, "learning_rate": 1.4260415832595656e-05, "loss": 2.9047, "step": 140910 }, { "epoch": 0.0010752, "grad_norm": 0.7279645204544067, "learning_rate": 1.4259687280461944e-05, "loss": 3.0124, "step": 140920 }, { "epoch": 0.0011008, "grad_norm": 0.7555878162384033, "learning_rate": 1.4258958700705292e-05, "loss": 2.8477, "step": 140930 }, { "epoch": 0.0011264, "grad_norm": 0.7468396425247192, "learning_rate": 1.4258230093330426e-05, "loss": 2.8003, "step": 140940 }, { "epoch": 0.001152, "grad_norm": 0.742798388004303, "learning_rate": 1.4257501458342069e-05, "loss": 2.8738, "step": 140950 }, { "epoch": 0.0011776, "grad_norm": 1.1826841831207275, "learning_rate": 1.4256772795744947e-05, "loss": 2.8434, "step": 140960 }, { "epoch": 0.0012032, "grad_norm": 0.7153862118721008, "learning_rate": 1.425604410554378e-05, "loss": 3.0787, "step": 140970 }, { "epoch": 0.0012288, "grad_norm": 0.731800377368927, "learning_rate": 1.4255315387743304e-05, "loss": 2.903, "step": 140980 }, { "epoch": 0.0012544, "grad_norm": 0.742514431476593, "learning_rate": 1.4254586642348236e-05, "loss": 2.8928, "step": 140990 }, { "epoch": 0.00128, "grad_norm": 0.7734864354133606, "learning_rate": 1.4253857869363306e-05, "loss": 2.9518, "step": 141000 }, { "epoch": 0.0013056, "grad_norm": 0.6938910484313965, "learning_rate": 1.4253129068793236e-05, "loss": 2.645, "step": 141010 }, { "epoch": 0.0013312, "grad_norm": 0.7527275681495667, "learning_rate": 1.4252400240642756e-05, "loss": 2.8946, "step": 141020 }, { "epoch": 0.0013568, "grad_norm": 0.9649232625961304, "learning_rate": 1.4251671384916593e-05, "loss": 3.0558, "step": 141030 }, { "epoch": 0.0013824, "grad_norm": 0.6598964333534241, "learning_rate": 1.4250942501619467e-05, "loss": 2.909, "step": 141040 }, { "epoch": 0.001408, "grad_norm": 0.6608702540397644, "learning_rate": 1.425021359075611e-05, "loss": 2.7936, "step": 141050 }, { "epoch": 0.0014336, "grad_norm": 0.7687805891036987, "learning_rate": 1.4249484652331246e-05, "loss": 2.9891, "step": 141060 }, { "epoch": 0.0014592, "grad_norm": 0.8315433263778687, "learning_rate": 1.4248755686349605e-05, "loss": 2.839, "step": 141070 }, { "epoch": 0.0014848, "grad_norm": 0.706589937210083, "learning_rate": 1.4248026692815913e-05, "loss": 2.5266, "step": 141080 }, { "epoch": 0.0015104, "grad_norm": 0.7002621293067932, "learning_rate": 1.4247297671734895e-05, "loss": 2.5575, "step": 141090 }, { "epoch": 0.001536, "grad_norm": 0.7565047740936279, "learning_rate": 1.4246568623111281e-05, "loss": 2.7061, "step": 141100 }, { "epoch": 2.56e-05, "grad_norm": 0.7982711791992188, "learning_rate": 1.4245839546949798e-05, "loss": 2.625, "step": 141110 }, { "epoch": 5.12e-05, "grad_norm": 0.7911417484283447, "learning_rate": 1.4245110443255176e-05, "loss": 2.9288, "step": 141120 }, { "epoch": 7.68e-05, "grad_norm": 1.7937456369400024, "learning_rate": 1.424438131203214e-05, "loss": 3.0671, "step": 141130 }, { "epoch": 0.0001024, "grad_norm": 0.844904363155365, "learning_rate": 1.4243652153285419e-05, "loss": 2.831, "step": 141140 }, { "epoch": 0.000128, "grad_norm": 0.7268857955932617, "learning_rate": 1.424292296701974e-05, "loss": 2.8286, "step": 141150 }, { "epoch": 0.0001536, "grad_norm": 0.6962952613830566, "learning_rate": 1.4242193753239835e-05, "loss": 2.9935, "step": 141160 }, { "epoch": 0.0001792, "grad_norm": 0.9585915207862854, "learning_rate": 1.4241464511950429e-05, "loss": 3.0322, "step": 141170 }, { "epoch": 0.0002048, "grad_norm": 0.7166131138801575, "learning_rate": 1.4240735243156255e-05, "loss": 2.5937, "step": 141180 }, { "epoch": 0.0002304, "grad_norm": 0.75483638048172, "learning_rate": 1.4240005946862041e-05, "loss": 2.9939, "step": 141190 }, { "epoch": 0.000256, "grad_norm": 0.918612003326416, "learning_rate": 1.4239276623072513e-05, "loss": 2.7916, "step": 141200 }, { "epoch": 0.0002816, "grad_norm": 1.5669434070587158, "learning_rate": 1.4238547271792405e-05, "loss": 2.9283, "step": 141210 }, { "epoch": 0.0003072, "grad_norm": 1.0324524641036987, "learning_rate": 1.4237817893026443e-05, "loss": 2.891, "step": 141220 }, { "epoch": 0.0003328, "grad_norm": 0.9960930347442627, "learning_rate": 1.4237088486779355e-05, "loss": 3.0479, "step": 141230 }, { "epoch": 0.0003584, "grad_norm": 0.7512034773826599, "learning_rate": 1.4236359053055876e-05, "loss": 2.8791, "step": 141240 }, { "epoch": 0.000384, "grad_norm": 0.7416692972183228, "learning_rate": 1.4235629591860737e-05, "loss": 3.0243, "step": 141250 }, { "epoch": 0.0004096, "grad_norm": 0.6695451140403748, "learning_rate": 1.4234900103198666e-05, "loss": 2.884, "step": 141260 }, { "epoch": 0.0004352, "grad_norm": 0.8698285222053528, "learning_rate": 1.4234170587074392e-05, "loss": 2.9916, "step": 141270 }, { "epoch": 0.0004608, "grad_norm": 0.7411278486251831, "learning_rate": 1.4233441043492649e-05, "loss": 2.8041, "step": 141280 }, { "epoch": 0.0004864, "grad_norm": 0.796816349029541, "learning_rate": 1.4232711472458163e-05, "loss": 3.0704, "step": 141290 }, { "epoch": 0.000512, "grad_norm": 0.6255020499229431, "learning_rate": 1.4231981873975666e-05, "loss": 2.7903, "step": 141300 }, { "epoch": 0.0005376, "grad_norm": 0.7089327573776245, "learning_rate": 1.4231252248049896e-05, "loss": 2.8531, "step": 141310 }, { "epoch": 0.0005632, "grad_norm": 0.7248948216438293, "learning_rate": 1.4230522594685577e-05, "loss": 2.7403, "step": 141320 }, { "epoch": 0.0005888, "grad_norm": 0.6854754090309143, "learning_rate": 1.4229792913887444e-05, "loss": 2.5749, "step": 141330 }, { "epoch": 0.0006144, "grad_norm": 0.7058699727058411, "learning_rate": 1.4229063205660228e-05, "loss": 2.7174, "step": 141340 }, { "epoch": 0.00064, "grad_norm": 0.7168949246406555, "learning_rate": 1.4228333470008661e-05, "loss": 3.0803, "step": 141350 }, { "epoch": 0.0006656, "grad_norm": 0.7622109651565552, "learning_rate": 1.4227603706937476e-05, "loss": 3.2288, "step": 141360 }, { "epoch": 0.0006912, "grad_norm": 0.6211451888084412, "learning_rate": 1.4226873916451404e-05, "loss": 2.7551, "step": 141370 }, { "epoch": 0.0007168, "grad_norm": 0.7481512427330017, "learning_rate": 1.4226144098555182e-05, "loss": 2.9205, "step": 141380 }, { "epoch": 0.0007424, "grad_norm": 0.7618716955184937, "learning_rate": 1.4225414253253533e-05, "loss": 2.8266, "step": 141390 }, { "epoch": 0.000768, "grad_norm": 0.8821691870689392, "learning_rate": 1.4224684380551197e-05, "loss": 2.8233, "step": 141400 }, { "epoch": 0.0007936, "grad_norm": 0.6927221417427063, "learning_rate": 1.4223954480452906e-05, "loss": 2.9865, "step": 141410 }, { "epoch": 0.0008192, "grad_norm": 0.7872273921966553, "learning_rate": 1.4223224552963393e-05, "loss": 2.7441, "step": 141420 }, { "epoch": 0.0008448, "grad_norm": 0.7962409257888794, "learning_rate": 1.4222494598087389e-05, "loss": 2.9508, "step": 141430 }, { "epoch": 0.0008704, "grad_norm": 0.6440117359161377, "learning_rate": 1.4221764615829632e-05, "loss": 2.7535, "step": 141440 }, { "epoch": 0.000896, "grad_norm": 0.8728603720664978, "learning_rate": 1.4221034606194854e-05, "loss": 2.86, "step": 141450 }, { "epoch": 0.0009216, "grad_norm": 0.7274469137191772, "learning_rate": 1.4220304569187788e-05, "loss": 2.8997, "step": 141460 }, { "epoch": 0.0009472, "grad_norm": 0.71063232421875, "learning_rate": 1.4219574504813167e-05, "loss": 2.8688, "step": 141470 }, { "epoch": 0.0009728, "grad_norm": 0.7359241843223572, "learning_rate": 1.421884441307573e-05, "loss": 2.9078, "step": 141480 }, { "epoch": 0.0009984, "grad_norm": 0.6999552845954895, "learning_rate": 1.4218114293980205e-05, "loss": 2.9835, "step": 141490 }, { "epoch": 0.001024, "grad_norm": 7.764564514160156, "learning_rate": 1.4217384147531328e-05, "loss": 3.0673, "step": 141500 }, { "epoch": 0.0010496, "grad_norm": 0.8005291819572449, "learning_rate": 1.421665397373384e-05, "loss": 3.0034, "step": 141510 }, { "epoch": 0.0010752, "grad_norm": 0.8081415295600891, "learning_rate": 1.421592377259247e-05, "loss": 2.8467, "step": 141520 }, { "epoch": 0.0011008, "grad_norm": 0.8638994693756104, "learning_rate": 1.4215193544111956e-05, "loss": 2.9202, "step": 141530 }, { "epoch": 0.0011264, "grad_norm": 0.7699359059333801, "learning_rate": 1.421446328829703e-05, "loss": 2.8077, "step": 141540 }, { "epoch": 0.001152, "grad_norm": 0.8051613569259644, "learning_rate": 1.421373300515243e-05, "loss": 2.806, "step": 141550 }, { "epoch": 0.0011776, "grad_norm": 0.8658273220062256, "learning_rate": 1.4213002694682894e-05, "loss": 2.8246, "step": 141560 }, { "epoch": 0.0012032, "grad_norm": 0.9012457132339478, "learning_rate": 1.4212272356893153e-05, "loss": 3.0584, "step": 141570 }, { "epoch": 0.0012288, "grad_norm": 0.8468641042709351, "learning_rate": 1.4211541991787945e-05, "loss": 2.8509, "step": 141580 }, { "epoch": 0.0012544, "grad_norm": 0.691803514957428, "learning_rate": 1.4210811599372006e-05, "loss": 2.9612, "step": 141590 }, { "epoch": 0.00128, "grad_norm": 0.7481704354286194, "learning_rate": 1.4210081179650073e-05, "loss": 3.0053, "step": 141600 }, { "epoch": 0.0013056, "grad_norm": 0.6937906742095947, "learning_rate": 1.4209350732626885e-05, "loss": 2.8154, "step": 141610 }, { "epoch": 0.0013312, "grad_norm": 0.9471346735954285, "learning_rate": 1.4208620258307172e-05, "loss": 2.8718, "step": 141620 }, { "epoch": 0.0013568, "grad_norm": 0.9100236892700195, "learning_rate": 1.4207889756695679e-05, "loss": 2.9132, "step": 141630 }, { "epoch": 0.0013824, "grad_norm": 0.6787776350975037, "learning_rate": 1.4207159227797139e-05, "loss": 2.7055, "step": 141640 }, { "epoch": 0.001408, "grad_norm": 0.7185527086257935, "learning_rate": 1.4206428671616288e-05, "loss": 2.8252, "step": 141650 }, { "epoch": 0.0014336, "grad_norm": 0.8056452870368958, "learning_rate": 1.4205698088157866e-05, "loss": 2.9273, "step": 141660 }, { "epoch": 0.0014592, "grad_norm": 0.7108494639396667, "learning_rate": 1.4204967477426607e-05, "loss": 2.9137, "step": 141670 }, { "epoch": 0.0014848, "grad_norm": 0.6803981065750122, "learning_rate": 1.4204236839427254e-05, "loss": 2.6722, "step": 141680 }, { "epoch": 0.0015104, "grad_norm": 0.7621194124221802, "learning_rate": 1.4203506174164543e-05, "loss": 2.5678, "step": 141690 }, { "epoch": 0.001536, "grad_norm": 0.7835968732833862, "learning_rate": 1.4202775481643209e-05, "loss": 2.7893, "step": 141700 }, { "epoch": 0.0015616, "grad_norm": 0.6815094947814941, "learning_rate": 1.4202044761867997e-05, "loss": 2.8307, "step": 141710 }, { "epoch": 0.0015872, "grad_norm": 0.7908379435539246, "learning_rate": 1.420131401484364e-05, "loss": 2.6583, "step": 141720 }, { "epoch": 0.0016128, "grad_norm": 0.7424555420875549, "learning_rate": 1.4200583240574875e-05, "loss": 2.9593, "step": 141730 }, { "epoch": 0.0016384, "grad_norm": 0.9902087450027466, "learning_rate": 1.4199852439066446e-05, "loss": 2.9885, "step": 141740 }, { "epoch": 0.001664, "grad_norm": 0.6982758641242981, "learning_rate": 1.4199121610323092e-05, "loss": 2.9641, "step": 141750 }, { "epoch": 0.0016896, "grad_norm": 0.8276152014732361, "learning_rate": 1.4198390754349547e-05, "loss": 2.7052, "step": 141760 }, { "epoch": 0.0017152, "grad_norm": 0.9229065775871277, "learning_rate": 1.4197659871150556e-05, "loss": 2.7051, "step": 141770 }, { "epoch": 0.0017408, "grad_norm": 0.7201256155967712, "learning_rate": 1.4196928960730855e-05, "loss": 2.9762, "step": 141780 }, { "epoch": 0.0017664, "grad_norm": 0.913431704044342, "learning_rate": 1.4196198023095186e-05, "loss": 3.0019, "step": 141790 }, { "epoch": 0.001792, "grad_norm": 0.8027333617210388, "learning_rate": 1.4195467058248289e-05, "loss": 2.701, "step": 141800 }, { "epoch": 0.0018176, "grad_norm": 0.7430017590522766, "learning_rate": 1.4194736066194903e-05, "loss": 2.9221, "step": 141810 }, { "epoch": 0.0018432, "grad_norm": 0.6975919008255005, "learning_rate": 1.4194005046939768e-05, "loss": 2.6051, "step": 141820 }, { "epoch": 0.0018688, "grad_norm": 0.8169081211090088, "learning_rate": 1.4193274000487622e-05, "loss": 2.9357, "step": 141830 }, { "epoch": 0.0018944, "grad_norm": 0.7712035775184631, "learning_rate": 1.4192542926843212e-05, "loss": 2.9843, "step": 141840 }, { "epoch": 0.00192, "grad_norm": 0.6562350392341614, "learning_rate": 1.4191811826011272e-05, "loss": 2.8517, "step": 141850 }, { "epoch": 0.0019456, "grad_norm": 0.6876917481422424, "learning_rate": 1.419108069799655e-05, "loss": 2.8472, "step": 141860 }, { "epoch": 0.0019712, "grad_norm": 1.2434996366500854, "learning_rate": 1.4190349542803782e-05, "loss": 3.0536, "step": 141870 }, { "epoch": 0.0019968, "grad_norm": 1.016282320022583, "learning_rate": 1.4189618360437713e-05, "loss": 3.0558, "step": 141880 }, { "epoch": 0.0020224, "grad_norm": 0.819394588470459, "learning_rate": 1.4188887150903079e-05, "loss": 3.062, "step": 141890 }, { "epoch": 0.002048, "grad_norm": 0.7358492612838745, "learning_rate": 1.4188155914204626e-05, "loss": 2.6821, "step": 141900 }, { "epoch": 0.0020736, "grad_norm": 0.7172924876213074, "learning_rate": 1.4187424650347099e-05, "loss": 2.6653, "step": 141910 }, { "epoch": 0.0020992, "grad_norm": 0.6943151354789734, "learning_rate": 1.4186693359335232e-05, "loss": 2.8895, "step": 141920 }, { "epoch": 0.0021248, "grad_norm": 0.9414620995521545, "learning_rate": 1.4185962041173769e-05, "loss": 2.8561, "step": 141930 }, { "epoch": 0.0021504, "grad_norm": 0.7797385454177856, "learning_rate": 1.418523069586746e-05, "loss": 3.1542, "step": 141940 }, { "epoch": 0.002176, "grad_norm": 0.6706773638725281, "learning_rate": 1.4184499323421039e-05, "loss": 3.1163, "step": 141950 }, { "epoch": 0.0022016, "grad_norm": 0.7942005395889282, "learning_rate": 1.4183767923839252e-05, "loss": 3.0022, "step": 141960 }, { "epoch": 0.0022272, "grad_norm": 0.792655348777771, "learning_rate": 1.4183036497126845e-05, "loss": 2.787, "step": 141970 }, { "epoch": 0.0022528, "grad_norm": 0.6820570826530457, "learning_rate": 1.4182305043288556e-05, "loss": 2.9207, "step": 141980 }, { "epoch": 0.0022784, "grad_norm": 0.697453498840332, "learning_rate": 1.418157356232913e-05, "loss": 2.8727, "step": 141990 }, { "epoch": 0.002304, "grad_norm": 0.8027113080024719, "learning_rate": 1.418084205425331e-05, "loss": 2.767, "step": 142000 }, { "epoch": 0.0023296, "grad_norm": 0.7700368762016296, "learning_rate": 1.4180110519065841e-05, "loss": 3.0452, "step": 142010 }, { "epoch": 0.0023552, "grad_norm": 0.7672551870346069, "learning_rate": 1.4179378956771466e-05, "loss": 2.9494, "step": 142020 }, { "epoch": 0.0023808, "grad_norm": 0.764110267162323, "learning_rate": 1.4178647367374929e-05, "loss": 2.8296, "step": 142030 }, { "epoch": 0.0024064, "grad_norm": 0.7629730105400085, "learning_rate": 1.4177915750880975e-05, "loss": 2.993, "step": 142040 }, { "epoch": 0.002432, "grad_norm": 0.6896069645881653, "learning_rate": 1.4177184107294348e-05, "loss": 2.9663, "step": 142050 }, { "epoch": 0.0024576, "grad_norm": 0.7499030232429504, "learning_rate": 1.4176452436619791e-05, "loss": 2.8215, "step": 142060 }, { "epoch": 0.0024832, "grad_norm": 0.73054039478302, "learning_rate": 1.4175720738862053e-05, "loss": 2.9632, "step": 142070 }, { "epoch": 0.0025088, "grad_norm": 0.8965749144554138, "learning_rate": 1.4174989014025875e-05, "loss": 3.081, "step": 142080 }, { "epoch": 0.0025344, "grad_norm": 0.7156023979187012, "learning_rate": 1.4174257262116e-05, "loss": 2.8729, "step": 142090 }, { "epoch": 0.00256, "grad_norm": 0.7219399809837341, "learning_rate": 1.4173525483137179e-05, "loss": 3.2253, "step": 142100 }, { "epoch": 0.0025856, "grad_norm": 0.7457805871963501, "learning_rate": 1.4172793677094152e-05, "loss": 2.9972, "step": 142110 }, { "epoch": 0.0026112, "grad_norm": 0.6942857503890991, "learning_rate": 1.4172061843991666e-05, "loss": 2.863, "step": 142120 }, { "epoch": 0.0026368, "grad_norm": 0.6516512036323547, "learning_rate": 1.4171329983834469e-05, "loss": 2.825, "step": 142130 }, { "epoch": 0.0026624, "grad_norm": 1.6177695989608765, "learning_rate": 1.4170598096627309e-05, "loss": 3.0049, "step": 142140 }, { "epoch": 0.002688, "grad_norm": 0.7404622435569763, "learning_rate": 1.4169866182374925e-05, "loss": 3.0847, "step": 142150 }, { "epoch": 0.0027136, "grad_norm": 1.0555000305175781, "learning_rate": 1.416913424108207e-05, "loss": 2.9275, "step": 142160 }, { "epoch": 0.0027392, "grad_norm": 0.759392499923706, "learning_rate": 1.4168402272753486e-05, "loss": 2.8897, "step": 142170 }, { "epoch": 0.0027648, "grad_norm": 0.6842783689498901, "learning_rate": 1.4167670277393918e-05, "loss": 2.5876, "step": 142180 }, { "epoch": 0.0027904, "grad_norm": 0.7349345684051514, "learning_rate": 1.4166938255008117e-05, "loss": 2.7843, "step": 142190 }, { "epoch": 0.002816, "grad_norm": 0.7820003628730774, "learning_rate": 1.4166206205600834e-05, "loss": 3.0331, "step": 142200 }, { "epoch": 0.0028416, "grad_norm": 0.6997767686843872, "learning_rate": 1.4165474129176806e-05, "loss": 3.1047, "step": 142210 }, { "epoch": 0.0028672, "grad_norm": 1.0214338302612305, "learning_rate": 1.4164742025740787e-05, "loss": 3.0131, "step": 142220 }, { "epoch": 0.0028928, "grad_norm": 0.7022421360015869, "learning_rate": 1.4164009895297523e-05, "loss": 2.9413, "step": 142230 }, { "epoch": 0.0029184, "grad_norm": 0.7405027151107788, "learning_rate": 1.4163277737851762e-05, "loss": 3.0169, "step": 142240 }, { "epoch": 0.002944, "grad_norm": 0.7395433783531189, "learning_rate": 1.416254555340825e-05, "loss": 3.0066, "step": 142250 }, { "epoch": 0.0029696, "grad_norm": 0.7352914810180664, "learning_rate": 1.4161813341971737e-05, "loss": 3.0912, "step": 142260 }, { "epoch": 0.0029952, "grad_norm": 0.7140201926231384, "learning_rate": 1.4161081103546973e-05, "loss": 3.0828, "step": 142270 }, { "epoch": 0.0030208, "grad_norm": 0.7601451277732849, "learning_rate": 1.4160348838138698e-05, "loss": 2.9695, "step": 142280 }, { "epoch": 0.0030464, "grad_norm": 1.2339825630187988, "learning_rate": 1.4159616545751669e-05, "loss": 2.8374, "step": 142290 }, { "epoch": 0.003072, "grad_norm": 0.7046846151351929, "learning_rate": 1.4158884226390635e-05, "loss": 3.0666, "step": 142300 }, { "epoch": 0.0030976, "grad_norm": 0.7105411887168884, "learning_rate": 1.4158151880060338e-05, "loss": 3.0492, "step": 142310 }, { "epoch": 0.0031232, "grad_norm": 0.7364956140518188, "learning_rate": 1.4157419506765533e-05, "loss": 3.2469, "step": 142320 }, { "epoch": 0.0031488, "grad_norm": 0.7018001079559326, "learning_rate": 1.4156687106510968e-05, "loss": 2.944, "step": 142330 }, { "epoch": 0.0031744, "grad_norm": 0.8160709142684937, "learning_rate": 1.4155954679301393e-05, "loss": 2.9256, "step": 142340 }, { "epoch": 0.0032, "grad_norm": 0.710280179977417, "learning_rate": 1.4155222225141554e-05, "loss": 2.9302, "step": 142350 }, { "epoch": 0.0032256, "grad_norm": 0.8248473405838013, "learning_rate": 1.4154489744036204e-05, "loss": 3.1, "step": 142360 }, { "epoch": 0.0032512, "grad_norm": 0.7982691526412964, "learning_rate": 1.4153757235990093e-05, "loss": 3.0219, "step": 142370 }, { "epoch": 0.0032768, "grad_norm": 0.7668217420578003, "learning_rate": 1.415302470100797e-05, "loss": 2.9868, "step": 142380 }, { "epoch": 0.0033024, "grad_norm": 0.7599076628684998, "learning_rate": 1.4152292139094583e-05, "loss": 2.9781, "step": 142390 }, { "epoch": 0.003328, "grad_norm": 0.7689213156700134, "learning_rate": 1.415155955025469e-05, "loss": 3.136, "step": 142400 }, { "epoch": 0.0033536, "grad_norm": 1.05978524684906, "learning_rate": 1.4150826934493032e-05, "loss": 3.0505, "step": 142410 }, { "epoch": 0.0033792, "grad_norm": 0.740631639957428, "learning_rate": 1.4150094291814367e-05, "loss": 2.8947, "step": 142420 }, { "epoch": 0.0034048, "grad_norm": 0.7429761290550232, "learning_rate": 1.4149361622223441e-05, "loss": 3.0193, "step": 142430 }, { "epoch": 0.0034304, "grad_norm": 0.7618638277053833, "learning_rate": 1.4148628925725012e-05, "loss": 2.7845, "step": 142440 }, { "epoch": 0.003456, "grad_norm": 0.7118881940841675, "learning_rate": 1.414789620232382e-05, "loss": 2.9424, "step": 142450 }, { "epoch": 0.0034816, "grad_norm": 0.7187985777854919, "learning_rate": 1.414716345202463e-05, "loss": 2.9369, "step": 142460 }, { "epoch": 0.0035072, "grad_norm": 0.7748449444770813, "learning_rate": 1.4146430674832185e-05, "loss": 2.8717, "step": 142470 }, { "epoch": 0.0035328, "grad_norm": 0.7246907353401184, "learning_rate": 1.4145697870751241e-05, "loss": 3.0959, "step": 142480 }, { "epoch": 0.0035584, "grad_norm": 1.0045931339263916, "learning_rate": 1.4144965039786544e-05, "loss": 3.0057, "step": 142490 }, { "epoch": 0.003584, "grad_norm": 0.86843341588974, "learning_rate": 1.4144232181942853e-05, "loss": 2.9297, "step": 142500 }, { "epoch": 0.0036096, "grad_norm": 0.7821683287620544, "learning_rate": 1.4143499297224918e-05, "loss": 3.0846, "step": 142510 }, { "epoch": 0.0036352, "grad_norm": 0.6855446696281433, "learning_rate": 1.414276638563749e-05, "loss": 2.905, "step": 142520 }, { "epoch": 0.0036608, "grad_norm": 0.8398676514625549, "learning_rate": 1.4142033447185323e-05, "loss": 3.0505, "step": 142530 }, { "epoch": 0.0036864, "grad_norm": 0.7430033683776855, "learning_rate": 1.414130048187317e-05, "loss": 2.9913, "step": 142540 }, { "epoch": 0.003712, "grad_norm": 0.7113761901855469, "learning_rate": 1.4140567489705786e-05, "loss": 3.0147, "step": 142550 }, { "epoch": 0.0037376, "grad_norm": 0.7197372317314148, "learning_rate": 1.413983447068792e-05, "loss": 2.9215, "step": 142560 }, { "epoch": 0.0037632, "grad_norm": 0.8762154579162598, "learning_rate": 1.4139101424824329e-05, "loss": 3.014, "step": 142570 }, { "epoch": 0.0037888, "grad_norm": 0.6750792264938354, "learning_rate": 1.4138368352119764e-05, "loss": 3.0542, "step": 142580 }, { "epoch": 0.0038144, "grad_norm": 0.7766822576522827, "learning_rate": 1.4137635252578981e-05, "loss": 2.9473, "step": 142590 }, { "epoch": 0.00384, "grad_norm": 0.7678459882736206, "learning_rate": 1.4136902126206732e-05, "loss": 3.0488, "step": 142600 }, { "epoch": 0.0038656, "grad_norm": 0.7460182905197144, "learning_rate": 1.4136168973007773e-05, "loss": 2.8756, "step": 142610 }, { "epoch": 0.0038912, "grad_norm": 0.7002955675125122, "learning_rate": 1.4135435792986857e-05, "loss": 3.1809, "step": 142620 }, { "epoch": 0.0039168, "grad_norm": 0.731963038444519, "learning_rate": 1.413470258614874e-05, "loss": 3.1846, "step": 142630 }, { "epoch": 0.0039424, "grad_norm": 0.8153712749481201, "learning_rate": 1.4133969352498174e-05, "loss": 3.0509, "step": 142640 }, { "epoch": 0.003968, "grad_norm": 0.7639964818954468, "learning_rate": 1.4133236092039917e-05, "loss": 3.0333, "step": 142650 }, { "epoch": 0.0039936, "grad_norm": 0.6986294388771057, "learning_rate": 1.4132502804778723e-05, "loss": 2.9659, "step": 142660 }, { "epoch": 0.0040192, "grad_norm": 0.9455350041389465, "learning_rate": 1.4131769490719345e-05, "loss": 3.1151, "step": 142670 }, { "epoch": 0.0040448, "grad_norm": 0.69926917552948, "learning_rate": 1.4131036149866541e-05, "loss": 3.2072, "step": 142680 }, { "epoch": 0.0040704, "grad_norm": 1.0069795846939087, "learning_rate": 1.4130302782225065e-05, "loss": 2.9418, "step": 142690 }, { "epoch": 0.004096, "grad_norm": 0.7201743125915527, "learning_rate": 1.4129569387799674e-05, "loss": 2.854, "step": 142700 }, { "epoch": 0.0041216, "grad_norm": 0.8648353219032288, "learning_rate": 1.412883596659512e-05, "loss": 2.9898, "step": 142710 }, { "epoch": 0.0041472, "grad_norm": 0.8102190494537354, "learning_rate": 1.4128102518616163e-05, "loss": 2.8732, "step": 142720 }, { "epoch": 0.0041728, "grad_norm": 0.660193681716919, "learning_rate": 1.412736904386756e-05, "loss": 2.9419, "step": 142730 }, { "epoch": 0.0041984, "grad_norm": 0.679689347743988, "learning_rate": 1.4126635542354066e-05, "loss": 3.1461, "step": 142740 }, { "epoch": 0.004224, "grad_norm": 0.7392802834510803, "learning_rate": 1.4125902014080436e-05, "loss": 3.2409, "step": 142750 }, { "epoch": 0.0042496, "grad_norm": 0.6933237910270691, "learning_rate": 1.4125168459051431e-05, "loss": 3.036, "step": 142760 }, { "epoch": 0.0042752, "grad_norm": 0.706977128982544, "learning_rate": 1.41244348772718e-05, "loss": 3.0782, "step": 142770 }, { "epoch": 0.0043008, "grad_norm": 0.6849939227104187, "learning_rate": 1.4123701268746306e-05, "loss": 3.0212, "step": 142780 }, { "epoch": 0.0043264, "grad_norm": 1.072235107421875, "learning_rate": 1.4122967633479707e-05, "loss": 2.972, "step": 142790 }, { "epoch": 0.004352, "grad_norm": 0.7551529407501221, "learning_rate": 1.4122233971476758e-05, "loss": 3.2136, "step": 142800 }, { "epoch": 0.0043776, "grad_norm": 0.7795946002006531, "learning_rate": 1.4121500282742215e-05, "loss": 3.1086, "step": 142810 }, { "epoch": 0.0044032, "grad_norm": 1.517099142074585, "learning_rate": 1.4120766567280842e-05, "loss": 2.9383, "step": 142820 }, { "epoch": 0.0044288, "grad_norm": 0.7739469408988953, "learning_rate": 1.4120032825097392e-05, "loss": 2.881, "step": 142830 }, { "epoch": 0.0044544, "grad_norm": 0.7198218107223511, "learning_rate": 1.4119299056196622e-05, "loss": 3.0937, "step": 142840 }, { "epoch": 0.00448, "grad_norm": 0.6969588994979858, "learning_rate": 1.4118565260583294e-05, "loss": 3.2327, "step": 142850 }, { "epoch": 0.0045056, "grad_norm": 0.7714828252792358, "learning_rate": 1.4117831438262166e-05, "loss": 2.922, "step": 142860 }, { "epoch": 0.0045312, "grad_norm": 0.6363928914070129, "learning_rate": 1.4117097589237993e-05, "loss": 2.9904, "step": 142870 }, { "epoch": 0.0045568, "grad_norm": 0.8273714184761047, "learning_rate": 1.4116363713515538e-05, "loss": 3.1, "step": 142880 }, { "epoch": 0.0045824, "grad_norm": 0.6992892026901245, "learning_rate": 1.4115629811099558e-05, "loss": 2.9294, "step": 142890 }, { "epoch": 0.004608, "grad_norm": 0.8000662922859192, "learning_rate": 1.4114895881994812e-05, "loss": 3.1799, "step": 142900 }, { "epoch": 0.0046336, "grad_norm": 0.7300844788551331, "learning_rate": 1.4114161926206058e-05, "loss": 3.0141, "step": 142910 }, { "epoch": 0.0046592, "grad_norm": 1.0715134143829346, "learning_rate": 1.411342794373806e-05, "loss": 2.9038, "step": 142920 }, { "epoch": 0.0046848, "grad_norm": 0.9893496632575989, "learning_rate": 1.4112693934595577e-05, "loss": 3.0887, "step": 142930 }, { "epoch": 0.0047104, "grad_norm": 0.7653165459632874, "learning_rate": 1.4111959898783363e-05, "loss": 3.193, "step": 142940 }, { "epoch": 0.004736, "grad_norm": 0.7290237545967102, "learning_rate": 1.4111225836306184e-05, "loss": 3.1067, "step": 142950 }, { "epoch": 0.0047616, "grad_norm": 0.8883979320526123, "learning_rate": 1.4110491747168798e-05, "loss": 3.0897, "step": 142960 }, { "epoch": 0.0047872, "grad_norm": 0.7198843359947205, "learning_rate": 1.4109757631375962e-05, "loss": 2.902, "step": 142970 }, { "epoch": 0.0048128, "grad_norm": 0.7930749654769897, "learning_rate": 1.4109023488932443e-05, "loss": 2.9232, "step": 142980 }, { "epoch": 0.0048384, "grad_norm": 0.7386233806610107, "learning_rate": 1.4108289319843e-05, "loss": 2.8817, "step": 142990 }, { "epoch": 0.004864, "grad_norm": 0.6919589042663574, "learning_rate": 1.4107555124112389e-05, "loss": 3.0752, "step": 143000 }, { "epoch": 0.0048896, "grad_norm": 0.762208878993988, "learning_rate": 1.4106820901745377e-05, "loss": 2.8977, "step": 143010 }, { "epoch": 0.0049152, "grad_norm": 0.7281240820884705, "learning_rate": 1.4106086652746722e-05, "loss": 2.9721, "step": 143020 }, { "epoch": 0.0049408, "grad_norm": 0.7259583473205566, "learning_rate": 1.4105352377121188e-05, "loss": 2.8537, "step": 143030 }, { "epoch": 0.0049664, "grad_norm": 0.816375195980072, "learning_rate": 1.4104618074873535e-05, "loss": 2.96, "step": 143040 }, { "epoch": 0.004992, "grad_norm": 0.6993662118911743, "learning_rate": 1.4103883746008523e-05, "loss": 2.9084, "step": 143050 }, { "epoch": 0.0050176, "grad_norm": 0.7356039881706238, "learning_rate": 1.4103149390530915e-05, "loss": 3.0841, "step": 143060 }, { "epoch": 0.0050432, "grad_norm": 0.889083743095398, "learning_rate": 1.4102415008445474e-05, "loss": 2.8905, "step": 143070 }, { "epoch": 0.0050688, "grad_norm": 1.3954850435256958, "learning_rate": 1.4101680599756963e-05, "loss": 3.0041, "step": 143080 }, { "epoch": 0.0050944, "grad_norm": 0.8726280331611633, "learning_rate": 1.4100946164470144e-05, "loss": 3.0154, "step": 143090 }, { "epoch": 0.00512, "grad_norm": 0.6873104572296143, "learning_rate": 1.4100211702589778e-05, "loss": 2.8714, "step": 143100 }, { "epoch": 0.0051456, "grad_norm": 0.6927973031997681, "learning_rate": 1.4099477214120628e-05, "loss": 2.9766, "step": 143110 }, { "epoch": 0.0051712, "grad_norm": 0.6536977291107178, "learning_rate": 1.409874269906746e-05, "loss": 2.943, "step": 143120 }, { "epoch": 0.0051968, "grad_norm": 0.7317422032356262, "learning_rate": 1.4098008157435035e-05, "loss": 3.1636, "step": 143130 }, { "epoch": 0.0052224, "grad_norm": 0.8957687616348267, "learning_rate": 1.4097273589228113e-05, "loss": 2.738, "step": 143140 }, { "epoch": 0.005248, "grad_norm": 0.754253625869751, "learning_rate": 1.4096538994451463e-05, "loss": 3.0953, "step": 143150 }, { "epoch": 0.0052736, "grad_norm": 0.7709265947341919, "learning_rate": 1.4095804373109847e-05, "loss": 3.0016, "step": 143160 }, { "epoch": 0.0052992, "grad_norm": 0.6545546650886536, "learning_rate": 1.4095069725208025e-05, "loss": 2.9578, "step": 143170 }, { "epoch": 0.0053248, "grad_norm": 0.7597678899765015, "learning_rate": 1.4094335050750767e-05, "loss": 3.1105, "step": 143180 }, { "epoch": 0.0053504, "grad_norm": 0.7477130889892578, "learning_rate": 1.4093600349742835e-05, "loss": 3.1632, "step": 143190 }, { "epoch": 0.005376, "grad_norm": 0.6919402480125427, "learning_rate": 1.4092865622188991e-05, "loss": 3.022, "step": 143200 }, { "epoch": 0.0054016, "grad_norm": 0.7094756960868835, "learning_rate": 1.4092130868094e-05, "loss": 2.8769, "step": 143210 }, { "epoch": 0.0054272, "grad_norm": 0.845346987247467, "learning_rate": 1.409139608746263e-05, "loss": 3.0453, "step": 143220 }, { "epoch": 0.0054528, "grad_norm": 0.6972144842147827, "learning_rate": 1.4090661280299645e-05, "loss": 2.9483, "step": 143230 }, { "epoch": 0.0054784, "grad_norm": 0.745604932308197, "learning_rate": 1.4089926446609805e-05, "loss": 3.1372, "step": 143240 }, { "epoch": 0.005504, "grad_norm": 0.7046521306037903, "learning_rate": 1.4089191586397881e-05, "loss": 2.8102, "step": 143250 }, { "epoch": 0.0055296, "grad_norm": 0.6589446663856506, "learning_rate": 1.4088456699668637e-05, "loss": 2.9033, "step": 143260 }, { "epoch": 0.0055552, "grad_norm": 0.7441034317016602, "learning_rate": 1.4087721786426835e-05, "loss": 3.0265, "step": 143270 }, { "epoch": 0.0055808, "grad_norm": 0.816439151763916, "learning_rate": 1.4086986846677247e-05, "loss": 3.004, "step": 143280 }, { "epoch": 0.0056064, "grad_norm": 0.743890643119812, "learning_rate": 1.4086251880424635e-05, "loss": 2.9726, "step": 143290 }, { "epoch": 0.005632, "grad_norm": 0.7696336507797241, "learning_rate": 1.4085516887673764e-05, "loss": 3.0308, "step": 143300 }, { "epoch": 0.0056576, "grad_norm": 0.8305509090423584, "learning_rate": 1.4084781868429402e-05, "loss": 2.8394, "step": 143310 }, { "epoch": 0.0056832, "grad_norm": 0.7708373069763184, "learning_rate": 1.4084046822696315e-05, "loss": 2.8345, "step": 143320 }, { "epoch": 0.0057088, "grad_norm": 0.7417104840278625, "learning_rate": 1.4083311750479272e-05, "loss": 2.9256, "step": 143330 }, { "epoch": 0.0057344, "grad_norm": 0.6923410892486572, "learning_rate": 1.4082576651783036e-05, "loss": 3.0503, "step": 143340 }, { "epoch": 0.00576, "grad_norm": 0.7863471508026123, "learning_rate": 1.4081841526612376e-05, "loss": 2.867, "step": 143350 }, { "epoch": 0.0057856, "grad_norm": 0.8702065944671631, "learning_rate": 1.4081106374972058e-05, "loss": 3.0468, "step": 143360 }, { "epoch": 0.0058112, "grad_norm": 0.6783581972122192, "learning_rate": 1.408037119686685e-05, "loss": 2.8669, "step": 143370 }, { "epoch": 0.0058368, "grad_norm": 0.8048560619354248, "learning_rate": 1.4079635992301517e-05, "loss": 2.9446, "step": 143380 }, { "epoch": 0.0058624, "grad_norm": 0.7576024532318115, "learning_rate": 1.4078900761280831e-05, "loss": 2.9551, "step": 143390 }, { "epoch": 0.005888, "grad_norm": 0.70101398229599, "learning_rate": 1.4078165503809557e-05, "loss": 2.7384, "step": 143400 }, { "epoch": 0.0059136, "grad_norm": 0.7943427562713623, "learning_rate": 1.4077430219892462e-05, "loss": 2.9479, "step": 143410 }, { "epoch": 0.0059392, "grad_norm": 0.6740132570266724, "learning_rate": 1.4076694909534321e-05, "loss": 2.9117, "step": 143420 }, { "epoch": 0.0059648, "grad_norm": 0.7433702349662781, "learning_rate": 1.4075959572739892e-05, "loss": 3.0523, "step": 143430 }, { "epoch": 0.0059904, "grad_norm": 0.6466106176376343, "learning_rate": 1.407522420951395e-05, "loss": 2.9911, "step": 143440 }, { "epoch": 0.006016, "grad_norm": 0.8032292127609253, "learning_rate": 1.4074488819861265e-05, "loss": 3.0771, "step": 143450 }, { "epoch": 0.0060416, "grad_norm": 0.7970643043518066, "learning_rate": 1.4073753403786598e-05, "loss": 3.194, "step": 143460 }, { "epoch": 0.0060672, "grad_norm": 0.7849502563476562, "learning_rate": 1.4073017961294724e-05, "loss": 2.9614, "step": 143470 }, { "epoch": 0.0060928, "grad_norm": 0.7281391024589539, "learning_rate": 1.4072282492390411e-05, "loss": 2.905, "step": 143480 }, { "epoch": 0.0061184, "grad_norm": 0.7429071068763733, "learning_rate": 1.4071546997078432e-05, "loss": 3.1691, "step": 143490 }, { "epoch": 0.006144, "grad_norm": 0.7078205347061157, "learning_rate": 1.4070811475363546e-05, "loss": 2.9102, "step": 143500 }, { "epoch": 0.0061696, "grad_norm": 0.7091627717018127, "learning_rate": 1.4070075927250534e-05, "loss": 3.0486, "step": 143510 }, { "epoch": 0.0061952, "grad_norm": 0.704142689704895, "learning_rate": 1.4069340352744162e-05, "loss": 3.037, "step": 143520 }, { "epoch": 0.0062208, "grad_norm": 0.6800131797790527, "learning_rate": 1.4068604751849197e-05, "loss": 3.1308, "step": 143530 }, { "epoch": 0.0062464, "grad_norm": 0.7289027571678162, "learning_rate": 1.4067869124570413e-05, "loss": 2.8537, "step": 143540 }, { "epoch": 0.006272, "grad_norm": 0.8869306445121765, "learning_rate": 1.406713347091258e-05, "loss": 2.9458, "step": 143550 }, { "epoch": 0.0062976, "grad_norm": 0.7831816077232361, "learning_rate": 1.4066397790880467e-05, "loss": 3.0383, "step": 143560 }, { "epoch": 0.0063232, "grad_norm": 0.8195732831954956, "learning_rate": 1.4065662084478842e-05, "loss": 2.9986, "step": 143570 }, { "epoch": 0.0063488, "grad_norm": 0.6693297624588013, "learning_rate": 1.406492635171248e-05, "loss": 2.7875, "step": 143580 }, { "epoch": 0.0063744, "grad_norm": 0.8151286840438843, "learning_rate": 1.4064190592586152e-05, "loss": 3.1054, "step": 143590 }, { "epoch": 0.0064, "grad_norm": 0.7241726517677307, "learning_rate": 1.4063454807104627e-05, "loss": 3.0382, "step": 143600 }, { "epoch": 0.0064256, "grad_norm": 0.8193961381912231, "learning_rate": 1.4062718995272677e-05, "loss": 2.9291, "step": 143610 }, { "epoch": 0.0064512, "grad_norm": 0.7401162385940552, "learning_rate": 1.4061983157095078e-05, "loss": 2.9502, "step": 143620 }, { "epoch": 0.0064768, "grad_norm": 0.6817061901092529, "learning_rate": 1.4061247292576595e-05, "loss": 2.8241, "step": 143630 }, { "epoch": 0.0065024, "grad_norm": 0.7818467617034912, "learning_rate": 1.4060511401722001e-05, "loss": 3.0116, "step": 143640 }, { "epoch": 0.006528, "grad_norm": 0.7543520331382751, "learning_rate": 1.4059775484536074e-05, "loss": 3.0239, "step": 143650 }, { "epoch": 0.0065536, "grad_norm": 0.8465763330459595, "learning_rate": 1.4059039541023578e-05, "loss": 3.1895, "step": 143660 }, { "epoch": 0.0065792, "grad_norm": 0.7181316614151001, "learning_rate": 1.4058303571189289e-05, "loss": 3.0716, "step": 143670 }, { "epoch": 0.0066048, "grad_norm": 0.7429556250572205, "learning_rate": 1.4057567575037983e-05, "loss": 3.0453, "step": 143680 }, { "epoch": 0.0066304, "grad_norm": 0.7865092158317566, "learning_rate": 1.4056831552574429e-05, "loss": 2.8145, "step": 143690 }, { "epoch": 0.006656, "grad_norm": 0.7080941200256348, "learning_rate": 1.4056095503803398e-05, "loss": 2.9245, "step": 143700 }, { "epoch": 0.0066816, "grad_norm": 0.7080429196357727, "learning_rate": 1.4055359428729667e-05, "loss": 3.0571, "step": 143710 }, { "epoch": 0.0067072, "grad_norm": 0.6985126733779907, "learning_rate": 1.4054623327358011e-05, "loss": 3.0369, "step": 143720 }, { "epoch": 0.0067328, "grad_norm": 1.0168430805206299, "learning_rate": 1.4053887199693197e-05, "loss": 3.1853, "step": 143730 }, { "epoch": 0.0067584, "grad_norm": 0.6978727579116821, "learning_rate": 1.4053151045740002e-05, "loss": 2.9787, "step": 143740 }, { "epoch": 0.006784, "grad_norm": 0.6924164295196533, "learning_rate": 1.4052414865503201e-05, "loss": 2.9864, "step": 143750 }, { "epoch": 0.0068096, "grad_norm": 0.9397156238555908, "learning_rate": 1.4051678658987565e-05, "loss": 2.9147, "step": 143760 }, { "epoch": 0.0068352, "grad_norm": 0.8686349391937256, "learning_rate": 1.405094242619787e-05, "loss": 3.1699, "step": 143770 }, { "epoch": 0.0068608, "grad_norm": 0.6713274121284485, "learning_rate": 1.4050206167138892e-05, "loss": 2.9143, "step": 143780 }, { "epoch": 0.0068864, "grad_norm": 0.7615916728973389, "learning_rate": 1.40494698818154e-05, "loss": 2.8533, "step": 143790 }, { "epoch": 0.006912, "grad_norm": 2.1831815242767334, "learning_rate": 1.4048733570232174e-05, "loss": 2.8658, "step": 143800 }, { "epoch": 0.0069376, "grad_norm": 0.7486753463745117, "learning_rate": 1.4047997232393986e-05, "loss": 3.0328, "step": 143810 }, { "epoch": 0.0069632, "grad_norm": 0.7121672630310059, "learning_rate": 1.4047260868305615e-05, "loss": 2.8755, "step": 143820 }, { "epoch": 0.0069888, "grad_norm": 0.7429726719856262, "learning_rate": 1.404652447797183e-05, "loss": 2.8518, "step": 143830 }, { "epoch": 0.0070144, "grad_norm": 1.0019724369049072, "learning_rate": 1.4045788061397408e-05, "loss": 3.2468, "step": 143840 }, { "epoch": 0.00704, "grad_norm": 0.7004295587539673, "learning_rate": 1.404505161858713e-05, "loss": 2.8686, "step": 143850 }, { "epoch": 0.0070656, "grad_norm": 0.8877153396606445, "learning_rate": 1.4044315149545762e-05, "loss": 2.9556, "step": 143860 }, { "epoch": 0.0070912, "grad_norm": 0.7175477147102356, "learning_rate": 1.4043578654278086e-05, "loss": 3.0748, "step": 143870 }, { "epoch": 0.0071168, "grad_norm": 0.6895048022270203, "learning_rate": 1.4042842132788883e-05, "loss": 3.268, "step": 143880 }, { "epoch": 0.0071424, "grad_norm": 0.7123662233352661, "learning_rate": 1.4042105585082917e-05, "loss": 3.0658, "step": 143890 }, { "epoch": 0.007168, "grad_norm": 0.6947184205055237, "learning_rate": 1.4041369011164974e-05, "loss": 3.0061, "step": 143900 }, { "epoch": 0.0071936, "grad_norm": 0.7715670466423035, "learning_rate": 1.4040632411039826e-05, "loss": 2.9701, "step": 143910 }, { "epoch": 0.0072192, "grad_norm": 0.7855663895606995, "learning_rate": 1.403989578471225e-05, "loss": 3.2666, "step": 143920 }, { "epoch": 0.0072448, "grad_norm": 0.7140313386917114, "learning_rate": 1.4039159132187022e-05, "loss": 3.0358, "step": 143930 }, { "epoch": 0.0072704, "grad_norm": 0.8466557264328003, "learning_rate": 1.4038422453468924e-05, "loss": 3.0071, "step": 143940 }, { "epoch": 0.007296, "grad_norm": 0.7256550788879395, "learning_rate": 1.4037685748562728e-05, "loss": 2.8266, "step": 143950 }, { "epoch": 0.0073216, "grad_norm": 0.7727591395378113, "learning_rate": 1.403694901747321e-05, "loss": 2.9509, "step": 143960 }, { "epoch": 0.0073472, "grad_norm": 0.823502779006958, "learning_rate": 1.4036212260205156e-05, "loss": 3.0076, "step": 143970 }, { "epoch": 0.0073728, "grad_norm": 0.717633843421936, "learning_rate": 1.4035475476763336e-05, "loss": 3.0619, "step": 143980 }, { "epoch": 0.0073984, "grad_norm": 0.6676013469696045, "learning_rate": 1.403473866715253e-05, "loss": 2.7637, "step": 143990 }, { "epoch": 0.007424, "grad_norm": 0.7450512051582336, "learning_rate": 1.4034001831377514e-05, "loss": 2.8929, "step": 144000 }, { "epoch": 0.0074496, "grad_norm": 0.7285844087600708, "learning_rate": 1.403326496944307e-05, "loss": 3.0903, "step": 144010 }, { "epoch": 0.0074752, "grad_norm": 0.7201297283172607, "learning_rate": 1.4032528081353976e-05, "loss": 3.2374, "step": 144020 }, { "epoch": 0.0075008, "grad_norm": 0.6950849294662476, "learning_rate": 1.4031791167115008e-05, "loss": 3.0243, "step": 144030 }, { "epoch": 0.0075264, "grad_norm": 0.8207499384880066, "learning_rate": 1.4031054226730945e-05, "loss": 3.193, "step": 144040 }, { "epoch": 0.007552, "grad_norm": 0.6735447645187378, "learning_rate": 1.4030317260206568e-05, "loss": 2.9093, "step": 144050 }, { "epoch": 0.0075776, "grad_norm": 0.7372346520423889, "learning_rate": 1.4029580267546651e-05, "loss": 2.9384, "step": 144060 }, { "epoch": 0.0076032, "grad_norm": 0.6868738532066345, "learning_rate": 1.4028843248755982e-05, "loss": 2.9482, "step": 144070 }, { "epoch": 0.0076288, "grad_norm": 0.7026614546775818, "learning_rate": 1.4028106203839333e-05, "loss": 3.1576, "step": 144080 }, { "epoch": 0.0076544, "grad_norm": 0.736286461353302, "learning_rate": 1.4027369132801487e-05, "loss": 2.8917, "step": 144090 }, { "epoch": 0.00768, "grad_norm": 0.7049217820167542, "learning_rate": 1.402663203564722e-05, "loss": 3.3442, "step": 144100 }, { "epoch": 0.0077056, "grad_norm": 0.6560079455375671, "learning_rate": 1.4025894912381315e-05, "loss": 3.0575, "step": 144110 }, { "epoch": 0.0077312, "grad_norm": 0.8596277832984924, "learning_rate": 1.4025157763008553e-05, "loss": 2.9995, "step": 144120 }, { "epoch": 0.0077568, "grad_norm": 0.7526292204856873, "learning_rate": 1.4024420587533711e-05, "loss": 3.2465, "step": 144130 }, { "epoch": 0.0077824, "grad_norm": 0.8736172914505005, "learning_rate": 1.4023683385961575e-05, "loss": 3.1806, "step": 144140 }, { "epoch": 0.007808, "grad_norm": 0.7854641079902649, "learning_rate": 1.4022946158296917e-05, "loss": 3.0064, "step": 144150 }, { "epoch": 0.0078336, "grad_norm": 0.9194599986076355, "learning_rate": 1.4022208904544524e-05, "loss": 2.9286, "step": 144160 }, { "epoch": 0.0078592, "grad_norm": 0.7204818725585938, "learning_rate": 1.4021471624709176e-05, "loss": 3.0736, "step": 144170 }, { "epoch": 0.0078848, "grad_norm": 0.8177036643028259, "learning_rate": 1.4020734318795652e-05, "loss": 3.0041, "step": 144180 }, { "epoch": 0.0079104, "grad_norm": 0.6972143650054932, "learning_rate": 1.4019996986808732e-05, "loss": 2.9656, "step": 144190 }, { "epoch": 0.007936, "grad_norm": 0.7367233633995056, "learning_rate": 1.4019259628753203e-05, "loss": 3.1869, "step": 144200 }, { "epoch": 0.0079616, "grad_norm": 0.7905002236366272, "learning_rate": 1.4018522244633845e-05, "loss": 3.0744, "step": 144210 }, { "epoch": 0.0079872, "grad_norm": 0.7635726928710938, "learning_rate": 1.4017784834455437e-05, "loss": 2.9478, "step": 144220 }, { "epoch": 0.0080128, "grad_norm": 0.7082076072692871, "learning_rate": 1.401704739822276e-05, "loss": 2.8699, "step": 144230 }, { "epoch": 0.0080384, "grad_norm": 0.6796163320541382, "learning_rate": 1.4016309935940601e-05, "loss": 2.9466, "step": 144240 }, { "epoch": 0.008064, "grad_norm": 0.967732846736908, "learning_rate": 1.401557244761374e-05, "loss": 3.0316, "step": 144250 }, { "epoch": 0.0080896, "grad_norm": 0.6454083919525146, "learning_rate": 1.4014834933246957e-05, "loss": 3.1957, "step": 144260 }, { "epoch": 0.0081152, "grad_norm": 2.0258138179779053, "learning_rate": 1.4014097392845037e-05, "loss": 3.072, "step": 144270 }, { "epoch": 0.0081408, "grad_norm": 0.7814035415649414, "learning_rate": 1.401335982641276e-05, "loss": 2.9061, "step": 144280 }, { "epoch": 0.0081664, "grad_norm": 0.7288528680801392, "learning_rate": 1.4012622233954914e-05, "loss": 3.0328, "step": 144290 }, { "epoch": 0.008192, "grad_norm": 0.7700785994529724, "learning_rate": 1.4011884615476276e-05, "loss": 2.947, "step": 144300 }, { "epoch": 0.0082176, "grad_norm": 0.6892662644386292, "learning_rate": 1.4011146970981638e-05, "loss": 3.0115, "step": 144310 }, { "epoch": 0.0082432, "grad_norm": 2.5627260208129883, "learning_rate": 1.4010409300475773e-05, "loss": 3.191, "step": 144320 }, { "epoch": 0.0082688, "grad_norm": 0.731621503829956, "learning_rate": 1.4009671603963473e-05, "loss": 3.1239, "step": 144330 }, { "epoch": 0.0082944, "grad_norm": 0.6825868487358093, "learning_rate": 1.4008933881449518e-05, "loss": 2.9135, "step": 144340 }, { "epoch": 0.00832, "grad_norm": 0.734065055847168, "learning_rate": 1.400819613293869e-05, "loss": 3.0267, "step": 144350 }, { "epoch": 0.0083456, "grad_norm": 0.9220870733261108, "learning_rate": 1.4007458358435777e-05, "loss": 3.1762, "step": 144360 }, { "epoch": 0.0083712, "grad_norm": 0.7185685038566589, "learning_rate": 1.4006720557945558e-05, "loss": 2.876, "step": 144370 }, { "epoch": 0.0083968, "grad_norm": 0.8504905700683594, "learning_rate": 1.4005982731472825e-05, "loss": 3.2048, "step": 144380 }, { "epoch": 0.0084224, "grad_norm": 1.1158885955810547, "learning_rate": 1.4005244879022357e-05, "loss": 2.8831, "step": 144390 }, { "epoch": 0.008448, "grad_norm": 0.8249319791793823, "learning_rate": 1.400450700059894e-05, "loss": 3.0656, "step": 144400 }, { "epoch": 0.0084736, "grad_norm": 1.1955206394195557, "learning_rate": 1.4003769096207362e-05, "loss": 3.0968, "step": 144410 }, { "epoch": 0.0084992, "grad_norm": 0.7324955463409424, "learning_rate": 1.4003031165852403e-05, "loss": 3.1995, "step": 144420 }, { "epoch": 0.0085248, "grad_norm": 0.7263059020042419, "learning_rate": 1.4002293209538854e-05, "loss": 3.0513, "step": 144430 }, { "epoch": 0.0085504, "grad_norm": 0.7676312327384949, "learning_rate": 1.4001555227271495e-05, "loss": 3.0889, "step": 144440 }, { "epoch": 0.008576, "grad_norm": 0.6721764206886292, "learning_rate": 1.4000817219055112e-05, "loss": 2.9961, "step": 144450 }, { "epoch": 0.0086016, "grad_norm": 0.7288126349449158, "learning_rate": 1.4000079184894493e-05, "loss": 3.078, "step": 144460 }, { "epoch": 0.0086272, "grad_norm": 0.6950892806053162, "learning_rate": 1.3999341124794425e-05, "loss": 2.9876, "step": 144470 }, { "epoch": 0.0086528, "grad_norm": 0.7241907119750977, "learning_rate": 1.3998603038759692e-05, "loss": 3.0109, "step": 144480 }, { "epoch": 0.0086784, "grad_norm": 0.753043532371521, "learning_rate": 1.3997864926795082e-05, "loss": 2.9729, "step": 144490 }, { "epoch": 0.008704, "grad_norm": 1.1758612394332886, "learning_rate": 1.399712678890538e-05, "loss": 2.8527, "step": 144500 }, { "epoch": 0.0087296, "grad_norm": 0.8917356729507446, "learning_rate": 1.3996388625095376e-05, "loss": 2.9138, "step": 144510 }, { "epoch": 0.0087552, "grad_norm": 0.7691230773925781, "learning_rate": 1.399565043536985e-05, "loss": 2.8876, "step": 144520 }, { "epoch": 0.0087808, "grad_norm": 0.6827847361564636, "learning_rate": 1.3994912219733592e-05, "loss": 2.9893, "step": 144530 }, { "epoch": 0.0088064, "grad_norm": 1.3535739183425903, "learning_rate": 1.3994173978191391e-05, "loss": 3.3657, "step": 144540 }, { "epoch": 0.008832, "grad_norm": 0.6386354565620422, "learning_rate": 1.3993435710748036e-05, "loss": 2.8241, "step": 144550 }, { "epoch": 0.0088576, "grad_norm": 0.8636192083358765, "learning_rate": 1.3992697417408308e-05, "loss": 3.0602, "step": 144560 }, { "epoch": 0.0088832, "grad_norm": 0.692557692527771, "learning_rate": 1.3991959098177e-05, "loss": 2.8995, "step": 144570 }, { "epoch": 0.0089088, "grad_norm": 0.753525972366333, "learning_rate": 1.3991220753058901e-05, "loss": 2.9856, "step": 144580 }, { "epoch": 0.0089344, "grad_norm": 0.6678280234336853, "learning_rate": 1.3990482382058792e-05, "loss": 2.8804, "step": 144590 }, { "epoch": 0.00896, "grad_norm": 0.770169734954834, "learning_rate": 1.3989743985181465e-05, "loss": 3.0787, "step": 144600 }, { "epoch": 0.0089856, "grad_norm": 0.736725389957428, "learning_rate": 1.3989005562431715e-05, "loss": 2.9728, "step": 144610 }, { "epoch": 0.0090112, "grad_norm": 0.7687088847160339, "learning_rate": 1.3988267113814317e-05, "loss": 3.1542, "step": 144620 }, { "epoch": 0.0090368, "grad_norm": 0.7034822106361389, "learning_rate": 1.3987528639334068e-05, "loss": 2.9663, "step": 144630 }, { "epoch": 0.0090624, "grad_norm": 0.756801426410675, "learning_rate": 1.3986790138995758e-05, "loss": 3.0503, "step": 144640 }, { "epoch": 0.009088, "grad_norm": 0.8213220834732056, "learning_rate": 1.3986051612804172e-05, "loss": 3.1202, "step": 144650 }, { "epoch": 0.0091136, "grad_norm": 0.7030706405639648, "learning_rate": 1.3985313060764101e-05, "loss": 3.0458, "step": 144660 }, { "epoch": 0.0091392, "grad_norm": 0.7379530072212219, "learning_rate": 1.3984574482880336e-05, "loss": 3.2326, "step": 144670 }, { "epoch": 0.0091648, "grad_norm": 0.8186582922935486, "learning_rate": 1.3983835879157662e-05, "loss": 2.9546, "step": 144680 }, { "epoch": 0.0091904, "grad_norm": 0.7824438214302063, "learning_rate": 1.398309724960087e-05, "loss": 3.1758, "step": 144690 }, { "epoch": 0.009216, "grad_norm": 0.7182568907737732, "learning_rate": 1.3982358594214751e-05, "loss": 3.3199, "step": 144700 }, { "epoch": 0.0092416, "grad_norm": 0.8063502311706543, "learning_rate": 1.3981619913004097e-05, "loss": 3.1528, "step": 144710 }, { "epoch": 0.0092672, "grad_norm": 0.8537460565567017, "learning_rate": 1.3980881205973695e-05, "loss": 2.9059, "step": 144720 }, { "epoch": 0.0092928, "grad_norm": 0.8868890404701233, "learning_rate": 1.3980142473128335e-05, "loss": 2.837, "step": 144730 }, { "epoch": 0.0093184, "grad_norm": 1.5546252727508545, "learning_rate": 1.3979403714472813e-05, "loss": 2.5534, "step": 144740 }, { "epoch": 0.009344, "grad_norm": 0.9204314351081848, "learning_rate": 1.3978664930011912e-05, "loss": 3.0113, "step": 144750 }, { "epoch": 0.0093696, "grad_norm": 0.7459883093833923, "learning_rate": 1.3977926119750426e-05, "loss": 3.1436, "step": 144760 }, { "epoch": 0.0093952, "grad_norm": 0.859115719795227, "learning_rate": 1.397718728369315e-05, "loss": 3.05, "step": 144770 }, { "epoch": 0.0094208, "grad_norm": 0.8098176717758179, "learning_rate": 1.3976448421844868e-05, "loss": 3.1254, "step": 144780 }, { "epoch": 0.0094464, "grad_norm": 0.8697461485862732, "learning_rate": 1.3975709534210378e-05, "loss": 2.9898, "step": 144790 }, { "epoch": 0.009472, "grad_norm": 0.6967079043388367, "learning_rate": 1.3974970620794466e-05, "loss": 3.1528, "step": 144800 }, { "epoch": 0.0094976, "grad_norm": 0.6929473876953125, "learning_rate": 1.3974231681601924e-05, "loss": 2.8918, "step": 144810 }, { "epoch": 0.0095232, "grad_norm": 0.8081632852554321, "learning_rate": 1.3973492716637549e-05, "loss": 2.9947, "step": 144820 }, { "epoch": 0.0095488, "grad_norm": 2.113647222518921, "learning_rate": 1.3972753725906127e-05, "loss": 3.1833, "step": 144830 }, { "epoch": 0.0095744, "grad_norm": 0.8310999870300293, "learning_rate": 1.3972014709412456e-05, "loss": 3.0535, "step": 144840 }, { "epoch": 0.0096, "grad_norm": 0.8106647729873657, "learning_rate": 1.3971275667161323e-05, "loss": 3.0474, "step": 144850 }, { "epoch": 0.0096256, "grad_norm": 0.7560297250747681, "learning_rate": 1.3970536599157525e-05, "loss": 3.0184, "step": 144860 }, { "epoch": 0.0096512, "grad_norm": 0.8387734889984131, "learning_rate": 1.3969797505405854e-05, "loss": 3.0712, "step": 144870 }, { "epoch": 0.0096768, "grad_norm": 0.7258874177932739, "learning_rate": 1.3969058385911096e-05, "loss": 2.8993, "step": 144880 }, { "epoch": 0.0097024, "grad_norm": 0.6925274729728699, "learning_rate": 1.3968319240678049e-05, "loss": 2.9788, "step": 144890 }, { "epoch": 0.009728, "grad_norm": 0.7591167688369751, "learning_rate": 1.396758006971151e-05, "loss": 3.0719, "step": 144900 }, { "epoch": 0.0097536, "grad_norm": 0.6938175559043884, "learning_rate": 1.3966840873016267e-05, "loss": 3.003, "step": 144910 }, { "epoch": 0.0097792, "grad_norm": 0.7750793695449829, "learning_rate": 1.3966101650597117e-05, "loss": 2.9031, "step": 144920 }, { "epoch": 0.0098048, "grad_norm": 0.6774130463600159, "learning_rate": 1.3965362402458848e-05, "loss": 3.0643, "step": 144930 }, { "epoch": 0.0098304, "grad_norm": 0.7419723868370056, "learning_rate": 1.3964623128606262e-05, "loss": 3.1848, "step": 144940 }, { "epoch": 0.009856, "grad_norm": 0.7381253242492676, "learning_rate": 1.396388382904415e-05, "loss": 2.8902, "step": 144950 }, { "epoch": 0.0098816, "grad_norm": 1.19251549243927, "learning_rate": 1.39631445037773e-05, "loss": 2.9689, "step": 144960 }, { "epoch": 0.0099072, "grad_norm": 0.7051481008529663, "learning_rate": 1.3962405152810514e-05, "loss": 3.0953, "step": 144970 }, { "epoch": 0.0099328, "grad_norm": 0.6745584607124329, "learning_rate": 1.3961665776148581e-05, "loss": 2.954, "step": 144980 }, { "epoch": 0.0099584, "grad_norm": 0.6619470715522766, "learning_rate": 1.39609263737963e-05, "loss": 2.9633, "step": 144990 }, { "epoch": 0.009984, "grad_norm": 2.094655990600586, "learning_rate": 1.3960186945758465e-05, "loss": 3.1549, "step": 145000 }, { "epoch": 0.0100096, "grad_norm": 0.6704065799713135, "learning_rate": 1.3959447492039868e-05, "loss": 2.9248, "step": 145010 }, { "epoch": 0.0100352, "grad_norm": 0.7041771411895752, "learning_rate": 1.3958708012645308e-05, "loss": 2.9533, "step": 145020 }, { "epoch": 0.0100608, "grad_norm": 0.6970328092575073, "learning_rate": 1.3957968507579577e-05, "loss": 3.0607, "step": 145030 }, { "epoch": 0.0100864, "grad_norm": 0.7038511633872986, "learning_rate": 1.3957228976847474e-05, "loss": 3.0896, "step": 145040 }, { "epoch": 0.010112, "grad_norm": 0.8368933200836182, "learning_rate": 1.3956489420453793e-05, "loss": 3.0135, "step": 145050 }, { "epoch": 0.0101376, "grad_norm": 0.7368367910385132, "learning_rate": 1.3955749838403328e-05, "loss": 3.0847, "step": 145060 }, { "epoch": 0.0101632, "grad_norm": 0.7367652058601379, "learning_rate": 1.3955010230700877e-05, "loss": 2.876, "step": 145070 }, { "epoch": 0.0101888, "grad_norm": 0.6840576529502869, "learning_rate": 1.3954270597351235e-05, "loss": 2.9867, "step": 145080 }, { "epoch": 0.0102144, "grad_norm": 0.7174962759017944, "learning_rate": 1.3953530938359201e-05, "loss": 3.0587, "step": 145090 }, { "epoch": 0.01024, "grad_norm": 0.7895544171333313, "learning_rate": 1.395279125372957e-05, "loss": 2.9276, "step": 145100 }, { "epoch": 0.0102656, "grad_norm": 0.9422256946563721, "learning_rate": 1.3952051543467138e-05, "loss": 3.0193, "step": 145110 }, { "epoch": 0.0102912, "grad_norm": 0.7664221525192261, "learning_rate": 1.3951311807576699e-05, "loss": 2.9552, "step": 145120 }, { "epoch": 0.0103168, "grad_norm": 0.7937126159667969, "learning_rate": 1.3950572046063058e-05, "loss": 3.0647, "step": 145130 }, { "epoch": 0.0103424, "grad_norm": 0.7881050705909729, "learning_rate": 1.3949832258931005e-05, "loss": 3.2148, "step": 145140 }, { "epoch": 0.010368, "grad_norm": 0.7934348583221436, "learning_rate": 1.3949092446185335e-05, "loss": 3.1868, "step": 145150 }, { "epoch": 0.0103936, "grad_norm": 0.6734258532524109, "learning_rate": 1.3948352607830855e-05, "loss": 2.9906, "step": 145160 }, { "epoch": 0.0104192, "grad_norm": 0.7283396124839783, "learning_rate": 1.3947612743872357e-05, "loss": 3.1097, "step": 145170 }, { "epoch": 0.0104448, "grad_norm": 0.8460701704025269, "learning_rate": 1.3946872854314639e-05, "loss": 2.9757, "step": 145180 }, { "epoch": 0.0104704, "grad_norm": 0.7282288670539856, "learning_rate": 1.3946132939162498e-05, "loss": 2.9892, "step": 145190 }, { "epoch": 0.010496, "grad_norm": 0.7551001310348511, "learning_rate": 1.3945392998420739e-05, "loss": 2.932, "step": 145200 }, { "epoch": 0.0105216, "grad_norm": 0.9452872276306152, "learning_rate": 1.3944653032094148e-05, "loss": 2.8926, "step": 145210 }, { "epoch": 0.0105472, "grad_norm": 1.2612309455871582, "learning_rate": 1.3943913040187535e-05, "loss": 3.1027, "step": 145220 }, { "epoch": 0.0105728, "grad_norm": 0.7189218401908875, "learning_rate": 1.3943173022705696e-05, "loss": 2.9257, "step": 145230 }, { "epoch": 0.0105984, "grad_norm": 0.7087399959564209, "learning_rate": 1.3942432979653422e-05, "loss": 3.0075, "step": 145240 }, { "epoch": 0.010624, "grad_norm": 0.8444010019302368, "learning_rate": 1.3941692911035518e-05, "loss": 2.9871, "step": 145250 }, { "epoch": 0.0106496, "grad_norm": 0.8721250891685486, "learning_rate": 1.394095281685679e-05, "loss": 2.9638, "step": 145260 }, { "epoch": 0.0106752, "grad_norm": 0.8099329471588135, "learning_rate": 1.3940212697122024e-05, "loss": 3.0434, "step": 145270 }, { "epoch": 0.0107008, "grad_norm": 0.6724770069122314, "learning_rate": 1.3939472551836028e-05, "loss": 2.9266, "step": 145280 }, { "epoch": 0.0107264, "grad_norm": 0.7497469782829285, "learning_rate": 1.3938732381003599e-05, "loss": 2.9597, "step": 145290 }, { "epoch": 0.010752, "grad_norm": 0.7112138271331787, "learning_rate": 1.393799218462954e-05, "loss": 2.8844, "step": 145300 }, { "epoch": 0.0107776, "grad_norm": 1.054614543914795, "learning_rate": 1.3937251962718646e-05, "loss": 3.0052, "step": 145310 }, { "epoch": 0.0108032, "grad_norm": 0.6988469362258911, "learning_rate": 1.3936511715275715e-05, "loss": 2.9105, "step": 145320 }, { "epoch": 0.0108288, "grad_norm": 0.761035680770874, "learning_rate": 1.3935771442305559e-05, "loss": 3.083, "step": 145330 }, { "epoch": 0.0108544, "grad_norm": 1.2305489778518677, "learning_rate": 1.3935031143812967e-05, "loss": 2.9942, "step": 145340 }, { "epoch": 0.01088, "grad_norm": 0.8025606274604797, "learning_rate": 1.3934290819802745e-05, "loss": 2.9675, "step": 145350 }, { "epoch": 0.0109056, "grad_norm": 0.695274829864502, "learning_rate": 1.3933550470279695e-05, "loss": 3.0377, "step": 145360 }, { "epoch": 0.0109312, "grad_norm": 0.7660502791404724, "learning_rate": 1.3932810095248611e-05, "loss": 3.2552, "step": 145370 }, { "epoch": 0.0109568, "grad_norm": 0.6751587390899658, "learning_rate": 1.3932069694714302e-05, "loss": 2.9904, "step": 145380 }, { "epoch": 0.0109824, "grad_norm": 0.7225292325019836, "learning_rate": 1.3931329268681566e-05, "loss": 3.0205, "step": 145390 }, { "epoch": 0.011008, "grad_norm": 0.8046196699142456, "learning_rate": 1.3930588817155204e-05, "loss": 2.9908, "step": 145400 }, { "epoch": 0.0110336, "grad_norm": 0.7507677674293518, "learning_rate": 1.3929848340140015e-05, "loss": 3.0972, "step": 145410 }, { "epoch": 0.0110592, "grad_norm": 0.7363468408584595, "learning_rate": 1.3929107837640808e-05, "loss": 3.1231, "step": 145420 }, { "epoch": 0.0110848, "grad_norm": 0.8602264523506165, "learning_rate": 1.392836730966238e-05, "loss": 2.922, "step": 145430 }, { "epoch": 0.0111104, "grad_norm": 0.8234158754348755, "learning_rate": 1.3927626756209534e-05, "loss": 3.0035, "step": 145440 }, { "epoch": 0.011136, "grad_norm": 0.8059276938438416, "learning_rate": 1.3926886177287071e-05, "loss": 3.1577, "step": 145450 }, { "epoch": 0.0111616, "grad_norm": 0.6598016619682312, "learning_rate": 1.3926145572899799e-05, "loss": 3.0569, "step": 145460 }, { "epoch": 0.0111872, "grad_norm": 0.70319002866745, "learning_rate": 1.3925404943052512e-05, "loss": 3.1327, "step": 145470 }, { "epoch": 0.0112128, "grad_norm": 0.7497625946998596, "learning_rate": 1.3924664287750018e-05, "loss": 3.0541, "step": 145480 }, { "epoch": 0.0112384, "grad_norm": 1.1792466640472412, "learning_rate": 1.3923923606997119e-05, "loss": 2.8143, "step": 145490 }, { "epoch": 0.011264, "grad_norm": 0.8667111992835999, "learning_rate": 1.392318290079862e-05, "loss": 3.0331, "step": 145500 }, { "epoch": 0.0112896, "grad_norm": 0.6667203307151794, "learning_rate": 1.392244216915932e-05, "loss": 2.9379, "step": 145510 }, { "epoch": 0.0113152, "grad_norm": 0.7462165355682373, "learning_rate": 1.3921701412084026e-05, "loss": 2.9848, "step": 145520 }, { "epoch": 0.0113408, "grad_norm": 0.7809635996818542, "learning_rate": 1.3920960629577543e-05, "loss": 3.2116, "step": 145530 }, { "epoch": 0.0113664, "grad_norm": 0.7127349972724915, "learning_rate": 1.3920219821644669e-05, "loss": 3.0159, "step": 145540 }, { "epoch": 0.011392, "grad_norm": 0.7281673550605774, "learning_rate": 1.3919478988290212e-05, "loss": 2.7941, "step": 145550 }, { "epoch": 0.0114176, "grad_norm": 0.710512101650238, "learning_rate": 1.3918738129518978e-05, "loss": 2.8299, "step": 145560 }, { "epoch": 0.0114432, "grad_norm": 0.8086144328117371, "learning_rate": 1.3917997245335766e-05, "loss": 3.1426, "step": 145570 }, { "epoch": 0.0114688, "grad_norm": 0.7013991475105286, "learning_rate": 1.391725633574538e-05, "loss": 3.0672, "step": 145580 }, { "epoch": 0.0114944, "grad_norm": 0.980908215045929, "learning_rate": 1.3916515400752635e-05, "loss": 2.9775, "step": 145590 }, { "epoch": 0.01152, "grad_norm": 0.9054400324821472, "learning_rate": 1.3915774440362326e-05, "loss": 2.9679, "step": 145600 }, { "epoch": 0.0115456, "grad_norm": 2.091383218765259, "learning_rate": 1.3915033454579259e-05, "loss": 2.8188, "step": 145610 }, { "epoch": 0.0115712, "grad_norm": 0.6852177381515503, "learning_rate": 1.391429244340824e-05, "loss": 3.1447, "step": 145620 }, { "epoch": 0.0115968, "grad_norm": 0.6566709280014038, "learning_rate": 1.3913551406854078e-05, "loss": 2.9145, "step": 145630 }, { "epoch": 0.0116224, "grad_norm": 0.8617481589317322, "learning_rate": 1.3912810344921573e-05, "loss": 2.9588, "step": 145640 }, { "epoch": 0.011648, "grad_norm": 0.6889113187789917, "learning_rate": 1.3912069257615532e-05, "loss": 3.0516, "step": 145650 }, { "epoch": 0.0116736, "grad_norm": 0.6974374651908875, "learning_rate": 1.3911328144940766e-05, "loss": 3.0662, "step": 145660 }, { "epoch": 0.0116992, "grad_norm": 0.7210676670074463, "learning_rate": 1.391058700690207e-05, "loss": 3.0291, "step": 145670 }, { "epoch": 0.0117248, "grad_norm": 0.9411457777023315, "learning_rate": 1.3909845843504259e-05, "loss": 3.163, "step": 145680 }, { "epoch": 0.0117504, "grad_norm": 0.7282448410987854, "learning_rate": 1.390910465475214e-05, "loss": 2.9905, "step": 145690 }, { "epoch": 0.011776, "grad_norm": 0.724960446357727, "learning_rate": 1.3908363440650511e-05, "loss": 3.0406, "step": 145700 }, { "epoch": 0.0118016, "grad_norm": 0.8038974404335022, "learning_rate": 1.3907622201204187e-05, "loss": 2.9122, "step": 145710 }, { "epoch": 0.0118272, "grad_norm": 1.4834791421890259, "learning_rate": 1.390688093641797e-05, "loss": 3.3683, "step": 145720 }, { "epoch": 0.0118528, "grad_norm": 0.7804247736930847, "learning_rate": 1.3906139646296673e-05, "loss": 2.9711, "step": 145730 }, { "epoch": 0.0118784, "grad_norm": 0.7430004477500916, "learning_rate": 1.3905398330845094e-05, "loss": 2.9674, "step": 145740 }, { "epoch": 0.011904, "grad_norm": 0.8943848013877869, "learning_rate": 1.3904656990068046e-05, "loss": 2.918, "step": 145750 }, { "epoch": 0.0119296, "grad_norm": 0.7383816838264465, "learning_rate": 1.3903915623970334e-05, "loss": 3.0206, "step": 145760 }, { "epoch": 0.0119552, "grad_norm": 0.8437849879264832, "learning_rate": 1.3903174232556768e-05, "loss": 3.0246, "step": 145770 }, { "epoch": 0.0119808, "grad_norm": 0.7708964943885803, "learning_rate": 1.3902432815832155e-05, "loss": 3.0259, "step": 145780 }, { "epoch": 0.0120064, "grad_norm": 0.9375954866409302, "learning_rate": 1.3901691373801301e-05, "loss": 2.9387, "step": 145790 }, { "epoch": 0.012032, "grad_norm": 0.7730780243873596, "learning_rate": 1.3900949906469014e-05, "loss": 3.048, "step": 145800 }, { "epoch": 0.0120576, "grad_norm": 0.6942618489265442, "learning_rate": 1.3900208413840106e-05, "loss": 3.1255, "step": 145810 }, { "epoch": 0.0120832, "grad_norm": 0.7629736065864563, "learning_rate": 1.3899466895919383e-05, "loss": 2.839, "step": 145820 }, { "epoch": 0.0121088, "grad_norm": 0.6810106039047241, "learning_rate": 1.3898725352711654e-05, "loss": 3.0478, "step": 145830 }, { "epoch": 0.0121344, "grad_norm": 0.8246899247169495, "learning_rate": 1.3897983784221727e-05, "loss": 3.142, "step": 145840 }, { "epoch": 0.01216, "grad_norm": 0.6936265826225281, "learning_rate": 1.3897242190454409e-05, "loss": 2.8964, "step": 145850 }, { "epoch": 0.0121856, "grad_norm": 0.8723527789115906, "learning_rate": 1.3896500571414513e-05, "loss": 3.0358, "step": 145860 }, { "epoch": 0.0122112, "grad_norm": 2.223757266998291, "learning_rate": 1.3895758927106845e-05, "loss": 2.7965, "step": 145870 }, { "epoch": 0.0122368, "grad_norm": 0.8618001341819763, "learning_rate": 1.3895017257536217e-05, "loss": 3.1485, "step": 145880 }, { "epoch": 0.0122624, "grad_norm": 0.6972246170043945, "learning_rate": 1.3894275562707438e-05, "loss": 3.0111, "step": 145890 }, { "epoch": 0.012288, "grad_norm": 0.8940045833587646, "learning_rate": 1.3893533842625315e-05, "loss": 3.0501, "step": 145900 }, { "epoch": 0.0123136, "grad_norm": 0.8088224530220032, "learning_rate": 1.3892792097294661e-05, "loss": 2.8881, "step": 145910 }, { "epoch": 0.0123392, "grad_norm": 0.8288060426712036, "learning_rate": 1.3892050326720283e-05, "loss": 3.0547, "step": 145920 }, { "epoch": 0.0123648, "grad_norm": 0.7268944978713989, "learning_rate": 1.3891308530906992e-05, "loss": 3.0991, "step": 145930 }, { "epoch": 0.0123904, "grad_norm": 0.8297809958457947, "learning_rate": 1.3890566709859603e-05, "loss": 3.2223, "step": 145940 }, { "epoch": 0.012416, "grad_norm": 0.866277277469635, "learning_rate": 1.3889824863582923e-05, "loss": 3.1081, "step": 145950 }, { "epoch": 0.0124416, "grad_norm": 0.6405873894691467, "learning_rate": 1.3889082992081762e-05, "loss": 2.9927, "step": 145960 }, { "epoch": 0.0124672, "grad_norm": 0.7783883213996887, "learning_rate": 1.3888341095360929e-05, "loss": 3.0922, "step": 145970 }, { "epoch": 0.0124928, "grad_norm": 0.8240606784820557, "learning_rate": 1.3887599173425239e-05, "loss": 3.1112, "step": 145980 }, { "epoch": 0.0125184, "grad_norm": 0.8167242407798767, "learning_rate": 1.3886857226279502e-05, "loss": 3.1696, "step": 145990 }, { "epoch": 0.012544, "grad_norm": 0.6814967393875122, "learning_rate": 1.388611525392853e-05, "loss": 2.8604, "step": 146000 }, { "epoch": 0.0125696, "grad_norm": 0.8069115877151489, "learning_rate": 1.3885373256377131e-05, "loss": 2.9573, "step": 146010 }, { "epoch": 0.0125952, "grad_norm": 0.8596985340118408, "learning_rate": 1.3884631233630119e-05, "loss": 2.9455, "step": 146020 }, { "epoch": 0.0126208, "grad_norm": 1.2566757202148438, "learning_rate": 1.3883889185692305e-05, "loss": 2.8639, "step": 146030 }, { "epoch": 0.0126464, "grad_norm": 0.7663188576698303, "learning_rate": 1.3883147112568505e-05, "loss": 3.0519, "step": 146040 }, { "epoch": 0.012672, "grad_norm": 0.6591998934745789, "learning_rate": 1.3882405014263526e-05, "loss": 2.9913, "step": 146050 }, { "epoch": 0.0126976, "grad_norm": 0.7221236228942871, "learning_rate": 1.3881662890782184e-05, "loss": 3.1243, "step": 146060 }, { "epoch": 0.0127232, "grad_norm": 0.8651473522186279, "learning_rate": 1.388092074212929e-05, "loss": 3.1495, "step": 146070 }, { "epoch": 0.0127488, "grad_norm": 0.6743771433830261, "learning_rate": 1.3880178568309654e-05, "loss": 3.126, "step": 146080 }, { "epoch": 0.0127744, "grad_norm": 0.7726635336875916, "learning_rate": 1.3879436369328093e-05, "loss": 3.0894, "step": 146090 }, { "epoch": 0.0128, "grad_norm": 0.6868703961372375, "learning_rate": 1.3878694145189418e-05, "loss": 3.0542, "step": 146100 }, { "epoch": 0.0128256, "grad_norm": 0.6100443601608276, "learning_rate": 1.3877951895898438e-05, "loss": 3.0017, "step": 146110 }, { "epoch": 0.0128512, "grad_norm": 0.7753400802612305, "learning_rate": 1.3877209621459977e-05, "loss": 3.0534, "step": 146120 }, { "epoch": 0.0128768, "grad_norm": 0.8853915929794312, "learning_rate": 1.3876467321878839e-05, "loss": 3.0386, "step": 146130 }, { "epoch": 0.0129024, "grad_norm": 0.8211409449577332, "learning_rate": 1.3875724997159841e-05, "loss": 3.0954, "step": 146140 }, { "epoch": 0.012928, "grad_norm": 0.752967894077301, "learning_rate": 1.3874982647307797e-05, "loss": 3.0405, "step": 146150 }, { "epoch": 0.0129536, "grad_norm": 0.8045150637626648, "learning_rate": 1.387424027232752e-05, "loss": 3.029, "step": 146160 }, { "epoch": 0.0129792, "grad_norm": 0.7088146805763245, "learning_rate": 1.3873497872223824e-05, "loss": 3.0574, "step": 146170 }, { "epoch": 0.0130048, "grad_norm": 0.8684143424034119, "learning_rate": 1.3872755447001525e-05, "loss": 3.1, "step": 146180 }, { "epoch": 0.0130304, "grad_norm": 0.8691788911819458, "learning_rate": 1.3872012996665434e-05, "loss": 3.3117, "step": 146190 }, { "epoch": 0.013056, "grad_norm": 0.8387829065322876, "learning_rate": 1.3871270521220367e-05, "loss": 2.8893, "step": 146200 }, { "epoch": 0.0130816, "grad_norm": 0.7421495318412781, "learning_rate": 1.3870528020671143e-05, "loss": 2.9841, "step": 146210 }, { "epoch": 0.0131072, "grad_norm": 0.9832843542098999, "learning_rate": 1.3869785495022571e-05, "loss": 3.057, "step": 146220 }, { "epoch": 0.0131328, "grad_norm": 0.8131888508796692, "learning_rate": 1.3869042944279472e-05, "loss": 3.042, "step": 146230 }, { "epoch": 0.0131584, "grad_norm": 0.7826391458511353, "learning_rate": 1.3868300368446654e-05, "loss": 3.2552, "step": 146240 }, { "epoch": 0.013184, "grad_norm": 0.8393590450286865, "learning_rate": 1.3867557767528939e-05, "loss": 3.0141, "step": 146250 }, { "epoch": 0.0132096, "grad_norm": 1.2026182413101196, "learning_rate": 1.3866815141531138e-05, "loss": 3.6859, "step": 146260 }, { "epoch": 0.0132352, "grad_norm": 0.8067667484283447, "learning_rate": 1.386607249045807e-05, "loss": 3.0159, "step": 146270 }, { "epoch": 0.0132608, "grad_norm": 0.7404128909111023, "learning_rate": 1.3865329814314545e-05, "loss": 2.9589, "step": 146280 }, { "epoch": 0.0132864, "grad_norm": 0.8165432214736938, "learning_rate": 1.3864587113105387e-05, "loss": 3.0293, "step": 146290 }, { "epoch": 0.013312, "grad_norm": 0.7613934874534607, "learning_rate": 1.3863844386835408e-05, "loss": 3.0623, "step": 146300 }, { "epoch": 0.0133376, "grad_norm": 0.8027778267860413, "learning_rate": 1.3863101635509426e-05, "loss": 2.9566, "step": 146310 }, { "epoch": 0.0133632, "grad_norm": 0.6749786138534546, "learning_rate": 1.3862358859132257e-05, "loss": 2.7832, "step": 146320 }, { "epoch": 0.0133888, "grad_norm": 1.0216357707977295, "learning_rate": 1.3861616057708716e-05, "loss": 2.6738, "step": 146330 }, { "epoch": 0.0134144, "grad_norm": 0.858916163444519, "learning_rate": 1.3860873231243619e-05, "loss": 2.9912, "step": 146340 }, { "epoch": 0.01344, "grad_norm": 0.7383489608764648, "learning_rate": 1.3860130379741789e-05, "loss": 3.317, "step": 146350 }, { "epoch": 0.0134656, "grad_norm": 0.7436265349388123, "learning_rate": 1.385938750320804e-05, "loss": 3.0661, "step": 146360 }, { "epoch": 0.0134912, "grad_norm": 0.7412686944007874, "learning_rate": 1.3858644601647182e-05, "loss": 2.9341, "step": 146370 }, { "epoch": 0.0135168, "grad_norm": 0.8343189358711243, "learning_rate": 1.3857901675064047e-05, "loss": 3.0337, "step": 146380 }, { "epoch": 0.0135424, "grad_norm": 0.7766076922416687, "learning_rate": 1.385715872346344e-05, "loss": 3.0716, "step": 146390 }, { "epoch": 0.013568, "grad_norm": 1.2034443616867065, "learning_rate": 1.3856415746850184e-05, "loss": 2.9317, "step": 146400 }, { "epoch": 0.0135936, "grad_norm": 1.1930532455444336, "learning_rate": 1.3855672745229101e-05, "loss": 3.0145, "step": 146410 }, { "epoch": 0.0136192, "grad_norm": 0.8767245411872864, "learning_rate": 1.3854929718605003e-05, "loss": 3.183, "step": 146420 }, { "epoch": 0.0136448, "grad_norm": 0.709552526473999, "learning_rate": 1.385418666698271e-05, "loss": 2.9982, "step": 146430 }, { "epoch": 0.0136704, "grad_norm": 1.6290082931518555, "learning_rate": 1.3853443590367039e-05, "loss": 2.9104, "step": 146440 }, { "epoch": 0.013696, "grad_norm": 0.784636914730072, "learning_rate": 1.3852700488762812e-05, "loss": 3.0297, "step": 146450 }, { "epoch": 0.0137216, "grad_norm": 0.9189484715461731, "learning_rate": 1.3851957362174845e-05, "loss": 3.03, "step": 146460 }, { "epoch": 0.0137472, "grad_norm": 0.7537113428115845, "learning_rate": 1.385121421060796e-05, "loss": 3.1394, "step": 146470 }, { "epoch": 0.0137728, "grad_norm": 0.6983779668807983, "learning_rate": 1.3850471034066975e-05, "loss": 3.0721, "step": 146480 }, { "epoch": 0.0137984, "grad_norm": 0.6917396187782288, "learning_rate": 1.3849727832556708e-05, "loss": 3.1846, "step": 146490 }, { "epoch": 0.013824, "grad_norm": 1.0908293724060059, "learning_rate": 1.3848984606081976e-05, "loss": 3.0594, "step": 146500 }, { "epoch": 0.0138496, "grad_norm": 0.7571679353713989, "learning_rate": 1.3848241354647604e-05, "loss": 3.0319, "step": 146510 }, { "epoch": 0.0138752, "grad_norm": 0.6986050009727478, "learning_rate": 1.3847498078258413e-05, "loss": 2.9728, "step": 146520 }, { "epoch": 0.0139008, "grad_norm": 0.6998432874679565, "learning_rate": 1.3846754776919216e-05, "loss": 3.1344, "step": 146530 }, { "epoch": 0.0139264, "grad_norm": 0.7536824941635132, "learning_rate": 1.3846011450634836e-05, "loss": 2.7567, "step": 146540 }, { "epoch": 0.013952, "grad_norm": 0.6836870908737183, "learning_rate": 1.3845268099410094e-05, "loss": 2.872, "step": 146550 }, { "epoch": 0.0139776, "grad_norm": 0.7920876145362854, "learning_rate": 1.3844524723249812e-05, "loss": 3.1157, "step": 146560 }, { "epoch": 0.0140032, "grad_norm": 0.7591642141342163, "learning_rate": 1.3843781322158806e-05, "loss": 2.9542, "step": 146570 }, { "epoch": 0.0140288, "grad_norm": 0.6792162656784058, "learning_rate": 1.3843037896141904e-05, "loss": 3.0637, "step": 146580 }, { "epoch": 0.0140544, "grad_norm": 0.7594730257987976, "learning_rate": 1.384229444520392e-05, "loss": 2.8216, "step": 146590 }, { "epoch": 0.01408, "grad_norm": 0.8148588538169861, "learning_rate": 1.384155096934968e-05, "loss": 3.0249, "step": 146600 }, { "epoch": 0.0141056, "grad_norm": 0.7218571305274963, "learning_rate": 1.3840807468583999e-05, "loss": 2.9773, "step": 146610 }, { "epoch": 0.0141312, "grad_norm": 0.7583289742469788, "learning_rate": 1.3840063942911705e-05, "loss": 2.7547, "step": 146620 }, { "epoch": 0.0141568, "grad_norm": 0.7308095693588257, "learning_rate": 1.3839320392337614e-05, "loss": 2.7474, "step": 146630 }, { "epoch": 0.0141824, "grad_norm": 0.8816370368003845, "learning_rate": 1.3838576816866556e-05, "loss": 3.1191, "step": 146640 }, { "epoch": 0.014208, "grad_norm": 0.8181881308555603, "learning_rate": 1.3837833216503343e-05, "loss": 3.3689, "step": 146650 }, { "epoch": 0.0142336, "grad_norm": 0.7182026505470276, "learning_rate": 1.3837089591252804e-05, "loss": 2.8397, "step": 146660 }, { "epoch": 0.0142592, "grad_norm": 0.6619818210601807, "learning_rate": 1.3836345941119756e-05, "loss": 2.9516, "step": 146670 }, { "epoch": 0.0142848, "grad_norm": 1.0891377925872803, "learning_rate": 1.3835602266109028e-05, "loss": 2.9441, "step": 146680 }, { "epoch": 0.0143104, "grad_norm": 0.7780723571777344, "learning_rate": 1.3834858566225435e-05, "loss": 2.9389, "step": 146690 }, { "epoch": 0.014336, "grad_norm": 0.7096871137619019, "learning_rate": 1.3834114841473808e-05, "loss": 2.9247, "step": 146700 }, { "epoch": 2.56e-05, "grad_norm": 0.7009669542312622, "learning_rate": 1.383337109185896e-05, "loss": 2.393, "step": 146710 }, { "epoch": 5.12e-05, "grad_norm": 0.8740160465240479, "learning_rate": 1.3832627317385721e-05, "loss": 2.8107, "step": 146720 }, { "epoch": 7.68e-05, "grad_norm": 0.903322696685791, "learning_rate": 1.3831883518058912e-05, "loss": 2.822, "step": 146730 }, { "epoch": 0.0001024, "grad_norm": 0.7453930974006653, "learning_rate": 1.383113969388336e-05, "loss": 2.8749, "step": 146740 }, { "epoch": 0.000128, "grad_norm": 0.7097765803337097, "learning_rate": 1.3830395844863882e-05, "loss": 2.764, "step": 146750 }, { "epoch": 0.0001536, "grad_norm": 0.6338069438934326, "learning_rate": 1.3829651971005306e-05, "loss": 2.9284, "step": 146760 }, { "epoch": 0.0001792, "grad_norm": 0.7601482272148132, "learning_rate": 1.3828908072312454e-05, "loss": 2.8445, "step": 146770 }, { "epoch": 0.0002048, "grad_norm": 0.6775023937225342, "learning_rate": 1.3828164148790154e-05, "loss": 3.0891, "step": 146780 }, { "epoch": 0.0002304, "grad_norm": 0.7435770630836487, "learning_rate": 1.3827420200443222e-05, "loss": 2.9924, "step": 146790 }, { "epoch": 0.000256, "grad_norm": 0.967919111251831, "learning_rate": 1.3826676227276487e-05, "loss": 2.9856, "step": 146800 }, { "epoch": 0.0002816, "grad_norm": 0.7288997173309326, "learning_rate": 1.3825932229294778e-05, "loss": 2.7589, "step": 146810 }, { "epoch": 0.0003072, "grad_norm": 0.8028984665870667, "learning_rate": 1.3825188206502916e-05, "loss": 2.7135, "step": 146820 }, { "epoch": 0.0003328, "grad_norm": 0.7625626921653748, "learning_rate": 1.3824444158905722e-05, "loss": 3.0785, "step": 146830 }, { "epoch": 0.0003584, "grad_norm": 0.7966890931129456, "learning_rate": 1.3823700086508025e-05, "loss": 2.7506, "step": 146840 }, { "epoch": 0.000384, "grad_norm": 0.6768530607223511, "learning_rate": 1.3822955989314648e-05, "loss": 2.7578, "step": 146850 }, { "epoch": 0.0004096, "grad_norm": 0.7799623012542725, "learning_rate": 1.3822211867330418e-05, "loss": 3.0709, "step": 146860 }, { "epoch": 0.0004352, "grad_norm": 0.7274339199066162, "learning_rate": 1.382146772056016e-05, "loss": 2.878, "step": 146870 }, { "epoch": 0.0004608, "grad_norm": 0.7029551267623901, "learning_rate": 1.3820723549008702e-05, "loss": 2.86, "step": 146880 }, { "epoch": 0.0004864, "grad_norm": 0.7305224537849426, "learning_rate": 1.381997935268086e-05, "loss": 2.8621, "step": 146890 }, { "epoch": 0.000512, "grad_norm": 0.6412774920463562, "learning_rate": 1.3819235131581471e-05, "loss": 2.8137, "step": 146900 }, { "epoch": 0.0005376, "grad_norm": 0.7634373903274536, "learning_rate": 1.381849088571536e-05, "loss": 2.7729, "step": 146910 }, { "epoch": 0.0005632, "grad_norm": 0.8380545377731323, "learning_rate": 1.3817746615087351e-05, "loss": 2.8677, "step": 146920 }, { "epoch": 0.0005888, "grad_norm": 0.6947778463363647, "learning_rate": 1.3817002319702266e-05, "loss": 2.6978, "step": 146930 }, { "epoch": 0.0006144, "grad_norm": 0.8329892754554749, "learning_rate": 1.3816257999564937e-05, "loss": 2.8628, "step": 146940 }, { "epoch": 0.00064, "grad_norm": 1.028887391090393, "learning_rate": 1.381551365468019e-05, "loss": 2.9086, "step": 146950 }, { "epoch": 0.0006656, "grad_norm": 0.7170112133026123, "learning_rate": 1.3814769285052849e-05, "loss": 2.8608, "step": 146960 }, { "epoch": 0.0006912, "grad_norm": 0.7494644522666931, "learning_rate": 1.3814024890687742e-05, "loss": 2.9574, "step": 146970 }, { "epoch": 0.0007168, "grad_norm": 0.6922876834869385, "learning_rate": 1.38132804715897e-05, "loss": 2.9703, "step": 146980 }, { "epoch": 0.0007424, "grad_norm": 0.8060649633407593, "learning_rate": 1.3812536027763545e-05, "loss": 3.0611, "step": 146990 }, { "epoch": 0.000768, "grad_norm": 0.7006008625030518, "learning_rate": 1.3811791559214109e-05, "loss": 2.7794, "step": 147000 }, { "epoch": 0.0007936, "grad_norm": 0.7732816338539124, "learning_rate": 1.381104706594622e-05, "loss": 2.9781, "step": 147010 }, { "epoch": 0.0008192, "grad_norm": 0.6669101715087891, "learning_rate": 1.38103025479647e-05, "loss": 2.9357, "step": 147020 }, { "epoch": 0.0008448, "grad_norm": 0.6854373812675476, "learning_rate": 1.3809558005274382e-05, "loss": 2.815, "step": 147030 }, { "epoch": 0.0008704, "grad_norm": 0.7245532274246216, "learning_rate": 1.3808813437880095e-05, "loss": 2.7928, "step": 147040 }, { "epoch": 0.000896, "grad_norm": 0.7005142569541931, "learning_rate": 1.3808068845786662e-05, "loss": 2.6007, "step": 147050 }, { "epoch": 0.0009216, "grad_norm": 0.7146377563476562, "learning_rate": 1.3807324228998913e-05, "loss": 2.6383, "step": 147060 }, { "epoch": 0.0009472, "grad_norm": 0.717065155506134, "learning_rate": 1.3806579587521683e-05, "loss": 2.8839, "step": 147070 }, { "epoch": 0.0009728, "grad_norm": 0.7202450633049011, "learning_rate": 1.3805834921359792e-05, "loss": 3.0046, "step": 147080 }, { "epoch": 0.0009984, "grad_norm": 0.7145574688911438, "learning_rate": 1.3805090230518074e-05, "loss": 2.848, "step": 147090 }, { "epoch": 0.001024, "grad_norm": 0.7212877869606018, "learning_rate": 1.3804345515001359e-05, "loss": 2.6594, "step": 147100 }, { "epoch": 0.0010496, "grad_norm": 0.8220937252044678, "learning_rate": 1.3803600774814474e-05, "loss": 2.9405, "step": 147110 }, { "epoch": 0.0010752, "grad_norm": 0.715415894985199, "learning_rate": 1.3802856009962247e-05, "loss": 2.8196, "step": 147120 }, { "epoch": 0.0011008, "grad_norm": 0.764664888381958, "learning_rate": 1.380211122044951e-05, "loss": 2.8211, "step": 147130 }, { "epoch": 0.0011264, "grad_norm": 0.7667043209075928, "learning_rate": 1.3801366406281096e-05, "loss": 2.7877, "step": 147140 }, { "epoch": 0.001152, "grad_norm": 0.7773059606552124, "learning_rate": 1.3800621567461826e-05, "loss": 3.041, "step": 147150 }, { "epoch": 0.0011776, "grad_norm": 1.0330883264541626, "learning_rate": 1.3799876703996536e-05, "loss": 3.0263, "step": 147160 }, { "epoch": 0.0012032, "grad_norm": 0.9702531695365906, "learning_rate": 1.3799131815890056e-05, "loss": 2.9497, "step": 147170 }, { "epoch": 0.0012288, "grad_norm": 0.7441454529762268, "learning_rate": 1.3798386903147218e-05, "loss": 2.7912, "step": 147180 }, { "epoch": 0.0012544, "grad_norm": 0.7700456380844116, "learning_rate": 1.3797641965772848e-05, "loss": 2.6362, "step": 147190 }, { "epoch": 0.00128, "grad_norm": 0.728385865688324, "learning_rate": 1.379689700377178e-05, "loss": 2.7847, "step": 147200 }, { "epoch": 2.56e-05, "grad_norm": 0.7800440192222595, "learning_rate": 1.3796152017148845e-05, "loss": 2.788, "step": 147210 }, { "epoch": 5.12e-05, "grad_norm": 0.7729989290237427, "learning_rate": 1.3795407005908872e-05, "loss": 2.9311, "step": 147220 }, { "epoch": 7.68e-05, "grad_norm": 0.8985741138458252, "learning_rate": 1.3794661970056694e-05, "loss": 3.0568, "step": 147230 }, { "epoch": 0.0001024, "grad_norm": 0.8183976411819458, "learning_rate": 1.3793916909597141e-05, "loss": 2.8757, "step": 147240 }, { "epoch": 0.000128, "grad_norm": 0.9126498103141785, "learning_rate": 1.3793171824535043e-05, "loss": 2.9591, "step": 147250 }, { "epoch": 0.0001536, "grad_norm": 0.7620256543159485, "learning_rate": 1.3792426714875236e-05, "loss": 2.8817, "step": 147260 }, { "epoch": 0.0001792, "grad_norm": 0.7700033783912659, "learning_rate": 1.379168158062255e-05, "loss": 2.9002, "step": 147270 }, { "epoch": 0.0002048, "grad_norm": 0.829311192035675, "learning_rate": 1.3790936421781815e-05, "loss": 3.0195, "step": 147280 }, { "epoch": 0.0002304, "grad_norm": 0.7199850082397461, "learning_rate": 1.3790191238357866e-05, "loss": 2.8997, "step": 147290 }, { "epoch": 0.000256, "grad_norm": 0.6985647678375244, "learning_rate": 1.3789446030355536e-05, "loss": 2.8197, "step": 147300 }, { "epoch": 0.0002816, "grad_norm": 0.6893442869186401, "learning_rate": 1.3788700797779656e-05, "loss": 2.8433, "step": 147310 }, { "epoch": 0.0003072, "grad_norm": 0.6500254273414612, "learning_rate": 1.3787955540635051e-05, "loss": 2.9058, "step": 147320 }, { "epoch": 0.0003328, "grad_norm": 0.7682217359542847, "learning_rate": 1.3787210258926569e-05, "loss": 3.0082, "step": 147330 }, { "epoch": 0.0003584, "grad_norm": 0.8845158815383911, "learning_rate": 1.378646495265903e-05, "loss": 3.0126, "step": 147340 }, { "epoch": 0.000384, "grad_norm": 0.7411803603172302, "learning_rate": 1.3785719621837274e-05, "loss": 2.8262, "step": 147350 }, { "epoch": 0.0004096, "grad_norm": 0.7723329663276672, "learning_rate": 1.3784974266466133e-05, "loss": 2.9118, "step": 147360 }, { "epoch": 0.0004352, "grad_norm": 0.6873086094856262, "learning_rate": 1.3784228886550442e-05, "loss": 2.6666, "step": 147370 }, { "epoch": 0.0004608, "grad_norm": 0.6801857948303223, "learning_rate": 1.378348348209503e-05, "loss": 3.0067, "step": 147380 }, { "epoch": 0.0004864, "grad_norm": 0.7819523811340332, "learning_rate": 1.3782738053104732e-05, "loss": 2.9148, "step": 147390 }, { "epoch": 0.000512, "grad_norm": 0.6656766533851624, "learning_rate": 1.3781992599584384e-05, "loss": 2.83, "step": 147400 }, { "epoch": 0.0005376, "grad_norm": 0.7283611297607422, "learning_rate": 1.378124712153882e-05, "loss": 2.7002, "step": 147410 }, { "epoch": 0.0005632, "grad_norm": 0.6990420818328857, "learning_rate": 1.3780501618972871e-05, "loss": 2.7636, "step": 147420 }, { "epoch": 0.0005888, "grad_norm": 0.6990112066268921, "learning_rate": 1.3779756091891377e-05, "loss": 2.9194, "step": 147430 }, { "epoch": 0.0006144, "grad_norm": 0.7716621160507202, "learning_rate": 1.3779010540299166e-05, "loss": 2.7784, "step": 147440 }, { "epoch": 0.00064, "grad_norm": 0.7655726671218872, "learning_rate": 1.3778264964201077e-05, "loss": 2.9077, "step": 147450 }, { "epoch": 0.0006656, "grad_norm": 0.6931673288345337, "learning_rate": 1.3777519363601944e-05, "loss": 2.7915, "step": 147460 }, { "epoch": 0.0006912, "grad_norm": 0.7029802799224854, "learning_rate": 1.3776773738506603e-05, "loss": 3.159, "step": 147470 }, { "epoch": 0.0007168, "grad_norm": 0.8267503380775452, "learning_rate": 1.3776028088919886e-05, "loss": 2.9176, "step": 147480 }, { "epoch": 0.0007424, "grad_norm": 1.0581640005111694, "learning_rate": 1.3775282414846633e-05, "loss": 3.3706, "step": 147490 }, { "epoch": 0.000768, "grad_norm": 0.7582218050956726, "learning_rate": 1.3774536716291676e-05, "loss": 2.8246, "step": 147500 }, { "epoch": 0.0007936, "grad_norm": 0.695014238357544, "learning_rate": 1.3773790993259848e-05, "loss": 2.8565, "step": 147510 }, { "epoch": 0.0008192, "grad_norm": 0.663722813129425, "learning_rate": 1.3773045245755991e-05, "loss": 2.7772, "step": 147520 }, { "epoch": 0.0008448, "grad_norm": 0.7351208329200745, "learning_rate": 1.3772299473784942e-05, "loss": 2.972, "step": 147530 }, { "epoch": 0.0008704, "grad_norm": 0.722156822681427, "learning_rate": 1.377155367735153e-05, "loss": 2.8701, "step": 147540 }, { "epoch": 0.000896, "grad_norm": 0.7417086958885193, "learning_rate": 1.3770807856460593e-05, "loss": 2.887, "step": 147550 }, { "epoch": 0.0009216, "grad_norm": 0.8600040674209595, "learning_rate": 1.3770062011116972e-05, "loss": 2.7694, "step": 147560 }, { "epoch": 0.0009472, "grad_norm": 0.7348060011863708, "learning_rate": 1.3769316141325502e-05, "loss": 2.6724, "step": 147570 }, { "epoch": 0.0009728, "grad_norm": 0.6664402484893799, "learning_rate": 1.3768570247091015e-05, "loss": 2.7922, "step": 147580 }, { "epoch": 0.0009984, "grad_norm": 0.7798592448234558, "learning_rate": 1.3767824328418351e-05, "loss": 2.8327, "step": 147590 }, { "epoch": 0.001024, "grad_norm": 0.7263541221618652, "learning_rate": 1.376707838531235e-05, "loss": 2.8664, "step": 147600 }, { "epoch": 0.0010496, "grad_norm": 0.8597992062568665, "learning_rate": 1.3766332417777847e-05, "loss": 2.9967, "step": 147610 }, { "epoch": 0.0010752, "grad_norm": 0.8280649185180664, "learning_rate": 1.3765586425819679e-05, "loss": 2.9577, "step": 147620 }, { "epoch": 0.0011008, "grad_norm": 0.6867247223854065, "learning_rate": 1.3764840409442686e-05, "loss": 2.8446, "step": 147630 }, { "epoch": 0.0011264, "grad_norm": 0.7929026484489441, "learning_rate": 1.3764094368651701e-05, "loss": 2.7281, "step": 147640 }, { "epoch": 0.001152, "grad_norm": 0.7048295736312866, "learning_rate": 1.3763348303451563e-05, "loss": 2.9555, "step": 147650 }, { "epoch": 0.0011776, "grad_norm": 0.8141499161720276, "learning_rate": 1.3762602213847113e-05, "loss": 3.0214, "step": 147660 }, { "epoch": 0.0012032, "grad_norm": 0.7490373253822327, "learning_rate": 1.3761856099843188e-05, "loss": 2.8024, "step": 147670 }, { "epoch": 0.0012288, "grad_norm": 0.6457375288009644, "learning_rate": 1.3761109961444622e-05, "loss": 2.9307, "step": 147680 }, { "epoch": 0.0012544, "grad_norm": 0.7410626411437988, "learning_rate": 1.3760363798656261e-05, "loss": 3.0334, "step": 147690 }, { "epoch": 0.00128, "grad_norm": 0.8019039034843445, "learning_rate": 1.3759617611482943e-05, "loss": 2.8448, "step": 147700 }, { "epoch": 0.0013056, "grad_norm": 0.7962800860404968, "learning_rate": 1.3758871399929502e-05, "loss": 2.8877, "step": 147710 }, { "epoch": 0.0013312, "grad_norm": 0.9015649557113647, "learning_rate": 1.3758125164000777e-05, "loss": 2.9792, "step": 147720 }, { "epoch": 0.0013568, "grad_norm": 0.7594185471534729, "learning_rate": 1.3757378903701613e-05, "loss": 2.8959, "step": 147730 }, { "epoch": 0.0013824, "grad_norm": 0.7103163003921509, "learning_rate": 1.375663261903684e-05, "loss": 2.8505, "step": 147740 }, { "epoch": 0.001408, "grad_norm": 0.7301872372627258, "learning_rate": 1.3755886310011306e-05, "loss": 2.9512, "step": 147750 }, { "epoch": 0.0014336, "grad_norm": 0.8479196429252625, "learning_rate": 1.3755139976629845e-05, "loss": 2.9273, "step": 147760 }, { "epoch": 0.0014592, "grad_norm": 0.8744624257087708, "learning_rate": 1.3754393618897304e-05, "loss": 2.7743, "step": 147770 }, { "epoch": 0.0014848, "grad_norm": 0.6613696813583374, "learning_rate": 1.3753647236818513e-05, "loss": 2.8023, "step": 147780 }, { "epoch": 0.0015104, "grad_norm": 0.7379962205886841, "learning_rate": 1.375290083039832e-05, "loss": 3.1216, "step": 147790 }, { "epoch": 0.001536, "grad_norm": 0.8022428750991821, "learning_rate": 1.3752154399641564e-05, "loss": 2.5809, "step": 147800 }, { "epoch": 2.56e-05, "grad_norm": 1.0703444480895996, "learning_rate": 1.3751407944553081e-05, "loss": 2.631, "step": 147810 }, { "epoch": 5.12e-05, "grad_norm": 0.9151116609573364, "learning_rate": 1.3750661465137716e-05, "loss": 2.8931, "step": 147820 }, { "epoch": 7.68e-05, "grad_norm": 0.6774001717567444, "learning_rate": 1.3749914961400308e-05, "loss": 2.7978, "step": 147830 }, { "epoch": 0.0001024, "grad_norm": 0.6772567629814148, "learning_rate": 1.37491684333457e-05, "loss": 2.7877, "step": 147840 }, { "epoch": 0.000128, "grad_norm": 0.9214804768562317, "learning_rate": 1.3748421880978724e-05, "loss": 3.0353, "step": 147850 }, { "epoch": 0.0001536, "grad_norm": 0.7942911982536316, "learning_rate": 1.3747675304304235e-05, "loss": 3.1723, "step": 147860 }, { "epoch": 0.0001792, "grad_norm": 0.7394864559173584, "learning_rate": 1.3746928703327064e-05, "loss": 3.0927, "step": 147870 }, { "epoch": 0.0002048, "grad_norm": 0.6979058384895325, "learning_rate": 1.374618207805206e-05, "loss": 2.407, "step": 147880 }, { "epoch": 0.0002304, "grad_norm": 0.7608799934387207, "learning_rate": 1.3745435428484059e-05, "loss": 3.1053, "step": 147890 }, { "epoch": 0.000256, "grad_norm": 0.8633912205696106, "learning_rate": 1.3744688754627907e-05, "loss": 2.7845, "step": 147900 }, { "epoch": 0.0002816, "grad_norm": 0.7191622257232666, "learning_rate": 1.374394205648844e-05, "loss": 2.797, "step": 147910 }, { "epoch": 0.0003072, "grad_norm": 0.6898598670959473, "learning_rate": 1.3743195334070505e-05, "loss": 2.5724, "step": 147920 }, { "epoch": 0.0003328, "grad_norm": 0.7219297885894775, "learning_rate": 1.3742448587378943e-05, "loss": 2.8635, "step": 147930 }, { "epoch": 0.0003584, "grad_norm": 0.9222599864006042, "learning_rate": 1.3741701816418595e-05, "loss": 2.9287, "step": 147940 }, { "epoch": 0.000384, "grad_norm": 0.6505687832832336, "learning_rate": 1.3740955021194307e-05, "loss": 2.8411, "step": 147950 }, { "epoch": 0.0004096, "grad_norm": 0.929923415184021, "learning_rate": 1.3740208201710921e-05, "loss": 2.913, "step": 147960 }, { "epoch": 0.0004352, "grad_norm": 0.7603358626365662, "learning_rate": 1.3739461357973279e-05, "loss": 2.995, "step": 147970 }, { "epoch": 0.0004608, "grad_norm": 0.7751281261444092, "learning_rate": 1.373871448998622e-05, "loss": 2.8153, "step": 147980 }, { "epoch": 0.0004864, "grad_norm": 0.8477267622947693, "learning_rate": 1.3737967597754594e-05, "loss": 3.2345, "step": 147990 }, { "epoch": 0.000512, "grad_norm": 0.7678060531616211, "learning_rate": 1.3737220681283244e-05, "loss": 2.7573, "step": 148000 }, { "epoch": 0.0005376, "grad_norm": 0.8035601377487183, "learning_rate": 1.3736473740577005e-05, "loss": 2.7886, "step": 148010 }, { "epoch": 0.0005632, "grad_norm": 0.675875723361969, "learning_rate": 1.3735726775640727e-05, "loss": 2.5274, "step": 148020 }, { "epoch": 0.0005888, "grad_norm": 0.7664881348609924, "learning_rate": 1.373497978647926e-05, "loss": 2.7406, "step": 148030 }, { "epoch": 0.0006144, "grad_norm": 0.7450348138809204, "learning_rate": 1.3734232773097438e-05, "loss": 2.7996, "step": 148040 }, { "epoch": 0.00064, "grad_norm": 0.6770084500312805, "learning_rate": 1.373348573550011e-05, "loss": 3.0366, "step": 148050 }, { "epoch": 0.0006656, "grad_norm": 0.7468043565750122, "learning_rate": 1.373273867369212e-05, "loss": 3.25, "step": 148060 }, { "epoch": 0.0006912, "grad_norm": 0.7790248990058899, "learning_rate": 1.373199158767831e-05, "loss": 3.0972, "step": 148070 }, { "epoch": 0.0007168, "grad_norm": 0.7147325277328491, "learning_rate": 1.3731244477463528e-05, "loss": 2.823, "step": 148080 }, { "epoch": 0.0007424, "grad_norm": 0.7204649448394775, "learning_rate": 1.3730497343052616e-05, "loss": 3.0981, "step": 148090 }, { "epoch": 0.000768, "grad_norm": 0.6850098967552185, "learning_rate": 1.3729750184450422e-05, "loss": 2.8223, "step": 148100 }, { "epoch": 0.0007936, "grad_norm": 0.7410989999771118, "learning_rate": 1.3729003001661782e-05, "loss": 2.8324, "step": 148110 }, { "epoch": 0.0008192, "grad_norm": 0.7408981919288635, "learning_rate": 1.3728255794691557e-05, "loss": 2.9018, "step": 148120 }, { "epoch": 0.0008448, "grad_norm": 0.7246842980384827, "learning_rate": 1.372750856354458e-05, "loss": 2.7869, "step": 148130 }, { "epoch": 0.0008704, "grad_norm": 0.6990874409675598, "learning_rate": 1.3726761308225702e-05, "loss": 2.7891, "step": 148140 }, { "epoch": 0.000896, "grad_norm": 0.6636192798614502, "learning_rate": 1.3726014028739767e-05, "loss": 2.7951, "step": 148150 }, { "epoch": 0.0009216, "grad_norm": 0.7149374485015869, "learning_rate": 1.3725266725091624e-05, "loss": 2.8496, "step": 148160 }, { "epoch": 0.0009472, "grad_norm": 0.7416369915008545, "learning_rate": 1.3724519397286114e-05, "loss": 2.9483, "step": 148170 }, { "epoch": 0.0009728, "grad_norm": 1.1643892526626587, "learning_rate": 1.3723772045328083e-05, "loss": 2.9559, "step": 148180 }, { "epoch": 0.0009984, "grad_norm": 0.8118160963058472, "learning_rate": 1.3723024669222383e-05, "loss": 2.8668, "step": 148190 }, { "epoch": 0.001024, "grad_norm": 0.7233357429504395, "learning_rate": 1.3722277268973857e-05, "loss": 2.7626, "step": 148200 }, { "epoch": 0.0010496, "grad_norm": 0.6737782955169678, "learning_rate": 1.372152984458735e-05, "loss": 2.8101, "step": 148210 }, { "epoch": 0.0010752, "grad_norm": 0.7093512415885925, "learning_rate": 1.3720782396067714e-05, "loss": 2.9447, "step": 148220 }, { "epoch": 0.0011008, "grad_norm": 0.9285646080970764, "learning_rate": 1.3720034923419792e-05, "loss": 2.9283, "step": 148230 }, { "epoch": 0.0011264, "grad_norm": 0.7516495585441589, "learning_rate": 1.371928742664843e-05, "loss": 2.6854, "step": 148240 }, { "epoch": 0.001152, "grad_norm": 0.837556779384613, "learning_rate": 1.3718539905758481e-05, "loss": 3.0453, "step": 148250 }, { "epoch": 0.0011776, "grad_norm": 0.8150685429573059, "learning_rate": 1.3717792360754789e-05, "loss": 3.0439, "step": 148260 }, { "epoch": 0.0012032, "grad_norm": 1.019733190536499, "learning_rate": 1.3717044791642203e-05, "loss": 2.9975, "step": 148270 }, { "epoch": 0.0012288, "grad_norm": 0.7184702157974243, "learning_rate": 1.3716297198425563e-05, "loss": 3.1018, "step": 148280 }, { "epoch": 0.0012544, "grad_norm": 0.6951852440834045, "learning_rate": 1.3715549581109728e-05, "loss": 2.8734, "step": 148290 }, { "epoch": 0.00128, "grad_norm": 0.7930635213851929, "learning_rate": 1.371480193969954e-05, "loss": 3.1234, "step": 148300 }, { "epoch": 0.0013056, "grad_norm": 0.7761464715003967, "learning_rate": 1.3714054274199848e-05, "loss": 2.7259, "step": 148310 }, { "epoch": 0.0013312, "grad_norm": 0.6884698867797852, "learning_rate": 1.3713306584615504e-05, "loss": 2.8602, "step": 148320 }, { "epoch": 0.0013568, "grad_norm": 0.8612653613090515, "learning_rate": 1.3712558870951352e-05, "loss": 2.8898, "step": 148330 }, { "epoch": 0.0013824, "grad_norm": 0.6741415858268738, "learning_rate": 1.3711811133212243e-05, "loss": 2.8888, "step": 148340 }, { "epoch": 0.001408, "grad_norm": 0.9594283103942871, "learning_rate": 1.3711063371403024e-05, "loss": 2.6794, "step": 148350 }, { "epoch": 0.0014336, "grad_norm": 0.8958351016044617, "learning_rate": 1.3710315585528546e-05, "loss": 2.8311, "step": 148360 }, { "epoch": 0.0014592, "grad_norm": 0.7185925245285034, "learning_rate": 1.3709567775593655e-05, "loss": 2.9772, "step": 148370 }, { "epoch": 0.0014848, "grad_norm": 0.7046003937721252, "learning_rate": 1.3708819941603207e-05, "loss": 2.5975, "step": 148380 }, { "epoch": 0.0015104, "grad_norm": 0.7832366228103638, "learning_rate": 1.3708072083562045e-05, "loss": 2.5279, "step": 148390 }, { "epoch": 0.001536, "grad_norm": 0.7368016839027405, "learning_rate": 1.3707324201475024e-05, "loss": 2.9016, "step": 148400 }, { "epoch": 0.0015616, "grad_norm": 0.7193725109100342, "learning_rate": 1.3706576295346986e-05, "loss": 2.9409, "step": 148410 }, { "epoch": 0.0015872, "grad_norm": 0.6888477802276611, "learning_rate": 1.370582836518279e-05, "loss": 3.103, "step": 148420 }, { "epoch": 0.0016128, "grad_norm": 0.7379840016365051, "learning_rate": 1.3705080410987281e-05, "loss": 2.9296, "step": 148430 }, { "epoch": 0.0016384, "grad_norm": 0.6795030832290649, "learning_rate": 1.3704332432765309e-05, "loss": 2.8106, "step": 148440 }, { "epoch": 0.001664, "grad_norm": 0.8466413021087646, "learning_rate": 1.3703584430521726e-05, "loss": 2.9156, "step": 148450 }, { "epoch": 0.0016896, "grad_norm": 0.7617866396903992, "learning_rate": 1.3702836404261382e-05, "loss": 2.8876, "step": 148460 }, { "epoch": 0.0017152, "grad_norm": 0.9209862351417542, "learning_rate": 1.3702088353989127e-05, "loss": 3.0599, "step": 148470 }, { "epoch": 0.0017408, "grad_norm": 0.7240144610404968, "learning_rate": 1.3701340279709813e-05, "loss": 2.8818, "step": 148480 }, { "epoch": 0.0017664, "grad_norm": 0.6407962441444397, "learning_rate": 1.3700592181428296e-05, "loss": 2.9143, "step": 148490 }, { "epoch": 0.001792, "grad_norm": 0.7051349878311157, "learning_rate": 1.3699844059149418e-05, "loss": 2.7328, "step": 148500 }, { "epoch": 0.0018176, "grad_norm": 0.7591454982757568, "learning_rate": 1.3699095912878035e-05, "loss": 2.6932, "step": 148510 }, { "epoch": 0.0018432, "grad_norm": 0.928862988948822, "learning_rate": 1.3698347742619e-05, "loss": 2.8265, "step": 148520 }, { "epoch": 0.0018688, "grad_norm": 0.7408100962638855, "learning_rate": 1.3697599548377161e-05, "loss": 2.9651, "step": 148530 }, { "epoch": 0.0018944, "grad_norm": 0.7071086168289185, "learning_rate": 1.3696851330157368e-05, "loss": 2.7863, "step": 148540 }, { "epoch": 0.00192, "grad_norm": 0.6966671347618103, "learning_rate": 1.3696103087964482e-05, "loss": 3.0521, "step": 148550 }, { "epoch": 0.0019456, "grad_norm": 0.8101327419281006, "learning_rate": 1.3695354821803348e-05, "loss": 2.9773, "step": 148560 }, { "epoch": 0.0019712, "grad_norm": 0.8779668211936951, "learning_rate": 1.3694606531678821e-05, "loss": 2.8448, "step": 148570 }, { "epoch": 0.0019968, "grad_norm": 0.8862952589988708, "learning_rate": 1.369385821759575e-05, "loss": 3.04, "step": 148580 }, { "epoch": 0.0020224, "grad_norm": 0.6750712394714355, "learning_rate": 1.3693109879558994e-05, "loss": 2.9755, "step": 148590 }, { "epoch": 0.002048, "grad_norm": 0.6694158315658569, "learning_rate": 1.36923615175734e-05, "loss": 2.6232, "step": 148600 }, { "epoch": 0.0020736, "grad_norm": 0.7088623642921448, "learning_rate": 1.3691613131643821e-05, "loss": 2.7674, "step": 148610 }, { "epoch": 0.0020992, "grad_norm": 0.7742458581924438, "learning_rate": 1.3690864721775117e-05, "loss": 2.851, "step": 148620 }, { "epoch": 0.0021248, "grad_norm": 0.8134743571281433, "learning_rate": 1.369011628797213e-05, "loss": 2.7755, "step": 148630 }, { "epoch": 0.0021504, "grad_norm": 0.8329054117202759, "learning_rate": 1.368936783023972e-05, "loss": 2.8658, "step": 148640 }, { "epoch": 0.002176, "grad_norm": 0.6685836315155029, "learning_rate": 1.3688619348582745e-05, "loss": 2.9316, "step": 148650 }, { "epoch": 0.0022016, "grad_norm": 1.0105751752853394, "learning_rate": 1.3687870843006052e-05, "loss": 2.8726, "step": 148660 }, { "epoch": 0.0022272, "grad_norm": 0.6405223608016968, "learning_rate": 1.3687122313514495e-05, "loss": 2.8131, "step": 148670 }, { "epoch": 0.0022528, "grad_norm": 0.819717288017273, "learning_rate": 1.3686373760112933e-05, "loss": 2.8535, "step": 148680 }, { "epoch": 0.0022784, "grad_norm": 0.7905990481376648, "learning_rate": 1.3685625182806215e-05, "loss": 3.0481, "step": 148690 }, { "epoch": 0.002304, "grad_norm": 0.7020528316497803, "learning_rate": 1.3684876581599196e-05, "loss": 2.6243, "step": 148700 }, { "epoch": 0.0023296, "grad_norm": 0.6753408312797546, "learning_rate": 1.3684127956496733e-05, "loss": 2.8889, "step": 148710 }, { "epoch": 0.0023552, "grad_norm": 0.7117926478385925, "learning_rate": 1.368337930750368e-05, "loss": 2.7471, "step": 148720 }, { "epoch": 0.0023808, "grad_norm": 0.79106605052948, "learning_rate": 1.3682630634624888e-05, "loss": 2.846, "step": 148730 }, { "epoch": 0.0024064, "grad_norm": 0.7474511861801147, "learning_rate": 1.3681881937865216e-05, "loss": 2.8374, "step": 148740 }, { "epoch": 0.002432, "grad_norm": 0.6875680685043335, "learning_rate": 1.3681133217229521e-05, "loss": 3.0882, "step": 148750 }, { "epoch": 0.0024576, "grad_norm": 0.7294420599937439, "learning_rate": 1.3680384472722654e-05, "loss": 3.0687, "step": 148760 }, { "epoch": 0.0024832, "grad_norm": 0.7193138599395752, "learning_rate": 1.3679635704349469e-05, "loss": 2.9256, "step": 148770 }, { "epoch": 0.0025088, "grad_norm": 0.7516984939575195, "learning_rate": 1.3678886912114828e-05, "loss": 2.8601, "step": 148780 }, { "epoch": 0.0025344, "grad_norm": 0.7192407846450806, "learning_rate": 1.3678138096023586e-05, "loss": 2.9595, "step": 148790 }, { "epoch": 0.00256, "grad_norm": 0.7397412061691284, "learning_rate": 1.3677389256080588e-05, "loss": 2.8681, "step": 148800 }, { "epoch": 0.0025856, "grad_norm": 0.7831892967224121, "learning_rate": 1.36766403922907e-05, "loss": 2.8734, "step": 148810 }, { "epoch": 0.0026112, "grad_norm": 0.7796103358268738, "learning_rate": 1.3675891504658782e-05, "loss": 3.0023, "step": 148820 }, { "epoch": 0.0026368, "grad_norm": 0.8320309519767761, "learning_rate": 1.367514259318968e-05, "loss": 3.1013, "step": 148830 }, { "epoch": 0.0026624, "grad_norm": 0.6357343792915344, "learning_rate": 1.3674393657888256e-05, "loss": 3.0084, "step": 148840 }, { "epoch": 0.002688, "grad_norm": 0.7344212532043457, "learning_rate": 1.3673644698759367e-05, "loss": 2.9388, "step": 148850 }, { "epoch": 0.0027136, "grad_norm": 0.9437594413757324, "learning_rate": 1.3672895715807866e-05, "loss": 2.997, "step": 148860 }, { "epoch": 0.0027392, "grad_norm": 0.7394600510597229, "learning_rate": 1.3672146709038612e-05, "loss": 3.0717, "step": 148870 }, { "epoch": 0.0027648, "grad_norm": 0.7479563355445862, "learning_rate": 1.3671397678456465e-05, "loss": 2.9749, "step": 148880 }, { "epoch": 0.0027904, "grad_norm": 0.7741900682449341, "learning_rate": 1.3670648624066276e-05, "loss": 2.8674, "step": 148890 }, { "epoch": 0.002816, "grad_norm": 1.2478344440460205, "learning_rate": 1.3669899545872909e-05, "loss": 2.8262, "step": 148900 }, { "epoch": 0.0028416, "grad_norm": 0.7028167247772217, "learning_rate": 1.3669150443881219e-05, "loss": 3.0104, "step": 148910 }, { "epoch": 0.0028672, "grad_norm": 0.769763708114624, "learning_rate": 1.3668401318096065e-05, "loss": 2.9241, "step": 148920 }, { "epoch": 0.0028928, "grad_norm": 0.9980193972587585, "learning_rate": 1.36676521685223e-05, "loss": 3.0238, "step": 148930 }, { "epoch": 0.0029184, "grad_norm": 0.6764048337936401, "learning_rate": 1.3666902995164788e-05, "loss": 2.8357, "step": 148940 }, { "epoch": 0.002944, "grad_norm": 0.9343173503875732, "learning_rate": 1.3666153798028386e-05, "loss": 2.8395, "step": 148950 }, { "epoch": 0.0029696, "grad_norm": 0.7476577162742615, "learning_rate": 1.3665404577117947e-05, "loss": 3.0998, "step": 148960 }, { "epoch": 0.0029952, "grad_norm": 0.6975662708282471, "learning_rate": 1.3664655332438334e-05, "loss": 2.8718, "step": 148970 }, { "epoch": 0.0030208, "grad_norm": 0.7432191371917725, "learning_rate": 1.3663906063994406e-05, "loss": 2.9398, "step": 148980 }, { "epoch": 0.0030464, "grad_norm": 0.8122135400772095, "learning_rate": 1.3663156771791022e-05, "loss": 3.0613, "step": 148990 }, { "epoch": 0.003072, "grad_norm": 0.8740618228912354, "learning_rate": 1.3662407455833038e-05, "loss": 3.1049, "step": 149000 }, { "epoch": 0.0030976, "grad_norm": 0.9788641929626465, "learning_rate": 1.3661658116125315e-05, "loss": 2.9793, "step": 149010 }, { "epoch": 0.0031232, "grad_norm": 0.8853468298912048, "learning_rate": 1.3660908752672715e-05, "loss": 2.8919, "step": 149020 }, { "epoch": 0.0031488, "grad_norm": 0.7240023612976074, "learning_rate": 1.3660159365480093e-05, "loss": 3.1398, "step": 149030 }, { "epoch": 0.0031744, "grad_norm": 0.7853969931602478, "learning_rate": 1.3659409954552309e-05, "loss": 3.029, "step": 149040 }, { "epoch": 0.0032, "grad_norm": 1.3278086185455322, "learning_rate": 1.3658660519894226e-05, "loss": 3.1238, "step": 149050 }, { "epoch": 0.0032256, "grad_norm": 0.7289062142372131, "learning_rate": 1.3657911061510701e-05, "loss": 2.815, "step": 149060 }, { "epoch": 0.0032512, "grad_norm": 0.6712574362754822, "learning_rate": 1.3657161579406594e-05, "loss": 3.0787, "step": 149070 }, { "epoch": 0.0032768, "grad_norm": 0.7458216547966003, "learning_rate": 1.365641207358677e-05, "loss": 3.223, "step": 149080 }, { "epoch": 0.0033024, "grad_norm": 0.7393764853477478, "learning_rate": 1.365566254405608e-05, "loss": 3.0704, "step": 149090 }, { "epoch": 0.003328, "grad_norm": 0.6819412708282471, "learning_rate": 1.3654912990819394e-05, "loss": 2.9818, "step": 149100 }, { "epoch": 0.0033536, "grad_norm": 0.7426139116287231, "learning_rate": 1.3654163413881566e-05, "loss": 2.8823, "step": 149110 }, { "epoch": 0.0033792, "grad_norm": 0.8052567839622498, "learning_rate": 1.3653413813247462e-05, "loss": 2.6376, "step": 149120 }, { "epoch": 0.0034048, "grad_norm": 0.7331191301345825, "learning_rate": 1.3652664188921938e-05, "loss": 2.7567, "step": 149130 }, { "epoch": 0.0034304, "grad_norm": 0.7134453654289246, "learning_rate": 1.3651914540909859e-05, "loss": 3.0919, "step": 149140 }, { "epoch": 0.003456, "grad_norm": 0.7256213426589966, "learning_rate": 1.3651164869216085e-05, "loss": 3.0688, "step": 149150 }, { "epoch": 0.0034816, "grad_norm": 0.7097216844558716, "learning_rate": 1.3650415173845474e-05, "loss": 2.8843, "step": 149160 }, { "epoch": 0.0035072, "grad_norm": 0.6918501853942871, "learning_rate": 1.3649665454802894e-05, "loss": 2.8396, "step": 149170 }, { "epoch": 0.0035328, "grad_norm": 0.7281439900398254, "learning_rate": 1.3648915712093204e-05, "loss": 2.9856, "step": 149180 }, { "epoch": 0.0035584, "grad_norm": 0.7640419602394104, "learning_rate": 1.3648165945721263e-05, "loss": 2.95, "step": 149190 }, { "epoch": 0.003584, "grad_norm": 0.7077479362487793, "learning_rate": 1.3647416155691936e-05, "loss": 2.9674, "step": 149200 }, { "epoch": 0.0036096, "grad_norm": 0.7585905194282532, "learning_rate": 1.3646666342010084e-05, "loss": 3.1817, "step": 149210 }, { "epoch": 0.0036352, "grad_norm": 0.691241443157196, "learning_rate": 1.3645916504680575e-05, "loss": 2.9757, "step": 149220 }, { "epoch": 0.0036608, "grad_norm": 0.7577566504478455, "learning_rate": 1.3645166643708261e-05, "loss": 3.1063, "step": 149230 }, { "epoch": 0.0036864, "grad_norm": 0.751715362071991, "learning_rate": 1.3644416759098012e-05, "loss": 3.1779, "step": 149240 }, { "epoch": 0.003712, "grad_norm": 0.7152424454689026, "learning_rate": 1.3643666850854687e-05, "loss": 2.8955, "step": 149250 }, { "epoch": 0.0037376, "grad_norm": 0.7696857452392578, "learning_rate": 1.3642916918983151e-05, "loss": 3.0718, "step": 149260 }, { "epoch": 0.0037632, "grad_norm": 0.7207740545272827, "learning_rate": 1.3642166963488268e-05, "loss": 2.9445, "step": 149270 }, { "epoch": 0.0037888, "grad_norm": 0.8166680932044983, "learning_rate": 1.3641416984374903e-05, "loss": 3.0671, "step": 149280 }, { "epoch": 0.0038144, "grad_norm": 0.7604951858520508, "learning_rate": 1.3640666981647913e-05, "loss": 2.8764, "step": 149290 }, { "epoch": 0.00384, "grad_norm": 0.8180694580078125, "learning_rate": 1.3639916955312167e-05, "loss": 2.8271, "step": 149300 }, { "epoch": 0.0038656, "grad_norm": 0.7391143441200256, "learning_rate": 1.3639166905372528e-05, "loss": 2.9576, "step": 149310 }, { "epoch": 0.0038912, "grad_norm": 0.834847092628479, "learning_rate": 1.3638416831833856e-05, "loss": 3.2557, "step": 149320 }, { "epoch": 0.0039168, "grad_norm": 0.8262297511100769, "learning_rate": 1.3637666734701017e-05, "loss": 2.9112, "step": 149330 }, { "epoch": 0.0039424, "grad_norm": 0.7215272188186646, "learning_rate": 1.363691661397888e-05, "loss": 2.8273, "step": 149340 }, { "epoch": 0.003968, "grad_norm": 0.7415220737457275, "learning_rate": 1.36361664696723e-05, "loss": 3.0942, "step": 149350 }, { "epoch": 0.0039936, "grad_norm": 0.8563736081123352, "learning_rate": 1.363541630178615e-05, "loss": 2.9671, "step": 149360 }, { "epoch": 0.0040192, "grad_norm": 0.8530844449996948, "learning_rate": 1.3634666110325291e-05, "loss": 3.0424, "step": 149370 }, { "epoch": 0.0040448, "grad_norm": 1.606966257095337, "learning_rate": 1.3633915895294592e-05, "loss": 3.1623, "step": 149380 }, { "epoch": 0.0040704, "grad_norm": 0.676544725894928, "learning_rate": 1.3633165656698907e-05, "loss": 3.0585, "step": 149390 }, { "epoch": 0.004096, "grad_norm": 0.7390878200531006, "learning_rate": 1.3632415394543115e-05, "loss": 2.919, "step": 149400 }, { "epoch": 0.0041216, "grad_norm": 0.7319850921630859, "learning_rate": 1.3631665108832071e-05, "loss": 3.0348, "step": 149410 }, { "epoch": 0.0041472, "grad_norm": 0.721907913684845, "learning_rate": 1.3630914799570642e-05, "loss": 2.9003, "step": 149420 }, { "epoch": 0.0041728, "grad_norm": 0.8007484078407288, "learning_rate": 1.3630164466763698e-05, "loss": 3.0174, "step": 149430 }, { "epoch": 0.0041984, "grad_norm": 0.7859363555908203, "learning_rate": 1.3629414110416103e-05, "loss": 3.1231, "step": 149440 }, { "epoch": 0.004224, "grad_norm": 0.7441596984863281, "learning_rate": 1.3628663730532723e-05, "loss": 3.1562, "step": 149450 }, { "epoch": 0.0042496, "grad_norm": 0.9077397584915161, "learning_rate": 1.3627913327118421e-05, "loss": 2.9462, "step": 149460 }, { "epoch": 0.0042752, "grad_norm": 0.7175806164741516, "learning_rate": 1.3627162900178066e-05, "loss": 2.9887, "step": 149470 }, { "epoch": 0.0043008, "grad_norm": 0.7048949599266052, "learning_rate": 1.3626412449716524e-05, "loss": 3.1468, "step": 149480 }, { "epoch": 0.0043264, "grad_norm": 0.7760501503944397, "learning_rate": 1.362566197573866e-05, "loss": 3.0313, "step": 149490 }, { "epoch": 0.004352, "grad_norm": 0.6871564388275146, "learning_rate": 1.3624911478249341e-05, "loss": 2.9747, "step": 149500 }, { "epoch": 0.0043776, "grad_norm": 0.639707624912262, "learning_rate": 1.3624160957253436e-05, "loss": 2.9093, "step": 149510 }, { "epoch": 0.0044032, "grad_norm": 0.8680580854415894, "learning_rate": 1.362341041275581e-05, "loss": 3.1684, "step": 149520 }, { "epoch": 0.0044288, "grad_norm": 0.9879869222640991, "learning_rate": 1.362265984476133e-05, "loss": 3.0451, "step": 149530 }, { "epoch": 0.0044544, "grad_norm": 0.8183565735816956, "learning_rate": 1.3621909253274866e-05, "loss": 2.8117, "step": 149540 }, { "epoch": 0.00448, "grad_norm": 0.8757680654525757, "learning_rate": 1.3621158638301282e-05, "loss": 3.0331, "step": 149550 }, { "epoch": 0.0045056, "grad_norm": 0.723071277141571, "learning_rate": 1.3620407999845445e-05, "loss": 3.0118, "step": 149560 }, { "epoch": 0.0045312, "grad_norm": 0.8380459547042847, "learning_rate": 1.3619657337912222e-05, "loss": 3.014, "step": 149570 }, { "epoch": 0.0045568, "grad_norm": 0.7377728819847107, "learning_rate": 1.361890665250649e-05, "loss": 3.0867, "step": 149580 }, { "epoch": 0.0045824, "grad_norm": 0.7657629251480103, "learning_rate": 1.3618155943633102e-05, "loss": 3.0709, "step": 149590 }, { "epoch": 0.004608, "grad_norm": 0.7381064891815186, "learning_rate": 1.3617405211296939e-05, "loss": 3.1034, "step": 149600 }, { "epoch": 0.0046336, "grad_norm": 0.9657368659973145, "learning_rate": 1.3616654455502867e-05, "loss": 2.994, "step": 149610 }, { "epoch": 0.0046592, "grad_norm": 1.5091923475265503, "learning_rate": 1.3615903676255748e-05, "loss": 3.0224, "step": 149620 }, { "epoch": 0.0046848, "grad_norm": 0.7608234882354736, "learning_rate": 1.3615152873560456e-05, "loss": 3.2247, "step": 149630 }, { "epoch": 0.0047104, "grad_norm": 0.8065701127052307, "learning_rate": 1.361440204742186e-05, "loss": 3.0042, "step": 149640 }, { "epoch": 0.004736, "grad_norm": 0.9221349954605103, "learning_rate": 1.3613651197844825e-05, "loss": 2.907, "step": 149650 }, { "epoch": 0.0047616, "grad_norm": 0.8605918884277344, "learning_rate": 1.3612900324834221e-05, "loss": 3.3002, "step": 149660 }, { "epoch": 0.0047872, "grad_norm": 0.6932876706123352, "learning_rate": 1.3612149428394923e-05, "loss": 3.0288, "step": 149670 }, { "epoch": 0.0048128, "grad_norm": 0.7701741456985474, "learning_rate": 1.3611398508531793e-05, "loss": 2.998, "step": 149680 }, { "epoch": 0.0048384, "grad_norm": 0.7375137805938721, "learning_rate": 1.3610647565249702e-05, "loss": 3.0398, "step": 149690 }, { "epoch": 0.004864, "grad_norm": 0.7576727867126465, "learning_rate": 1.3609896598553522e-05, "loss": 3.0118, "step": 149700 }, { "epoch": 0.0048896, "grad_norm": 0.6981201767921448, "learning_rate": 1.3609145608448126e-05, "loss": 2.9635, "step": 149710 }, { "epoch": 0.0049152, "grad_norm": 0.7329005002975464, "learning_rate": 1.3608394594938377e-05, "loss": 2.9593, "step": 149720 }, { "epoch": 0.0049408, "grad_norm": 0.6493296027183533, "learning_rate": 1.3607643558029149e-05, "loss": 2.8173, "step": 149730 }, { "epoch": 0.0049664, "grad_norm": 0.6562065482139587, "learning_rate": 1.360689249772531e-05, "loss": 2.8974, "step": 149740 }, { "epoch": 0.004992, "grad_norm": 0.7002583146095276, "learning_rate": 1.3606141414031733e-05, "loss": 2.9625, "step": 149750 }, { "epoch": 0.0050176, "grad_norm": 0.7658461928367615, "learning_rate": 1.3605390306953283e-05, "loss": 2.8465, "step": 149760 }, { "epoch": 0.0050432, "grad_norm": 0.6711487174034119, "learning_rate": 1.360463917649484e-05, "loss": 3.0248, "step": 149770 }, { "epoch": 0.0050688, "grad_norm": 1.218629240989685, "learning_rate": 1.360388802266127e-05, "loss": 3.5192, "step": 149780 }, { "epoch": 0.0050944, "grad_norm": 0.7170151472091675, "learning_rate": 1.3603136845457441e-05, "loss": 2.8753, "step": 149790 }, { "epoch": 0.00512, "grad_norm": 0.7133424878120422, "learning_rate": 1.3602385644888227e-05, "loss": 2.8954, "step": 149800 }, { "epoch": 0.0051456, "grad_norm": 0.647879421710968, "learning_rate": 1.3601634420958505e-05, "loss": 2.8146, "step": 149810 }, { "epoch": 0.0051712, "grad_norm": 0.7162283658981323, "learning_rate": 1.3600883173673136e-05, "loss": 2.9405, "step": 149820 }, { "epoch": 0.0051968, "grad_norm": 0.8170652389526367, "learning_rate": 1.3600131903036997e-05, "loss": 3.0292, "step": 149830 }, { "epoch": 0.0052224, "grad_norm": 0.7080100178718567, "learning_rate": 1.3599380609054962e-05, "loss": 2.8154, "step": 149840 }, { "epoch": 0.005248, "grad_norm": 0.7593753933906555, "learning_rate": 1.3598629291731898e-05, "loss": 3.1431, "step": 149850 }, { "epoch": 0.0052736, "grad_norm": 0.6946579217910767, "learning_rate": 1.3597877951072677e-05, "loss": 3.0284, "step": 149860 }, { "epoch": 0.0052992, "grad_norm": 0.7407926321029663, "learning_rate": 1.3597126587082178e-05, "loss": 2.8647, "step": 149870 }, { "epoch": 0.0053248, "grad_norm": 0.7383136749267578, "learning_rate": 1.3596375199765267e-05, "loss": 3.0726, "step": 149880 }, { "epoch": 0.0053504, "grad_norm": 0.7992581725120544, "learning_rate": 1.3595623789126821e-05, "loss": 3.0018, "step": 149890 }, { "epoch": 0.005376, "grad_norm": 0.6860992908477783, "learning_rate": 1.3594872355171709e-05, "loss": 2.7361, "step": 149900 }, { "epoch": 0.0054016, "grad_norm": 0.8448092937469482, "learning_rate": 1.3594120897904806e-05, "loss": 2.9888, "step": 149910 }, { "epoch": 0.0054272, "grad_norm": 0.7445347905158997, "learning_rate": 1.3593369417330981e-05, "loss": 2.8941, "step": 149920 }, { "epoch": 0.0054528, "grad_norm": 0.7517738342285156, "learning_rate": 1.3592617913455114e-05, "loss": 3.0199, "step": 149930 }, { "epoch": 0.0054784, "grad_norm": 0.876920759677887, "learning_rate": 1.359186638628207e-05, "loss": 3.0846, "step": 149940 }, { "epoch": 0.005504, "grad_norm": 0.7228550910949707, "learning_rate": 1.359111483581673e-05, "loss": 2.7714, "step": 149950 }, { "epoch": 0.0055296, "grad_norm": 0.7555521130561829, "learning_rate": 1.3590363262063966e-05, "loss": 3.057, "step": 149960 }, { "epoch": 0.0055552, "grad_norm": 0.7392433285713196, "learning_rate": 1.358961166502865e-05, "loss": 2.9948, "step": 149970 }, { "epoch": 0.0055808, "grad_norm": 0.7582632303237915, "learning_rate": 1.3588860044715656e-05, "loss": 3.1424, "step": 149980 }, { "epoch": 0.0056064, "grad_norm": 0.719718337059021, "learning_rate": 1.3588108401129855e-05, "loss": 3.031, "step": 149990 }, { "epoch": 0.005632, "grad_norm": 0.749502956867218, "learning_rate": 1.3587356734276129e-05, "loss": 3.007, "step": 150000 }, { "epoch": 0.0056576, "grad_norm": 0.8178825378417969, "learning_rate": 1.358660504415935e-05, "loss": 3.0754, "step": 150010 }, { "epoch": 0.0056832, "grad_norm": 0.7738874554634094, "learning_rate": 1.3585853330784384e-05, "loss": 2.8068, "step": 150020 }, { "epoch": 0.0057088, "grad_norm": 0.7829034328460693, "learning_rate": 1.3585101594156115e-05, "loss": 3.2355, "step": 150030 }, { "epoch": 0.0057344, "grad_norm": 0.7213264107704163, "learning_rate": 1.3584349834279415e-05, "loss": 2.8373, "step": 150040 }, { "epoch": 0.00576, "grad_norm": 0.6953907012939453, "learning_rate": 1.358359805115916e-05, "loss": 2.9608, "step": 150050 }, { "epoch": 0.0057856, "grad_norm": 0.6804588437080383, "learning_rate": 1.3582846244800224e-05, "loss": 3.0596, "step": 150060 }, { "epoch": 0.0058112, "grad_norm": 0.9034935235977173, "learning_rate": 1.358209441520748e-05, "loss": 2.8725, "step": 150070 }, { "epoch": 0.0058368, "grad_norm": 0.8576529622077942, "learning_rate": 1.3581342562385809e-05, "loss": 2.8911, "step": 150080 }, { "epoch": 0.0058624, "grad_norm": 0.7472614645957947, "learning_rate": 1.358059068634008e-05, "loss": 3.1642, "step": 150090 }, { "epoch": 0.005888, "grad_norm": 0.6860098242759705, "learning_rate": 1.3579838787075175e-05, "loss": 2.7582, "step": 150100 }, { "epoch": 0.0059136, "grad_norm": 0.7314273118972778, "learning_rate": 1.3579086864595966e-05, "loss": 2.8192, "step": 150110 }, { "epoch": 0.0059392, "grad_norm": 0.7662492394447327, "learning_rate": 1.3578334918907325e-05, "loss": 2.8465, "step": 150120 }, { "epoch": 0.0059648, "grad_norm": 0.8309227824211121, "learning_rate": 1.3577582950014142e-05, "loss": 3.0456, "step": 150130 }, { "epoch": 0.0059904, "grad_norm": 4.2456183433532715, "learning_rate": 1.3576830957921278e-05, "loss": 2.8078, "step": 150140 }, { "epoch": 0.006016, "grad_norm": 0.8131573796272278, "learning_rate": 1.3576078942633617e-05, "loss": 3.1473, "step": 150150 }, { "epoch": 0.0060416, "grad_norm": 0.6927598118782043, "learning_rate": 1.3575326904156034e-05, "loss": 2.9188, "step": 150160 }, { "epoch": 0.0060672, "grad_norm": 0.7302212119102478, "learning_rate": 1.3574574842493409e-05, "loss": 2.8592, "step": 150170 }, { "epoch": 0.0060928, "grad_norm": 0.7476082444190979, "learning_rate": 1.3573822757650614e-05, "loss": 2.9779, "step": 150180 }, { "epoch": 0.0061184, "grad_norm": 0.7361941933631897, "learning_rate": 1.3573070649632529e-05, "loss": 2.9815, "step": 150190 }, { "epoch": 0.006144, "grad_norm": 9.628172874450684, "learning_rate": 1.357231851844403e-05, "loss": 3.1527, "step": 150200 }, { "epoch": 0.0061696, "grad_norm": 0.7626611590385437, "learning_rate": 1.3571566364089995e-05, "loss": 3.0659, "step": 150210 }, { "epoch": 0.0061952, "grad_norm": 0.6417592167854309, "learning_rate": 1.35708141865753e-05, "loss": 2.9848, "step": 150220 }, { "epoch": 0.0062208, "grad_norm": 0.742063581943512, "learning_rate": 1.3570061985904828e-05, "loss": 2.878, "step": 150230 }, { "epoch": 0.0062464, "grad_norm": 0.7634857892990112, "learning_rate": 1.3569309762083449e-05, "loss": 3.2163, "step": 150240 }, { "epoch": 0.006272, "grad_norm": 0.651519775390625, "learning_rate": 1.3568557515116046e-05, "loss": 2.8847, "step": 150250 }, { "epoch": 0.0062976, "grad_norm": 0.832029402256012, "learning_rate": 1.3567805245007495e-05, "loss": 3.0203, "step": 150260 }, { "epoch": 0.0063232, "grad_norm": 0.8496564626693726, "learning_rate": 1.3567052951762679e-05, "loss": 3.0981, "step": 150270 }, { "epoch": 0.0063488, "grad_norm": 0.8802273869514465, "learning_rate": 1.3566300635386464e-05, "loss": 3.023, "step": 150280 }, { "epoch": 0.0063744, "grad_norm": 0.6970723867416382, "learning_rate": 1.3565548295883744e-05, "loss": 2.9256, "step": 150290 }, { "epoch": 0.0064, "grad_norm": 0.7001450061798096, "learning_rate": 1.356479593325939e-05, "loss": 2.8413, "step": 150300 }, { "epoch": 0.0064256, "grad_norm": 0.7635965347290039, "learning_rate": 1.3564043547518282e-05, "loss": 3.1373, "step": 150310 }, { "epoch": 0.0064512, "grad_norm": 0.745324969291687, "learning_rate": 1.3563291138665297e-05, "loss": 2.8585, "step": 150320 }, { "epoch": 0.0064768, "grad_norm": 0.6950260996818542, "learning_rate": 1.3562538706705319e-05, "loss": 2.752, "step": 150330 }, { "epoch": 0.0065024, "grad_norm": 0.7623053789138794, "learning_rate": 1.3561786251643224e-05, "loss": 2.8496, "step": 150340 }, { "epoch": 0.006528, "grad_norm": 0.8236132264137268, "learning_rate": 1.3561033773483889e-05, "loss": 3.1546, "step": 150350 }, { "epoch": 0.0065536, "grad_norm": 0.7111491560935974, "learning_rate": 1.3560281272232199e-05, "loss": 3.0151, "step": 150360 }, { "epoch": 0.0065792, "grad_norm": 0.7597302794456482, "learning_rate": 1.3559528747893029e-05, "loss": 2.9887, "step": 150370 }, { "epoch": 0.0066048, "grad_norm": 0.8508371710777283, "learning_rate": 1.3558776200471263e-05, "loss": 2.9723, "step": 150380 }, { "epoch": 0.0066304, "grad_norm": 0.7260265946388245, "learning_rate": 1.3558023629971775e-05, "loss": 2.8653, "step": 150390 }, { "epoch": 0.006656, "grad_norm": 0.7849753499031067, "learning_rate": 1.3557271036399456e-05, "loss": 2.9879, "step": 150400 }, { "epoch": 0.0066816, "grad_norm": 0.7251006364822388, "learning_rate": 1.3556518419759175e-05, "loss": 2.9555, "step": 150410 }, { "epoch": 0.0067072, "grad_norm": 0.8172610402107239, "learning_rate": 1.3555765780055818e-05, "loss": 3.0648, "step": 150420 }, { "epoch": 0.0067328, "grad_norm": 0.6733943819999695, "learning_rate": 1.3555013117294267e-05, "loss": 3.0333, "step": 150430 }, { "epoch": 0.0067584, "grad_norm": 0.7136964797973633, "learning_rate": 1.35542604314794e-05, "loss": 2.8748, "step": 150440 }, { "epoch": 0.006784, "grad_norm": 0.6507987976074219, "learning_rate": 1.3553507722616095e-05, "loss": 2.8178, "step": 150450 }, { "epoch": 0.0068096, "grad_norm": 0.7554211020469666, "learning_rate": 1.355275499070924e-05, "loss": 3.0492, "step": 150460 }, { "epoch": 0.0068352, "grad_norm": 0.7219202518463135, "learning_rate": 1.3552002235763713e-05, "loss": 2.9464, "step": 150470 }, { "epoch": 0.0068608, "grad_norm": 0.841537594795227, "learning_rate": 1.3551249457784395e-05, "loss": 2.9074, "step": 150480 }, { "epoch": 0.0068864, "grad_norm": 0.783624529838562, "learning_rate": 1.3550496656776168e-05, "loss": 3.1251, "step": 150490 }, { "epoch": 0.006912, "grad_norm": 0.6615675091743469, "learning_rate": 1.3549743832743916e-05, "loss": 2.8675, "step": 150500 }, { "epoch": 0.0069376, "grad_norm": 0.7738893032073975, "learning_rate": 1.3548990985692516e-05, "loss": 3.087, "step": 150510 }, { "epoch": 0.0069632, "grad_norm": 0.6930506825447083, "learning_rate": 1.3548238115626852e-05, "loss": 2.7852, "step": 150520 }, { "epoch": 0.0069888, "grad_norm": 2.077134132385254, "learning_rate": 1.3547485222551811e-05, "loss": 3.2673, "step": 150530 }, { "epoch": 0.0070144, "grad_norm": 0.734059751033783, "learning_rate": 1.3546732306472266e-05, "loss": 2.9393, "step": 150540 }, { "epoch": 0.00704, "grad_norm": 0.7850819826126099, "learning_rate": 1.3545979367393107e-05, "loss": 2.9311, "step": 150550 }, { "epoch": 0.0070656, "grad_norm": 0.6766937375068665, "learning_rate": 1.3545226405319216e-05, "loss": 2.9119, "step": 150560 }, { "epoch": 0.0070912, "grad_norm": 0.7031638026237488, "learning_rate": 1.3544473420255471e-05, "loss": 2.7748, "step": 150570 }, { "epoch": 0.0071168, "grad_norm": 0.7286983728408813, "learning_rate": 1.354372041220676e-05, "loss": 3.0767, "step": 150580 }, { "epoch": 0.0071424, "grad_norm": 0.8704716563224792, "learning_rate": 1.3542967381177962e-05, "loss": 3.0536, "step": 150590 }, { "epoch": 0.007168, "grad_norm": 0.6791532039642334, "learning_rate": 1.3542214327173966e-05, "loss": 3.0676, "step": 150600 }, { "epoch": 0.0071936, "grad_norm": 0.7378979325294495, "learning_rate": 1.3541461250199647e-05, "loss": 2.8433, "step": 150610 }, { "epoch": 0.0072192, "grad_norm": 0.7813800573348999, "learning_rate": 1.3540708150259895e-05, "loss": 3.0647, "step": 150620 }, { "epoch": 0.0072448, "grad_norm": 0.7096885442733765, "learning_rate": 1.353995502735959e-05, "loss": 3.1445, "step": 150630 }, { "epoch": 0.0072704, "grad_norm": 0.8536785840988159, "learning_rate": 1.3539201881503619e-05, "loss": 3.075, "step": 150640 }, { "epoch": 0.007296, "grad_norm": 0.6805723905563354, "learning_rate": 1.3538448712696863e-05, "loss": 3.0951, "step": 150650 }, { "epoch": 0.0073216, "grad_norm": 0.8969608545303345, "learning_rate": 1.3537695520944211e-05, "loss": 3.0494, "step": 150660 }, { "epoch": 0.0073472, "grad_norm": 0.8111087679862976, "learning_rate": 1.3536942306250543e-05, "loss": 2.8691, "step": 150670 }, { "epoch": 0.0073728, "grad_norm": 0.7484108209609985, "learning_rate": 1.353618906862074e-05, "loss": 3.0815, "step": 150680 }, { "epoch": 0.0073984, "grad_norm": 0.8659794926643372, "learning_rate": 1.3535435808059694e-05, "loss": 3.0302, "step": 150690 }, { "epoch": 0.007424, "grad_norm": 0.7401551604270935, "learning_rate": 1.3534682524572287e-05, "loss": 2.8671, "step": 150700 }, { "epoch": 2.56e-05, "grad_norm": 1.3122862577438354, "learning_rate": 1.3533929218163402e-05, "loss": 2.0538, "step": 150710 }, { "epoch": 5.12e-05, "grad_norm": 0.8779069185256958, "learning_rate": 1.3533175888837924e-05, "loss": 2.8035, "step": 150720 }, { "epoch": 7.68e-05, "grad_norm": 0.7132169008255005, "learning_rate": 1.353242253660074e-05, "loss": 2.9056, "step": 150730 }, { "epoch": 0.0001024, "grad_norm": 1.5305137634277344, "learning_rate": 1.3531669161456736e-05, "loss": 2.695, "step": 150740 }, { "epoch": 0.000128, "grad_norm": 0.7896769642829895, "learning_rate": 1.3530915763410794e-05, "loss": 2.679, "step": 150750 }, { "epoch": 0.0001536, "grad_norm": 0.7605023980140686, "learning_rate": 1.3530162342467805e-05, "loss": 2.6467, "step": 150760 }, { "epoch": 0.0001792, "grad_norm": 0.7692933678627014, "learning_rate": 1.3529408898632648e-05, "loss": 2.9054, "step": 150770 }, { "epoch": 0.0002048, "grad_norm": 0.9053195118904114, "learning_rate": 1.3528655431910214e-05, "loss": 2.6928, "step": 150780 }, { "epoch": 0.0002304, "grad_norm": 0.7177989482879639, "learning_rate": 1.3527901942305386e-05, "loss": 2.6542, "step": 150790 }, { "epoch": 0.000256, "grad_norm": 0.9707161784172058, "learning_rate": 1.3527148429823053e-05, "loss": 2.8529, "step": 150800 }, { "epoch": 0.0002816, "grad_norm": 0.7239887714385986, "learning_rate": 1.3526394894468098e-05, "loss": 2.8784, "step": 150810 }, { "epoch": 0.0003072, "grad_norm": 0.7852709889411926, "learning_rate": 1.3525641336245412e-05, "loss": 2.6601, "step": 150820 }, { "epoch": 0.0003328, "grad_norm": 1.159462571144104, "learning_rate": 1.3524887755159877e-05, "loss": 2.9636, "step": 150830 }, { "epoch": 0.0003584, "grad_norm": 0.8084529042243958, "learning_rate": 1.3524134151216384e-05, "loss": 3.0671, "step": 150840 }, { "epoch": 0.000384, "grad_norm": 0.7349879145622253, "learning_rate": 1.3523380524419814e-05, "loss": 2.8232, "step": 150850 }, { "epoch": 0.0004096, "grad_norm": 0.9598551392555237, "learning_rate": 1.3522626874775063e-05, "loss": 2.7881, "step": 150860 }, { "epoch": 0.0004352, "grad_norm": 0.9627137184143066, "learning_rate": 1.3521873202287009e-05, "loss": 2.8623, "step": 150870 }, { "epoch": 0.0004608, "grad_norm": 0.7027792930603027, "learning_rate": 1.3521119506960542e-05, "loss": 3.4475, "step": 150880 }, { "epoch": 0.0004864, "grad_norm": 0.8022890090942383, "learning_rate": 1.3520365788800551e-05, "loss": 3.0188, "step": 150890 }, { "epoch": 0.000512, "grad_norm": 0.7770183682441711, "learning_rate": 1.3519612047811925e-05, "loss": 2.8096, "step": 150900 }, { "epoch": 0.0005376, "grad_norm": 0.6690129637718201, "learning_rate": 1.3518858283999551e-05, "loss": 2.7534, "step": 150910 }, { "epoch": 0.0005632, "grad_norm": 0.777775228023529, "learning_rate": 1.3518104497368319e-05, "loss": 2.7103, "step": 150920 }, { "epoch": 0.0005888, "grad_norm": 0.7731890082359314, "learning_rate": 1.3517350687923107e-05, "loss": 2.7033, "step": 150930 }, { "epoch": 0.0006144, "grad_norm": 0.6800152063369751, "learning_rate": 1.3516596855668815e-05, "loss": 2.6907, "step": 150940 }, { "epoch": 0.00064, "grad_norm": 0.707027792930603, "learning_rate": 1.3515843000610327e-05, "loss": 2.7015, "step": 150950 }, { "epoch": 0.0006656, "grad_norm": 0.7492449879646301, "learning_rate": 1.3515089122752532e-05, "loss": 3.0135, "step": 150960 }, { "epoch": 0.0006912, "grad_norm": 0.6911273002624512, "learning_rate": 1.3514335222100319e-05, "loss": 3.3089, "step": 150970 }, { "epoch": 0.0007168, "grad_norm": 0.7053622007369995, "learning_rate": 1.3513581298658571e-05, "loss": 3.116, "step": 150980 }, { "epoch": 0.0007424, "grad_norm": 0.7446701526641846, "learning_rate": 1.3512827352432187e-05, "loss": 2.9655, "step": 150990 }, { "epoch": 0.000768, "grad_norm": 0.919447660446167, "learning_rate": 1.351207338342605e-05, "loss": 2.8362, "step": 151000 }, { "epoch": 0.0007936, "grad_norm": 0.7127045392990112, "learning_rate": 1.3511319391645047e-05, "loss": 2.8113, "step": 151010 }, { "epoch": 0.0008192, "grad_norm": 0.732657790184021, "learning_rate": 1.3510565377094077e-05, "loss": 2.8495, "step": 151020 }, { "epoch": 0.0008448, "grad_norm": 0.6358503699302673, "learning_rate": 1.3509811339778021e-05, "loss": 2.7859, "step": 151030 }, { "epoch": 0.0008704, "grad_norm": 0.774202823638916, "learning_rate": 1.3509057279701772e-05, "loss": 2.826, "step": 151040 }, { "epoch": 0.000896, "grad_norm": 0.7313756942749023, "learning_rate": 1.3508303196870215e-05, "loss": 2.8633, "step": 151050 }, { "epoch": 0.0009216, "grad_norm": 0.6780301928520203, "learning_rate": 1.350754909128825e-05, "loss": 2.8401, "step": 151060 }, { "epoch": 0.0009472, "grad_norm": 0.6675123572349548, "learning_rate": 1.3506794962960756e-05, "loss": 2.9034, "step": 151070 }, { "epoch": 0.0009728, "grad_norm": 0.7753456234931946, "learning_rate": 1.3506040811892632e-05, "loss": 2.8756, "step": 151080 }, { "epoch": 0.0009984, "grad_norm": 0.6762017011642456, "learning_rate": 1.3505286638088767e-05, "loss": 2.6463, "step": 151090 }, { "epoch": 0.001024, "grad_norm": 0.7376872301101685, "learning_rate": 1.3504532441554044e-05, "loss": 2.6828, "step": 151100 }, { "epoch": 0.0010496, "grad_norm": 0.8347638249397278, "learning_rate": 1.3503778222293362e-05, "loss": 2.9707, "step": 151110 }, { "epoch": 0.0010752, "grad_norm": 0.7925242185592651, "learning_rate": 1.3503023980311612e-05, "loss": 2.8563, "step": 151120 }, { "epoch": 0.0011008, "grad_norm": 0.7163968086242676, "learning_rate": 1.3502269715613681e-05, "loss": 3.0114, "step": 151130 }, { "epoch": 0.0011264, "grad_norm": 0.6442728638648987, "learning_rate": 1.350151542820446e-05, "loss": 2.8571, "step": 151140 }, { "epoch": 0.001152, "grad_norm": 0.6782138347625732, "learning_rate": 1.3500761118088844e-05, "loss": 2.7859, "step": 151150 }, { "epoch": 0.0011776, "grad_norm": 0.7909648418426514, "learning_rate": 1.350000678527172e-05, "loss": 3.1366, "step": 151160 }, { "epoch": 0.0012032, "grad_norm": 0.7769089341163635, "learning_rate": 1.3499252429757983e-05, "loss": 2.9907, "step": 151170 }, { "epoch": 0.0012288, "grad_norm": 0.6828466057777405, "learning_rate": 1.3498498051552527e-05, "loss": 3.0935, "step": 151180 }, { "epoch": 0.0012544, "grad_norm": 0.8592311143875122, "learning_rate": 1.3497743650660238e-05, "loss": 2.9327, "step": 151190 }, { "epoch": 0.00128, "grad_norm": 0.7156491875648499, "learning_rate": 1.3496989227086012e-05, "loss": 2.794, "step": 151200 }, { "epoch": 0.0013056, "grad_norm": 0.731674313545227, "learning_rate": 1.349623478083474e-05, "loss": 2.8947, "step": 151210 }, { "epoch": 0.0013312, "grad_norm": 0.7114081382751465, "learning_rate": 1.3495480311911316e-05, "loss": 2.5502, "step": 151220 }, { "epoch": 0.0013568, "grad_norm": 0.7310649752616882, "learning_rate": 1.3494725820320631e-05, "loss": 2.6525, "step": 151230 }, { "epoch": 0.0013824, "grad_norm": 0.8016746044158936, "learning_rate": 1.3493971306067572e-05, "loss": 3.2063, "step": 151240 }, { "epoch": 0.001408, "grad_norm": 0.8108059167861938, "learning_rate": 1.3493216769157045e-05, "loss": 2.8625, "step": 151250 }, { "epoch": 0.0014336, "grad_norm": 0.7735768556594849, "learning_rate": 1.3492462209593933e-05, "loss": 2.8374, "step": 151260 }, { "epoch": 0.0014592, "grad_norm": 0.7368446588516235, "learning_rate": 1.349170762738313e-05, "loss": 2.8191, "step": 151270 }, { "epoch": 0.0014848, "grad_norm": 0.7450923919677734, "learning_rate": 1.3490953022529533e-05, "loss": 2.669, "step": 151280 }, { "epoch": 0.0015104, "grad_norm": 0.7251847982406616, "learning_rate": 1.3490198395038036e-05, "loss": 2.5856, "step": 151290 }, { "epoch": 0.001536, "grad_norm": 0.7992613911628723, "learning_rate": 1.3489443744913526e-05, "loss": 3.0668, "step": 151300 }, { "epoch": 2.56e-05, "grad_norm": 0.7946096062660217, "learning_rate": 1.3488689072160904e-05, "loss": 2.2994, "step": 151310 }, { "epoch": 5.12e-05, "grad_norm": 0.7469306588172913, "learning_rate": 1.348793437678506e-05, "loss": 2.8464, "step": 151320 }, { "epoch": 7.68e-05, "grad_norm": 0.6847732067108154, "learning_rate": 1.3487179658790884e-05, "loss": 2.8681, "step": 151330 }, { "epoch": 0.0001024, "grad_norm": 0.7415357232093811, "learning_rate": 1.3486424918183277e-05, "loss": 3.1572, "step": 151340 }, { "epoch": 0.000128, "grad_norm": 0.7574542164802551, "learning_rate": 1.3485670154967132e-05, "loss": 2.6277, "step": 151350 }, { "epoch": 0.0001536, "grad_norm": 0.8392378687858582, "learning_rate": 1.3484915369147343e-05, "loss": 2.932, "step": 151360 }, { "epoch": 0.0001792, "grad_norm": 0.7188478708267212, "learning_rate": 1.3484160560728802e-05, "loss": 2.764, "step": 151370 }, { "epoch": 0.0002048, "grad_norm": 1.2680573463439941, "learning_rate": 1.3483405729716408e-05, "loss": 3.0387, "step": 151380 }, { "epoch": 0.0002304, "grad_norm": 0.9700928926467896, "learning_rate": 1.3482650876115053e-05, "loss": 3.0767, "step": 151390 }, { "epoch": 0.000256, "grad_norm": 0.7615792155265808, "learning_rate": 1.3481895999929631e-05, "loss": 2.7307, "step": 151400 }, { "epoch": 0.0002816, "grad_norm": 0.9118419885635376, "learning_rate": 1.3481141101165039e-05, "loss": 2.615, "step": 151410 }, { "epoch": 0.0003072, "grad_norm": 0.720130205154419, "learning_rate": 1.3480386179826174e-05, "loss": 2.8022, "step": 151420 }, { "epoch": 0.0003328, "grad_norm": 1.0961211919784546, "learning_rate": 1.347963123591793e-05, "loss": 2.9641, "step": 151430 }, { "epoch": 0.0003584, "grad_norm": 0.8292062878608704, "learning_rate": 1.3478876269445199e-05, "loss": 2.827, "step": 151440 }, { "epoch": 0.000384, "grad_norm": 1.0773701667785645, "learning_rate": 1.3478121280412884e-05, "loss": 2.8794, "step": 151450 }, { "epoch": 0.0004096, "grad_norm": 0.7169719934463501, "learning_rate": 1.3477366268825873e-05, "loss": 2.8135, "step": 151460 }, { "epoch": 0.0004352, "grad_norm": 0.7169512510299683, "learning_rate": 1.3476611234689068e-05, "loss": 2.8523, "step": 151470 }, { "epoch": 0.0004608, "grad_norm": 0.7183415293693542, "learning_rate": 1.3475856178007362e-05, "loss": 2.7516, "step": 151480 }, { "epoch": 0.0004864, "grad_norm": 0.739408552646637, "learning_rate": 1.3475101098785656e-05, "loss": 2.9076, "step": 151490 }, { "epoch": 0.000512, "grad_norm": 0.7203744649887085, "learning_rate": 1.3474345997028837e-05, "loss": 2.948, "step": 151500 }, { "epoch": 0.0005376, "grad_norm": 0.8000878095626831, "learning_rate": 1.3473590872741814e-05, "loss": 2.711, "step": 151510 }, { "epoch": 0.0005632, "grad_norm": 0.7230044603347778, "learning_rate": 1.3472835725929472e-05, "loss": 2.5412, "step": 151520 }, { "epoch": 0.0005888, "grad_norm": 0.7565807104110718, "learning_rate": 1.3472080556596716e-05, "loss": 2.8508, "step": 151530 }, { "epoch": 0.0006144, "grad_norm": 0.679448664188385, "learning_rate": 1.347132536474844e-05, "loss": 2.7904, "step": 151540 }, { "epoch": 0.00064, "grad_norm": 0.7232788801193237, "learning_rate": 1.3470570150389542e-05, "loss": 3.065, "step": 151550 }, { "epoch": 0.0006656, "grad_norm": 0.7107081413269043, "learning_rate": 1.3469814913524916e-05, "loss": 3.2462, "step": 151560 }, { "epoch": 0.0006912, "grad_norm": 0.7213209271430969, "learning_rate": 1.3469059654159467e-05, "loss": 2.9444, "step": 151570 }, { "epoch": 0.0007168, "grad_norm": 0.7190524339675903, "learning_rate": 1.3468304372298084e-05, "loss": 2.9087, "step": 151580 }, { "epoch": 0.0007424, "grad_norm": 0.7979332208633423, "learning_rate": 1.3467549067945671e-05, "loss": 2.8255, "step": 151590 }, { "epoch": 0.000768, "grad_norm": 0.8551862239837646, "learning_rate": 1.3466793741107122e-05, "loss": 3.1286, "step": 151600 }, { "epoch": 2.56e-05, "grad_norm": 0.7467971444129944, "learning_rate": 1.3466038391787341e-05, "loss": 2.3713, "step": 151610 }, { "epoch": 5.12e-05, "grad_norm": 0.7912241816520691, "learning_rate": 1.3465283019991219e-05, "loss": 2.8712, "step": 151620 }, { "epoch": 7.68e-05, "grad_norm": 0.8291336297988892, "learning_rate": 1.346452762572366e-05, "loss": 2.725, "step": 151630 }, { "epoch": 0.0001024, "grad_norm": 0.7827492356300354, "learning_rate": 1.3463772208989557e-05, "loss": 2.9146, "step": 151640 }, { "epoch": 0.000128, "grad_norm": 0.7071890830993652, "learning_rate": 1.3463016769793815e-05, "loss": 2.5815, "step": 151650 }, { "epoch": 0.0001536, "grad_norm": 0.6939826011657715, "learning_rate": 1.3462261308141326e-05, "loss": 2.9502, "step": 151660 }, { "epoch": 0.0001792, "grad_norm": 0.9298929572105408, "learning_rate": 1.3461505824036996e-05, "loss": 2.8176, "step": 151670 }, { "epoch": 0.0002048, "grad_norm": 1.3818434476852417, "learning_rate": 1.3460750317485718e-05, "loss": 2.8446, "step": 151680 }, { "epoch": 0.0002304, "grad_norm": 0.8658328652381897, "learning_rate": 1.3459994788492395e-05, "loss": 2.8038, "step": 151690 }, { "epoch": 0.000256, "grad_norm": 0.7653952836990356, "learning_rate": 1.3459239237061925e-05, "loss": 2.7447, "step": 151700 }, { "epoch": 0.0002816, "grad_norm": 0.7590574026107788, "learning_rate": 1.3458483663199208e-05, "loss": 2.7805, "step": 151710 }, { "epoch": 0.0003072, "grad_norm": 0.8169260621070862, "learning_rate": 1.3457728066909147e-05, "loss": 2.6301, "step": 151720 }, { "epoch": 0.0003328, "grad_norm": 0.8340070247650146, "learning_rate": 1.3456972448196632e-05, "loss": 2.9649, "step": 151730 }, { "epoch": 0.0003584, "grad_norm": 0.8072484731674194, "learning_rate": 1.3456216807066574e-05, "loss": 2.9071, "step": 151740 }, { "epoch": 0.000384, "grad_norm": 0.7798853516578674, "learning_rate": 1.345546114352387e-05, "loss": 2.6449, "step": 151750 }, { "epoch": 0.0004096, "grad_norm": 0.711077094078064, "learning_rate": 1.3454705457573414e-05, "loss": 2.8434, "step": 151760 }, { "epoch": 0.0004352, "grad_norm": 0.7440125346183777, "learning_rate": 1.3453949749220113e-05, "loss": 2.9067, "step": 151770 }, { "epoch": 0.0004608, "grad_norm": 0.8454040288925171, "learning_rate": 1.3453194018468868e-05, "loss": 2.8782, "step": 151780 }, { "epoch": 0.0004864, "grad_norm": 0.7775565981864929, "learning_rate": 1.3452438265324574e-05, "loss": 2.9905, "step": 151790 }, { "epoch": 0.000512, "grad_norm": 0.7238280177116394, "learning_rate": 1.3451682489792137e-05, "loss": 2.6487, "step": 151800 }, { "epoch": 0.0005376, "grad_norm": 0.7640389204025269, "learning_rate": 1.3450926691876459e-05, "loss": 2.6954, "step": 151810 }, { "epoch": 0.0005632, "grad_norm": 0.744608461856842, "learning_rate": 1.3450170871582435e-05, "loss": 2.6384, "step": 151820 }, { "epoch": 0.0005888, "grad_norm": 0.8112512826919556, "learning_rate": 1.3449415028914969e-05, "loss": 2.6924, "step": 151830 }, { "epoch": 0.0006144, "grad_norm": 0.8624919652938843, "learning_rate": 1.3448659163878968e-05, "loss": 2.7357, "step": 151840 }, { "epoch": 0.00064, "grad_norm": 0.7322348356246948, "learning_rate": 1.3447903276479323e-05, "loss": 2.6565, "step": 151850 }, { "epoch": 0.0006656, "grad_norm": 0.7376341223716736, "learning_rate": 1.3447147366720945e-05, "loss": 3.2542, "step": 151860 }, { "epoch": 0.0006912, "grad_norm": 0.6372274160385132, "learning_rate": 1.3446391434608732e-05, "loss": 2.6699, "step": 151870 }, { "epoch": 0.0007168, "grad_norm": 0.90547776222229, "learning_rate": 1.344563548014759e-05, "loss": 2.7232, "step": 151880 }, { "epoch": 0.0007424, "grad_norm": 0.7356379628181458, "learning_rate": 1.3444879503342412e-05, "loss": 2.9184, "step": 151890 }, { "epoch": 0.000768, "grad_norm": 0.9730448126792908, "learning_rate": 1.3444123504198107e-05, "loss": 2.8067, "step": 151900 }, { "epoch": 0.0007936, "grad_norm": 0.7094584703445435, "learning_rate": 1.3443367482719581e-05, "loss": 2.8909, "step": 151910 }, { "epoch": 0.0008192, "grad_norm": 0.8174322843551636, "learning_rate": 1.3442611438911729e-05, "loss": 2.9894, "step": 151920 }, { "epoch": 0.0008448, "grad_norm": 0.7139805555343628, "learning_rate": 1.3441855372779455e-05, "loss": 2.7343, "step": 151930 }, { "epoch": 0.0008704, "grad_norm": 0.6673411726951599, "learning_rate": 1.3441099284327666e-05, "loss": 2.5828, "step": 151940 }, { "epoch": 0.000896, "grad_norm": 0.6894097328186035, "learning_rate": 1.344034317356126e-05, "loss": 2.649, "step": 151950 }, { "epoch": 0.0009216, "grad_norm": 0.6458552479743958, "learning_rate": 1.3439587040485145e-05, "loss": 2.8514, "step": 151960 }, { "epoch": 0.0009472, "grad_norm": 0.6766388416290283, "learning_rate": 1.3438830885104222e-05, "loss": 2.8984, "step": 151970 }, { "epoch": 0.0009728, "grad_norm": 0.8119925856590271, "learning_rate": 1.3438074707423396e-05, "loss": 2.7181, "step": 151980 }, { "epoch": 0.0009984, "grad_norm": 0.7551708817481995, "learning_rate": 1.3437318507447568e-05, "loss": 2.7446, "step": 151990 }, { "epoch": 0.001024, "grad_norm": 0.8039174675941467, "learning_rate": 1.3436562285181642e-05, "loss": 2.7408, "step": 152000 }, { "epoch": 0.0010496, "grad_norm": 0.7960211038589478, "learning_rate": 1.3435806040630523e-05, "loss": 2.8931, "step": 152010 }, { "epoch": 0.0010752, "grad_norm": 0.7694637775421143, "learning_rate": 1.3435049773799116e-05, "loss": 2.8182, "step": 152020 }, { "epoch": 0.0011008, "grad_norm": 0.6944838762283325, "learning_rate": 1.3434293484692322e-05, "loss": 2.9218, "step": 152030 }, { "epoch": 0.0011264, "grad_norm": 0.6948977112770081, "learning_rate": 1.3433537173315049e-05, "loss": 2.7618, "step": 152040 }, { "epoch": 0.001152, "grad_norm": 0.858274519443512, "learning_rate": 1.3432780839672202e-05, "loss": 2.962, "step": 152050 }, { "epoch": 0.0011776, "grad_norm": 0.7821109890937805, "learning_rate": 1.3432024483768682e-05, "loss": 3.0516, "step": 152060 }, { "epoch": 0.0012032, "grad_norm": 0.7028017640113831, "learning_rate": 1.3431268105609395e-05, "loss": 2.7798, "step": 152070 }, { "epoch": 0.0012288, "grad_norm": 0.6812482476234436, "learning_rate": 1.3430511705199247e-05, "loss": 3.0354, "step": 152080 }, { "epoch": 0.0012544, "grad_norm": 0.8078558444976807, "learning_rate": 1.3429755282543142e-05, "loss": 2.8182, "step": 152090 }, { "epoch": 0.00128, "grad_norm": 0.7274227738380432, "learning_rate": 1.3428998837645983e-05, "loss": 2.7, "step": 152100 }, { "epoch": 0.0013056, "grad_norm": 0.7376326322555542, "learning_rate": 1.342824237051268e-05, "loss": 2.7968, "step": 152110 }, { "epoch": 0.0013312, "grad_norm": 0.9699748754501343, "learning_rate": 1.3427485881148136e-05, "loss": 2.8664, "step": 152120 }, { "epoch": 0.0013568, "grad_norm": 1.2728017568588257, "learning_rate": 1.3426729369557259e-05, "loss": 3.1731, "step": 152130 }, { "epoch": 0.0013824, "grad_norm": 0.6470882296562195, "learning_rate": 1.342597283574495e-05, "loss": 2.8977, "step": 152140 }, { "epoch": 0.001408, "grad_norm": 0.6698575615882874, "learning_rate": 1.342521627971612e-05, "loss": 2.8699, "step": 152150 }, { "epoch": 0.0014336, "grad_norm": 0.6659584641456604, "learning_rate": 1.3424459701475671e-05, "loss": 2.7245, "step": 152160 }, { "epoch": 0.0014592, "grad_norm": 0.7176017761230469, "learning_rate": 1.3423703101028511e-05, "loss": 2.704, "step": 152170 }, { "epoch": 0.0014848, "grad_norm": 0.7167057991027832, "learning_rate": 1.3422946478379549e-05, "loss": 2.95, "step": 152180 }, { "epoch": 0.0015104, "grad_norm": 0.7188296318054199, "learning_rate": 1.3422189833533685e-05, "loss": 2.6969, "step": 152190 }, { "epoch": 0.001536, "grad_norm": 0.706889271736145, "learning_rate": 1.342143316649583e-05, "loss": 2.7915, "step": 152200 }, { "epoch": 2.56e-05, "grad_norm": 0.7227145433425903, "learning_rate": 1.3420676477270893e-05, "loss": 2.2976, "step": 152210 }, { "epoch": 5.12e-05, "grad_norm": 0.8055666089057922, "learning_rate": 1.3419919765863776e-05, "loss": 2.888, "step": 152220 }, { "epoch": 7.68e-05, "grad_norm": 0.6577663421630859, "learning_rate": 1.3419163032279388e-05, "loss": 2.8093, "step": 152230 }, { "epoch": 0.0001024, "grad_norm": 0.7716017365455627, "learning_rate": 1.341840627652264e-05, "loss": 2.7739, "step": 152240 }, { "epoch": 0.000128, "grad_norm": 0.7636255025863647, "learning_rate": 1.341764949859843e-05, "loss": 2.8636, "step": 152250 }, { "epoch": 0.0001536, "grad_norm": 0.8759190440177917, "learning_rate": 1.3416892698511675e-05, "loss": 2.8261, "step": 152260 }, { "epoch": 0.0001792, "grad_norm": 0.7179985046386719, "learning_rate": 1.3416135876267277e-05, "loss": 2.888, "step": 152270 }, { "epoch": 0.0002048, "grad_norm": 0.8249112367630005, "learning_rate": 1.3415379031870148e-05, "loss": 3.1941, "step": 152280 }, { "epoch": 0.0002304, "grad_norm": 0.7492528557777405, "learning_rate": 1.341462216532519e-05, "loss": 2.799, "step": 152290 }, { "epoch": 0.000256, "grad_norm": 0.7223359942436218, "learning_rate": 1.341386527663732e-05, "loss": 2.7207, "step": 152300 }, { "epoch": 0.0002816, "grad_norm": 0.754362165927887, "learning_rate": 1.3413108365811438e-05, "loss": 2.8203, "step": 152310 }, { "epoch": 0.0003072, "grad_norm": 0.8369652628898621, "learning_rate": 1.3412351432852455e-05, "loss": 2.7294, "step": 152320 }, { "epoch": 0.0003328, "grad_norm": 0.6850730180740356, "learning_rate": 1.3411594477765279e-05, "loss": 2.7408, "step": 152330 }, { "epoch": 0.0003584, "grad_norm": 0.8030862808227539, "learning_rate": 1.3410837500554825e-05, "loss": 3.0925, "step": 152340 }, { "epoch": 0.000384, "grad_norm": 1.0760188102722168, "learning_rate": 1.3410080501225991e-05, "loss": 2.8147, "step": 152350 }, { "epoch": 0.0004096, "grad_norm": 0.7164019346237183, "learning_rate": 1.3409323479783692e-05, "loss": 2.804, "step": 152360 }, { "epoch": 0.0004352, "grad_norm": 0.8366197347640991, "learning_rate": 1.3408566436232838e-05, "loss": 3.0954, "step": 152370 }, { "epoch": 0.0004608, "grad_norm": 0.6972530484199524, "learning_rate": 1.3407809370578333e-05, "loss": 3.3986, "step": 152380 }, { "epoch": 0.0004864, "grad_norm": 0.8165735006332397, "learning_rate": 1.3407052282825093e-05, "loss": 2.8707, "step": 152390 }, { "epoch": 0.000512, "grad_norm": 0.7018740773200989, "learning_rate": 1.3406295172978023e-05, "loss": 2.8154, "step": 152400 }, { "epoch": 0.0005376, "grad_norm": 0.7808170914649963, "learning_rate": 1.3405538041042036e-05, "loss": 2.6846, "step": 152410 }, { "epoch": 0.0005632, "grad_norm": 0.7872294187545776, "learning_rate": 1.340478088702204e-05, "loss": 2.7315, "step": 152420 }, { "epoch": 0.0005888, "grad_norm": 0.8036305904388428, "learning_rate": 1.3404023710922942e-05, "loss": 2.7582, "step": 152430 }, { "epoch": 0.0006144, "grad_norm": 0.7703967690467834, "learning_rate": 1.3403266512749658e-05, "loss": 2.7881, "step": 152440 }, { "epoch": 0.00064, "grad_norm": 0.7307029962539673, "learning_rate": 1.3402509292507093e-05, "loss": 2.78, "step": 152450 }, { "epoch": 0.0006656, "grad_norm": 0.7945950031280518, "learning_rate": 1.3401752050200156e-05, "loss": 2.7654, "step": 152460 }, { "epoch": 0.0006912, "grad_norm": 0.7924906015396118, "learning_rate": 1.340099478583377e-05, "loss": 2.8355, "step": 152470 }, { "epoch": 0.0007168, "grad_norm": 0.6739295125007629, "learning_rate": 1.3400237499412828e-05, "loss": 2.7601, "step": 152480 }, { "epoch": 0.0007424, "grad_norm": 0.727257251739502, "learning_rate": 1.3399480190942256e-05, "loss": 2.6184, "step": 152490 }, { "epoch": 0.000768, "grad_norm": 1.0141901969909668, "learning_rate": 1.3398722860426957e-05, "loss": 2.7403, "step": 152500 }, { "epoch": 0.0007936, "grad_norm": 0.6992701888084412, "learning_rate": 1.3397965507871841e-05, "loss": 2.8414, "step": 152510 }, { "epoch": 0.0008192, "grad_norm": 0.7555119395256042, "learning_rate": 1.3397208133281823e-05, "loss": 3.0578, "step": 152520 }, { "epoch": 0.0008448, "grad_norm": 0.6531882286071777, "learning_rate": 1.3396450736661813e-05, "loss": 2.6958, "step": 152530 }, { "epoch": 0.0008704, "grad_norm": 0.6881319284439087, "learning_rate": 1.3395693318016725e-05, "loss": 2.7603, "step": 152540 }, { "epoch": 0.000896, "grad_norm": 0.6416264176368713, "learning_rate": 1.3394935877351462e-05, "loss": 2.7794, "step": 152550 }, { "epoch": 0.0009216, "grad_norm": 0.7420856952667236, "learning_rate": 1.3394178414670947e-05, "loss": 2.8095, "step": 152560 }, { "epoch": 0.0009472, "grad_norm": 0.7615488171577454, "learning_rate": 1.3393420929980088e-05, "loss": 2.9645, "step": 152570 }, { "epoch": 0.0009728, "grad_norm": 1.0068343877792358, "learning_rate": 1.3392663423283794e-05, "loss": 3.1528, "step": 152580 }, { "epoch": 0.0009984, "grad_norm": 0.7472988367080688, "learning_rate": 1.3391905894586978e-05, "loss": 2.8452, "step": 152590 }, { "epoch": 0.001024, "grad_norm": 0.7260757088661194, "learning_rate": 1.3391148343894558e-05, "loss": 2.9525, "step": 152600 }, { "epoch": 0.0010496, "grad_norm": 0.8645016551017761, "learning_rate": 1.3390390771211439e-05, "loss": 2.9701, "step": 152610 }, { "epoch": 0.0010752, "grad_norm": 0.6899016499519348, "learning_rate": 1.3389633176542539e-05, "loss": 2.9414, "step": 152620 }, { "epoch": 0.0011008, "grad_norm": 0.6621735692024231, "learning_rate": 1.3388875559892767e-05, "loss": 2.7018, "step": 152630 }, { "epoch": 0.0011264, "grad_norm": 0.6960548162460327, "learning_rate": 1.3388117921267036e-05, "loss": 2.626, "step": 152640 }, { "epoch": 0.001152, "grad_norm": 0.7703298926353455, "learning_rate": 1.3387360260670263e-05, "loss": 2.9708, "step": 152650 }, { "epoch": 0.0011776, "grad_norm": 0.6697927713394165, "learning_rate": 1.3386602578107358e-05, "loss": 2.7325, "step": 152660 }, { "epoch": 0.0012032, "grad_norm": 0.8500475883483887, "learning_rate": 1.3385844873583238e-05, "loss": 3.1587, "step": 152670 }, { "epoch": 0.0012288, "grad_norm": 0.6795507073402405, "learning_rate": 1.338508714710281e-05, "loss": 2.7177, "step": 152680 }, { "epoch": 0.0012544, "grad_norm": 0.7432292103767395, "learning_rate": 1.3384329398670994e-05, "loss": 3.0793, "step": 152690 }, { "epoch": 0.00128, "grad_norm": 0.671795666217804, "learning_rate": 1.3383571628292702e-05, "loss": 3.0507, "step": 152700 }, { "epoch": 0.0013056, "grad_norm": 0.7900012135505676, "learning_rate": 1.3382813835972846e-05, "loss": 2.6137, "step": 152710 }, { "epoch": 0.0013312, "grad_norm": 0.7512438893318176, "learning_rate": 1.3382056021716339e-05, "loss": 3.0139, "step": 152720 }, { "epoch": 0.0013568, "grad_norm": 0.8096258044242859, "learning_rate": 1.33812981855281e-05, "loss": 2.6073, "step": 152730 }, { "epoch": 0.0013824, "grad_norm": 0.8042914271354675, "learning_rate": 1.3380540327413042e-05, "loss": 2.6362, "step": 152740 }, { "epoch": 0.001408, "grad_norm": 0.6661328077316284, "learning_rate": 1.3379782447376078e-05, "loss": 2.9237, "step": 152750 }, { "epoch": 0.0014336, "grad_norm": 0.7410399317741394, "learning_rate": 1.3379024545422122e-05, "loss": 2.82, "step": 152760 }, { "epoch": 0.0014592, "grad_norm": 0.6953751444816589, "learning_rate": 1.337826662155609e-05, "loss": 2.9496, "step": 152770 }, { "epoch": 0.0014848, "grad_norm": 0.7084395289421082, "learning_rate": 1.3377508675782896e-05, "loss": 2.7828, "step": 152780 }, { "epoch": 0.0015104, "grad_norm": 0.7303781509399414, "learning_rate": 1.3376750708107457e-05, "loss": 2.5701, "step": 152790 }, { "epoch": 0.001536, "grad_norm": 0.8055418133735657, "learning_rate": 1.3375992718534687e-05, "loss": 2.6547, "step": 152800 }, { "epoch": 0.0015616, "grad_norm": 0.7886579632759094, "learning_rate": 1.3375234707069499e-05, "loss": 2.7967, "step": 152810 }, { "epoch": 0.0015872, "grad_norm": 0.7347304224967957, "learning_rate": 1.3374476673716814e-05, "loss": 3.033, "step": 152820 }, { "epoch": 0.0016128, "grad_norm": 0.8395349979400635, "learning_rate": 1.3373718618481543e-05, "loss": 2.9433, "step": 152830 }, { "epoch": 0.0016384, "grad_norm": 0.7227303981781006, "learning_rate": 1.3372960541368607e-05, "loss": 2.9305, "step": 152840 }, { "epoch": 0.001664, "grad_norm": 0.7494807243347168, "learning_rate": 1.3372202442382914e-05, "loss": 2.89, "step": 152850 }, { "epoch": 0.0016896, "grad_norm": 0.7374147772789001, "learning_rate": 1.3371444321529386e-05, "loss": 2.991, "step": 152860 }, { "epoch": 0.0017152, "grad_norm": 0.8813654184341431, "learning_rate": 1.3370686178812937e-05, "loss": 2.9446, "step": 152870 }, { "epoch": 0.0017408, "grad_norm": 1.0319311618804932, "learning_rate": 1.3369928014238485e-05, "loss": 2.755, "step": 152880 }, { "epoch": 0.0017664, "grad_norm": 0.8786868453025818, "learning_rate": 1.3369169827810943e-05, "loss": 2.933, "step": 152890 }, { "epoch": 0.001792, "grad_norm": 0.822763204574585, "learning_rate": 1.3368411619535233e-05, "loss": 3.0784, "step": 152900 }, { "epoch": 0.0018176, "grad_norm": 0.7497866153717041, "learning_rate": 1.3367653389416268e-05, "loss": 2.6054, "step": 152910 }, { "epoch": 0.0018432, "grad_norm": 0.7631179690361023, "learning_rate": 1.3366895137458963e-05, "loss": 2.7805, "step": 152920 }, { "epoch": 0.0018688, "grad_norm": 0.6904217004776001, "learning_rate": 1.3366136863668245e-05, "loss": 2.7507, "step": 152930 }, { "epoch": 0.0018944, "grad_norm": 0.6772449016571045, "learning_rate": 1.336537856804902e-05, "loss": 2.5984, "step": 152940 }, { "epoch": 0.00192, "grad_norm": 0.8853394389152527, "learning_rate": 1.3364620250606206e-05, "loss": 2.8924, "step": 152950 }, { "epoch": 0.0019456, "grad_norm": 0.7337802052497864, "learning_rate": 1.3363861911344726e-05, "loss": 2.7634, "step": 152960 }, { "epoch": 0.0019712, "grad_norm": 0.6568397283554077, "learning_rate": 1.3363103550269498e-05, "loss": 3.0907, "step": 152970 }, { "epoch": 0.0019968, "grad_norm": 0.7915749549865723, "learning_rate": 1.3362345167385434e-05, "loss": 2.7699, "step": 152980 }, { "epoch": 0.0020224, "grad_norm": 0.9796453714370728, "learning_rate": 1.3361586762697456e-05, "loss": 2.8141, "step": 152990 }, { "epoch": 0.002048, "grad_norm": 0.8886174559593201, "learning_rate": 1.3360828336210485e-05, "loss": 2.8356, "step": 153000 }, { "epoch": 0.0020736, "grad_norm": 0.805650532245636, "learning_rate": 1.3360069887929432e-05, "loss": 2.7386, "step": 153010 }, { "epoch": 0.0020992, "grad_norm": 0.8949053883552551, "learning_rate": 1.3359311417859218e-05, "loss": 2.8308, "step": 153020 }, { "epoch": 0.0021248, "grad_norm": 1.0818723440170288, "learning_rate": 1.3358552926004765e-05, "loss": 2.6837, "step": 153030 }, { "epoch": 0.0021504, "grad_norm": 0.8820096850395203, "learning_rate": 1.3357794412370988e-05, "loss": 3.0175, "step": 153040 }, { "epoch": 0.002176, "grad_norm": 0.8693305253982544, "learning_rate": 1.3357035876962806e-05, "loss": 3.0089, "step": 153050 }, { "epoch": 0.0022016, "grad_norm": 0.7059523463249207, "learning_rate": 1.3356277319785136e-05, "loss": 2.9654, "step": 153060 }, { "epoch": 0.0022272, "grad_norm": 0.8329213261604309, "learning_rate": 1.3355518740842904e-05, "loss": 3.1241, "step": 153070 }, { "epoch": 0.0022528, "grad_norm": 0.8387805223464966, "learning_rate": 1.335476014014102e-05, "loss": 3.0636, "step": 153080 }, { "epoch": 0.0022784, "grad_norm": 0.8041932582855225, "learning_rate": 1.3354001517684412e-05, "loss": 2.7562, "step": 153090 }, { "epoch": 0.002304, "grad_norm": 0.8168085813522339, "learning_rate": 1.3353242873477996e-05, "loss": 2.9594, "step": 153100 }, { "epoch": 0.0023296, "grad_norm": 0.7255269289016724, "learning_rate": 1.3352484207526688e-05, "loss": 2.6495, "step": 153110 }, { "epoch": 0.0023552, "grad_norm": 1.5619840621948242, "learning_rate": 1.3351725519835412e-05, "loss": 2.8143, "step": 153120 }, { "epoch": 0.0023808, "grad_norm": 0.9707084894180298, "learning_rate": 1.335096681040909e-05, "loss": 2.8329, "step": 153130 }, { "epoch": 0.0024064, "grad_norm": 0.7337956428527832, "learning_rate": 1.3350208079252636e-05, "loss": 2.5666, "step": 153140 }, { "epoch": 0.002432, "grad_norm": 0.727333128452301, "learning_rate": 1.3349449326370972e-05, "loss": 2.7159, "step": 153150 }, { "epoch": 0.0024576, "grad_norm": 0.7869760394096375, "learning_rate": 1.3348690551769021e-05, "loss": 2.9795, "step": 153160 }, { "epoch": 0.0024832, "grad_norm": 0.6847649216651917, "learning_rate": 1.3347931755451699e-05, "loss": 2.6829, "step": 153170 }, { "epoch": 0.0025088, "grad_norm": 0.7714400291442871, "learning_rate": 1.3347172937423931e-05, "loss": 2.7571, "step": 153180 }, { "epoch": 0.0025344, "grad_norm": 0.732970654964447, "learning_rate": 1.3346414097690637e-05, "loss": 2.8907, "step": 153190 }, { "epoch": 0.00256, "grad_norm": 1.0939440727233887, "learning_rate": 1.3345655236256739e-05, "loss": 3.0939, "step": 153200 }, { "epoch": 0.0025856, "grad_norm": 0.7543057799339294, "learning_rate": 1.3344896353127152e-05, "loss": 2.8357, "step": 153210 }, { "epoch": 0.0026112, "grad_norm": 0.7994116544723511, "learning_rate": 1.3344137448306805e-05, "loss": 2.9194, "step": 153220 }, { "epoch": 0.0026368, "grad_norm": 0.8410994410514832, "learning_rate": 1.3343378521800613e-05, "loss": 3.0282, "step": 153230 }, { "epoch": 0.0026624, "grad_norm": 0.8606271743774414, "learning_rate": 1.3342619573613497e-05, "loss": 2.7959, "step": 153240 }, { "epoch": 0.002688, "grad_norm": 0.766679048538208, "learning_rate": 1.3341860603750383e-05, "loss": 3.0579, "step": 153250 }, { "epoch": 0.0027136, "grad_norm": 0.7017667889595032, "learning_rate": 1.3341101612216196e-05, "loss": 2.8888, "step": 153260 }, { "epoch": 0.0027392, "grad_norm": 0.7024810314178467, "learning_rate": 1.3340342599015847e-05, "loss": 2.8668, "step": 153270 }, { "epoch": 0.0027648, "grad_norm": 0.6614726185798645, "learning_rate": 1.3339583564154265e-05, "loss": 2.8803, "step": 153280 }, { "epoch": 0.0027904, "grad_norm": 0.722578227519989, "learning_rate": 1.3338824507636373e-05, "loss": 2.5852, "step": 153290 }, { "epoch": 0.002816, "grad_norm": 0.7736412882804871, "learning_rate": 1.3338065429467095e-05, "loss": 2.8949, "step": 153300 }, { "epoch": 0.0028416, "grad_norm": 0.7661463618278503, "learning_rate": 1.3337306329651345e-05, "loss": 3.0014, "step": 153310 }, { "epoch": 0.0028672, "grad_norm": 0.6511965990066528, "learning_rate": 1.3336547208194051e-05, "loss": 2.9301, "step": 153320 }, { "epoch": 0.0028928, "grad_norm": 0.7584457993507385, "learning_rate": 1.3335788065100134e-05, "loss": 2.9817, "step": 153330 }, { "epoch": 0.0029184, "grad_norm": 0.7364270091056824, "learning_rate": 1.333502890037452e-05, "loss": 2.8605, "step": 153340 }, { "epoch": 0.002944, "grad_norm": 0.8630871176719666, "learning_rate": 1.3334269714022127e-05, "loss": 3.1867, "step": 153350 }, { "epoch": 0.0029696, "grad_norm": 0.7842271327972412, "learning_rate": 1.3333510506047887e-05, "loss": 2.9849, "step": 153360 }, { "epoch": 0.0029952, "grad_norm": 0.7343931794166565, "learning_rate": 1.3332751276456712e-05, "loss": 3.0336, "step": 153370 }, { "epoch": 0.0030208, "grad_norm": 0.794678807258606, "learning_rate": 1.333199202525353e-05, "loss": 2.9796, "step": 153380 }, { "epoch": 0.0030464, "grad_norm": 0.8125028014183044, "learning_rate": 1.3331232752443268e-05, "loss": 2.7028, "step": 153390 }, { "epoch": 0.003072, "grad_norm": 0.696661651134491, "learning_rate": 1.333047345803085e-05, "loss": 3.1196, "step": 153400 }, { "epoch": 0.0030976, "grad_norm": 0.7749840021133423, "learning_rate": 1.3329714142021191e-05, "loss": 3.0517, "step": 153410 }, { "epoch": 0.0031232, "grad_norm": 0.8906959891319275, "learning_rate": 1.3328954804419223e-05, "loss": 2.6911, "step": 153420 }, { "epoch": 0.0031488, "grad_norm": 0.7166550159454346, "learning_rate": 1.3328195445229869e-05, "loss": 2.9871, "step": 153430 }, { "epoch": 0.0031744, "grad_norm": 0.7506685256958008, "learning_rate": 1.3327436064458051e-05, "loss": 2.8848, "step": 153440 }, { "epoch": 0.0032, "grad_norm": 0.7302100658416748, "learning_rate": 1.3326676662108692e-05, "loss": 3.0076, "step": 153450 }, { "epoch": 0.0032256, "grad_norm": 0.7551884651184082, "learning_rate": 1.3325917238186722e-05, "loss": 2.9195, "step": 153460 }, { "epoch": 0.0032512, "grad_norm": 0.7347272634506226, "learning_rate": 1.3325157792697062e-05, "loss": 2.9174, "step": 153470 }, { "epoch": 0.0032768, "grad_norm": 0.8114956021308899, "learning_rate": 1.3324398325644636e-05, "loss": 3.0959, "step": 153480 }, { "epoch": 0.0033024, "grad_norm": 0.7583616971969604, "learning_rate": 1.3323638837034377e-05, "loss": 3.0672, "step": 153490 }, { "epoch": 0.003328, "grad_norm": 0.6943502426147461, "learning_rate": 1.3322879326871193e-05, "loss": 2.8702, "step": 153500 }, { "epoch": 0.0033536, "grad_norm": 0.7177296280860901, "learning_rate": 1.3322119795160024e-05, "loss": 2.889, "step": 153510 }, { "epoch": 0.0033792, "grad_norm": 0.727215588092804, "learning_rate": 1.3321360241905794e-05, "loss": 2.8216, "step": 153520 }, { "epoch": 0.0034048, "grad_norm": 0.7037586569786072, "learning_rate": 1.3320600667113423e-05, "loss": 2.8835, "step": 153530 }, { "epoch": 0.0034304, "grad_norm": 0.7097554802894592, "learning_rate": 1.331984107078784e-05, "loss": 2.9657, "step": 153540 }, { "epoch": 0.003456, "grad_norm": 0.7814300060272217, "learning_rate": 1.331908145293397e-05, "loss": 3.1045, "step": 153550 }, { "epoch": 0.0034816, "grad_norm": 0.6826186180114746, "learning_rate": 1.331832181355674e-05, "loss": 3.049, "step": 153560 }, { "epoch": 0.0035072, "grad_norm": 0.6906809210777283, "learning_rate": 1.3317562152661072e-05, "loss": 2.8754, "step": 153570 }, { "epoch": 0.0035328, "grad_norm": 0.7181533575057983, "learning_rate": 1.3316802470251897e-05, "loss": 2.8479, "step": 153580 }, { "epoch": 0.0035584, "grad_norm": 0.8462674617767334, "learning_rate": 1.3316042766334139e-05, "loss": 2.8472, "step": 153590 }, { "epoch": 0.003584, "grad_norm": 0.977535605430603, "learning_rate": 1.3315283040912723e-05, "loss": 2.9987, "step": 153600 }, { "epoch": 0.0036096, "grad_norm": 0.7576768398284912, "learning_rate": 1.3314523293992578e-05, "loss": 3.0976, "step": 153610 }, { "epoch": 0.0036352, "grad_norm": 1.1657965183258057, "learning_rate": 1.3313763525578636e-05, "loss": 3.0986, "step": 153620 }, { "epoch": 0.0036608, "grad_norm": 0.7441853284835815, "learning_rate": 1.3313003735675814e-05, "loss": 3.0263, "step": 153630 }, { "epoch": 0.0036864, "grad_norm": 1.2774291038513184, "learning_rate": 1.331224392428904e-05, "loss": 2.9844, "step": 153640 }, { "epoch": 0.003712, "grad_norm": 0.7376899719238281, "learning_rate": 1.3311484091423248e-05, "loss": 2.8184, "step": 153650 }, { "epoch": 0.0037376, "grad_norm": 0.7363936901092529, "learning_rate": 1.3310724237083363e-05, "loss": 3.0306, "step": 153660 }, { "epoch": 0.0037632, "grad_norm": 0.7016220092773438, "learning_rate": 1.330996436127431e-05, "loss": 2.8486, "step": 153670 }, { "epoch": 0.0037888, "grad_norm": 0.7449378371238708, "learning_rate": 1.3309204464001015e-05, "loss": 2.9499, "step": 153680 }, { "epoch": 0.0038144, "grad_norm": 0.7303327322006226, "learning_rate": 1.3308444545268415e-05, "loss": 2.9321, "step": 153690 }, { "epoch": 0.00384, "grad_norm": 0.8227167725563049, "learning_rate": 1.3307684605081426e-05, "loss": 3.0099, "step": 153700 }, { "epoch": 0.0038656, "grad_norm": 0.6903060674667358, "learning_rate": 1.3306924643444982e-05, "loss": 2.8729, "step": 153710 }, { "epoch": 0.0038912, "grad_norm": 0.6848516464233398, "learning_rate": 1.3306164660364014e-05, "loss": 2.9325, "step": 153720 }, { "epoch": 0.0039168, "grad_norm": 0.7471410632133484, "learning_rate": 1.3305404655843444e-05, "loss": 3.0107, "step": 153730 }, { "epoch": 0.0039424, "grad_norm": 0.7224570512771606, "learning_rate": 1.3304644629888204e-05, "loss": 2.846, "step": 153740 }, { "epoch": 0.003968, "grad_norm": 0.676620602607727, "learning_rate": 1.3303884582503219e-05, "loss": 3.0405, "step": 153750 }, { "epoch": 0.0039936, "grad_norm": 0.7913346290588379, "learning_rate": 1.3303124513693424e-05, "loss": 2.9797, "step": 153760 }, { "epoch": 0.0040192, "grad_norm": 0.788415789604187, "learning_rate": 1.3302364423463742e-05, "loss": 2.8147, "step": 153770 }, { "epoch": 0.0040448, "grad_norm": 0.6827661991119385, "learning_rate": 1.3301604311819105e-05, "loss": 3.152, "step": 153780 }, { "epoch": 0.0040704, "grad_norm": 0.7785497307777405, "learning_rate": 1.3300844178764445e-05, "loss": 2.9083, "step": 153790 }, { "epoch": 0.004096, "grad_norm": 0.8485366106033325, "learning_rate": 1.3300084024304685e-05, "loss": 3.0448, "step": 153800 }, { "epoch": 0.0041216, "grad_norm": 0.7518506646156311, "learning_rate": 1.3299323848444756e-05, "loss": 2.9612, "step": 153810 }, { "epoch": 0.0041472, "grad_norm": 1.0999683141708374, "learning_rate": 1.3298563651189591e-05, "loss": 3.2474, "step": 153820 }, { "epoch": 0.0041728, "grad_norm": 0.9203945994377136, "learning_rate": 1.3297803432544117e-05, "loss": 2.9407, "step": 153830 }, { "epoch": 0.0041984, "grad_norm": 0.6322221755981445, "learning_rate": 1.3297043192513263e-05, "loss": 2.7762, "step": 153840 }, { "epoch": 0.004224, "grad_norm": 1.0681616067886353, "learning_rate": 1.329628293110196e-05, "loss": 2.7523, "step": 153850 }, { "epoch": 0.0042496, "grad_norm": 0.6858962774276733, "learning_rate": 1.3295522648315139e-05, "loss": 2.9898, "step": 153860 }, { "epoch": 0.0042752, "grad_norm": 0.7853872179985046, "learning_rate": 1.3294762344157728e-05, "loss": 2.9186, "step": 153870 }, { "epoch": 0.0043008, "grad_norm": 0.6966865062713623, "learning_rate": 1.329400201863466e-05, "loss": 3.0447, "step": 153880 }, { "epoch": 0.0043264, "grad_norm": 0.6969542503356934, "learning_rate": 1.3293241671750867e-05, "loss": 3.0263, "step": 153890 }, { "epoch": 0.004352, "grad_norm": 0.8920428156852722, "learning_rate": 1.3292481303511273e-05, "loss": 2.9253, "step": 153900 }, { "epoch": 0.0043776, "grad_norm": 0.7037938833236694, "learning_rate": 1.3291720913920813e-05, "loss": 3.2985, "step": 153910 }, { "epoch": 0.0044032, "grad_norm": 0.7530844211578369, "learning_rate": 1.3290960502984423e-05, "loss": 2.9063, "step": 153920 }, { "epoch": 0.0044288, "grad_norm": 0.8338096141815186, "learning_rate": 1.3290200070707023e-05, "loss": 2.9878, "step": 153930 }, { "epoch": 0.0044544, "grad_norm": 0.7064024806022644, "learning_rate": 1.328943961709355e-05, "loss": 2.8837, "step": 153940 }, { "epoch": 0.00448, "grad_norm": 0.6653213500976562, "learning_rate": 1.328867914214894e-05, "loss": 2.8611, "step": 153950 }, { "epoch": 0.0045056, "grad_norm": 0.7396346926689148, "learning_rate": 1.3287918645878116e-05, "loss": 2.9492, "step": 153960 }, { "epoch": 0.0045312, "grad_norm": 0.8143625259399414, "learning_rate": 1.3287158128286014e-05, "loss": 2.9968, "step": 153970 }, { "epoch": 0.0045568, "grad_norm": 0.8367869853973389, "learning_rate": 1.3286397589377565e-05, "loss": 2.7977, "step": 153980 }, { "epoch": 0.0045824, "grad_norm": 0.7124364972114563, "learning_rate": 1.3285637029157701e-05, "loss": 3.0446, "step": 153990 }, { "epoch": 0.004608, "grad_norm": 0.7357916235923767, "learning_rate": 1.3284876447631354e-05, "loss": 3.0484, "step": 154000 }, { "epoch": 0.0046336, "grad_norm": 0.7721680402755737, "learning_rate": 1.3284115844803458e-05, "loss": 2.9563, "step": 154010 }, { "epoch": 0.0046592, "grad_norm": 0.973122239112854, "learning_rate": 1.3283355220678941e-05, "loss": 2.9743, "step": 154020 }, { "epoch": 0.0046848, "grad_norm": 0.770258367061615, "learning_rate": 1.3282594575262739e-05, "loss": 3.1227, "step": 154030 }, { "epoch": 0.0047104, "grad_norm": 0.697741687297821, "learning_rate": 1.3281833908559785e-05, "loss": 3.144, "step": 154040 }, { "epoch": 0.004736, "grad_norm": 1.3237820863723755, "learning_rate": 1.328107322057501e-05, "loss": 2.7505, "step": 154050 }, { "epoch": 0.0047616, "grad_norm": 0.970626711845398, "learning_rate": 1.3280312511313346e-05, "loss": 3.1404, "step": 154060 }, { "epoch": 0.0047872, "grad_norm": 0.864391565322876, "learning_rate": 1.3279551780779728e-05, "loss": 2.942, "step": 154070 }, { "epoch": 0.0048128, "grad_norm": 0.712597668170929, "learning_rate": 1.3278791028979087e-05, "loss": 2.9048, "step": 154080 }, { "epoch": 0.0048384, "grad_norm": 0.6978810429573059, "learning_rate": 1.3278030255916359e-05, "loss": 3.1287, "step": 154090 }, { "epoch": 0.004864, "grad_norm": 0.6951286196708679, "learning_rate": 1.3277269461596476e-05, "loss": 2.9575, "step": 154100 }, { "epoch": 0.0048896, "grad_norm": 0.8597800731658936, "learning_rate": 1.3276508646024371e-05, "loss": 2.8783, "step": 154110 }, { "epoch": 0.0049152, "grad_norm": 0.7657076120376587, "learning_rate": 1.3275747809204977e-05, "loss": 2.7565, "step": 154120 }, { "epoch": 0.0049408, "grad_norm": 0.7347002029418945, "learning_rate": 1.327498695114323e-05, "loss": 2.7977, "step": 154130 }, { "epoch": 0.0049664, "grad_norm": 0.8370612263679504, "learning_rate": 1.3274226071844063e-05, "loss": 2.9573, "step": 154140 }, { "epoch": 0.004992, "grad_norm": 0.7696663737297058, "learning_rate": 1.3273465171312412e-05, "loss": 3.1112, "step": 154150 }, { "epoch": 0.0050176, "grad_norm": 0.7368503212928772, "learning_rate": 1.3272704249553208e-05, "loss": 2.9328, "step": 154160 }, { "epoch": 0.0050432, "grad_norm": 0.7149988412857056, "learning_rate": 1.3271943306571387e-05, "loss": 2.9287, "step": 154170 }, { "epoch": 0.0050688, "grad_norm": 0.7602733373641968, "learning_rate": 1.327118234237188e-05, "loss": 2.8579, "step": 154180 }, { "epoch": 0.0050944, "grad_norm": 0.9796506762504578, "learning_rate": 1.3270421356959632e-05, "loss": 2.9385, "step": 154190 }, { "epoch": 0.00512, "grad_norm": 0.7499924302101135, "learning_rate": 1.326966035033956e-05, "loss": 2.9579, "step": 154200 }, { "epoch": 0.0051456, "grad_norm": 0.6890515089035034, "learning_rate": 1.3268899322516619e-05, "loss": 2.9232, "step": 154210 }, { "epoch": 0.0051712, "grad_norm": 0.6978850960731506, "learning_rate": 1.3268138273495732e-05, "loss": 2.6927, "step": 154220 }, { "epoch": 0.0051968, "grad_norm": 0.7503034472465515, "learning_rate": 1.3267377203281836e-05, "loss": 3.0669, "step": 154230 }, { "epoch": 0.0052224, "grad_norm": 0.7340383529663086, "learning_rate": 1.3266616111879867e-05, "loss": 2.885, "step": 154240 }, { "epoch": 0.005248, "grad_norm": 0.7228344678878784, "learning_rate": 1.3265854999294763e-05, "loss": 2.9541, "step": 154250 }, { "epoch": 0.0052736, "grad_norm": 0.8094726800918579, "learning_rate": 1.3265093865531456e-05, "loss": 2.8722, "step": 154260 }, { "epoch": 0.0052992, "grad_norm": 0.7070287466049194, "learning_rate": 1.3264332710594881e-05, "loss": 3.0128, "step": 154270 }, { "epoch": 0.0053248, "grad_norm": 0.7009731531143188, "learning_rate": 1.3263571534489979e-05, "loss": 2.9244, "step": 154280 }, { "epoch": 0.0053504, "grad_norm": 0.8252706527709961, "learning_rate": 1.3262810337221679e-05, "loss": 3.2814, "step": 154290 }, { "epoch": 0.005376, "grad_norm": 0.7603222131729126, "learning_rate": 1.3262049118794924e-05, "loss": 3.0879, "step": 154300 }, { "epoch": 0.0054016, "grad_norm": 0.7361925840377808, "learning_rate": 1.3261287879214648e-05, "loss": 2.8312, "step": 154310 }, { "epoch": 0.0054272, "grad_norm": 0.7424972653388977, "learning_rate": 1.3260526618485785e-05, "loss": 2.911, "step": 154320 }, { "epoch": 0.0054528, "grad_norm": 0.8402000069618225, "learning_rate": 1.3259765336613276e-05, "loss": 3.0378, "step": 154330 }, { "epoch": 0.0054784, "grad_norm": 1.5456044673919678, "learning_rate": 1.3259004033602052e-05, "loss": 2.7365, "step": 154340 }, { "epoch": 0.005504, "grad_norm": 0.9096757173538208, "learning_rate": 1.3258242709457054e-05, "loss": 2.912, "step": 154350 }, { "epoch": 0.0055296, "grad_norm": 0.8004801869392395, "learning_rate": 1.325748136418322e-05, "loss": 2.8984, "step": 154360 }, { "epoch": 0.0055552, "grad_norm": 0.7616493105888367, "learning_rate": 1.325671999778548e-05, "loss": 3.0612, "step": 154370 }, { "epoch": 0.0055808, "grad_norm": 0.6711608171463013, "learning_rate": 1.3255958610268782e-05, "loss": 3.0505, "step": 154380 }, { "epoch": 0.0056064, "grad_norm": 0.8640995025634766, "learning_rate": 1.3255197201638055e-05, "loss": 3.0896, "step": 154390 }, { "epoch": 0.005632, "grad_norm": 0.6580069661140442, "learning_rate": 1.325443577189824e-05, "loss": 3.2538, "step": 154400 }, { "epoch": 0.0056576, "grad_norm": 0.6961845755577087, "learning_rate": 1.3253674321054275e-05, "loss": 2.9108, "step": 154410 }, { "epoch": 0.0056832, "grad_norm": 0.674903154373169, "learning_rate": 1.3252912849111094e-05, "loss": 2.8548, "step": 154420 }, { "epoch": 0.0057088, "grad_norm": 0.8548702001571655, "learning_rate": 1.325215135607364e-05, "loss": 2.985, "step": 154430 }, { "epoch": 0.0057344, "grad_norm": 0.7824594378471375, "learning_rate": 1.3251389841946846e-05, "loss": 2.9166, "step": 154440 }, { "epoch": 0.00576, "grad_norm": 0.6654881834983826, "learning_rate": 1.3250628306735658e-05, "loss": 3.1125, "step": 154450 }, { "epoch": 0.0057856, "grad_norm": 0.8064141273498535, "learning_rate": 1.3249866750445005e-05, "loss": 2.8094, "step": 154460 }, { "epoch": 0.0058112, "grad_norm": 0.7534526586532593, "learning_rate": 1.3249105173079831e-05, "loss": 2.8442, "step": 154470 }, { "epoch": 0.0058368, "grad_norm": 0.7430852055549622, "learning_rate": 1.3248343574645075e-05, "loss": 2.9874, "step": 154480 }, { "epoch": 0.0058624, "grad_norm": 0.7260296940803528, "learning_rate": 1.3247581955145673e-05, "loss": 2.8317, "step": 154490 }, { "epoch": 0.005888, "grad_norm": 0.8305237293243408, "learning_rate": 1.3246820314586564e-05, "loss": 2.8463, "step": 154500 }, { "epoch": 0.0059136, "grad_norm": 0.7471860647201538, "learning_rate": 1.3246058652972692e-05, "loss": 2.9771, "step": 154510 }, { "epoch": 0.0059392, "grad_norm": 0.6936290860176086, "learning_rate": 1.3245296970308986e-05, "loss": 2.8306, "step": 154520 }, { "epoch": 0.0059648, "grad_norm": 0.763027012348175, "learning_rate": 1.3244535266600395e-05, "loss": 3.0867, "step": 154530 }, { "epoch": 0.0059904, "grad_norm": 0.7795389294624329, "learning_rate": 1.3243773541851855e-05, "loss": 3.1192, "step": 154540 }, { "epoch": 0.006016, "grad_norm": 0.7009979486465454, "learning_rate": 1.3243011796068307e-05, "loss": 2.9832, "step": 154550 }, { "epoch": 0.0060416, "grad_norm": 0.8302760124206543, "learning_rate": 1.3242250029254687e-05, "loss": 3.0398, "step": 154560 }, { "epoch": 0.0060672, "grad_norm": 0.6985628604888916, "learning_rate": 1.3241488241415939e-05, "loss": 2.7963, "step": 154570 }, { "epoch": 0.0060928, "grad_norm": 0.7140093445777893, "learning_rate": 1.3240726432557e-05, "loss": 2.9869, "step": 154580 }, { "epoch": 0.0061184, "grad_norm": 0.7387544512748718, "learning_rate": 1.3239964602682814e-05, "loss": 3.5978, "step": 154590 }, { "epoch": 0.006144, "grad_norm": 0.7655576467514038, "learning_rate": 1.3239202751798315e-05, "loss": 2.934, "step": 154600 }, { "epoch": 2.56e-05, "grad_norm": 0.6921613812446594, "learning_rate": 1.323844087990845e-05, "loss": 2.2269, "step": 154610 }, { "epoch": 5.12e-05, "grad_norm": 0.7942327260971069, "learning_rate": 1.3237678987018154e-05, "loss": 3.0438, "step": 154620 }, { "epoch": 7.68e-05, "grad_norm": 0.6883183121681213, "learning_rate": 1.323691707313237e-05, "loss": 2.817, "step": 154630 }, { "epoch": 0.0001024, "grad_norm": 0.7089741230010986, "learning_rate": 1.3236155138256042e-05, "loss": 2.7766, "step": 154640 }, { "epoch": 0.000128, "grad_norm": 0.7034642100334167, "learning_rate": 1.3235393182394108e-05, "loss": 2.9749, "step": 154650 }, { "epoch": 0.0001536, "grad_norm": 0.7205371260643005, "learning_rate": 1.3234631205551506e-05, "loss": 2.8175, "step": 154660 }, { "epoch": 0.0001792, "grad_norm": 0.7789135575294495, "learning_rate": 1.3233869207733181e-05, "loss": 2.9695, "step": 154670 }, { "epoch": 0.0002048, "grad_norm": 0.6841922402381897, "learning_rate": 1.3233107188944076e-05, "loss": 2.7738, "step": 154680 }, { "epoch": 0.0002304, "grad_norm": 1.8161890506744385, "learning_rate": 1.323234514918913e-05, "loss": 2.7566, "step": 154690 }, { "epoch": 0.000256, "grad_norm": 0.9273061156272888, "learning_rate": 1.3231583088473283e-05, "loss": 2.7849, "step": 154700 }, { "epoch": 0.0002816, "grad_norm": 1.231113076210022, "learning_rate": 1.323082100680148e-05, "loss": 2.7503, "step": 154710 }, { "epoch": 0.0003072, "grad_norm": 0.8893477916717529, "learning_rate": 1.3230058904178658e-05, "loss": 2.8785, "step": 154720 }, { "epoch": 0.0003328, "grad_norm": 0.7712924480438232, "learning_rate": 1.3229296780609767e-05, "loss": 2.9019, "step": 154730 }, { "epoch": 0.0003584, "grad_norm": 0.6988312005996704, "learning_rate": 1.3228534636099741e-05, "loss": 2.7609, "step": 154740 }, { "epoch": 0.000384, "grad_norm": 0.6913323402404785, "learning_rate": 1.3227772470653528e-05, "loss": 2.9069, "step": 154750 }, { "epoch": 0.0004096, "grad_norm": 0.7083433866500854, "learning_rate": 1.3227010284276068e-05, "loss": 2.8425, "step": 154760 }, { "epoch": 0.0004352, "grad_norm": 0.715545654296875, "learning_rate": 1.3226248076972303e-05, "loss": 2.8092, "step": 154770 }, { "epoch": 0.0004608, "grad_norm": 0.8820390105247498, "learning_rate": 1.3225485848747179e-05, "loss": 2.8565, "step": 154780 }, { "epoch": 0.0004864, "grad_norm": 0.7018926739692688, "learning_rate": 1.3224723599605635e-05, "loss": 2.9097, "step": 154790 }, { "epoch": 0.000512, "grad_norm": 0.6904200315475464, "learning_rate": 1.3223961329552614e-05, "loss": 2.6303, "step": 154800 }, { "epoch": 0.0005376, "grad_norm": 0.7703801393508911, "learning_rate": 1.3223199038593062e-05, "loss": 2.7165, "step": 154810 }, { "epoch": 0.0005632, "grad_norm": 0.7993183732032776, "learning_rate": 1.3222436726731921e-05, "loss": 2.8681, "step": 154820 }, { "epoch": 0.0005888, "grad_norm": 0.681306004524231, "learning_rate": 1.3221674393974133e-05, "loss": 2.689, "step": 154830 }, { "epoch": 0.0006144, "grad_norm": 0.8011993765830994, "learning_rate": 1.3220912040324647e-05, "loss": 2.7265, "step": 154840 }, { "epoch": 0.00064, "grad_norm": 0.7362048029899597, "learning_rate": 1.3220149665788398e-05, "loss": 2.9982, "step": 154850 }, { "epoch": 0.0006656, "grad_norm": 1.1611385345458984, "learning_rate": 1.3219387270370337e-05, "loss": 2.859, "step": 154860 }, { "epoch": 0.0006912, "grad_norm": 0.7225836515426636, "learning_rate": 1.3218624854075401e-05, "loss": 2.9279, "step": 154870 }, { "epoch": 0.0007168, "grad_norm": 0.8936278223991394, "learning_rate": 1.3217862416908545e-05, "loss": 2.8364, "step": 154880 }, { "epoch": 0.0007424, "grad_norm": 0.6677219271659851, "learning_rate": 1.3217099958874698e-05, "loss": 2.7536, "step": 154890 }, { "epoch": 0.000768, "grad_norm": 0.7520290017127991, "learning_rate": 1.3216337479978818e-05, "loss": 2.9758, "step": 154900 }, { "epoch": 0.0007936, "grad_norm": 0.6540368795394897, "learning_rate": 1.3215574980225844e-05, "loss": 2.9202, "step": 154910 }, { "epoch": 0.0008192, "grad_norm": 0.7333931922912598, "learning_rate": 1.321481245962072e-05, "loss": 2.7861, "step": 154920 }, { "epoch": 0.0008448, "grad_norm": 0.6692076921463013, "learning_rate": 1.3214049918168392e-05, "loss": 2.9356, "step": 154930 }, { "epoch": 0.0008704, "grad_norm": 0.9724552631378174, "learning_rate": 1.3213287355873805e-05, "loss": 2.7303, "step": 154940 }, { "epoch": 0.000896, "grad_norm": 1.7251945734024048, "learning_rate": 1.3212524772741903e-05, "loss": 2.8793, "step": 154950 }, { "epoch": 0.0009216, "grad_norm": 0.6103620529174805, "learning_rate": 1.321176216877763e-05, "loss": 2.7914, "step": 154960 }, { "epoch": 0.0009472, "grad_norm": 0.7223263382911682, "learning_rate": 1.3210999543985934e-05, "loss": 2.6647, "step": 154970 }, { "epoch": 0.0009728, "grad_norm": 0.799348771572113, "learning_rate": 1.321023689837176e-05, "loss": 2.9278, "step": 154980 }, { "epoch": 0.0009984, "grad_norm": 0.6950348019599915, "learning_rate": 1.3209474231940052e-05, "loss": 2.7498, "step": 154990 }, { "epoch": 0.001024, "grad_norm": 0.7398308515548706, "learning_rate": 1.3208711544695758e-05, "loss": 2.9286, "step": 155000 }, { "epoch": 0.0010496, "grad_norm": 0.6743210554122925, "learning_rate": 1.3207948836643823e-05, "loss": 2.7192, "step": 155010 }, { "epoch": 0.0010752, "grad_norm": 0.8326336145401001, "learning_rate": 1.3207186107789189e-05, "loss": 3.1796, "step": 155020 }, { "epoch": 0.0011008, "grad_norm": 0.7319444417953491, "learning_rate": 1.3206423358136808e-05, "loss": 2.892, "step": 155030 }, { "epoch": 0.0011264, "grad_norm": 1.5014623403549194, "learning_rate": 1.3205660587691625e-05, "loss": 2.8832, "step": 155040 }, { "epoch": 0.001152, "grad_norm": 0.704390287399292, "learning_rate": 1.3204897796458582e-05, "loss": 3.0325, "step": 155050 }, { "epoch": 0.0011776, "grad_norm": 0.7178816199302673, "learning_rate": 1.3204134984442629e-05, "loss": 2.8795, "step": 155060 }, { "epoch": 0.0012032, "grad_norm": 0.768094539642334, "learning_rate": 1.3203372151648713e-05, "loss": 2.8952, "step": 155070 }, { "epoch": 0.0012288, "grad_norm": 0.733726441860199, "learning_rate": 1.320260929808178e-05, "loss": 2.9366, "step": 155080 }, { "epoch": 0.0012544, "grad_norm": 0.7802292704582214, "learning_rate": 1.3201846423746775e-05, "loss": 2.7815, "step": 155090 }, { "epoch": 0.00128, "grad_norm": 0.7179306149482727, "learning_rate": 1.3201083528648651e-05, "loss": 3.0698, "step": 155100 }, { "epoch": 0.0013056, "grad_norm": 0.8429141640663147, "learning_rate": 1.3200320612792348e-05, "loss": 2.766, "step": 155110 }, { "epoch": 0.0013312, "grad_norm": 0.7675527930259705, "learning_rate": 1.3199557676182818e-05, "loss": 3.1242, "step": 155120 }, { "epoch": 0.0013568, "grad_norm": 0.8469274640083313, "learning_rate": 1.3198794718825005e-05, "loss": 2.9742, "step": 155130 }, { "epoch": 0.0013824, "grad_norm": 0.6826106309890747, "learning_rate": 1.319803174072386e-05, "loss": 2.7818, "step": 155140 }, { "epoch": 0.001408, "grad_norm": 0.7645906209945679, "learning_rate": 1.3197268741884328e-05, "loss": 2.7758, "step": 155150 }, { "epoch": 0.0014336, "grad_norm": 0.9957227110862732, "learning_rate": 1.3196505722311357e-05, "loss": 2.8448, "step": 155160 }, { "epoch": 0.0014592, "grad_norm": 1.4151618480682373, "learning_rate": 1.3195742682009897e-05, "loss": 2.7827, "step": 155170 }, { "epoch": 0.0014848, "grad_norm": 0.9931414127349854, "learning_rate": 1.3194979620984896e-05, "loss": 2.63, "step": 155180 }, { "epoch": 0.0015104, "grad_norm": 0.7088366746902466, "learning_rate": 1.31942165392413e-05, "loss": 2.8189, "step": 155190 }, { "epoch": 0.001536, "grad_norm": 0.9527180790901184, "learning_rate": 1.319345343678406e-05, "loss": 2.8025, "step": 155200 }, { "epoch": 2.56e-05, "grad_norm": 0.7709183096885681, "learning_rate": 1.3192690313618122e-05, "loss": 2.3023, "step": 155210 }, { "epoch": 5.12e-05, "grad_norm": 0.7962323427200317, "learning_rate": 1.3191927169748437e-05, "loss": 2.8455, "step": 155220 }, { "epoch": 7.68e-05, "grad_norm": 0.725540816783905, "learning_rate": 1.319116400517995e-05, "loss": 2.9404, "step": 155230 }, { "epoch": 0.0001024, "grad_norm": 0.696476399898529, "learning_rate": 1.3190400819917615e-05, "loss": 2.5748, "step": 155240 }, { "epoch": 0.000128, "grad_norm": 0.9147282838821411, "learning_rate": 1.3189637613966378e-05, "loss": 2.7926, "step": 155250 }, { "epoch": 0.0001536, "grad_norm": 0.7785531878471375, "learning_rate": 1.3188874387331188e-05, "loss": 2.8565, "step": 155260 }, { "epoch": 0.0001792, "grad_norm": 1.079111933708191, "learning_rate": 1.3188111140016998e-05, "loss": 2.84, "step": 155270 }, { "epoch": 0.0002048, "grad_norm": 1.9366906881332397, "learning_rate": 1.318734787202875e-05, "loss": 2.6724, "step": 155280 }, { "epoch": 0.0002304, "grad_norm": 0.985965371131897, "learning_rate": 1.31865845833714e-05, "loss": 2.9958, "step": 155290 }, { "epoch": 0.000256, "grad_norm": 0.7072895169258118, "learning_rate": 1.31858212740499e-05, "loss": 2.9454, "step": 155300 }, { "epoch": 0.0002816, "grad_norm": 0.687254011631012, "learning_rate": 1.318505794406919e-05, "loss": 2.6485, "step": 155310 }, { "epoch": 0.0003072, "grad_norm": 0.6931094527244568, "learning_rate": 1.3184294593434228e-05, "loss": 2.7505, "step": 155320 }, { "epoch": 0.0003328, "grad_norm": 0.7306064963340759, "learning_rate": 1.318353122214996e-05, "loss": 3.0127, "step": 155330 }, { "epoch": 0.0003584, "grad_norm": 0.7741953134536743, "learning_rate": 1.318276783022134e-05, "loss": 3.0589, "step": 155340 }, { "epoch": 0.000384, "grad_norm": 0.7126598358154297, "learning_rate": 1.3182004417653316e-05, "loss": 2.8568, "step": 155350 }, { "epoch": 0.0004096, "grad_norm": 0.7611620426177979, "learning_rate": 1.3181240984450839e-05, "loss": 3.295, "step": 155360 }, { "epoch": 0.0004352, "grad_norm": 0.7213660478591919, "learning_rate": 1.3180477530618862e-05, "loss": 2.8758, "step": 155370 }, { "epoch": 0.0004608, "grad_norm": 1.0255721807479858, "learning_rate": 1.3179714056162332e-05, "loss": 2.9948, "step": 155380 }, { "epoch": 0.0004864, "grad_norm": 0.9547633528709412, "learning_rate": 1.3178950561086198e-05, "loss": 3.0128, "step": 155390 }, { "epoch": 0.000512, "grad_norm": 0.7993968725204468, "learning_rate": 1.3178187045395421e-05, "loss": 2.9009, "step": 155400 }, { "epoch": 0.0005376, "grad_norm": 0.8267178535461426, "learning_rate": 1.317742350909494e-05, "loss": 2.8598, "step": 155410 }, { "epoch": 0.0005632, "grad_norm": 0.7571094036102295, "learning_rate": 1.3176659952189712e-05, "loss": 2.7313, "step": 155420 }, { "epoch": 0.0005888, "grad_norm": 0.9144321084022522, "learning_rate": 1.3175896374684693e-05, "loss": 2.9023, "step": 155430 }, { "epoch": 0.0006144, "grad_norm": 0.7138877511024475, "learning_rate": 1.3175132776584827e-05, "loss": 2.6636, "step": 155440 }, { "epoch": 0.00064, "grad_norm": 0.7420457601547241, "learning_rate": 1.3174369157895068e-05, "loss": 2.8606, "step": 155450 }, { "epoch": 0.0006656, "grad_norm": 0.7838029861450195, "learning_rate": 1.317360551862037e-05, "loss": 3.2407, "step": 155460 }, { "epoch": 0.0006912, "grad_norm": 0.71327805519104, "learning_rate": 1.3172841858765686e-05, "loss": 2.8331, "step": 155470 }, { "epoch": 0.0007168, "grad_norm": 1.7113604545593262, "learning_rate": 1.3172078178335964e-05, "loss": 2.9815, "step": 155480 }, { "epoch": 0.0007424, "grad_norm": 0.9492096900939941, "learning_rate": 1.3171314477336155e-05, "loss": 2.8603, "step": 155490 }, { "epoch": 0.000768, "grad_norm": 0.8853720426559448, "learning_rate": 1.317055075577122e-05, "loss": 2.9574, "step": 155500 }, { "epoch": 0.0007936, "grad_norm": 0.7068458199501038, "learning_rate": 1.31697870136461e-05, "loss": 2.6953, "step": 155510 }, { "epoch": 0.0008192, "grad_norm": 0.7223622798919678, "learning_rate": 1.3169023250965756e-05, "loss": 2.8303, "step": 155520 }, { "epoch": 0.0008448, "grad_norm": 0.7385770678520203, "learning_rate": 1.3168259467735141e-05, "loss": 2.7799, "step": 155530 }, { "epoch": 0.0008704, "grad_norm": 0.7840859889984131, "learning_rate": 1.3167495663959202e-05, "loss": 2.9514, "step": 155540 }, { "epoch": 0.000896, "grad_norm": 0.7657705545425415, "learning_rate": 1.3166731839642897e-05, "loss": 2.6029, "step": 155550 }, { "epoch": 0.0009216, "grad_norm": 0.8680262565612793, "learning_rate": 1.3165967994791177e-05, "loss": 3.0992, "step": 155560 }, { "epoch": 0.0009472, "grad_norm": 0.6899589896202087, "learning_rate": 1.3165204129408999e-05, "loss": 2.8628, "step": 155570 }, { "epoch": 0.0009728, "grad_norm": 0.8506712317466736, "learning_rate": 1.3164440243501311e-05, "loss": 2.9562, "step": 155580 }, { "epoch": 0.0009984, "grad_norm": 0.9716579914093018, "learning_rate": 1.3163676337073063e-05, "loss": 2.822, "step": 155590 }, { "epoch": 0.001024, "grad_norm": 0.9526202082633972, "learning_rate": 1.3162912410129225e-05, "loss": 2.9239, "step": 155600 }, { "epoch": 0.0010496, "grad_norm": 0.8106932640075684, "learning_rate": 1.3162148462674736e-05, "loss": 3.1323, "step": 155610 }, { "epoch": 0.0010752, "grad_norm": 0.762961745262146, "learning_rate": 1.3161384494714553e-05, "loss": 2.7968, "step": 155620 }, { "epoch": 0.0011008, "grad_norm": 0.8069424629211426, "learning_rate": 1.3160620506253636e-05, "loss": 2.9648, "step": 155630 }, { "epoch": 0.0011264, "grad_norm": 0.6805785298347473, "learning_rate": 1.3159856497296932e-05, "loss": 2.6506, "step": 155640 }, { "epoch": 0.001152, "grad_norm": 0.7864617109298706, "learning_rate": 1.31590924678494e-05, "loss": 2.9487, "step": 155650 }, { "epoch": 0.0011776, "grad_norm": 0.8497716784477234, "learning_rate": 1.315832841791599e-05, "loss": 3.0, "step": 155660 }, { "epoch": 0.0012032, "grad_norm": 0.6816596984863281, "learning_rate": 1.3157564347501664e-05, "loss": 2.9152, "step": 155670 }, { "epoch": 0.0012288, "grad_norm": 0.6507987976074219, "learning_rate": 1.3156800256611366e-05, "loss": 2.8766, "step": 155680 }, { "epoch": 0.0012544, "grad_norm": 0.7339760661125183, "learning_rate": 1.3156036145250065e-05, "loss": 3.0045, "step": 155690 }, { "epoch": 0.00128, "grad_norm": 0.8435063362121582, "learning_rate": 1.3155272013422703e-05, "loss": 2.718, "step": 155700 }, { "epoch": 0.0013056, "grad_norm": 0.992586076259613, "learning_rate": 1.3154507861134243e-05, "loss": 2.8793, "step": 155710 }, { "epoch": 0.0013312, "grad_norm": 0.7489669919013977, "learning_rate": 1.3153743688389636e-05, "loss": 2.7989, "step": 155720 }, { "epoch": 0.0013568, "grad_norm": 0.828183650970459, "learning_rate": 1.3152979495193843e-05, "loss": 2.8354, "step": 155730 }, { "epoch": 0.0013824, "grad_norm": 0.6852625012397766, "learning_rate": 1.3152215281551812e-05, "loss": 2.9777, "step": 155740 }, { "epoch": 0.001408, "grad_norm": 0.7544513940811157, "learning_rate": 1.3151451047468503e-05, "loss": 2.7672, "step": 155750 }, { "epoch": 0.0014336, "grad_norm": 0.677753210067749, "learning_rate": 1.3150686792948873e-05, "loss": 2.9007, "step": 155760 }, { "epoch": 0.0014592, "grad_norm": 0.7975491285324097, "learning_rate": 1.3149922517997875e-05, "loss": 2.8769, "step": 155770 }, { "epoch": 0.0014848, "grad_norm": 0.9497395753860474, "learning_rate": 1.3149158222620467e-05, "loss": 2.7533, "step": 155780 }, { "epoch": 0.0015104, "grad_norm": 0.7696306705474854, "learning_rate": 1.3148393906821608e-05, "loss": 2.8117, "step": 155790 }, { "epoch": 0.001536, "grad_norm": 0.9336252808570862, "learning_rate": 1.3147629570606249e-05, "loss": 2.6018, "step": 155800 }, { "epoch": 0.0015616, "grad_norm": 0.7592140436172485, "learning_rate": 1.3146865213979347e-05, "loss": 2.8685, "step": 155810 }, { "epoch": 0.0015872, "grad_norm": 1.046487808227539, "learning_rate": 1.3146100836945862e-05, "loss": 2.6732, "step": 155820 }, { "epoch": 0.0016128, "grad_norm": 0.7436508536338806, "learning_rate": 1.3145336439510749e-05, "loss": 2.9775, "step": 155830 }, { "epoch": 0.0016384, "grad_norm": 0.8042546510696411, "learning_rate": 1.3144572021678968e-05, "loss": 3.008, "step": 155840 }, { "epoch": 0.001664, "grad_norm": 0.7934517860412598, "learning_rate": 1.3143807583455466e-05, "loss": 3.0609, "step": 155850 }, { "epoch": 0.0016896, "grad_norm": 0.7098757028579712, "learning_rate": 1.3143043124845215e-05, "loss": 2.9053, "step": 155860 }, { "epoch": 0.0017152, "grad_norm": 0.9844384789466858, "learning_rate": 1.3142278645853159e-05, "loss": 3.0369, "step": 155870 }, { "epoch": 0.0017408, "grad_norm": 0.810023844242096, "learning_rate": 1.3141514146484265e-05, "loss": 2.8574, "step": 155880 }, { "epoch": 0.0017664, "grad_norm": 1.1225000619888306, "learning_rate": 1.3140749626743484e-05, "loss": 3.1424, "step": 155890 }, { "epoch": 0.001792, "grad_norm": 0.8099989295005798, "learning_rate": 1.3139985086635779e-05, "loss": 2.9633, "step": 155900 }, { "epoch": 0.0018176, "grad_norm": 0.7761035561561584, "learning_rate": 1.3139220526166104e-05, "loss": 2.7677, "step": 155910 }, { "epoch": 0.0018432, "grad_norm": 0.7705764770507812, "learning_rate": 1.3138455945339417e-05, "loss": 2.7172, "step": 155920 }, { "epoch": 0.0018688, "grad_norm": 0.6742528080940247, "learning_rate": 1.3137691344160679e-05, "loss": 2.7952, "step": 155930 }, { "epoch": 0.0018944, "grad_norm": 1.2314075231552124, "learning_rate": 1.3136926722634843e-05, "loss": 2.2314, "step": 155940 }, { "epoch": 0.00192, "grad_norm": 0.9430000185966492, "learning_rate": 1.3136162080766874e-05, "loss": 2.9836, "step": 155950 }, { "epoch": 0.0019456, "grad_norm": 0.6656598448753357, "learning_rate": 1.313539741856173e-05, "loss": 2.8089, "step": 155960 }, { "epoch": 0.0019712, "grad_norm": 0.6452435255050659, "learning_rate": 1.3134632736024364e-05, "loss": 2.6823, "step": 155970 }, { "epoch": 0.0019968, "grad_norm": 0.6945540904998779, "learning_rate": 1.3133868033159738e-05, "loss": 2.7907, "step": 155980 }, { "epoch": 0.0020224, "grad_norm": 0.8627620935440063, "learning_rate": 1.3133103309972813e-05, "loss": 3.057, "step": 155990 }, { "epoch": 0.002048, "grad_norm": 0.6896421909332275, "learning_rate": 1.313233856646854e-05, "loss": 2.9247, "step": 156000 }, { "epoch": 0.0020736, "grad_norm": 1.0901693105697632, "learning_rate": 1.313157380265189e-05, "loss": 2.7166, "step": 156010 }, { "epoch": 0.0020992, "grad_norm": 0.9973311424255371, "learning_rate": 1.3130809018527813e-05, "loss": 2.7991, "step": 156020 }, { "epoch": 0.0021248, "grad_norm": 0.7433181405067444, "learning_rate": 1.3130044214101269e-05, "loss": 2.8157, "step": 156030 }, { "epoch": 0.0021504, "grad_norm": 0.7485822439193726, "learning_rate": 1.3129279389377223e-05, "loss": 2.8516, "step": 156040 }, { "epoch": 0.002176, "grad_norm": 0.7910752296447754, "learning_rate": 1.3128514544360633e-05, "loss": 2.6795, "step": 156050 }, { "epoch": 0.0022016, "grad_norm": 0.7930727601051331, "learning_rate": 1.3127749679056458e-05, "loss": 2.9912, "step": 156060 }, { "epoch": 0.0022272, "grad_norm": 0.7741997838020325, "learning_rate": 1.3126984793469654e-05, "loss": 3.0162, "step": 156070 }, { "epoch": 0.0022528, "grad_norm": 1.4313288927078247, "learning_rate": 1.3126219887605187e-05, "loss": 2.738, "step": 156080 }, { "epoch": 0.0022784, "grad_norm": 0.6997891664505005, "learning_rate": 1.3125454961468016e-05, "loss": 2.8416, "step": 156090 }, { "epoch": 0.002304, "grad_norm": 0.8882743120193481, "learning_rate": 1.31246900150631e-05, "loss": 3.0159, "step": 156100 }, { "epoch": 0.0023296, "grad_norm": 0.8216678500175476, "learning_rate": 1.3123925048395397e-05, "loss": 2.7855, "step": 156110 }, { "epoch": 0.0023552, "grad_norm": 0.9823518395423889, "learning_rate": 1.3123160061469873e-05, "loss": 2.827, "step": 156120 }, { "epoch": 0.0023808, "grad_norm": 0.747160792350769, "learning_rate": 1.3122395054291484e-05, "loss": 2.8177, "step": 156130 }, { "epoch": 0.0024064, "grad_norm": 0.8329748511314392, "learning_rate": 1.3121630026865193e-05, "loss": 3.0289, "step": 156140 }, { "epoch": 0.002432, "grad_norm": 0.8024886250495911, "learning_rate": 1.3120864979195962e-05, "loss": 2.8306, "step": 156150 }, { "epoch": 0.0024576, "grad_norm": 0.8240295052528381, "learning_rate": 1.3120099911288752e-05, "loss": 2.9364, "step": 156160 }, { "epoch": 0.0024832, "grad_norm": 0.6850224137306213, "learning_rate": 1.3119334823148522e-05, "loss": 3.1147, "step": 156170 }, { "epoch": 0.0025088, "grad_norm": 0.792198896408081, "learning_rate": 1.3118569714780235e-05, "loss": 2.6876, "step": 156180 }, { "epoch": 0.0025344, "grad_norm": 2.9541821479797363, "learning_rate": 1.3117804586188854e-05, "loss": 2.8298, "step": 156190 }, { "epoch": 0.00256, "grad_norm": 1.3159551620483398, "learning_rate": 1.3117039437379333e-05, "loss": 2.7209, "step": 156200 }, { "epoch": 0.0025856, "grad_norm": 0.9440057873725891, "learning_rate": 1.3116274268356644e-05, "loss": 3.0381, "step": 156210 }, { "epoch": 0.0026112, "grad_norm": 0.7188708782196045, "learning_rate": 1.3115509079125745e-05, "loss": 3.0271, "step": 156220 }, { "epoch": 0.0026368, "grad_norm": 0.7065588235855103, "learning_rate": 1.3114743869691598e-05, "loss": 2.9825, "step": 156230 }, { "epoch": 0.0026624, "grad_norm": 0.7376793622970581, "learning_rate": 1.3113978640059162e-05, "loss": 3.0126, "step": 156240 }, { "epoch": 0.002688, "grad_norm": 0.8964505195617676, "learning_rate": 1.3113213390233404e-05, "loss": 2.9173, "step": 156250 }, { "epoch": 0.0027136, "grad_norm": 0.7599084973335266, "learning_rate": 1.3112448120219286e-05, "loss": 3.1442, "step": 156260 }, { "epoch": 0.0027392, "grad_norm": 0.9576349258422852, "learning_rate": 1.3111682830021769e-05, "loss": 2.9402, "step": 156270 }, { "epoch": 0.0027648, "grad_norm": 0.7273979187011719, "learning_rate": 1.311091751964581e-05, "loss": 3.0413, "step": 156280 }, { "epoch": 0.0027904, "grad_norm": 0.6907287836074829, "learning_rate": 1.3110152189096384e-05, "loss": 2.9175, "step": 156290 }, { "epoch": 0.002816, "grad_norm": 0.8504545092582703, "learning_rate": 1.3109386838378444e-05, "loss": 2.9679, "step": 156300 }, { "epoch": 0.0028416, "grad_norm": 0.7860512733459473, "learning_rate": 1.3108621467496958e-05, "loss": 2.9938, "step": 156310 }, { "epoch": 0.0028672, "grad_norm": 0.7180687785148621, "learning_rate": 1.310785607645689e-05, "loss": 3.1241, "step": 156320 }, { "epoch": 0.0028928, "grad_norm": 0.8618840575218201, "learning_rate": 1.3107090665263198e-05, "loss": 3.1109, "step": 156330 }, { "epoch": 0.0029184, "grad_norm": 0.6895686984062195, "learning_rate": 1.310632523392085e-05, "loss": 2.8613, "step": 156340 }, { "epoch": 0.002944, "grad_norm": 0.8362485766410828, "learning_rate": 1.310555978243481e-05, "loss": 3.0097, "step": 156350 }, { "epoch": 0.0029696, "grad_norm": 0.9320666790008545, "learning_rate": 1.310479431081004e-05, "loss": 2.8762, "step": 156360 }, { "epoch": 0.0029952, "grad_norm": 0.7563979625701904, "learning_rate": 1.31040288190515e-05, "loss": 3.1054, "step": 156370 }, { "epoch": 0.0030208, "grad_norm": 0.7140699625015259, "learning_rate": 1.310326330716416e-05, "loss": 3.0917, "step": 156380 }, { "epoch": 0.0030464, "grad_norm": 0.8474112749099731, "learning_rate": 1.3102497775152986e-05, "loss": 3.0612, "step": 156390 }, { "epoch": 0.003072, "grad_norm": 0.7188971638679504, "learning_rate": 1.3101732223022936e-05, "loss": 2.9558, "step": 156400 }, { "epoch": 0.0030976, "grad_norm": 0.8084739446640015, "learning_rate": 1.3100966650778975e-05, "loss": 3.1073, "step": 156410 }, { "epoch": 0.0031232, "grad_norm": 1.3429865837097168, "learning_rate": 1.3100201058426074e-05, "loss": 3.1439, "step": 156420 }, { "epoch": 0.0031488, "grad_norm": 0.8536660075187683, "learning_rate": 1.3099435445969189e-05, "loss": 2.9609, "step": 156430 }, { "epoch": 0.0031744, "grad_norm": 0.7411107420921326, "learning_rate": 1.3098669813413289e-05, "loss": 3.1584, "step": 156440 }, { "epoch": 0.0032, "grad_norm": 0.6966136693954468, "learning_rate": 1.309790416076334e-05, "loss": 2.8856, "step": 156450 }, { "epoch": 0.0032256, "grad_norm": 0.7835645079612732, "learning_rate": 1.3097138488024305e-05, "loss": 3.2906, "step": 156460 }, { "epoch": 0.0032512, "grad_norm": 0.7109268307685852, "learning_rate": 1.309637279520115e-05, "loss": 3.0612, "step": 156470 }, { "epoch": 0.0032768, "grad_norm": 0.7360507249832153, "learning_rate": 1.309560708229884e-05, "loss": 3.0262, "step": 156480 }, { "epoch": 0.0033024, "grad_norm": 0.9745792150497437, "learning_rate": 1.3094841349322343e-05, "loss": 3.3486, "step": 156490 }, { "epoch": 0.003328, "grad_norm": 0.7118180394172668, "learning_rate": 1.3094075596276621e-05, "loss": 3.105, "step": 156500 }, { "epoch": 0.0033536, "grad_norm": 0.7073668241500854, "learning_rate": 1.3093309823166642e-05, "loss": 3.0486, "step": 156510 }, { "epoch": 0.0033792, "grad_norm": 0.6707995533943176, "learning_rate": 1.3092544029997371e-05, "loss": 3.1165, "step": 156520 }, { "epoch": 0.0034048, "grad_norm": 0.8487180471420288, "learning_rate": 1.3091778216773772e-05, "loss": 2.9258, "step": 156530 }, { "epoch": 0.0034304, "grad_norm": 0.8387988209724426, "learning_rate": 1.3091012383500814e-05, "loss": 2.8373, "step": 156540 }, { "epoch": 0.003456, "grad_norm": 0.784358024597168, "learning_rate": 1.3090246530183461e-05, "loss": 3.3247, "step": 156550 }, { "epoch": 0.0034816, "grad_norm": 0.7675186395645142, "learning_rate": 1.3089480656826684e-05, "loss": 3.11, "step": 156560 }, { "epoch": 0.0035072, "grad_norm": 0.7000558972358704, "learning_rate": 1.3088714763435441e-05, "loss": 2.8463, "step": 156570 }, { "epoch": 0.0035328, "grad_norm": 0.6878183484077454, "learning_rate": 1.3087948850014707e-05, "loss": 2.9866, "step": 156580 }, { "epoch": 0.0035584, "grad_norm": 0.7454559803009033, "learning_rate": 1.3087182916569446e-05, "loss": 2.8786, "step": 156590 }, { "epoch": 0.003584, "grad_norm": 0.9809410572052002, "learning_rate": 1.3086416963104623e-05, "loss": 3.209, "step": 156600 }, { "epoch": 0.0036096, "grad_norm": 0.9941956996917725, "learning_rate": 1.3085650989625205e-05, "loss": 3.0877, "step": 156610 }, { "epoch": 0.0036352, "grad_norm": 1.9774123430252075, "learning_rate": 1.3084884996136162e-05, "loss": 3.0101, "step": 156620 }, { "epoch": 0.0036608, "grad_norm": 0.8236140608787537, "learning_rate": 1.3084118982642458e-05, "loss": 3.1044, "step": 156630 }, { "epoch": 0.0036864, "grad_norm": 0.8107465505599976, "learning_rate": 1.3083352949149062e-05, "loss": 3.1417, "step": 156640 }, { "epoch": 0.003712, "grad_norm": 0.8218506574630737, "learning_rate": 1.3082586895660945e-05, "loss": 2.922, "step": 156650 }, { "epoch": 0.0037376, "grad_norm": 0.8353140950202942, "learning_rate": 1.308182082218307e-05, "loss": 3.2549, "step": 156660 }, { "epoch": 0.0037632, "grad_norm": 0.7824971675872803, "learning_rate": 1.3081054728720403e-05, "loss": 3.1183, "step": 156670 }, { "epoch": 0.0037888, "grad_norm": 0.8752866387367249, "learning_rate": 1.3080288615277919e-05, "loss": 3.1811, "step": 156680 }, { "epoch": 0.0038144, "grad_norm": 0.8572845458984375, "learning_rate": 1.3079522481860578e-05, "loss": 3.093, "step": 156690 }, { "epoch": 0.00384, "grad_norm": 1.0470550060272217, "learning_rate": 1.3078756328473355e-05, "loss": 3.0439, "step": 156700 }, { "epoch": 0.0038656, "grad_norm": 0.7855179309844971, "learning_rate": 1.3077990155121215e-05, "loss": 3.0588, "step": 156710 }, { "epoch": 0.0038912, "grad_norm": 1.4850919246673584, "learning_rate": 1.3077223961809124e-05, "loss": 2.9575, "step": 156720 }, { "epoch": 0.0039168, "grad_norm": 0.7664437294006348, "learning_rate": 1.3076457748542057e-05, "loss": 3.1692, "step": 156730 }, { "epoch": 0.0039424, "grad_norm": 0.8203683495521545, "learning_rate": 1.3075691515324977e-05, "loss": 2.9329, "step": 156740 }, { "epoch": 0.003968, "grad_norm": 0.7372188568115234, "learning_rate": 1.307492526216286e-05, "loss": 2.7338, "step": 156750 }, { "epoch": 0.0039936, "grad_norm": 0.7390174865722656, "learning_rate": 1.3074158989060663e-05, "loss": 3.2052, "step": 156760 }, { "epoch": 0.0040192, "grad_norm": 0.7602416276931763, "learning_rate": 1.3073392696023365e-05, "loss": 2.8633, "step": 156770 }, { "epoch": 0.0040448, "grad_norm": 0.814864456653595, "learning_rate": 1.3072626383055933e-05, "loss": 3.1726, "step": 156780 }, { "epoch": 0.0040704, "grad_norm": 1.3002867698669434, "learning_rate": 1.3071860050163334e-05, "loss": 3.0288, "step": 156790 }, { "epoch": 0.004096, "grad_norm": 0.7825523614883423, "learning_rate": 1.3071093697350538e-05, "loss": 3.1359, "step": 156800 }, { "epoch": 0.0041216, "grad_norm": 0.6858330965042114, "learning_rate": 1.3070327324622519e-05, "loss": 2.7658, "step": 156810 }, { "epoch": 0.0041472, "grad_norm": 0.9409710168838501, "learning_rate": 1.3069560931984241e-05, "loss": 3.2222, "step": 156820 }, { "epoch": 0.0041728, "grad_norm": 0.7983074188232422, "learning_rate": 1.3068794519440676e-05, "loss": 3.0042, "step": 156830 }, { "epoch": 0.0041984, "grad_norm": 0.677839457988739, "learning_rate": 1.3068028086996793e-05, "loss": 2.9855, "step": 156840 }, { "epoch": 0.004224, "grad_norm": 0.7307679653167725, "learning_rate": 1.3067261634657568e-05, "loss": 3.135, "step": 156850 }, { "epoch": 0.0042496, "grad_norm": 0.7176709175109863, "learning_rate": 1.3066495162427964e-05, "loss": 2.9821, "step": 156860 }, { "epoch": 0.0042752, "grad_norm": 0.7034798264503479, "learning_rate": 1.3065728670312951e-05, "loss": 2.9616, "step": 156870 }, { "epoch": 0.0043008, "grad_norm": 0.8386746644973755, "learning_rate": 1.3064962158317507e-05, "loss": 3.1217, "step": 156880 }, { "epoch": 0.0043264, "grad_norm": 0.7849603891372681, "learning_rate": 1.3064195626446595e-05, "loss": 3.0227, "step": 156890 }, { "epoch": 0.004352, "grad_norm": 0.7551283240318298, "learning_rate": 1.3063429074705187e-05, "loss": 2.9224, "step": 156900 }, { "epoch": 0.0043776, "grad_norm": 0.8632886409759521, "learning_rate": 1.3062662503098259e-05, "loss": 3.0615, "step": 156910 }, { "epoch": 0.0044032, "grad_norm": 1.058627724647522, "learning_rate": 1.3061895911630777e-05, "loss": 3.0197, "step": 156920 }, { "epoch": 0.0044288, "grad_norm": 0.9330770373344421, "learning_rate": 1.3061129300307714e-05, "loss": 3.2417, "step": 156930 }, { "epoch": 0.0044544, "grad_norm": 0.7230471968650818, "learning_rate": 1.3060362669134042e-05, "loss": 3.0778, "step": 156940 }, { "epoch": 0.00448, "grad_norm": 0.6584292054176331, "learning_rate": 1.3059596018114731e-05, "loss": 2.687, "step": 156950 }, { "epoch": 0.0045056, "grad_norm": 0.716713547706604, "learning_rate": 1.3058829347254754e-05, "loss": 2.8989, "step": 156960 }, { "epoch": 0.0045312, "grad_norm": 0.7051933407783508, "learning_rate": 1.3058062656559079e-05, "loss": 2.9397, "step": 156970 }, { "epoch": 0.0045568, "grad_norm": 0.7752386927604675, "learning_rate": 1.305729594603268e-05, "loss": 2.999, "step": 156980 }, { "epoch": 0.0045824, "grad_norm": 0.8791157007217407, "learning_rate": 1.3056529215680532e-05, "loss": 2.955, "step": 156990 }, { "epoch": 0.004608, "grad_norm": 0.9825304746627808, "learning_rate": 1.3055762465507602e-05, "loss": 3.1398, "step": 157000 }, { "epoch": 0.0046336, "grad_norm": 0.9926678538322449, "learning_rate": 1.305499569551887e-05, "loss": 2.9338, "step": 157010 }, { "epoch": 0.0046592, "grad_norm": 0.6961506605148315, "learning_rate": 1.3054228905719297e-05, "loss": 3.0382, "step": 157020 }, { "epoch": 0.0046848, "grad_norm": 0.7577587962150574, "learning_rate": 1.3053462096113863e-05, "loss": 3.1888, "step": 157030 }, { "epoch": 0.0047104, "grad_norm": 0.9336861968040466, "learning_rate": 1.3052695266707538e-05, "loss": 2.9333, "step": 157040 }, { "epoch": 0.004736, "grad_norm": 0.6711830496788025, "learning_rate": 1.3051928417505297e-05, "loss": 3.0353, "step": 157050 }, { "epoch": 0.0047616, "grad_norm": 1.0649679899215698, "learning_rate": 1.305116154851211e-05, "loss": 3.0285, "step": 157060 }, { "epoch": 0.0047872, "grad_norm": 1.0283151865005493, "learning_rate": 1.305039465973295e-05, "loss": 3.1476, "step": 157070 }, { "epoch": 0.0048128, "grad_norm": 1.1821082830429077, "learning_rate": 1.3049627751172796e-05, "loss": 3.0784, "step": 157080 }, { "epoch": 0.0048384, "grad_norm": 0.7324677109718323, "learning_rate": 1.3048860822836613e-05, "loss": 2.9615, "step": 157090 }, { "epoch": 0.004864, "grad_norm": 0.6903595328330994, "learning_rate": 1.304809387472938e-05, "loss": 2.9178, "step": 157100 }, { "epoch": 0.0048896, "grad_norm": 0.682474672794342, "learning_rate": 1.3047326906856066e-05, "loss": 2.8531, "step": 157110 }, { "epoch": 0.0049152, "grad_norm": 0.7922346591949463, "learning_rate": 1.3046559919221648e-05, "loss": 3.3007, "step": 157120 }, { "epoch": 0.0049408, "grad_norm": 0.8252784609794617, "learning_rate": 1.30457929118311e-05, "loss": 2.8148, "step": 157130 }, { "epoch": 0.0049664, "grad_norm": 0.6703030467033386, "learning_rate": 1.304502588468939e-05, "loss": 3.0954, "step": 157140 }, { "epoch": 0.004992, "grad_norm": 1.0228477716445923, "learning_rate": 1.3044258837801503e-05, "loss": 2.9984, "step": 157150 }, { "epoch": 0.0050176, "grad_norm": 0.746637225151062, "learning_rate": 1.3043491771172399e-05, "loss": 2.9879, "step": 157160 }, { "epoch": 0.0050432, "grad_norm": 0.7399764060974121, "learning_rate": 1.3042724684807065e-05, "loss": 3.0889, "step": 157170 }, { "epoch": 0.0050688, "grad_norm": 0.921311616897583, "learning_rate": 1.3041957578710469e-05, "loss": 3.0151, "step": 157180 }, { "epoch": 0.0050944, "grad_norm": 0.8488785624504089, "learning_rate": 1.3041190452887586e-05, "loss": 2.8613, "step": 157190 }, { "epoch": 0.00512, "grad_norm": 0.799111008644104, "learning_rate": 1.304042330734339e-05, "loss": 3.1297, "step": 157200 }, { "epoch": 0.0051456, "grad_norm": 0.7366091012954712, "learning_rate": 1.3039656142082861e-05, "loss": 2.7186, "step": 157210 }, { "epoch": 0.0051712, "grad_norm": 0.7586222290992737, "learning_rate": 1.3038888957110968e-05, "loss": 2.834, "step": 157220 }, { "epoch": 0.0051968, "grad_norm": 0.768435001373291, "learning_rate": 1.3038121752432685e-05, "loss": 2.9844, "step": 157230 }, { "epoch": 0.0052224, "grad_norm": 0.716440737247467, "learning_rate": 1.3037354528052993e-05, "loss": 2.9421, "step": 157240 }, { "epoch": 0.005248, "grad_norm": 0.7416959404945374, "learning_rate": 1.303658728397686e-05, "loss": 3.1098, "step": 157250 }, { "epoch": 0.0052736, "grad_norm": 0.8854365944862366, "learning_rate": 1.3035820020209269e-05, "loss": 2.9923, "step": 157260 }, { "epoch": 0.0052992, "grad_norm": 0.774517834186554, "learning_rate": 1.3035052736755189e-05, "loss": 3.0531, "step": 157270 }, { "epoch": 0.0053248, "grad_norm": 0.9468046426773071, "learning_rate": 1.3034285433619605e-05, "loss": 2.9102, "step": 157280 }, { "epoch": 0.0053504, "grad_norm": 0.8953983187675476, "learning_rate": 1.3033518110807482e-05, "loss": 2.9629, "step": 157290 }, { "epoch": 0.005376, "grad_norm": 0.7402786612510681, "learning_rate": 1.3032750768323798e-05, "loss": 3.218, "step": 157300 }, { "epoch": 0.0054016, "grad_norm": 0.8163862824440002, "learning_rate": 1.3031983406173537e-05, "loss": 2.9467, "step": 157310 }, { "epoch": 0.0054272, "grad_norm": 0.7583193182945251, "learning_rate": 1.3031216024361664e-05, "loss": 2.9222, "step": 157320 }, { "epoch": 0.0054528, "grad_norm": 0.6989954113960266, "learning_rate": 1.3030448622893162e-05, "loss": 2.8617, "step": 157330 }, { "epoch": 0.0054784, "grad_norm": 0.9537332057952881, "learning_rate": 1.3029681201773007e-05, "loss": 2.8647, "step": 157340 }, { "epoch": 0.005504, "grad_norm": 0.7281283140182495, "learning_rate": 1.3028913761006174e-05, "loss": 2.8576, "step": 157350 }, { "epoch": 0.0055296, "grad_norm": 0.7808654308319092, "learning_rate": 1.3028146300597642e-05, "loss": 2.9082, "step": 157360 }, { "epoch": 0.0055552, "grad_norm": 0.8925688862800598, "learning_rate": 1.3027378820552383e-05, "loss": 3.0106, "step": 157370 }, { "epoch": 0.0055808, "grad_norm": 0.7387989163398743, "learning_rate": 1.3026611320875383e-05, "loss": 2.9549, "step": 157380 }, { "epoch": 0.0056064, "grad_norm": 0.7554405331611633, "learning_rate": 1.3025843801571607e-05, "loss": 3.0161, "step": 157390 }, { "epoch": 0.005632, "grad_norm": 0.7752603888511658, "learning_rate": 1.302507626264604e-05, "loss": 2.9238, "step": 157400 }, { "epoch": 0.0056576, "grad_norm": 0.7731270790100098, "learning_rate": 1.302430870410366e-05, "loss": 2.8625, "step": 157410 }, { "epoch": 0.0056832, "grad_norm": 0.7567758560180664, "learning_rate": 1.3023541125949434e-05, "loss": 3.0631, "step": 157420 }, { "epoch": 0.0057088, "grad_norm": 0.8045272827148438, "learning_rate": 1.3022773528188353e-05, "loss": 2.9217, "step": 157430 }, { "epoch": 0.0057344, "grad_norm": 0.6889452338218689, "learning_rate": 1.3022005910825391e-05, "loss": 2.9547, "step": 157440 }, { "epoch": 0.00576, "grad_norm": 0.8323574662208557, "learning_rate": 1.302123827386552e-05, "loss": 2.8415, "step": 157450 }, { "epoch": 0.0057856, "grad_norm": 0.9881680607795715, "learning_rate": 1.3020470617313724e-05, "loss": 2.8466, "step": 157460 }, { "epoch": 0.0058112, "grad_norm": 0.8045130968093872, "learning_rate": 1.3019702941174976e-05, "loss": 2.9185, "step": 157470 }, { "epoch": 0.0058368, "grad_norm": 0.7288213968276978, "learning_rate": 1.301893524545426e-05, "loss": 2.8666, "step": 157480 }, { "epoch": 0.0058624, "grad_norm": 0.8346871733665466, "learning_rate": 1.301816753015655e-05, "loss": 3.0549, "step": 157490 }, { "epoch": 0.005888, "grad_norm": 0.7177994847297668, "learning_rate": 1.3017399795286824e-05, "loss": 2.8284, "step": 157500 }, { "epoch": 0.0059136, "grad_norm": 0.8289815187454224, "learning_rate": 1.3016632040850066e-05, "loss": 3.2362, "step": 157510 }, { "epoch": 0.0059392, "grad_norm": 0.7139100432395935, "learning_rate": 1.3015864266851249e-05, "loss": 2.9873, "step": 157520 }, { "epoch": 0.0059648, "grad_norm": 0.7384265661239624, "learning_rate": 1.3015096473295352e-05, "loss": 3.1276, "step": 157530 }, { "epoch": 0.0059904, "grad_norm": 0.670535147190094, "learning_rate": 1.3014328660187359e-05, "loss": 2.9037, "step": 157540 }, { "epoch": 0.006016, "grad_norm": 0.7515684366226196, "learning_rate": 1.3013560827532244e-05, "loss": 3.0926, "step": 157550 }, { "epoch": 0.0060416, "grad_norm": 0.7458747625350952, "learning_rate": 1.3012792975334986e-05, "loss": 2.9947, "step": 157560 }, { "epoch": 0.0060672, "grad_norm": 0.765641450881958, "learning_rate": 1.301202510360057e-05, "loss": 2.8821, "step": 157570 }, { "epoch": 0.0060928, "grad_norm": 0.9736091494560242, "learning_rate": 1.301125721233397e-05, "loss": 2.903, "step": 157580 }, { "epoch": 0.0061184, "grad_norm": 0.7786723375320435, "learning_rate": 1.3010489301540165e-05, "loss": 2.9058, "step": 157590 }, { "epoch": 0.006144, "grad_norm": 0.6802018880844116, "learning_rate": 1.3009721371224142e-05, "loss": 3.0389, "step": 157600 }, { "epoch": 0.0061696, "grad_norm": 0.6872896552085876, "learning_rate": 1.3008953421390873e-05, "loss": 2.8486, "step": 157610 }, { "epoch": 0.0061952, "grad_norm": 0.8714033365249634, "learning_rate": 1.300818545204534e-05, "loss": 3.0715, "step": 157620 }, { "epoch": 0.0062208, "grad_norm": 0.6964694261550903, "learning_rate": 1.3007417463192523e-05, "loss": 2.9426, "step": 157630 }, { "epoch": 0.0062464, "grad_norm": 0.7638799548149109, "learning_rate": 1.3006649454837408e-05, "loss": 2.9328, "step": 157640 }, { "epoch": 0.006272, "grad_norm": 0.8834753036499023, "learning_rate": 1.3005881426984965e-05, "loss": 2.8044, "step": 157650 }, { "epoch": 0.0062976, "grad_norm": 0.691158652305603, "learning_rate": 1.3005113379640185e-05, "loss": 2.7722, "step": 157660 }, { "epoch": 0.0063232, "grad_norm": 0.6960089206695557, "learning_rate": 1.3004345312808039e-05, "loss": 3.1416, "step": 157670 }, { "epoch": 0.0063488, "grad_norm": 0.704704999923706, "learning_rate": 1.300357722649351e-05, "loss": 2.7955, "step": 157680 }, { "epoch": 0.0063744, "grad_norm": 0.7987824082374573, "learning_rate": 1.3002809120701585e-05, "loss": 2.8306, "step": 157690 }, { "epoch": 0.0064, "grad_norm": 0.7940008044242859, "learning_rate": 1.3002040995437243e-05, "loss": 2.8879, "step": 157700 }, { "epoch": 0.0064256, "grad_norm": 1.2280902862548828, "learning_rate": 1.3001272850705459e-05, "loss": 3.1012, "step": 157710 }, { "epoch": 0.0064512, "grad_norm": 0.7304947376251221, "learning_rate": 1.300050468651122e-05, "loss": 2.8788, "step": 157720 }, { "epoch": 0.0064768, "grad_norm": 0.8550583720207214, "learning_rate": 1.2999736502859504e-05, "loss": 2.8958, "step": 157730 }, { "epoch": 0.0065024, "grad_norm": 0.8416258096694946, "learning_rate": 1.2998968299755297e-05, "loss": 2.8677, "step": 157740 }, { "epoch": 0.006528, "grad_norm": 4.395725250244141, "learning_rate": 1.2998200077203574e-05, "loss": 2.9146, "step": 157750 }, { "epoch": 0.0065536, "grad_norm": 0.7366019487380981, "learning_rate": 1.2997431835209323e-05, "loss": 2.8903, "step": 157760 }, { "epoch": 0.0065792, "grad_norm": 0.6696361303329468, "learning_rate": 1.2996663573777522e-05, "loss": 2.9828, "step": 157770 }, { "epoch": 0.0066048, "grad_norm": 1.0106079578399658, "learning_rate": 1.2995895292913151e-05, "loss": 2.7886, "step": 157780 }, { "epoch": 0.0066304, "grad_norm": 0.7394365072250366, "learning_rate": 1.2995126992621199e-05, "loss": 2.8882, "step": 157790 }, { "epoch": 0.006656, "grad_norm": 0.7748926877975464, "learning_rate": 1.2994358672906647e-05, "loss": 3.1552, "step": 157800 }, { "epoch": 0.0066816, "grad_norm": 0.7769863605499268, "learning_rate": 1.2993590333774471e-05, "loss": 2.9403, "step": 157810 }, { "epoch": 0.0067072, "grad_norm": 0.7572225332260132, "learning_rate": 1.2992821975229657e-05, "loss": 3.0228, "step": 157820 }, { "epoch": 0.0067328, "grad_norm": 0.9107100367546082, "learning_rate": 1.2992053597277187e-05, "loss": 2.8836, "step": 157830 }, { "epoch": 0.0067584, "grad_norm": 0.853088915348053, "learning_rate": 1.299128519992205e-05, "loss": 3.1032, "step": 157840 }, { "epoch": 0.006784, "grad_norm": 0.8334886431694031, "learning_rate": 1.2990516783169219e-05, "loss": 2.9697, "step": 157850 }, { "epoch": 0.0068096, "grad_norm": 0.808171272277832, "learning_rate": 1.2989748347023681e-05, "loss": 2.9379, "step": 157860 }, { "epoch": 0.0068352, "grad_norm": 0.7143435478210449, "learning_rate": 1.2988979891490422e-05, "loss": 3.0853, "step": 157870 }, { "epoch": 0.0068608, "grad_norm": 0.714730978012085, "learning_rate": 1.298821141657442e-05, "loss": 3.0521, "step": 157880 }, { "epoch": 0.0068864, "grad_norm": 0.7379032969474792, "learning_rate": 1.2987442922280664e-05, "loss": 2.9826, "step": 157890 }, { "epoch": 0.006912, "grad_norm": 0.7951793074607849, "learning_rate": 1.2986674408614134e-05, "loss": 3.1404, "step": 157900 }, { "epoch": 0.0069376, "grad_norm": 1.3112714290618896, "learning_rate": 1.2985905875579813e-05, "loss": 3.1755, "step": 157910 }, { "epoch": 0.0069632, "grad_norm": 0.832281768321991, "learning_rate": 1.2985137323182687e-05, "loss": 2.7081, "step": 157920 }, { "epoch": 0.0069888, "grad_norm": 0.743211030960083, "learning_rate": 1.2984368751427738e-05, "loss": 2.9818, "step": 157930 }, { "epoch": 0.0070144, "grad_norm": 0.842400312423706, "learning_rate": 1.298360016031995e-05, "loss": 3.2284, "step": 157940 }, { "epoch": 0.00704, "grad_norm": 0.8054184913635254, "learning_rate": 1.298283154986431e-05, "loss": 2.9228, "step": 157950 }, { "epoch": 0.0070656, "grad_norm": 0.7924818992614746, "learning_rate": 1.2982062920065798e-05, "loss": 2.9773, "step": 157960 }, { "epoch": 0.0070912, "grad_norm": 0.8226363062858582, "learning_rate": 1.2981294270929405e-05, "loss": 2.843, "step": 157970 }, { "epoch": 0.0071168, "grad_norm": 0.7069033980369568, "learning_rate": 1.2980525602460107e-05, "loss": 3.0391, "step": 157980 }, { "epoch": 0.0071424, "grad_norm": 0.8968921899795532, "learning_rate": 1.2979756914662894e-05, "loss": 3.0106, "step": 157990 }, { "epoch": 0.007168, "grad_norm": 0.7836350202560425, "learning_rate": 1.2978988207542748e-05, "loss": 3.0335, "step": 158000 }, { "epoch": 0.0071936, "grad_norm": 0.9205614328384399, "learning_rate": 1.2978219481104657e-05, "loss": 3.2503, "step": 158010 }, { "epoch": 0.0072192, "grad_norm": 0.7343874573707581, "learning_rate": 1.2977450735353604e-05, "loss": 3.0514, "step": 158020 }, { "epoch": 0.0072448, "grad_norm": 0.7298985123634338, "learning_rate": 1.2976681970294571e-05, "loss": 3.2837, "step": 158030 }, { "epoch": 0.0072704, "grad_norm": 0.7194582223892212, "learning_rate": 1.2975913185932549e-05, "loss": 3.0351, "step": 158040 }, { "epoch": 0.007296, "grad_norm": 0.8200139403343201, "learning_rate": 1.297514438227252e-05, "loss": 3.1078, "step": 158050 }, { "epoch": 0.0073216, "grad_norm": 0.736244261264801, "learning_rate": 1.2974375559319473e-05, "loss": 3.043, "step": 158060 }, { "epoch": 0.0073472, "grad_norm": 0.8073552250862122, "learning_rate": 1.2973606717078388e-05, "loss": 3.2216, "step": 158070 }, { "epoch": 0.0073728, "grad_norm": 0.908913791179657, "learning_rate": 1.2972837855554256e-05, "loss": 3.0968, "step": 158080 }, { "epoch": 0.0073984, "grad_norm": 0.7577901482582092, "learning_rate": 1.297206897475206e-05, "loss": 2.989, "step": 158090 }, { "epoch": 0.007424, "grad_norm": 0.6750307083129883, "learning_rate": 1.2971300074676787e-05, "loss": 3.0996, "step": 158100 }, { "epoch": 0.0074496, "grad_norm": 0.7931209206581116, "learning_rate": 1.2970531155333419e-05, "loss": 2.9387, "step": 158110 }, { "epoch": 0.0074752, "grad_norm": 0.8058444261550903, "learning_rate": 1.2969762216726949e-05, "loss": 3.1221, "step": 158120 }, { "epoch": 0.0075008, "grad_norm": 0.8157444596290588, "learning_rate": 1.296899325886236e-05, "loss": 3.0075, "step": 158130 }, { "epoch": 0.0075264, "grad_norm": 0.6580027341842651, "learning_rate": 1.2968224281744639e-05, "loss": 3.0261, "step": 158140 }, { "epoch": 0.007552, "grad_norm": 0.7619990110397339, "learning_rate": 1.2967455285378772e-05, "loss": 2.9869, "step": 158150 }, { "epoch": 0.0075776, "grad_norm": 0.7030697464942932, "learning_rate": 1.2966686269769745e-05, "loss": 3.0046, "step": 158160 }, { "epoch": 0.0076032, "grad_norm": 0.8819656372070312, "learning_rate": 1.2965917234922549e-05, "loss": 3.1496, "step": 158170 }, { "epoch": 0.0076288, "grad_norm": 0.7753493189811707, "learning_rate": 1.2965148180842166e-05, "loss": 3.1288, "step": 158180 }, { "epoch": 0.0076544, "grad_norm": 0.7428973913192749, "learning_rate": 1.2964379107533587e-05, "loss": 2.9506, "step": 158190 }, { "epoch": 0.00768, "grad_norm": 0.6670449376106262, "learning_rate": 1.2963610015001795e-05, "loss": 3.0139, "step": 158200 }, { "epoch": 0.0077056, "grad_norm": 0.9028905034065247, "learning_rate": 1.296284090325178e-05, "loss": 2.9858, "step": 158210 }, { "epoch": 0.0077312, "grad_norm": 0.7545561790466309, "learning_rate": 1.2962071772288529e-05, "loss": 2.9166, "step": 158220 }, { "epoch": 0.0077568, "grad_norm": 0.7280651926994324, "learning_rate": 1.296130262211703e-05, "loss": 2.9648, "step": 158230 }, { "epoch": 0.0077824, "grad_norm": 0.8229560256004333, "learning_rate": 1.2960533452742272e-05, "loss": 3.0892, "step": 158240 }, { "epoch": 0.007808, "grad_norm": 0.7222123146057129, "learning_rate": 1.2959764264169241e-05, "loss": 2.879, "step": 158250 }, { "epoch": 0.0078336, "grad_norm": 0.7029297947883606, "learning_rate": 1.2958995056402923e-05, "loss": 2.9964, "step": 158260 }, { "epoch": 0.0078592, "grad_norm": 0.735322892665863, "learning_rate": 1.2958225829448312e-05, "loss": 2.9529, "step": 158270 }, { "epoch": 0.0078848, "grad_norm": 0.7041569352149963, "learning_rate": 1.295745658331039e-05, "loss": 2.9017, "step": 158280 }, { "epoch": 0.0079104, "grad_norm": 0.7318933606147766, "learning_rate": 1.295668731799415e-05, "loss": 2.9498, "step": 158290 }, { "epoch": 0.007936, "grad_norm": 0.6838696002960205, "learning_rate": 1.2955918033504575e-05, "loss": 3.1447, "step": 158300 }, { "epoch": 0.0079616, "grad_norm": 0.8654678463935852, "learning_rate": 1.2955148729846661e-05, "loss": 2.9674, "step": 158310 }, { "epoch": 0.0079872, "grad_norm": 0.7502714991569519, "learning_rate": 1.2954379407025391e-05, "loss": 2.9674, "step": 158320 }, { "epoch": 0.0080128, "grad_norm": 0.7646512985229492, "learning_rate": 1.2953610065045757e-05, "loss": 2.9139, "step": 158330 }, { "epoch": 0.0080384, "grad_norm": 1.0022064447402954, "learning_rate": 1.2952840703912746e-05, "loss": 3.045, "step": 158340 }, { "epoch": 0.008064, "grad_norm": 0.6897217035293579, "learning_rate": 1.2952071323631345e-05, "loss": 3.0355, "step": 158350 }, { "epoch": 0.0080896, "grad_norm": 0.6891476511955261, "learning_rate": 1.295130192420655e-05, "loss": 2.9302, "step": 158360 }, { "epoch": 0.0081152, "grad_norm": 0.8627844452857971, "learning_rate": 1.2950532505643345e-05, "loss": 3.1553, "step": 158370 }, { "epoch": 0.0081408, "grad_norm": 0.8721771836280823, "learning_rate": 1.294976306794672e-05, "loss": 2.9676, "step": 158380 }, { "epoch": 0.0081664, "grad_norm": 1.512261986732483, "learning_rate": 1.2948993611121668e-05, "loss": 3.0201, "step": 158390 }, { "epoch": 0.008192, "grad_norm": 0.7961944341659546, "learning_rate": 1.2948224135173173e-05, "loss": 2.9329, "step": 158400 }, { "epoch": 0.0082176, "grad_norm": 0.6924070119857788, "learning_rate": 1.2947454640106229e-05, "loss": 3.0044, "step": 158410 }, { "epoch": 0.0082432, "grad_norm": 1.0833500623703003, "learning_rate": 1.2946685125925827e-05, "loss": 3.2726, "step": 158420 }, { "epoch": 0.0082688, "grad_norm": 1.3636233806610107, "learning_rate": 1.2945915592636956e-05, "loss": 2.985, "step": 158430 }, { "epoch": 0.0082944, "grad_norm": 0.6384313106536865, "learning_rate": 1.29451460402446e-05, "loss": 2.9304, "step": 158440 }, { "epoch": 0.00832, "grad_norm": 0.7613054513931274, "learning_rate": 1.294437646875376e-05, "loss": 2.9936, "step": 158450 }, { "epoch": 0.0083456, "grad_norm": 0.8427127599716187, "learning_rate": 1.2943606878169417e-05, "loss": 3.0807, "step": 158460 }, { "epoch": 0.0083712, "grad_norm": 0.6935970187187195, "learning_rate": 1.2942837268496566e-05, "loss": 2.9458, "step": 158470 }, { "epoch": 0.0083968, "grad_norm": 1.1900993585586548, "learning_rate": 1.29420676397402e-05, "loss": 2.9307, "step": 158480 }, { "epoch": 0.0084224, "grad_norm": 0.883732259273529, "learning_rate": 1.2941297991905306e-05, "loss": 3.1519, "step": 158490 }, { "epoch": 0.008448, "grad_norm": 0.8768525719642639, "learning_rate": 1.2940528324996878e-05, "loss": 3.0867, "step": 158500 }, { "epoch": 0.0084736, "grad_norm": 0.7951394319534302, "learning_rate": 1.2939758639019902e-05, "loss": 3.0528, "step": 158510 }, { "epoch": 0.0084992, "grad_norm": 0.7999143004417419, "learning_rate": 1.2938988933979373e-05, "loss": 3.144, "step": 158520 }, { "epoch": 0.0085248, "grad_norm": 0.7702014446258545, "learning_rate": 1.2938219209880285e-05, "loss": 3.0122, "step": 158530 }, { "epoch": 0.0085504, "grad_norm": 0.737062394618988, "learning_rate": 1.2937449466727622e-05, "loss": 2.88, "step": 158540 }, { "epoch": 0.008576, "grad_norm": 2.1713430881500244, "learning_rate": 1.2936679704526383e-05, "loss": 2.9118, "step": 158550 }, { "epoch": 0.0086016, "grad_norm": 0.9030700325965881, "learning_rate": 1.2935909923281558e-05, "loss": 3.0941, "step": 158560 }, { "epoch": 0.0086272, "grad_norm": 0.747414767742157, "learning_rate": 1.2935140122998134e-05, "loss": 2.8773, "step": 158570 }, { "epoch": 0.0086528, "grad_norm": 0.7154574394226074, "learning_rate": 1.2934370303681107e-05, "loss": 3.04, "step": 158580 }, { "epoch": 0.0086784, "grad_norm": 0.7889642119407654, "learning_rate": 1.293360046533547e-05, "loss": 2.9734, "step": 158590 }, { "epoch": 0.008704, "grad_norm": 0.9209099411964417, "learning_rate": 1.2932830607966214e-05, "loss": 2.9961, "step": 158600 }, { "epoch": 0.0087296, "grad_norm": 0.8950599431991577, "learning_rate": 1.2932060731578328e-05, "loss": 2.9553, "step": 158610 }, { "epoch": 0.0087552, "grad_norm": 0.7305883765220642, "learning_rate": 1.293129083617681e-05, "loss": 2.8583, "step": 158620 }, { "epoch": 0.0087808, "grad_norm": 0.680558443069458, "learning_rate": 1.293052092176665e-05, "loss": 3.0965, "step": 158630 }, { "epoch": 0.0088064, "grad_norm": 0.7504251003265381, "learning_rate": 1.2929750988352837e-05, "loss": 2.9672, "step": 158640 }, { "epoch": 0.008832, "grad_norm": 0.8175002932548523, "learning_rate": 1.292898103594037e-05, "loss": 2.8999, "step": 158650 }, { "epoch": 0.0088576, "grad_norm": 0.6855626702308655, "learning_rate": 1.292821106453424e-05, "loss": 3.0122, "step": 158660 }, { "epoch": 0.0088832, "grad_norm": 1.0603234767913818, "learning_rate": 1.292744107413944e-05, "loss": 2.8872, "step": 158670 }, { "epoch": 0.0089088, "grad_norm": 0.6946086883544922, "learning_rate": 1.2926671064760961e-05, "loss": 2.9945, "step": 158680 }, { "epoch": 0.0089344, "grad_norm": 0.8112022876739502, "learning_rate": 1.2925901036403801e-05, "loss": 3.0545, "step": 158690 }, { "epoch": 0.00896, "grad_norm": 0.6968417763710022, "learning_rate": 1.2925130989072947e-05, "loss": 3.2002, "step": 158700 }, { "epoch": 0.0089856, "grad_norm": 0.73453289270401, "learning_rate": 1.2924360922773395e-05, "loss": 3.0793, "step": 158710 }, { "epoch": 0.0090112, "grad_norm": 1.0840176343917847, "learning_rate": 1.2923590837510142e-05, "loss": 3.0285, "step": 158720 }, { "epoch": 0.0090368, "grad_norm": 0.6787621378898621, "learning_rate": 1.292282073328818e-05, "loss": 3.0249, "step": 158730 }, { "epoch": 0.0090624, "grad_norm": 0.8205118775367737, "learning_rate": 1.29220506101125e-05, "loss": 3.149, "step": 158740 }, { "epoch": 0.009088, "grad_norm": 0.7990975975990295, "learning_rate": 1.29212804679881e-05, "loss": 3.0557, "step": 158750 }, { "epoch": 0.0091136, "grad_norm": 0.7632776498794556, "learning_rate": 1.2920510306919973e-05, "loss": 2.9589, "step": 158760 }, { "epoch": 0.0091392, "grad_norm": 0.7674446105957031, "learning_rate": 1.2919740126913112e-05, "loss": 2.9994, "step": 158770 }, { "epoch": 0.0091648, "grad_norm": 0.7632391452789307, "learning_rate": 1.2918969927972512e-05, "loss": 2.9842, "step": 158780 }, { "epoch": 0.0091904, "grad_norm": 1.3782461881637573, "learning_rate": 1.291819971010317e-05, "loss": 2.8865, "step": 158790 }, { "epoch": 0.009216, "grad_norm": 1.599236011505127, "learning_rate": 1.2917429473310077e-05, "loss": 3.1444, "step": 158800 }, { "epoch": 0.0092416, "grad_norm": 0.7619290351867676, "learning_rate": 1.2916659217598225e-05, "loss": 2.8845, "step": 158810 }, { "epoch": 0.0092672, "grad_norm": 1.0910731554031372, "learning_rate": 1.2915888942972618e-05, "loss": 2.7313, "step": 158820 }, { "epoch": 0.0092928, "grad_norm": 0.6974194049835205, "learning_rate": 1.2915118649438245e-05, "loss": 3.1289, "step": 158830 }, { "epoch": 0.0093184, "grad_norm": 0.6724995374679565, "learning_rate": 1.2914348337000102e-05, "loss": 2.8733, "step": 158840 }, { "epoch": 0.009344, "grad_norm": 0.6729476451873779, "learning_rate": 1.2913578005663185e-05, "loss": 3.294, "step": 158850 }, { "epoch": 0.0093696, "grad_norm": 0.7744200825691223, "learning_rate": 1.291280765543249e-05, "loss": 3.2071, "step": 158860 }, { "epoch": 0.0093952, "grad_norm": 0.6862038969993591, "learning_rate": 1.291203728631301e-05, "loss": 2.8075, "step": 158870 }, { "epoch": 0.0094208, "grad_norm": 0.6863150596618652, "learning_rate": 1.2911266898309745e-05, "loss": 3.1324, "step": 158880 }, { "epoch": 0.0094464, "grad_norm": 0.749868631362915, "learning_rate": 1.2910496491427686e-05, "loss": 2.9483, "step": 158890 }, { "epoch": 0.009472, "grad_norm": 0.8320629000663757, "learning_rate": 1.2909726065671829e-05, "loss": 3.0952, "step": 158900 }, { "epoch": 0.0094976, "grad_norm": 0.788653552532196, "learning_rate": 1.290895562104717e-05, "loss": 2.9297, "step": 158910 }, { "epoch": 0.0095232, "grad_norm": 0.7723089456558228, "learning_rate": 1.2908185157558713e-05, "loss": 3.039, "step": 158920 }, { "epoch": 0.0095488, "grad_norm": 0.7135122418403625, "learning_rate": 1.2907414675211444e-05, "loss": 3.1918, "step": 158930 }, { "epoch": 0.0095744, "grad_norm": 0.6797303557395935, "learning_rate": 1.2906644174010366e-05, "loss": 3.025, "step": 158940 }, { "epoch": 0.0096, "grad_norm": 0.7914223074913025, "learning_rate": 1.290587365396047e-05, "loss": 3.0456, "step": 158950 }, { "epoch": 0.0096256, "grad_norm": 1.2285068035125732, "learning_rate": 1.2905103115066759e-05, "loss": 2.9177, "step": 158960 }, { "epoch": 0.0096512, "grad_norm": 0.7490269541740417, "learning_rate": 1.2904332557334222e-05, "loss": 3.0462, "step": 158970 }, { "epoch": 0.0096768, "grad_norm": 0.6924014091491699, "learning_rate": 1.2903561980767865e-05, "loss": 2.8845, "step": 158980 }, { "epoch": 0.0097024, "grad_norm": 0.750555694103241, "learning_rate": 1.2902791385372677e-05, "loss": 2.9291, "step": 158990 }, { "epoch": 0.009728, "grad_norm": 0.7784501314163208, "learning_rate": 1.290202077115366e-05, "loss": 3.0285, "step": 159000 }, { "epoch": 0.0097536, "grad_norm": 0.8438336849212646, "learning_rate": 1.2901250138115808e-05, "loss": 2.9855, "step": 159010 }, { "epoch": 0.0097792, "grad_norm": 0.7412826418876648, "learning_rate": 1.2900479486264121e-05, "loss": 2.9914, "step": 159020 }, { "epoch": 0.0098048, "grad_norm": 0.8094550967216492, "learning_rate": 1.2899708815603598e-05, "loss": 2.942, "step": 159030 }, { "epoch": 0.0098304, "grad_norm": 0.7325878143310547, "learning_rate": 1.2898938126139228e-05, "loss": 2.9324, "step": 159040 }, { "epoch": 0.009856, "grad_norm": 0.8741501569747925, "learning_rate": 1.289816741787602e-05, "loss": 2.9405, "step": 159050 }, { "epoch": 0.0098816, "grad_norm": 0.7228857278823853, "learning_rate": 1.2897396690818966e-05, "loss": 3.1236, "step": 159060 }, { "epoch": 0.0099072, "grad_norm": 0.8808927536010742, "learning_rate": 1.2896625944973061e-05, "loss": 2.9648, "step": 159070 }, { "epoch": 0.0099328, "grad_norm": 0.7807949781417847, "learning_rate": 1.289585518034331e-05, "loss": 3.1224, "step": 159080 }, { "epoch": 0.0099584, "grad_norm": 0.699545681476593, "learning_rate": 1.2895084396934705e-05, "loss": 2.8694, "step": 159090 }, { "epoch": 0.009984, "grad_norm": 1.589388370513916, "learning_rate": 1.2894313594752248e-05, "loss": 3.0663, "step": 159100 }, { "epoch": 0.0100096, "grad_norm": 0.7960229516029358, "learning_rate": 1.2893542773800936e-05, "loss": 3.0531, "step": 159110 }, { "epoch": 0.0100352, "grad_norm": 0.7880611419677734, "learning_rate": 1.2892771934085771e-05, "loss": 3.0328, "step": 159120 }, { "epoch": 0.0100608, "grad_norm": 0.8088954091072083, "learning_rate": 1.2892001075611749e-05, "loss": 3.1164, "step": 159130 }, { "epoch": 0.0100864, "grad_norm": 0.6695725917816162, "learning_rate": 1.2891230198383864e-05, "loss": 3.19, "step": 159140 }, { "epoch": 0.010112, "grad_norm": 0.9373601078987122, "learning_rate": 1.2890459302407123e-05, "loss": 3.0421, "step": 159150 }, { "epoch": 0.0101376, "grad_norm": 0.7346605062484741, "learning_rate": 1.2889688387686521e-05, "loss": 3.1234, "step": 159160 }, { "epoch": 0.0101632, "grad_norm": 0.9905429482460022, "learning_rate": 1.2888917454227057e-05, "loss": 2.9562, "step": 159170 }, { "epoch": 0.0101888, "grad_norm": 0.6361174583435059, "learning_rate": 1.2888146502033733e-05, "loss": 2.8132, "step": 159180 }, { "epoch": 0.0102144, "grad_norm": 0.7271152138710022, "learning_rate": 1.2887375531111543e-05, "loss": 2.9426, "step": 159190 }, { "epoch": 0.01024, "grad_norm": 0.7464186549186707, "learning_rate": 1.2886604541465494e-05, "loss": 3.0153, "step": 159200 }, { "epoch": 0.0102656, "grad_norm": 0.6904612183570862, "learning_rate": 1.2885833533100578e-05, "loss": 3.1581, "step": 159210 }, { "epoch": 0.0102912, "grad_norm": 0.7624492049217224, "learning_rate": 1.2885062506021803e-05, "loss": 2.9676, "step": 159220 }, { "epoch": 0.0103168, "grad_norm": 0.7502874135971069, "learning_rate": 1.2884291460234161e-05, "loss": 3.1931, "step": 159230 }, { "epoch": 0.0103424, "grad_norm": 0.9597269296646118, "learning_rate": 1.2883520395742654e-05, "loss": 3.152, "step": 159240 }, { "epoch": 0.010368, "grad_norm": 0.8296472430229187, "learning_rate": 1.2882749312552288e-05, "loss": 3.0121, "step": 159250 }, { "epoch": 0.0103936, "grad_norm": 1.186745047569275, "learning_rate": 1.2881978210668056e-05, "loss": 3.0847, "step": 159260 }, { "epoch": 0.0104192, "grad_norm": 0.9124724268913269, "learning_rate": 1.2881207090094963e-05, "loss": 3.1815, "step": 159270 }, { "epoch": 0.0104448, "grad_norm": 0.9483534693717957, "learning_rate": 1.2880435950838006e-05, "loss": 3.2739, "step": 159280 }, { "epoch": 0.0104704, "grad_norm": 0.7152683138847351, "learning_rate": 1.2879664792902188e-05, "loss": 2.7153, "step": 159290 }, { "epoch": 0.010496, "grad_norm": 0.7768272161483765, "learning_rate": 1.287889361629251e-05, "loss": 3.0036, "step": 159300 }, { "epoch": 0.0105216, "grad_norm": 0.8308396935462952, "learning_rate": 1.287812242101397e-05, "loss": 2.9128, "step": 159310 }, { "epoch": 0.0105472, "grad_norm": 0.6679433584213257, "learning_rate": 1.2877351207071574e-05, "loss": 2.9889, "step": 159320 }, { "epoch": 0.0105728, "grad_norm": 1.0085344314575195, "learning_rate": 1.2876579974470315e-05, "loss": 3.0705, "step": 159330 }, { "epoch": 0.0105984, "grad_norm": 0.7354730367660522, "learning_rate": 1.2875808723215204e-05, "loss": 2.8722, "step": 159340 }, { "epoch": 0.010624, "grad_norm": 0.8119280338287354, "learning_rate": 1.287503745331124e-05, "loss": 2.877, "step": 159350 }, { "epoch": 0.0106496, "grad_norm": 0.7937350869178772, "learning_rate": 1.2874266164763416e-05, "loss": 2.9445, "step": 159360 }, { "epoch": 0.0106752, "grad_norm": 1.267627477645874, "learning_rate": 1.2873494857576742e-05, "loss": 3.1108, "step": 159370 }, { "epoch": 0.0107008, "grad_norm": 0.723141610622406, "learning_rate": 1.287272353175622e-05, "loss": 3.0141, "step": 159380 }, { "epoch": 0.0107264, "grad_norm": 0.7421491146087646, "learning_rate": 1.2871952187306849e-05, "loss": 3.1938, "step": 159390 }, { "epoch": 0.010752, "grad_norm": 0.7825025320053101, "learning_rate": 1.2871180824233629e-05, "loss": 2.8276, "step": 159400 }, { "epoch": 0.0107776, "grad_norm": 0.7112756967544556, "learning_rate": 1.2870409442541565e-05, "loss": 3.0625, "step": 159410 }, { "epoch": 0.0108032, "grad_norm": 0.6824522018432617, "learning_rate": 1.2869638042235657e-05, "loss": 2.8148, "step": 159420 }, { "epoch": 0.0108288, "grad_norm": 0.8187894225120544, "learning_rate": 1.286886662332091e-05, "loss": 3.099, "step": 159430 }, { "epoch": 0.0108544, "grad_norm": 0.7275810241699219, "learning_rate": 1.2868095185802324e-05, "loss": 2.9263, "step": 159440 }, { "epoch": 0.01088, "grad_norm": 0.832947850227356, "learning_rate": 1.2867323729684909e-05, "loss": 3.1137, "step": 159450 }, { "epoch": 0.0109056, "grad_norm": 0.6797857880592346, "learning_rate": 1.2866552254973658e-05, "loss": 2.9568, "step": 159460 }, { "epoch": 0.0109312, "grad_norm": 0.7150548696517944, "learning_rate": 1.2865780761673576e-05, "loss": 3.0047, "step": 159470 }, { "epoch": 0.0109568, "grad_norm": 0.7176756262779236, "learning_rate": 1.286500924978967e-05, "loss": 3.1126, "step": 159480 }, { "epoch": 0.0109824, "grad_norm": 1.0133949518203735, "learning_rate": 1.286423771932694e-05, "loss": 2.9348, "step": 159490 }, { "epoch": 0.011008, "grad_norm": 0.8166414499282837, "learning_rate": 1.2863466170290389e-05, "loss": 3.0052, "step": 159500 }, { "epoch": 0.0110336, "grad_norm": 0.6645814776420593, "learning_rate": 1.286269460268502e-05, "loss": 3.1652, "step": 159510 }, { "epoch": 0.0110592, "grad_norm": 0.7426900863647461, "learning_rate": 1.2861923016515836e-05, "loss": 3.0368, "step": 159520 }, { "epoch": 0.0110848, "grad_norm": 0.7054117918014526, "learning_rate": 1.2861151411787845e-05, "loss": 2.9387, "step": 159530 }, { "epoch": 0.0111104, "grad_norm": 0.9146630167961121, "learning_rate": 1.2860379788506046e-05, "loss": 3.3706, "step": 159540 }, { "epoch": 0.011136, "grad_norm": 0.734809160232544, "learning_rate": 1.2859608146675444e-05, "loss": 2.91, "step": 159550 }, { "epoch": 0.0111616, "grad_norm": 0.773043155670166, "learning_rate": 1.2858836486301046e-05, "loss": 3.0729, "step": 159560 }, { "epoch": 0.0111872, "grad_norm": 0.7601718306541443, "learning_rate": 1.285806480738785e-05, "loss": 3.0134, "step": 159570 }, { "epoch": 0.0112128, "grad_norm": 1.2227025032043457, "learning_rate": 1.2857293109940866e-05, "loss": 3.2945, "step": 159580 }, { "epoch": 0.0112384, "grad_norm": 1.1461427211761475, "learning_rate": 1.2856521393965089e-05, "loss": 3.1275, "step": 159590 }, { "epoch": 0.011264, "grad_norm": 0.7715551257133484, "learning_rate": 1.2855749659465535e-05, "loss": 2.9927, "step": 159600 }, { "epoch": 0.0112896, "grad_norm": 0.9218350052833557, "learning_rate": 1.2854977906447206e-05, "loss": 2.866, "step": 159610 }, { "epoch": 0.0113152, "grad_norm": 0.766235888004303, "learning_rate": 1.2854206134915099e-05, "loss": 3.0572, "step": 159620 }, { "epoch": 0.0113408, "grad_norm": 0.7818238139152527, "learning_rate": 1.2853434344874227e-05, "loss": 2.9379, "step": 159630 }, { "epoch": 0.0113664, "grad_norm": 0.738613486289978, "learning_rate": 1.285266253632959e-05, "loss": 2.9289, "step": 159640 }, { "epoch": 0.011392, "grad_norm": 0.7144317030906677, "learning_rate": 1.2851890709286197e-05, "loss": 3.0217, "step": 159650 }, { "epoch": 0.0114176, "grad_norm": 0.722210705280304, "learning_rate": 1.2851118863749048e-05, "loss": 2.9901, "step": 159660 }, { "epoch": 0.0114432, "grad_norm": 0.8550010919570923, "learning_rate": 1.285034699972315e-05, "loss": 3.1098, "step": 159670 }, { "epoch": 0.0114688, "grad_norm": 0.8076212406158447, "learning_rate": 1.2849575117213512e-05, "loss": 3.0964, "step": 159680 }, { "epoch": 0.0114944, "grad_norm": 0.869040310382843, "learning_rate": 1.2848803216225134e-05, "loss": 2.945, "step": 159690 }, { "epoch": 0.01152, "grad_norm": 0.8795031309127808, "learning_rate": 1.2848031296763029e-05, "loss": 2.9852, "step": 159700 }, { "epoch": 0.0115456, "grad_norm": 0.7475602626800537, "learning_rate": 1.2847259358832194e-05, "loss": 2.9982, "step": 159710 }, { "epoch": 0.0115712, "grad_norm": 0.7654808759689331, "learning_rate": 1.284648740243764e-05, "loss": 2.9744, "step": 159720 }, { "epoch": 0.0115968, "grad_norm": 0.7657656073570251, "learning_rate": 1.284571542758437e-05, "loss": 2.9228, "step": 159730 }, { "epoch": 0.0116224, "grad_norm": 0.9973211288452148, "learning_rate": 1.2844943434277393e-05, "loss": 3.0534, "step": 159740 }, { "epoch": 0.011648, "grad_norm": 0.8024255633354187, "learning_rate": 1.2844171422521717e-05, "loss": 3.1499, "step": 159750 }, { "epoch": 0.0116736, "grad_norm": 0.8728197813034058, "learning_rate": 1.284339939232234e-05, "loss": 3.0817, "step": 159760 }, { "epoch": 0.0116992, "grad_norm": 0.8391112089157104, "learning_rate": 1.2842627343684276e-05, "loss": 3.2089, "step": 159770 }, { "epoch": 0.0117248, "grad_norm": 0.7883065342903137, "learning_rate": 1.2841855276612529e-05, "loss": 3.0595, "step": 159780 }, { "epoch": 0.0117504, "grad_norm": 0.9453813433647156, "learning_rate": 1.2841083191112107e-05, "loss": 2.984, "step": 159790 }, { "epoch": 0.011776, "grad_norm": 1.602652668952942, "learning_rate": 1.2840311087188013e-05, "loss": 2.7713, "step": 159800 }, { "epoch": 0.0118016, "grad_norm": 1.2052311897277832, "learning_rate": 1.283953896484526e-05, "loss": 3.186, "step": 159810 }, { "epoch": 0.0118272, "grad_norm": 0.8407978415489197, "learning_rate": 1.283876682408885e-05, "loss": 2.862, "step": 159820 }, { "epoch": 0.0118528, "grad_norm": 0.8051459193229675, "learning_rate": 1.283799466492379e-05, "loss": 3.0945, "step": 159830 }, { "epoch": 0.0118784, "grad_norm": 0.9039620161056519, "learning_rate": 1.2837222487355089e-05, "loss": 3.1919, "step": 159840 }, { "epoch": 0.011904, "grad_norm": 0.7876933217048645, "learning_rate": 1.2836450291387755e-05, "loss": 2.8796, "step": 159850 }, { "epoch": 0.0119296, "grad_norm": 0.7765511274337769, "learning_rate": 1.2835678077026794e-05, "loss": 3.1916, "step": 159860 }, { "epoch": 0.0119552, "grad_norm": 0.7642792463302612, "learning_rate": 1.2834905844277217e-05, "loss": 3.1093, "step": 159870 }, { "epoch": 0.0119808, "grad_norm": 0.8862224221229553, "learning_rate": 1.2834133593144027e-05, "loss": 2.9578, "step": 159880 }, { "epoch": 0.0120064, "grad_norm": 0.7327105402946472, "learning_rate": 1.2833361323632232e-05, "loss": 3.0435, "step": 159890 }, { "epoch": 0.012032, "grad_norm": 0.7586178779602051, "learning_rate": 1.2832589035746843e-05, "loss": 2.9488, "step": 159900 }, { "epoch": 0.0120576, "grad_norm": 0.7282135486602783, "learning_rate": 1.283181672949287e-05, "loss": 2.9411, "step": 159910 }, { "epoch": 0.0120832, "grad_norm": 0.7623345255851746, "learning_rate": 1.2831044404875316e-05, "loss": 2.9113, "step": 159920 }, { "epoch": 0.0121088, "grad_norm": 0.9279289841651917, "learning_rate": 1.2830272061899192e-05, "loss": 3.0543, "step": 159930 }, { "epoch": 0.0121344, "grad_norm": 0.7990336418151855, "learning_rate": 1.2829499700569503e-05, "loss": 3.0949, "step": 159940 }, { "epoch": 0.01216, "grad_norm": 0.8122403025627136, "learning_rate": 1.2828727320891261e-05, "loss": 2.9646, "step": 159950 }, { "epoch": 0.0121856, "grad_norm": 0.7752038240432739, "learning_rate": 1.2827954922869476e-05, "loss": 2.8501, "step": 159960 }, { "epoch": 0.0122112, "grad_norm": 0.8715935349464417, "learning_rate": 1.2827182506509157e-05, "loss": 3.0639, "step": 159970 }, { "epoch": 0.0122368, "grad_norm": 0.7618805170059204, "learning_rate": 1.2826410071815308e-05, "loss": 3.0302, "step": 159980 }, { "epoch": 0.0122624, "grad_norm": 0.8062857389450073, "learning_rate": 1.2825637618792941e-05, "loss": 3.0276, "step": 159990 }, { "epoch": 0.012288, "grad_norm": 0.9154497981071472, "learning_rate": 1.2824865147447065e-05, "loss": 3.215, "step": 160000 }, { "epoch": 0.0123136, "grad_norm": 0.7330613136291504, "learning_rate": 1.2824092657782691e-05, "loss": 2.9999, "step": 160010 }, { "epoch": 0.0123392, "grad_norm": 1.0688722133636475, "learning_rate": 1.2823320149804825e-05, "loss": 2.9537, "step": 160020 }, { "epoch": 0.0123648, "grad_norm": 0.7361816167831421, "learning_rate": 1.2822547623518476e-05, "loss": 2.7873, "step": 160030 }, { "epoch": 0.0123904, "grad_norm": 4.275111675262451, "learning_rate": 1.282177507892866e-05, "loss": 3.0368, "step": 160040 }, { "epoch": 0.012416, "grad_norm": 0.7812186479568481, "learning_rate": 1.282100251604038e-05, "loss": 2.9523, "step": 160050 }, { "epoch": 0.0124416, "grad_norm": 0.7198354601860046, "learning_rate": 1.2820229934858648e-05, "loss": 3.0484, "step": 160060 }, { "epoch": 0.0124672, "grad_norm": 0.6853269338607788, "learning_rate": 1.2819457335388476e-05, "loss": 2.9493, "step": 160070 }, { "epoch": 0.0124928, "grad_norm": 0.7479275465011597, "learning_rate": 1.2818684717634872e-05, "loss": 2.9424, "step": 160080 }, { "epoch": 0.0125184, "grad_norm": 0.8418031930923462, "learning_rate": 1.2817912081602848e-05, "loss": 3.0046, "step": 160090 }, { "epoch": 0.012544, "grad_norm": 0.7696672677993774, "learning_rate": 1.2817139427297412e-05, "loss": 3.3382, "step": 160100 }, { "epoch": 0.0125696, "grad_norm": 2.895810842514038, "learning_rate": 1.2816366754723575e-05, "loss": 3.1162, "step": 160110 }, { "epoch": 0.0125952, "grad_norm": 0.8102920055389404, "learning_rate": 1.2815594063886347e-05, "loss": 2.9714, "step": 160120 }, { "epoch": 0.0126208, "grad_norm": 0.9745992422103882, "learning_rate": 1.2814821354790742e-05, "loss": 2.9344, "step": 160130 }, { "epoch": 0.0126464, "grad_norm": 3.7244415283203125, "learning_rate": 1.2814048627441769e-05, "loss": 2.9759, "step": 160140 }, { "epoch": 0.012672, "grad_norm": 0.9078335762023926, "learning_rate": 1.2813275881844436e-05, "loss": 3.187, "step": 160150 }, { "epoch": 0.0126976, "grad_norm": 0.9327425956726074, "learning_rate": 1.2812503118003758e-05, "loss": 3.1891, "step": 160160 }, { "epoch": 0.0127232, "grad_norm": 0.8407724499702454, "learning_rate": 1.2811730335924746e-05, "loss": 3.1007, "step": 160170 }, { "epoch": 0.0127488, "grad_norm": 0.7979499101638794, "learning_rate": 1.2810957535612408e-05, "loss": 2.9448, "step": 160180 }, { "epoch": 0.0127744, "grad_norm": 0.9486956596374512, "learning_rate": 1.2810184717071758e-05, "loss": 3.2306, "step": 160190 }, { "epoch": 0.0128, "grad_norm": 0.8355842232704163, "learning_rate": 1.2809411880307806e-05, "loss": 2.8916, "step": 160200 }, { "epoch": 0.0128256, "grad_norm": 0.7018094658851624, "learning_rate": 1.2808639025325567e-05, "loss": 3.149, "step": 160210 }, { "epoch": 0.0128512, "grad_norm": 0.7831176519393921, "learning_rate": 1.2807866152130048e-05, "loss": 3.0715, "step": 160220 }, { "epoch": 0.0128768, "grad_norm": 1.082188606262207, "learning_rate": 1.2807093260726264e-05, "loss": 2.8528, "step": 160230 }, { "epoch": 0.0129024, "grad_norm": 0.8127033114433289, "learning_rate": 1.2806320351119229e-05, "loss": 3.2692, "step": 160240 }, { "epoch": 0.012928, "grad_norm": 0.6952275037765503, "learning_rate": 1.2805547423313948e-05, "loss": 3.095, "step": 160250 }, { "epoch": 0.0129536, "grad_norm": 0.7990983724594116, "learning_rate": 1.280477447731544e-05, "loss": 3.318, "step": 160260 }, { "epoch": 0.0129792, "grad_norm": 0.8015307784080505, "learning_rate": 1.2804001513128716e-05, "loss": 2.9934, "step": 160270 }, { "epoch": 0.0130048, "grad_norm": 2.367438793182373, "learning_rate": 1.2803228530758784e-05, "loss": 3.2193, "step": 160280 }, { "epoch": 0.0130304, "grad_norm": 0.8102415800094604, "learning_rate": 1.2802455530210663e-05, "loss": 2.8942, "step": 160290 }, { "epoch": 0.013056, "grad_norm": 0.8986193537712097, "learning_rate": 1.2801682511489362e-05, "loss": 2.8834, "step": 160300 }, { "epoch": 0.0130816, "grad_norm": 1.1987823247909546, "learning_rate": 1.2800909474599892e-05, "loss": 2.8287, "step": 160310 }, { "epoch": 0.0131072, "grad_norm": 1.143073558807373, "learning_rate": 1.280013641954727e-05, "loss": 2.9928, "step": 160320 }, { "epoch": 0.0131328, "grad_norm": 0.7416387796401978, "learning_rate": 1.2799363346336509e-05, "loss": 2.8516, "step": 160330 }, { "epoch": 0.0131584, "grad_norm": 0.8129456043243408, "learning_rate": 1.279859025497262e-05, "loss": 3.2238, "step": 160340 }, { "epoch": 0.013184, "grad_norm": 0.7072232365608215, "learning_rate": 1.2797817145460619e-05, "loss": 3.0924, "step": 160350 }, { "epoch": 0.0132096, "grad_norm": 0.7502305507659912, "learning_rate": 1.2797044017805513e-05, "loss": 3.0988, "step": 160360 }, { "epoch": 0.0132352, "grad_norm": 0.8158115744590759, "learning_rate": 1.2796270872012324e-05, "loss": 2.9711, "step": 160370 }, { "epoch": 0.0132608, "grad_norm": 0.873762309551239, "learning_rate": 1.2795497708086057e-05, "loss": 2.8118, "step": 160380 }, { "epoch": 0.0132864, "grad_norm": 0.8000418543815613, "learning_rate": 1.2794724526031731e-05, "loss": 3.0269, "step": 160390 }, { "epoch": 0.013312, "grad_norm": 0.7205066084861755, "learning_rate": 1.2793951325854363e-05, "loss": 3.0417, "step": 160400 }, { "epoch": 0.0133376, "grad_norm": 1.0553308725357056, "learning_rate": 1.279317810755896e-05, "loss": 3.0678, "step": 160410 }, { "epoch": 0.0133632, "grad_norm": 0.7348912358283997, "learning_rate": 1.2792404871150544e-05, "loss": 2.9149, "step": 160420 }, { "epoch": 0.0133888, "grad_norm": 1.624765396118164, "learning_rate": 1.2791631616634118e-05, "loss": 3.1672, "step": 160430 }, { "epoch": 0.0134144, "grad_norm": 0.7840121984481812, "learning_rate": 1.279085834401471e-05, "loss": 2.9113, "step": 160440 }, { "epoch": 0.01344, "grad_norm": 0.7271910309791565, "learning_rate": 1.2790085053297321e-05, "loss": 3.4061, "step": 160450 }, { "epoch": 0.0134656, "grad_norm": 1.032725214958191, "learning_rate": 1.2789311744486973e-05, "loss": 2.8621, "step": 160460 }, { "epoch": 0.0134912, "grad_norm": 0.9582712054252625, "learning_rate": 1.2788538417588683e-05, "loss": 3.0757, "step": 160470 }, { "epoch": 0.0135168, "grad_norm": 0.8253062963485718, "learning_rate": 1.2787765072607458e-05, "loss": 3.0759, "step": 160480 }, { "epoch": 0.0135424, "grad_norm": 0.7220808863639832, "learning_rate": 1.278699170954832e-05, "loss": 2.9518, "step": 160490 }, { "epoch": 0.013568, "grad_norm": 1.062535047531128, "learning_rate": 1.2786218328416284e-05, "loss": 2.9946, "step": 160500 }, { "epoch": 0.0135936, "grad_norm": 0.7164386510848999, "learning_rate": 1.2785444929216358e-05, "loss": 3.0592, "step": 160510 }, { "epoch": 0.0136192, "grad_norm": 0.8381034135818481, "learning_rate": 1.2784671511953563e-05, "loss": 3.1767, "step": 160520 }, { "epoch": 0.0136448, "grad_norm": 0.7054333090782166, "learning_rate": 1.2783898076632915e-05, "loss": 3.1918, "step": 160530 }, { "epoch": 0.0136704, "grad_norm": 0.8267545700073242, "learning_rate": 1.2783124623259428e-05, "loss": 2.9388, "step": 160540 }, { "epoch": 0.013696, "grad_norm": 0.7869579792022705, "learning_rate": 1.2782351151838116e-05, "loss": 2.993, "step": 160550 }, { "epoch": 0.0137216, "grad_norm": 0.7456094622612, "learning_rate": 1.2781577662373993e-05, "loss": 3.0523, "step": 160560 }, { "epoch": 0.0137472, "grad_norm": 0.7329692244529724, "learning_rate": 1.2780804154872084e-05, "loss": 3.0827, "step": 160570 }, { "epoch": 0.0137728, "grad_norm": 0.7654909491539001, "learning_rate": 1.2780030629337396e-05, "loss": 3.0327, "step": 160580 }, { "epoch": 0.0137984, "grad_norm": 0.8370616436004639, "learning_rate": 1.277925708577495e-05, "loss": 3.0437, "step": 160590 }, { "epoch": 0.013824, "grad_norm": 0.7578884959220886, "learning_rate": 1.2778483524189761e-05, "loss": 3.009, "step": 160600 }, { "epoch": 0.0138496, "grad_norm": 0.7246184349060059, "learning_rate": 1.2777709944586842e-05, "loss": 2.9543, "step": 160610 }, { "epoch": 0.0138752, "grad_norm": 0.6134545803070068, "learning_rate": 1.2776936346971214e-05, "loss": 2.8123, "step": 160620 }, { "epoch": 0.0139008, "grad_norm": 0.7821166515350342, "learning_rate": 1.2776162731347892e-05, "loss": 3.0336, "step": 160630 }, { "epoch": 0.0139264, "grad_norm": 0.8111178278923035, "learning_rate": 1.2775389097721891e-05, "loss": 3.1465, "step": 160640 }, { "epoch": 0.013952, "grad_norm": 0.7878761887550354, "learning_rate": 1.277461544609823e-05, "loss": 3.377, "step": 160650 }, { "epoch": 0.0139776, "grad_norm": 0.8389997482299805, "learning_rate": 1.2773841776481925e-05, "loss": 2.9943, "step": 160660 }, { "epoch": 0.0140032, "grad_norm": 0.7388045191764832, "learning_rate": 1.2773068088877995e-05, "loss": 2.957, "step": 160670 }, { "epoch": 0.0140288, "grad_norm": 0.692157506942749, "learning_rate": 1.2772294383291455e-05, "loss": 2.9369, "step": 160680 }, { "epoch": 0.0140544, "grad_norm": 0.6823390126228333, "learning_rate": 1.2771520659727321e-05, "loss": 2.7588, "step": 160690 }, { "epoch": 0.01408, "grad_norm": 0.7185901999473572, "learning_rate": 1.2770746918190615e-05, "loss": 3.0073, "step": 160700 }, { "epoch": 0.0141056, "grad_norm": 0.7777443528175354, "learning_rate": 1.2769973158686348e-05, "loss": 2.9071, "step": 160710 }, { "epoch": 0.0141312, "grad_norm": 0.7930980324745178, "learning_rate": 1.2769199381219541e-05, "loss": 2.9506, "step": 160720 }, { "epoch": 0.0141568, "grad_norm": 0.776965320110321, "learning_rate": 1.2768425585795216e-05, "loss": 3.2537, "step": 160730 }, { "epoch": 0.0141824, "grad_norm": 0.7841219305992126, "learning_rate": 1.2767651772418383e-05, "loss": 2.9254, "step": 160740 }, { "epoch": 0.014208, "grad_norm": 1.0630871057510376, "learning_rate": 1.2766877941094067e-05, "loss": 2.9447, "step": 160750 }, { "epoch": 0.0142336, "grad_norm": 0.7082913517951965, "learning_rate": 1.2766104091827281e-05, "loss": 2.9255, "step": 160760 }, { "epoch": 0.0142592, "grad_norm": 0.8676468133926392, "learning_rate": 1.2765330224623048e-05, "loss": 3.0056, "step": 160770 }, { "epoch": 0.0142848, "grad_norm": 0.9701303243637085, "learning_rate": 1.2764556339486378e-05, "loss": 3.0836, "step": 160780 }, { "epoch": 0.0143104, "grad_norm": 0.8284077048301697, "learning_rate": 1.2763782436422298e-05, "loss": 2.9405, "step": 160790 }, { "epoch": 0.014336, "grad_norm": 0.707990288734436, "learning_rate": 1.2763008515435825e-05, "loss": 2.9587, "step": 160800 }, { "epoch": 0.0143616, "grad_norm": 0.6764119267463684, "learning_rate": 1.2762234576531972e-05, "loss": 2.9393, "step": 160810 }, { "epoch": 0.0143872, "grad_norm": 0.6605439782142639, "learning_rate": 1.2761460619715764e-05, "loss": 2.9785, "step": 160820 }, { "epoch": 0.0144128, "grad_norm": 0.6807273030281067, "learning_rate": 1.276068664499222e-05, "loss": 3.0469, "step": 160830 }, { "epoch": 0.0144384, "grad_norm": 0.7039932012557983, "learning_rate": 1.2759912652366354e-05, "loss": 3.0894, "step": 160840 }, { "epoch": 0.014464, "grad_norm": 0.6524353623390198, "learning_rate": 1.2759138641843187e-05, "loss": 2.8097, "step": 160850 }, { "epoch": 0.0144896, "grad_norm": 0.8854445219039917, "learning_rate": 1.2758364613427743e-05, "loss": 3.0385, "step": 160860 }, { "epoch": 0.0145152, "grad_norm": 0.7492115497589111, "learning_rate": 1.2757590567125035e-05, "loss": 2.8584, "step": 160870 }, { "epoch": 0.0145408, "grad_norm": 0.8797497749328613, "learning_rate": 1.2756816502940087e-05, "loss": 3.0208, "step": 160880 }, { "epoch": 0.0145664, "grad_norm": 0.7278574109077454, "learning_rate": 1.2756042420877913e-05, "loss": 2.9162, "step": 160890 }, { "epoch": 0.014592, "grad_norm": 1.0688140392303467, "learning_rate": 1.2755268320943539e-05, "loss": 2.9235, "step": 160900 }, { "epoch": 0.0146176, "grad_norm": 1.1723616123199463, "learning_rate": 1.275449420314198e-05, "loss": 3.0548, "step": 160910 }, { "epoch": 0.0146432, "grad_norm": 0.7000181078910828, "learning_rate": 1.2753720067478259e-05, "loss": 3.0325, "step": 160920 }, { "epoch": 0.0146688, "grad_norm": 0.7476467490196228, "learning_rate": 1.2752945913957399e-05, "loss": 2.7934, "step": 160930 }, { "epoch": 0.0146944, "grad_norm": 0.7767773270606995, "learning_rate": 1.2752171742584414e-05, "loss": 3.1021, "step": 160940 }, { "epoch": 0.01472, "grad_norm": 1.2075669765472412, "learning_rate": 1.2751397553364324e-05, "loss": 3.1153, "step": 160950 }, { "epoch": 0.0147456, "grad_norm": 1.3726530075073242, "learning_rate": 1.2750623346302157e-05, "loss": 3.0611, "step": 160960 }, { "epoch": 0.0147712, "grad_norm": 0.6882036924362183, "learning_rate": 1.2749849121402927e-05, "loss": 3.0466, "step": 160970 }, { "epoch": 0.0147968, "grad_norm": 0.7436124682426453, "learning_rate": 1.2749074878671652e-05, "loss": 2.9611, "step": 160980 }, { "epoch": 0.0148224, "grad_norm": 0.6729011535644531, "learning_rate": 1.274830061811336e-05, "loss": 3.0699, "step": 160990 }, { "epoch": 0.014848, "grad_norm": 0.7350685596466064, "learning_rate": 1.274752633973307e-05, "loss": 2.8299, "step": 161000 }, { "epoch": 0.0148736, "grad_norm": 0.753442645072937, "learning_rate": 1.2746752043535801e-05, "loss": 2.9613, "step": 161010 }, { "epoch": 0.0148992, "grad_norm": 0.8250007629394531, "learning_rate": 1.2745977729526575e-05, "loss": 2.7354, "step": 161020 }, { "epoch": 0.0149248, "grad_norm": 0.7355549335479736, "learning_rate": 1.2745203397710415e-05, "loss": 3.0086, "step": 161030 }, { "epoch": 0.0149504, "grad_norm": 0.7243404984474182, "learning_rate": 1.2744429048092337e-05, "loss": 3.09, "step": 161040 }, { "epoch": 0.014976, "grad_norm": 0.7624366879463196, "learning_rate": 1.274365468067737e-05, "loss": 3.2222, "step": 161050 }, { "epoch": 0.0150016, "grad_norm": 0.7929800152778625, "learning_rate": 1.2742880295470528e-05, "loss": 2.8743, "step": 161060 }, { "epoch": 0.0150272, "grad_norm": 0.9736834764480591, "learning_rate": 1.2742105892476837e-05, "loss": 3.194, "step": 161070 }, { "epoch": 0.0150528, "grad_norm": 0.8131333589553833, "learning_rate": 1.2741331471701318e-05, "loss": 3.0382, "step": 161080 }, { "epoch": 0.0150784, "grad_norm": 0.6662613749504089, "learning_rate": 1.2740557033148996e-05, "loss": 2.4771, "step": 161090 }, { "epoch": 0.015104, "grad_norm": 0.8114410042762756, "learning_rate": 1.2739782576824885e-05, "loss": 2.8323, "step": 161100 }, { "epoch": 0.0151296, "grad_norm": 0.8725372552871704, "learning_rate": 1.2739008102734017e-05, "loss": 3.0332, "step": 161110 }, { "epoch": 0.0151552, "grad_norm": 0.6796423196792603, "learning_rate": 1.2738233610881405e-05, "loss": 3.2675, "step": 161120 }, { "epoch": 0.0151808, "grad_norm": 0.775292694568634, "learning_rate": 1.2737459101272078e-05, "loss": 3.008, "step": 161130 }, { "epoch": 0.0152064, "grad_norm": 0.9898127317428589, "learning_rate": 1.2736684573911057e-05, "loss": 2.9167, "step": 161140 }, { "epoch": 0.015232, "grad_norm": 0.7579036355018616, "learning_rate": 1.2735910028803362e-05, "loss": 3.372, "step": 161150 }, { "epoch": 0.0152576, "grad_norm": 0.7949566841125488, "learning_rate": 1.2735135465954019e-05, "loss": 2.9261, "step": 161160 }, { "epoch": 0.0152832, "grad_norm": 0.6776793599128723, "learning_rate": 1.273436088536805e-05, "loss": 3.0592, "step": 161170 }, { "epoch": 0.0153088, "grad_norm": 0.8408713340759277, "learning_rate": 1.2733586287050475e-05, "loss": 2.9037, "step": 161180 }, { "epoch": 0.0153344, "grad_norm": 0.7671926617622375, "learning_rate": 1.2732811671006322e-05, "loss": 2.9636, "step": 161190 }, { "epoch": 0.01536, "grad_norm": 0.8117569088935852, "learning_rate": 1.273203703724061e-05, "loss": 2.8883, "step": 161200 }, { "epoch": 0.0153856, "grad_norm": 0.9122995138168335, "learning_rate": 1.2731262385758364e-05, "loss": 3.0388, "step": 161210 }, { "epoch": 0.0154112, "grad_norm": 0.7769036293029785, "learning_rate": 1.2730487716564609e-05, "loss": 3.2237, "step": 161220 }, { "epoch": 0.0154368, "grad_norm": 1.1490756273269653, "learning_rate": 1.2729713029664366e-05, "loss": 3.0347, "step": 161230 }, { "epoch": 0.0154624, "grad_norm": 0.7761332392692566, "learning_rate": 1.2728938325062658e-05, "loss": 3.2524, "step": 161240 }, { "epoch": 0.015488, "grad_norm": 0.7188382744789124, "learning_rate": 1.272816360276451e-05, "loss": 3.1108, "step": 161250 }, { "epoch": 0.0155136, "grad_norm": 0.9372606873512268, "learning_rate": 1.2727388862774948e-05, "loss": 3.0614, "step": 161260 }, { "epoch": 0.0155392, "grad_norm": 0.8095247149467468, "learning_rate": 1.2726614105098994e-05, "loss": 3.1028, "step": 161270 }, { "epoch": 0.0155648, "grad_norm": 0.8873169422149658, "learning_rate": 1.2725839329741669e-05, "loss": 3.0086, "step": 161280 }, { "epoch": 0.0155904, "grad_norm": 0.7810564041137695, "learning_rate": 1.2725064536708005e-05, "loss": 2.9984, "step": 161290 }, { "epoch": 0.015616, "grad_norm": 1.0517501831054688, "learning_rate": 1.2724289726003021e-05, "loss": 2.8468, "step": 161300 }, { "epoch": 0.0156416, "grad_norm": 0.7522770166397095, "learning_rate": 1.2723514897631739e-05, "loss": 2.8829, "step": 161310 }, { "epoch": 0.0156672, "grad_norm": 0.8001958131790161, "learning_rate": 1.272274005159919e-05, "loss": 2.892, "step": 161320 }, { "epoch": 0.0156928, "grad_norm": 0.6362606883049011, "learning_rate": 1.2721965187910392e-05, "loss": 2.8998, "step": 161330 }, { "epoch": 0.0157184, "grad_norm": 1.0428820848464966, "learning_rate": 1.2721190306570374e-05, "loss": 2.9018, "step": 161340 }, { "epoch": 0.015744, "grad_norm": 0.7353734374046326, "learning_rate": 1.2720415407584162e-05, "loss": 3.0185, "step": 161350 }, { "epoch": 0.0157696, "grad_norm": 0.6934482455253601, "learning_rate": 1.271964049095678e-05, "loss": 2.987, "step": 161360 }, { "epoch": 0.0157952, "grad_norm": 0.9664925336837769, "learning_rate": 1.271886555669325e-05, "loss": 3.102, "step": 161370 }, { "epoch": 0.0158208, "grad_norm": 0.7151677012443542, "learning_rate": 1.27180906047986e-05, "loss": 2.9206, "step": 161380 }, { "epoch": 0.0158464, "grad_norm": 0.6738695502281189, "learning_rate": 1.2717315635277857e-05, "loss": 3.0774, "step": 161390 }, { "epoch": 0.015872, "grad_norm": 0.7482864856719971, "learning_rate": 1.271654064813604e-05, "loss": 2.7543, "step": 161400 }, { "epoch": 0.0158976, "grad_norm": 0.8441137075424194, "learning_rate": 1.271576564337818e-05, "loss": 3.1772, "step": 161410 }, { "epoch": 0.0159232, "grad_norm": 1.5935746431350708, "learning_rate": 1.2714990621009302e-05, "loss": 3.166, "step": 161420 }, { "epoch": 0.0159488, "grad_norm": 0.6822569370269775, "learning_rate": 1.2714215581034433e-05, "loss": 2.9522, "step": 161430 }, { "epoch": 0.0159744, "grad_norm": 1.1382882595062256, "learning_rate": 1.2713440523458593e-05, "loss": 2.8354, "step": 161440 }, { "epoch": 0.016, "grad_norm": 0.9507063627243042, "learning_rate": 1.2712665448286816e-05, "loss": 3.147, "step": 161450 }, { "epoch": 0.0160256, "grad_norm": 0.7607306838035583, "learning_rate": 1.2711890355524126e-05, "loss": 2.9693, "step": 161460 }, { "epoch": 0.0160512, "grad_norm": 0.7410914301872253, "learning_rate": 1.2711115245175545e-05, "loss": 2.9981, "step": 161470 }, { "epoch": 0.0160768, "grad_norm": 0.8249974250793457, "learning_rate": 1.2710340117246103e-05, "loss": 2.884, "step": 161480 }, { "epoch": 0.0161024, "grad_norm": 0.8594032526016235, "learning_rate": 1.2709564971740826e-05, "loss": 2.9511, "step": 161490 }, { "epoch": 0.016128, "grad_norm": 0.7597438097000122, "learning_rate": 1.2708789808664741e-05, "loss": 2.8429, "step": 161500 }, { "epoch": 0.0161536, "grad_norm": 0.9994359612464905, "learning_rate": 1.270801462802287e-05, "loss": 3.1588, "step": 161510 }, { "epoch": 0.0161792, "grad_norm": 0.7745078206062317, "learning_rate": 1.2707239429820248e-05, "loss": 3.0304, "step": 161520 }, { "epoch": 0.0162048, "grad_norm": 0.7939670085906982, "learning_rate": 1.2706464214061898e-05, "loss": 2.9419, "step": 161530 }, { "epoch": 0.0162304, "grad_norm": 0.7803778648376465, "learning_rate": 1.2705688980752844e-05, "loss": 2.9735, "step": 161540 }, { "epoch": 0.016256, "grad_norm": 0.7570012807846069, "learning_rate": 1.2704913729898116e-05, "loss": 3.251, "step": 161550 }, { "epoch": 0.0162816, "grad_norm": 0.6940667629241943, "learning_rate": 1.2704138461502743e-05, "loss": 3.1318, "step": 161560 }, { "epoch": 0.0163072, "grad_norm": 0.7889423966407776, "learning_rate": 1.2703363175571752e-05, "loss": 2.9754, "step": 161570 }, { "epoch": 0.0163328, "grad_norm": 0.8364319205284119, "learning_rate": 1.2702587872110166e-05, "loss": 2.918, "step": 161580 }, { "epoch": 0.0163584, "grad_norm": 0.7476851344108582, "learning_rate": 1.2701812551123018e-05, "loss": 2.9887, "step": 161590 }, { "epoch": 0.016384, "grad_norm": 0.9918330311775208, "learning_rate": 1.2701037212615333e-05, "loss": 3.0881, "step": 161600 }, { "epoch": 0.0164096, "grad_norm": 0.750256359577179, "learning_rate": 1.270026185659214e-05, "loss": 2.9659, "step": 161610 }, { "epoch": 0.0164352, "grad_norm": 0.7455964088439941, "learning_rate": 1.2699486483058468e-05, "loss": 3.0656, "step": 161620 }, { "epoch": 0.0164608, "grad_norm": 0.7253100275993347, "learning_rate": 1.269871109201934e-05, "loss": 2.8523, "step": 161630 }, { "epoch": 0.0164864, "grad_norm": 1.7237942218780518, "learning_rate": 1.2697935683479789e-05, "loss": 2.9133, "step": 161640 }, { "epoch": 0.016512, "grad_norm": 0.8466028571128845, "learning_rate": 1.2697160257444843e-05, "loss": 3.216, "step": 161650 }, { "epoch": 0.0165376, "grad_norm": 0.8150230050086975, "learning_rate": 1.2696384813919531e-05, "loss": 2.95, "step": 161660 }, { "epoch": 0.0165632, "grad_norm": 0.7682480812072754, "learning_rate": 1.269560935290888e-05, "loss": 3.0413, "step": 161670 }, { "epoch": 0.0165888, "grad_norm": 0.738714873790741, "learning_rate": 1.2694833874417915e-05, "loss": 3.1801, "step": 161680 }, { "epoch": 0.0166144, "grad_norm": 0.7485424280166626, "learning_rate": 1.269405837845167e-05, "loss": 3.0394, "step": 161690 }, { "epoch": 0.01664, "grad_norm": 0.8301598429679871, "learning_rate": 1.2693282865015176e-05, "loss": 3.2367, "step": 161700 }, { "epoch": 0.0166656, "grad_norm": 0.8052005767822266, "learning_rate": 1.2692507334113454e-05, "loss": 3.0834, "step": 161710 }, { "epoch": 0.0166912, "grad_norm": 0.6809080839157104, "learning_rate": 1.269173178575154e-05, "loss": 3.2346, "step": 161720 }, { "epoch": 0.0167168, "grad_norm": 0.795150637626648, "learning_rate": 1.269095621993446e-05, "loss": 3.0965, "step": 161730 }, { "epoch": 0.0167424, "grad_norm": 0.8117408156394958, "learning_rate": 1.2690180636667244e-05, "loss": 3.0697, "step": 161740 }, { "epoch": 0.016768, "grad_norm": 0.7595096826553345, "learning_rate": 1.268940503595492e-05, "loss": 3.2433, "step": 161750 }, { "epoch": 0.0167936, "grad_norm": 0.808198869228363, "learning_rate": 1.2688629417802523e-05, "loss": 2.9151, "step": 161760 }, { "epoch": 0.0168192, "grad_norm": 1.4525179862976074, "learning_rate": 1.2687853782215073e-05, "loss": 3.0128, "step": 161770 }, { "epoch": 0.0168448, "grad_norm": 0.79277104139328, "learning_rate": 1.268707812919761e-05, "loss": 3.0853, "step": 161780 }, { "epoch": 0.0168704, "grad_norm": 0.9046763777732849, "learning_rate": 1.268630245875516e-05, "loss": 2.9304, "step": 161790 }, { "epoch": 0.016896, "grad_norm": 0.8681610822677612, "learning_rate": 1.268552677089275e-05, "loss": 2.9862, "step": 161800 }, { "epoch": 0.0169216, "grad_norm": 0.9793742895126343, "learning_rate": 1.2684751065615413e-05, "loss": 3.0568, "step": 161810 }, { "epoch": 0.0169472, "grad_norm": 3.238636016845703, "learning_rate": 1.2683975342928181e-05, "loss": 3.0281, "step": 161820 }, { "epoch": 0.0169728, "grad_norm": 0.7347465753555298, "learning_rate": 1.268319960283608e-05, "loss": 2.8462, "step": 161830 }, { "epoch": 0.0169984, "grad_norm": 0.7735986709594727, "learning_rate": 1.2682423845344143e-05, "loss": 2.951, "step": 161840 }, { "epoch": 0.017024, "grad_norm": 0.9293426871299744, "learning_rate": 1.26816480704574e-05, "loss": 3.0147, "step": 161850 }, { "epoch": 0.0170496, "grad_norm": 0.8351374864578247, "learning_rate": 1.2680872278180883e-05, "loss": 2.9002, "step": 161860 }, { "epoch": 0.0170752, "grad_norm": 0.716023862361908, "learning_rate": 1.268009646851962e-05, "loss": 2.8962, "step": 161870 }, { "epoch": 0.0171008, "grad_norm": 0.9973973035812378, "learning_rate": 1.2679320641478648e-05, "loss": 3.1424, "step": 161880 }, { "epoch": 0.0171264, "grad_norm": 0.7453780770301819, "learning_rate": 1.2678544797062988e-05, "loss": 3.1156, "step": 161890 }, { "epoch": 0.017152, "grad_norm": 0.7233790755271912, "learning_rate": 1.2677768935277678e-05, "loss": 2.9578, "step": 161900 }, { "epoch": 0.0171776, "grad_norm": 0.8108129501342773, "learning_rate": 1.267699305612775e-05, "loss": 2.9955, "step": 161910 }, { "epoch": 0.0172032, "grad_norm": 0.8432162404060364, "learning_rate": 1.2676217159618237e-05, "loss": 2.8678, "step": 161920 }, { "epoch": 0.0172288, "grad_norm": 0.9139381051063538, "learning_rate": 1.2675441245754163e-05, "loss": 3.0844, "step": 161930 }, { "epoch": 0.0172544, "grad_norm": 2.054584264755249, "learning_rate": 1.2674665314540559e-05, "loss": 2.9346, "step": 161940 }, { "epoch": 0.01728, "grad_norm": 0.7036446928977966, "learning_rate": 1.2673889365982468e-05, "loss": 2.9653, "step": 161950 }, { "epoch": 0.0173056, "grad_norm": 1.1681970357894897, "learning_rate": 1.2673113400084913e-05, "loss": 3.1727, "step": 161960 }, { "epoch": 0.0173312, "grad_norm": 0.6757882833480835, "learning_rate": 1.2672337416852928e-05, "loss": 2.9462, "step": 161970 }, { "epoch": 0.0173568, "grad_norm": 0.715634822845459, "learning_rate": 1.2671561416291548e-05, "loss": 3.0143, "step": 161980 }, { "epoch": 0.0173824, "grad_norm": 0.876500129699707, "learning_rate": 1.2670785398405799e-05, "loss": 3.0463, "step": 161990 }, { "epoch": 0.017408, "grad_norm": 0.9315961599349976, "learning_rate": 1.2670009363200715e-05, "loss": 2.9275, "step": 162000 }, { "epoch": 0.0174336, "grad_norm": 1.0121111869812012, "learning_rate": 1.2669233310681332e-05, "loss": 3.1862, "step": 162010 }, { "epoch": 0.0174592, "grad_norm": 0.7738348245620728, "learning_rate": 1.2668457240852683e-05, "loss": 2.7356, "step": 162020 }, { "epoch": 0.0174848, "grad_norm": 0.6674397587776184, "learning_rate": 1.266768115371979e-05, "loss": 3.1085, "step": 162030 }, { "epoch": 0.0175104, "grad_norm": 0.8565925359725952, "learning_rate": 1.26669050492877e-05, "loss": 3.0861, "step": 162040 }, { "epoch": 0.017536, "grad_norm": 1.2490431070327759, "learning_rate": 1.266612892756144e-05, "loss": 2.9522, "step": 162050 }, { "epoch": 0.0175616, "grad_norm": 13.668892860412598, "learning_rate": 1.2665352788546042e-05, "loss": 3.1646, "step": 162060 }, { "epoch": 0.0175872, "grad_norm": 0.8124579787254333, "learning_rate": 1.2664576632246537e-05, "loss": 3.1532, "step": 162070 }, { "epoch": 0.0176128, "grad_norm": 1.1811280250549316, "learning_rate": 1.2663800458667964e-05, "loss": 3.1075, "step": 162080 }, { "epoch": 0.0176384, "grad_norm": 0.9141339063644409, "learning_rate": 1.266302426781535e-05, "loss": 3.0014, "step": 162090 }, { "epoch": 0.017664, "grad_norm": 0.6969896554946899, "learning_rate": 1.266224805969373e-05, "loss": 2.9855, "step": 162100 }, { "epoch": 0.0176896, "grad_norm": 0.9749575853347778, "learning_rate": 1.2661471834308144e-05, "loss": 3.1497, "step": 162110 }, { "epoch": 0.0177152, "grad_norm": 0.7670280933380127, "learning_rate": 1.2660695591663617e-05, "loss": 2.8193, "step": 162120 }, { "epoch": 0.0177408, "grad_norm": 0.6970888376235962, "learning_rate": 1.2659919331765185e-05, "loss": 3.1343, "step": 162130 }, { "epoch": 0.0177664, "grad_norm": 0.6936991810798645, "learning_rate": 1.2659143054617884e-05, "loss": 2.9492, "step": 162140 }, { "epoch": 0.017792, "grad_norm": 0.9469363689422607, "learning_rate": 1.265836676022675e-05, "loss": 3.0242, "step": 162150 }, { "epoch": 0.0178176, "grad_norm": 0.7407118082046509, "learning_rate": 1.2657590448596812e-05, "loss": 3.113, "step": 162160 }, { "epoch": 0.0178432, "grad_norm": 0.8662333488464355, "learning_rate": 1.2656814119733107e-05, "loss": 2.8593, "step": 162170 }, { "epoch": 0.0178688, "grad_norm": 0.8362329006195068, "learning_rate": 1.265603777364067e-05, "loss": 3.089, "step": 162180 }, { "epoch": 0.0178944, "grad_norm": 0.872886061668396, "learning_rate": 1.265526141032453e-05, "loss": 2.8859, "step": 162190 }, { "epoch": 0.01792, "grad_norm": 0.7847909927368164, "learning_rate": 1.2654485029789725e-05, "loss": 3.0217, "step": 162200 }, { "epoch": 0.0179456, "grad_norm": 0.9815323352813721, "learning_rate": 1.2653708632041295e-05, "loss": 3.056, "step": 162210 }, { "epoch": 0.0179712, "grad_norm": 0.7019100189208984, "learning_rate": 1.2652932217084266e-05, "loss": 2.9143, "step": 162220 }, { "epoch": 0.0179968, "grad_norm": 0.7092252969741821, "learning_rate": 1.2652155784923678e-05, "loss": 3.1241, "step": 162230 }, { "epoch": 0.0180224, "grad_norm": 1.1685210466384888, "learning_rate": 1.2651379335564565e-05, "loss": 2.9389, "step": 162240 }, { "epoch": 0.018048, "grad_norm": 0.7494313716888428, "learning_rate": 1.265060286901196e-05, "loss": 2.9992, "step": 162250 }, { "epoch": 0.0180736, "grad_norm": 0.7580208778381348, "learning_rate": 1.2649826385270903e-05, "loss": 3.1098, "step": 162260 }, { "epoch": 0.0180992, "grad_norm": 0.7341330647468567, "learning_rate": 1.2649049884346423e-05, "loss": 3.0813, "step": 162270 }, { "epoch": 0.0181248, "grad_norm": 0.8352303504943848, "learning_rate": 1.264827336624356e-05, "loss": 3.3048, "step": 162280 }, { "epoch": 0.0181504, "grad_norm": 0.7492456436157227, "learning_rate": 1.2647496830967346e-05, "loss": 3.0581, "step": 162290 }, { "epoch": 0.018176, "grad_norm": 1.2216728925704956, "learning_rate": 1.2646720278522819e-05, "loss": 3.0524, "step": 162300 }, { "epoch": 0.0182016, "grad_norm": 0.9162569046020508, "learning_rate": 1.2645943708915018e-05, "loss": 2.9604, "step": 162310 }, { "epoch": 0.0182272, "grad_norm": 0.7524341344833374, "learning_rate": 1.2645167122148973e-05, "loss": 3.087, "step": 162320 }, { "epoch": 0.0182528, "grad_norm": 0.8165287375450134, "learning_rate": 1.264439051822972e-05, "loss": 3.2006, "step": 162330 }, { "epoch": 0.0182784, "grad_norm": 0.8732429146766663, "learning_rate": 1.2643613897162301e-05, "loss": 3.1312, "step": 162340 }, { "epoch": 0.018304, "grad_norm": 1.0658824443817139, "learning_rate": 1.2642837258951748e-05, "loss": 3.0844, "step": 162350 }, { "epoch": 0.0183296, "grad_norm": 0.7458089590072632, "learning_rate": 1.2642060603603094e-05, "loss": 2.9466, "step": 162360 }, { "epoch": 0.0183552, "grad_norm": 0.7941561937332153, "learning_rate": 1.2641283931121385e-05, "loss": 2.8658, "step": 162370 }, { "epoch": 0.0183808, "grad_norm": 0.9096815586090088, "learning_rate": 1.2640507241511646e-05, "loss": 2.7444, "step": 162380 }, { "epoch": 0.0184064, "grad_norm": 1.0411241054534912, "learning_rate": 1.2639730534778922e-05, "loss": 2.9843, "step": 162390 }, { "epoch": 0.018432, "grad_norm": 0.9924826622009277, "learning_rate": 1.2638953810928248e-05, "loss": 2.9935, "step": 162400 }, { "epoch": 0.0184576, "grad_norm": 0.6795765161514282, "learning_rate": 1.263817706996466e-05, "loss": 3.0894, "step": 162410 }, { "epoch": 0.0184832, "grad_norm": 0.8427343368530273, "learning_rate": 1.2637400311893191e-05, "loss": 2.9695, "step": 162420 }, { "epoch": 0.0185088, "grad_norm": 1.0614529848098755, "learning_rate": 1.2636623536718885e-05, "loss": 3.0587, "step": 162430 }, { "epoch": 0.0185344, "grad_norm": 0.7328135967254639, "learning_rate": 1.2635846744446774e-05, "loss": 2.9184, "step": 162440 }, { "epoch": 0.01856, "grad_norm": 0.8707526326179504, "learning_rate": 1.26350699350819e-05, "loss": 2.9039, "step": 162450 }, { "epoch": 0.0185856, "grad_norm": 0.8811509013175964, "learning_rate": 1.2634293108629297e-05, "loss": 3.0295, "step": 162460 }, { "epoch": 0.0186112, "grad_norm": 0.7017319202423096, "learning_rate": 1.2633516265094003e-05, "loss": 2.9243, "step": 162470 }, { "epoch": 0.0186368, "grad_norm": 0.810367226600647, "learning_rate": 1.2632739404481057e-05, "loss": 3.1669, "step": 162480 }, { "epoch": 0.0186624, "grad_norm": 0.7024931907653809, "learning_rate": 1.2631962526795494e-05, "loss": 2.9546, "step": 162490 }, { "epoch": 0.018688, "grad_norm": 0.8219224214553833, "learning_rate": 1.2631185632042352e-05, "loss": 3.0583, "step": 162500 }, { "epoch": 0.0187136, "grad_norm": 0.7279183268547058, "learning_rate": 1.2630408720226674e-05, "loss": 2.946, "step": 162510 }, { "epoch": 0.0187392, "grad_norm": 0.9423285722732544, "learning_rate": 1.2629631791353494e-05, "loss": 3.0199, "step": 162520 }, { "epoch": 0.0187648, "grad_norm": 0.6933878064155579, "learning_rate": 1.262885484542785e-05, "loss": 2.9102, "step": 162530 }, { "epoch": 0.0187904, "grad_norm": 0.75311279296875, "learning_rate": 1.2628077882454778e-05, "loss": 2.7654, "step": 162540 }, { "epoch": 0.018816, "grad_norm": 0.7470889091491699, "learning_rate": 1.2627300902439324e-05, "loss": 2.9335, "step": 162550 }, { "epoch": 0.0188416, "grad_norm": 0.998354971408844, "learning_rate": 1.2626523905386519e-05, "loss": 3.0027, "step": 162560 }, { "epoch": 0.0188672, "grad_norm": 0.8533899784088135, "learning_rate": 1.2625746891301407e-05, "loss": 2.9772, "step": 162570 }, { "epoch": 0.0188928, "grad_norm": 0.8138185143470764, "learning_rate": 1.2624969860189023e-05, "loss": 3.1032, "step": 162580 }, { "epoch": 0.0189184, "grad_norm": 0.7755266427993774, "learning_rate": 1.2624192812054408e-05, "loss": 2.9985, "step": 162590 }, { "epoch": 0.018944, "grad_norm": 0.8090701103210449, "learning_rate": 1.2623415746902597e-05, "loss": 3.0228, "step": 162600 }, { "epoch": 0.0189696, "grad_norm": 0.8501218557357788, "learning_rate": 1.2622638664738635e-05, "loss": 3.0609, "step": 162610 }, { "epoch": 0.0189952, "grad_norm": 0.926106870174408, "learning_rate": 1.262186156556756e-05, "loss": 2.9566, "step": 162620 }, { "epoch": 0.0190208, "grad_norm": 0.8115736246109009, "learning_rate": 1.2621084449394406e-05, "loss": 3.0676, "step": 162630 }, { "epoch": 0.0190464, "grad_norm": 0.7251251339912415, "learning_rate": 1.2620307316224214e-05, "loss": 3.0939, "step": 162640 }, { "epoch": 0.019072, "grad_norm": 0.8645637631416321, "learning_rate": 1.2619530166062028e-05, "loss": 2.9832, "step": 162650 }, { "epoch": 0.0190976, "grad_norm": 0.8019978404045105, "learning_rate": 1.2618752998912886e-05, "loss": 3.2352, "step": 162660 }, { "epoch": 0.0191232, "grad_norm": 0.7238855361938477, "learning_rate": 1.2617975814781827e-05, "loss": 2.9165, "step": 162670 }, { "epoch": 0.0191488, "grad_norm": 0.7848620414733887, "learning_rate": 1.261719861367389e-05, "loss": 3.0008, "step": 162680 }, { "epoch": 0.0191744, "grad_norm": 1.162273645401001, "learning_rate": 1.2616421395594115e-05, "loss": 2.9699, "step": 162690 }, { "epoch": 0.0192, "grad_norm": 1.0728120803833008, "learning_rate": 1.2615644160547542e-05, "loss": 3.1851, "step": 162700 }, { "epoch": 0.0192256, "grad_norm": 0.7828640341758728, "learning_rate": 1.2614866908539213e-05, "loss": 3.2044, "step": 162710 }, { "epoch": 0.0192512, "grad_norm": 0.8066052794456482, "learning_rate": 1.2614089639574165e-05, "loss": 3.1081, "step": 162720 }, { "epoch": 0.0192768, "grad_norm": 0.7269130349159241, "learning_rate": 1.2613312353657443e-05, "loss": 2.2447, "step": 162730 }, { "epoch": 0.0193024, "grad_norm": 0.7651394009590149, "learning_rate": 1.2612535050794086e-05, "loss": 2.8471, "step": 162740 }, { "epoch": 0.019328, "grad_norm": 0.6755054593086243, "learning_rate": 1.2611757730989131e-05, "loss": 2.8234, "step": 162750 }, { "epoch": 0.0193536, "grad_norm": 0.7572638392448425, "learning_rate": 1.2610980394247623e-05, "loss": 2.7778, "step": 162760 }, { "epoch": 0.0193792, "grad_norm": 0.8627349138259888, "learning_rate": 1.2610203040574601e-05, "loss": 2.7708, "step": 162770 }, { "epoch": 0.0194048, "grad_norm": 0.8936994075775146, "learning_rate": 1.2609425669975104e-05, "loss": 2.8181, "step": 162780 }, { "epoch": 0.0194304, "grad_norm": 0.7164081335067749, "learning_rate": 1.2608648282454174e-05, "loss": 2.6711, "step": 162790 }, { "epoch": 0.019456, "grad_norm": 0.8460490107536316, "learning_rate": 1.2607870878016856e-05, "loss": 2.9708, "step": 162800 }, { "epoch": 0.0194816, "grad_norm": 0.7127835154533386, "learning_rate": 1.2607093456668187e-05, "loss": 2.8045, "step": 162810 }, { "epoch": 0.0195072, "grad_norm": 0.8948394060134888, "learning_rate": 1.2606316018413209e-05, "loss": 3.1369, "step": 162820 }, { "epoch": 0.0195328, "grad_norm": 0.8279277086257935, "learning_rate": 1.2605538563256968e-05, "loss": 2.7738, "step": 162830 }, { "epoch": 0.0195584, "grad_norm": 0.9035027027130127, "learning_rate": 1.26047610912045e-05, "loss": 2.719, "step": 162840 }, { "epoch": 0.019584, "grad_norm": 0.963765561580658, "learning_rate": 1.2603983602260848e-05, "loss": 3.0065, "step": 162850 }, { "epoch": 0.0196096, "grad_norm": 0.7422772645950317, "learning_rate": 1.2603206096431053e-05, "loss": 3.142, "step": 162860 }, { "epoch": 0.0196352, "grad_norm": 0.668605387210846, "learning_rate": 1.2602428573720162e-05, "loss": 2.7843, "step": 162870 }, { "epoch": 0.0196608, "grad_norm": 0.7268433570861816, "learning_rate": 1.2601651034133211e-05, "loss": 2.7189, "step": 162880 }, { "epoch": 0.0196864, "grad_norm": 0.7348839044570923, "learning_rate": 1.2600873477675245e-05, "loss": 2.749, "step": 162890 }, { "epoch": 0.019712, "grad_norm": 0.7015781998634338, "learning_rate": 1.2600095904351304e-05, "loss": 2.8119, "step": 162900 }, { "epoch": 0.0197376, "grad_norm": 0.7020540237426758, "learning_rate": 1.2599318314166433e-05, "loss": 2.6709, "step": 162910 }, { "epoch": 0.0197632, "grad_norm": 0.8326773047447205, "learning_rate": 1.2598540707125675e-05, "loss": 2.956, "step": 162920 }, { "epoch": 0.0197888, "grad_norm": 0.7609766721725464, "learning_rate": 1.2597763083234069e-05, "loss": 2.8627, "step": 162930 }, { "epoch": 0.0198144, "grad_norm": 0.796384871006012, "learning_rate": 1.2596985442496663e-05, "loss": 2.7483, "step": 162940 }, { "epoch": 0.01984, "grad_norm": 0.6459380388259888, "learning_rate": 1.2596207784918492e-05, "loss": 2.8276, "step": 162950 }, { "epoch": 0.0198656, "grad_norm": 0.6863076090812683, "learning_rate": 1.2595430110504608e-05, "loss": 2.6612, "step": 162960 }, { "epoch": 0.0198912, "grad_norm": 0.8266708254814148, "learning_rate": 1.2594652419260049e-05, "loss": 2.7833, "step": 162970 }, { "epoch": 0.0199168, "grad_norm": 0.7330851554870605, "learning_rate": 1.2593874711189857e-05, "loss": 2.7104, "step": 162980 }, { "epoch": 0.0199424, "grad_norm": 0.7486229538917542, "learning_rate": 1.2593096986299074e-05, "loss": 3.0746, "step": 162990 }, { "epoch": 0.019968, "grad_norm": 0.7290875315666199, "learning_rate": 1.2592319244592752e-05, "loss": 2.7551, "step": 163000 }, { "epoch": 0.0199936, "grad_norm": 0.7850679755210876, "learning_rate": 1.2591541486075925e-05, "loss": 2.8569, "step": 163010 }, { "epoch": 0.0200192, "grad_norm": 0.742603600025177, "learning_rate": 1.2590763710753642e-05, "loss": 2.8482, "step": 163020 }, { "epoch": 0.0200448, "grad_norm": 0.7276847958564758, "learning_rate": 1.2589985918630942e-05, "loss": 2.7621, "step": 163030 }, { "epoch": 0.0200704, "grad_norm": 0.7687313556671143, "learning_rate": 1.2589208109712876e-05, "loss": 2.8958, "step": 163040 }, { "epoch": 0.020096, "grad_norm": 0.6996604800224304, "learning_rate": 1.258843028400448e-05, "loss": 2.9591, "step": 163050 }, { "epoch": 0.0201216, "grad_norm": 0.7181794047355652, "learning_rate": 1.2587652441510803e-05, "loss": 2.9869, "step": 163060 }, { "epoch": 0.0201472, "grad_norm": 0.7359869480133057, "learning_rate": 1.2586874582236887e-05, "loss": 2.927, "step": 163070 }, { "epoch": 0.0201728, "grad_norm": 0.746745228767395, "learning_rate": 1.2586096706187779e-05, "loss": 2.5343, "step": 163080 }, { "epoch": 0.0201984, "grad_norm": 0.7846253514289856, "learning_rate": 1.2585318813368516e-05, "loss": 3.0083, "step": 163090 }, { "epoch": 0.020224, "grad_norm": 0.708909273147583, "learning_rate": 1.2584540903784155e-05, "loss": 2.8234, "step": 163100 }, { "epoch": 0.0202496, "grad_norm": 0.7158833146095276, "learning_rate": 1.2583762977439728e-05, "loss": 2.777, "step": 163110 }, { "epoch": 0.0202752, "grad_norm": 0.7648003697395325, "learning_rate": 1.2582985034340285e-05, "loss": 2.7427, "step": 163120 }, { "epoch": 0.0203008, "grad_norm": 0.7248725295066833, "learning_rate": 1.2582207074490871e-05, "loss": 2.828, "step": 163130 }, { "epoch": 0.0203264, "grad_norm": 0.7281235456466675, "learning_rate": 1.2581429097896533e-05, "loss": 2.8884, "step": 163140 }, { "epoch": 0.020352, "grad_norm": 0.695122480392456, "learning_rate": 1.2580651104562311e-05, "loss": 2.8477, "step": 163150 }, { "epoch": 0.0203776, "grad_norm": 0.7977341413497925, "learning_rate": 1.2579873094493251e-05, "loss": 2.7011, "step": 163160 }, { "epoch": 0.0204032, "grad_norm": 0.6887531280517578, "learning_rate": 1.25790950676944e-05, "loss": 2.913, "step": 163170 }, { "epoch": 0.0204288, "grad_norm": 0.7347843050956726, "learning_rate": 1.2578317024170804e-05, "loss": 3.1768, "step": 163180 }, { "epoch": 0.0204544, "grad_norm": 0.707101047039032, "learning_rate": 1.2577538963927509e-05, "loss": 2.8893, "step": 163190 }, { "epoch": 0.02048, "grad_norm": 0.6998326778411865, "learning_rate": 1.2576760886969558e-05, "loss": 2.8046, "step": 163200 }, { "epoch": 0.0205056, "grad_norm": 0.6885679960250854, "learning_rate": 1.2575982793301997e-05, "loss": 2.8601, "step": 163210 }, { "epoch": 0.0205312, "grad_norm": 0.7522342801094055, "learning_rate": 1.2575204682929872e-05, "loss": 2.7789, "step": 163220 }, { "epoch": 0.0205568, "grad_norm": 0.7875891327857971, "learning_rate": 1.2574426555858228e-05, "loss": 2.8942, "step": 163230 }, { "epoch": 0.0205824, "grad_norm": 0.6633599400520325, "learning_rate": 1.2573648412092116e-05, "loss": 3.0252, "step": 163240 }, { "epoch": 0.020608, "grad_norm": 0.7433841228485107, "learning_rate": 1.2572870251636572e-05, "loss": 3.0373, "step": 163250 }, { "epoch": 0.0206336, "grad_norm": 0.9398143291473389, "learning_rate": 1.2572092074496654e-05, "loss": 2.7203, "step": 163260 }, { "epoch": 0.0206592, "grad_norm": 0.7511386871337891, "learning_rate": 1.2571313880677397e-05, "loss": 2.9124, "step": 163270 }, { "epoch": 0.0206848, "grad_norm": 0.6971763372421265, "learning_rate": 1.2570535670183857e-05, "loss": 2.7509, "step": 163280 }, { "epoch": 0.0207104, "grad_norm": 0.7260473966598511, "learning_rate": 1.2569757443021073e-05, "loss": 2.917, "step": 163290 }, { "epoch": 0.020736, "grad_norm": 0.7611877918243408, "learning_rate": 1.2568979199194101e-05, "loss": 2.8851, "step": 163300 }, { "epoch": 0.0207616, "grad_norm": 0.6976643204689026, "learning_rate": 1.2568200938707975e-05, "loss": 2.737, "step": 163310 }, { "epoch": 0.0207872, "grad_norm": 0.7463003396987915, "learning_rate": 1.2567422661567753e-05, "loss": 2.5919, "step": 163320 }, { "epoch": 0.0208128, "grad_norm": 0.7533705830574036, "learning_rate": 1.2566644367778473e-05, "loss": 2.5146, "step": 163330 }, { "epoch": 0.0208384, "grad_norm": 0.7613591551780701, "learning_rate": 1.2565866057345189e-05, "loss": 2.6534, "step": 163340 }, { "epoch": 0.020864, "grad_norm": 0.8456147313117981, "learning_rate": 1.2565087730272944e-05, "loss": 2.7795, "step": 163350 }, { "epoch": 0.0208896, "grad_norm": 0.8059525489807129, "learning_rate": 1.2564309386566791e-05, "loss": 2.9994, "step": 163360 }, { "epoch": 0.0209152, "grad_norm": 0.7409974932670593, "learning_rate": 1.256353102623177e-05, "loss": 2.8646, "step": 163370 }, { "epoch": 0.0209408, "grad_norm": 0.7541608214378357, "learning_rate": 1.256275264927293e-05, "loss": 2.925, "step": 163380 }, { "epoch": 0.0209664, "grad_norm": 0.7319989204406738, "learning_rate": 1.2561974255695325e-05, "loss": 2.7644, "step": 163390 }, { "epoch": 0.020992, "grad_norm": 0.8318505883216858, "learning_rate": 1.2561195845503996e-05, "loss": 2.9036, "step": 163400 }, { "epoch": 0.0210176, "grad_norm": 0.927903950214386, "learning_rate": 1.2560417418703989e-05, "loss": 2.9039, "step": 163410 }, { "epoch": 0.0210432, "grad_norm": 0.7100863456726074, "learning_rate": 1.2559638975300357e-05, "loss": 3.1003, "step": 163420 }, { "epoch": 0.0210688, "grad_norm": 0.8993958830833435, "learning_rate": 1.2558860515298148e-05, "loss": 3.0819, "step": 163430 }, { "epoch": 0.0210944, "grad_norm": 0.6824213862419128, "learning_rate": 1.2558082038702408e-05, "loss": 2.8249, "step": 163440 }, { "epoch": 0.02112, "grad_norm": 0.8560311794281006, "learning_rate": 1.2557303545518184e-05, "loss": 2.8179, "step": 163450 }, { "epoch": 0.0211456, "grad_norm": 0.6710889339447021, "learning_rate": 1.2556525035750532e-05, "loss": 2.5895, "step": 163460 }, { "epoch": 0.0211712, "grad_norm": 0.7756181955337524, "learning_rate": 1.255574650940449e-05, "loss": 2.889, "step": 163470 }, { "epoch": 0.0211968, "grad_norm": 0.794623076915741, "learning_rate": 1.2554967966485112e-05, "loss": 3.0133, "step": 163480 }, { "epoch": 0.0212224, "grad_norm": 0.7640734910964966, "learning_rate": 1.2554189406997446e-05, "loss": 3.0236, "step": 163490 }, { "epoch": 0.021248, "grad_norm": 0.8866276144981384, "learning_rate": 1.255341083094654e-05, "loss": 2.7968, "step": 163500 }, { "epoch": 0.0212736, "grad_norm": 0.9544615149497986, "learning_rate": 1.2552632238337441e-05, "loss": 2.6653, "step": 163510 }, { "epoch": 0.0212992, "grad_norm": 1.0927720069885254, "learning_rate": 1.2551853629175203e-05, "loss": 2.8038, "step": 163520 }, { "epoch": 0.0213248, "grad_norm": 0.6997183561325073, "learning_rate": 1.2551075003464874e-05, "loss": 3.1124, "step": 163530 }, { "epoch": 0.0213504, "grad_norm": 0.7735721468925476, "learning_rate": 1.25502963612115e-05, "loss": 2.7115, "step": 163540 }, { "epoch": 0.021376, "grad_norm": 1.109253168106079, "learning_rate": 1.254951770242013e-05, "loss": 2.7606, "step": 163550 }, { "epoch": 0.0214016, "grad_norm": 0.9968019723892212, "learning_rate": 1.254873902709582e-05, "loss": 2.7163, "step": 163560 }, { "epoch": 0.0214272, "grad_norm": 0.7567253708839417, "learning_rate": 1.2547960335243612e-05, "loss": 2.8499, "step": 163570 }, { "epoch": 0.0214528, "grad_norm": 0.8199518918991089, "learning_rate": 1.2547181626868556e-05, "loss": 3.0301, "step": 163580 }, { "epoch": 0.0214784, "grad_norm": 0.8130943179130554, "learning_rate": 1.2546402901975707e-05, "loss": 3.0422, "step": 163590 }, { "epoch": 0.021504, "grad_norm": 0.7483816742897034, "learning_rate": 1.2545624160570111e-05, "loss": 2.8517, "step": 163600 }, { "epoch": 0.0215296, "grad_norm": 0.7220596671104431, "learning_rate": 1.2544845402656817e-05, "loss": 2.6887, "step": 163610 }, { "epoch": 0.0215552, "grad_norm": 1.099342942237854, "learning_rate": 1.254406662824088e-05, "loss": 3.0627, "step": 163620 }, { "epoch": 0.0215808, "grad_norm": 0.8224278092384338, "learning_rate": 1.2543287837327349e-05, "loss": 2.8514, "step": 163630 }, { "epoch": 0.0216064, "grad_norm": 0.6961601376533508, "learning_rate": 1.2542509029921268e-05, "loss": 2.6793, "step": 163640 }, { "epoch": 0.021632, "grad_norm": 0.638500988483429, "learning_rate": 1.2541730206027694e-05, "loss": 2.7706, "step": 163650 }, { "epoch": 0.0216576, "grad_norm": 0.97780841588974, "learning_rate": 1.2540951365651675e-05, "loss": 2.8785, "step": 163660 }, { "epoch": 0.0216832, "grad_norm": 0.7367458343505859, "learning_rate": 1.254017250879826e-05, "loss": 2.869, "step": 163670 }, { "epoch": 0.0217088, "grad_norm": 0.8583669066429138, "learning_rate": 1.2539393635472501e-05, "loss": 2.9451, "step": 163680 }, { "epoch": 0.0217344, "grad_norm": 0.7673109769821167, "learning_rate": 1.2538614745679452e-05, "loss": 2.8405, "step": 163690 }, { "epoch": 0.02176, "grad_norm": 0.7204912304878235, "learning_rate": 1.2537835839424157e-05, "loss": 2.6594, "step": 163700 }, { "epoch": 0.0217856, "grad_norm": 0.7123575210571289, "learning_rate": 1.2537056916711673e-05, "loss": 3.0241, "step": 163710 }, { "epoch": 0.0218112, "grad_norm": 0.8312537670135498, "learning_rate": 1.2536277977547051e-05, "loss": 2.9144, "step": 163720 }, { "epoch": 0.0218368, "grad_norm": 0.7241026759147644, "learning_rate": 1.253549902193534e-05, "loss": 2.7189, "step": 163730 }, { "epoch": 0.0218624, "grad_norm": 0.7071126103401184, "learning_rate": 1.2534720049881593e-05, "loss": 2.738, "step": 163740 }, { "epoch": 0.021888, "grad_norm": 0.9543094635009766, "learning_rate": 1.2533941061390855e-05, "loss": 3.0444, "step": 163750 }, { "epoch": 0.0219136, "grad_norm": 0.7108399271965027, "learning_rate": 1.2533162056468188e-05, "loss": 3.0372, "step": 163760 }, { "epoch": 0.0219392, "grad_norm": 0.7252488732337952, "learning_rate": 1.2532383035118634e-05, "loss": 3.0331, "step": 163770 }, { "epoch": 0.0219648, "grad_norm": 0.7828452587127686, "learning_rate": 1.253160399734725e-05, "loss": 3.1027, "step": 163780 }, { "epoch": 0.0219904, "grad_norm": 0.7162184715270996, "learning_rate": 1.253082494315909e-05, "loss": 3.0453, "step": 163790 }, { "epoch": 0.022016, "grad_norm": 0.7209524512290955, "learning_rate": 1.25300458725592e-05, "loss": 3.0044, "step": 163800 }, { "epoch": 0.0220416, "grad_norm": 0.813176155090332, "learning_rate": 1.2529266785552635e-05, "loss": 2.8427, "step": 163810 }, { "epoch": 0.0220672, "grad_norm": 0.7479166388511658, "learning_rate": 1.2528487682144448e-05, "loss": 2.8657, "step": 163820 }, { "epoch": 0.0220928, "grad_norm": 0.7236971259117126, "learning_rate": 1.252770856233969e-05, "loss": 3.0599, "step": 163830 }, { "epoch": 0.0221184, "grad_norm": 0.758402943611145, "learning_rate": 1.2526929426143415e-05, "loss": 2.9412, "step": 163840 }, { "epoch": 0.022144, "grad_norm": 1.5653116703033447, "learning_rate": 1.2526150273560673e-05, "loss": 2.9185, "step": 163850 }, { "epoch": 0.0221696, "grad_norm": 0.7046721577644348, "learning_rate": 1.2525371104596515e-05, "loss": 2.9882, "step": 163860 }, { "epoch": 0.0221952, "grad_norm": 0.7036281824111938, "learning_rate": 1.2524591919256e-05, "loss": 2.7977, "step": 163870 }, { "epoch": 0.0222208, "grad_norm": 0.6562994122505188, "learning_rate": 1.2523812717544178e-05, "loss": 2.8495, "step": 163880 }, { "epoch": 0.0222464, "grad_norm": 0.6788927912712097, "learning_rate": 1.25230334994661e-05, "loss": 2.851, "step": 163890 }, { "epoch": 0.022272, "grad_norm": 0.6925287842750549, "learning_rate": 1.252225426502682e-05, "loss": 2.9006, "step": 163900 }, { "epoch": 0.0222976, "grad_norm": 0.8564067482948303, "learning_rate": 1.2521475014231392e-05, "loss": 3.041, "step": 163910 }, { "epoch": 0.0223232, "grad_norm": 0.7239643335342407, "learning_rate": 1.2520695747084867e-05, "loss": 3.064, "step": 163920 }, { "epoch": 0.0223488, "grad_norm": 0.8117056488990784, "learning_rate": 1.2519916463592305e-05, "loss": 2.9326, "step": 163930 }, { "epoch": 0.0223744, "grad_norm": 0.8211484551429749, "learning_rate": 1.2519137163758748e-05, "loss": 2.7763, "step": 163940 }, { "epoch": 0.0224, "grad_norm": 0.7084312438964844, "learning_rate": 1.2518357847589257e-05, "loss": 3.1244, "step": 163950 }, { "epoch": 0.0224256, "grad_norm": 0.746609091758728, "learning_rate": 1.2517578515088889e-05, "loss": 3.1441, "step": 163960 }, { "epoch": 0.0224512, "grad_norm": 0.7084263563156128, "learning_rate": 1.2516799166262692e-05, "loss": 2.7539, "step": 163970 }, { "epoch": 0.0224768, "grad_norm": 0.7626562714576721, "learning_rate": 1.251601980111572e-05, "loss": 3.0293, "step": 163980 }, { "epoch": 0.0225024, "grad_norm": 0.7852655053138733, "learning_rate": 1.251524041965303e-05, "loss": 3.1524, "step": 163990 }, { "epoch": 0.022528, "grad_norm": 0.8048540949821472, "learning_rate": 1.2514461021879671e-05, "loss": 2.8254, "step": 164000 }, { "epoch": 0.0225536, "grad_norm": 0.7460691928863525, "learning_rate": 1.2513681607800704e-05, "loss": 2.7265, "step": 164010 }, { "epoch": 0.0225792, "grad_norm": 1.4142998456954956, "learning_rate": 1.2512902177421178e-05, "loss": 2.7561, "step": 164020 }, { "epoch": 0.0226048, "grad_norm": 0.6850709915161133, "learning_rate": 1.2512122730746147e-05, "loss": 2.958, "step": 164030 }, { "epoch": 0.0226304, "grad_norm": 0.7392207384109497, "learning_rate": 1.2511343267780672e-05, "loss": 2.8136, "step": 164040 }, { "epoch": 0.022656, "grad_norm": 0.6919724941253662, "learning_rate": 1.2510563788529802e-05, "loss": 3.1193, "step": 164050 }, { "epoch": 0.0226816, "grad_norm": 0.9290418028831482, "learning_rate": 1.2509784292998593e-05, "loss": 3.0138, "step": 164060 }, { "epoch": 0.0227072, "grad_norm": 0.6842784881591797, "learning_rate": 1.2509004781192098e-05, "loss": 2.9692, "step": 164070 }, { "epoch": 0.0227328, "grad_norm": 0.8062970042228699, "learning_rate": 1.2508225253115376e-05, "loss": 2.7731, "step": 164080 }, { "epoch": 0.0227584, "grad_norm": 0.8368182182312012, "learning_rate": 1.250744570877348e-05, "loss": 2.8221, "step": 164090 }, { "epoch": 0.022784, "grad_norm": 0.7173618674278259, "learning_rate": 1.2506666148171465e-05, "loss": 2.7829, "step": 164100 }, { "epoch": 0.0228096, "grad_norm": 0.9017552137374878, "learning_rate": 1.2505886571314384e-05, "loss": 2.9731, "step": 164110 }, { "epoch": 0.0228352, "grad_norm": 0.8578366637229919, "learning_rate": 1.2505106978207295e-05, "loss": 3.0239, "step": 164120 }, { "epoch": 0.0228608, "grad_norm": 0.8553779721260071, "learning_rate": 1.2504327368855254e-05, "loss": 3.0089, "step": 164130 }, { "epoch": 0.0228864, "grad_norm": 0.8025388121604919, "learning_rate": 1.2503547743263315e-05, "loss": 3.1681, "step": 164140 }, { "epoch": 0.022912, "grad_norm": 0.8896942734718323, "learning_rate": 1.2502768101436538e-05, "loss": 2.9452, "step": 164150 }, { "epoch": 0.0229376, "grad_norm": 0.8317633271217346, "learning_rate": 1.2501988443379971e-05, "loss": 3.0252, "step": 164160 }, { "epoch": 0.0229632, "grad_norm": 1.168853521347046, "learning_rate": 1.2501208769098674e-05, "loss": 2.8873, "step": 164170 }, { "epoch": 0.0229888, "grad_norm": 0.7164713144302368, "learning_rate": 1.2500429078597702e-05, "loss": 3.001, "step": 164180 }, { "epoch": 0.0230144, "grad_norm": 0.7749810814857483, "learning_rate": 1.2499649371882116e-05, "loss": 2.7996, "step": 164190 }, { "epoch": 0.02304, "grad_norm": 0.845707893371582, "learning_rate": 1.2498869648956964e-05, "loss": 3.1202, "step": 164200 }, { "epoch": 0.0230656, "grad_norm": 0.793245255947113, "learning_rate": 1.2498089909827306e-05, "loss": 2.8132, "step": 164210 }, { "epoch": 0.0230912, "grad_norm": 0.7347394824028015, "learning_rate": 1.2497310154498204e-05, "loss": 2.9551, "step": 164220 }, { "epoch": 0.0231168, "grad_norm": 0.7434432506561279, "learning_rate": 1.2496530382974705e-05, "loss": 3.031, "step": 164230 }, { "epoch": 0.0231424, "grad_norm": 0.6708064079284668, "learning_rate": 1.2495750595261872e-05, "loss": 2.9756, "step": 164240 }, { "epoch": 0.023168, "grad_norm": 0.7563039660453796, "learning_rate": 1.249497079136476e-05, "loss": 2.8721, "step": 164250 }, { "epoch": 0.0231936, "grad_norm": 0.8564332127571106, "learning_rate": 1.2494190971288424e-05, "loss": 2.8509, "step": 164260 }, { "epoch": 0.0232192, "grad_norm": 0.7753869891166687, "learning_rate": 1.2493411135037922e-05, "loss": 2.6697, "step": 164270 }, { "epoch": 0.0232448, "grad_norm": 1.066203236579895, "learning_rate": 1.2492631282618312e-05, "loss": 2.9508, "step": 164280 }, { "epoch": 0.0232704, "grad_norm": 0.8718221783638, "learning_rate": 1.2491851414034652e-05, "loss": 2.9718, "step": 164290 }, { "epoch": 0.023296, "grad_norm": 0.751764178276062, "learning_rate": 1.2491071529291995e-05, "loss": 2.9926, "step": 164300 }, { "epoch": 0.0233216, "grad_norm": 0.9284178614616394, "learning_rate": 1.2490291628395404e-05, "loss": 3.169, "step": 164310 }, { "epoch": 0.0233472, "grad_norm": 0.6895351409912109, "learning_rate": 1.2489511711349932e-05, "loss": 2.9078, "step": 164320 }, { "epoch": 0.0233728, "grad_norm": 0.687593936920166, "learning_rate": 1.2488731778160642e-05, "loss": 3.0437, "step": 164330 }, { "epoch": 0.0233984, "grad_norm": 0.742205023765564, "learning_rate": 1.2487951828832583e-05, "loss": 2.9383, "step": 164340 }, { "epoch": 0.023424, "grad_norm": 0.7545583248138428, "learning_rate": 1.2487171863370819e-05, "loss": 2.8878, "step": 164350 }, { "epoch": 0.0234496, "grad_norm": 0.9321815371513367, "learning_rate": 1.2486391881780407e-05, "loss": 2.8129, "step": 164360 }, { "epoch": 0.0234752, "grad_norm": 0.7405126690864563, "learning_rate": 1.2485611884066405e-05, "loss": 3.0628, "step": 164370 }, { "epoch": 0.0235008, "grad_norm": 0.7709720134735107, "learning_rate": 1.2484831870233868e-05, "loss": 2.9379, "step": 164380 }, { "epoch": 0.0235264, "grad_norm": 0.8002142310142517, "learning_rate": 1.2484051840287858e-05, "loss": 2.9388, "step": 164390 }, { "epoch": 0.023552, "grad_norm": 1.0918101072311401, "learning_rate": 1.2483271794233433e-05, "loss": 3.0198, "step": 164400 }, { "epoch": 0.0235776, "grad_norm": 0.7278318405151367, "learning_rate": 1.248249173207565e-05, "loss": 2.8861, "step": 164410 }, { "epoch": 0.0236032, "grad_norm": 0.8457884788513184, "learning_rate": 1.2481711653819568e-05, "loss": 3.0797, "step": 164420 }, { "epoch": 0.0236288, "grad_norm": 0.710144579410553, "learning_rate": 1.2480931559470246e-05, "loss": 2.9262, "step": 164430 }, { "epoch": 0.0236544, "grad_norm": 0.7520105242729187, "learning_rate": 1.248015144903274e-05, "loss": 2.88, "step": 164440 }, { "epoch": 0.02368, "grad_norm": 0.7665188908576965, "learning_rate": 1.2479371322512115e-05, "loss": 3.0273, "step": 164450 }, { "epoch": 0.0237056, "grad_norm": 0.6675719618797302, "learning_rate": 1.247859117991342e-05, "loss": 2.9355, "step": 164460 }, { "epoch": 0.0237312, "grad_norm": 0.7226279377937317, "learning_rate": 1.2477811021241724e-05, "loss": 2.8013, "step": 164470 }, { "epoch": 0.0237568, "grad_norm": 1.047497272491455, "learning_rate": 1.2477030846502083e-05, "loss": 2.962, "step": 164480 }, { "epoch": 0.0237824, "grad_norm": 0.7397101521492004, "learning_rate": 1.247625065569955e-05, "loss": 3.0125, "step": 164490 }, { "epoch": 0.023808, "grad_norm": 0.7714818716049194, "learning_rate": 1.2475470448839195e-05, "loss": 2.7916, "step": 164500 }, { "epoch": 0.0238336, "grad_norm": 0.7364087700843811, "learning_rate": 1.247469022592607e-05, "loss": 2.9747, "step": 164510 }, { "epoch": 0.0238592, "grad_norm": 2.066758155822754, "learning_rate": 1.2473909986965239e-05, "loss": 2.9098, "step": 164520 }, { "epoch": 0.0238848, "grad_norm": 0.8319358825683594, "learning_rate": 1.2473129731961755e-05, "loss": 3.0622, "step": 164530 }, { "epoch": 0.0239104, "grad_norm": 0.7004555463790894, "learning_rate": 1.2472349460920684e-05, "loss": 3.1131, "step": 164540 }, { "epoch": 0.023936, "grad_norm": 0.8143203854560852, "learning_rate": 1.2471569173847083e-05, "loss": 2.9207, "step": 164550 }, { "epoch": 0.0239616, "grad_norm": 0.8367255926132202, "learning_rate": 1.2470788870746014e-05, "loss": 2.9457, "step": 164560 }, { "epoch": 0.0239872, "grad_norm": 0.8238577842712402, "learning_rate": 1.2470008551622534e-05, "loss": 2.9706, "step": 164570 }, { "epoch": 0.0240128, "grad_norm": 0.9479941725730896, "learning_rate": 1.2469228216481708e-05, "loss": 2.9095, "step": 164580 }, { "epoch": 0.0240384, "grad_norm": 0.8874319195747375, "learning_rate": 1.2468447865328592e-05, "loss": 3.0367, "step": 164590 }, { "epoch": 0.024064, "grad_norm": 0.7040230631828308, "learning_rate": 1.2467667498168248e-05, "loss": 2.9541, "step": 164600 }, { "epoch": 0.0240896, "grad_norm": 0.7034757733345032, "learning_rate": 1.2466887115005734e-05, "loss": 2.8743, "step": 164610 }, { "epoch": 0.0241152, "grad_norm": 0.8424984812736511, "learning_rate": 1.2466106715846116e-05, "loss": 3.0352, "step": 164620 }, { "epoch": 0.0241408, "grad_norm": 0.7448604702949524, "learning_rate": 1.2465326300694448e-05, "loss": 2.9058, "step": 164630 }, { "epoch": 0.0241664, "grad_norm": 0.7435645461082458, "learning_rate": 1.2464545869555793e-05, "loss": 3.0531, "step": 164640 }, { "epoch": 0.024192, "grad_norm": 0.7380478382110596, "learning_rate": 1.2463765422435218e-05, "loss": 2.8935, "step": 164650 }, { "epoch": 0.0242176, "grad_norm": 0.8327507972717285, "learning_rate": 1.2462984959337777e-05, "loss": 2.7538, "step": 164660 }, { "epoch": 0.0242432, "grad_norm": 0.8425449132919312, "learning_rate": 1.2462204480268532e-05, "loss": 2.939, "step": 164670 }, { "epoch": 0.0242688, "grad_norm": 0.7126723527908325, "learning_rate": 1.2461423985232548e-05, "loss": 2.8839, "step": 164680 }, { "epoch": 0.0242944, "grad_norm": 2.1186187267303467, "learning_rate": 1.2460643474234879e-05, "loss": 3.1105, "step": 164690 }, { "epoch": 0.02432, "grad_norm": 0.8325498700141907, "learning_rate": 1.2459862947280594e-05, "loss": 2.8928, "step": 164700 }, { "epoch": 0.0243456, "grad_norm": 0.7459079027175903, "learning_rate": 1.2459082404374752e-05, "loss": 3.0078, "step": 164710 }, { "epoch": 0.0243712, "grad_norm": 0.7967675924301147, "learning_rate": 1.2458301845522413e-05, "loss": 3.0247, "step": 164720 }, { "epoch": 0.0243968, "grad_norm": 0.7550130486488342, "learning_rate": 1.2457521270728638e-05, "loss": 2.9469, "step": 164730 }, { "epoch": 0.0244224, "grad_norm": 0.708958625793457, "learning_rate": 1.2456740679998494e-05, "loss": 3.0124, "step": 164740 }, { "epoch": 0.024448, "grad_norm": 0.6872243285179138, "learning_rate": 1.2455960073337039e-05, "loss": 2.8396, "step": 164750 }, { "epoch": 0.0244736, "grad_norm": 0.9559994339942932, "learning_rate": 1.2455179450749335e-05, "loss": 2.9632, "step": 164760 }, { "epoch": 0.0244992, "grad_norm": 0.8522434830665588, "learning_rate": 1.2454398812240444e-05, "loss": 2.9952, "step": 164770 }, { "epoch": 0.0245248, "grad_norm": 0.7202666401863098, "learning_rate": 1.2453618157815433e-05, "loss": 2.8071, "step": 164780 }, { "epoch": 0.0245504, "grad_norm": 0.8699986934661865, "learning_rate": 1.2452837487479357e-05, "loss": 3.0325, "step": 164790 }, { "epoch": 0.024576, "grad_norm": 0.8043888807296753, "learning_rate": 1.2452056801237283e-05, "loss": 2.9709, "step": 164800 }, { "epoch": 0.0246016, "grad_norm": 0.7113257050514221, "learning_rate": 1.245127609909427e-05, "loss": 2.9151, "step": 164810 }, { "epoch": 0.0246272, "grad_norm": 0.7220120429992676, "learning_rate": 1.2450495381055383e-05, "loss": 2.9387, "step": 164820 }, { "epoch": 0.0246528, "grad_norm": 0.6984817981719971, "learning_rate": 1.2449714647125686e-05, "loss": 3.1002, "step": 164830 }, { "epoch": 0.0246784, "grad_norm": 0.7207285165786743, "learning_rate": 1.244893389731024e-05, "loss": 2.8144, "step": 164840 }, { "epoch": 0.024704, "grad_norm": 0.8291991353034973, "learning_rate": 1.244815313161411e-05, "loss": 2.8228, "step": 164850 }, { "epoch": 0.0247296, "grad_norm": 0.8760736584663391, "learning_rate": 1.2447372350042358e-05, "loss": 2.8284, "step": 164860 }, { "epoch": 0.0247552, "grad_norm": 0.9484257102012634, "learning_rate": 1.2446591552600045e-05, "loss": 3.016, "step": 164870 }, { "epoch": 0.0247808, "grad_norm": 0.755029559135437, "learning_rate": 1.244581073929224e-05, "loss": 2.9802, "step": 164880 }, { "epoch": 0.0248064, "grad_norm": 0.716140627861023, "learning_rate": 1.2445029910123996e-05, "loss": 2.8851, "step": 164890 }, { "epoch": 0.024832, "grad_norm": 0.8642842173576355, "learning_rate": 1.2444249065100385e-05, "loss": 2.8438, "step": 164900 }, { "epoch": 0.0248576, "grad_norm": 0.95639967918396, "learning_rate": 1.244346820422647e-05, "loss": 3.0362, "step": 164910 }, { "epoch": 0.0248832, "grad_norm": 0.6505314111709595, "learning_rate": 1.2442687327507312e-05, "loss": 2.8899, "step": 164920 }, { "epoch": 0.0249088, "grad_norm": 1.0484639406204224, "learning_rate": 1.2441906434947974e-05, "loss": 2.7256, "step": 164930 }, { "epoch": 0.0249344, "grad_norm": 0.8747625946998596, "learning_rate": 1.2441125526553526e-05, "loss": 2.9276, "step": 164940 }, { "epoch": 0.02496, "grad_norm": 0.7763429880142212, "learning_rate": 1.2440344602329026e-05, "loss": 3.0567, "step": 164950 }, { "epoch": 0.0249856, "grad_norm": 0.6855756640434265, "learning_rate": 1.243956366227954e-05, "loss": 2.8481, "step": 164960 }, { "epoch": 0.0250112, "grad_norm": 0.740422248840332, "learning_rate": 1.243878270641013e-05, "loss": 2.8363, "step": 164970 }, { "epoch": 0.0250368, "grad_norm": 0.6904656291007996, "learning_rate": 1.2438001734725866e-05, "loss": 2.8183, "step": 164980 }, { "epoch": 0.0250624, "grad_norm": 0.7536830902099609, "learning_rate": 1.2437220747231803e-05, "loss": 2.7287, "step": 164990 }, { "epoch": 0.025088, "grad_norm": 0.7287561893463135, "learning_rate": 1.2436439743933015e-05, "loss": 2.8121, "step": 165000 }, { "epoch": 0.0251136, "grad_norm": 0.7074523568153381, "learning_rate": 1.2435658724834562e-05, "loss": 2.9273, "step": 165010 }, { "epoch": 0.0251392, "grad_norm": 0.7532923817634583, "learning_rate": 1.2434877689941511e-05, "loss": 2.807, "step": 165020 }, { "epoch": 0.0251648, "grad_norm": 0.7171104550361633, "learning_rate": 1.2434096639258924e-05, "loss": 2.883, "step": 165030 }, { "epoch": 0.0251904, "grad_norm": 0.7215018272399902, "learning_rate": 1.243331557279187e-05, "loss": 2.9197, "step": 165040 }, { "epoch": 0.025216, "grad_norm": 0.888873815536499, "learning_rate": 1.2432534490545407e-05, "loss": 2.9869, "step": 165050 }, { "epoch": 0.0252416, "grad_norm": 0.9270246028900146, "learning_rate": 1.2431753392524605e-05, "loss": 2.9445, "step": 165060 }, { "epoch": 0.0252672, "grad_norm": 0.806422770023346, "learning_rate": 1.243097227873453e-05, "loss": 2.8967, "step": 165070 }, { "epoch": 0.0252928, "grad_norm": 0.7489637136459351, "learning_rate": 1.2430191149180244e-05, "loss": 2.8523, "step": 165080 }, { "epoch": 0.0253184, "grad_norm": 0.7565411329269409, "learning_rate": 1.2429410003866814e-05, "loss": 2.8916, "step": 165090 }, { "epoch": 0.025344, "grad_norm": 0.9152825474739075, "learning_rate": 1.2428628842799309e-05, "loss": 2.9068, "step": 165100 }, { "epoch": 0.0253696, "grad_norm": 0.7999425530433655, "learning_rate": 1.242784766598279e-05, "loss": 2.8467, "step": 165110 }, { "epoch": 0.0253952, "grad_norm": 0.707503616809845, "learning_rate": 1.2427066473422325e-05, "loss": 2.86, "step": 165120 }, { "epoch": 0.0254208, "grad_norm": 0.6818287372589111, "learning_rate": 1.2426285265122977e-05, "loss": 2.748, "step": 165130 }, { "epoch": 0.0254464, "grad_norm": 0.8229036927223206, "learning_rate": 1.2425504041089818e-05, "loss": 3.0479, "step": 165140 }, { "epoch": 0.025472, "grad_norm": 0.8741743564605713, "learning_rate": 1.2424722801327905e-05, "loss": 3.1475, "step": 165150 }, { "epoch": 0.0254976, "grad_norm": 1.3036706447601318, "learning_rate": 1.2423941545842309e-05, "loss": 2.8093, "step": 165160 }, { "epoch": 0.0255232, "grad_norm": 0.7402436137199402, "learning_rate": 1.2423160274638097e-05, "loss": 3.1069, "step": 165170 }, { "epoch": 0.0255488, "grad_norm": 0.7299261093139648, "learning_rate": 1.2422378987720335e-05, "loss": 3.2349, "step": 165180 }, { "epoch": 0.0255744, "grad_norm": 0.7131541967391968, "learning_rate": 1.2421597685094092e-05, "loss": 3.0716, "step": 165190 }, { "epoch": 0.0256, "grad_norm": 0.918722927570343, "learning_rate": 1.2420816366764428e-05, "loss": 2.8322, "step": 165200 }, { "epoch": 0.0256256, "grad_norm": 0.7119694352149963, "learning_rate": 1.2420035032736414e-05, "loss": 2.8177, "step": 165210 }, { "epoch": 0.0256512, "grad_norm": 0.7573862075805664, "learning_rate": 1.2419253683015118e-05, "loss": 2.936, "step": 165220 }, { "epoch": 0.0256768, "grad_norm": 0.8486272096633911, "learning_rate": 1.2418472317605603e-05, "loss": 3.0381, "step": 165230 }, { "epoch": 0.0257024, "grad_norm": 0.9092274904251099, "learning_rate": 1.2417690936512941e-05, "loss": 3.0592, "step": 165240 }, { "epoch": 0.025728, "grad_norm": 0.7638285756111145, "learning_rate": 1.241690953974219e-05, "loss": 2.8514, "step": 165250 }, { "epoch": 0.0257536, "grad_norm": 0.8022599816322327, "learning_rate": 1.2416128127298426e-05, "loss": 2.8544, "step": 165260 }, { "epoch": 0.0257792, "grad_norm": 0.9043650031089783, "learning_rate": 1.2415346699186714e-05, "loss": 2.9799, "step": 165270 }, { "epoch": 0.0258048, "grad_norm": 0.7394779920578003, "learning_rate": 1.2414565255412118e-05, "loss": 2.9066, "step": 165280 }, { "epoch": 0.0258304, "grad_norm": 0.7932813167572021, "learning_rate": 1.241378379597971e-05, "loss": 2.9872, "step": 165290 }, { "epoch": 0.025856, "grad_norm": 0.9243188500404358, "learning_rate": 1.2413002320894556e-05, "loss": 2.9222, "step": 165300 }, { "epoch": 0.0258816, "grad_norm": 0.7745926380157471, "learning_rate": 1.2412220830161724e-05, "loss": 2.9342, "step": 165310 }, { "epoch": 0.0259072, "grad_norm": 0.760385274887085, "learning_rate": 1.2411439323786279e-05, "loss": 2.9303, "step": 165320 }, { "epoch": 0.0259328, "grad_norm": 0.8157394528388977, "learning_rate": 1.2410657801773287e-05, "loss": 2.8771, "step": 165330 }, { "epoch": 0.0259584, "grad_norm": 0.7337501645088196, "learning_rate": 1.2409876264127826e-05, "loss": 2.7097, "step": 165340 }, { "epoch": 0.025984, "grad_norm": 0.7485678791999817, "learning_rate": 1.2409094710854957e-05, "loss": 2.9, "step": 165350 }, { "epoch": 0.0260096, "grad_norm": 0.7228420376777649, "learning_rate": 1.2408313141959747e-05, "loss": 2.8616, "step": 165360 }, { "epoch": 0.0260352, "grad_norm": 0.6592890024185181, "learning_rate": 1.240753155744727e-05, "loss": 2.7636, "step": 165370 }, { "epoch": 0.0260608, "grad_norm": 0.8554831743240356, "learning_rate": 1.2406749957322588e-05, "loss": 3.1414, "step": 165380 }, { "epoch": 0.0260864, "grad_norm": 0.7663716077804565, "learning_rate": 1.2405968341590771e-05, "loss": 2.8229, "step": 165390 }, { "epoch": 0.026112, "grad_norm": 1.019506573677063, "learning_rate": 1.2405186710256892e-05, "loss": 3.0471, "step": 165400 }, { "epoch": 0.0261376, "grad_norm": 0.7612788677215576, "learning_rate": 1.2404405063326014e-05, "loss": 2.8838, "step": 165410 }, { "epoch": 0.0261632, "grad_norm": 0.955372154712677, "learning_rate": 1.2403623400803206e-05, "loss": 2.9803, "step": 165420 }, { "epoch": 0.0261888, "grad_norm": 0.9252245426177979, "learning_rate": 1.2402841722693542e-05, "loss": 3.0832, "step": 165430 }, { "epoch": 0.0262144, "grad_norm": 0.7836432456970215, "learning_rate": 1.240206002900209e-05, "loss": 2.6558, "step": 165440 }, { "epoch": 0.02624, "grad_norm": 0.7283932566642761, "learning_rate": 1.2401278319733913e-05, "loss": 3.1148, "step": 165450 }, { "epoch": 0.0262656, "grad_norm": 0.7565057277679443, "learning_rate": 1.2400496594894087e-05, "loss": 2.8875, "step": 165460 }, { "epoch": 0.0262912, "grad_norm": 0.663768470287323, "learning_rate": 1.2399714854487678e-05, "loss": 2.928, "step": 165470 }, { "epoch": 0.0263168, "grad_norm": 1.1326186656951904, "learning_rate": 1.2398933098519757e-05, "loss": 3.0436, "step": 165480 }, { "epoch": 0.0263424, "grad_norm": 0.7111271023750305, "learning_rate": 1.239815132699539e-05, "loss": 2.7545, "step": 165490 }, { "epoch": 0.026368, "grad_norm": 0.8005551695823669, "learning_rate": 1.239736953991965e-05, "loss": 3.281, "step": 165500 }, { "epoch": 0.0263936, "grad_norm": 0.7613652348518372, "learning_rate": 1.2396587737297605e-05, "loss": 2.9036, "step": 165510 }, { "epoch": 0.0264192, "grad_norm": 0.7449299097061157, "learning_rate": 1.2395805919134324e-05, "loss": 2.8638, "step": 165520 }, { "epoch": 0.0264448, "grad_norm": 0.7307286858558655, "learning_rate": 1.2395024085434882e-05, "loss": 3.0265, "step": 165530 }, { "epoch": 0.0264704, "grad_norm": 1.0039937496185303, "learning_rate": 1.2394242236204345e-05, "loss": 2.9984, "step": 165540 }, { "epoch": 0.026496, "grad_norm": 0.7001950740814209, "learning_rate": 1.2393460371447783e-05, "loss": 2.9018, "step": 165550 }, { "epoch": 0.0265216, "grad_norm": 0.7897709012031555, "learning_rate": 1.2392678491170265e-05, "loss": 2.9419, "step": 165560 }, { "epoch": 0.0265472, "grad_norm": 1.153572678565979, "learning_rate": 1.2391896595376865e-05, "loss": 2.9742, "step": 165570 }, { "epoch": 0.0265728, "grad_norm": 0.7359186410903931, "learning_rate": 1.2391114684072648e-05, "loss": 3.1995, "step": 165580 }, { "epoch": 0.0265984, "grad_norm": 0.6942355036735535, "learning_rate": 1.2390332757262691e-05, "loss": 3.1203, "step": 165590 }, { "epoch": 0.026624, "grad_norm": 0.7250762581825256, "learning_rate": 1.2389550814952061e-05, "loss": 2.9071, "step": 165600 }, { "epoch": 0.0266496, "grad_norm": 1.3177624940872192, "learning_rate": 1.2388768857145828e-05, "loss": 2.9732, "step": 165610 }, { "epoch": 0.0266752, "grad_norm": 0.7370818853378296, "learning_rate": 1.238798688384906e-05, "loss": 2.9628, "step": 165620 }, { "epoch": 0.0267008, "grad_norm": 0.7562016844749451, "learning_rate": 1.2387204895066837e-05, "loss": 2.9887, "step": 165630 }, { "epoch": 0.0267264, "grad_norm": 0.8398001194000244, "learning_rate": 1.2386422890804226e-05, "loss": 3.1252, "step": 165640 }, { "epoch": 0.026752, "grad_norm": 2.511749267578125, "learning_rate": 1.2385640871066294e-05, "loss": 2.9806, "step": 165650 }, { "epoch": 0.0267776, "grad_norm": 0.9174619317054749, "learning_rate": 1.2384858835858113e-05, "loss": 3.0422, "step": 165660 }, { "epoch": 0.0268032, "grad_norm": 0.855276346206665, "learning_rate": 1.2384076785184761e-05, "loss": 2.8194, "step": 165670 }, { "epoch": 0.0268288, "grad_norm": 0.7450501322746277, "learning_rate": 1.23832947190513e-05, "loss": 3.0133, "step": 165680 }, { "epoch": 0.0268544, "grad_norm": 0.7151066064834595, "learning_rate": 1.2382512637462808e-05, "loss": 2.9685, "step": 165690 }, { "epoch": 0.02688, "grad_norm": 1.0886566638946533, "learning_rate": 1.2381730540424357e-05, "loss": 3.0106, "step": 165700 }, { "epoch": 0.0269056, "grad_norm": 0.7901213765144348, "learning_rate": 1.2380948427941013e-05, "loss": 3.0324, "step": 165710 }, { "epoch": 0.0269312, "grad_norm": 0.7215688824653625, "learning_rate": 1.2380166300017853e-05, "loss": 3.1188, "step": 165720 }, { "epoch": 0.0269568, "grad_norm": 0.8004478216171265, "learning_rate": 1.2379384156659946e-05, "loss": 3.1561, "step": 165730 }, { "epoch": 0.0269824, "grad_norm": 0.7107883095741272, "learning_rate": 1.2378601997872368e-05, "loss": 3.025, "step": 165740 }, { "epoch": 0.027008, "grad_norm": 0.7879767417907715, "learning_rate": 1.2377819823660185e-05, "loss": 3.1864, "step": 165750 }, { "epoch": 0.0270336, "grad_norm": 0.6919664144515991, "learning_rate": 1.2377037634028472e-05, "loss": 3.0059, "step": 165760 }, { "epoch": 0.0270592, "grad_norm": 0.6773558259010315, "learning_rate": 1.2376255428982305e-05, "loss": 2.9999, "step": 165770 }, { "epoch": 0.0270848, "grad_norm": 0.8571470975875854, "learning_rate": 1.237547320852675e-05, "loss": 2.882, "step": 165780 }, { "epoch": 0.0271104, "grad_norm": 0.9815887808799744, "learning_rate": 1.2374690972666883e-05, "loss": 2.9005, "step": 165790 }, { "epoch": 0.027136, "grad_norm": 0.7072755098342896, "learning_rate": 1.2373908721407779e-05, "loss": 3.0443, "step": 165800 }, { "epoch": 0.0271616, "grad_norm": 0.7253057360649109, "learning_rate": 1.2373126454754505e-05, "loss": 2.9424, "step": 165810 }, { "epoch": 0.0271872, "grad_norm": 0.8221508860588074, "learning_rate": 1.2372344172712136e-05, "loss": 2.8756, "step": 165820 }, { "epoch": 0.0272128, "grad_norm": 0.7142670750617981, "learning_rate": 1.2371561875285747e-05, "loss": 3.0779, "step": 165830 }, { "epoch": 0.0272384, "grad_norm": 0.8789436221122742, "learning_rate": 1.2370779562480412e-05, "loss": 3.2068, "step": 165840 }, { "epoch": 0.027264, "grad_norm": 0.8046290278434753, "learning_rate": 1.2369997234301199e-05, "loss": 3.095, "step": 165850 }, { "epoch": 0.0272896, "grad_norm": 0.7633087635040283, "learning_rate": 1.2369214890753183e-05, "loss": 3.157, "step": 165860 }, { "epoch": 0.0273152, "grad_norm": 1.7602157592773438, "learning_rate": 1.236843253184144e-05, "loss": 2.9978, "step": 165870 }, { "epoch": 0.0273408, "grad_norm": 1.1426671743392944, "learning_rate": 1.2367650157571039e-05, "loss": 3.0987, "step": 165880 }, { "epoch": 0.0273664, "grad_norm": 0.9355602264404297, "learning_rate": 1.2366867767947058e-05, "loss": 3.0297, "step": 165890 }, { "epoch": 0.027392, "grad_norm": 0.910973310470581, "learning_rate": 1.2366085362974569e-05, "loss": 3.068, "step": 165900 }, { "epoch": 0.0274176, "grad_norm": 0.7641189098358154, "learning_rate": 1.2365302942658644e-05, "loss": 3.005, "step": 165910 }, { "epoch": 0.0274432, "grad_norm": 0.7398766875267029, "learning_rate": 1.2364520507004359e-05, "loss": 3.1654, "step": 165920 }, { "epoch": 0.0274688, "grad_norm": 0.7621681690216064, "learning_rate": 1.2363738056016785e-05, "loss": 3.0717, "step": 165930 }, { "epoch": 0.0274944, "grad_norm": 0.8259392380714417, "learning_rate": 1.2362955589700997e-05, "loss": 3.1948, "step": 165940 }, { "epoch": 0.02752, "grad_norm": 0.9792870879173279, "learning_rate": 1.2362173108062073e-05, "loss": 2.7622, "step": 165950 }, { "epoch": 0.0275456, "grad_norm": 0.798247754573822, "learning_rate": 1.2361390611105084e-05, "loss": 3.0095, "step": 165960 }, { "epoch": 0.0275712, "grad_norm": 0.8057528138160706, "learning_rate": 1.2360608098835104e-05, "loss": 2.8597, "step": 165970 }, { "epoch": 0.0275968, "grad_norm": 0.7408746480941772, "learning_rate": 1.2359825571257207e-05, "loss": 3.0174, "step": 165980 }, { "epoch": 0.0276224, "grad_norm": 0.7787286639213562, "learning_rate": 1.235904302837647e-05, "loss": 2.9356, "step": 165990 }, { "epoch": 0.027648, "grad_norm": 0.7623046040534973, "learning_rate": 1.2358260470197968e-05, "loss": 2.7449, "step": 166000 }, { "epoch": 0.0276736, "grad_norm": 0.6460602879524231, "learning_rate": 1.2357477896726768e-05, "loss": 2.9461, "step": 166010 }, { "epoch": 0.0276992, "grad_norm": 0.796856164932251, "learning_rate": 1.2356695307967952e-05, "loss": 3.1785, "step": 166020 }, { "epoch": 0.0277248, "grad_norm": 0.7031287550926208, "learning_rate": 1.2355912703926593e-05, "loss": 2.8529, "step": 166030 }, { "epoch": 0.0277504, "grad_norm": 0.818893313407898, "learning_rate": 1.235513008460777e-05, "loss": 2.9579, "step": 166040 }, { "epoch": 0.027776, "grad_norm": 0.6793913841247559, "learning_rate": 1.235434745001655e-05, "loss": 3.3307, "step": 166050 }, { "epoch": 0.0278016, "grad_norm": 0.7770230770111084, "learning_rate": 1.2353564800158015e-05, "loss": 3.0256, "step": 166060 }, { "epoch": 0.0278272, "grad_norm": 0.7777014970779419, "learning_rate": 1.2352782135037237e-05, "loss": 2.9078, "step": 166070 }, { "epoch": 0.0278528, "grad_norm": 0.8442689180374146, "learning_rate": 1.2351999454659292e-05, "loss": 3.0118, "step": 166080 }, { "epoch": 0.0278784, "grad_norm": 0.8978269696235657, "learning_rate": 1.2351216759029256e-05, "loss": 3.3035, "step": 166090 }, { "epoch": 0.027904, "grad_norm": 0.804988443851471, "learning_rate": 1.2350434048152204e-05, "loss": 2.9255, "step": 166100 }, { "epoch": 0.0279296, "grad_norm": 0.8239240050315857, "learning_rate": 1.2349651322033211e-05, "loss": 2.9032, "step": 166110 }, { "epoch": 0.0279552, "grad_norm": 0.7324630618095398, "learning_rate": 1.2348868580677354e-05, "loss": 2.8864, "step": 166120 }, { "epoch": 0.0279808, "grad_norm": 0.8179672956466675, "learning_rate": 1.234808582408971e-05, "loss": 2.8683, "step": 166130 }, { "epoch": 0.0280064, "grad_norm": 1.1679621934890747, "learning_rate": 1.2347303052275353e-05, "loss": 2.833, "step": 166140 }, { "epoch": 0.028032, "grad_norm": 0.8242415189743042, "learning_rate": 1.2346520265239358e-05, "loss": 3.0269, "step": 166150 }, { "epoch": 0.0280576, "grad_norm": 0.713559091091156, "learning_rate": 1.2345737462986804e-05, "loss": 2.933, "step": 166160 }, { "epoch": 0.0280832, "grad_norm": 0.9499161243438721, "learning_rate": 1.2344954645522766e-05, "loss": 3.0718, "step": 166170 }, { "epoch": 0.0281088, "grad_norm": 0.8252747654914856, "learning_rate": 1.2344171812852319e-05, "loss": 2.9709, "step": 166180 }, { "epoch": 0.0281344, "grad_norm": 0.8380191326141357, "learning_rate": 1.2343388964980542e-05, "loss": 3.0696, "step": 166190 }, { "epoch": 0.02816, "grad_norm": 0.8485730886459351, "learning_rate": 1.2342606101912511e-05, "loss": 2.7502, "step": 166200 }, { "epoch": 0.0281856, "grad_norm": 0.788121223449707, "learning_rate": 1.2341823223653299e-05, "loss": 2.9685, "step": 166210 }, { "epoch": 0.0282112, "grad_norm": 0.7473651170730591, "learning_rate": 1.234104033020799e-05, "loss": 2.9687, "step": 166220 }, { "epoch": 0.0282368, "grad_norm": 0.8265305757522583, "learning_rate": 1.2340257421581656e-05, "loss": 3.0382, "step": 166230 }, { "epoch": 0.0282624, "grad_norm": 0.6791077256202698, "learning_rate": 1.233947449777937e-05, "loss": 2.8838, "step": 166240 }, { "epoch": 0.028288, "grad_norm": 0.7092932462692261, "learning_rate": 1.2338691558806217e-05, "loss": 2.9696, "step": 166250 }, { "epoch": 0.0283136, "grad_norm": 0.9984268546104431, "learning_rate": 1.2337908604667274e-05, "loss": 3.0504, "step": 166260 }, { "epoch": 0.0283392, "grad_norm": 1.2537246942520142, "learning_rate": 1.233712563536761e-05, "loss": 2.8627, "step": 166270 }, { "epoch": 0.0283648, "grad_norm": 0.8343387842178345, "learning_rate": 1.233634265091231e-05, "loss": 2.9397, "step": 166280 }, { "epoch": 0.0283904, "grad_norm": 0.7089143395423889, "learning_rate": 1.2335559651306448e-05, "loss": 2.9538, "step": 166290 }, { "epoch": 0.028416, "grad_norm": 0.7575404644012451, "learning_rate": 1.2334776636555101e-05, "loss": 2.8724, "step": 166300 }, { "epoch": 0.0284416, "grad_norm": 0.723747968673706, "learning_rate": 1.2333993606663349e-05, "loss": 3.2785, "step": 166310 }, { "epoch": 0.0284672, "grad_norm": 0.6648651957511902, "learning_rate": 1.2333210561636268e-05, "loss": 3.011, "step": 166320 }, { "epoch": 0.0284928, "grad_norm": 1.1767876148223877, "learning_rate": 1.233242750147894e-05, "loss": 3.0774, "step": 166330 }, { "epoch": 0.0285184, "grad_norm": 0.9811416268348694, "learning_rate": 1.2331644426196436e-05, "loss": 2.9558, "step": 166340 }, { "epoch": 0.028544, "grad_norm": 0.7982310056686401, "learning_rate": 1.233086133579384e-05, "loss": 2.8936, "step": 166350 }, { "epoch": 0.0285696, "grad_norm": 0.7350375652313232, "learning_rate": 1.2330078230276228e-05, "loss": 2.9931, "step": 166360 }, { "epoch": 0.0285952, "grad_norm": 0.7670161128044128, "learning_rate": 1.2329295109648676e-05, "loss": 2.9827, "step": 166370 }, { "epoch": 0.0286208, "grad_norm": 0.7417252063751221, "learning_rate": 1.2328511973916263e-05, "loss": 2.963, "step": 166380 }, { "epoch": 0.0286464, "grad_norm": 0.7298921346664429, "learning_rate": 1.232772882308407e-05, "loss": 2.9516, "step": 166390 }, { "epoch": 0.028672, "grad_norm": 0.8312608003616333, "learning_rate": 1.2326945657157175e-05, "loss": 2.9534, "step": 166400 }, { "epoch": 0.0286976, "grad_norm": 0.7781735062599182, "learning_rate": 1.2326162476140655e-05, "loss": 2.9106, "step": 166410 }, { "epoch": 0.0287232, "grad_norm": 0.7259693145751953, "learning_rate": 1.2325379280039589e-05, "loss": 2.9846, "step": 166420 }, { "epoch": 0.0287488, "grad_norm": 0.8426761031150818, "learning_rate": 1.232459606885906e-05, "loss": 2.8379, "step": 166430 }, { "epoch": 0.0287744, "grad_norm": 0.8157395720481873, "learning_rate": 1.2323812842604139e-05, "loss": 3.0641, "step": 166440 }, { "epoch": 0.0288, "grad_norm": 1.0716662406921387, "learning_rate": 1.2323029601279912e-05, "loss": 3.0225, "step": 166450 }, { "epoch": 0.0288256, "grad_norm": 0.6860570907592773, "learning_rate": 1.2322246344891455e-05, "loss": 3.0054, "step": 166460 }, { "epoch": 0.0288512, "grad_norm": 0.7852005362510681, "learning_rate": 1.2321463073443844e-05, "loss": 3.0675, "step": 166470 }, { "epoch": 0.0288768, "grad_norm": 0.7043718695640564, "learning_rate": 1.2320679786942166e-05, "loss": 2.9039, "step": 166480 }, { "epoch": 0.0289024, "grad_norm": 0.8422663807868958, "learning_rate": 1.2319896485391496e-05, "loss": 2.9682, "step": 166490 }, { "epoch": 0.028928, "grad_norm": 0.7871642708778381, "learning_rate": 1.2319113168796913e-05, "loss": 2.985, "step": 166500 }, { "epoch": 0.0289536, "grad_norm": 0.8225905895233154, "learning_rate": 1.2318329837163499e-05, "loss": 3.0204, "step": 166510 }, { "epoch": 0.0289792, "grad_norm": 0.7729621529579163, "learning_rate": 1.2317546490496328e-05, "loss": 3.1014, "step": 166520 }, { "epoch": 0.0290048, "grad_norm": 0.8161130547523499, "learning_rate": 1.2316763128800488e-05, "loss": 3.1118, "step": 166530 }, { "epoch": 0.0290304, "grad_norm": 0.7640846967697144, "learning_rate": 1.2315979752081055e-05, "loss": 2.8528, "step": 166540 }, { "epoch": 0.029056, "grad_norm": 0.7703695297241211, "learning_rate": 1.2315196360343107e-05, "loss": 2.876, "step": 166550 }, { "epoch": 0.0290816, "grad_norm": 0.8935678005218506, "learning_rate": 1.2314412953591726e-05, "loss": 3.0966, "step": 166560 }, { "epoch": 0.0291072, "grad_norm": 0.7310304045677185, "learning_rate": 1.2313629531831992e-05, "loss": 2.8649, "step": 166570 }, { "epoch": 0.0291328, "grad_norm": 0.7946735620498657, "learning_rate": 1.2312846095068985e-05, "loss": 3.0985, "step": 166580 }, { "epoch": 0.0291584, "grad_norm": 0.7366984486579895, "learning_rate": 1.2312062643307788e-05, "loss": 2.9343, "step": 166590 }, { "epoch": 0.029184, "grad_norm": 0.8237199187278748, "learning_rate": 1.2311279176553477e-05, "loss": 2.828, "step": 166600 }, { "epoch": 0.0292096, "grad_norm": 1.123819351196289, "learning_rate": 1.2310495694811135e-05, "loss": 3.0624, "step": 166610 }, { "epoch": 0.0292352, "grad_norm": 1.0456584692001343, "learning_rate": 1.2309712198085844e-05, "loss": 3.3838, "step": 166620 }, { "epoch": 0.0292608, "grad_norm": 0.8239085674285889, "learning_rate": 1.2308928686382684e-05, "loss": 2.9727, "step": 166630 }, { "epoch": 0.0292864, "grad_norm": 0.9408020973205566, "learning_rate": 1.230814515970673e-05, "loss": 2.9021, "step": 166640 }, { "epoch": 0.029312, "grad_norm": 0.7641359567642212, "learning_rate": 1.2307361618063074e-05, "loss": 3.1409, "step": 166650 }, { "epoch": 0.0293376, "grad_norm": 0.7564322352409363, "learning_rate": 1.2306578061456787e-05, "loss": 3.1969, "step": 166660 }, { "epoch": 0.0293632, "grad_norm": 0.7537514567375183, "learning_rate": 1.2305794489892955e-05, "loss": 2.978, "step": 166670 }, { "epoch": 0.0293888, "grad_norm": 0.8010739684104919, "learning_rate": 1.230501090337666e-05, "loss": 2.9678, "step": 166680 }, { "epoch": 0.0294144, "grad_norm": 0.7807368040084839, "learning_rate": 1.2304227301912983e-05, "loss": 2.9358, "step": 166690 }, { "epoch": 0.02944, "grad_norm": 0.7686527371406555, "learning_rate": 1.2303443685507002e-05, "loss": 3.2765, "step": 166700 }, { "epoch": 0.0294656, "grad_norm": 0.656791090965271, "learning_rate": 1.2302660054163802e-05, "loss": 2.8757, "step": 166710 }, { "epoch": 0.0294912, "grad_norm": 0.7010819911956787, "learning_rate": 1.2301876407888463e-05, "loss": 2.9469, "step": 166720 }, { "epoch": 0.0295168, "grad_norm": 0.7900651097297668, "learning_rate": 1.2301092746686065e-05, "loss": 2.9999, "step": 166730 }, { "epoch": 0.0295424, "grad_norm": 1.7909326553344727, "learning_rate": 1.2300309070561693e-05, "loss": 3.1476, "step": 166740 }, { "epoch": 0.029568, "grad_norm": 0.7146904468536377, "learning_rate": 1.2299525379520432e-05, "loss": 2.9547, "step": 166750 }, { "epoch": 0.0295936, "grad_norm": 0.7676268815994263, "learning_rate": 1.2298741673567357e-05, "loss": 2.9533, "step": 166760 }, { "epoch": 0.0296192, "grad_norm": 0.7589269876480103, "learning_rate": 1.2297957952707553e-05, "loss": 3.08, "step": 166770 }, { "epoch": 0.0296448, "grad_norm": 0.7636048197746277, "learning_rate": 1.2297174216946104e-05, "loss": 3.0314, "step": 166780 }, { "epoch": 0.0296704, "grad_norm": 0.8858301639556885, "learning_rate": 1.229639046628809e-05, "loss": 3.1405, "step": 166790 }, { "epoch": 0.029696, "grad_norm": 1.3247493505477905, "learning_rate": 1.2295606700738593e-05, "loss": 2.9573, "step": 166800 }, { "epoch": 0.0297216, "grad_norm": 0.7550074458122253, "learning_rate": 1.2294822920302696e-05, "loss": 2.9835, "step": 166810 }, { "epoch": 0.0297472, "grad_norm": 1.0033167600631714, "learning_rate": 1.2294039124985486e-05, "loss": 3.0677, "step": 166820 }, { "epoch": 0.0297728, "grad_norm": 0.767516016960144, "learning_rate": 1.229325531479204e-05, "loss": 3.2099, "step": 166830 }, { "epoch": 0.0297984, "grad_norm": 0.7546026706695557, "learning_rate": 1.229247148972744e-05, "loss": 2.8801, "step": 166840 }, { "epoch": 0.029824, "grad_norm": 0.7989012598991394, "learning_rate": 1.2291687649796778e-05, "loss": 2.8264, "step": 166850 }, { "epoch": 0.0298496, "grad_norm": 0.761600911617279, "learning_rate": 1.2290903795005125e-05, "loss": 3.1408, "step": 166860 }, { "epoch": 0.0298752, "grad_norm": 0.7754215002059937, "learning_rate": 1.229011992535757e-05, "loss": 2.9263, "step": 166870 }, { "epoch": 0.0299008, "grad_norm": 0.7348731160163879, "learning_rate": 1.2289336040859199e-05, "loss": 2.9794, "step": 166880 }, { "epoch": 0.0299264, "grad_norm": 1.3273496627807617, "learning_rate": 1.228855214151509e-05, "loss": 2.9498, "step": 166890 }, { "epoch": 0.029952, "grad_norm": 0.6777273416519165, "learning_rate": 1.2287768227330328e-05, "loss": 2.982, "step": 166900 }, { "epoch": 0.0299776, "grad_norm": 0.6911039352416992, "learning_rate": 1.2286984298309999e-05, "loss": 3.2393, "step": 166910 }, { "epoch": 0.0300032, "grad_norm": 0.7450918555259705, "learning_rate": 1.2286200354459186e-05, "loss": 3.2008, "step": 166920 }, { "epoch": 0.0300288, "grad_norm": 0.8275279402732849, "learning_rate": 1.2285416395782968e-05, "loss": 3.0463, "step": 166930 }, { "epoch": 0.0300544, "grad_norm": 0.7342904210090637, "learning_rate": 1.2284632422286433e-05, "loss": 2.8448, "step": 166940 }, { "epoch": 0.03008, "grad_norm": 0.8387961983680725, "learning_rate": 1.2283848433974666e-05, "loss": 2.9935, "step": 166950 }, { "epoch": 0.0301056, "grad_norm": 0.7714235782623291, "learning_rate": 1.2283064430852747e-05, "loss": 2.9595, "step": 166960 }, { "epoch": 0.0301312, "grad_norm": 0.867396891117096, "learning_rate": 1.2282280412925763e-05, "loss": 2.8272, "step": 166970 }, { "epoch": 0.0301568, "grad_norm": 0.680462121963501, "learning_rate": 1.2281496380198794e-05, "loss": 3.0403, "step": 166980 }, { "epoch": 0.0301824, "grad_norm": 0.6796178221702576, "learning_rate": 1.228071233267693e-05, "loss": 2.945, "step": 166990 }, { "epoch": 0.030208, "grad_norm": 0.7146211862564087, "learning_rate": 1.2279928270365252e-05, "loss": 3.0907, "step": 167000 }, { "epoch": 0.0302336, "grad_norm": 0.8125022649765015, "learning_rate": 1.2279144193268844e-05, "loss": 2.9679, "step": 167010 }, { "epoch": 0.0302592, "grad_norm": 0.7494077086448669, "learning_rate": 1.2278360101392795e-05, "loss": 2.9536, "step": 167020 }, { "epoch": 0.0302848, "grad_norm": 0.6992656588554382, "learning_rate": 1.2277575994742184e-05, "loss": 3.1152, "step": 167030 }, { "epoch": 0.0303104, "grad_norm": 0.8402556777000427, "learning_rate": 1.2276791873322097e-05, "loss": 3.1205, "step": 167040 }, { "epoch": 0.030336, "grad_norm": 0.8002217411994934, "learning_rate": 1.2276007737137624e-05, "loss": 2.9465, "step": 167050 }, { "epoch": 0.0303616, "grad_norm": 0.7224688529968262, "learning_rate": 1.227522358619384e-05, "loss": 3.0171, "step": 167060 }, { "epoch": 0.0303872, "grad_norm": 0.8572763800621033, "learning_rate": 1.2274439420495836e-05, "loss": 2.8805, "step": 167070 }, { "epoch": 0.0304128, "grad_norm": 0.6971174478530884, "learning_rate": 1.2273655240048701e-05, "loss": 2.8595, "step": 167080 }, { "epoch": 0.0304384, "grad_norm": 0.7923457026481628, "learning_rate": 1.2272871044857513e-05, "loss": 2.9745, "step": 167090 }, { "epoch": 0.030464, "grad_norm": 0.9527232646942139, "learning_rate": 1.2272086834927361e-05, "loss": 3.167, "step": 167100 }, { "epoch": 0.0304896, "grad_norm": 0.7540363669395447, "learning_rate": 1.2271302610263331e-05, "loss": 2.9702, "step": 167110 }, { "epoch": 0.0305152, "grad_norm": 0.7640970349311829, "learning_rate": 1.2270518370870508e-05, "loss": 2.9838, "step": 167120 }, { "epoch": 0.0305408, "grad_norm": 1.0048514604568481, "learning_rate": 1.2269734116753975e-05, "loss": 2.6506, "step": 167130 }, { "epoch": 0.0305664, "grad_norm": 0.7580425143241882, "learning_rate": 1.2268949847918818e-05, "loss": 3.073, "step": 167140 }, { "epoch": 0.030592, "grad_norm": 0.7866840958595276, "learning_rate": 1.226816556437013e-05, "loss": 2.8645, "step": 167150 }, { "epoch": 0.0306176, "grad_norm": 1.6518728733062744, "learning_rate": 1.2267381266112983e-05, "loss": 3.1697, "step": 167160 }, { "epoch": 0.0306432, "grad_norm": 0.7773293256759644, "learning_rate": 1.2266596953152473e-05, "loss": 2.796, "step": 167170 }, { "epoch": 0.0306688, "grad_norm": 0.8046136498451233, "learning_rate": 1.2265812625493689e-05, "loss": 2.7999, "step": 167180 }, { "epoch": 0.0306944, "grad_norm": 0.7788482308387756, "learning_rate": 1.2265028283141712e-05, "loss": 3.0491, "step": 167190 }, { "epoch": 0.03072, "grad_norm": 0.69981449842453, "learning_rate": 1.2264243926101624e-05, "loss": 2.9927, "step": 167200 }, { "epoch": 0.0307456, "grad_norm": 0.8597198724746704, "learning_rate": 1.2263459554378519e-05, "loss": 2.8814, "step": 167210 }, { "epoch": 0.0307712, "grad_norm": 0.8035961985588074, "learning_rate": 1.226267516797748e-05, "loss": 2.7449, "step": 167220 }, { "epoch": 0.0307968, "grad_norm": 0.7686842679977417, "learning_rate": 1.2261890766903593e-05, "loss": 2.9336, "step": 167230 }, { "epoch": 0.0308224, "grad_norm": 2.9586546421051025, "learning_rate": 1.2261106351161945e-05, "loss": 3.523, "step": 167240 }, { "epoch": 0.030848, "grad_norm": 0.7042770981788635, "learning_rate": 1.2260321920757625e-05, "loss": 2.9075, "step": 167250 }, { "epoch": 0.0308736, "grad_norm": 0.7057177424430847, "learning_rate": 1.2259537475695715e-05, "loss": 3.1169, "step": 167260 }, { "epoch": 0.0308992, "grad_norm": 1.095282793045044, "learning_rate": 1.2258753015981308e-05, "loss": 2.923, "step": 167270 }, { "epoch": 0.0309248, "grad_norm": 0.7881894707679749, "learning_rate": 1.2257968541619486e-05, "loss": 3.0491, "step": 167280 }, { "epoch": 0.0309504, "grad_norm": 1.0278434753417969, "learning_rate": 1.225718405261534e-05, "loss": 3.1088, "step": 167290 }, { "epoch": 0.030976, "grad_norm": 0.7607501149177551, "learning_rate": 1.2256399548973956e-05, "loss": 3.0047, "step": 167300 }, { "epoch": 0.0310016, "grad_norm": 0.8783835768699646, "learning_rate": 1.2255615030700419e-05, "loss": 3.5122, "step": 167310 }, { "epoch": 0.0310272, "grad_norm": 0.7688623666763306, "learning_rate": 1.2254830497799818e-05, "loss": 2.88, "step": 167320 }, { "epoch": 0.0310528, "grad_norm": 0.8118489384651184, "learning_rate": 1.225404595027724e-05, "loss": 3.0458, "step": 167330 }, { "epoch": 0.0310784, "grad_norm": 1.2582391500473022, "learning_rate": 1.2253261388137774e-05, "loss": 2.9943, "step": 167340 }, { "epoch": 0.031104, "grad_norm": 0.6907967329025269, "learning_rate": 1.2252476811386507e-05, "loss": 3.1976, "step": 167350 }, { "epoch": 0.0311296, "grad_norm": 0.962096631526947, "learning_rate": 1.2251692220028524e-05, "loss": 2.699, "step": 167360 }, { "epoch": 0.0311552, "grad_norm": 0.7717009782791138, "learning_rate": 1.225090761406892e-05, "loss": 2.9738, "step": 167370 }, { "epoch": 0.0311808, "grad_norm": 1.0942130088806152, "learning_rate": 1.2250122993512776e-05, "loss": 3.0203, "step": 167380 }, { "epoch": 0.0312064, "grad_norm": 0.7686832547187805, "learning_rate": 1.2249338358365182e-05, "loss": 2.8186, "step": 167390 }, { "epoch": 0.031232, "grad_norm": 0.7188810706138611, "learning_rate": 1.2248553708631226e-05, "loss": 3.0028, "step": 167400 }, { "epoch": 0.0312576, "grad_norm": 0.7574512958526611, "learning_rate": 1.2247769044315996e-05, "loss": 2.6154, "step": 167410 }, { "epoch": 0.0312832, "grad_norm": 0.7383067607879639, "learning_rate": 1.2246984365424583e-05, "loss": 2.9452, "step": 167420 }, { "epoch": 0.0313088, "grad_norm": 0.6793121695518494, "learning_rate": 1.2246199671962071e-05, "loss": 3.064, "step": 167430 }, { "epoch": 0.0313344, "grad_norm": 1.1580290794372559, "learning_rate": 1.2245414963933555e-05, "loss": 3.0059, "step": 167440 }, { "epoch": 0.03136, "grad_norm": 0.72611403465271, "learning_rate": 1.2244630241344117e-05, "loss": 2.9966, "step": 167450 }, { "epoch": 0.0313856, "grad_norm": 0.7086089253425598, "learning_rate": 1.224384550419885e-05, "loss": 3.018, "step": 167460 }, { "epoch": 0.0314112, "grad_norm": 0.6698803305625916, "learning_rate": 1.224306075250284e-05, "loss": 2.993, "step": 167470 }, { "epoch": 0.0314368, "grad_norm": 0.7120078206062317, "learning_rate": 1.224227598626118e-05, "loss": 3.1672, "step": 167480 }, { "epoch": 0.0314624, "grad_norm": 0.7900748252868652, "learning_rate": 1.2241491205478955e-05, "loss": 2.7847, "step": 167490 }, { "epoch": 0.031488, "grad_norm": 0.9977065920829773, "learning_rate": 1.2240706410161252e-05, "loss": 2.6476, "step": 167500 }, { "epoch": 0.0315136, "grad_norm": 0.7756576538085938, "learning_rate": 1.2239921600313165e-05, "loss": 3.1027, "step": 167510 }, { "epoch": 0.0315392, "grad_norm": 0.7367790937423706, "learning_rate": 1.2239136775939781e-05, "loss": 2.9147, "step": 167520 }, { "epoch": 0.0315648, "grad_norm": 0.8437474966049194, "learning_rate": 1.2238351937046194e-05, "loss": 2.9333, "step": 167530 }, { "epoch": 0.0315904, "grad_norm": 0.8778507113456726, "learning_rate": 1.2237567083637488e-05, "loss": 2.9742, "step": 167540 }, { "epoch": 0.031616, "grad_norm": 0.6954713463783264, "learning_rate": 1.2236782215718751e-05, "loss": 3.2022, "step": 167550 }, { "epoch": 0.0316416, "grad_norm": 0.6936032176017761, "learning_rate": 1.2235997333295077e-05, "loss": 2.948, "step": 167560 }, { "epoch": 0.0316672, "grad_norm": 0.6792190670967102, "learning_rate": 1.2235212436371558e-05, "loss": 2.8849, "step": 167570 }, { "epoch": 0.0316928, "grad_norm": 0.9562922716140747, "learning_rate": 1.2234427524953278e-05, "loss": 3.174, "step": 167580 }, { "epoch": 0.0317184, "grad_norm": 0.8937747478485107, "learning_rate": 1.2233642599045328e-05, "loss": 3.4188, "step": 167590 }, { "epoch": 0.031744, "grad_norm": 0.7930257320404053, "learning_rate": 1.22328576586528e-05, "loss": 2.9446, "step": 167600 }, { "epoch": 0.0317696, "grad_norm": 0.7927343249320984, "learning_rate": 1.2232072703780785e-05, "loss": 2.8265, "step": 167610 }, { "epoch": 0.0317952, "grad_norm": 0.7108399271965027, "learning_rate": 1.223128773443437e-05, "loss": 2.9361, "step": 167620 }, { "epoch": 0.0318208, "grad_norm": 0.8430777192115784, "learning_rate": 1.2230502750618648e-05, "loss": 3.1675, "step": 167630 }, { "epoch": 0.0318464, "grad_norm": 0.7337055802345276, "learning_rate": 1.222971775233871e-05, "loss": 2.9442, "step": 167640 }, { "epoch": 0.031872, "grad_norm": 0.7713225483894348, "learning_rate": 1.2228932739599642e-05, "loss": 2.8845, "step": 167650 }, { "epoch": 0.0318976, "grad_norm": 0.799058735370636, "learning_rate": 1.222814771240654e-05, "loss": 3.0071, "step": 167660 }, { "epoch": 0.0319232, "grad_norm": 0.6942851543426514, "learning_rate": 1.2227362670764488e-05, "loss": 2.7233, "step": 167670 }, { "epoch": 0.0319488, "grad_norm": 0.719082772731781, "learning_rate": 1.2226577614678585e-05, "loss": 2.7796, "step": 167680 }, { "epoch": 0.0319744, "grad_norm": 0.7747388482093811, "learning_rate": 1.2225792544153919e-05, "loss": 3.0207, "step": 167690 }, { "epoch": 0.032, "grad_norm": 1.2482446432113647, "learning_rate": 1.2225007459195575e-05, "loss": 3.1784, "step": 167700 }, { "epoch": 0.0320256, "grad_norm": 0.8015198707580566, "learning_rate": 1.2224222359808655e-05, "loss": 3.0412, "step": 167710 }, { "epoch": 0.0320512, "grad_norm": 0.7263438105583191, "learning_rate": 1.222343724599824e-05, "loss": 3.0304, "step": 167720 }, { "epoch": 0.0320768, "grad_norm": 0.9216289520263672, "learning_rate": 1.2222652117769428e-05, "loss": 2.9345, "step": 167730 }, { "epoch": 0.0321024, "grad_norm": 0.7084512710571289, "learning_rate": 1.2221866975127305e-05, "loss": 2.926, "step": 167740 }, { "epoch": 0.032128, "grad_norm": 0.9483878016471863, "learning_rate": 1.2221081818076967e-05, "loss": 2.8281, "step": 167750 }, { "epoch": 0.0321536, "grad_norm": 0.840840756893158, "learning_rate": 1.2220296646623501e-05, "loss": 2.8519, "step": 167760 }, { "epoch": 0.0321792, "grad_norm": 0.8221477270126343, "learning_rate": 1.2219511460772004e-05, "loss": 2.9464, "step": 167770 }, { "epoch": 0.0322048, "grad_norm": 0.7132425308227539, "learning_rate": 1.2218726260527565e-05, "loss": 3.0016, "step": 167780 }, { "epoch": 0.0322304, "grad_norm": 0.9712709784507751, "learning_rate": 1.2217941045895273e-05, "loss": 2.9445, "step": 167790 }, { "epoch": 0.032256, "grad_norm": 0.718955397605896, "learning_rate": 1.2217155816880227e-05, "loss": 2.8987, "step": 167800 }, { "epoch": 0.0322816, "grad_norm": 0.8136925101280212, "learning_rate": 1.2216370573487514e-05, "loss": 3.0626, "step": 167810 }, { "epoch": 0.0323072, "grad_norm": 0.9198264479637146, "learning_rate": 1.2215585315722225e-05, "loss": 2.9919, "step": 167820 }, { "epoch": 0.0323328, "grad_norm": 1.134775996208191, "learning_rate": 1.2214800043589454e-05, "loss": 2.8904, "step": 167830 }, { "epoch": 0.0323584, "grad_norm": 1.062509536743164, "learning_rate": 1.2214014757094296e-05, "loss": 2.9069, "step": 167840 }, { "epoch": 0.032384, "grad_norm": 0.9217603206634521, "learning_rate": 1.221322945624184e-05, "loss": 3.0273, "step": 167850 }, { "epoch": 0.0324096, "grad_norm": 0.8697993755340576, "learning_rate": 1.2212444141037175e-05, "loss": 2.8617, "step": 167860 }, { "epoch": 0.0324352, "grad_norm": 0.7057008147239685, "learning_rate": 1.22116588114854e-05, "loss": 3.1057, "step": 167870 }, { "epoch": 0.0324608, "grad_norm": 0.8031547665596008, "learning_rate": 1.221087346759161e-05, "loss": 2.7995, "step": 167880 }, { "epoch": 0.0324864, "grad_norm": 0.7203283905982971, "learning_rate": 1.2210088109360887e-05, "loss": 2.862, "step": 167890 }, { "epoch": 0.032512, "grad_norm": 0.7217540740966797, "learning_rate": 1.2209302736798332e-05, "loss": 3.0396, "step": 167900 }, { "epoch": 0.0325376, "grad_norm": 0.7530310750007629, "learning_rate": 1.2208517349909038e-05, "loss": 2.8895, "step": 167910 }, { "epoch": 0.0325632, "grad_norm": 0.8142457604408264, "learning_rate": 1.2207731948698094e-05, "loss": 2.8557, "step": 167920 }, { "epoch": 0.0325888, "grad_norm": 0.668989360332489, "learning_rate": 1.2206946533170593e-05, "loss": 3.0149, "step": 167930 }, { "epoch": 0.0326144, "grad_norm": 0.8723989129066467, "learning_rate": 1.2206161103331633e-05, "loss": 2.8664, "step": 167940 }, { "epoch": 0.03264, "grad_norm": 0.9265856146812439, "learning_rate": 1.2205375659186303e-05, "loss": 2.8845, "step": 167950 }, { "epoch": 0.0326656, "grad_norm": 0.7000398635864258, "learning_rate": 1.22045902007397e-05, "loss": 2.8079, "step": 167960 }, { "epoch": 0.0326912, "grad_norm": 0.7895260453224182, "learning_rate": 1.2203804727996914e-05, "loss": 2.8342, "step": 167970 }, { "epoch": 0.0327168, "grad_norm": 0.8244807720184326, "learning_rate": 1.220301924096304e-05, "loss": 2.8675, "step": 167980 }, { "epoch": 0.0327424, "grad_norm": 0.6994374394416809, "learning_rate": 1.2202233739643173e-05, "loss": 3.0647, "step": 167990 }, { "epoch": 0.032768, "grad_norm": 0.8661199808120728, "learning_rate": 1.2201448224042404e-05, "loss": 2.9443, "step": 168000 }, { "epoch": 0.0327936, "grad_norm": 0.8356528878211975, "learning_rate": 1.220066269416583e-05, "loss": 3.106, "step": 168010 }, { "epoch": 0.0328192, "grad_norm": 0.8613563179969788, "learning_rate": 1.2199877150018544e-05, "loss": 3.245, "step": 168020 }, { "epoch": 0.0328448, "grad_norm": 0.7806801795959473, "learning_rate": 1.2199091591605634e-05, "loss": 2.8428, "step": 168030 }, { "epoch": 0.0328704, "grad_norm": 0.7513066530227661, "learning_rate": 1.2198306018932204e-05, "loss": 2.9771, "step": 168040 }, { "epoch": 0.032896, "grad_norm": 0.8596505522727966, "learning_rate": 1.2197520432003342e-05, "loss": 2.7751, "step": 168050 }, { "epoch": 0.0329216, "grad_norm": 0.8001241683959961, "learning_rate": 1.2196734830824146e-05, "loss": 3.0237, "step": 168060 }, { "epoch": 0.0329472, "grad_norm": 1.9531008005142212, "learning_rate": 1.219594921539971e-05, "loss": 3.0485, "step": 168070 }, { "epoch": 0.0329728, "grad_norm": 1.2441197633743286, "learning_rate": 1.2195163585735123e-05, "loss": 3.0289, "step": 168080 }, { "epoch": 0.0329984, "grad_norm": 1.3449668884277344, "learning_rate": 1.2194377941835485e-05, "loss": 2.8721, "step": 168090 }, { "epoch": 0.033024, "grad_norm": 0.9966877698898315, "learning_rate": 1.2193592283705889e-05, "loss": 3.1543, "step": 168100 }, { "epoch": 0.0330496, "grad_norm": 0.7120347619056702, "learning_rate": 1.2192806611351431e-05, "loss": 3.016, "step": 168110 }, { "epoch": 0.0330752, "grad_norm": 0.7914604544639587, "learning_rate": 1.2192020924777201e-05, "loss": 3.065, "step": 168120 }, { "epoch": 0.0331008, "grad_norm": 0.8190692663192749, "learning_rate": 1.21912352239883e-05, "loss": 3.1715, "step": 168130 }, { "epoch": 0.0331264, "grad_norm": 0.8252336382865906, "learning_rate": 1.2190449508989826e-05, "loss": 3.0375, "step": 168140 }, { "epoch": 0.033152, "grad_norm": 0.8909265995025635, "learning_rate": 1.2189663779786863e-05, "loss": 3.1432, "step": 168150 }, { "epoch": 0.0331776, "grad_norm": 0.642023503780365, "learning_rate": 1.2188878036384515e-05, "loss": 2.7962, "step": 168160 }, { "epoch": 0.0332032, "grad_norm": 0.7084571123123169, "learning_rate": 1.2188092278787875e-05, "loss": 2.8732, "step": 168170 }, { "epoch": 0.0332288, "grad_norm": 0.7749373316764832, "learning_rate": 1.2187306507002037e-05, "loss": 3.0546, "step": 168180 }, { "epoch": 0.0332544, "grad_norm": 0.7869170308113098, "learning_rate": 1.2186520721032096e-05, "loss": 3.1272, "step": 168190 }, { "epoch": 0.03328, "grad_norm": 0.7891371846199036, "learning_rate": 1.218573492088315e-05, "loss": 2.8098, "step": 168200 }, { "epoch": 0.0333056, "grad_norm": 1.3439435958862305, "learning_rate": 1.2184949106560295e-05, "loss": 3.1276, "step": 168210 }, { "epoch": 0.0333312, "grad_norm": 0.6862838864326477, "learning_rate": 1.2184163278068626e-05, "loss": 2.7745, "step": 168220 }, { "epoch": 0.0333568, "grad_norm": 0.8556116819381714, "learning_rate": 1.2183377435413239e-05, "loss": 3.1066, "step": 168230 }, { "epoch": 0.0333824, "grad_norm": 0.7932107448577881, "learning_rate": 1.2182591578599227e-05, "loss": 3.293, "step": 168240 }, { "epoch": 0.033408, "grad_norm": 0.8682769536972046, "learning_rate": 1.2181805707631691e-05, "loss": 2.7038, "step": 168250 }, { "epoch": 0.0334336, "grad_norm": 0.7443700432777405, "learning_rate": 1.2181019822515725e-05, "loss": 3.1791, "step": 168260 }, { "epoch": 0.0334592, "grad_norm": 0.7121846675872803, "learning_rate": 1.2180233923256427e-05, "loss": 3.2255, "step": 168270 }, { "epoch": 0.0334848, "grad_norm": 0.8056380748748779, "learning_rate": 1.2179448009858888e-05, "loss": 3.0501, "step": 168280 }, { "epoch": 0.0335104, "grad_norm": 0.7888695001602173, "learning_rate": 1.2178662082328206e-05, "loss": 2.8965, "step": 168290 }, { "epoch": 0.033536, "grad_norm": 0.977628231048584, "learning_rate": 1.2177876140669482e-05, "loss": 2.9776, "step": 168300 }, { "epoch": 0.0335616, "grad_norm": 0.7558168768882751, "learning_rate": 1.217709018488781e-05, "loss": 2.8828, "step": 168310 }, { "epoch": 0.0335872, "grad_norm": 0.923981785774231, "learning_rate": 1.2176304214988287e-05, "loss": 2.9836, "step": 168320 }, { "epoch": 0.0336128, "grad_norm": 0.9251224994659424, "learning_rate": 1.2175518230976011e-05, "loss": 3.0992, "step": 168330 }, { "epoch": 0.0336384, "grad_norm": 0.7564983367919922, "learning_rate": 1.2174732232856074e-05, "loss": 2.8848, "step": 168340 }, { "epoch": 0.033664, "grad_norm": 0.9841768741607666, "learning_rate": 1.217394622063358e-05, "loss": 3.0172, "step": 168350 }, { "epoch": 0.0336896, "grad_norm": 0.7291246056556702, "learning_rate": 1.217316019431362e-05, "loss": 2.8704, "step": 168360 }, { "epoch": 0.0337152, "grad_norm": 0.982746422290802, "learning_rate": 1.2172374153901297e-05, "loss": 2.8321, "step": 168370 }, { "epoch": 0.0337408, "grad_norm": 0.9661498069763184, "learning_rate": 1.21715880994017e-05, "loss": 2.9538, "step": 168380 }, { "epoch": 0.0337664, "grad_norm": 0.7373771071434021, "learning_rate": 1.2170802030819933e-05, "loss": 3.0491, "step": 168390 }, { "epoch": 0.033792, "grad_norm": 0.7085177898406982, "learning_rate": 1.2170015948161093e-05, "loss": 2.8761, "step": 168400 }, { "epoch": 0.0338176, "grad_norm": 0.9798237085342407, "learning_rate": 1.2169229851430276e-05, "loss": 3.2183, "step": 168410 }, { "epoch": 0.0338432, "grad_norm": 0.8595061898231506, "learning_rate": 1.2168443740632581e-05, "loss": 2.8238, "step": 168420 }, { "epoch": 0.0338688, "grad_norm": 0.8340810537338257, "learning_rate": 1.2167657615773103e-05, "loss": 2.9895, "step": 168430 }, { "epoch": 0.0338944, "grad_norm": 0.9427556991577148, "learning_rate": 1.2166871476856943e-05, "loss": 2.9595, "step": 168440 }, { "epoch": 0.03392, "grad_norm": 0.9744496941566467, "learning_rate": 1.2166085323889196e-05, "loss": 3.0563, "step": 168450 }, { "epoch": 0.0339456, "grad_norm": 0.7500432133674622, "learning_rate": 1.2165299156874961e-05, "loss": 2.8982, "step": 168460 }, { "epoch": 0.0339712, "grad_norm": 0.7502318024635315, "learning_rate": 1.2164512975819336e-05, "loss": 2.846, "step": 168470 }, { "epoch": 0.0339968, "grad_norm": 0.8452950716018677, "learning_rate": 1.2163726780727421e-05, "loss": 2.6614, "step": 168480 }, { "epoch": 0.0340224, "grad_norm": 0.877834141254425, "learning_rate": 1.2162940571604312e-05, "loss": 2.8778, "step": 168490 }, { "epoch": 0.034048, "grad_norm": 0.8545234799385071, "learning_rate": 1.216215434845511e-05, "loss": 2.9699, "step": 168500 }, { "epoch": 0.0340736, "grad_norm": 0.6691372990608215, "learning_rate": 1.216136811128491e-05, "loss": 3.0747, "step": 168510 }, { "epoch": 0.0340992, "grad_norm": 0.8205648064613342, "learning_rate": 1.2160581860098813e-05, "loss": 3.0031, "step": 168520 }, { "epoch": 0.0341248, "grad_norm": 0.7915347814559937, "learning_rate": 1.2159795594901918e-05, "loss": 3.2785, "step": 168530 }, { "epoch": 0.0341504, "grad_norm": 0.7314020395278931, "learning_rate": 1.215900931569932e-05, "loss": 2.9002, "step": 168540 }, { "epoch": 0.034176, "grad_norm": 0.8141231536865234, "learning_rate": 1.2158223022496119e-05, "loss": 3.0292, "step": 168550 }, { "epoch": 0.0342016, "grad_norm": 0.9443104863166809, "learning_rate": 1.215743671529742e-05, "loss": 3.0648, "step": 168560 }, { "epoch": 0.0342272, "grad_norm": 0.6867789030075073, "learning_rate": 1.2156650394108316e-05, "loss": 2.7861, "step": 168570 }, { "epoch": 0.0342528, "grad_norm": 0.7942747473716736, "learning_rate": 1.2155864058933904e-05, "loss": 2.7403, "step": 168580 }, { "epoch": 0.0342784, "grad_norm": 0.7642173171043396, "learning_rate": 1.2155077709779288e-05, "loss": 3.0263, "step": 168590 }, { "epoch": 0.034304, "grad_norm": 0.7606890797615051, "learning_rate": 1.2154291346649569e-05, "loss": 2.9822, "step": 168600 }, { "epoch": 0.0343296, "grad_norm": 0.7663426995277405, "learning_rate": 1.2153504969549842e-05, "loss": 2.8582, "step": 168610 }, { "epoch": 0.0343552, "grad_norm": 0.777625322341919, "learning_rate": 1.2152718578485206e-05, "loss": 3.0673, "step": 168620 }, { "epoch": 0.0343808, "grad_norm": 0.7675381898880005, "learning_rate": 1.2151932173460762e-05, "loss": 2.712, "step": 168630 }, { "epoch": 0.0344064, "grad_norm": 0.7123093008995056, "learning_rate": 1.215114575448161e-05, "loss": 2.6039, "step": 168640 }, { "epoch": 0.034432, "grad_norm": 0.7480199337005615, "learning_rate": 1.2150359321552849e-05, "loss": 3.0971, "step": 168650 }, { "epoch": 0.0344576, "grad_norm": 0.773407518863678, "learning_rate": 1.2149572874679582e-05, "loss": 3.0516, "step": 168660 }, { "epoch": 0.0344832, "grad_norm": 1.009102463722229, "learning_rate": 1.2148786413866905e-05, "loss": 2.8832, "step": 168670 }, { "epoch": 0.0345088, "grad_norm": 0.6746913194656372, "learning_rate": 1.2147999939119916e-05, "loss": 3.0308, "step": 168680 }, { "epoch": 0.0345344, "grad_norm": 0.8937622904777527, "learning_rate": 1.2147213450443722e-05, "loss": 3.1024, "step": 168690 }, { "epoch": 0.03456, "grad_norm": 0.9163482189178467, "learning_rate": 1.214642694784342e-05, "loss": 3.144, "step": 168700 }, { "epoch": 0.0345856, "grad_norm": 0.7493677735328674, "learning_rate": 1.2145640431324109e-05, "loss": 3.1321, "step": 168710 }, { "epoch": 0.0346112, "grad_norm": 0.7408660650253296, "learning_rate": 1.2144853900890888e-05, "loss": 3.001, "step": 168720 }, { "epoch": 0.0346368, "grad_norm": 0.7491288781166077, "learning_rate": 1.2144067356548859e-05, "loss": 2.943, "step": 168730 }, { "epoch": 0.0346624, "grad_norm": 0.7651917934417725, "learning_rate": 1.2143280798303125e-05, "loss": 2.9761, "step": 168740 }, { "epoch": 0.034688, "grad_norm": 0.8179588913917542, "learning_rate": 1.2142494226158782e-05, "loss": 2.9889, "step": 168750 }, { "epoch": 0.0347136, "grad_norm": 0.7610487341880798, "learning_rate": 1.2141707640120938e-05, "loss": 3.0349, "step": 168760 }, { "epoch": 0.0347392, "grad_norm": 1.0748748779296875, "learning_rate": 1.2140921040194686e-05, "loss": 3.0483, "step": 168770 }, { "epoch": 0.0347648, "grad_norm": 0.8937400579452515, "learning_rate": 1.214013442638513e-05, "loss": 3.1155, "step": 168780 }, { "epoch": 0.0347904, "grad_norm": 0.8217286467552185, "learning_rate": 1.2139347798697372e-05, "loss": 3.1028, "step": 168790 }, { "epoch": 0.034816, "grad_norm": 0.7112692594528198, "learning_rate": 1.2138561157136515e-05, "loss": 3.2817, "step": 168800 }, { "epoch": 0.0348416, "grad_norm": 0.6985200643539429, "learning_rate": 1.213777450170765e-05, "loss": 2.9294, "step": 168810 }, { "epoch": 0.0348672, "grad_norm": 0.6884267330169678, "learning_rate": 1.213698783241589e-05, "loss": 2.9998, "step": 168820 }, { "epoch": 0.0348928, "grad_norm": 2.284637451171875, "learning_rate": 1.2136201149266333e-05, "loss": 2.945, "step": 168830 }, { "epoch": 0.0349184, "grad_norm": 0.7688633799552917, "learning_rate": 1.2135414452264078e-05, "loss": 2.8397, "step": 168840 }, { "epoch": 0.034944, "grad_norm": 0.9090843796730042, "learning_rate": 1.2134627741414227e-05, "loss": 2.9432, "step": 168850 }, { "epoch": 0.0349696, "grad_norm": 0.7624059319496155, "learning_rate": 1.2133841016721883e-05, "loss": 3.0898, "step": 168860 }, { "epoch": 0.0349952, "grad_norm": 0.7759736776351929, "learning_rate": 1.2133054278192149e-05, "loss": 2.9461, "step": 168870 }, { "epoch": 0.0350208, "grad_norm": 0.7353754043579102, "learning_rate": 1.213226752583012e-05, "loss": 3.0072, "step": 168880 }, { "epoch": 0.0350464, "grad_norm": 0.6874367594718933, "learning_rate": 1.2131480759640905e-05, "loss": 2.9191, "step": 168890 }, { "epoch": 0.035072, "grad_norm": 0.7979167699813843, "learning_rate": 1.2130693979629606e-05, "loss": 3.0127, "step": 168900 }, { "epoch": 0.0350976, "grad_norm": 0.7401205897331238, "learning_rate": 1.2129907185801322e-05, "loss": 2.7712, "step": 168910 }, { "epoch": 0.0351232, "grad_norm": 0.7495757937431335, "learning_rate": 1.2129120378161154e-05, "loss": 2.8727, "step": 168920 }, { "epoch": 0.0351488, "grad_norm": 1.0492268800735474, "learning_rate": 1.212833355671421e-05, "loss": 2.9426, "step": 168930 }, { "epoch": 0.0351744, "grad_norm": 0.8949528932571411, "learning_rate": 1.2127546721465584e-05, "loss": 3.0167, "step": 168940 }, { "epoch": 0.0352, "grad_norm": 0.6765617728233337, "learning_rate": 1.2126759872420386e-05, "loss": 2.9262, "step": 168950 }, { "epoch": 0.0352256, "grad_norm": 0.7588096857070923, "learning_rate": 1.2125973009583715e-05, "loss": 2.8814, "step": 168960 }, { "epoch": 0.0352512, "grad_norm": 0.6937249898910522, "learning_rate": 1.2125186132960673e-05, "loss": 2.8707, "step": 168970 }, { "epoch": 0.0352768, "grad_norm": 0.923653781414032, "learning_rate": 1.2124399242556364e-05, "loss": 3.0213, "step": 168980 }, { "epoch": 0.0353024, "grad_norm": 0.7831915616989136, "learning_rate": 1.212361233837589e-05, "loss": 2.8582, "step": 168990 }, { "epoch": 0.035328, "grad_norm": 0.9240729808807373, "learning_rate": 1.2122825420424356e-05, "loss": 2.9886, "step": 169000 }, { "epoch": 0.0353536, "grad_norm": 0.7582876086235046, "learning_rate": 1.2122038488706863e-05, "loss": 3.3045, "step": 169010 }, { "epoch": 0.0353792, "grad_norm": 0.7905002236366272, "learning_rate": 1.2121251543228512e-05, "loss": 3.0143, "step": 169020 }, { "epoch": 0.0354048, "grad_norm": 0.7418729662895203, "learning_rate": 1.2120464583994411e-05, "loss": 3.0467, "step": 169030 }, { "epoch": 0.0354304, "grad_norm": 0.7413344979286194, "learning_rate": 1.211967761100966e-05, "loss": 2.9493, "step": 169040 }, { "epoch": 0.035456, "grad_norm": 0.7174448370933533, "learning_rate": 1.2118890624279362e-05, "loss": 2.7593, "step": 169050 }, { "epoch": 0.0354816, "grad_norm": 0.7300994396209717, "learning_rate": 1.2118103623808625e-05, "loss": 3.092, "step": 169060 }, { "epoch": 0.0355072, "grad_norm": 0.7719937562942505, "learning_rate": 1.2117316609602545e-05, "loss": 3.0087, "step": 169070 }, { "epoch": 0.0355328, "grad_norm": 0.8246413469314575, "learning_rate": 1.211652958166623e-05, "loss": 3.1202, "step": 169080 }, { "epoch": 0.0355584, "grad_norm": 0.7843719124794006, "learning_rate": 1.2115742540004785e-05, "loss": 3.1528, "step": 169090 }, { "epoch": 0.035584, "grad_norm": 0.8083593249320984, "learning_rate": 1.2114955484623311e-05, "loss": 2.9774, "step": 169100 }, { "epoch": 0.0356096, "grad_norm": 0.7846943140029907, "learning_rate": 1.2114168415526912e-05, "loss": 3.0038, "step": 169110 }, { "epoch": 0.0356352, "grad_norm": 0.7351388335227966, "learning_rate": 1.2113381332720692e-05, "loss": 3.103, "step": 169120 }, { "epoch": 0.0356608, "grad_norm": 0.7901125550270081, "learning_rate": 1.211259423620976e-05, "loss": 2.9494, "step": 169130 }, { "epoch": 0.0356864, "grad_norm": 0.712273895740509, "learning_rate": 1.2111807125999213e-05, "loss": 3.1125, "step": 169140 }, { "epoch": 0.035712, "grad_norm": 0.7901937365531921, "learning_rate": 1.2111020002094158e-05, "loss": 2.8601, "step": 169150 }, { "epoch": 0.0357376, "grad_norm": 0.7803787589073181, "learning_rate": 1.2110232864499697e-05, "loss": 3.1625, "step": 169160 }, { "epoch": 0.0357632, "grad_norm": 0.8635037541389465, "learning_rate": 1.210944571322094e-05, "loss": 3.0061, "step": 169170 }, { "epoch": 0.0357888, "grad_norm": 0.756010115146637, "learning_rate": 1.2108658548262988e-05, "loss": 3.0522, "step": 169180 }, { "epoch": 0.0358144, "grad_norm": 1.212026596069336, "learning_rate": 1.2107871369630945e-05, "loss": 2.9925, "step": 169190 }, { "epoch": 0.03584, "grad_norm": 0.8088088035583496, "learning_rate": 1.2107084177329918e-05, "loss": 3.0119, "step": 169200 }, { "epoch": 0.0358656, "grad_norm": 0.9118105173110962, "learning_rate": 1.2106296971365007e-05, "loss": 3.2926, "step": 169210 }, { "epoch": 0.0358912, "grad_norm": 0.7116754651069641, "learning_rate": 1.2105509751741321e-05, "loss": 2.9657, "step": 169220 }, { "epoch": 0.0359168, "grad_norm": 0.8999078273773193, "learning_rate": 1.2104722518463967e-05, "loss": 2.8501, "step": 169230 }, { "epoch": 0.0359424, "grad_norm": 0.7874734401702881, "learning_rate": 1.2103935271538043e-05, "loss": 3.0333, "step": 169240 }, { "epoch": 0.035968, "grad_norm": 0.7297267913818359, "learning_rate": 1.2103148010968659e-05, "loss": 2.9242, "step": 169250 }, { "epoch": 0.0359936, "grad_norm": 0.8701591491699219, "learning_rate": 1.2102360736760918e-05, "loss": 2.9578, "step": 169260 }, { "epoch": 0.0360192, "grad_norm": 0.709636926651001, "learning_rate": 1.2101573448919927e-05, "loss": 2.9572, "step": 169270 }, { "epoch": 0.0360448, "grad_norm": 0.8278937935829163, "learning_rate": 1.2100786147450792e-05, "loss": 3.1311, "step": 169280 }, { "epoch": 0.0360704, "grad_norm": 0.7322900891304016, "learning_rate": 1.2099998832358617e-05, "loss": 2.9979, "step": 169290 }, { "epoch": 0.036096, "grad_norm": 0.7176112532615662, "learning_rate": 1.2099211503648507e-05, "loss": 3.0764, "step": 169300 }, { "epoch": 0.0361216, "grad_norm": 0.889128565788269, "learning_rate": 1.2098424161325568e-05, "loss": 2.9037, "step": 169310 }, { "epoch": 0.0361472, "grad_norm": 0.7610548734664917, "learning_rate": 1.2097636805394908e-05, "loss": 3.0793, "step": 169320 }, { "epoch": 0.0361728, "grad_norm": 0.722619354724884, "learning_rate": 1.2096849435861628e-05, "loss": 2.9747, "step": 169330 }, { "epoch": 0.0361984, "grad_norm": 1.1749686002731323, "learning_rate": 1.2096062052730835e-05, "loss": 3.0749, "step": 169340 }, { "epoch": 0.036224, "grad_norm": 0.7779594659805298, "learning_rate": 1.2095274656007642e-05, "loss": 3.1023, "step": 169350 }, { "epoch": 0.0362496, "grad_norm": 0.7972935438156128, "learning_rate": 1.2094487245697146e-05, "loss": 3.1259, "step": 169360 }, { "epoch": 0.0362752, "grad_norm": 0.8055484294891357, "learning_rate": 1.2093699821804459e-05, "loss": 3.038, "step": 169370 }, { "epoch": 0.0363008, "grad_norm": 0.7667295336723328, "learning_rate": 1.2092912384334683e-05, "loss": 3.0533, "step": 169380 }, { "epoch": 0.0363264, "grad_norm": 0.920808732509613, "learning_rate": 1.209212493329293e-05, "loss": 3.2141, "step": 169390 }, { "epoch": 0.036352, "grad_norm": 1.567270040512085, "learning_rate": 1.2091337468684299e-05, "loss": 3.1046, "step": 169400 }, { "epoch": 0.0363776, "grad_norm": 0.8431521058082581, "learning_rate": 1.2090549990513902e-05, "loss": 2.7515, "step": 169410 }, { "epoch": 0.0364032, "grad_norm": 1.7408185005187988, "learning_rate": 1.2089762498786844e-05, "loss": 2.8772, "step": 169420 }, { "epoch": 0.0364288, "grad_norm": 0.9755913019180298, "learning_rate": 1.2088974993508228e-05, "loss": 2.8906, "step": 169430 }, { "epoch": 0.0364544, "grad_norm": 0.7607773542404175, "learning_rate": 1.208818747468317e-05, "loss": 2.8117, "step": 169440 }, { "epoch": 0.03648, "grad_norm": 0.7713410258293152, "learning_rate": 1.208739994231677e-05, "loss": 2.8705, "step": 169450 }, { "epoch": 0.0365056, "grad_norm": 0.7519956827163696, "learning_rate": 1.2086612396414135e-05, "loss": 3.0827, "step": 169460 }, { "epoch": 0.0365312, "grad_norm": 0.9111257791519165, "learning_rate": 1.2085824836980371e-05, "loss": 2.9565, "step": 169470 }, { "epoch": 0.0365568, "grad_norm": 0.8600939512252808, "learning_rate": 1.208503726402059e-05, "loss": 2.8924, "step": 169480 }, { "epoch": 0.0365824, "grad_norm": 0.7049847841262817, "learning_rate": 1.2084249677539899e-05, "loss": 2.9241, "step": 169490 }, { "epoch": 0.036608, "grad_norm": 0.8495995998382568, "learning_rate": 1.2083462077543398e-05, "loss": 2.9203, "step": 169500 }, { "epoch": 0.0366336, "grad_norm": 0.8572457432746887, "learning_rate": 1.2082674464036196e-05, "loss": 2.907, "step": 169510 }, { "epoch": 0.0366592, "grad_norm": 0.6541683673858643, "learning_rate": 1.2081886837023414e-05, "loss": 3.1827, "step": 169520 }, { "epoch": 0.0366848, "grad_norm": 0.9101367592811584, "learning_rate": 1.2081099196510142e-05, "loss": 2.968, "step": 169530 }, { "epoch": 0.0367104, "grad_norm": 0.8699093461036682, "learning_rate": 1.2080311542501494e-05, "loss": 3.1277, "step": 169540 }, { "epoch": 0.036736, "grad_norm": 0.7835769653320312, "learning_rate": 1.2079523875002583e-05, "loss": 3.0263, "step": 169550 }, { "epoch": 0.0367616, "grad_norm": 0.7867380976676941, "learning_rate": 1.207873619401851e-05, "loss": 3.0527, "step": 169560 }, { "epoch": 0.0367872, "grad_norm": 0.8139405846595764, "learning_rate": 1.2077948499554385e-05, "loss": 2.9476, "step": 169570 }, { "epoch": 0.0368128, "grad_norm": 1.2005524635314941, "learning_rate": 1.2077160791615316e-05, "loss": 2.9278, "step": 169580 }, { "epoch": 0.0368384, "grad_norm": 1.3324098587036133, "learning_rate": 1.2076373070206415e-05, "loss": 3.1104, "step": 169590 }, { "epoch": 0.036864, "grad_norm": 0.777259349822998, "learning_rate": 1.207558533533278e-05, "loss": 3.0603, "step": 169600 }, { "epoch": 0.0368896, "grad_norm": 0.7484896183013916, "learning_rate": 1.207479758699953e-05, "loss": 3.1499, "step": 169610 }, { "epoch": 0.0369152, "grad_norm": 0.9323729276657104, "learning_rate": 1.207400982521177e-05, "loss": 2.9829, "step": 169620 }, { "epoch": 0.0369408, "grad_norm": 0.7255876064300537, "learning_rate": 1.2073222049974605e-05, "loss": 3.0365, "step": 169630 }, { "epoch": 0.0369664, "grad_norm": 1.0085092782974243, "learning_rate": 1.2072434261293147e-05, "loss": 3.2118, "step": 169640 }, { "epoch": 0.036992, "grad_norm": 0.9167107939720154, "learning_rate": 1.2071646459172504e-05, "loss": 2.9524, "step": 169650 }, { "epoch": 0.0370176, "grad_norm": 0.8403063416481018, "learning_rate": 1.2070858643617785e-05, "loss": 3.0587, "step": 169660 }, { "epoch": 0.0370432, "grad_norm": 0.8006026744842529, "learning_rate": 1.2070070814634094e-05, "loss": 3.0531, "step": 169670 }, { "epoch": 0.0370688, "grad_norm": 0.8956020474433899, "learning_rate": 1.2069282972226545e-05, "loss": 2.9856, "step": 169680 }, { "epoch": 0.0370944, "grad_norm": 0.7348983287811279, "learning_rate": 1.206849511640025e-05, "loss": 2.9776, "step": 169690 }, { "epoch": 0.03712, "grad_norm": 0.7421283721923828, "learning_rate": 1.206770724716031e-05, "loss": 3.0551, "step": 169700 }, { "epoch": 0.0371456, "grad_norm": 0.7376253008842468, "learning_rate": 1.206691936451184e-05, "loss": 3.0448, "step": 169710 }, { "epoch": 0.0371712, "grad_norm": 0.7250006198883057, "learning_rate": 1.2066131468459948e-05, "loss": 3.0542, "step": 169720 }, { "epoch": 0.0371968, "grad_norm": 0.971698522567749, "learning_rate": 1.2065343559009741e-05, "loss": 2.9325, "step": 169730 }, { "epoch": 0.0372224, "grad_norm": 0.951154351234436, "learning_rate": 1.2064555636166328e-05, "loss": 3.1477, "step": 169740 }, { "epoch": 0.037248, "grad_norm": 0.7593223452568054, "learning_rate": 1.2063767699934825e-05, "loss": 2.9094, "step": 169750 }, { "epoch": 0.0372736, "grad_norm": 0.8368145823478699, "learning_rate": 1.2062979750320333e-05, "loss": 3.0567, "step": 169760 }, { "epoch": 0.0372992, "grad_norm": 0.7089905738830566, "learning_rate": 1.2062191787327966e-05, "loss": 2.8393, "step": 169770 }, { "epoch": 0.0373248, "grad_norm": 1.6198550462722778, "learning_rate": 1.2061403810962835e-05, "loss": 3.096, "step": 169780 }, { "epoch": 0.0373504, "grad_norm": 0.7752892971038818, "learning_rate": 1.2060615821230049e-05, "loss": 3.006, "step": 169790 }, { "epoch": 0.037376, "grad_norm": 1.4727050065994263, "learning_rate": 1.2059827818134716e-05, "loss": 2.9955, "step": 169800 }, { "epoch": 0.0374016, "grad_norm": 1.0709779262542725, "learning_rate": 1.2059039801681944e-05, "loss": 2.8931, "step": 169810 }, { "epoch": 0.0374272, "grad_norm": 0.7483014464378357, "learning_rate": 1.205825177187685e-05, "loss": 2.9388, "step": 169820 }, { "epoch": 0.0374528, "grad_norm": 0.8329728841781616, "learning_rate": 1.2057463728724538e-05, "loss": 3.2691, "step": 169830 }, { "epoch": 0.0374784, "grad_norm": 0.7172961831092834, "learning_rate": 1.2056675672230122e-05, "loss": 3.2201, "step": 169840 }, { "epoch": 0.037504, "grad_norm": 0.9224380850791931, "learning_rate": 1.2055887602398713e-05, "loss": 2.9381, "step": 169850 }, { "epoch": 0.0375296, "grad_norm": 0.705219566822052, "learning_rate": 1.2055099519235413e-05, "loss": 2.7871, "step": 169860 }, { "epoch": 0.0375552, "grad_norm": 0.9283183813095093, "learning_rate": 1.2054311422745344e-05, "loss": 3.0413, "step": 169870 }, { "epoch": 0.0375808, "grad_norm": 0.9139036536216736, "learning_rate": 1.205352331293361e-05, "loss": 2.7965, "step": 169880 }, { "epoch": 0.0376064, "grad_norm": 0.7477394342422485, "learning_rate": 1.2052735189805323e-05, "loss": 3.145, "step": 169890 }, { "epoch": 0.037632, "grad_norm": 1.7218071222305298, "learning_rate": 1.2051947053365594e-05, "loss": 3.0525, "step": 169900 }, { "epoch": 0.0376576, "grad_norm": 1.9420957565307617, "learning_rate": 1.2051158903619533e-05, "loss": 3.0801, "step": 169910 }, { "epoch": 0.0376832, "grad_norm": 0.7718296647071838, "learning_rate": 1.2050370740572253e-05, "loss": 2.9584, "step": 169920 }, { "epoch": 0.0377088, "grad_norm": 0.8574616312980652, "learning_rate": 1.2049582564228863e-05, "loss": 3.0433, "step": 169930 }, { "epoch": 0.0377344, "grad_norm": 0.7283958196640015, "learning_rate": 1.2048794374594474e-05, "loss": 2.9256, "step": 169940 }, { "epoch": 0.03776, "grad_norm": 0.8094336986541748, "learning_rate": 1.2048006171674199e-05, "loss": 2.8163, "step": 169950 }, { "epoch": 0.0377856, "grad_norm": 0.7437400221824646, "learning_rate": 1.2047217955473148e-05, "loss": 3.0435, "step": 169960 }, { "epoch": 0.0378112, "grad_norm": 0.7008181214332581, "learning_rate": 1.2046429725996431e-05, "loss": 2.9273, "step": 169970 }, { "epoch": 0.0378368, "grad_norm": 0.7241178750991821, "learning_rate": 1.2045641483249163e-05, "loss": 3.0295, "step": 169980 }, { "epoch": 0.0378624, "grad_norm": 0.7037479877471924, "learning_rate": 1.2044853227236455e-05, "loss": 3.0581, "step": 169990 }, { "epoch": 0.037888, "grad_norm": 0.9512315988540649, "learning_rate": 1.2044064957963415e-05, "loss": 3.1367, "step": 170000 }, { "epoch": 0.0379136, "grad_norm": 0.7728565335273743, "learning_rate": 1.2043276675435156e-05, "loss": 2.9171, "step": 170010 }, { "epoch": 0.0379392, "grad_norm": 0.7334672808647156, "learning_rate": 1.2042488379656793e-05, "loss": 2.9363, "step": 170020 }, { "epoch": 0.0379648, "grad_norm": 0.8345179557800293, "learning_rate": 1.2041700070633432e-05, "loss": 3.0504, "step": 170030 }, { "epoch": 0.0379904, "grad_norm": 0.9070752263069153, "learning_rate": 1.2040911748370196e-05, "loss": 3.1161, "step": 170040 }, { "epoch": 0.038016, "grad_norm": 0.7442013621330261, "learning_rate": 1.2040123412872185e-05, "loss": 2.9888, "step": 170050 }, { "epoch": 0.0380416, "grad_norm": 0.7890026569366455, "learning_rate": 1.2039335064144515e-05, "loss": 2.9461, "step": 170060 }, { "epoch": 0.0380672, "grad_norm": 0.7679912447929382, "learning_rate": 1.20385467021923e-05, "loss": 2.9656, "step": 170070 }, { "epoch": 0.0380928, "grad_norm": 0.8567817211151123, "learning_rate": 1.2037758327020654e-05, "loss": 3.0312, "step": 170080 }, { "epoch": 0.0381184, "grad_norm": 0.7093268632888794, "learning_rate": 1.2036969938634684e-05, "loss": 2.8647, "step": 170090 }, { "epoch": 0.038144, "grad_norm": 0.7246065139770508, "learning_rate": 1.2036181537039507e-05, "loss": 2.9109, "step": 170100 }, { "epoch": 0.0381696, "grad_norm": 0.987937331199646, "learning_rate": 1.2035393122240231e-05, "loss": 2.9197, "step": 170110 }, { "epoch": 0.0381952, "grad_norm": 0.75640469789505, "learning_rate": 1.2034604694241974e-05, "loss": 3.0842, "step": 170120 }, { "epoch": 0.0382208, "grad_norm": 0.9684905409812927, "learning_rate": 1.2033816253049843e-05, "loss": 2.9624, "step": 170130 }, { "epoch": 0.0382464, "grad_norm": 0.8756424784660339, "learning_rate": 1.2033027798668959e-05, "loss": 2.9735, "step": 170140 }, { "epoch": 0.038272, "grad_norm": 0.7019726037979126, "learning_rate": 1.2032239331104427e-05, "loss": 3.0869, "step": 170150 }, { "epoch": 0.0382976, "grad_norm": 2.611459732055664, "learning_rate": 1.2031450850361363e-05, "loss": 2.9679, "step": 170160 }, { "epoch": 0.0383232, "grad_norm": 0.737953782081604, "learning_rate": 1.2030662356444882e-05, "loss": 2.8675, "step": 170170 }, { "epoch": 0.0383488, "grad_norm": 1.6628024578094482, "learning_rate": 1.2029873849360097e-05, "loss": 3.0305, "step": 170180 }, { "epoch": 0.0383744, "grad_norm": 0.8914567828178406, "learning_rate": 1.2029085329112116e-05, "loss": 2.8103, "step": 170190 }, { "epoch": 0.0384, "grad_norm": 0.7344534993171692, "learning_rate": 1.2028296795706056e-05, "loss": 3.0173, "step": 170200 }, { "epoch": 0.0384256, "grad_norm": 0.8665632605552673, "learning_rate": 1.2027508249147032e-05, "loss": 3.1581, "step": 170210 }, { "epoch": 0.0384512, "grad_norm": 0.7825149297714233, "learning_rate": 1.2026719689440153e-05, "loss": 2.9272, "step": 170220 }, { "epoch": 0.0384768, "grad_norm": 0.8384028077125549, "learning_rate": 1.2025931116590537e-05, "loss": 3.2262, "step": 170230 }, { "epoch": 0.0385024, "grad_norm": 0.8110758066177368, "learning_rate": 1.2025142530603298e-05, "loss": 3.0011, "step": 170240 }, { "epoch": 0.038528, "grad_norm": 0.7813648581504822, "learning_rate": 1.2024353931483546e-05, "loss": 2.9901, "step": 170250 }, { "epoch": 0.0385536, "grad_norm": 0.9959779381752014, "learning_rate": 1.2023565319236396e-05, "loss": 3.0827, "step": 170260 }, { "epoch": 0.0385792, "grad_norm": 0.7790793776512146, "learning_rate": 1.2022776693866965e-05, "loss": 3.0622, "step": 170270 }, { "epoch": 0.0386048, "grad_norm": 0.7582787871360779, "learning_rate": 1.2021988055380364e-05, "loss": 2.8561, "step": 170280 }, { "epoch": 0.0386304, "grad_norm": 0.8321682810783386, "learning_rate": 1.2021199403781705e-05, "loss": 2.8002, "step": 170290 }, { "epoch": 0.038656, "grad_norm": 0.7367854118347168, "learning_rate": 1.2020410739076108e-05, "loss": 2.7898, "step": 170300 }, { "epoch": 0.0386816, "grad_norm": 0.8906142115592957, "learning_rate": 1.2019622061268686e-05, "loss": 2.9719, "step": 170310 }, { "epoch": 0.0387072, "grad_norm": 0.7848893404006958, "learning_rate": 1.201883337036455e-05, "loss": 2.9616, "step": 170320 }, { "epoch": 0.0387328, "grad_norm": 1.2360656261444092, "learning_rate": 1.2018044666368813e-05, "loss": 2.6894, "step": 170330 }, { "epoch": 0.0387584, "grad_norm": 0.8888070583343506, "learning_rate": 1.2017255949286596e-05, "loss": 2.9121, "step": 170340 }, { "epoch": 0.038784, "grad_norm": 0.7599682211875916, "learning_rate": 1.2016467219123011e-05, "loss": 2.928, "step": 170350 }, { "epoch": 0.0388096, "grad_norm": 0.782758891582489, "learning_rate": 1.201567847588317e-05, "loss": 2.8582, "step": 170360 }, { "epoch": 0.0388352, "grad_norm": 0.7606122493743896, "learning_rate": 1.201488971957219e-05, "loss": 2.8228, "step": 170370 }, { "epoch": 0.0388608, "grad_norm": 0.7503918409347534, "learning_rate": 1.2014100950195186e-05, "loss": 2.724, "step": 170380 }, { "epoch": 0.0388864, "grad_norm": 0.6549776792526245, "learning_rate": 1.2013312167757272e-05, "loss": 2.8935, "step": 170390 }, { "epoch": 0.038912, "grad_norm": 1.1020742654800415, "learning_rate": 1.2012523372263566e-05, "loss": 3.0572, "step": 170400 }, { "epoch": 0.0389376, "grad_norm": 0.7082906365394592, "learning_rate": 1.201173456371918e-05, "loss": 2.7453, "step": 170410 }, { "epoch": 0.0389632, "grad_norm": 0.8130098581314087, "learning_rate": 1.2010945742129228e-05, "loss": 2.7746, "step": 170420 }, { "epoch": 0.0389888, "grad_norm": 0.7521194219589233, "learning_rate": 1.2010156907498826e-05, "loss": 2.8959, "step": 170430 }, { "epoch": 0.0390144, "grad_norm": 0.6995358467102051, "learning_rate": 1.2009368059833098e-05, "loss": 2.9187, "step": 170440 }, { "epoch": 0.03904, "grad_norm": 0.8992771506309509, "learning_rate": 1.2008579199137144e-05, "loss": 2.7408, "step": 170450 }, { "epoch": 0.0390656, "grad_norm": 0.8820354342460632, "learning_rate": 1.2007790325416094e-05, "loss": 2.8094, "step": 170460 }, { "epoch": 0.0390912, "grad_norm": 0.7473173141479492, "learning_rate": 1.2007001438675053e-05, "loss": 2.6955, "step": 170470 }, { "epoch": 0.0391168, "grad_norm": 0.7535742521286011, "learning_rate": 1.2006212538919142e-05, "loss": 2.6266, "step": 170480 }, { "epoch": 0.0391424, "grad_norm": 0.6989601254463196, "learning_rate": 1.2005423626153473e-05, "loss": 2.6287, "step": 170490 }, { "epoch": 0.039168, "grad_norm": 0.6788511872291565, "learning_rate": 1.2004634700383168e-05, "loss": 2.6323, "step": 170500 }, { "epoch": 0.0391936, "grad_norm": 0.7860579490661621, "learning_rate": 1.2003845761613344e-05, "loss": 3.0615, "step": 170510 }, { "epoch": 0.0392192, "grad_norm": 0.7929598689079285, "learning_rate": 1.2003056809849107e-05, "loss": 2.9377, "step": 170520 }, { "epoch": 0.0392448, "grad_norm": 0.9325815439224243, "learning_rate": 1.2002267845095581e-05, "loss": 2.7012, "step": 170530 }, { "epoch": 0.0392704, "grad_norm": 0.8407998085021973, "learning_rate": 1.200147886735788e-05, "loss": 2.8626, "step": 170540 }, { "epoch": 0.039296, "grad_norm": 0.9271493554115295, "learning_rate": 1.2000689876641118e-05, "loss": 3.0923, "step": 170550 }, { "epoch": 0.0393216, "grad_norm": 0.705650269985199, "learning_rate": 1.1999900872950416e-05, "loss": 2.6876, "step": 170560 }, { "epoch": 0.0393472, "grad_norm": 0.6759563088417053, "learning_rate": 1.199911185629089e-05, "loss": 2.6437, "step": 170570 }, { "epoch": 0.0393728, "grad_norm": 0.7393131256103516, "learning_rate": 1.1998322826667653e-05, "loss": 2.5366, "step": 170580 }, { "epoch": 0.0393984, "grad_norm": 0.6697709560394287, "learning_rate": 1.1997533784085824e-05, "loss": 2.7582, "step": 170590 }, { "epoch": 0.039424, "grad_norm": 0.6735674738883972, "learning_rate": 1.1996744728550518e-05, "loss": 2.8796, "step": 170600 }, { "epoch": 0.0394496, "grad_norm": 0.8138843178749084, "learning_rate": 1.1995955660066855e-05, "loss": 2.5305, "step": 170610 }, { "epoch": 0.0394752, "grad_norm": 0.7983009219169617, "learning_rate": 1.199516657863995e-05, "loss": 2.7523, "step": 170620 }, { "epoch": 0.0395008, "grad_norm": 0.6764093637466431, "learning_rate": 1.199437748427492e-05, "loss": 2.7574, "step": 170630 }, { "epoch": 0.0395264, "grad_norm": 0.8142428398132324, "learning_rate": 1.1993588376976879e-05, "loss": 2.7679, "step": 170640 }, { "epoch": 0.039552, "grad_norm": 0.9411346316337585, "learning_rate": 1.199279925675095e-05, "loss": 2.648, "step": 170650 }, { "epoch": 0.0395776, "grad_norm": 0.801419734954834, "learning_rate": 1.1992010123602246e-05, "loss": 2.9503, "step": 170660 }, { "epoch": 0.0396032, "grad_norm": 0.7982240319252014, "learning_rate": 1.1991220977535889e-05, "loss": 2.9408, "step": 170670 }, { "epoch": 0.0396288, "grad_norm": 0.7381760478019714, "learning_rate": 1.1990431818556992e-05, "loss": 2.8963, "step": 170680 }, { "epoch": 0.0396544, "grad_norm": 0.9208093881607056, "learning_rate": 1.1989642646670669e-05, "loss": 2.8264, "step": 170690 }, { "epoch": 0.03968, "grad_norm": 0.8163942098617554, "learning_rate": 1.1988853461882046e-05, "loss": 2.6311, "step": 170700 }, { "epoch": 0.0397056, "grad_norm": 0.7548791170120239, "learning_rate": 1.198806426419624e-05, "loss": 2.9172, "step": 170710 }, { "epoch": 0.0397312, "grad_norm": 0.7684751152992249, "learning_rate": 1.198727505361836e-05, "loss": 3.0537, "step": 170720 }, { "epoch": 0.0397568, "grad_norm": 0.8073747158050537, "learning_rate": 1.198648583015353e-05, "loss": 2.9237, "step": 170730 }, { "epoch": 0.0397824, "grad_norm": 0.7994379997253418, "learning_rate": 1.1985696593806871e-05, "loss": 3.0454, "step": 170740 }, { "epoch": 0.039808, "grad_norm": 0.7774279117584229, "learning_rate": 1.1984907344583495e-05, "loss": 2.7303, "step": 170750 }, { "epoch": 0.0398336, "grad_norm": 0.812553346157074, "learning_rate": 1.1984118082488522e-05, "loss": 3.0745, "step": 170760 }, { "epoch": 0.0398592, "grad_norm": 0.6986249685287476, "learning_rate": 1.1983328807527072e-05, "loss": 2.9517, "step": 170770 }, { "epoch": 0.0398848, "grad_norm": 1.0279380083084106, "learning_rate": 1.1982539519704262e-05, "loss": 2.578, "step": 170780 }, { "epoch": 0.0399104, "grad_norm": 0.7795369029045105, "learning_rate": 1.1981750219025208e-05, "loss": 2.7107, "step": 170790 }, { "epoch": 0.039936, "grad_norm": 0.7000385522842407, "learning_rate": 1.198096090549503e-05, "loss": 2.8063, "step": 170800 }, { "epoch": 0.0399616, "grad_norm": 0.8169569373130798, "learning_rate": 1.1980171579118852e-05, "loss": 2.7675, "step": 170810 }, { "epoch": 0.0399872, "grad_norm": 0.824174165725708, "learning_rate": 1.1979382239901782e-05, "loss": 2.9049, "step": 170820 }, { "epoch": 0.0400128, "grad_norm": 0.6942928433418274, "learning_rate": 1.1978592887848948e-05, "loss": 2.7876, "step": 170830 }, { "epoch": 0.0400384, "grad_norm": 0.7556196451187134, "learning_rate": 1.1977803522965464e-05, "loss": 2.7641, "step": 170840 }, { "epoch": 0.040064, "grad_norm": 0.7661706209182739, "learning_rate": 1.1977014145256448e-05, "loss": 2.3208, "step": 170850 }, { "epoch": 0.0400896, "grad_norm": 0.8032887578010559, "learning_rate": 1.1976224754727022e-05, "loss": 3.0011, "step": 170860 }, { "epoch": 0.0401152, "grad_norm": 0.7368884682655334, "learning_rate": 1.1975435351382306e-05, "loss": 2.8949, "step": 170870 }, { "epoch": 0.0401408, "grad_norm": 0.7556254267692566, "learning_rate": 1.1974645935227414e-05, "loss": 2.9247, "step": 170880 }, { "epoch": 0.0401664, "grad_norm": 0.7233942151069641, "learning_rate": 1.1973856506267468e-05, "loss": 2.9853, "step": 170890 }, { "epoch": 0.040192, "grad_norm": 0.7662217020988464, "learning_rate": 1.1973067064507587e-05, "loss": 2.8882, "step": 170900 }, { "epoch": 2.56e-05, "grad_norm": 0.7664540410041809, "learning_rate": 1.1972277609952888e-05, "loss": 2.252, "step": 170910 }, { "epoch": 5.12e-05, "grad_norm": 0.8032822012901306, "learning_rate": 1.1971488142608496e-05, "loss": 2.816, "step": 170920 }, { "epoch": 7.68e-05, "grad_norm": 0.7084815502166748, "learning_rate": 1.197069866247953e-05, "loss": 2.6903, "step": 170930 }, { "epoch": 0.0001024, "grad_norm": 0.7833142876625061, "learning_rate": 1.1969909169571104e-05, "loss": 2.5661, "step": 170940 }, { "epoch": 0.000128, "grad_norm": 0.7155617475509644, "learning_rate": 1.196911966388834e-05, "loss": 2.5149, "step": 170950 }, { "epoch": 0.0001536, "grad_norm": 0.735083281993866, "learning_rate": 1.1968330145436357e-05, "loss": 2.776, "step": 170960 }, { "epoch": 0.0001792, "grad_norm": 0.7438260316848755, "learning_rate": 1.1967540614220281e-05, "loss": 3.0402, "step": 170970 }, { "epoch": 0.0002048, "grad_norm": 0.8191271424293518, "learning_rate": 1.1966751070245225e-05, "loss": 2.9096, "step": 170980 }, { "epoch": 0.0002304, "grad_norm": 0.7492700219154358, "learning_rate": 1.1965961513516306e-05, "loss": 2.8298, "step": 170990 }, { "epoch": 0.000256, "grad_norm": 0.7985707521438599, "learning_rate": 1.1965171944038656e-05, "loss": 2.8778, "step": 171000 }, { "epoch": 0.0002816, "grad_norm": 0.7130568027496338, "learning_rate": 1.1964382361817383e-05, "loss": 2.7673, "step": 171010 }, { "epoch": 0.0003072, "grad_norm": 0.8161627650260925, "learning_rate": 1.1963592766857614e-05, "loss": 2.8744, "step": 171020 }, { "epoch": 0.0003328, "grad_norm": 0.7219778895378113, "learning_rate": 1.196280315916447e-05, "loss": 2.7369, "step": 171030 }, { "epoch": 0.0003584, "grad_norm": 0.8409712314605713, "learning_rate": 1.1962013538743069e-05, "loss": 3.1385, "step": 171040 }, { "epoch": 0.000384, "grad_norm": 0.8277954459190369, "learning_rate": 1.196122390559853e-05, "loss": 2.8302, "step": 171050 }, { "epoch": 0.0004096, "grad_norm": 0.7638432383537292, "learning_rate": 1.1960434259735976e-05, "loss": 2.748, "step": 171060 }, { "epoch": 0.0004352, "grad_norm": 0.815001368522644, "learning_rate": 1.1959644601160526e-05, "loss": 2.9901, "step": 171070 }, { "epoch": 0.0004608, "grad_norm": 0.6732355952262878, "learning_rate": 1.1958854929877302e-05, "loss": 2.9204, "step": 171080 }, { "epoch": 0.0004864, "grad_norm": 1.3651931285858154, "learning_rate": 1.1958065245891424e-05, "loss": 3.1199, "step": 171090 }, { "epoch": 0.000512, "grad_norm": 0.7957640886306763, "learning_rate": 1.1957275549208014e-05, "loss": 2.8319, "step": 171100 }, { "epoch": 0.0005376, "grad_norm": 0.8462172746658325, "learning_rate": 1.1956485839832192e-05, "loss": 2.7934, "step": 171110 }, { "epoch": 0.0005632, "grad_norm": 0.8090543746948242, "learning_rate": 1.1955696117769078e-05, "loss": 3.0069, "step": 171120 }, { "epoch": 0.0005888, "grad_norm": 0.7732002139091492, "learning_rate": 1.1954906383023799e-05, "loss": 2.7486, "step": 171130 }, { "epoch": 0.0006144, "grad_norm": 0.7003718018531799, "learning_rate": 1.1954116635601467e-05, "loss": 2.6313, "step": 171140 }, { "epoch": 0.00064, "grad_norm": 0.7633545398712158, "learning_rate": 1.1953326875507209e-05, "loss": 2.8409, "step": 171150 }, { "epoch": 0.0006656, "grad_norm": 0.7032375335693359, "learning_rate": 1.1952537102746144e-05, "loss": 2.9891, "step": 171160 }, { "epoch": 0.0006912, "grad_norm": 0.6876550912857056, "learning_rate": 1.1951747317323397e-05, "loss": 2.7853, "step": 171170 }, { "epoch": 0.0007168, "grad_norm": 1.5311959981918335, "learning_rate": 1.1950957519244087e-05, "loss": 3.1214, "step": 171180 }, { "epoch": 0.0007424, "grad_norm": 2.5446228981018066, "learning_rate": 1.1950167708513334e-05, "loss": 2.7739, "step": 171190 }, { "epoch": 0.000768, "grad_norm": 0.7492824792861938, "learning_rate": 1.1949377885136267e-05, "loss": 2.6355, "step": 171200 }, { "epoch": 0.0007936, "grad_norm": 0.7420294880867004, "learning_rate": 1.1948588049117997e-05, "loss": 2.745, "step": 171210 }, { "epoch": 0.0008192, "grad_norm": 0.7748954892158508, "learning_rate": 1.1947798200463654e-05, "loss": 2.8012, "step": 171220 }, { "epoch": 0.0008448, "grad_norm": 0.7283935546875, "learning_rate": 1.1947008339178358e-05, "loss": 2.8857, "step": 171230 }, { "epoch": 0.0008704, "grad_norm": 0.8800570368766785, "learning_rate": 1.194621846526723e-05, "loss": 2.7759, "step": 171240 }, { "epoch": 0.000896, "grad_norm": 0.7479312419891357, "learning_rate": 1.1945428578735387e-05, "loss": 2.8952, "step": 171250 }, { "epoch": 0.0009216, "grad_norm": 1.8012471199035645, "learning_rate": 1.1944638679587964e-05, "loss": 2.8589, "step": 171260 }, { "epoch": 0.0009472, "grad_norm": 0.8192050457000732, "learning_rate": 1.1943848767830073e-05, "loss": 2.729, "step": 171270 }, { "epoch": 0.0009728, "grad_norm": 0.6737633347511292, "learning_rate": 1.194305884346684e-05, "loss": 2.9485, "step": 171280 }, { "epoch": 0.0009984, "grad_norm": 0.8112813830375671, "learning_rate": 1.1942268906503385e-05, "loss": 2.8685, "step": 171290 }, { "epoch": 0.001024, "grad_norm": 0.7871214747428894, "learning_rate": 1.1941478956944836e-05, "loss": 2.8031, "step": 171300 }, { "epoch": 0.0010496, "grad_norm": 0.690331757068634, "learning_rate": 1.194068899479631e-05, "loss": 2.884, "step": 171310 }, { "epoch": 0.0010752, "grad_norm": 0.744398832321167, "learning_rate": 1.193989902006293e-05, "loss": 2.8139, "step": 171320 }, { "epoch": 0.0011008, "grad_norm": 0.8241642117500305, "learning_rate": 1.1939109032749823e-05, "loss": 2.7968, "step": 171330 }, { "epoch": 0.0011264, "grad_norm": 0.7524031400680542, "learning_rate": 1.1938319032862106e-05, "loss": 2.8003, "step": 171340 }, { "epoch": 0.001152, "grad_norm": 0.8452348709106445, "learning_rate": 1.1937529020404908e-05, "loss": 2.8243, "step": 171350 }, { "epoch": 0.0011776, "grad_norm": 0.7466590404510498, "learning_rate": 1.1936738995383347e-05, "loss": 2.7327, "step": 171360 }, { "epoch": 0.0012032, "grad_norm": 0.7297242283821106, "learning_rate": 1.193594895780255e-05, "loss": 2.9679, "step": 171370 }, { "epoch": 0.0012288, "grad_norm": 0.8111708164215088, "learning_rate": 1.1935158907667639e-05, "loss": 2.962, "step": 171380 }, { "epoch": 0.0012544, "grad_norm": 0.786648690700531, "learning_rate": 1.1934368844983735e-05, "loss": 2.8604, "step": 171390 }, { "epoch": 0.00128, "grad_norm": 0.7705619931221008, "learning_rate": 1.1933578769755967e-05, "loss": 2.9079, "step": 171400 }, { "epoch": 0.0013056, "grad_norm": 0.8190808892250061, "learning_rate": 1.193278868198945e-05, "loss": 2.8515, "step": 171410 }, { "epoch": 0.0013312, "grad_norm": 0.7216054201126099, "learning_rate": 1.1931998581689316e-05, "loss": 2.8174, "step": 171420 }, { "epoch": 0.0013568, "grad_norm": 0.7908394932746887, "learning_rate": 1.193120846886068e-05, "loss": 2.5298, "step": 171430 }, { "epoch": 0.0013824, "grad_norm": 0.7377322316169739, "learning_rate": 1.1930418343508673e-05, "loss": 3.0183, "step": 171440 }, { "epoch": 0.001408, "grad_norm": 0.8118821978569031, "learning_rate": 1.1929628205638415e-05, "loss": 2.8737, "step": 171450 }, { "epoch": 0.0014336, "grad_norm": 0.6975714564323425, "learning_rate": 1.1928838055255035e-05, "loss": 3.0207, "step": 171460 }, { "epoch": 0.0014592, "grad_norm": 0.8610931634902954, "learning_rate": 1.192804789236365e-05, "loss": 2.8238, "step": 171470 }, { "epoch": 0.0014848, "grad_norm": 0.7803717851638794, "learning_rate": 1.1927257716969386e-05, "loss": 2.5684, "step": 171480 }, { "epoch": 0.0015104, "grad_norm": 0.7624993920326233, "learning_rate": 1.1926467529077368e-05, "loss": 2.4463, "step": 171490 }, { "epoch": 0.001536, "grad_norm": 0.954696774482727, "learning_rate": 1.1925677328692721e-05, "loss": 2.8078, "step": 171500 }, { "epoch": 0.0015616, "grad_norm": 0.6522114872932434, "learning_rate": 1.1924887115820566e-05, "loss": 2.885, "step": 171510 }, { "epoch": 0.0015872, "grad_norm": 0.7744719982147217, "learning_rate": 1.1924096890466033e-05, "loss": 2.8961, "step": 171520 }, { "epoch": 0.0016128, "grad_norm": 0.7355860471725464, "learning_rate": 1.192330665263424e-05, "loss": 2.6983, "step": 171530 }, { "epoch": 0.0016384, "grad_norm": 0.8321810960769653, "learning_rate": 1.1922516402330318e-05, "loss": 2.788, "step": 171540 }, { "epoch": 0.001664, "grad_norm": 0.7513644099235535, "learning_rate": 1.1921726139559384e-05, "loss": 2.6743, "step": 171550 }, { "epoch": 0.0016896, "grad_norm": 0.7595493197441101, "learning_rate": 1.192093586432657e-05, "loss": 2.9321, "step": 171560 }, { "epoch": 0.0017152, "grad_norm": 0.7797627449035645, "learning_rate": 1.1920145576636997e-05, "loss": 2.821, "step": 171570 }, { "epoch": 0.0017408, "grad_norm": 0.8071648478507996, "learning_rate": 1.191935527649579e-05, "loss": 2.9667, "step": 171580 }, { "epoch": 0.0017664, "grad_norm": 0.8219982981681824, "learning_rate": 1.1918564963908072e-05, "loss": 2.8297, "step": 171590 }, { "epoch": 0.001792, "grad_norm": 0.9543573260307312, "learning_rate": 1.1917774638878971e-05, "loss": 2.6812, "step": 171600 }, { "epoch": 0.0018176, "grad_norm": 0.7279067635536194, "learning_rate": 1.1916984301413612e-05, "loss": 2.9357, "step": 171610 }, { "epoch": 0.0018432, "grad_norm": 0.8803426623344421, "learning_rate": 1.1916193951517121e-05, "loss": 2.533, "step": 171620 }, { "epoch": 0.0018688, "grad_norm": 0.6604522466659546, "learning_rate": 1.1915403589194618e-05, "loss": 2.6139, "step": 171630 }, { "epoch": 0.0018944, "grad_norm": 0.8878922462463379, "learning_rate": 1.1914613214451234e-05, "loss": 2.9249, "step": 171640 }, { "epoch": 0.00192, "grad_norm": 0.7764707207679749, "learning_rate": 1.1913822827292092e-05, "loss": 2.7767, "step": 171650 }, { "epoch": 0.0019456, "grad_norm": 0.6884385347366333, "learning_rate": 1.1913032427722318e-05, "loss": 2.8822, "step": 171660 }, { "epoch": 0.0019712, "grad_norm": 0.7968587875366211, "learning_rate": 1.1912242015747038e-05, "loss": 2.8066, "step": 171670 }, { "epoch": 0.0019968, "grad_norm": 1.4527106285095215, "learning_rate": 1.1911451591371372e-05, "loss": 3.0222, "step": 171680 }, { "epoch": 0.0020224, "grad_norm": 0.7410345673561096, "learning_rate": 1.1910661154600453e-05, "loss": 2.9335, "step": 171690 }, { "epoch": 0.002048, "grad_norm": 0.8445099592208862, "learning_rate": 1.1909870705439405e-05, "loss": 2.9025, "step": 171700 }, { "epoch": 0.0020736, "grad_norm": 0.6841070652008057, "learning_rate": 1.1909080243893352e-05, "loss": 2.6736, "step": 171710 }, { "epoch": 0.0020992, "grad_norm": 0.7898637056350708, "learning_rate": 1.1908289769967426e-05, "loss": 2.9927, "step": 171720 }, { "epoch": 0.0021248, "grad_norm": 1.1392992734909058, "learning_rate": 1.1907499283666741e-05, "loss": 3.0052, "step": 171730 }, { "epoch": 0.0021504, "grad_norm": 0.7461122274398804, "learning_rate": 1.1906708784996433e-05, "loss": 2.9045, "step": 171740 }, { "epoch": 0.002176, "grad_norm": 0.7620013952255249, "learning_rate": 1.1905918273961626e-05, "loss": 2.8738, "step": 171750 }, { "epoch": 0.0022016, "grad_norm": 0.7322381734848022, "learning_rate": 1.1905127750567444e-05, "loss": 2.9011, "step": 171760 }, { "epoch": 0.0022272, "grad_norm": 0.6940298080444336, "learning_rate": 1.1904337214819016e-05, "loss": 2.9137, "step": 171770 }, { "epoch": 0.0022528, "grad_norm": 0.6345421671867371, "learning_rate": 1.1903546666721464e-05, "loss": 2.6978, "step": 171780 }, { "epoch": 0.0022784, "grad_norm": 0.8571527600288391, "learning_rate": 1.1902756106279923e-05, "loss": 2.7155, "step": 171790 }, { "epoch": 0.002304, "grad_norm": 0.7304227352142334, "learning_rate": 1.190196553349951e-05, "loss": 2.5814, "step": 171800 }, { "epoch": 0.0023296, "grad_norm": 0.9717293977737427, "learning_rate": 1.1901174948385358e-05, "loss": 3.0249, "step": 171810 }, { "epoch": 0.0023552, "grad_norm": 0.6659586429595947, "learning_rate": 1.1900384350942592e-05, "loss": 2.5222, "step": 171820 }, { "epoch": 0.0023808, "grad_norm": 0.7488079071044922, "learning_rate": 1.1899593741176339e-05, "loss": 2.7362, "step": 171830 }, { "epoch": 0.0024064, "grad_norm": 0.8189420700073242, "learning_rate": 1.1898803119091725e-05, "loss": 2.9541, "step": 171840 }, { "epoch": 0.002432, "grad_norm": 0.7809364199638367, "learning_rate": 1.1898012484693877e-05, "loss": 2.8165, "step": 171850 }, { "epoch": 0.0024576, "grad_norm": 0.8634269833564758, "learning_rate": 1.1897221837987922e-05, "loss": 2.7976, "step": 171860 }, { "epoch": 0.0024832, "grad_norm": 0.7789600491523743, "learning_rate": 1.1896431178978987e-05, "loss": 2.859, "step": 171870 }, { "epoch": 0.0025088, "grad_norm": 0.8836341500282288, "learning_rate": 1.1895640507672202e-05, "loss": 2.9429, "step": 171880 }, { "epoch": 0.0025344, "grad_norm": 0.7269927263259888, "learning_rate": 1.1894849824072692e-05, "loss": 2.7702, "step": 171890 }, { "epoch": 0.00256, "grad_norm": 0.880806565284729, "learning_rate": 1.1894059128185582e-05, "loss": 2.7006, "step": 171900 }, { "epoch": 0.0025856, "grad_norm": 0.8217308521270752, "learning_rate": 1.1893268420016004e-05, "loss": 2.9178, "step": 171910 }, { "epoch": 0.0026112, "grad_norm": 0.7170425653457642, "learning_rate": 1.1892477699569086e-05, "loss": 2.9649, "step": 171920 }, { "epoch": 0.0026368, "grad_norm": 0.787686288356781, "learning_rate": 1.1891686966849947e-05, "loss": 3.0793, "step": 171930 }, { "epoch": 0.0026624, "grad_norm": 0.7923747301101685, "learning_rate": 1.1890896221863723e-05, "loss": 3.0419, "step": 171940 }, { "epoch": 0.002688, "grad_norm": 0.7303363680839539, "learning_rate": 1.189010546461554e-05, "loss": 2.8463, "step": 171950 }, { "epoch": 0.0027136, "grad_norm": 0.7636951804161072, "learning_rate": 1.1889314695110525e-05, "loss": 2.8958, "step": 171960 }, { "epoch": 0.0027392, "grad_norm": 0.8608322143554688, "learning_rate": 1.1888523913353809e-05, "loss": 3.0063, "step": 171970 }, { "epoch": 0.0027648, "grad_norm": 0.7841907739639282, "learning_rate": 1.1887733119350513e-05, "loss": 3.0128, "step": 171980 }, { "epoch": 0.0027904, "grad_norm": 0.7047268152236938, "learning_rate": 1.1886942313105774e-05, "loss": 2.952, "step": 171990 }, { "epoch": 0.002816, "grad_norm": 0.967607855796814, "learning_rate": 1.1886151494624714e-05, "loss": 3.0347, "step": 172000 }, { "epoch": 0.0028416, "grad_norm": 0.7373393774032593, "learning_rate": 1.1885360663912463e-05, "loss": 2.981, "step": 172010 }, { "epoch": 0.0028672, "grad_norm": 0.7582761645317078, "learning_rate": 1.1884569820974148e-05, "loss": 2.8765, "step": 172020 }, { "epoch": 0.0028928, "grad_norm": 0.7557874321937561, "learning_rate": 1.1883778965814896e-05, "loss": 2.9663, "step": 172030 }, { "epoch": 0.0029184, "grad_norm": 0.746644914150238, "learning_rate": 1.188298809843984e-05, "loss": 2.9379, "step": 172040 }, { "epoch": 0.002944, "grad_norm": 0.7055810689926147, "learning_rate": 1.1882197218854111e-05, "loss": 2.8278, "step": 172050 }, { "epoch": 0.0029696, "grad_norm": 0.7572097778320312, "learning_rate": 1.1881406327062827e-05, "loss": 2.9796, "step": 172060 }, { "epoch": 0.0029952, "grad_norm": 0.8880619406700134, "learning_rate": 1.1880615423071126e-05, "loss": 3.0136, "step": 172070 }, { "epoch": 0.0030208, "grad_norm": 0.6812332272529602, "learning_rate": 1.1879824506884135e-05, "loss": 2.9629, "step": 172080 }, { "epoch": 0.0030464, "grad_norm": 0.8327699303627014, "learning_rate": 1.1879033578506982e-05, "loss": 2.882, "step": 172090 }, { "epoch": 0.003072, "grad_norm": 0.7286043167114258, "learning_rate": 1.1878242637944793e-05, "loss": 2.9657, "step": 172100 }, { "epoch": 0.0030976, "grad_norm": 0.6994619965553284, "learning_rate": 1.1877451685202699e-05, "loss": 2.8815, "step": 172110 }, { "epoch": 0.0031232, "grad_norm": 0.7607470154762268, "learning_rate": 1.1876660720285834e-05, "loss": 2.9138, "step": 172120 }, { "epoch": 0.0031488, "grad_norm": 0.7632052898406982, "learning_rate": 1.1875869743199318e-05, "loss": 2.9022, "step": 172130 }, { "epoch": 0.0031744, "grad_norm": 0.717603325843811, "learning_rate": 1.1875078753948288e-05, "loss": 3.0539, "step": 172140 }, { "epoch": 0.0032, "grad_norm": 0.7088153958320618, "learning_rate": 1.1874287752537872e-05, "loss": 2.7852, "step": 172150 }, { "epoch": 0.0032256, "grad_norm": 0.8273375630378723, "learning_rate": 1.1873496738973197e-05, "loss": 3.1405, "step": 172160 }, { "epoch": 0.0032512, "grad_norm": 0.6810125112533569, "learning_rate": 1.1872705713259392e-05, "loss": 2.981, "step": 172170 }, { "epoch": 0.0032768, "grad_norm": 0.7331036329269409, "learning_rate": 1.1871914675401587e-05, "loss": 2.8126, "step": 172180 }, { "epoch": 0.0033024, "grad_norm": 0.8281006217002869, "learning_rate": 1.1871123625404917e-05, "loss": 2.9917, "step": 172190 }, { "epoch": 0.003328, "grad_norm": 0.8248384594917297, "learning_rate": 1.1870332563274504e-05, "loss": 3.2972, "step": 172200 }, { "epoch": 0.0033536, "grad_norm": 0.8269795775413513, "learning_rate": 1.1869541489015481e-05, "loss": 2.9849, "step": 172210 }, { "epoch": 0.0033792, "grad_norm": 0.7958652377128601, "learning_rate": 1.1868750402632982e-05, "loss": 3.0378, "step": 172220 }, { "epoch": 0.0034048, "grad_norm": 0.7241684794425964, "learning_rate": 1.186795930413213e-05, "loss": 2.6833, "step": 172230 }, { "epoch": 0.0034304, "grad_norm": 0.7193166613578796, "learning_rate": 1.186716819351806e-05, "loss": 2.8501, "step": 172240 }, { "epoch": 0.003456, "grad_norm": 0.7326870560646057, "learning_rate": 1.1866377070795902e-05, "loss": 3.064, "step": 172250 }, { "epoch": 0.0034816, "grad_norm": 0.7779173254966736, "learning_rate": 1.1865585935970783e-05, "loss": 3.1548, "step": 172260 }, { "epoch": 0.0035072, "grad_norm": 0.746058464050293, "learning_rate": 1.1864794789047835e-05, "loss": 2.8141, "step": 172270 }, { "epoch": 0.0035328, "grad_norm": 0.7740813493728638, "learning_rate": 1.1864003630032188e-05, "loss": 2.9142, "step": 172280 }, { "epoch": 0.0035584, "grad_norm": 0.9892716407775879, "learning_rate": 1.1863212458928973e-05, "loss": 2.8492, "step": 172290 }, { "epoch": 0.003584, "grad_norm": 0.9412796497344971, "learning_rate": 1.1862421275743318e-05, "loss": 3.1551, "step": 172300 }, { "epoch": 0.0036096, "grad_norm": 0.8756728768348694, "learning_rate": 1.186163008048036e-05, "loss": 3.1224, "step": 172310 }, { "epoch": 0.0036352, "grad_norm": 0.8105581998825073, "learning_rate": 1.1860838873145225e-05, "loss": 3.0945, "step": 172320 }, { "epoch": 0.0036608, "grad_norm": 0.8158262968063354, "learning_rate": 1.1860047653743046e-05, "loss": 3.0464, "step": 172330 }, { "epoch": 0.0036864, "grad_norm": 0.8351215720176697, "learning_rate": 1.1859256422278948e-05, "loss": 3.0741, "step": 172340 }, { "epoch": 0.003712, "grad_norm": 0.747581958770752, "learning_rate": 1.185846517875807e-05, "loss": 3.1227, "step": 172350 }, { "epoch": 0.0037376, "grad_norm": 0.9651317000389099, "learning_rate": 1.1857673923185537e-05, "loss": 2.9882, "step": 172360 }, { "epoch": 0.0037632, "grad_norm": 0.7625432014465332, "learning_rate": 1.1856882655566481e-05, "loss": 2.8249, "step": 172370 }, { "epoch": 0.0037888, "grad_norm": 0.8869432210922241, "learning_rate": 1.1856091375906038e-05, "loss": 2.9515, "step": 172380 }, { "epoch": 0.0038144, "grad_norm": 0.8484205603599548, "learning_rate": 1.1855300084209334e-05, "loss": 2.9665, "step": 172390 }, { "epoch": 0.00384, "grad_norm": 0.8366653919219971, "learning_rate": 1.1854508780481502e-05, "loss": 2.9722, "step": 172400 }, { "epoch": 0.0038656, "grad_norm": 1.204750895500183, "learning_rate": 1.1853717464727674e-05, "loss": 2.7378, "step": 172410 }, { "epoch": 0.0038912, "grad_norm": 0.8320466876029968, "learning_rate": 1.185292613695298e-05, "loss": 3.0461, "step": 172420 }, { "epoch": 0.0039168, "grad_norm": 0.7527263164520264, "learning_rate": 1.1852134797162554e-05, "loss": 3.0947, "step": 172430 }, { "epoch": 0.0039424, "grad_norm": 0.9024736881256104, "learning_rate": 1.1851343445361523e-05, "loss": 2.663, "step": 172440 }, { "epoch": 0.003968, "grad_norm": 0.7691287398338318, "learning_rate": 1.1850552081555027e-05, "loss": 2.941, "step": 172450 }, { "epoch": 0.0039936, "grad_norm": 0.6893355846405029, "learning_rate": 1.1849760705748187e-05, "loss": 2.8463, "step": 172460 }, { "epoch": 0.0040192, "grad_norm": 0.7226272821426392, "learning_rate": 1.1848969317946143e-05, "loss": 2.9581, "step": 172470 }, { "epoch": 0.0040448, "grad_norm": 0.7008537650108337, "learning_rate": 1.1848177918154025e-05, "loss": 3.116, "step": 172480 }, { "epoch": 0.0040704, "grad_norm": 0.8968163132667542, "learning_rate": 1.1847386506376961e-05, "loss": 3.1644, "step": 172490 }, { "epoch": 0.004096, "grad_norm": 0.6995431184768677, "learning_rate": 1.1846595082620089e-05, "loss": 2.9512, "step": 172500 }, { "epoch": 2.56e-05, "grad_norm": 0.837833046913147, "learning_rate": 1.1845803646888541e-05, "loss": 2.9688, "step": 172510 }, { "epoch": 5.12e-05, "grad_norm": 0.8323366045951843, "learning_rate": 1.1845012199187445e-05, "loss": 2.7303, "step": 172520 }, { "epoch": 7.68e-05, "grad_norm": 0.660396158695221, "learning_rate": 1.1844220739521934e-05, "loss": 2.8675, "step": 172530 }, { "epoch": 0.0001024, "grad_norm": 0.7659946084022522, "learning_rate": 1.1843429267897141e-05, "loss": 3.307, "step": 172540 }, { "epoch": 0.000128, "grad_norm": 0.832397997379303, "learning_rate": 1.1842637784318203e-05, "loss": 2.8629, "step": 172550 }, { "epoch": 0.0001536, "grad_norm": 0.7814621329307556, "learning_rate": 1.1841846288790244e-05, "loss": 2.8833, "step": 172560 }, { "epoch": 0.0001792, "grad_norm": 0.7609577775001526, "learning_rate": 1.1841054781318402e-05, "loss": 2.7195, "step": 172570 }, { "epoch": 0.0002048, "grad_norm": 0.7250059843063354, "learning_rate": 1.1840263261907815e-05, "loss": 2.9005, "step": 172580 }, { "epoch": 0.0002304, "grad_norm": 0.7944498658180237, "learning_rate": 1.1839471730563603e-05, "loss": 2.6797, "step": 172590 }, { "epoch": 0.000256, "grad_norm": 0.8269703388214111, "learning_rate": 1.1838680187290907e-05, "loss": 2.6776, "step": 172600 }, { "epoch": 0.0002816, "grad_norm": 0.7465259432792664, "learning_rate": 1.1837888632094863e-05, "loss": 2.746, "step": 172610 }, { "epoch": 0.0003072, "grad_norm": 0.6586175560951233, "learning_rate": 1.1837097064980596e-05, "loss": 2.6688, "step": 172620 }, { "epoch": 0.0003328, "grad_norm": 0.8428567051887512, "learning_rate": 1.1836305485953243e-05, "loss": 2.8019, "step": 172630 }, { "epoch": 0.0003584, "grad_norm": 0.8697051405906677, "learning_rate": 1.1835513895017936e-05, "loss": 2.9514, "step": 172640 }, { "epoch": 0.000384, "grad_norm": 0.8065400719642639, "learning_rate": 1.183472229217981e-05, "loss": 2.7793, "step": 172650 }, { "epoch": 0.0004096, "grad_norm": 0.9413745403289795, "learning_rate": 1.1833930677443997e-05, "loss": 3.0075, "step": 172660 }, { "epoch": 0.0004352, "grad_norm": 0.9003576040267944, "learning_rate": 1.1833139050815631e-05, "loss": 3.0242, "step": 172670 }, { "epoch": 0.0004608, "grad_norm": 0.7908284068107605, "learning_rate": 1.1832347412299849e-05, "loss": 2.7628, "step": 172680 }, { "epoch": 0.0004864, "grad_norm": 0.8951981067657471, "learning_rate": 1.1831555761901778e-05, "loss": 2.7129, "step": 172690 }, { "epoch": 0.000512, "grad_norm": 0.7857206463813782, "learning_rate": 1.1830764099626555e-05, "loss": 2.6423, "step": 172700 }, { "epoch": 0.0005376, "grad_norm": 0.8118544816970825, "learning_rate": 1.1829972425479315e-05, "loss": 2.7459, "step": 172710 }, { "epoch": 0.0005632, "grad_norm": 0.7545457482337952, "learning_rate": 1.1829180739465186e-05, "loss": 2.7882, "step": 172720 }, { "epoch": 0.0005888, "grad_norm": 0.7164615988731384, "learning_rate": 1.1828389041589308e-05, "loss": 2.7937, "step": 172730 }, { "epoch": 0.0006144, "grad_norm": 0.7682875990867615, "learning_rate": 1.1827597331856814e-05, "loss": 2.8095, "step": 172740 }, { "epoch": 0.00064, "grad_norm": 0.7498032450675964, "learning_rate": 1.1826805610272838e-05, "loss": 2.7424, "step": 172750 }, { "epoch": 0.0006656, "grad_norm": 0.7856879830360413, "learning_rate": 1.182601387684251e-05, "loss": 2.9362, "step": 172760 }, { "epoch": 0.0006912, "grad_norm": 0.7415906190872192, "learning_rate": 1.1825222131570971e-05, "loss": 2.817, "step": 172770 }, { "epoch": 0.0007168, "grad_norm": 0.8445479273796082, "learning_rate": 1.1824430374463349e-05, "loss": 2.6888, "step": 172780 }, { "epoch": 0.0007424, "grad_norm": 0.7565027475357056, "learning_rate": 1.1823638605524781e-05, "loss": 2.9777, "step": 172790 }, { "epoch": 0.000768, "grad_norm": 0.7568451166152954, "learning_rate": 1.1822846824760403e-05, "loss": 2.8538, "step": 172800 }, { "epoch": 0.0007936, "grad_norm": 0.6996409893035889, "learning_rate": 1.1822055032175348e-05, "loss": 2.7496, "step": 172810 }, { "epoch": 0.0008192, "grad_norm": 0.692755937576294, "learning_rate": 1.1821263227774746e-05, "loss": 2.7906, "step": 172820 }, { "epoch": 0.0008448, "grad_norm": 0.8050832152366638, "learning_rate": 1.1820471411563738e-05, "loss": 2.8381, "step": 172830 }, { "epoch": 0.0008704, "grad_norm": 0.7639116644859314, "learning_rate": 1.1819679583547461e-05, "loss": 2.82, "step": 172840 }, { "epoch": 0.000896, "grad_norm": 0.7055280208587646, "learning_rate": 1.181888774373104e-05, "loss": 2.8268, "step": 172850 }, { "epoch": 0.0009216, "grad_norm": 0.7977848649024963, "learning_rate": 1.181809589211962e-05, "loss": 2.6065, "step": 172860 }, { "epoch": 0.0009472, "grad_norm": 0.7990344762802124, "learning_rate": 1.1817304028718326e-05, "loss": 2.7716, "step": 172870 }, { "epoch": 0.0009728, "grad_norm": 0.6798332929611206, "learning_rate": 1.1816512153532305e-05, "loss": 2.8665, "step": 172880 }, { "epoch": 0.0009984, "grad_norm": 0.7078158259391785, "learning_rate": 1.1815720266566678e-05, "loss": 2.6719, "step": 172890 }, { "epoch": 0.001024, "grad_norm": 0.7559341788291931, "learning_rate": 1.1814928367826591e-05, "loss": 2.609, "step": 172900 }, { "epoch": 0.0010496, "grad_norm": 0.7410348057746887, "learning_rate": 1.1814136457317179e-05, "loss": 2.6586, "step": 172910 }, { "epoch": 0.0010752, "grad_norm": 0.7332859039306641, "learning_rate": 1.181334453504357e-05, "loss": 2.9757, "step": 172920 }, { "epoch": 0.0011008, "grad_norm": 0.6815427541732788, "learning_rate": 1.1812552601010904e-05, "loss": 2.8808, "step": 172930 }, { "epoch": 0.0011264, "grad_norm": 0.759851336479187, "learning_rate": 1.1811760655224318e-05, "loss": 2.9109, "step": 172940 }, { "epoch": 0.001152, "grad_norm": 0.8892029523849487, "learning_rate": 1.1810968697688945e-05, "loss": 2.8403, "step": 172950 }, { "epoch": 0.0011776, "grad_norm": 0.8457830548286438, "learning_rate": 1.1810176728409922e-05, "loss": 2.6574, "step": 172960 }, { "epoch": 0.0012032, "grad_norm": 0.7790572047233582, "learning_rate": 1.1809384747392383e-05, "loss": 2.8095, "step": 172970 }, { "epoch": 0.0012288, "grad_norm": 0.7802683115005493, "learning_rate": 1.1808592754641465e-05, "loss": 2.8654, "step": 172980 }, { "epoch": 0.0012544, "grad_norm": 0.7640780806541443, "learning_rate": 1.1807800750162301e-05, "loss": 2.8982, "step": 172990 }, { "epoch": 0.00128, "grad_norm": 0.9965572357177734, "learning_rate": 1.1807008733960033e-05, "loss": 2.9409, "step": 173000 }, { "epoch": 0.0013056, "grad_norm": 0.833570659160614, "learning_rate": 1.1806216706039794e-05, "loss": 2.7364, "step": 173010 }, { "epoch": 0.0013312, "grad_norm": 0.7783163785934448, "learning_rate": 1.1805424666406718e-05, "loss": 2.9744, "step": 173020 }, { "epoch": 0.0013568, "grad_norm": 0.7624675631523132, "learning_rate": 1.1804632615065942e-05, "loss": 2.7744, "step": 173030 }, { "epoch": 0.0013824, "grad_norm": 0.6983879208564758, "learning_rate": 1.1803840552022607e-05, "loss": 2.7522, "step": 173040 }, { "epoch": 0.001408, "grad_norm": 0.7070256471633911, "learning_rate": 1.1803048477281842e-05, "loss": 2.7517, "step": 173050 }, { "epoch": 0.0014336, "grad_norm": 0.7967227697372437, "learning_rate": 1.1802256390848788e-05, "loss": 2.9451, "step": 173060 }, { "epoch": 0.0014592, "grad_norm": 0.7055456042289734, "learning_rate": 1.180146429272858e-05, "loss": 2.9323, "step": 173070 }, { "epoch": 0.0014848, "grad_norm": 0.7604727745056152, "learning_rate": 1.1800672182926354e-05, "loss": 2.8592, "step": 173080 }, { "epoch": 0.0015104, "grad_norm": 1.0946123600006104, "learning_rate": 1.1799880061447247e-05, "loss": 2.6139, "step": 173090 }, { "epoch": 0.001536, "grad_norm": 0.6634453535079956, "learning_rate": 1.1799087928296399e-05, "loss": 2.7687, "step": 173100 }, { "epoch": 0.0015616, "grad_norm": 0.8091483116149902, "learning_rate": 1.1798295783478944e-05, "loss": 2.6521, "step": 173110 }, { "epoch": 0.0015872, "grad_norm": 0.8218347430229187, "learning_rate": 1.1797503627000015e-05, "loss": 2.7226, "step": 173120 }, { "epoch": 0.0016128, "grad_norm": 0.7118636965751648, "learning_rate": 1.1796711458864753e-05, "loss": 2.8165, "step": 173130 }, { "epoch": 0.0016384, "grad_norm": 0.8369912505149841, "learning_rate": 1.17959192790783e-05, "loss": 3.0795, "step": 173140 }, { "epoch": 0.001664, "grad_norm": 0.7145464420318604, "learning_rate": 1.1795127087645784e-05, "loss": 2.9278, "step": 173150 }, { "epoch": 0.0016896, "grad_norm": 0.7679559588432312, "learning_rate": 1.1794334884572344e-05, "loss": 2.8487, "step": 173160 }, { "epoch": 0.0017152, "grad_norm": 0.7055689096450806, "learning_rate": 1.1793542669863123e-05, "loss": 2.7538, "step": 173170 }, { "epoch": 0.0017408, "grad_norm": 0.7685969471931458, "learning_rate": 1.1792750443523251e-05, "loss": 2.8964, "step": 173180 }, { "epoch": 0.0017664, "grad_norm": 0.8188390135765076, "learning_rate": 1.1791958205557873e-05, "loss": 2.9224, "step": 173190 }, { "epoch": 0.001792, "grad_norm": 2.0015597343444824, "learning_rate": 1.1791165955972117e-05, "loss": 2.9797, "step": 173200 }, { "epoch": 2.56e-05, "grad_norm": 0.8252999782562256, "learning_rate": 1.179037369477113e-05, "loss": 2.8462, "step": 173210 }, { "epoch": 5.12e-05, "grad_norm": 0.961331844329834, "learning_rate": 1.1789581421960045e-05, "loss": 2.6327, "step": 173220 }, { "epoch": 7.68e-05, "grad_norm": 0.6547598242759705, "learning_rate": 1.1788789137544e-05, "loss": 2.771, "step": 173230 }, { "epoch": 0.0001024, "grad_norm": 0.7481229901313782, "learning_rate": 1.1787996841528133e-05, "loss": 3.1375, "step": 173240 }, { "epoch": 0.000128, "grad_norm": 0.8589295148849487, "learning_rate": 1.1787204533917578e-05, "loss": 2.7684, "step": 173250 }, { "epoch": 0.0001536, "grad_norm": 0.7556331753730774, "learning_rate": 1.1786412214717478e-05, "loss": 2.7817, "step": 173260 }, { "epoch": 0.0001792, "grad_norm": 0.7900488972663879, "learning_rate": 1.1785619883932973e-05, "loss": 2.628, "step": 173270 }, { "epoch": 0.0002048, "grad_norm": 0.7277190685272217, "learning_rate": 1.1784827541569195e-05, "loss": 2.7997, "step": 173280 }, { "epoch": 0.0002304, "grad_norm": 0.7306280136108398, "learning_rate": 1.1784035187631284e-05, "loss": 2.5849, "step": 173290 }, { "epoch": 0.000256, "grad_norm": 0.8203924894332886, "learning_rate": 1.178324282212438e-05, "loss": 2.5747, "step": 173300 }, { "epoch": 0.0002816, "grad_norm": 0.75213223695755, "learning_rate": 1.1782450445053624e-05, "loss": 2.6605, "step": 173310 }, { "epoch": 0.0003072, "grad_norm": 0.6805530190467834, "learning_rate": 1.1781658056424145e-05, "loss": 2.565, "step": 173320 }, { "epoch": 0.0003328, "grad_norm": 0.8921489119529724, "learning_rate": 1.178086565624109e-05, "loss": 2.7021, "step": 173330 }, { "epoch": 0.0003584, "grad_norm": 0.8477033376693726, "learning_rate": 1.1780073244509591e-05, "loss": 2.8344, "step": 173340 }, { "epoch": 0.000384, "grad_norm": 0.7766164541244507, "learning_rate": 1.1779280821234793e-05, "loss": 2.6724, "step": 173350 }, { "epoch": 0.0004096, "grad_norm": 0.9692688584327698, "learning_rate": 1.1778488386421831e-05, "loss": 2.8895, "step": 173360 }, { "epoch": 0.0004352, "grad_norm": 0.7998868227005005, "learning_rate": 1.1777695940075846e-05, "loss": 2.8969, "step": 173370 }, { "epoch": 0.0004608, "grad_norm": 0.8007900714874268, "learning_rate": 1.1776903482201976e-05, "loss": 2.6713, "step": 173380 }, { "epoch": 0.0004864, "grad_norm": 0.9697636365890503, "learning_rate": 1.1776111012805355e-05, "loss": 2.6006, "step": 173390 }, { "epoch": 0.000512, "grad_norm": 0.780389130115509, "learning_rate": 1.177531853189113e-05, "loss": 2.5477, "step": 173400 }, { "epoch": 0.0005376, "grad_norm": 1.1342930793762207, "learning_rate": 1.1774526039464436e-05, "loss": 2.6635, "step": 173410 }, { "epoch": 0.0005632, "grad_norm": 0.7507990002632141, "learning_rate": 1.177373353553041e-05, "loss": 2.6821, "step": 173420 }, { "epoch": 0.0005888, "grad_norm": 0.7292525172233582, "learning_rate": 1.1772941020094192e-05, "loss": 2.6807, "step": 173430 }, { "epoch": 0.0006144, "grad_norm": 0.7728913426399231, "learning_rate": 1.1772148493160927e-05, "loss": 2.6903, "step": 173440 }, { "epoch": 0.00064, "grad_norm": 0.7690333724021912, "learning_rate": 1.1771355954735745e-05, "loss": 2.6461, "step": 173450 }, { "epoch": 0.0006656, "grad_norm": 0.7947357892990112, "learning_rate": 1.1770563404823794e-05, "loss": 2.8265, "step": 173460 }, { "epoch": 0.0006912, "grad_norm": 0.7456632256507874, "learning_rate": 1.1769770843430211e-05, "loss": 2.7143, "step": 173470 }, { "epoch": 0.0007168, "grad_norm": 0.879496157169342, "learning_rate": 1.1768978270560132e-05, "loss": 2.5925, "step": 173480 }, { "epoch": 0.0007424, "grad_norm": 0.8305849432945251, "learning_rate": 1.17681856862187e-05, "loss": 2.8699, "step": 173490 }, { "epoch": 0.000768, "grad_norm": 0.723537027835846, "learning_rate": 1.1767393090411054e-05, "loss": 2.737, "step": 173500 }, { "epoch": 0.0007936, "grad_norm": 0.7214834094047546, "learning_rate": 1.1766600483142331e-05, "loss": 2.6516, "step": 173510 }, { "epoch": 0.0008192, "grad_norm": 0.7061360478401184, "learning_rate": 1.1765807864417676e-05, "loss": 2.6876, "step": 173520 }, { "epoch": 0.0008448, "grad_norm": 0.8220206499099731, "learning_rate": 1.1765015234242228e-05, "loss": 2.7406, "step": 173530 }, { "epoch": 0.0008704, "grad_norm": 0.8413786888122559, "learning_rate": 1.1764222592621121e-05, "loss": 2.7091, "step": 173540 }, { "epoch": 0.000896, "grad_norm": 0.714245617389679, "learning_rate": 1.17634299395595e-05, "loss": 2.7101, "step": 173550 }, { "epoch": 0.0009216, "grad_norm": 0.8945775032043457, "learning_rate": 1.1762637275062506e-05, "loss": 2.4957, "step": 173560 }, { "epoch": 0.0009472, "grad_norm": 0.7822365164756775, "learning_rate": 1.1761844599135279e-05, "loss": 2.675, "step": 173570 }, { "epoch": 0.0009728, "grad_norm": 0.6592868566513062, "learning_rate": 1.1761051911782957e-05, "loss": 2.7579, "step": 173580 }, { "epoch": 0.0009984, "grad_norm": 0.7284269332885742, "learning_rate": 1.1760259213010681e-05, "loss": 2.5734, "step": 173590 }, { "epoch": 0.001024, "grad_norm": 0.7780434489250183, "learning_rate": 1.1759466502823591e-05, "loss": 2.5153, "step": 173600 }, { "epoch": 0.0010496, "grad_norm": 0.7441065311431885, "learning_rate": 1.1758673781226828e-05, "loss": 2.5312, "step": 173610 }, { "epoch": 0.0010752, "grad_norm": 0.7621614336967468, "learning_rate": 1.1757881048225532e-05, "loss": 2.874, "step": 173620 }, { "epoch": 0.0011008, "grad_norm": 0.705966055393219, "learning_rate": 1.175708830382485e-05, "loss": 2.7849, "step": 173630 }, { "epoch": 0.0011264, "grad_norm": 0.8003607392311096, "learning_rate": 1.1756295548029915e-05, "loss": 2.8013, "step": 173640 }, { "epoch": 0.001152, "grad_norm": 0.8179855942726135, "learning_rate": 1.1755502780845867e-05, "loss": 2.6927, "step": 173650 }, { "epoch": 0.0011776, "grad_norm": 0.807626485824585, "learning_rate": 1.1754710002277851e-05, "loss": 2.5645, "step": 173660 }, { "epoch": 0.0012032, "grad_norm": 0.7896914482116699, "learning_rate": 1.175391721233101e-05, "loss": 2.7042, "step": 173670 }, { "epoch": 0.0012288, "grad_norm": 0.7476792335510254, "learning_rate": 1.175312441101048e-05, "loss": 2.7709, "step": 173680 }, { "epoch": 0.0012544, "grad_norm": 0.7918075323104858, "learning_rate": 1.1752331598321403e-05, "loss": 2.8017, "step": 173690 }, { "epoch": 0.00128, "grad_norm": 0.9064738750457764, "learning_rate": 1.1751538774268922e-05, "loss": 2.8213, "step": 173700 }, { "epoch": 0.0013056, "grad_norm": 0.8082424402236938, "learning_rate": 1.1750745938858179e-05, "loss": 2.6452, "step": 173710 }, { "epoch": 0.0013312, "grad_norm": 0.783677875995636, "learning_rate": 1.1749953092094314e-05, "loss": 2.8591, "step": 173720 }, { "epoch": 0.0013568, "grad_norm": 0.7654567360877991, "learning_rate": 1.1749160233982467e-05, "loss": 2.6828, "step": 173730 }, { "epoch": 0.0013824, "grad_norm": 0.720088005065918, "learning_rate": 1.174836736452778e-05, "loss": 2.6474, "step": 173740 }, { "epoch": 0.001408, "grad_norm": 0.7097371816635132, "learning_rate": 1.1747574483735396e-05, "loss": 2.6493, "step": 173750 }, { "epoch": 0.0014336, "grad_norm": 0.7806673049926758, "learning_rate": 1.1746781591610457e-05, "loss": 2.8224, "step": 173760 }, { "epoch": 0.0014592, "grad_norm": 0.7399827837944031, "learning_rate": 1.17459886881581e-05, "loss": 2.8251, "step": 173770 }, { "epoch": 0.0014848, "grad_norm": 0.7599501013755798, "learning_rate": 1.1745195773383475e-05, "loss": 2.773, "step": 173780 }, { "epoch": 0.0015104, "grad_norm": 0.8269316554069519, "learning_rate": 1.1744402847291715e-05, "loss": 2.5272, "step": 173790 }, { "epoch": 0.001536, "grad_norm": 0.6904733180999756, "learning_rate": 1.1743609909887973e-05, "loss": 2.6706, "step": 173800 }, { "epoch": 0.0015616, "grad_norm": 0.8134756684303284, "learning_rate": 1.1742816961177378e-05, "loss": 2.5398, "step": 173810 }, { "epoch": 0.0015872, "grad_norm": 0.8299903869628906, "learning_rate": 1.174202400116508e-05, "loss": 2.6341, "step": 173820 }, { "epoch": 0.0016128, "grad_norm": 0.739424467086792, "learning_rate": 1.1741231029856221e-05, "loss": 2.7237, "step": 173830 }, { "epoch": 0.0016384, "grad_norm": 1.0570639371871948, "learning_rate": 1.174043804725594e-05, "loss": 2.9737, "step": 173840 }, { "epoch": 0.001664, "grad_norm": 0.7471712231636047, "learning_rate": 1.173964505336938e-05, "loss": 2.8379, "step": 173850 }, { "epoch": 0.0016896, "grad_norm": 0.7705503702163696, "learning_rate": 1.1738852048201683e-05, "loss": 2.7547, "step": 173860 }, { "epoch": 0.0017152, "grad_norm": 0.7243708372116089, "learning_rate": 1.1738059031757994e-05, "loss": 2.6639, "step": 173870 }, { "epoch": 0.0017408, "grad_norm": 0.7750471234321594, "learning_rate": 1.1737266004043455e-05, "loss": 2.7954, "step": 173880 }, { "epoch": 0.0017664, "grad_norm": 0.7837400436401367, "learning_rate": 1.1736472965063206e-05, "loss": 2.8273, "step": 173890 }, { "epoch": 0.001792, "grad_norm": 2.173328399658203, "learning_rate": 1.1735679914822393e-05, "loss": 2.8869, "step": 173900 }, { "epoch": 2.56e-05, "grad_norm": 0.8104168772697449, "learning_rate": 1.1734886853326157e-05, "loss": 2.751, "step": 173910 }, { "epoch": 5.12e-05, "grad_norm": 0.96109539270401, "learning_rate": 1.173409378057964e-05, "loss": 2.5332, "step": 173920 }, { "epoch": 7.68e-05, "grad_norm": 0.7246716618537903, "learning_rate": 1.1733300696587988e-05, "loss": 2.6896, "step": 173930 }, { "epoch": 0.0001024, "grad_norm": 0.8162938356399536, "learning_rate": 1.173250760135634e-05, "loss": 3.0792, "step": 173940 }, { "epoch": 0.000128, "grad_norm": 0.8910471200942993, "learning_rate": 1.1731714494889836e-05, "loss": 2.6968, "step": 173950 }, { "epoch": 0.0001536, "grad_norm": 0.8117607235908508, "learning_rate": 1.173092137719363e-05, "loss": 2.7002, "step": 173960 }, { "epoch": 0.0001792, "grad_norm": 0.7742542028427124, "learning_rate": 1.1730128248272856e-05, "loss": 2.5485, "step": 173970 }, { "epoch": 0.0002048, "grad_norm": 0.7549847960472107, "learning_rate": 1.172933510813266e-05, "loss": 2.7123, "step": 173980 }, { "epoch": 0.0002304, "grad_norm": 0.7694359421730042, "learning_rate": 1.1728541956778188e-05, "loss": 2.5025, "step": 173990 }, { "epoch": 0.000256, "grad_norm": 0.8226665258407593, "learning_rate": 1.1727748794214579e-05, "loss": 2.4899, "step": 174000 }, { "epoch": 0.0002816, "grad_norm": 0.7971057295799255, "learning_rate": 1.172695562044698e-05, "loss": 2.5727, "step": 174010 }, { "epoch": 0.0003072, "grad_norm": 0.7278082966804504, "learning_rate": 1.172616243548053e-05, "loss": 2.4779, "step": 174020 }, { "epoch": 0.0003328, "grad_norm": 0.8503307104110718, "learning_rate": 1.1725369239320379e-05, "loss": 2.5829, "step": 174030 }, { "epoch": 0.0003584, "grad_norm": 0.8713273406028748, "learning_rate": 1.1724576031971662e-05, "loss": 2.7327, "step": 174040 }, { "epoch": 0.000384, "grad_norm": 0.8159201741218567, "learning_rate": 1.1723782813439531e-05, "loss": 2.5851, "step": 174050 }, { "epoch": 0.0004096, "grad_norm": 1.035557508468628, "learning_rate": 1.1722989583729128e-05, "loss": 2.7974, "step": 174060 }, { "epoch": 0.0004352, "grad_norm": 0.8319981098175049, "learning_rate": 1.1722196342845591e-05, "loss": 2.791, "step": 174070 }, { "epoch": 0.0004608, "grad_norm": 0.8083121180534363, "learning_rate": 1.1721403090794072e-05, "loss": 2.5923, "step": 174080 }, { "epoch": 0.0004864, "grad_norm": 1.0174126625061035, "learning_rate": 1.172060982757971e-05, "loss": 2.5127, "step": 174090 }, { "epoch": 0.000512, "grad_norm": 0.8033115863800049, "learning_rate": 1.1719816553207653e-05, "loss": 2.4648, "step": 174100 }, { "epoch": 0.0005376, "grad_norm": 0.9741674661636353, "learning_rate": 1.1719023267683039e-05, "loss": 2.5869, "step": 174110 }, { "epoch": 0.0005632, "grad_norm": 0.7940669059753418, "learning_rate": 1.1718229971011017e-05, "loss": 2.5893, "step": 174120 }, { "epoch": 0.0005888, "grad_norm": 0.7625023126602173, "learning_rate": 1.171743666319673e-05, "loss": 2.5799, "step": 174130 }, { "epoch": 0.0006144, "grad_norm": 0.8184512853622437, "learning_rate": 1.1716643344245325e-05, "loss": 2.5925, "step": 174140 }, { "epoch": 0.00064, "grad_norm": 0.8047292232513428, "learning_rate": 1.1715850014161941e-05, "loss": 2.5583, "step": 174150 }, { "epoch": 0.0006656, "grad_norm": 0.8286426663398743, "learning_rate": 1.171505667295173e-05, "loss": 2.7282, "step": 174160 }, { "epoch": 0.0006912, "grad_norm": 0.8812823295593262, "learning_rate": 1.1714263320619827e-05, "loss": 2.6188, "step": 174170 }, { "epoch": 0.0007168, "grad_norm": 0.8818685412406921, "learning_rate": 1.1713469957171387e-05, "loss": 2.5091, "step": 174180 }, { "epoch": 0.0007424, "grad_norm": 0.8404036164283752, "learning_rate": 1.1712676582611546e-05, "loss": 2.7665, "step": 174190 }, { "epoch": 0.000768, "grad_norm": 0.759715735912323, "learning_rate": 1.1711883196945456e-05, "loss": 2.6462, "step": 174200 }, { "epoch": 0.0007936, "grad_norm": 0.761411190032959, "learning_rate": 1.1711089800178253e-05, "loss": 2.5694, "step": 174210 }, { "epoch": 0.0008192, "grad_norm": 0.7347497344017029, "learning_rate": 1.1710296392315093e-05, "loss": 2.5992, "step": 174220 }, { "epoch": 0.0008448, "grad_norm": 0.8878333568572998, "learning_rate": 1.1709502973361114e-05, "loss": 2.6563, "step": 174230 }, { "epoch": 0.0008704, "grad_norm": 0.8240652084350586, "learning_rate": 1.1708709543321459e-05, "loss": 2.6227, "step": 174240 }, { "epoch": 0.000896, "grad_norm": 0.743211567401886, "learning_rate": 1.170791610220128e-05, "loss": 2.6014, "step": 174250 }, { "epoch": 0.0009216, "grad_norm": 0.9280749559402466, "learning_rate": 1.170712265000572e-05, "loss": 2.4004, "step": 174260 }, { "epoch": 0.0009472, "grad_norm": 0.8706360459327698, "learning_rate": 1.1706329186739922e-05, "loss": 2.5871, "step": 174270 }, { "epoch": 0.0009728, "grad_norm": 0.6983007788658142, "learning_rate": 1.1705535712409031e-05, "loss": 2.6618, "step": 174280 }, { "epoch": 0.0009984, "grad_norm": 0.7667189240455627, "learning_rate": 1.1704742227018196e-05, "loss": 2.4812, "step": 174290 }, { "epoch": 0.001024, "grad_norm": 0.796097993850708, "learning_rate": 1.1703948730572559e-05, "loss": 2.4326, "step": 174300 }, { "epoch": 0.0010496, "grad_norm": 0.7675474286079407, "learning_rate": 1.1703155223077269e-05, "loss": 2.4255, "step": 174310 }, { "epoch": 0.0010752, "grad_norm": 0.7908011674880981, "learning_rate": 1.170236170453747e-05, "loss": 2.7733, "step": 174320 }, { "epoch": 0.0011008, "grad_norm": 0.7327082753181458, "learning_rate": 1.1701568174958308e-05, "loss": 2.7016, "step": 174330 }, { "epoch": 0.0011264, "grad_norm": 0.8725582361221313, "learning_rate": 1.1700774634344926e-05, "loss": 2.7057, "step": 174340 }, { "epoch": 0.001152, "grad_norm": 0.882472813129425, "learning_rate": 1.1699981082702477e-05, "loss": 2.5906, "step": 174350 }, { "epoch": 0.0011776, "grad_norm": 0.8586829900741577, "learning_rate": 1.16991875200361e-05, "loss": 2.4892, "step": 174360 }, { "epoch": 0.0012032, "grad_norm": 0.8162894248962402, "learning_rate": 1.1698393946350944e-05, "loss": 2.6096, "step": 174370 }, { "epoch": 0.0012288, "grad_norm": 0.7594168186187744, "learning_rate": 1.1697600361652153e-05, "loss": 2.6868, "step": 174380 }, { "epoch": 0.0012544, "grad_norm": 0.8124955296516418, "learning_rate": 1.1696806765944877e-05, "loss": 2.7168, "step": 174390 }, { "epoch": 0.00128, "grad_norm": 0.9372060894966125, "learning_rate": 1.1696013159234259e-05, "loss": 2.7253, "step": 174400 }, { "epoch": 0.0013056, "grad_norm": 0.881633460521698, "learning_rate": 1.1695219541525446e-05, "loss": 2.5657, "step": 174410 }, { "epoch": 0.0013312, "grad_norm": 0.8216814398765564, "learning_rate": 1.1694425912823586e-05, "loss": 2.7587, "step": 174420 }, { "epoch": 0.0013568, "grad_norm": 0.8086392879486084, "learning_rate": 1.1693632273133824e-05, "loss": 2.6025, "step": 174430 }, { "epoch": 0.0013824, "grad_norm": 0.7519819140434265, "learning_rate": 1.1692838622461306e-05, "loss": 2.5704, "step": 174440 }, { "epoch": 0.001408, "grad_norm": 0.7648795247077942, "learning_rate": 1.169204496081118e-05, "loss": 2.5555, "step": 174450 }, { "epoch": 0.0014336, "grad_norm": 0.8219456672668457, "learning_rate": 1.1691251288188595e-05, "loss": 2.723, "step": 174460 }, { "epoch": 0.0014592, "grad_norm": 0.7984724640846252, "learning_rate": 1.1690457604598689e-05, "loss": 2.7341, "step": 174470 }, { "epoch": 0.0014848, "grad_norm": 0.8342920541763306, "learning_rate": 1.1689663910046616e-05, "loss": 2.695, "step": 174480 }, { "epoch": 0.0015104, "grad_norm": 0.8372483253479004, "learning_rate": 1.1688870204537525e-05, "loss": 2.4505, "step": 174490 }, { "epoch": 0.001536, "grad_norm": 0.7297271490097046, "learning_rate": 1.1688076488076556e-05, "loss": 2.579, "step": 174500 }, { "epoch": 0.0015616, "grad_norm": 0.8603392839431763, "learning_rate": 1.1687282760668861e-05, "loss": 2.4389, "step": 174510 }, { "epoch": 0.0015872, "grad_norm": 0.8460466265678406, "learning_rate": 1.168648902231959e-05, "loss": 2.5534, "step": 174520 }, { "epoch": 0.0016128, "grad_norm": 0.9986941814422607, "learning_rate": 1.1685695273033881e-05, "loss": 2.6337, "step": 174530 }, { "epoch": 0.0016384, "grad_norm": 0.9342180490493774, "learning_rate": 1.1684901512816887e-05, "loss": 2.8768, "step": 174540 }, { "epoch": 0.001664, "grad_norm": 0.8300243616104126, "learning_rate": 1.1684107741673754e-05, "loss": 2.7533, "step": 174550 }, { "epoch": 0.0016896, "grad_norm": 0.8226402997970581, "learning_rate": 1.168331395960963e-05, "loss": 2.6828, "step": 174560 }, { "epoch": 0.0017152, "grad_norm": 0.7759517431259155, "learning_rate": 1.1682520166629663e-05, "loss": 2.5843, "step": 174570 }, { "epoch": 0.0017408, "grad_norm": 0.8324217796325684, "learning_rate": 1.1681726362738998e-05, "loss": 2.7066, "step": 174580 }, { "epoch": 0.0017664, "grad_norm": 0.8736914396286011, "learning_rate": 1.1680932547942789e-05, "loss": 2.7533, "step": 174590 }, { "epoch": 0.001792, "grad_norm": 33.07537841796875, "learning_rate": 1.1680138722246176e-05, "loss": 3.2207, "step": 174600 }, { "epoch": 2.56e-05, "grad_norm": 0.8718136548995972, "learning_rate": 1.167934488565431e-05, "loss": 2.6518, "step": 174610 }, { "epoch": 5.12e-05, "grad_norm": 0.8869025111198425, "learning_rate": 1.167855103817234e-05, "loss": 2.4385, "step": 174620 }, { "epoch": 7.68e-05, "grad_norm": 0.7694931626319885, "learning_rate": 1.167775717980541e-05, "loss": 2.6125, "step": 174630 }, { "epoch": 0.0001024, "grad_norm": 0.8260440826416016, "learning_rate": 1.1676963310558671e-05, "loss": 3.0079, "step": 174640 }, { "epoch": 0.000128, "grad_norm": 0.8748237490653992, "learning_rate": 1.1676169430437274e-05, "loss": 2.6161, "step": 174650 }, { "epoch": 0.0001536, "grad_norm": 0.8593302369117737, "learning_rate": 1.1675375539446361e-05, "loss": 2.609, "step": 174660 }, { "epoch": 0.0001792, "grad_norm": 0.8349483609199524, "learning_rate": 1.1674581637591083e-05, "loss": 2.4723, "step": 174670 }, { "epoch": 0.0002048, "grad_norm": 0.7999831438064575, "learning_rate": 1.167378772487659e-05, "loss": 2.6224, "step": 174680 }, { "epoch": 0.0002304, "grad_norm": 0.8203142285346985, "learning_rate": 1.1672993801308027e-05, "loss": 2.4195, "step": 174690 }, { "epoch": 0.000256, "grad_norm": 0.8289633989334106, "learning_rate": 1.1672199866890544e-05, "loss": 2.3944, "step": 174700 }, { "epoch": 0.0002816, "grad_norm": 0.914984405040741, "learning_rate": 1.1671405921629289e-05, "loss": 2.4882, "step": 174710 }, { "epoch": 0.0003072, "grad_norm": 0.7880131006240845, "learning_rate": 1.1670611965529412e-05, "loss": 2.3859, "step": 174720 }, { "epoch": 0.0003328, "grad_norm": 0.995449960231781, "learning_rate": 1.1669817998596059e-05, "loss": 2.481, "step": 174730 }, { "epoch": 0.0003584, "grad_norm": 0.9085299372673035, "learning_rate": 1.1669024020834378e-05, "loss": 2.6271, "step": 174740 }, { "epoch": 0.000384, "grad_norm": 0.8595489859580994, "learning_rate": 1.1668230032249524e-05, "loss": 2.4965, "step": 174750 }, { "epoch": 0.0004096, "grad_norm": 1.062538743019104, "learning_rate": 1.166743603284664e-05, "loss": 2.7055, "step": 174760 }, { "epoch": 0.0004352, "grad_norm": 0.8705061674118042, "learning_rate": 1.1666642022630876e-05, "loss": 2.6847, "step": 174770 }, { "epoch": 0.0004608, "grad_norm": 0.8783756494522095, "learning_rate": 1.1665848001607381e-05, "loss": 2.5138, "step": 174780 }, { "epoch": 0.0004864, "grad_norm": 1.0159835815429688, "learning_rate": 1.1665053969781305e-05, "loss": 2.4199, "step": 174790 }, { "epoch": 0.000512, "grad_norm": 0.8640912175178528, "learning_rate": 1.1664259927157797e-05, "loss": 2.3785, "step": 174800 }, { "epoch": 0.0005376, "grad_norm": 0.9586338996887207, "learning_rate": 1.1663465873742006e-05, "loss": 2.5007, "step": 174810 }, { "epoch": 0.0005632, "grad_norm": 0.8343967795372009, "learning_rate": 1.166267180953908e-05, "loss": 2.4896, "step": 174820 }, { "epoch": 0.0005888, "grad_norm": 0.8151038885116577, "learning_rate": 1.166187773455417e-05, "loss": 2.477, "step": 174830 }, { "epoch": 0.0006144, "grad_norm": 0.8889661431312561, "learning_rate": 1.1661083648792422e-05, "loss": 2.4832, "step": 174840 }, { "epoch": 0.00064, "grad_norm": 0.8653803467750549, "learning_rate": 1.1660289552258992e-05, "loss": 2.4638, "step": 174850 }, { "epoch": 0.0006656, "grad_norm": 0.9097771048545837, "learning_rate": 1.1659495444959023e-05, "loss": 2.6278, "step": 174860 }, { "epoch": 0.0006912, "grad_norm": 0.921151876449585, "learning_rate": 1.1658701326897668e-05, "loss": 2.5198, "step": 174870 }, { "epoch": 0.0007168, "grad_norm": 0.9501931071281433, "learning_rate": 1.1657907198080074e-05, "loss": 2.42, "step": 174880 }, { "epoch": 0.0007424, "grad_norm": 0.9290995597839355, "learning_rate": 1.1657113058511396e-05, "loss": 2.662, "step": 174890 }, { "epoch": 0.000768, "grad_norm": 0.8094062805175781, "learning_rate": 1.1656318908196778e-05, "loss": 2.5479, "step": 174900 }, { "epoch": 0.0007936, "grad_norm": 0.8316280245780945, "learning_rate": 1.1655524747141372e-05, "loss": 2.4828, "step": 174910 }, { "epoch": 0.0008192, "grad_norm": 0.8025264143943787, "learning_rate": 1.1654730575350329e-05, "loss": 2.5077, "step": 174920 }, { "epoch": 0.0008448, "grad_norm": 0.9510561227798462, "learning_rate": 1.1653936392828796e-05, "loss": 2.5673, "step": 174930 }, { "epoch": 0.0008704, "grad_norm": 0.8864812254905701, "learning_rate": 1.1653142199581927e-05, "loss": 2.5268, "step": 174940 }, { "epoch": 0.000896, "grad_norm": 0.8201943039894104, "learning_rate": 1.1652347995614868e-05, "loss": 2.4959, "step": 174950 }, { "epoch": 0.0009216, "grad_norm": 0.9976449012756348, "learning_rate": 1.1651553780932775e-05, "loss": 2.3098, "step": 174960 }, { "epoch": 0.0009472, "grad_norm": 0.909758448600769, "learning_rate": 1.165075955554079e-05, "loss": 2.5013, "step": 174970 }, { "epoch": 0.0009728, "grad_norm": 0.7707372307777405, "learning_rate": 1.164996531944407e-05, "loss": 2.5612, "step": 174980 }, { "epoch": 0.0009984, "grad_norm": 0.837324321269989, "learning_rate": 1.1649171072647762e-05, "loss": 2.3869, "step": 174990 }, { "epoch": 0.001024, "grad_norm": 0.8812177181243896, "learning_rate": 1.1648376815157022e-05, "loss": 2.3471, "step": 175000 }, { "epoch": 0.0010496, "grad_norm": 0.8222870230674744, "learning_rate": 1.1647582546976995e-05, "loss": 2.326, "step": 175010 }, { "epoch": 0.0010752, "grad_norm": 0.8761847019195557, "learning_rate": 1.1646788268112832e-05, "loss": 2.6759, "step": 175020 }, { "epoch": 0.0011008, "grad_norm": 0.8008686900138855, "learning_rate": 1.1645993978569682e-05, "loss": 2.6147, "step": 175030 }, { "epoch": 0.0011264, "grad_norm": 0.9457492828369141, "learning_rate": 1.1645199678352703e-05, "loss": 2.6057, "step": 175040 }, { "epoch": 0.001152, "grad_norm": 1.0553078651428223, "learning_rate": 1.1644405367467042e-05, "loss": 2.4866, "step": 175050 }, { "epoch": 0.0011776, "grad_norm": 0.9009864926338196, "learning_rate": 1.1643611045917846e-05, "loss": 2.4046, "step": 175060 }, { "epoch": 0.0012032, "grad_norm": 0.8672531247138977, "learning_rate": 1.164281671371027e-05, "loss": 2.5119, "step": 175070 }, { "epoch": 0.0012288, "grad_norm": 0.8232078552246094, "learning_rate": 1.1642022370849464e-05, "loss": 2.6036, "step": 175080 }, { "epoch": 0.0012544, "grad_norm": 0.8797309398651123, "learning_rate": 1.164122801734058e-05, "loss": 2.6285, "step": 175090 }, { "epoch": 0.00128, "grad_norm": 0.9540970921516418, "learning_rate": 1.1640433653188768e-05, "loss": 2.6272, "step": 175100 }, { "epoch": 0.0013056, "grad_norm": 0.9914212822914124, "learning_rate": 1.1639639278399182e-05, "loss": 2.4836, "step": 175110 }, { "epoch": 0.0013312, "grad_norm": 0.8892649412155151, "learning_rate": 1.1638844892976968e-05, "loss": 2.6642, "step": 175120 }, { "epoch": 0.0013568, "grad_norm": 0.8660133481025696, "learning_rate": 1.163805049692728e-05, "loss": 2.5139, "step": 175130 }, { "epoch": 0.0013824, "grad_norm": 0.8154740333557129, "learning_rate": 1.1637256090255271e-05, "loss": 2.4781, "step": 175140 }, { "epoch": 0.001408, "grad_norm": 0.8059491515159607, "learning_rate": 1.1636461672966092e-05, "loss": 2.4628, "step": 175150 }, { "epoch": 0.0014336, "grad_norm": 0.8585731387138367, "learning_rate": 1.1635667245064894e-05, "loss": 2.6266, "step": 175160 }, { "epoch": 0.0014592, "grad_norm": 0.8831412196159363, "learning_rate": 1.1634872806556827e-05, "loss": 2.6328, "step": 175170 }, { "epoch": 0.0014848, "grad_norm": 0.8830903172492981, "learning_rate": 1.1634078357447047e-05, "loss": 2.6111, "step": 175180 }, { "epoch": 0.0015104, "grad_norm": 0.8500996232032776, "learning_rate": 1.1633283897740701e-05, "loss": 2.3659, "step": 175190 }, { "epoch": 0.001536, "grad_norm": 0.8076556921005249, "learning_rate": 1.1632489427442943e-05, "loss": 2.4871, "step": 175200 }, { "epoch": 0.0015616, "grad_norm": 0.9059053659439087, "learning_rate": 1.1631694946558928e-05, "loss": 2.3374, "step": 175210 }, { "epoch": 0.0015872, "grad_norm": 0.9047465920448303, "learning_rate": 1.1630900455093803e-05, "loss": 2.47, "step": 175220 }, { "epoch": 0.0016128, "grad_norm": 0.8671374917030334, "learning_rate": 1.1630105953052722e-05, "loss": 2.5399, "step": 175230 }, { "epoch": 0.0016384, "grad_norm": 0.9660854935646057, "learning_rate": 1.1629311440440836e-05, "loss": 2.7737, "step": 175240 }, { "epoch": 0.001664, "grad_norm": 0.8750063180923462, "learning_rate": 1.1628516917263298e-05, "loss": 2.6557, "step": 175250 }, { "epoch": 0.0016896, "grad_norm": 0.8735054135322571, "learning_rate": 1.1627722383525262e-05, "loss": 2.6151, "step": 175260 }, { "epoch": 0.0017152, "grad_norm": 0.8483597040176392, "learning_rate": 1.1626927839231876e-05, "loss": 2.4986, "step": 175270 }, { "epoch": 0.0017408, "grad_norm": 0.8747251033782959, "learning_rate": 1.16261332843883e-05, "loss": 2.6176, "step": 175280 }, { "epoch": 0.0017664, "grad_norm": 0.9646210670471191, "learning_rate": 1.162533871899968e-05, "loss": 2.6795, "step": 175290 }, { "epoch": 0.001792, "grad_norm": 4.492051124572754, "learning_rate": 1.162454414307117e-05, "loss": 2.7863, "step": 175300 }, { "epoch": 2.56e-05, "grad_norm": 0.8858876824378967, "learning_rate": 1.1623749556607924e-05, "loss": 2.5521, "step": 175310 }, { "epoch": 5.12e-05, "grad_norm": 1.0715711116790771, "learning_rate": 1.1622954959615094e-05, "loss": 2.3547, "step": 175320 }, { "epoch": 7.68e-05, "grad_norm": 0.854969322681427, "learning_rate": 1.162216035209783e-05, "loss": 2.5201, "step": 175330 }, { "epoch": 0.0001024, "grad_norm": 0.9179050922393799, "learning_rate": 1.1621365734061288e-05, "loss": 2.9087, "step": 175340 }, { "epoch": 0.000128, "grad_norm": 0.9295458197593689, "learning_rate": 1.1620571105510621e-05, "loss": 2.5387, "step": 175350 }, { "epoch": 0.0001536, "grad_norm": 1.0106277465820312, "learning_rate": 1.1619776466450979e-05, "loss": 2.5151, "step": 175360 }, { "epoch": 0.0001792, "grad_norm": 0.893126904964447, "learning_rate": 1.1618981816887519e-05, "loss": 2.3924, "step": 175370 }, { "epoch": 0.0002048, "grad_norm": 0.8595192432403564, "learning_rate": 1.1618187156825393e-05, "loss": 2.5238, "step": 175380 }, { "epoch": 0.0002304, "grad_norm": 0.90450519323349, "learning_rate": 1.1617392486269751e-05, "loss": 2.329, "step": 175390 }, { "epoch": 0.000256, "grad_norm": 0.9004784226417542, "learning_rate": 1.161659780522575e-05, "loss": 2.2974, "step": 175400 }, { "epoch": 0.0002816, "grad_norm": 0.9680680632591248, "learning_rate": 1.1615803113698543e-05, "loss": 2.3953, "step": 175410 }, { "epoch": 0.0003072, "grad_norm": 0.8644226789474487, "learning_rate": 1.1615008411693283e-05, "loss": 2.2852, "step": 175420 }, { "epoch": 0.0003328, "grad_norm": 1.0695770978927612, "learning_rate": 1.1614213699215115e-05, "loss": 2.3819, "step": 175430 }, { "epoch": 0.0003584, "grad_norm": 0.9714519381523132, "learning_rate": 1.1613418976269208e-05, "loss": 2.5184, "step": 175440 }, { "epoch": 0.000384, "grad_norm": 0.9341350793838501, "learning_rate": 1.1612624242860706e-05, "loss": 2.3992, "step": 175450 }, { "epoch": 0.0004096, "grad_norm": 1.153281331062317, "learning_rate": 1.1611829498994765e-05, "loss": 2.614, "step": 175460 }, { "epoch": 0.0004352, "grad_norm": 1.0476542711257935, "learning_rate": 1.1611034744676535e-05, "loss": 2.5782, "step": 175470 }, { "epoch": 0.0004608, "grad_norm": 0.9195758104324341, "learning_rate": 1.1610239979911178e-05, "loss": 2.4299, "step": 175480 }, { "epoch": 0.0004864, "grad_norm": 1.012023687362671, "learning_rate": 1.1609445204703837e-05, "loss": 2.3225, "step": 175490 }, { "epoch": 0.000512, "grad_norm": 1.006791114807129, "learning_rate": 1.1608650419059675e-05, "loss": 2.2874, "step": 175500 }, { "epoch": 0.0005376, "grad_norm": 1.0289982557296753, "learning_rate": 1.1607855622983842e-05, "loss": 2.4137, "step": 175510 }, { "epoch": 0.0005632, "grad_norm": 0.9047911763191223, "learning_rate": 1.160706081648149e-05, "loss": 2.3862, "step": 175520 }, { "epoch": 0.0005888, "grad_norm": 0.8827734589576721, "learning_rate": 1.1606265999557777e-05, "loss": 2.3696, "step": 175530 }, { "epoch": 0.0006144, "grad_norm": 0.9718815684318542, "learning_rate": 1.160547117221786e-05, "loss": 2.3743, "step": 175540 }, { "epoch": 0.00064, "grad_norm": 0.9412439465522766, "learning_rate": 1.1604676334466884e-05, "loss": 2.3608, "step": 175550 }, { "epoch": 0.0006656, "grad_norm": 0.9554172158241272, "learning_rate": 1.1603881486310011e-05, "loss": 2.5148, "step": 175560 }, { "epoch": 0.0006912, "grad_norm": 0.9697608351707458, "learning_rate": 1.1603086627752391e-05, "loss": 2.4093, "step": 175570 }, { "epoch": 0.0007168, "grad_norm": 1.035419225692749, "learning_rate": 1.1602291758799185e-05, "loss": 2.3273, "step": 175580 }, { "epoch": 0.0007424, "grad_norm": 1.0409835577011108, "learning_rate": 1.1601496879455539e-05, "loss": 2.5481, "step": 175590 }, { "epoch": 0.000768, "grad_norm": 0.8813816905021667, "learning_rate": 1.1600701989726611e-05, "loss": 2.4458, "step": 175600 }, { "epoch": 0.0007936, "grad_norm": 0.9034520983695984, "learning_rate": 1.1599907089617554e-05, "loss": 2.3875, "step": 175610 }, { "epoch": 0.0008192, "grad_norm": 0.8760473132133484, "learning_rate": 1.159911217913353e-05, "loss": 2.4109, "step": 175620 }, { "epoch": 0.0008448, "grad_norm": 1.033850073814392, "learning_rate": 1.1598317258279684e-05, "loss": 2.4689, "step": 175630 }, { "epoch": 0.0008704, "grad_norm": 0.9730675220489502, "learning_rate": 1.1597522327061178e-05, "loss": 2.4313, "step": 175640 }, { "epoch": 0.000896, "grad_norm": 0.9245584011077881, "learning_rate": 1.1596727385483164e-05, "loss": 2.3872, "step": 175650 }, { "epoch": 0.0009216, "grad_norm": 0.8912810683250427, "learning_rate": 1.1595932433550796e-05, "loss": 2.2064, "step": 175660 }, { "epoch": 0.0009472, "grad_norm": 0.9687070250511169, "learning_rate": 1.1595137471269231e-05, "loss": 2.3914, "step": 175670 }, { "epoch": 0.0009728, "grad_norm": 0.8118566870689392, "learning_rate": 1.1594342498643626e-05, "loss": 2.4505, "step": 175680 }, { "epoch": 0.0009984, "grad_norm": 0.9180454015731812, "learning_rate": 1.1593547515679129e-05, "loss": 2.2843, "step": 175690 }, { "epoch": 0.001024, "grad_norm": 0.9666879177093506, "learning_rate": 1.1592752522380903e-05, "loss": 2.2516, "step": 175700 }, { "epoch": 0.0010496, "grad_norm": 0.8921622633934021, "learning_rate": 1.15919575187541e-05, "loss": 2.2204, "step": 175710 }, { "epoch": 0.0010752, "grad_norm": 0.9526209831237793, "learning_rate": 1.1591162504803872e-05, "loss": 2.5717, "step": 175720 }, { "epoch": 0.0011008, "grad_norm": 0.8725219964981079, "learning_rate": 1.1590367480535382e-05, "loss": 2.5183, "step": 175730 }, { "epoch": 0.0011264, "grad_norm": 0.9775106310844421, "learning_rate": 1.1589572445953784e-05, "loss": 2.5006, "step": 175740 }, { "epoch": 0.001152, "grad_norm": 1.012105107307434, "learning_rate": 1.1588777401064224e-05, "loss": 2.3763, "step": 175750 }, { "epoch": 0.0011776, "grad_norm": 0.9871156811714172, "learning_rate": 1.158798234587187e-05, "loss": 2.3145, "step": 175760 }, { "epoch": 0.0012032, "grad_norm": 0.9501256942749023, "learning_rate": 1.158718728038187e-05, "loss": 2.407, "step": 175770 }, { "epoch": 0.0012288, "grad_norm": 0.9148171544075012, "learning_rate": 1.1586392204599382e-05, "loss": 2.5082, "step": 175780 }, { "epoch": 0.0012544, "grad_norm": 0.9660540223121643, "learning_rate": 1.1585597118529564e-05, "loss": 2.5336, "step": 175790 }, { "epoch": 0.00128, "grad_norm": 1.0217968225479126, "learning_rate": 1.158480202217757e-05, "loss": 2.5234, "step": 175800 }, { "epoch": 0.0013056, "grad_norm": 1.0155316591262817, "learning_rate": 1.1584006915548556e-05, "loss": 2.3947, "step": 175810 }, { "epoch": 0.0013312, "grad_norm": 0.962711751461029, "learning_rate": 1.1583211798647677e-05, "loss": 2.5537, "step": 175820 }, { "epoch": 0.0013568, "grad_norm": 0.9394251108169556, "learning_rate": 1.158241667148009e-05, "loss": 2.4224, "step": 175830 }, { "epoch": 0.0013824, "grad_norm": 0.8980616331100464, "learning_rate": 1.1581621534050955e-05, "loss": 2.3876, "step": 175840 }, { "epoch": 0.001408, "grad_norm": 0.8876811861991882, "learning_rate": 1.1580826386365423e-05, "loss": 2.3635, "step": 175850 }, { "epoch": 0.0014336, "grad_norm": 0.9525896310806274, "learning_rate": 1.1580031228428647e-05, "loss": 2.5277, "step": 175860 }, { "epoch": 0.0014592, "grad_norm": 0.9788523316383362, "learning_rate": 1.1579236060245794e-05, "loss": 2.5277, "step": 175870 }, { "epoch": 0.0014848, "grad_norm": 0.9248085618019104, "learning_rate": 1.1578440881822014e-05, "loss": 2.5198, "step": 175880 }, { "epoch": 0.0015104, "grad_norm": 0.9027215242385864, "learning_rate": 1.1577645693162462e-05, "loss": 2.2802, "step": 175890 }, { "epoch": 0.001536, "grad_norm": 0.8721632361412048, "learning_rate": 1.1576850494272301e-05, "loss": 2.3861, "step": 175900 }, { "epoch": 0.0015616, "grad_norm": 0.9930867552757263, "learning_rate": 1.157605528515668e-05, "loss": 2.2329, "step": 175910 }, { "epoch": 0.0015872, "grad_norm": 0.9851894378662109, "learning_rate": 1.1575260065820762e-05, "loss": 2.3765, "step": 175920 }, { "epoch": 0.0016128, "grad_norm": 0.9744214415550232, "learning_rate": 1.1574464836269698e-05, "loss": 2.437, "step": 175930 }, { "epoch": 0.0016384, "grad_norm": 1.0185705423355103, "learning_rate": 1.1573669596508653e-05, "loss": 2.6541, "step": 175940 }, { "epoch": 0.001664, "grad_norm": 0.9564975500106812, "learning_rate": 1.1572874346542772e-05, "loss": 2.5508, "step": 175950 }, { "epoch": 0.0016896, "grad_norm": 0.9020068049430847, "learning_rate": 1.1572079086377221e-05, "loss": 2.5146, "step": 175960 }, { "epoch": 0.0017152, "grad_norm": 0.9447435140609741, "learning_rate": 1.1571283816017158e-05, "loss": 2.4041, "step": 175970 }, { "epoch": 0.0017408, "grad_norm": 0.9478213787078857, "learning_rate": 1.1570488535467737e-05, "loss": 2.5156, "step": 175980 }, { "epoch": 0.0017664, "grad_norm": 1.0221803188323975, "learning_rate": 1.1569693244734113e-05, "loss": 2.5817, "step": 175990 }, { "epoch": 0.001792, "grad_norm": 5.273014068603516, "learning_rate": 1.1568897943821446e-05, "loss": 2.6907, "step": 176000 }, { "epoch": 2.56e-05, "grad_norm": 1.008777141571045, "learning_rate": 1.1568102632734895e-05, "loss": 2.1469, "step": 176010 }, { "epoch": 5.12e-05, "grad_norm": 0.9457123875617981, "learning_rate": 1.1567307311479611e-05, "loss": 2.8383, "step": 176020 }, { "epoch": 7.68e-05, "grad_norm": 0.9373971819877625, "learning_rate": 1.1566511980060758e-05, "loss": 2.8955, "step": 176030 }, { "epoch": 0.0001024, "grad_norm": 0.8316510915756226, "learning_rate": 1.156571663848349e-05, "loss": 2.7587, "step": 176040 }, { "epoch": 0.000128, "grad_norm": 0.988446056842804, "learning_rate": 1.1564921286752968e-05, "loss": 2.9554, "step": 176050 }, { "epoch": 0.0001536, "grad_norm": 0.9288784265518188, "learning_rate": 1.1564125924874345e-05, "loss": 2.9345, "step": 176060 }, { "epoch": 0.0001792, "grad_norm": 0.9555906057357788, "learning_rate": 1.1563330552852782e-05, "loss": 2.8153, "step": 176070 }, { "epoch": 0.0002048, "grad_norm": 1.2099683284759521, "learning_rate": 1.1562535170693438e-05, "loss": 3.0632, "step": 176080 }, { "epoch": 0.0002304, "grad_norm": 0.938945472240448, "learning_rate": 1.1561739778401466e-05, "loss": 2.9628, "step": 176090 }, { "epoch": 0.000256, "grad_norm": 1.059974193572998, "learning_rate": 1.1560944375982028e-05, "loss": 2.9333, "step": 176100 }, { "epoch": 0.0002816, "grad_norm": 1.1185914278030396, "learning_rate": 1.1560148963440281e-05, "loss": 2.6055, "step": 176110 }, { "epoch": 0.0003072, "grad_norm": 1.0914720296859741, "learning_rate": 1.1559353540781378e-05, "loss": 2.741, "step": 176120 }, { "epoch": 0.0003328, "grad_norm": 0.9048447012901306, "learning_rate": 1.1558558108010487e-05, "loss": 2.9426, "step": 176130 }, { "epoch": 0.0003584, "grad_norm": 0.8376767039299011, "learning_rate": 1.155776266513276e-05, "loss": 2.7728, "step": 176140 }, { "epoch": 0.000384, "grad_norm": 0.9336244463920593, "learning_rate": 1.1556967212153353e-05, "loss": 2.7574, "step": 176150 }, { "epoch": 0.0004096, "grad_norm": 0.8681533932685852, "learning_rate": 1.155617174907743e-05, "loss": 2.6917, "step": 176160 }, { "epoch": 0.0004352, "grad_norm": 0.8507768511772156, "learning_rate": 1.1555376275910148e-05, "loss": 2.717, "step": 176170 }, { "epoch": 0.0004608, "grad_norm": 1.0348881483078003, "learning_rate": 1.1554580792656662e-05, "loss": 2.8117, "step": 176180 }, { "epoch": 0.0004864, "grad_norm": 0.8441411852836609, "learning_rate": 1.1553785299322133e-05, "loss": 2.699, "step": 176190 }, { "epoch": 0.000512, "grad_norm": 0.8167259693145752, "learning_rate": 1.155298979591172e-05, "loss": 2.9778, "step": 176200 }, { "epoch": 0.0005376, "grad_norm": 0.8469694256782532, "learning_rate": 1.155219428243058e-05, "loss": 2.606, "step": 176210 }, { "epoch": 0.0005632, "grad_norm": 0.8618794083595276, "learning_rate": 1.1551398758883872e-05, "loss": 2.7015, "step": 176220 }, { "epoch": 0.0005888, "grad_norm": 0.9252408146858215, "learning_rate": 1.1550603225276758e-05, "loss": 2.8085, "step": 176230 }, { "epoch": 0.0006144, "grad_norm": 0.9508523344993591, "learning_rate": 1.154980768161439e-05, "loss": 2.9342, "step": 176240 }, { "epoch": 0.00064, "grad_norm": 0.8473820686340332, "learning_rate": 1.1549012127901935e-05, "loss": 2.9329, "step": 176250 }, { "epoch": 0.0006656, "grad_norm": 0.8013142943382263, "learning_rate": 1.1548216564144547e-05, "loss": 3.3365, "step": 176260 }, { "epoch": 0.0006912, "grad_norm": 0.7918533086776733, "learning_rate": 1.1547420990347387e-05, "loss": 2.7901, "step": 176270 }, { "epoch": 0.0007168, "grad_norm": 0.8693552017211914, "learning_rate": 1.1546625406515613e-05, "loss": 2.6394, "step": 176280 }, { "epoch": 0.0007424, "grad_norm": 0.9772925972938538, "learning_rate": 1.1545829812654382e-05, "loss": 3.0947, "step": 176290 }, { "epoch": 0.000768, "grad_norm": 0.8584876656532288, "learning_rate": 1.1545034208768854e-05, "loss": 2.8378, "step": 176300 }, { "epoch": 0.0007936, "grad_norm": 0.8287798762321472, "learning_rate": 1.1544238594864194e-05, "loss": 2.9628, "step": 176310 }, { "epoch": 0.0008192, "grad_norm": 0.8738986849784851, "learning_rate": 1.1543442970945555e-05, "loss": 2.6714, "step": 176320 }, { "epoch": 0.0008448, "grad_norm": 0.948123037815094, "learning_rate": 1.15426473370181e-05, "loss": 2.6222, "step": 176330 }, { "epoch": 0.0008704, "grad_norm": 0.8124473690986633, "learning_rate": 1.1541851693086987e-05, "loss": 2.8973, "step": 176340 }, { "epoch": 0.000896, "grad_norm": 1.9979033470153809, "learning_rate": 1.1541056039157374e-05, "loss": 3.0553, "step": 176350 }, { "epoch": 0.0009216, "grad_norm": 0.8267486691474915, "learning_rate": 1.1540260375234423e-05, "loss": 2.8835, "step": 176360 }, { "epoch": 0.0009472, "grad_norm": 0.9142788052558899, "learning_rate": 1.1539464701323296e-05, "loss": 2.8557, "step": 176370 }, { "epoch": 0.0009728, "grad_norm": 1.1086044311523438, "learning_rate": 1.153866901742914e-05, "loss": 2.7259, "step": 176380 }, { "epoch": 0.0009984, "grad_norm": 0.7773061990737915, "learning_rate": 1.153787332355713e-05, "loss": 2.6868, "step": 176390 }, { "epoch": 0.001024, "grad_norm": 0.73861163854599, "learning_rate": 1.1537077619712425e-05, "loss": 2.5533, "step": 176400 }, { "epoch": 0.0010496, "grad_norm": 0.7726515531539917, "learning_rate": 1.1536281905900175e-05, "loss": 2.6867, "step": 176410 }, { "epoch": 0.0010752, "grad_norm": 0.8464676737785339, "learning_rate": 1.1535486182125546e-05, "loss": 2.8575, "step": 176420 }, { "epoch": 0.0011008, "grad_norm": 0.7673640251159668, "learning_rate": 1.15346904483937e-05, "loss": 2.6557, "step": 176430 }, { "epoch": 0.0011264, "grad_norm": 0.8242387771606445, "learning_rate": 1.1533894704709789e-05, "loss": 2.7503, "step": 176440 }, { "epoch": 0.001152, "grad_norm": 0.7906450033187866, "learning_rate": 1.1533098951078982e-05, "loss": 3.0004, "step": 176450 }, { "epoch": 0.0011776, "grad_norm": 0.9540255665779114, "learning_rate": 1.1532303187506435e-05, "loss": 2.9717, "step": 176460 }, { "epoch": 0.0012032, "grad_norm": 0.832810640335083, "learning_rate": 1.1531507413997308e-05, "loss": 3.2768, "step": 176470 }, { "epoch": 0.0012288, "grad_norm": 0.7281011939048767, "learning_rate": 1.1530711630556762e-05, "loss": 2.7437, "step": 176480 }, { "epoch": 0.0012544, "grad_norm": 0.6871960163116455, "learning_rate": 1.152991583718996e-05, "loss": 2.6252, "step": 176490 }, { "epoch": 0.00128, "grad_norm": 0.8951466679573059, "learning_rate": 1.1529120033902061e-05, "loss": 2.9861, "step": 176500 }, { "epoch": 0.0013056, "grad_norm": 0.7605875730514526, "learning_rate": 1.1528324220698224e-05, "loss": 2.8443, "step": 176510 }, { "epoch": 0.0013312, "grad_norm": 0.9019201993942261, "learning_rate": 1.1527528397583607e-05, "loss": 2.6128, "step": 176520 }, { "epoch": 0.0013568, "grad_norm": 0.8205854296684265, "learning_rate": 1.152673256456338e-05, "loss": 3.0472, "step": 176530 }, { "epoch": 0.0013824, "grad_norm": 0.8199158310890198, "learning_rate": 1.1525936721642697e-05, "loss": 3.0232, "step": 176540 }, { "epoch": 0.001408, "grad_norm": 0.7757709622383118, "learning_rate": 1.1525140868826716e-05, "loss": 2.8535, "step": 176550 }, { "epoch": 0.0014336, "grad_norm": 0.7929774522781372, "learning_rate": 1.1524345006120605e-05, "loss": 2.8863, "step": 176560 }, { "epoch": 0.0014592, "grad_norm": 0.7571427226066589, "learning_rate": 1.1523549133529521e-05, "loss": 2.9099, "step": 176570 }, { "epoch": 0.0014848, "grad_norm": 0.8591910004615784, "learning_rate": 1.1522753251058625e-05, "loss": 2.5448, "step": 176580 }, { "epoch": 0.0015104, "grad_norm": 0.8047811985015869, "learning_rate": 1.1521957358713081e-05, "loss": 2.3106, "step": 176590 }, { "epoch": 0.001536, "grad_norm": 1.0384327173233032, "learning_rate": 1.1521161456498047e-05, "loss": 2.6868, "step": 176600 }, { "epoch": 0.0015616, "grad_norm": 0.931768536567688, "learning_rate": 1.1520365544418684e-05, "loss": 2.84, "step": 176610 }, { "epoch": 0.0015872, "grad_norm": 0.807777464389801, "learning_rate": 1.1519569622480155e-05, "loss": 2.9702, "step": 176620 }, { "epoch": 0.0016128, "grad_norm": 0.9602682590484619, "learning_rate": 1.1518773690687623e-05, "loss": 2.8653, "step": 176630 }, { "epoch": 0.0016384, "grad_norm": 0.8369147777557373, "learning_rate": 1.1517977749046246e-05, "loss": 2.7105, "step": 176640 }, { "epoch": 0.001664, "grad_norm": 0.781491219997406, "learning_rate": 1.1517181797561183e-05, "loss": 2.8642, "step": 176650 }, { "epoch": 0.0016896, "grad_norm": 0.8577233552932739, "learning_rate": 1.1516385836237605e-05, "loss": 3.1278, "step": 176660 }, { "epoch": 0.0017152, "grad_norm": 0.8410885334014893, "learning_rate": 1.1515589865080665e-05, "loss": 2.7793, "step": 176670 }, { "epoch": 0.0017408, "grad_norm": 0.8590471744537354, "learning_rate": 1.1514793884095527e-05, "loss": 2.9511, "step": 176680 }, { "epoch": 0.0017664, "grad_norm": 0.7953317165374756, "learning_rate": 1.1513997893287357e-05, "loss": 2.4851, "step": 176690 }, { "epoch": 0.001792, "grad_norm": 0.7878415584564209, "learning_rate": 1.151320189266131e-05, "loss": 2.7152, "step": 176700 }, { "epoch": 0.0018176, "grad_norm": 0.9111788868904114, "learning_rate": 1.151240588222255e-05, "loss": 2.6845, "step": 176710 }, { "epoch": 0.0018432, "grad_norm": 0.7325723171234131, "learning_rate": 1.1511609861976241e-05, "loss": 2.5689, "step": 176720 }, { "epoch": 0.0018688, "grad_norm": 1.2923541069030762, "learning_rate": 1.1510813831927542e-05, "loss": 2.1461, "step": 176730 }, { "epoch": 0.0018944, "grad_norm": 1.1274945735931396, "learning_rate": 1.1510017792081618e-05, "loss": 2.986, "step": 176740 }, { "epoch": 0.00192, "grad_norm": 0.8578786849975586, "learning_rate": 1.150922174244363e-05, "loss": 2.8043, "step": 176750 }, { "epoch": 0.0019456, "grad_norm": 0.7264650464057922, "learning_rate": 1.1508425683018742e-05, "loss": 2.6933, "step": 176760 }, { "epoch": 0.0019712, "grad_norm": 0.6953710913658142, "learning_rate": 1.1507629613812113e-05, "loss": 2.7579, "step": 176770 }, { "epoch": 0.0019968, "grad_norm": 0.7766081690788269, "learning_rate": 1.1506833534828905e-05, "loss": 2.6507, "step": 176780 }, { "epoch": 0.0020224, "grad_norm": 0.8562648892402649, "learning_rate": 1.1506037446074285e-05, "loss": 2.9689, "step": 176790 }, { "epoch": 0.002048, "grad_norm": 0.7877646088600159, "learning_rate": 1.150524134755341e-05, "loss": 2.8398, "step": 176800 }, { "epoch": 0.0020736, "grad_norm": 0.8743738532066345, "learning_rate": 1.1504445239271447e-05, "loss": 2.7455, "step": 176810 }, { "epoch": 0.0020992, "grad_norm": 0.8038202524185181, "learning_rate": 1.1503649121233556e-05, "loss": 2.9193, "step": 176820 }, { "epoch": 0.0021248, "grad_norm": 0.8440250158309937, "learning_rate": 1.1502852993444898e-05, "loss": 2.7444, "step": 176830 }, { "epoch": 0.0021504, "grad_norm": 0.8175899386405945, "learning_rate": 1.150205685591064e-05, "loss": 2.6921, "step": 176840 }, { "epoch": 0.002176, "grad_norm": 0.8362565636634827, "learning_rate": 1.150126070863594e-05, "loss": 3.1246, "step": 176850 }, { "epoch": 0.0022016, "grad_norm": 0.8424209356307983, "learning_rate": 1.1500464551625968e-05, "loss": 2.9107, "step": 176860 }, { "epoch": 0.0022272, "grad_norm": 0.7597900629043579, "learning_rate": 1.1499668384885878e-05, "loss": 2.6149, "step": 176870 }, { "epoch": 0.0022528, "grad_norm": 0.8801723718643188, "learning_rate": 1.1498872208420837e-05, "loss": 3.3129, "step": 176880 }, { "epoch": 0.0022784, "grad_norm": 0.781288743019104, "learning_rate": 1.1498076022236011e-05, "loss": 2.7205, "step": 176890 }, { "epoch": 0.002304, "grad_norm": 0.8333078622817993, "learning_rate": 1.1497279826336558e-05, "loss": 2.9136, "step": 176900 }, { "epoch": 0.0023296, "grad_norm": 0.9351137280464172, "learning_rate": 1.149648362072764e-05, "loss": 2.9192, "step": 176910 }, { "epoch": 0.0023552, "grad_norm": 0.7798454761505127, "learning_rate": 1.1495687405414431e-05, "loss": 2.653, "step": 176920 }, { "epoch": 0.0023808, "grad_norm": 1.0446890592575073, "learning_rate": 1.1494891180402081e-05, "loss": 2.7454, "step": 176930 }, { "epoch": 0.0024064, "grad_norm": 0.7280722260475159, "learning_rate": 1.149409494569576e-05, "loss": 2.8129, "step": 176940 }, { "epoch": 0.002432, "grad_norm": 0.8016515970230103, "learning_rate": 1.149329870130063e-05, "loss": 2.9746, "step": 176950 }, { "epoch": 0.0024576, "grad_norm": 1.8866057395935059, "learning_rate": 1.1492502447221859e-05, "loss": 3.0591, "step": 176960 }, { "epoch": 0.0024832, "grad_norm": 0.8529635071754456, "learning_rate": 1.1491706183464602e-05, "loss": 2.9636, "step": 176970 }, { "epoch": 0.0025088, "grad_norm": 0.7191368341445923, "learning_rate": 1.1490909910034026e-05, "loss": 3.069, "step": 176980 }, { "epoch": 0.0025344, "grad_norm": 0.784189760684967, "learning_rate": 1.14901136269353e-05, "loss": 3.067, "step": 176990 }, { "epoch": 0.00256, "grad_norm": 0.7635249495506287, "learning_rate": 1.148931733417358e-05, "loss": 2.9111, "step": 177000 }, { "epoch": 0.0025856, "grad_norm": 0.794507622718811, "learning_rate": 1.1488521031754031e-05, "loss": 2.7276, "step": 177010 }, { "epoch": 0.0026112, "grad_norm": 0.8655429482460022, "learning_rate": 1.1487724719681826e-05, "loss": 3.0366, "step": 177020 }, { "epoch": 0.0026368, "grad_norm": 0.8748533725738525, "learning_rate": 1.1486928397962113e-05, "loss": 2.996, "step": 177030 }, { "epoch": 0.0026624, "grad_norm": 0.7133491635322571, "learning_rate": 1.148613206660007e-05, "loss": 2.8967, "step": 177040 }, { "epoch": 0.002688, "grad_norm": 0.8087611198425293, "learning_rate": 1.1485335725600853e-05, "loss": 3.028, "step": 177050 }, { "epoch": 0.0027136, "grad_norm": 0.7971720695495605, "learning_rate": 1.1484539374969632e-05, "loss": 2.9874, "step": 177060 }, { "epoch": 0.0027392, "grad_norm": 0.8992428183555603, "learning_rate": 1.1483743014711564e-05, "loss": 3.0238, "step": 177070 }, { "epoch": 0.0027648, "grad_norm": 0.9719247817993164, "learning_rate": 1.1482946644831816e-05, "loss": 2.811, "step": 177080 }, { "epoch": 0.0027904, "grad_norm": 1.3922507762908936, "learning_rate": 1.1482150265335556e-05, "loss": 2.7357, "step": 177090 }, { "epoch": 0.002816, "grad_norm": 0.8279901742935181, "learning_rate": 1.1481353876227945e-05, "loss": 3.021, "step": 177100 }, { "epoch": 0.0028416, "grad_norm": 0.8018562197685242, "learning_rate": 1.1480557477514147e-05, "loss": 2.8225, "step": 177110 }, { "epoch": 0.0028672, "grad_norm": 0.807930588722229, "learning_rate": 1.1479761069199329e-05, "loss": 2.9837, "step": 177120 }, { "epoch": 0.0028928, "grad_norm": 0.7792941927909851, "learning_rate": 1.147896465128865e-05, "loss": 3.1236, "step": 177130 }, { "epoch": 0.0029184, "grad_norm": 0.7632566094398499, "learning_rate": 1.1478168223787282e-05, "loss": 3.1955, "step": 177140 }, { "epoch": 0.002944, "grad_norm": 0.8141444325447083, "learning_rate": 1.1477371786700385e-05, "loss": 2.9618, "step": 177150 }, { "epoch": 0.0029696, "grad_norm": 0.7851940393447876, "learning_rate": 1.1476575340033124e-05, "loss": 2.9072, "step": 177160 }, { "epoch": 0.0029952, "grad_norm": 0.7762718796730042, "learning_rate": 1.1475778883790663e-05, "loss": 2.9026, "step": 177170 }, { "epoch": 0.0030208, "grad_norm": 0.8402253985404968, "learning_rate": 1.147498241797817e-05, "loss": 3.0343, "step": 177180 }, { "epoch": 0.0030464, "grad_norm": 0.7681818604469299, "learning_rate": 1.1474185942600808e-05, "loss": 3.1164, "step": 177190 }, { "epoch": 0.003072, "grad_norm": 0.8411984443664551, "learning_rate": 1.1473389457663742e-05, "loss": 2.9266, "step": 177200 }, { "epoch": 0.0030976, "grad_norm": 0.9086307883262634, "learning_rate": 1.1472592963172136e-05, "loss": 2.9256, "step": 177210 }, { "epoch": 0.0031232, "grad_norm": 1.4148894548416138, "learning_rate": 1.147179645913116e-05, "loss": 3.0502, "step": 177220 }, { "epoch": 0.0031488, "grad_norm": 0.7943236231803894, "learning_rate": 1.147099994554597e-05, "loss": 2.8994, "step": 177230 }, { "epoch": 0.0031744, "grad_norm": 1.6876612901687622, "learning_rate": 1.1470203422421737e-05, "loss": 3.0269, "step": 177240 }, { "epoch": 0.0032, "grad_norm": 1.9043986797332764, "learning_rate": 1.1469406889763629e-05, "loss": 2.8086, "step": 177250 }, { "epoch": 0.0032256, "grad_norm": 0.9173686504364014, "learning_rate": 1.1468610347576803e-05, "loss": 3.0153, "step": 177260 }, { "epoch": 0.0032512, "grad_norm": 0.7987806797027588, "learning_rate": 1.146781379586643e-05, "loss": 3.2017, "step": 177270 }, { "epoch": 0.0032768, "grad_norm": 0.8830957412719727, "learning_rate": 1.1467017234637676e-05, "loss": 2.8154, "step": 177280 }, { "epoch": 0.0033024, "grad_norm": 0.8444089889526367, "learning_rate": 1.1466220663895708e-05, "loss": 3.1515, "step": 177290 }, { "epoch": 0.003328, "grad_norm": 0.8027430772781372, "learning_rate": 1.1465424083645687e-05, "loss": 3.1449, "step": 177300 }, { "epoch": 0.0033536, "grad_norm": 0.7790316343307495, "learning_rate": 1.1464627493892778e-05, "loss": 2.8419, "step": 177310 }, { "epoch": 0.0033792, "grad_norm": 0.8097103238105774, "learning_rate": 1.1463830894642152e-05, "loss": 2.8971, "step": 177320 }, { "epoch": 0.0034048, "grad_norm": 0.7750105261802673, "learning_rate": 1.1463034285898968e-05, "loss": 2.8653, "step": 177330 }, { "epoch": 0.0034304, "grad_norm": 0.7911513447761536, "learning_rate": 1.1462237667668396e-05, "loss": 2.8635, "step": 177340 }, { "epoch": 0.003456, "grad_norm": 0.8196858167648315, "learning_rate": 1.1461441039955605e-05, "loss": 3.1156, "step": 177350 }, { "epoch": 0.0034816, "grad_norm": 0.7451420426368713, "learning_rate": 1.1460644402765754e-05, "loss": 2.8251, "step": 177360 }, { "epoch": 0.0035072, "grad_norm": 0.7407277226448059, "learning_rate": 1.1459847756104013e-05, "loss": 2.8586, "step": 177370 }, { "epoch": 0.0035328, "grad_norm": 0.8677040934562683, "learning_rate": 1.1459051099975547e-05, "loss": 2.9473, "step": 177380 }, { "epoch": 0.0035584, "grad_norm": 0.8061270117759705, "learning_rate": 1.1458254434385525e-05, "loss": 2.9442, "step": 177390 }, { "epoch": 0.003584, "grad_norm": 0.8415855169296265, "learning_rate": 1.1457457759339109e-05, "loss": 2.9119, "step": 177400 }, { "epoch": 0.0036096, "grad_norm": 0.8742506504058838, "learning_rate": 1.1456661074841465e-05, "loss": 3.1597, "step": 177410 }, { "epoch": 0.0036352, "grad_norm": 0.7169381380081177, "learning_rate": 1.1455864380897763e-05, "loss": 3.1731, "step": 177420 }, { "epoch": 0.0036608, "grad_norm": 0.8143844604492188, "learning_rate": 1.1455067677513165e-05, "loss": 3.1131, "step": 177430 }, { "epoch": 0.0036864, "grad_norm": 1.047279715538025, "learning_rate": 1.1454270964692841e-05, "loss": 2.9814, "step": 177440 }, { "epoch": 0.003712, "grad_norm": 0.6729635000228882, "learning_rate": 1.145347424244196e-05, "loss": 2.9688, "step": 177450 }, { "epoch": 0.0037376, "grad_norm": 0.8404088020324707, "learning_rate": 1.1452677510765682e-05, "loss": 2.8747, "step": 177460 }, { "epoch": 0.0037632, "grad_norm": 0.7161473631858826, "learning_rate": 1.1451880769669174e-05, "loss": 2.7992, "step": 177470 }, { "epoch": 0.0037888, "grad_norm": 0.815553605556488, "learning_rate": 1.1451084019157608e-05, "loss": 3.0756, "step": 177480 }, { "epoch": 0.0038144, "grad_norm": 0.8041364550590515, "learning_rate": 1.1450287259236148e-05, "loss": 2.9202, "step": 177490 }, { "epoch": 0.00384, "grad_norm": 0.7528536915779114, "learning_rate": 1.1449490489909957e-05, "loss": 2.9493, "step": 177500 }, { "epoch": 0.0038656, "grad_norm": 0.8539830446243286, "learning_rate": 1.144869371118421e-05, "loss": 2.833, "step": 177510 }, { "epoch": 0.0038912, "grad_norm": 0.8902786374092102, "learning_rate": 1.1447896923064066e-05, "loss": 2.9925, "step": 177520 }, { "epoch": 0.0039168, "grad_norm": 0.9566497206687927, "learning_rate": 1.1447100125554695e-05, "loss": 3.0997, "step": 177530 }, { "epoch": 0.0039424, "grad_norm": 1.06929349899292, "learning_rate": 1.1446303318661266e-05, "loss": 2.8442, "step": 177540 }, { "epoch": 0.003968, "grad_norm": 0.7976888418197632, "learning_rate": 1.1445506502388945e-05, "loss": 2.9381, "step": 177550 }, { "epoch": 0.0039936, "grad_norm": 0.9437931180000305, "learning_rate": 1.1444709676742898e-05, "loss": 2.9614, "step": 177560 }, { "epoch": 0.0040192, "grad_norm": 0.7463863492012024, "learning_rate": 1.1443912841728293e-05, "loss": 3.1267, "step": 177570 }, { "epoch": 0.0040448, "grad_norm": 0.8366419076919556, "learning_rate": 1.1443115997350297e-05, "loss": 3.0532, "step": 177580 }, { "epoch": 0.0040704, "grad_norm": 0.7611911296844482, "learning_rate": 1.1442319143614078e-05, "loss": 2.9877, "step": 177590 }, { "epoch": 0.004096, "grad_norm": 0.7974886298179626, "learning_rate": 1.1441522280524797e-05, "loss": 2.9228, "step": 177600 }, { "epoch": 0.0041216, "grad_norm": 0.8009863495826721, "learning_rate": 1.1440725408087633e-05, "loss": 2.9347, "step": 177610 }, { "epoch": 0.0041472, "grad_norm": 0.783744215965271, "learning_rate": 1.1439928526307746e-05, "loss": 2.8611, "step": 177620 }, { "epoch": 0.0041728, "grad_norm": 0.7994371056556702, "learning_rate": 1.1439131635190305e-05, "loss": 2.9542, "step": 177630 }, { "epoch": 0.0041984, "grad_norm": 0.8095774054527283, "learning_rate": 1.1438334734740476e-05, "loss": 2.9344, "step": 177640 }, { "epoch": 0.004224, "grad_norm": 0.8835508823394775, "learning_rate": 1.1437537824963434e-05, "loss": 3.1842, "step": 177650 }, { "epoch": 0.0042496, "grad_norm": 0.8983892202377319, "learning_rate": 1.1436740905864338e-05, "loss": 3.1097, "step": 177660 }, { "epoch": 0.0042752, "grad_norm": 0.8262011408805847, "learning_rate": 1.1435943977448358e-05, "loss": 2.9057, "step": 177670 }, { "epoch": 0.0043008, "grad_norm": 0.7563495635986328, "learning_rate": 1.1435147039720667e-05, "loss": 2.9996, "step": 177680 }, { "epoch": 0.0043264, "grad_norm": 0.7974626421928406, "learning_rate": 1.1434350092686421e-05, "loss": 2.9606, "step": 177690 }, { "epoch": 0.004352, "grad_norm": 0.7527391314506531, "learning_rate": 1.1433553136350802e-05, "loss": 2.8954, "step": 177700 }, { "epoch": 0.0043776, "grad_norm": 0.8013245463371277, "learning_rate": 1.143275617071897e-05, "loss": 3.0099, "step": 177710 }, { "epoch": 0.0044032, "grad_norm": 0.8253337144851685, "learning_rate": 1.1431959195796097e-05, "loss": 2.945, "step": 177720 }, { "epoch": 0.0044288, "grad_norm": 0.875028133392334, "learning_rate": 1.1431162211587346e-05, "loss": 2.8338, "step": 177730 }, { "epoch": 0.0044544, "grad_norm": 0.8696004748344421, "learning_rate": 1.143036521809789e-05, "loss": 2.6314, "step": 177740 }, { "epoch": 0.00448, "grad_norm": 0.9623648524284363, "learning_rate": 1.1429568215332898e-05, "loss": 2.8122, "step": 177750 }, { "epoch": 0.0045056, "grad_norm": 1.0771230459213257, "learning_rate": 1.1428771203297533e-05, "loss": 3.004, "step": 177760 }, { "epoch": 0.0045312, "grad_norm": 0.9382220506668091, "learning_rate": 1.1427974181996965e-05, "loss": 2.8643, "step": 177770 }, { "epoch": 0.0045568, "grad_norm": 0.7022601962089539, "learning_rate": 1.1427177151436369e-05, "loss": 3.0438, "step": 177780 }, { "epoch": 0.0045824, "grad_norm": 0.8477071523666382, "learning_rate": 1.1426380111620902e-05, "loss": 3.0376, "step": 177790 }, { "epoch": 0.004608, "grad_norm": 0.7384995818138123, "learning_rate": 1.1425583062555744e-05, "loss": 2.9141, "step": 177800 }, { "epoch": 0.0046336, "grad_norm": 0.8447135090827942, "learning_rate": 1.142478600424606e-05, "loss": 3.2367, "step": 177810 }, { "epoch": 0.0046592, "grad_norm": 0.9311054944992065, "learning_rate": 1.1423988936697014e-05, "loss": 2.953, "step": 177820 }, { "epoch": 0.0046848, "grad_norm": 0.7676343321800232, "learning_rate": 1.1423191859913779e-05, "loss": 2.9928, "step": 177830 }, { "epoch": 0.0047104, "grad_norm": 1.3100107908248901, "learning_rate": 1.1422394773901521e-05, "loss": 3.2578, "step": 177840 }, { "epoch": 0.004736, "grad_norm": 0.7464354634284973, "learning_rate": 1.1421597678665418e-05, "loss": 2.8677, "step": 177850 }, { "epoch": 0.0047616, "grad_norm": 1.024067759513855, "learning_rate": 1.1420800574210622e-05, "loss": 2.9792, "step": 177860 }, { "epoch": 0.0047872, "grad_norm": 0.785135805606842, "learning_rate": 1.1420003460542317e-05, "loss": 2.8688, "step": 177870 }, { "epoch": 0.0048128, "grad_norm": 0.8819915056228638, "learning_rate": 1.141920633766567e-05, "loss": 2.8859, "step": 177880 }, { "epoch": 0.0048384, "grad_norm": 0.7956554293632507, "learning_rate": 1.141840920558584e-05, "loss": 2.943, "step": 177890 }, { "epoch": 0.004864, "grad_norm": 1.0591520071029663, "learning_rate": 1.1417612064308009e-05, "loss": 2.8093, "step": 177900 }, { "epoch": 0.0048896, "grad_norm": 0.7729935050010681, "learning_rate": 1.141681491383734e-05, "loss": 3.0303, "step": 177910 }, { "epoch": 0.0049152, "grad_norm": 0.7908639907836914, "learning_rate": 1.1416017754179e-05, "loss": 2.9512, "step": 177920 }, { "epoch": 0.0049408, "grad_norm": 1.0726169347763062, "learning_rate": 1.1415220585338163e-05, "loss": 3.106, "step": 177930 }, { "epoch": 0.0049664, "grad_norm": 0.8336561322212219, "learning_rate": 1.1414423407319997e-05, "loss": 2.8083, "step": 177940 }, { "epoch": 0.004992, "grad_norm": 0.7871261835098267, "learning_rate": 1.141362622012967e-05, "loss": 2.9151, "step": 177950 }, { "epoch": 0.0050176, "grad_norm": 0.711874783039093, "learning_rate": 1.1412829023772352e-05, "loss": 2.9598, "step": 177960 }, { "epoch": 0.0050432, "grad_norm": 0.8384661078453064, "learning_rate": 1.1412031818253212e-05, "loss": 2.895, "step": 177970 }, { "epoch": 0.0050688, "grad_norm": 0.9305762052536011, "learning_rate": 1.1411234603577426e-05, "loss": 2.8951, "step": 177980 }, { "epoch": 0.0050944, "grad_norm": 0.7952817678451538, "learning_rate": 1.1410437379750157e-05, "loss": 2.9697, "step": 177990 }, { "epoch": 0.00512, "grad_norm": 0.8518325686454773, "learning_rate": 1.1409640146776572e-05, "loss": 3.0343, "step": 178000 }, { "epoch": 0.0051456, "grad_norm": 0.7511232495307922, "learning_rate": 1.1408842904661852e-05, "loss": 2.9344, "step": 178010 }, { "epoch": 0.0051712, "grad_norm": 0.9643879532814026, "learning_rate": 1.1408045653411158e-05, "loss": 2.9259, "step": 178020 }, { "epoch": 0.0051968, "grad_norm": 0.751527726650238, "learning_rate": 1.140724839302966e-05, "loss": 2.9982, "step": 178030 }, { "epoch": 0.0052224, "grad_norm": 1.2196993827819824, "learning_rate": 1.1406451123522531e-05, "loss": 2.9913, "step": 178040 }, { "epoch": 0.005248, "grad_norm": 0.9429224729537964, "learning_rate": 1.140565384489494e-05, "loss": 3.1689, "step": 178050 }, { "epoch": 0.0052736, "grad_norm": 0.7430213093757629, "learning_rate": 1.1404856557152055e-05, "loss": 2.8897, "step": 178060 }, { "epoch": 0.0052992, "grad_norm": 1.0108846426010132, "learning_rate": 1.1404059260299052e-05, "loss": 2.8266, "step": 178070 }, { "epoch": 0.0053248, "grad_norm": 0.8778771162033081, "learning_rate": 1.1403261954341098e-05, "loss": 3.1718, "step": 178080 }, { "epoch": 0.0053504, "grad_norm": 1.0193414688110352, "learning_rate": 1.1402464639283361e-05, "loss": 3.0165, "step": 178090 }, { "epoch": 0.005376, "grad_norm": 0.7050647139549255, "learning_rate": 1.1401667315131015e-05, "loss": 2.7598, "step": 178100 }, { "epoch": 0.0054016, "grad_norm": 0.7870612740516663, "learning_rate": 1.1400869981889234e-05, "loss": 2.9733, "step": 178110 }, { "epoch": 0.0054272, "grad_norm": 0.7509704828262329, "learning_rate": 1.1400072639563173e-05, "loss": 2.9069, "step": 178120 }, { "epoch": 0.0054528, "grad_norm": 1.076228380203247, "learning_rate": 1.1399275288158018e-05, "loss": 3.1289, "step": 178130 }, { "epoch": 0.0054784, "grad_norm": 0.8622605204582214, "learning_rate": 1.1398477927678937e-05, "loss": 2.8508, "step": 178140 }, { "epoch": 0.005504, "grad_norm": 0.7284674048423767, "learning_rate": 1.1397680558131096e-05, "loss": 2.7223, "step": 178150 }, { "epoch": 0.0055296, "grad_norm": 0.8925106525421143, "learning_rate": 1.139688317951967e-05, "loss": 2.8341, "step": 178160 }, { "epoch": 0.0055552, "grad_norm": 0.8576605319976807, "learning_rate": 1.1396085791849825e-05, "loss": 3.0038, "step": 178170 }, { "epoch": 0.0055808, "grad_norm": 1.0504295825958252, "learning_rate": 1.1395288395126738e-05, "loss": 2.927, "step": 178180 }, { "epoch": 0.0056064, "grad_norm": 0.882957398891449, "learning_rate": 1.1394490989355574e-05, "loss": 3.051, "step": 178190 }, { "epoch": 0.005632, "grad_norm": 0.7446844577789307, "learning_rate": 1.1393693574541506e-05, "loss": 3.036, "step": 178200 }, { "epoch": 0.0056576, "grad_norm": 0.9385851621627808, "learning_rate": 1.1392896150689705e-05, "loss": 2.9737, "step": 178210 }, { "epoch": 0.0056832, "grad_norm": 0.7339939475059509, "learning_rate": 1.1392098717805344e-05, "loss": 3.1969, "step": 178220 }, { "epoch": 0.0057088, "grad_norm": 0.7741355299949646, "learning_rate": 1.1391301275893595e-05, "loss": 2.888, "step": 178230 }, { "epoch": 0.0057344, "grad_norm": 0.79500812292099, "learning_rate": 1.1390503824959625e-05, "loss": 3.1731, "step": 178240 }, { "epoch": 0.00576, "grad_norm": 0.7515828013420105, "learning_rate": 1.1389706365008606e-05, "loss": 2.8636, "step": 178250 }, { "epoch": 0.0057856, "grad_norm": 0.802098274230957, "learning_rate": 1.1388908896045711e-05, "loss": 2.7849, "step": 178260 }, { "epoch": 0.0058112, "grad_norm": 0.740307092666626, "learning_rate": 1.1388111418076114e-05, "loss": 3.0365, "step": 178270 }, { "epoch": 0.0058368, "grad_norm": 0.8740142583847046, "learning_rate": 1.1387313931104981e-05, "loss": 2.9698, "step": 178280 }, { "epoch": 0.0058624, "grad_norm": 0.9749943017959595, "learning_rate": 1.1386516435137487e-05, "loss": 2.781, "step": 178290 }, { "epoch": 0.005888, "grad_norm": 0.7834630608558655, "learning_rate": 1.13857189301788e-05, "loss": 2.8976, "step": 178300 }, { "epoch": 0.0059136, "grad_norm": 0.7247830629348755, "learning_rate": 1.1384921416234095e-05, "loss": 2.9473, "step": 178310 }, { "epoch": 0.0059392, "grad_norm": 0.8134961128234863, "learning_rate": 1.1384123893308544e-05, "loss": 3.1671, "step": 178320 }, { "epoch": 0.0059648, "grad_norm": 0.7218001484870911, "learning_rate": 1.1383326361407315e-05, "loss": 2.934, "step": 178330 }, { "epoch": 0.0059904, "grad_norm": 0.9012013077735901, "learning_rate": 1.1382528820535585e-05, "loss": 2.907, "step": 178340 }, { "epoch": 0.006016, "grad_norm": 0.8488770723342896, "learning_rate": 1.1381731270698521e-05, "loss": 2.7894, "step": 178350 }, { "epoch": 0.0060416, "grad_norm": 0.8250787258148193, "learning_rate": 1.1380933711901299e-05, "loss": 2.949, "step": 178360 }, { "epoch": 0.0060672, "grad_norm": 0.7125406265258789, "learning_rate": 1.1380136144149085e-05, "loss": 2.7592, "step": 178370 }, { "epoch": 0.0060928, "grad_norm": 0.7141785621643066, "learning_rate": 1.137933856744706e-05, "loss": 2.8636, "step": 178380 }, { "epoch": 0.0061184, "grad_norm": 0.7694665789604187, "learning_rate": 1.1378540981800386e-05, "loss": 2.8599, "step": 178390 }, { "epoch": 0.006144, "grad_norm": 0.7531274557113647, "learning_rate": 1.1377743387214245e-05, "loss": 2.7715, "step": 178400 }, { "epoch": 0.0061696, "grad_norm": 0.6977613568305969, "learning_rate": 1.13769457836938e-05, "loss": 2.8242, "step": 178410 }, { "epoch": 0.0061952, "grad_norm": 0.8306766152381897, "learning_rate": 1.137614817124423e-05, "loss": 3.0449, "step": 178420 }, { "epoch": 0.0062208, "grad_norm": 0.8460814952850342, "learning_rate": 1.1375350549870702e-05, "loss": 2.9621, "step": 178430 }, { "epoch": 0.0062464, "grad_norm": 0.7372650504112244, "learning_rate": 1.1374552919578397e-05, "loss": 2.9792, "step": 178440 }, { "epoch": 0.006272, "grad_norm": 0.6960386633872986, "learning_rate": 1.1373755280372477e-05, "loss": 2.7656, "step": 178450 }, { "epoch": 0.0062976, "grad_norm": 0.9942585825920105, "learning_rate": 1.137295763225812e-05, "loss": 2.7658, "step": 178460 }, { "epoch": 0.0063232, "grad_norm": 0.7354550361633301, "learning_rate": 1.1372159975240498e-05, "loss": 2.9186, "step": 178470 }, { "epoch": 0.0063488, "grad_norm": 0.9393681287765503, "learning_rate": 1.1371362309324783e-05, "loss": 2.9641, "step": 178480 }, { "epoch": 0.0063744, "grad_norm": 0.8252468705177307, "learning_rate": 1.1370564634516146e-05, "loss": 2.9381, "step": 178490 }, { "epoch": 0.0064, "grad_norm": 0.8329770565032959, "learning_rate": 1.1369766950819767e-05, "loss": 2.8037, "step": 178500 }, { "epoch": 0.0064256, "grad_norm": 0.7315084934234619, "learning_rate": 1.1368969258240807e-05, "loss": 2.828, "step": 178510 }, { "epoch": 0.0064512, "grad_norm": 0.7072600722312927, "learning_rate": 1.1368171556784448e-05, "loss": 2.9463, "step": 178520 }, { "epoch": 0.0064768, "grad_norm": 0.7572991847991943, "learning_rate": 1.136737384645586e-05, "loss": 2.9808, "step": 178530 }, { "epoch": 0.0065024, "grad_norm": 0.8099971413612366, "learning_rate": 1.1366576127260219e-05, "loss": 2.7505, "step": 178540 }, { "epoch": 0.006528, "grad_norm": 0.7692347168922424, "learning_rate": 1.1365778399202691e-05, "loss": 2.9925, "step": 178550 }, { "epoch": 0.0065536, "grad_norm": 1.490023136138916, "learning_rate": 1.1364980662288452e-05, "loss": 2.7337, "step": 178560 }, { "epoch": 0.0065792, "grad_norm": 0.7161414623260498, "learning_rate": 1.1364182916522681e-05, "loss": 2.9348, "step": 178570 }, { "epoch": 0.0066048, "grad_norm": 0.7461245656013489, "learning_rate": 1.1363385161910544e-05, "loss": 2.8321, "step": 178580 }, { "epoch": 0.0066304, "grad_norm": 0.9445036053657532, "learning_rate": 1.1362587398457214e-05, "loss": 2.8257, "step": 178590 }, { "epoch": 0.006656, "grad_norm": 0.7768218517303467, "learning_rate": 1.1361789626167871e-05, "loss": 2.9385, "step": 178600 }, { "epoch": 0.0066816, "grad_norm": 0.8907334208488464, "learning_rate": 1.1360991845047682e-05, "loss": 2.9507, "step": 178610 }, { "epoch": 0.0067072, "grad_norm": 0.7699339985847473, "learning_rate": 1.1360194055101823e-05, "loss": 2.895, "step": 178620 }, { "epoch": 0.0067328, "grad_norm": 0.8706409931182861, "learning_rate": 1.1359396256335466e-05, "loss": 2.9364, "step": 178630 }, { "epoch": 0.0067584, "grad_norm": 0.7490491271018982, "learning_rate": 1.1358598448753788e-05, "loss": 3.0489, "step": 178640 }, { "epoch": 0.006784, "grad_norm": 0.8060686588287354, "learning_rate": 1.1357800632361957e-05, "loss": 2.9025, "step": 178650 }, { "epoch": 0.0068096, "grad_norm": 0.7733566164970398, "learning_rate": 1.1357002807165151e-05, "loss": 2.8758, "step": 178660 }, { "epoch": 0.0068352, "grad_norm": 0.951117753982544, "learning_rate": 1.1356204973168546e-05, "loss": 3.0274, "step": 178670 }, { "epoch": 0.0068608, "grad_norm": 0.7431973218917847, "learning_rate": 1.1355407130377307e-05, "loss": 3.011, "step": 178680 }, { "epoch": 0.0068864, "grad_norm": 0.736910879611969, "learning_rate": 1.1354609278796616e-05, "loss": 2.9237, "step": 178690 }, { "epoch": 0.006912, "grad_norm": 0.840933084487915, "learning_rate": 1.1353811418431643e-05, "loss": 2.9422, "step": 178700 }, { "epoch": 2.56e-05, "grad_norm": 1.0321344137191772, "learning_rate": 1.1353013549287563e-05, "loss": 2.4242, "step": 178710 }, { "epoch": 5.12e-05, "grad_norm": 0.7768599390983582, "learning_rate": 1.1352215671369548e-05, "loss": 2.8886, "step": 178720 }, { "epoch": 7.68e-05, "grad_norm": 0.9135515689849854, "learning_rate": 1.1351417784682777e-05, "loss": 2.8706, "step": 178730 }, { "epoch": 0.0001024, "grad_norm": 0.7046143412590027, "learning_rate": 1.1350619889232416e-05, "loss": 2.5388, "step": 178740 }, { "epoch": 0.000128, "grad_norm": 0.8216621279716492, "learning_rate": 1.1349821985023647e-05, "loss": 2.7795, "step": 178750 }, { "epoch": 0.0001536, "grad_norm": 0.7864163517951965, "learning_rate": 1.1349024072061641e-05, "loss": 2.9725, "step": 178760 }, { "epoch": 0.0001792, "grad_norm": 0.8257717490196228, "learning_rate": 1.1348226150351576e-05, "loss": 2.7225, "step": 178770 }, { "epoch": 0.0002048, "grad_norm": 0.7820620536804199, "learning_rate": 1.1347428219898617e-05, "loss": 2.6675, "step": 178780 }, { "epoch": 0.0002304, "grad_norm": 0.7438197731971741, "learning_rate": 1.1346630280707946e-05, "loss": 2.759, "step": 178790 }, { "epoch": 0.000256, "grad_norm": 0.8857191801071167, "learning_rate": 1.1345832332784739e-05, "loss": 2.7849, "step": 178800 }, { "epoch": 0.0002816, "grad_norm": 0.7827448844909668, "learning_rate": 1.1345034376134162e-05, "loss": 2.6676, "step": 178810 }, { "epoch": 0.0003072, "grad_norm": 0.8424334526062012, "learning_rate": 1.1344236410761393e-05, "loss": 2.9508, "step": 178820 }, { "epoch": 0.0003328, "grad_norm": 1.0730551481246948, "learning_rate": 1.1343438436671615e-05, "loss": 2.5366, "step": 178830 }, { "epoch": 0.0003584, "grad_norm": 0.884127676486969, "learning_rate": 1.134264045386999e-05, "loss": 2.8429, "step": 178840 }, { "epoch": 0.000384, "grad_norm": 1.0081220865249634, "learning_rate": 1.1341842462361703e-05, "loss": 2.5905, "step": 178850 }, { "epoch": 0.0004096, "grad_norm": 0.8280529379844666, "learning_rate": 1.134104446215192e-05, "loss": 2.7151, "step": 178860 }, { "epoch": 0.0004352, "grad_norm": 0.8088805675506592, "learning_rate": 1.1340246453245825e-05, "loss": 2.8197, "step": 178870 }, { "epoch": 0.0004608, "grad_norm": 0.7569994330406189, "learning_rate": 1.1339448435648584e-05, "loss": 2.8195, "step": 178880 }, { "epoch": 0.0004864, "grad_norm": 0.9213008880615234, "learning_rate": 1.1338650409365377e-05, "loss": 2.9798, "step": 178890 }, { "epoch": 0.000512, "grad_norm": 0.6547340750694275, "learning_rate": 1.1337852374401379e-05, "loss": 2.5423, "step": 178900 }, { "epoch": 0.0005376, "grad_norm": 0.7898364067077637, "learning_rate": 1.1337054330761761e-05, "loss": 2.7993, "step": 178910 }, { "epoch": 0.0005632, "grad_norm": 0.8736489415168762, "learning_rate": 1.1336256278451703e-05, "loss": 2.7594, "step": 178920 }, { "epoch": 0.0005888, "grad_norm": 0.7882075905799866, "learning_rate": 1.1335458217476379e-05, "loss": 2.5745, "step": 178930 }, { "epoch": 0.0006144, "grad_norm": 0.8188495635986328, "learning_rate": 1.1334660147840962e-05, "loss": 2.7614, "step": 178940 }, { "epoch": 0.00064, "grad_norm": 0.8725834488868713, "learning_rate": 1.1333862069550627e-05, "loss": 2.6589, "step": 178950 }, { "epoch": 0.0006656, "grad_norm": 0.7796096205711365, "learning_rate": 1.1333063982610552e-05, "loss": 2.9207, "step": 178960 }, { "epoch": 0.0006912, "grad_norm": 1.0044376850128174, "learning_rate": 1.1332265887025915e-05, "loss": 2.9421, "step": 178970 }, { "epoch": 0.0007168, "grad_norm": 0.7788074612617493, "learning_rate": 1.1331467782801886e-05, "loss": 2.7641, "step": 178980 }, { "epoch": 0.0007424, "grad_norm": 0.888286828994751, "learning_rate": 1.1330669669943639e-05, "loss": 2.7144, "step": 178990 }, { "epoch": 0.000768, "grad_norm": 0.8970246911048889, "learning_rate": 1.1329871548456355e-05, "loss": 2.8088, "step": 179000 }, { "epoch": 0.0007936, "grad_norm": 0.9158724546432495, "learning_rate": 1.132907341834521e-05, "loss": 2.7262, "step": 179010 }, { "epoch": 0.0008192, "grad_norm": 0.8133058547973633, "learning_rate": 1.1328275279615372e-05, "loss": 2.4102, "step": 179020 }, { "epoch": 0.0008448, "grad_norm": 0.7081503868103027, "learning_rate": 1.1327477132272026e-05, "loss": 2.8833, "step": 179030 }, { "epoch": 0.0008704, "grad_norm": 0.8659404516220093, "learning_rate": 1.1326678976320342e-05, "loss": 2.9156, "step": 179040 }, { "epoch": 0.000896, "grad_norm": 0.7317468523979187, "learning_rate": 1.1325880811765497e-05, "loss": 2.7078, "step": 179050 }, { "epoch": 0.0009216, "grad_norm": 0.7749430537223816, "learning_rate": 1.1325082638612668e-05, "loss": 2.8002, "step": 179060 }, { "epoch": 0.0009472, "grad_norm": 0.7374236583709717, "learning_rate": 1.1324284456867031e-05, "loss": 2.6093, "step": 179070 }, { "epoch": 0.0009728, "grad_norm": 0.7528843283653259, "learning_rate": 1.1323486266533758e-05, "loss": 2.6978, "step": 179080 }, { "epoch": 0.0009984, "grad_norm": 0.8143858909606934, "learning_rate": 1.1322688067618033e-05, "loss": 2.8177, "step": 179090 }, { "epoch": 0.001024, "grad_norm": 0.7538518309593201, "learning_rate": 1.1321889860125026e-05, "loss": 2.7163, "step": 179100 }, { "epoch": 0.0010496, "grad_norm": 0.978838324546814, "learning_rate": 1.1321091644059912e-05, "loss": 2.9538, "step": 179110 }, { "epoch": 0.0010752, "grad_norm": 0.7939143180847168, "learning_rate": 1.1320293419427873e-05, "loss": 2.746, "step": 179120 }, { "epoch": 0.0011008, "grad_norm": 0.7400130033493042, "learning_rate": 1.1319495186234082e-05, "loss": 2.8157, "step": 179130 }, { "epoch": 0.0011264, "grad_norm": 0.7967342734336853, "learning_rate": 1.1318696944483713e-05, "loss": 3.0322, "step": 179140 }, { "epoch": 0.001152, "grad_norm": 0.8118176460266113, "learning_rate": 1.1317898694181944e-05, "loss": 2.8561, "step": 179150 }, { "epoch": 0.0011776, "grad_norm": 0.8739185929298401, "learning_rate": 1.1317100435333954e-05, "loss": 3.0587, "step": 179160 }, { "epoch": 0.0012032, "grad_norm": 0.7159085869789124, "learning_rate": 1.1316302167944916e-05, "loss": 2.9945, "step": 179170 }, { "epoch": 0.0012288, "grad_norm": 0.7891978621482849, "learning_rate": 1.1315503892020012e-05, "loss": 3.0149, "step": 179180 }, { "epoch": 0.0012544, "grad_norm": 0.7870365977287292, "learning_rate": 1.1314705607564413e-05, "loss": 2.8985, "step": 179190 }, { "epoch": 0.00128, "grad_norm": 0.7634742856025696, "learning_rate": 1.1313907314583296e-05, "loss": 2.6981, "step": 179200 }, { "epoch": 0.0013056, "grad_norm": 0.8804885745048523, "learning_rate": 1.1313109013081843e-05, "loss": 2.8958, "step": 179210 }, { "epoch": 0.0013312, "grad_norm": 0.84999018907547, "learning_rate": 1.1312310703065223e-05, "loss": 3.0556, "step": 179220 }, { "epoch": 0.0013568, "grad_norm": 0.9917230010032654, "learning_rate": 1.131151238453862e-05, "loss": 2.7264, "step": 179230 }, { "epoch": 0.0013824, "grad_norm": 0.9593727588653564, "learning_rate": 1.1310714057507206e-05, "loss": 2.9769, "step": 179240 }, { "epoch": 0.001408, "grad_norm": 0.8082655668258667, "learning_rate": 1.1309915721976159e-05, "loss": 2.8395, "step": 179250 }, { "epoch": 0.0014336, "grad_norm": 0.7341845035552979, "learning_rate": 1.1309117377950658e-05, "loss": 2.9249, "step": 179260 }, { "epoch": 0.0014592, "grad_norm": 0.804740846157074, "learning_rate": 1.130831902543588e-05, "loss": 2.8044, "step": 179270 }, { "epoch": 0.0014848, "grad_norm": 0.820080578327179, "learning_rate": 1.1307520664436997e-05, "loss": 2.897, "step": 179280 }, { "epoch": 0.0015104, "grad_norm": 0.8725882172584534, "learning_rate": 1.1306722294959193e-05, "loss": 2.502, "step": 179290 }, { "epoch": 0.001536, "grad_norm": 0.8599993586540222, "learning_rate": 1.1305923917007643e-05, "loss": 2.5345, "step": 179300 }, { "epoch": 0.0015616, "grad_norm": 0.8927634954452515, "learning_rate": 1.130512553058752e-05, "loss": 2.9526, "step": 179310 }, { "epoch": 0.0015872, "grad_norm": 0.8187386393547058, "learning_rate": 1.1304327135704007e-05, "loss": 2.7261, "step": 179320 }, { "epoch": 0.0016128, "grad_norm": 0.7248399257659912, "learning_rate": 1.1303528732362282e-05, "loss": 2.7712, "step": 179330 }, { "epoch": 0.0016384, "grad_norm": 0.6773858070373535, "learning_rate": 1.1302730320567511e-05, "loss": 2.7258, "step": 179340 }, { "epoch": 0.001664, "grad_norm": 0.8592636585235596, "learning_rate": 1.1301931900324885e-05, "loss": 2.8451, "step": 179350 }, { "epoch": 0.0016896, "grad_norm": 0.728992223739624, "learning_rate": 1.1301133471639581e-05, "loss": 2.9404, "step": 179360 }, { "epoch": 0.0017152, "grad_norm": 0.7238315939903259, "learning_rate": 1.1300335034516767e-05, "loss": 2.962, "step": 179370 }, { "epoch": 0.0017408, "grad_norm": 0.6796764731407166, "learning_rate": 1.1299536588961626e-05, "loss": 2.5024, "step": 179380 }, { "epoch": 0.0017664, "grad_norm": 0.7555723190307617, "learning_rate": 1.1298738134979338e-05, "loss": 3.1179, "step": 179390 }, { "epoch": 0.001792, "grad_norm": 0.7718579173088074, "learning_rate": 1.1297939672575078e-05, "loss": 2.813, "step": 179400 }, { "epoch": 0.0018176, "grad_norm": 0.9382299184799194, "learning_rate": 1.1297141201754023e-05, "loss": 2.5692, "step": 179410 }, { "epoch": 0.0018432, "grad_norm": 0.880297064781189, "learning_rate": 1.1296342722521353e-05, "loss": 2.7011, "step": 179420 }, { "epoch": 0.0018688, "grad_norm": 0.7514759302139282, "learning_rate": 1.1295544234882242e-05, "loss": 2.6334, "step": 179430 }, { "epoch": 0.0018944, "grad_norm": 0.8068806529045105, "learning_rate": 1.1294745738841875e-05, "loss": 3.0989, "step": 179440 }, { "epoch": 0.00192, "grad_norm": 0.961610734462738, "learning_rate": 1.1293947234405422e-05, "loss": 2.9382, "step": 179450 }, { "epoch": 0.0019456, "grad_norm": 0.7403169274330139, "learning_rate": 1.129314872157807e-05, "loss": 2.8332, "step": 179460 }, { "epoch": 0.0019712, "grad_norm": 0.7739865183830261, "learning_rate": 1.129235020036499e-05, "loss": 2.8171, "step": 179470 }, { "epoch": 0.0019968, "grad_norm": 0.6932883262634277, "learning_rate": 1.129155167077136e-05, "loss": 2.3548, "step": 179480 }, { "epoch": 0.0020224, "grad_norm": 0.7734315395355225, "learning_rate": 1.1290753132802366e-05, "loss": 2.8375, "step": 179490 }, { "epoch": 0.002048, "grad_norm": 1.9287558794021606, "learning_rate": 1.1289954586463176e-05, "loss": 3.0115, "step": 179500 }, { "epoch": 0.0020736, "grad_norm": 0.7739068865776062, "learning_rate": 1.1289156031758977e-05, "loss": 2.7533, "step": 179510 }, { "epoch": 0.0020992, "grad_norm": 0.7975166440010071, "learning_rate": 1.1288357468694939e-05, "loss": 2.6357, "step": 179520 }, { "epoch": 0.0021248, "grad_norm": 1.0895479917526245, "learning_rate": 1.128755889727625e-05, "loss": 3.1443, "step": 179530 }, { "epoch": 0.0021504, "grad_norm": 0.7566304802894592, "learning_rate": 1.128676031750808e-05, "loss": 2.9112, "step": 179540 }, { "epoch": 0.002176, "grad_norm": 0.7334862351417542, "learning_rate": 1.1285961729395614e-05, "loss": 3.0187, "step": 179550 }, { "epoch": 0.0022016, "grad_norm": 0.8113537430763245, "learning_rate": 1.1285163132944029e-05, "loss": 3.0149, "step": 179560 }, { "epoch": 0.0022272, "grad_norm": 0.7332924604415894, "learning_rate": 1.12843645281585e-05, "loss": 3.0757, "step": 179570 }, { "epoch": 0.0022528, "grad_norm": 0.7286558151245117, "learning_rate": 1.1283565915044208e-05, "loss": 2.9665, "step": 179580 }, { "epoch": 0.0022784, "grad_norm": 0.7130041718482971, "learning_rate": 1.1282767293606334e-05, "loss": 2.897, "step": 179590 }, { "epoch": 0.002304, "grad_norm": 0.7398561835289001, "learning_rate": 1.1281968663850052e-05, "loss": 2.7446, "step": 179600 }, { "epoch": 0.0023296, "grad_norm": 0.7528486847877502, "learning_rate": 1.1281170025780546e-05, "loss": 2.8151, "step": 179610 }, { "epoch": 0.0023552, "grad_norm": 0.7700749039649963, "learning_rate": 1.1280371379402995e-05, "loss": 2.977, "step": 179620 }, { "epoch": 0.0023808, "grad_norm": 0.8994340300559998, "learning_rate": 1.1279572724722573e-05, "loss": 2.7002, "step": 179630 }, { "epoch": 0.0024064, "grad_norm": 0.7398996353149414, "learning_rate": 1.1278774061744463e-05, "loss": 2.883, "step": 179640 }, { "epoch": 0.002432, "grad_norm": 0.7444579601287842, "learning_rate": 1.1277975390473843e-05, "loss": 2.9164, "step": 179650 }, { "epoch": 0.0024576, "grad_norm": 1.0532699823379517, "learning_rate": 1.1277176710915893e-05, "loss": 2.9566, "step": 179660 }, { "epoch": 0.0024832, "grad_norm": 1.895294189453125, "learning_rate": 1.127637802307579e-05, "loss": 2.956, "step": 179670 }, { "epoch": 0.0025088, "grad_norm": 0.6936531662940979, "learning_rate": 1.1275579326958716e-05, "loss": 2.6803, "step": 179680 }, { "epoch": 0.0025344, "grad_norm": 0.7697333693504333, "learning_rate": 1.1274780622569848e-05, "loss": 2.8648, "step": 179690 }, { "epoch": 0.00256, "grad_norm": 0.7518815994262695, "learning_rate": 1.1273981909914366e-05, "loss": 2.7295, "step": 179700 }, { "epoch": 0.0025856, "grad_norm": 0.7385385632514954, "learning_rate": 1.1273183188997451e-05, "loss": 2.9825, "step": 179710 }, { "epoch": 0.0026112, "grad_norm": 0.7747321128845215, "learning_rate": 1.1272384459824282e-05, "loss": 2.8932, "step": 179720 }, { "epoch": 0.0026368, "grad_norm": 0.9515610337257385, "learning_rate": 1.1271585722400038e-05, "loss": 3.0427, "step": 179730 }, { "epoch": 0.0026624, "grad_norm": 0.7786464095115662, "learning_rate": 1.1270786976729898e-05, "loss": 2.8744, "step": 179740 }, { "epoch": 0.002688, "grad_norm": 1.082594633102417, "learning_rate": 1.126998822281904e-05, "loss": 2.8977, "step": 179750 }, { "epoch": 0.0027136, "grad_norm": 0.8031436800956726, "learning_rate": 1.126918946067265e-05, "loss": 3.0453, "step": 179760 }, { "epoch": 0.0027392, "grad_norm": 0.7932531237602234, "learning_rate": 1.12683906902959e-05, "loss": 3.1054, "step": 179770 }, { "epoch": 0.0027648, "grad_norm": 0.7237130999565125, "learning_rate": 1.1267591911693976e-05, "loss": 2.9617, "step": 179780 }, { "epoch": 0.0027904, "grad_norm": 0.7769494652748108, "learning_rate": 1.1266793124872053e-05, "loss": 2.7504, "step": 179790 }, { "epoch": 0.002816, "grad_norm": 0.7556609511375427, "learning_rate": 1.1265994329835315e-05, "loss": 2.8087, "step": 179800 }, { "epoch": 0.0028416, "grad_norm": 0.8093754053115845, "learning_rate": 1.126519552658894e-05, "loss": 3.0419, "step": 179810 }, { "epoch": 0.0028672, "grad_norm": 0.7913612127304077, "learning_rate": 1.1264396715138109e-05, "loss": 2.9259, "step": 179820 }, { "epoch": 0.0028928, "grad_norm": 0.770091712474823, "learning_rate": 1.1263597895487999e-05, "loss": 3.0484, "step": 179830 }, { "epoch": 0.0029184, "grad_norm": 0.827130138874054, "learning_rate": 1.1262799067643794e-05, "loss": 3.1128, "step": 179840 }, { "epoch": 0.002944, "grad_norm": 0.7670475244522095, "learning_rate": 1.1262000231610673e-05, "loss": 2.8995, "step": 179850 }, { "epoch": 0.0029696, "grad_norm": 0.8481507301330566, "learning_rate": 1.1261201387393814e-05, "loss": 2.9077, "step": 179860 }, { "epoch": 0.0029952, "grad_norm": 0.8363949060440063, "learning_rate": 1.12604025349984e-05, "loss": 2.9052, "step": 179870 }, { "epoch": 0.0030208, "grad_norm": 0.7569262385368347, "learning_rate": 1.1259603674429613e-05, "loss": 2.9418, "step": 179880 }, { "epoch": 0.0030464, "grad_norm": 0.7418984770774841, "learning_rate": 1.1258804805692629e-05, "loss": 2.9346, "step": 179890 }, { "epoch": 0.003072, "grad_norm": 0.8109654188156128, "learning_rate": 1.125800592879263e-05, "loss": 2.9874, "step": 179900 }, { "epoch": 0.0030976, "grad_norm": 0.7607152462005615, "learning_rate": 1.1257207043734797e-05, "loss": 2.9515, "step": 179910 }, { "epoch": 0.0031232, "grad_norm": 0.8639597296714783, "learning_rate": 1.1256408150524312e-05, "loss": 3.2923, "step": 179920 }, { "epoch": 0.0031488, "grad_norm": 0.7745692729949951, "learning_rate": 1.1255609249166353e-05, "loss": 2.7185, "step": 179930 }, { "epoch": 0.0031744, "grad_norm": 0.7714041471481323, "learning_rate": 1.12548103396661e-05, "loss": 3.031, "step": 179940 }, { "epoch": 0.0032, "grad_norm": 0.7822355628013611, "learning_rate": 1.1254011422028736e-05, "loss": 3.1022, "step": 179950 }, { "epoch": 0.0032256, "grad_norm": 0.836159348487854, "learning_rate": 1.1253212496259442e-05, "loss": 3.0505, "step": 179960 }, { "epoch": 0.0032512, "grad_norm": 0.85896235704422, "learning_rate": 1.1252413562363398e-05, "loss": 2.812, "step": 179970 }, { "epoch": 0.0032768, "grad_norm": 0.7763145565986633, "learning_rate": 1.1251614620345788e-05, "loss": 3.1245, "step": 179980 }, { "epoch": 0.0033024, "grad_norm": 0.7665755152702332, "learning_rate": 1.1250815670211786e-05, "loss": 2.9577, "step": 179990 }, { "epoch": 0.003328, "grad_norm": 0.8736397624015808, "learning_rate": 1.1250016711966576e-05, "loss": 3.0201, "step": 180000 }, { "epoch": 0.0033536, "grad_norm": 0.7873408198356628, "learning_rate": 1.1249217745615342e-05, "loss": 3.2371, "step": 180010 }, { "epoch": 0.0033792, "grad_norm": 1.0636975765228271, "learning_rate": 1.1248418771163266e-05, "loss": 3.1319, "step": 180020 }, { "epoch": 0.0034048, "grad_norm": 0.7809575796127319, "learning_rate": 1.124761978861552e-05, "loss": 3.0194, "step": 180030 }, { "epoch": 0.0034304, "grad_norm": 0.8595772385597229, "learning_rate": 1.1246820797977292e-05, "loss": 2.8931, "step": 180040 }, { "epoch": 0.003456, "grad_norm": 1.1366807222366333, "learning_rate": 1.1246021799253766e-05, "loss": 3.1034, "step": 180050 }, { "epoch": 0.0034816, "grad_norm": 0.706566572189331, "learning_rate": 1.1245222792450118e-05, "loss": 2.9592, "step": 180060 }, { "epoch": 0.0035072, "grad_norm": 0.9198355674743652, "learning_rate": 1.124442377757153e-05, "loss": 2.8802, "step": 180070 }, { "epoch": 0.0035328, "grad_norm": 0.7272351384162903, "learning_rate": 1.1243624754623188e-05, "loss": 2.9166, "step": 180080 }, { "epoch": 0.0035584, "grad_norm": 0.7235783934593201, "learning_rate": 1.1242825723610267e-05, "loss": 2.9782, "step": 180090 }, { "epoch": 0.003584, "grad_norm": 0.7449129223823547, "learning_rate": 1.1242026684537951e-05, "loss": 3.1824, "step": 180100 }, { "epoch": 0.0036096, "grad_norm": 1.233141541481018, "learning_rate": 1.1241227637411423e-05, "loss": 2.8655, "step": 180110 }, { "epoch": 0.0036352, "grad_norm": 0.7517179250717163, "learning_rate": 1.1240428582235867e-05, "loss": 3.0726, "step": 180120 }, { "epoch": 0.0036608, "grad_norm": 1.17909574508667, "learning_rate": 1.1239629519016454e-05, "loss": 2.9794, "step": 180130 }, { "epoch": 0.0036864, "grad_norm": 0.8423622846603394, "learning_rate": 1.1238830447758377e-05, "loss": 2.9822, "step": 180140 }, { "epoch": 0.003712, "grad_norm": 0.8225858211517334, "learning_rate": 1.1238031368466818e-05, "loss": 2.9998, "step": 180150 }, { "epoch": 0.0037376, "grad_norm": 0.7842171788215637, "learning_rate": 1.1237232281146948e-05, "loss": 2.9804, "step": 180160 }, { "epoch": 0.0037632, "grad_norm": 1.000328540802002, "learning_rate": 1.1236433185803958e-05, "loss": 3.2195, "step": 180170 }, { "epoch": 0.0037888, "grad_norm": 0.7368493676185608, "learning_rate": 1.123563408244303e-05, "loss": 2.9785, "step": 180180 }, { "epoch": 0.0038144, "grad_norm": 0.8100875020027161, "learning_rate": 1.123483497106934e-05, "loss": 3.0901, "step": 180190 }, { "epoch": 0.00384, "grad_norm": 0.8356369137763977, "learning_rate": 1.1234035851688075e-05, "loss": 2.9266, "step": 180200 }, { "epoch": 0.0038656, "grad_norm": 0.7334678173065186, "learning_rate": 1.1233236724304412e-05, "loss": 3.1451, "step": 180210 }, { "epoch": 0.0038912, "grad_norm": 0.8871488571166992, "learning_rate": 1.123243758892354e-05, "loss": 2.7885, "step": 180220 }, { "epoch": 0.0039168, "grad_norm": 0.9621161818504333, "learning_rate": 1.1231638445550636e-05, "loss": 3.0737, "step": 180230 }, { "epoch": 0.0039424, "grad_norm": 0.8722413182258606, "learning_rate": 1.1230839294190885e-05, "loss": 2.8306, "step": 180240 }, { "epoch": 0.003968, "grad_norm": 1.049694299697876, "learning_rate": 1.1230040134849472e-05, "loss": 2.8889, "step": 180250 }, { "epoch": 0.0039936, "grad_norm": 0.7500121593475342, "learning_rate": 1.122924096753157e-05, "loss": 2.7826, "step": 180260 }, { "epoch": 0.0040192, "grad_norm": 0.8492777943611145, "learning_rate": 1.1228441792242368e-05, "loss": 2.9642, "step": 180270 }, { "epoch": 0.0040448, "grad_norm": 0.8904128670692444, "learning_rate": 1.122764260898705e-05, "loss": 2.9875, "step": 180280 }, { "epoch": 0.0040704, "grad_norm": 1.4354273080825806, "learning_rate": 1.1226843417770793e-05, "loss": 3.169, "step": 180290 }, { "epoch": 0.004096, "grad_norm": 0.852963387966156, "learning_rate": 1.1226044218598784e-05, "loss": 3.0089, "step": 180300 }, { "epoch": 0.0041216, "grad_norm": 0.7616632580757141, "learning_rate": 1.1225245011476204e-05, "loss": 2.7442, "step": 180310 }, { "epoch": 0.0041472, "grad_norm": 0.9580668210983276, "learning_rate": 1.1224445796408236e-05, "loss": 2.9418, "step": 180320 }, { "epoch": 0.0041728, "grad_norm": 0.7597030401229858, "learning_rate": 1.1223646573400063e-05, "loss": 3.0613, "step": 180330 }, { "epoch": 0.0041984, "grad_norm": 0.7258002161979675, "learning_rate": 1.1222847342456866e-05, "loss": 2.8529, "step": 180340 }, { "epoch": 0.004224, "grad_norm": 0.9574311375617981, "learning_rate": 1.1222048103583832e-05, "loss": 2.848, "step": 180350 }, { "epoch": 0.0042496, "grad_norm": 0.7657601237297058, "learning_rate": 1.1221248856786135e-05, "loss": 3.0684, "step": 180360 }, { "epoch": 0.0042752, "grad_norm": 0.8768625259399414, "learning_rate": 1.1220449602068968e-05, "loss": 3.2018, "step": 180370 }, { "epoch": 0.0043008, "grad_norm": 0.7426198124885559, "learning_rate": 1.1219650339437508e-05, "loss": 2.9015, "step": 180380 }, { "epoch": 0.0043264, "grad_norm": 0.7456268668174744, "learning_rate": 1.121885106889694e-05, "loss": 2.975, "step": 180390 }, { "epoch": 0.004352, "grad_norm": 0.7169418931007385, "learning_rate": 1.1218051790452447e-05, "loss": 2.9979, "step": 180400 }, { "epoch": 0.0043776, "grad_norm": 0.756630539894104, "learning_rate": 1.1217252504109213e-05, "loss": 2.8249, "step": 180410 }, { "epoch": 0.0044032, "grad_norm": 0.7703995704650879, "learning_rate": 1.1216453209872418e-05, "loss": 2.8511, "step": 180420 }, { "epoch": 0.0044288, "grad_norm": 1.0523701906204224, "learning_rate": 1.1215653907747249e-05, "loss": 2.9123, "step": 180430 }, { "epoch": 0.0044544, "grad_norm": 0.8918944001197815, "learning_rate": 1.1214854597738885e-05, "loss": 3.1635, "step": 180440 }, { "epoch": 0.00448, "grad_norm": 0.9046383500099182, "learning_rate": 1.1214055279852517e-05, "loss": 3.0024, "step": 180450 }, { "epoch": 0.0045056, "grad_norm": 0.8465500473976135, "learning_rate": 1.1213255954093318e-05, "loss": 2.718, "step": 180460 }, { "epoch": 0.0045312, "grad_norm": 0.7646885514259338, "learning_rate": 1.1212456620466475e-05, "loss": 2.9166, "step": 180470 }, { "epoch": 0.0045568, "grad_norm": 0.7020695805549622, "learning_rate": 1.1211657278977178e-05, "loss": 3.0552, "step": 180480 }, { "epoch": 0.0045824, "grad_norm": 0.8489379286766052, "learning_rate": 1.1210857929630603e-05, "loss": 2.82, "step": 180490 }, { "epoch": 0.004608, "grad_norm": 0.7627686262130737, "learning_rate": 1.1210058572431937e-05, "loss": 3.0781, "step": 180500 }, { "epoch": 0.0046336, "grad_norm": 0.880966305732727, "learning_rate": 1.1209259207386365e-05, "loss": 2.8633, "step": 180510 }, { "epoch": 0.0046592, "grad_norm": 0.7200230956077576, "learning_rate": 1.1208459834499066e-05, "loss": 2.9897, "step": 180520 }, { "epoch": 0.0046848, "grad_norm": 0.7643254995346069, "learning_rate": 1.1207660453775225e-05, "loss": 2.88, "step": 180530 }, { "epoch": 0.0047104, "grad_norm": 0.8617703914642334, "learning_rate": 1.1206861065220027e-05, "loss": 3.0032, "step": 180540 }, { "epoch": 0.004736, "grad_norm": 0.7662879228591919, "learning_rate": 1.120606166883866e-05, "loss": 2.9217, "step": 180550 }, { "epoch": 0.0047616, "grad_norm": 0.8392367959022522, "learning_rate": 1.1205262264636298e-05, "loss": 3.0063, "step": 180560 }, { "epoch": 0.0047872, "grad_norm": 0.7717578411102295, "learning_rate": 1.1204462852618132e-05, "loss": 2.8904, "step": 180570 }, { "epoch": 0.0048128, "grad_norm": 0.8579752445220947, "learning_rate": 1.1203663432789346e-05, "loss": 3.1754, "step": 180580 }, { "epoch": 0.0048384, "grad_norm": 0.7459044456481934, "learning_rate": 1.1202864005155123e-05, "loss": 3.1134, "step": 180590 }, { "epoch": 0.004864, "grad_norm": 0.8407599329948425, "learning_rate": 1.1202064569720646e-05, "loss": 2.8978, "step": 180600 }, { "epoch": 0.0048896, "grad_norm": 0.8336502313613892, "learning_rate": 1.1201265126491101e-05, "loss": 2.9241, "step": 180610 }, { "epoch": 0.0049152, "grad_norm": 3.661181926727295, "learning_rate": 1.1200465675471668e-05, "loss": 2.9339, "step": 180620 }, { "epoch": 0.0049408, "grad_norm": 0.7120161652565002, "learning_rate": 1.1199666216667536e-05, "loss": 2.9844, "step": 180630 }, { "epoch": 0.0049664, "grad_norm": 0.6968246102333069, "learning_rate": 1.1198866750083886e-05, "loss": 2.8603, "step": 180640 }, { "epoch": 0.004992, "grad_norm": 0.7947421073913574, "learning_rate": 1.1198067275725904e-05, "loss": 2.9596, "step": 180650 }, { "epoch": 0.0050176, "grad_norm": 0.7692615985870361, "learning_rate": 1.1197267793598773e-05, "loss": 2.9624, "step": 180660 }, { "epoch": 0.0050432, "grad_norm": 0.7905416488647461, "learning_rate": 1.1196468303707681e-05, "loss": 3.1506, "step": 180670 }, { "epoch": 0.0050688, "grad_norm": 0.7982426285743713, "learning_rate": 1.119566880605781e-05, "loss": 2.7832, "step": 180680 }, { "epoch": 0.0050944, "grad_norm": 0.7856884002685547, "learning_rate": 1.1194869300654345e-05, "loss": 2.8168, "step": 180690 }, { "epoch": 0.00512, "grad_norm": 1.0598028898239136, "learning_rate": 1.1194069787502469e-05, "loss": 2.694, "step": 180700 }, { "epoch": 0.0051456, "grad_norm": 0.8205692172050476, "learning_rate": 1.119327026660737e-05, "loss": 2.9623, "step": 180710 }, { "epoch": 0.0051712, "grad_norm": 0.8676571249961853, "learning_rate": 1.1192470737974228e-05, "loss": 3.1509, "step": 180720 }, { "epoch": 0.0051968, "grad_norm": 0.8388112783432007, "learning_rate": 1.1191671201608228e-05, "loss": 3.0377, "step": 180730 }, { "epoch": 0.0052224, "grad_norm": 0.7473390698432922, "learning_rate": 1.1190871657514562e-05, "loss": 2.904, "step": 180740 }, { "epoch": 0.005248, "grad_norm": 0.7781198024749756, "learning_rate": 1.1190072105698406e-05, "loss": 3.0001, "step": 180750 }, { "epoch": 0.0052736, "grad_norm": 0.7936059832572937, "learning_rate": 1.1189272546164949e-05, "loss": 2.9045, "step": 180760 }, { "epoch": 0.0052992, "grad_norm": 0.7637633681297302, "learning_rate": 1.1188472978919379e-05, "loss": 2.8742, "step": 180770 }, { "epoch": 0.0053248, "grad_norm": 0.7884093523025513, "learning_rate": 1.1187673403966874e-05, "loss": 2.8771, "step": 180780 }, { "epoch": 0.0053504, "grad_norm": 1.5761895179748535, "learning_rate": 1.1186873821312623e-05, "loss": 2.9846, "step": 180790 }, { "epoch": 0.005376, "grad_norm": 0.9276547431945801, "learning_rate": 1.1186074230961812e-05, "loss": 3.0423, "step": 180800 }, { "epoch": 0.0054016, "grad_norm": 0.8187463879585266, "learning_rate": 1.1185274632919625e-05, "loss": 2.9483, "step": 180810 }, { "epoch": 0.0054272, "grad_norm": 0.7413440346717834, "learning_rate": 1.1184475027191244e-05, "loss": 2.9119, "step": 180820 }, { "epoch": 0.0054528, "grad_norm": 0.8063270449638367, "learning_rate": 1.1183675413781859e-05, "loss": 2.6695, "step": 180830 }, { "epoch": 0.0054784, "grad_norm": 0.8687254786491394, "learning_rate": 1.1182875792696657e-05, "loss": 2.9881, "step": 180840 }, { "epoch": 0.005504, "grad_norm": 0.8578570485115051, "learning_rate": 1.1182076163940817e-05, "loss": 2.8445, "step": 180850 }, { "epoch": 0.0055296, "grad_norm": 0.7394763827323914, "learning_rate": 1.1181276527519524e-05, "loss": 2.9198, "step": 180860 }, { "epoch": 0.0055552, "grad_norm": 0.8343993425369263, "learning_rate": 1.1180476883437975e-05, "loss": 2.7539, "step": 180870 }, { "epoch": 0.0055808, "grad_norm": 0.8177229762077332, "learning_rate": 1.117967723170134e-05, "loss": 2.9073, "step": 180880 }, { "epoch": 0.0056064, "grad_norm": 0.7522717714309692, "learning_rate": 1.1178877572314812e-05, "loss": 2.7573, "step": 180890 }, { "epoch": 0.005632, "grad_norm": 0.7637367844581604, "learning_rate": 1.1178077905283577e-05, "loss": 3.064, "step": 180900 }, { "epoch": 0.0056576, "grad_norm": 0.9032487273216248, "learning_rate": 1.1177278230612823e-05, "loss": 2.8505, "step": 180910 }, { "epoch": 0.0056832, "grad_norm": 0.7287074327468872, "learning_rate": 1.1176478548307727e-05, "loss": 2.7762, "step": 180920 }, { "epoch": 0.0057088, "grad_norm": 0.8053156137466431, "learning_rate": 1.1175678858373484e-05, "loss": 2.8678, "step": 180930 }, { "epoch": 0.0057344, "grad_norm": 0.7723068594932556, "learning_rate": 1.1174879160815278e-05, "loss": 2.9373, "step": 180940 }, { "epoch": 0.00576, "grad_norm": 1.0739632844924927, "learning_rate": 1.1174079455638288e-05, "loss": 3.104, "step": 180950 }, { "epoch": 0.0057856, "grad_norm": 0.7357975840568542, "learning_rate": 1.117327974284771e-05, "loss": 3.0087, "step": 180960 }, { "epoch": 0.0058112, "grad_norm": 0.9279378652572632, "learning_rate": 1.1172480022448723e-05, "loss": 2.7298, "step": 180970 }, { "epoch": 0.0058368, "grad_norm": 0.8572146892547607, "learning_rate": 1.1171680294446512e-05, "loss": 2.9612, "step": 180980 }, { "epoch": 0.0058624, "grad_norm": 0.778934121131897, "learning_rate": 1.1170880558846269e-05, "loss": 2.7478, "step": 180990 }, { "epoch": 0.005888, "grad_norm": 0.795319676399231, "learning_rate": 1.1170080815653175e-05, "loss": 2.7261, "step": 181000 }, { "epoch": 0.0059136, "grad_norm": 0.7912237048149109, "learning_rate": 1.1169281064872417e-05, "loss": 2.9441, "step": 181010 }, { "epoch": 0.0059392, "grad_norm": 0.7235998511314392, "learning_rate": 1.1168481306509182e-05, "loss": 2.7628, "step": 181020 }, { "epoch": 0.0059648, "grad_norm": 0.7180615067481995, "learning_rate": 1.1167681540568657e-05, "loss": 2.9837, "step": 181030 }, { "epoch": 0.0059904, "grad_norm": 0.8027224540710449, "learning_rate": 1.116688176705603e-05, "loss": 3.0658, "step": 181040 }, { "epoch": 0.006016, "grad_norm": 0.7743443250656128, "learning_rate": 1.1166081985976482e-05, "loss": 2.987, "step": 181050 }, { "epoch": 0.0060416, "grad_norm": 0.7389100193977356, "learning_rate": 1.1165282197335202e-05, "loss": 2.8491, "step": 181060 }, { "epoch": 0.0060672, "grad_norm": 0.9244176149368286, "learning_rate": 1.1164482401137379e-05, "loss": 3.0389, "step": 181070 }, { "epoch": 0.0060928, "grad_norm": 0.9315748810768127, "learning_rate": 1.1163682597388194e-05, "loss": 2.7879, "step": 181080 }, { "epoch": 0.0061184, "grad_norm": 0.8785526752471924, "learning_rate": 1.1162882786092837e-05, "loss": 2.8319, "step": 181090 }, { "epoch": 0.006144, "grad_norm": 0.7211942076683044, "learning_rate": 1.1162082967256498e-05, "loss": 2.8689, "step": 181100 }, { "epoch": 0.0061696, "grad_norm": 0.8269826769828796, "learning_rate": 1.1161283140884355e-05, "loss": 2.854, "step": 181110 }, { "epoch": 0.0061952, "grad_norm": 0.7860661149024963, "learning_rate": 1.1160483306981604e-05, "loss": 2.8389, "step": 181120 }, { "epoch": 0.0062208, "grad_norm": 0.7578710913658142, "learning_rate": 1.1159683465553424e-05, "loss": 2.8751, "step": 181130 }, { "epoch": 0.0062464, "grad_norm": 0.9972310662269592, "learning_rate": 1.1158883616605008e-05, "loss": 2.9727, "step": 181140 }, { "epoch": 0.006272, "grad_norm": 1.1213080883026123, "learning_rate": 1.1158083760141535e-05, "loss": 2.8968, "step": 181150 }, { "epoch": 0.0062976, "grad_norm": 0.9545733332633972, "learning_rate": 1.1157283896168201e-05, "loss": 2.9219, "step": 181160 }, { "epoch": 0.0063232, "grad_norm": 0.862927258014679, "learning_rate": 1.1156484024690186e-05, "loss": 3.0094, "step": 181170 }, { "epoch": 0.0063488, "grad_norm": 0.8002909421920776, "learning_rate": 1.115568414571268e-05, "loss": 3.0202, "step": 181180 }, { "epoch": 0.0063744, "grad_norm": 0.904219925403595, "learning_rate": 1.1154884259240868e-05, "loss": 3.0082, "step": 181190 }, { "epoch": 0.0064, "grad_norm": 1.0513391494750977, "learning_rate": 1.1154084365279943e-05, "loss": 2.9639, "step": 181200 }, { "epoch": 0.0064256, "grad_norm": 0.8819170594215393, "learning_rate": 1.1153284463835083e-05, "loss": 2.9935, "step": 181210 }, { "epoch": 0.0064512, "grad_norm": 0.685364842414856, "learning_rate": 1.1152484554911482e-05, "loss": 2.8716, "step": 181220 }, { "epoch": 0.0064768, "grad_norm": 0.7844609618186951, "learning_rate": 1.1151684638514323e-05, "loss": 2.9095, "step": 181230 }, { "epoch": 0.0065024, "grad_norm": 0.8620690107345581, "learning_rate": 1.1150884714648799e-05, "loss": 2.8296, "step": 181240 }, { "epoch": 0.006528, "grad_norm": 0.8633244037628174, "learning_rate": 1.1150084783320093e-05, "loss": 2.9846, "step": 181250 }, { "epoch": 0.0065536, "grad_norm": 0.8688656091690063, "learning_rate": 1.1149284844533388e-05, "loss": 2.9057, "step": 181260 }, { "epoch": 0.0065792, "grad_norm": 0.782691478729248, "learning_rate": 1.1148484898293882e-05, "loss": 2.9504, "step": 181270 }, { "epoch": 0.0066048, "grad_norm": 0.7988200187683105, "learning_rate": 1.1147684944606751e-05, "loss": 2.8184, "step": 181280 }, { "epoch": 0.0066304, "grad_norm": 0.7690287232398987, "learning_rate": 1.1146884983477191e-05, "loss": 2.8024, "step": 181290 }, { "epoch": 0.006656, "grad_norm": 0.7340044975280762, "learning_rate": 1.1146085014910388e-05, "loss": 2.9872, "step": 181300 }, { "epoch": 0.0066816, "grad_norm": 0.7783244252204895, "learning_rate": 1.1145285038911527e-05, "loss": 2.8751, "step": 181310 }, { "epoch": 0.0067072, "grad_norm": 0.8737711906433105, "learning_rate": 1.1144485055485797e-05, "loss": 2.8435, "step": 181320 }, { "epoch": 0.0067328, "grad_norm": 0.9239711165428162, "learning_rate": 1.1143685064638386e-05, "loss": 2.9583, "step": 181330 }, { "epoch": 0.0067584, "grad_norm": 0.7356252074241638, "learning_rate": 1.114288506637448e-05, "loss": 2.9881, "step": 181340 }, { "epoch": 0.006784, "grad_norm": 0.8447418808937073, "learning_rate": 1.1142085060699269e-05, "loss": 3.233, "step": 181350 }, { "epoch": 0.0068096, "grad_norm": 0.7650920152664185, "learning_rate": 1.1141285047617938e-05, "loss": 2.8262, "step": 181360 }, { "epoch": 0.0068352, "grad_norm": 0.898368239402771, "learning_rate": 1.1140485027135681e-05, "loss": 3.0441, "step": 181370 }, { "epoch": 0.0068608, "grad_norm": 0.7947142124176025, "learning_rate": 1.1139684999257679e-05, "loss": 2.8916, "step": 181380 }, { "epoch": 0.0068864, "grad_norm": 0.8988890051841736, "learning_rate": 1.1138884963989123e-05, "loss": 2.8949, "step": 181390 }, { "epoch": 0.006912, "grad_norm": 0.7918432950973511, "learning_rate": 1.1138084921335201e-05, "loss": 2.6896, "step": 181400 }, { "epoch": 0.0069376, "grad_norm": 0.7782970666885376, "learning_rate": 1.1137284871301099e-05, "loss": 2.9857, "step": 181410 }, { "epoch": 0.0069632, "grad_norm": 0.7948915362358093, "learning_rate": 1.113648481389201e-05, "loss": 2.8946, "step": 181420 }, { "epoch": 0.0069888, "grad_norm": 0.7497816681861877, "learning_rate": 1.1135684749113117e-05, "loss": 3.022, "step": 181430 }, { "epoch": 0.0070144, "grad_norm": 0.7222577333450317, "learning_rate": 1.1134884676969608e-05, "loss": 2.8698, "step": 181440 }, { "epoch": 0.00704, "grad_norm": 0.8113381266593933, "learning_rate": 1.1134084597466677e-05, "loss": 2.6836, "step": 181450 }, { "epoch": 0.0070656, "grad_norm": 0.8011505603790283, "learning_rate": 1.1133284510609508e-05, "loss": 2.694, "step": 181460 }, { "epoch": 0.0070912, "grad_norm": 0.8089151978492737, "learning_rate": 1.1132484416403291e-05, "loss": 3.1176, "step": 181470 }, { "epoch": 0.0071168, "grad_norm": 0.7732824087142944, "learning_rate": 1.1131684314853212e-05, "loss": 3.0022, "step": 181480 }, { "epoch": 0.0071424, "grad_norm": 0.8158890604972839, "learning_rate": 1.113088420596446e-05, "loss": 3.0434, "step": 181490 }, { "epoch": 0.007168, "grad_norm": 0.8028304576873779, "learning_rate": 1.1130084089742228e-05, "loss": 2.8768, "step": 181500 }, { "epoch": 0.0071936, "grad_norm": 0.8411784172058105, "learning_rate": 1.11292839661917e-05, "loss": 2.759, "step": 181510 }, { "epoch": 0.0072192, "grad_norm": 0.8070506453514099, "learning_rate": 1.112848383531806e-05, "loss": 3.0219, "step": 181520 }, { "epoch": 0.0072448, "grad_norm": 0.9871681332588196, "learning_rate": 1.112768369712651e-05, "loss": 3.0456, "step": 181530 }, { "epoch": 0.0072704, "grad_norm": 0.7939916849136353, "learning_rate": 1.1126883551622228e-05, "loss": 3.0185, "step": 181540 }, { "epoch": 0.007296, "grad_norm": 0.7958455681800842, "learning_rate": 1.1126083398810403e-05, "loss": 2.9443, "step": 181550 }, { "epoch": 0.0073216, "grad_norm": 0.9284798502922058, "learning_rate": 1.112528323869623e-05, "loss": 3.0295, "step": 181560 }, { "epoch": 0.0073472, "grad_norm": 0.774563729763031, "learning_rate": 1.1124483071284895e-05, "loss": 3.0472, "step": 181570 }, { "epoch": 0.0073728, "grad_norm": 0.8857006430625916, "learning_rate": 1.1123682896581582e-05, "loss": 2.9745, "step": 181580 }, { "epoch": 0.0073984, "grad_norm": 0.9724896550178528, "learning_rate": 1.1122882714591488e-05, "loss": 2.9567, "step": 181590 }, { "epoch": 0.007424, "grad_norm": 0.7567624449729919, "learning_rate": 1.1122082525319795e-05, "loss": 2.9948, "step": 181600 }, { "epoch": 0.0074496, "grad_norm": 0.708784282207489, "learning_rate": 1.1121282328771697e-05, "loss": 2.8584, "step": 181610 }, { "epoch": 0.0074752, "grad_norm": 0.7037550806999207, "learning_rate": 1.1120482124952378e-05, "loss": 2.8722, "step": 181620 }, { "epoch": 0.0075008, "grad_norm": 0.7394558191299438, "learning_rate": 1.1119681913867036e-05, "loss": 2.8149, "step": 181630 }, { "epoch": 0.0075264, "grad_norm": 0.8138246536254883, "learning_rate": 1.111888169552085e-05, "loss": 2.9709, "step": 181640 }, { "epoch": 0.007552, "grad_norm": 0.731019914150238, "learning_rate": 1.1118081469919013e-05, "loss": 3.1217, "step": 181650 }, { "epoch": 0.0075776, "grad_norm": 0.7211238145828247, "learning_rate": 1.1117281237066716e-05, "loss": 2.9851, "step": 181660 }, { "epoch": 0.0076032, "grad_norm": 0.9083707928657532, "learning_rate": 1.111648099696915e-05, "loss": 3.08, "step": 181670 }, { "epoch": 0.0076288, "grad_norm": 0.8530842065811157, "learning_rate": 1.1115680749631498e-05, "loss": 3.0495, "step": 181680 }, { "epoch": 0.0076544, "grad_norm": 0.7342842817306519, "learning_rate": 1.1114880495058955e-05, "loss": 2.9349, "step": 181690 }, { "epoch": 0.00768, "grad_norm": 0.851687490940094, "learning_rate": 1.1114080233256705e-05, "loss": 2.9147, "step": 181700 }, { "epoch": 0.0077056, "grad_norm": 0.7911025881767273, "learning_rate": 1.1113279964229944e-05, "loss": 3.0238, "step": 181710 }, { "epoch": 0.0077312, "grad_norm": 0.6697909235954285, "learning_rate": 1.1112479687983856e-05, "loss": 2.8316, "step": 181720 }, { "epoch": 0.0077568, "grad_norm": 0.7806631326675415, "learning_rate": 1.1111679404523637e-05, "loss": 2.9223, "step": 181730 }, { "epoch": 0.0077824, "grad_norm": 0.8072689175605774, "learning_rate": 1.1110879113854469e-05, "loss": 3.0982, "step": 181740 }, { "epoch": 0.007808, "grad_norm": 0.8845444917678833, "learning_rate": 1.1110078815981545e-05, "loss": 3.0538, "step": 181750 }, { "epoch": 0.0078336, "grad_norm": 0.7213308215141296, "learning_rate": 1.1109278510910057e-05, "loss": 2.8922, "step": 181760 }, { "epoch": 0.0078592, "grad_norm": 0.8943037986755371, "learning_rate": 1.1108478198645191e-05, "loss": 3.0105, "step": 181770 }, { "epoch": 0.0078848, "grad_norm": 0.7742500901222229, "learning_rate": 1.1107677879192136e-05, "loss": 2.9408, "step": 181780 }, { "epoch": 0.0079104, "grad_norm": 0.8047808408737183, "learning_rate": 1.1106877552556091e-05, "loss": 3.1371, "step": 181790 }, { "epoch": 0.007936, "grad_norm": 0.8027930855751038, "learning_rate": 1.1106077218742233e-05, "loss": 2.8548, "step": 181800 }, { "epoch": 0.0079616, "grad_norm": 0.6731576919555664, "learning_rate": 1.110527687775576e-05, "loss": 2.843, "step": 181810 }, { "epoch": 0.0079872, "grad_norm": 0.7775254249572754, "learning_rate": 1.1104476529601863e-05, "loss": 2.8732, "step": 181820 }, { "epoch": 0.0080128, "grad_norm": 0.7870674729347229, "learning_rate": 1.1103676174285727e-05, "loss": 2.8743, "step": 181830 }, { "epoch": 0.0080384, "grad_norm": 0.7920750975608826, "learning_rate": 1.1102875811812543e-05, "loss": 2.9754, "step": 181840 }, { "epoch": 0.008064, "grad_norm": 0.7883992195129395, "learning_rate": 1.1102075442187504e-05, "loss": 3.0971, "step": 181850 }, { "epoch": 0.0080896, "grad_norm": 0.8114142417907715, "learning_rate": 1.11012750654158e-05, "loss": 2.8986, "step": 181860 }, { "epoch": 0.0081152, "grad_norm": 0.8127996921539307, "learning_rate": 1.1100474681502615e-05, "loss": 3.2738, "step": 181870 }, { "epoch": 0.0081408, "grad_norm": 0.7317397594451904, "learning_rate": 1.1099674290453148e-05, "loss": 2.8977, "step": 181880 }, { "epoch": 0.0081664, "grad_norm": 0.7461296319961548, "learning_rate": 1.1098873892272588e-05, "loss": 3.0838, "step": 181890 }, { "epoch": 0.008192, "grad_norm": 0.8942358493804932, "learning_rate": 1.1098073486966118e-05, "loss": 3.1813, "step": 181900 }, { "epoch": 0.0082176, "grad_norm": 0.7745600938796997, "learning_rate": 1.1097273074538934e-05, "loss": 2.9313, "step": 181910 }, { "epoch": 0.0082432, "grad_norm": 0.9391730427742004, "learning_rate": 1.1096472654996229e-05, "loss": 2.8243, "step": 181920 }, { "epoch": 0.0082688, "grad_norm": 0.7480381727218628, "learning_rate": 1.1095672228343189e-05, "loss": 2.9227, "step": 181930 }, { "epoch": 0.0082944, "grad_norm": 0.7358924746513367, "learning_rate": 1.1094871794585004e-05, "loss": 3.0948, "step": 181940 }, { "epoch": 0.00832, "grad_norm": 0.8321477174758911, "learning_rate": 1.1094071353726864e-05, "loss": 2.9362, "step": 181950 }, { "epoch": 0.0083456, "grad_norm": 0.8375934958457947, "learning_rate": 1.1093270905773968e-05, "loss": 3.0024, "step": 181960 }, { "epoch": 0.0083712, "grad_norm": 0.8183842301368713, "learning_rate": 1.1092470450731499e-05, "loss": 2.8826, "step": 181970 }, { "epoch": 0.0083968, "grad_norm": 0.7572598457336426, "learning_rate": 1.1091669988604648e-05, "loss": 3.0159, "step": 181980 }, { "epoch": 0.0084224, "grad_norm": 0.8015252947807312, "learning_rate": 1.1090869519398609e-05, "loss": 2.8361, "step": 181990 }, { "epoch": 0.008448, "grad_norm": 0.8110958337783813, "learning_rate": 1.109006904311857e-05, "loss": 3.0113, "step": 182000 }, { "epoch": 0.0084736, "grad_norm": 0.8056759238243103, "learning_rate": 1.108926855976972e-05, "loss": 2.789, "step": 182010 }, { "epoch": 0.0084992, "grad_norm": 0.7721429467201233, "learning_rate": 1.1088468069357256e-05, "loss": 3.033, "step": 182020 }, { "epoch": 0.0085248, "grad_norm": 0.7194506525993347, "learning_rate": 1.108766757188637e-05, "loss": 2.8295, "step": 182030 }, { "epoch": 0.0085504, "grad_norm": 0.7484130859375, "learning_rate": 1.1086867067362238e-05, "loss": 3.015, "step": 182040 }, { "epoch": 0.008576, "grad_norm": 0.9267652630805969, "learning_rate": 1.108606655579007e-05, "loss": 2.9637, "step": 182050 }, { "epoch": 0.0086016, "grad_norm": 0.8454791903495789, "learning_rate": 1.1085266037175049e-05, "loss": 3.0629, "step": 182060 }, { "epoch": 0.0086272, "grad_norm": 0.8250845670700073, "learning_rate": 1.1084465511522365e-05, "loss": 2.9403, "step": 182070 }, { "epoch": 0.0086528, "grad_norm": 0.9809964299201965, "learning_rate": 1.108366497883721e-05, "loss": 3.0167, "step": 182080 }, { "epoch": 0.0086784, "grad_norm": 0.7376616597175598, "learning_rate": 1.1082864439124777e-05, "loss": 2.8629, "step": 182090 }, { "epoch": 0.008704, "grad_norm": 0.8457229137420654, "learning_rate": 1.1082063892390253e-05, "loss": 2.9172, "step": 182100 }, { "epoch": 0.0087296, "grad_norm": 0.6918458938598633, "learning_rate": 1.1081263338638833e-05, "loss": 3.018, "step": 182110 }, { "epoch": 0.0087552, "grad_norm": 0.838036835193634, "learning_rate": 1.1080462777875708e-05, "loss": 3.1235, "step": 182120 }, { "epoch": 0.0087808, "grad_norm": 0.7792806029319763, "learning_rate": 1.107966221010607e-05, "loss": 2.9564, "step": 182130 }, { "epoch": 0.0088064, "grad_norm": 0.9218632578849792, "learning_rate": 1.1078861635335107e-05, "loss": 2.9039, "step": 182140 }, { "epoch": 0.008832, "grad_norm": 0.7397317290306091, "learning_rate": 1.1078061053568015e-05, "loss": 2.9044, "step": 182150 }, { "epoch": 0.0088576, "grad_norm": 0.7077484130859375, "learning_rate": 1.1077260464809984e-05, "loss": 2.8552, "step": 182160 }, { "epoch": 0.0088832, "grad_norm": 0.986209511756897, "learning_rate": 1.1076459869066205e-05, "loss": 3.2263, "step": 182170 }, { "epoch": 0.0089088, "grad_norm": 1.0808534622192383, "learning_rate": 1.107565926634187e-05, "loss": 3.1158, "step": 182180 }, { "epoch": 0.0089344, "grad_norm": 0.8484922647476196, "learning_rate": 1.1074858656642172e-05, "loss": 2.983, "step": 182190 }, { "epoch": 0.00896, "grad_norm": 0.862973690032959, "learning_rate": 1.1074058039972298e-05, "loss": 2.6738, "step": 182200 }, { "epoch": 0.0089856, "grad_norm": 0.8636783957481384, "learning_rate": 1.1073257416337441e-05, "loss": 2.8225, "step": 182210 }, { "epoch": 0.0090112, "grad_norm": 0.6994662880897522, "learning_rate": 1.1072456785742803e-05, "loss": 2.8937, "step": 182220 }, { "epoch": 0.0090368, "grad_norm": 0.9072562456130981, "learning_rate": 1.1071656148193561e-05, "loss": 2.9152, "step": 182230 }, { "epoch": 0.0090624, "grad_norm": 0.8316283822059631, "learning_rate": 1.1070855503694915e-05, "loss": 3.296, "step": 182240 }, { "epoch": 0.009088, "grad_norm": 0.8506913781166077, "learning_rate": 1.1070054852252056e-05, "loss": 2.8426, "step": 182250 }, { "epoch": 0.0091136, "grad_norm": 0.9180954694747925, "learning_rate": 1.1069254193870178e-05, "loss": 2.9625, "step": 182260 }, { "epoch": 0.0091392, "grad_norm": 0.820547878742218, "learning_rate": 1.1068453528554469e-05, "loss": 3.2819, "step": 182270 }, { "epoch": 0.0091648, "grad_norm": 0.8861138224601746, "learning_rate": 1.106765285631012e-05, "loss": 2.9966, "step": 182280 }, { "epoch": 0.0091904, "grad_norm": 0.835882306098938, "learning_rate": 1.106685217714233e-05, "loss": 3.0542, "step": 182290 }, { "epoch": 0.009216, "grad_norm": 0.8358424305915833, "learning_rate": 1.1066051491056284e-05, "loss": 2.8636, "step": 182300 }, { "epoch": 0.0092416, "grad_norm": 0.7852123379707336, "learning_rate": 1.106525079805718e-05, "loss": 2.996, "step": 182310 }, { "epoch": 0.0092672, "grad_norm": 0.856354296207428, "learning_rate": 1.1064450098150207e-05, "loss": 2.9887, "step": 182320 }, { "epoch": 0.0092928, "grad_norm": 0.8125008940696716, "learning_rate": 1.106364939134056e-05, "loss": 3.0546, "step": 182330 }, { "epoch": 0.0093184, "grad_norm": 0.7629084587097168, "learning_rate": 1.1062848677633427e-05, "loss": 2.9933, "step": 182340 }, { "epoch": 0.009344, "grad_norm": 0.7660284638404846, "learning_rate": 1.1062047957034001e-05, "loss": 2.9258, "step": 182350 }, { "epoch": 0.0093696, "grad_norm": 0.9126073122024536, "learning_rate": 1.106124722954748e-05, "loss": 2.6268, "step": 182360 }, { "epoch": 0.0093952, "grad_norm": 0.860480785369873, "learning_rate": 1.1060446495179052e-05, "loss": 2.9146, "step": 182370 }, { "epoch": 0.0094208, "grad_norm": 0.9064407348632812, "learning_rate": 1.1059645753933907e-05, "loss": 3.1116, "step": 182380 }, { "epoch": 0.0094464, "grad_norm": 0.7248519062995911, "learning_rate": 1.1058845005817242e-05, "loss": 2.8115, "step": 182390 }, { "epoch": 0.009472, "grad_norm": 0.7487176656723022, "learning_rate": 1.105804425083425e-05, "loss": 2.9201, "step": 182400 }, { "epoch": 0.0094976, "grad_norm": 0.7174089550971985, "learning_rate": 1.1057243488990124e-05, "loss": 2.8081, "step": 182410 }, { "epoch": 0.0095232, "grad_norm": 0.7283899188041687, "learning_rate": 1.1056442720290053e-05, "loss": 3.0195, "step": 182420 }, { "epoch": 0.0095488, "grad_norm": 1.9542518854141235, "learning_rate": 1.1055641944739233e-05, "loss": 3.0284, "step": 182430 }, { "epoch": 0.0095744, "grad_norm": 0.7717635631561279, "learning_rate": 1.1054841162342853e-05, "loss": 3.0553, "step": 182440 }, { "epoch": 0.0096, "grad_norm": 0.7778887152671814, "learning_rate": 1.105404037310611e-05, "loss": 2.7788, "step": 182450 }, { "epoch": 0.0096256, "grad_norm": 0.7746303081512451, "learning_rate": 1.1053239577034197e-05, "loss": 2.8605, "step": 182460 }, { "epoch": 0.0096512, "grad_norm": 0.8862619400024414, "learning_rate": 1.10524387741323e-05, "loss": 2.95, "step": 182470 }, { "epoch": 0.0096768, "grad_norm": 0.7266414761543274, "learning_rate": 1.1051637964405622e-05, "loss": 2.8884, "step": 182480 }, { "epoch": 0.0097024, "grad_norm": 0.8322254419326782, "learning_rate": 1.1050837147859347e-05, "loss": 2.9213, "step": 182490 }, { "epoch": 0.009728, "grad_norm": 0.8300888538360596, "learning_rate": 1.1050036324498676e-05, "loss": 2.974, "step": 182500 }, { "epoch": 0.0097536, "grad_norm": 1.8328063488006592, "learning_rate": 1.1049235494328799e-05, "loss": 3.0636, "step": 182510 }, { "epoch": 0.0097792, "grad_norm": 0.7597084045410156, "learning_rate": 1.1048434657354907e-05, "loss": 2.8106, "step": 182520 }, { "epoch": 0.0098048, "grad_norm": 0.7726423740386963, "learning_rate": 1.1047633813582194e-05, "loss": 2.9859, "step": 182530 }, { "epoch": 0.0098304, "grad_norm": 0.7708535194396973, "learning_rate": 1.1046832963015857e-05, "loss": 3.0141, "step": 182540 }, { "epoch": 0.009856, "grad_norm": 0.8466053605079651, "learning_rate": 1.1046032105661083e-05, "loss": 2.7119, "step": 182550 }, { "epoch": 0.0098816, "grad_norm": 0.7321096062660217, "learning_rate": 1.1045231241523072e-05, "loss": 2.929, "step": 182560 }, { "epoch": 0.0099072, "grad_norm": 0.897915244102478, "learning_rate": 1.104443037060701e-05, "loss": 3.0645, "step": 182570 }, { "epoch": 0.0099328, "grad_norm": 0.9244039058685303, "learning_rate": 1.10436294929181e-05, "loss": 2.881, "step": 182580 }, { "epoch": 0.0099584, "grad_norm": 1.5334947109222412, "learning_rate": 1.1042828608461529e-05, "loss": 3.4498, "step": 182590 }, { "epoch": 0.009984, "grad_norm": 0.6998804807662964, "learning_rate": 1.104202771724249e-05, "loss": 2.7111, "step": 182600 }, { "epoch": 0.0100096, "grad_norm": 0.8614511489868164, "learning_rate": 1.1041226819266177e-05, "loss": 2.8806, "step": 182610 }, { "epoch": 0.0100352, "grad_norm": 0.8023002743721008, "learning_rate": 1.1040425914537787e-05, "loss": 2.9644, "step": 182620 }, { "epoch": 0.0100608, "grad_norm": 0.8092998266220093, "learning_rate": 1.103962500306251e-05, "loss": 2.7365, "step": 182630 }, { "epoch": 0.0100864, "grad_norm": 0.790992259979248, "learning_rate": 1.1038824084845543e-05, "loss": 3.0018, "step": 182640 }, { "epoch": 0.010112, "grad_norm": 0.7040772438049316, "learning_rate": 1.1038023159892075e-05, "loss": 3.101, "step": 182650 }, { "epoch": 0.0101376, "grad_norm": 0.6955679655075073, "learning_rate": 1.1037222228207305e-05, "loss": 2.9469, "step": 182660 }, { "epoch": 0.0101632, "grad_norm": 0.7149716019630432, "learning_rate": 1.1036421289796421e-05, "loss": 2.9657, "step": 182670 }, { "epoch": 0.0101888, "grad_norm": 0.7244393229484558, "learning_rate": 1.1035620344664626e-05, "loss": 3.0447, "step": 182680 }, { "epoch": 0.0102144, "grad_norm": 0.7608321309089661, "learning_rate": 1.1034819392817106e-05, "loss": 3.0967, "step": 182690 }, { "epoch": 0.01024, "grad_norm": 0.7211750149726868, "learning_rate": 1.1034018434259054e-05, "loss": 2.8788, "step": 182700 }, { "epoch": 0.0102656, "grad_norm": 0.7805671095848083, "learning_rate": 1.1033217468995671e-05, "loss": 2.7862, "step": 182710 }, { "epoch": 0.0102912, "grad_norm": 0.7866308093070984, "learning_rate": 1.1032416497032147e-05, "loss": 2.9331, "step": 182720 }, { "epoch": 0.0103168, "grad_norm": 0.7442153692245483, "learning_rate": 1.1031615518373678e-05, "loss": 2.9624, "step": 182730 }, { "epoch": 0.0103424, "grad_norm": 0.7976658344268799, "learning_rate": 1.1030814533025451e-05, "loss": 3.0181, "step": 182740 }, { "epoch": 0.010368, "grad_norm": 1.0129802227020264, "learning_rate": 1.103001354099267e-05, "loss": 2.9995, "step": 182750 }, { "epoch": 0.0103936, "grad_norm": 0.7364873886108398, "learning_rate": 1.1029212542280526e-05, "loss": 2.9898, "step": 182760 }, { "epoch": 0.0104192, "grad_norm": 0.8107487559318542, "learning_rate": 1.1028411536894208e-05, "loss": 2.6499, "step": 182770 }, { "epoch": 0.0104448, "grad_norm": 0.7528256773948669, "learning_rate": 1.1027610524838919e-05, "loss": 2.8447, "step": 182780 }, { "epoch": 0.0104704, "grad_norm": 0.7751381993293762, "learning_rate": 1.1026809506119844e-05, "loss": 2.9463, "step": 182790 }, { "epoch": 0.010496, "grad_norm": 0.8479676842689514, "learning_rate": 1.1026008480742185e-05, "loss": 2.8367, "step": 182800 }, { "epoch": 0.0105216, "grad_norm": 0.6874504089355469, "learning_rate": 1.1025207448711133e-05, "loss": 2.8446, "step": 182810 }, { "epoch": 0.0105472, "grad_norm": 0.8666626214981079, "learning_rate": 1.1024406410031884e-05, "loss": 3.1641, "step": 182820 }, { "epoch": 0.0105728, "grad_norm": 0.7758886814117432, "learning_rate": 1.102360536470963e-05, "loss": 3.0158, "step": 182830 }, { "epoch": 0.0105984, "grad_norm": 0.7485368847846985, "learning_rate": 1.1022804312749568e-05, "loss": 2.829, "step": 182840 }, { "epoch": 0.010624, "grad_norm": 0.7836526036262512, "learning_rate": 1.1022003254156895e-05, "loss": 3.0265, "step": 182850 }, { "epoch": 0.0106496, "grad_norm": 0.9168951511383057, "learning_rate": 1.1021202188936798e-05, "loss": 2.8342, "step": 182860 }, { "epoch": 0.0106752, "grad_norm": 0.7252046465873718, "learning_rate": 1.102040111709448e-05, "loss": 2.7767, "step": 182870 }, { "epoch": 0.0107008, "grad_norm": 0.8314619064331055, "learning_rate": 1.1019600038635131e-05, "loss": 3.0163, "step": 182880 }, { "epoch": 0.0107264, "grad_norm": 0.908734917640686, "learning_rate": 1.1018798953563945e-05, "loss": 3.0957, "step": 182890 }, { "epoch": 0.010752, "grad_norm": 0.7141382694244385, "learning_rate": 1.1017997861886118e-05, "loss": 2.9166, "step": 182900 }, { "epoch": 0.0107776, "grad_norm": 0.8304566144943237, "learning_rate": 1.1017196763606847e-05, "loss": 2.8749, "step": 182910 }, { "epoch": 0.0108032, "grad_norm": 0.8105591535568237, "learning_rate": 1.1016395658731324e-05, "loss": 2.8512, "step": 182920 }, { "epoch": 0.0108288, "grad_norm": 0.8151239156723022, "learning_rate": 1.1015594547264747e-05, "loss": 2.9452, "step": 182930 }, { "epoch": 0.0108544, "grad_norm": 0.7391579747200012, "learning_rate": 1.1014793429212309e-05, "loss": 3.0804, "step": 182940 }, { "epoch": 0.01088, "grad_norm": 0.778017520904541, "learning_rate": 1.1013992304579207e-05, "loss": 3.0135, "step": 182950 }, { "epoch": 0.0109056, "grad_norm": 0.8385372757911682, "learning_rate": 1.101319117337063e-05, "loss": 3.0521, "step": 182960 }, { "epoch": 0.0109312, "grad_norm": 0.8089011311531067, "learning_rate": 1.101239003559178e-05, "loss": 2.9602, "step": 182970 }, { "epoch": 0.0109568, "grad_norm": 0.7248254418373108, "learning_rate": 1.101158889124785e-05, "loss": 3.0731, "step": 182980 }, { "epoch": 0.0109824, "grad_norm": 0.7274981737136841, "learning_rate": 1.1010787740344036e-05, "loss": 2.8994, "step": 182990 }, { "epoch": 0.011008, "grad_norm": 0.8142207860946655, "learning_rate": 1.1009986582885527e-05, "loss": 3.0996, "step": 183000 }, { "epoch": 0.0110336, "grad_norm": 0.9080842137336731, "learning_rate": 1.1009185418877528e-05, "loss": 2.9238, "step": 183010 }, { "epoch": 0.0110592, "grad_norm": 0.8588999509811401, "learning_rate": 1.1008384248325231e-05, "loss": 3.2809, "step": 183020 }, { "epoch": 0.0110848, "grad_norm": 0.9236361384391785, "learning_rate": 1.1007583071233826e-05, "loss": 3.0509, "step": 183030 }, { "epoch": 0.0111104, "grad_norm": 0.7764927744865417, "learning_rate": 1.1006781887608513e-05, "loss": 2.8438, "step": 183040 }, { "epoch": 0.011136, "grad_norm": 0.8354451656341553, "learning_rate": 1.100598069745449e-05, "loss": 2.9977, "step": 183050 }, { "epoch": 0.0111616, "grad_norm": 0.8902186155319214, "learning_rate": 1.1005179500776948e-05, "loss": 2.9019, "step": 183060 }, { "epoch": 0.0111872, "grad_norm": 0.7037246823310852, "learning_rate": 1.1004378297581084e-05, "loss": 3.076, "step": 183070 }, { "epoch": 0.0112128, "grad_norm": 0.9357329607009888, "learning_rate": 1.1003577087872095e-05, "loss": 2.9806, "step": 183080 }, { "epoch": 0.0112384, "grad_norm": 0.7361576557159424, "learning_rate": 1.1002775871655174e-05, "loss": 2.7977, "step": 183090 }, { "epoch": 0.011264, "grad_norm": 0.7873843908309937, "learning_rate": 1.1001974648935517e-05, "loss": 3.0589, "step": 183100 }, { "epoch": 0.0112896, "grad_norm": 0.8914549350738525, "learning_rate": 1.1001173419718322e-05, "loss": 3.0957, "step": 183110 }, { "epoch": 0.0113152, "grad_norm": 0.9514195919036865, "learning_rate": 1.1000372184008783e-05, "loss": 3.0118, "step": 183120 }, { "epoch": 0.0113408, "grad_norm": 0.8008498549461365, "learning_rate": 1.0999570941812097e-05, "loss": 2.922, "step": 183130 }, { "epoch": 0.0113664, "grad_norm": 0.8971365094184875, "learning_rate": 1.0998769693133458e-05, "loss": 3.1088, "step": 183140 }, { "epoch": 0.011392, "grad_norm": 0.834219753742218, "learning_rate": 1.0997968437978067e-05, "loss": 2.7788, "step": 183150 }, { "epoch": 0.0114176, "grad_norm": 0.7181866765022278, "learning_rate": 1.099716717635111e-05, "loss": 2.8692, "step": 183160 }, { "epoch": 0.0114432, "grad_norm": 0.895719051361084, "learning_rate": 1.0996365908257791e-05, "loss": 3.0774, "step": 183170 }, { "epoch": 0.0114688, "grad_norm": 0.7863770127296448, "learning_rate": 1.0995564633703303e-05, "loss": 2.8962, "step": 183180 }, { "epoch": 0.0114944, "grad_norm": 0.8840596079826355, "learning_rate": 1.0994763352692846e-05, "loss": 2.8726, "step": 183190 }, { "epoch": 0.01152, "grad_norm": 0.9037365913391113, "learning_rate": 1.099396206523161e-05, "loss": 2.9717, "step": 183200 }, { "epoch": 0.0115456, "grad_norm": 0.8634172081947327, "learning_rate": 1.0993160771324796e-05, "loss": 3.0115, "step": 183210 }, { "epoch": 0.0115712, "grad_norm": 0.7175366878509521, "learning_rate": 1.0992359470977598e-05, "loss": 3.0193, "step": 183220 }, { "epoch": 0.0115968, "grad_norm": 0.8222169280052185, "learning_rate": 1.099155816419521e-05, "loss": 3.0893, "step": 183230 }, { "epoch": 0.0116224, "grad_norm": 0.9846743941307068, "learning_rate": 1.0990756850982832e-05, "loss": 3.0384, "step": 183240 }, { "epoch": 0.011648, "grad_norm": 1.0291149616241455, "learning_rate": 1.0989955531345664e-05, "loss": 2.8644, "step": 183250 }, { "epoch": 0.0116736, "grad_norm": 0.9902107119560242, "learning_rate": 1.0989154205288893e-05, "loss": 2.9775, "step": 183260 }, { "epoch": 0.0116992, "grad_norm": 0.7901538610458374, "learning_rate": 1.098835287281772e-05, "loss": 3.3185, "step": 183270 }, { "epoch": 0.0117248, "grad_norm": 1.165738582611084, "learning_rate": 1.0987551533937341e-05, "loss": 2.854, "step": 183280 }, { "epoch": 0.0117504, "grad_norm": 0.7563243508338928, "learning_rate": 1.0986750188652955e-05, "loss": 3.0281, "step": 183290 }, { "epoch": 0.011776, "grad_norm": 1.2099896669387817, "learning_rate": 1.0985948836969753e-05, "loss": 2.9481, "step": 183300 }, { "epoch": 0.0118016, "grad_norm": 0.7041154503822327, "learning_rate": 1.098514747889294e-05, "loss": 3.0014, "step": 183310 }, { "epoch": 0.0118272, "grad_norm": 0.7309106588363647, "learning_rate": 1.0984346114427702e-05, "loss": 3.0001, "step": 183320 }, { "epoch": 0.0118528, "grad_norm": 1.4315173625946045, "learning_rate": 1.0983544743579244e-05, "loss": 2.9965, "step": 183330 }, { "epoch": 0.0118784, "grad_norm": 0.8157833218574524, "learning_rate": 1.0982743366352758e-05, "loss": 3.0218, "step": 183340 }, { "epoch": 0.011904, "grad_norm": 0.7382116317749023, "learning_rate": 1.0981941982753444e-05, "loss": 2.8568, "step": 183350 }, { "epoch": 0.0119296, "grad_norm": 0.7480598092079163, "learning_rate": 1.0981140592786494e-05, "loss": 2.8947, "step": 183360 }, { "epoch": 0.0119552, "grad_norm": 1.1110069751739502, "learning_rate": 1.0980339196457112e-05, "loss": 3.1731, "step": 183370 }, { "epoch": 0.0119808, "grad_norm": 0.9179084300994873, "learning_rate": 1.0979537793770492e-05, "loss": 2.9989, "step": 183380 }, { "epoch": 0.0120064, "grad_norm": 0.875126838684082, "learning_rate": 1.0978736384731825e-05, "loss": 2.4224, "step": 183390 }, { "epoch": 0.012032, "grad_norm": 0.8643067479133606, "learning_rate": 1.0977934969346316e-05, "loss": 2.532, "step": 183400 }, { "epoch": 0.0120576, "grad_norm": 0.8035053014755249, "learning_rate": 1.097713354761916e-05, "loss": 2.813, "step": 183410 }, { "epoch": 0.0120832, "grad_norm": 0.899473249912262, "learning_rate": 1.0976332119555549e-05, "loss": 2.973, "step": 183420 }, { "epoch": 0.0121088, "grad_norm": 0.7096781134605408, "learning_rate": 1.0975530685160681e-05, "loss": 2.7232, "step": 183430 }, { "epoch": 0.0121344, "grad_norm": 0.7977221608161926, "learning_rate": 1.097472924443976e-05, "loss": 2.9691, "step": 183440 }, { "epoch": 0.01216, "grad_norm": 0.772750735282898, "learning_rate": 1.097392779739798e-05, "loss": 2.8642, "step": 183450 }, { "epoch": 0.0121856, "grad_norm": 0.7834232449531555, "learning_rate": 1.0973126344040534e-05, "loss": 3.6039, "step": 183460 }, { "epoch": 0.0122112, "grad_norm": 0.8485148549079895, "learning_rate": 1.0972324884372627e-05, "loss": 3.2413, "step": 183470 }, { "epoch": 0.0122368, "grad_norm": 1.1934709548950195, "learning_rate": 1.0971523418399447e-05, "loss": 3.0963, "step": 183480 }, { "epoch": 0.0122624, "grad_norm": 0.7639839053153992, "learning_rate": 1.0970721946126197e-05, "loss": 2.9927, "step": 183490 }, { "epoch": 0.012288, "grad_norm": 1.0020688772201538, "learning_rate": 1.0969920467558073e-05, "loss": 2.9595, "step": 183500 }, { "epoch": 0.0123136, "grad_norm": 0.9173044562339783, "learning_rate": 1.0969118982700275e-05, "loss": 3.0328, "step": 183510 }, { "epoch": 0.0123392, "grad_norm": 0.8281455039978027, "learning_rate": 1.0968317491557995e-05, "loss": 3.1007, "step": 183520 }, { "epoch": 0.0123648, "grad_norm": 0.9998189806938171, "learning_rate": 1.0967515994136433e-05, "loss": 3.0603, "step": 183530 }, { "epoch": 0.0123904, "grad_norm": 0.902275562286377, "learning_rate": 1.0966714490440789e-05, "loss": 3.0096, "step": 183540 }, { "epoch": 0.012416, "grad_norm": 0.7060209512710571, "learning_rate": 1.0965912980476257e-05, "loss": 2.8202, "step": 183550 }, { "epoch": 0.0124416, "grad_norm": 0.8180386424064636, "learning_rate": 1.0965111464248038e-05, "loss": 2.886, "step": 183560 }, { "epoch": 0.0124672, "grad_norm": 0.7082303762435913, "learning_rate": 1.0964309941761327e-05, "loss": 2.8847, "step": 183570 }, { "epoch": 0.0124928, "grad_norm": 1.0201915502548218, "learning_rate": 1.096350841302132e-05, "loss": 3.1467, "step": 183580 }, { "epoch": 0.0125184, "grad_norm": 0.8199949264526367, "learning_rate": 1.0962706878033219e-05, "loss": 2.9611, "step": 183590 }, { "epoch": 0.012544, "grad_norm": 0.848825216293335, "learning_rate": 1.0961905336802219e-05, "loss": 2.8393, "step": 183600 }, { "epoch": 0.0125696, "grad_norm": 0.797758936882019, "learning_rate": 1.0961103789333518e-05, "loss": 3.3493, "step": 183610 }, { "epoch": 0.0125952, "grad_norm": 1.0173068046569824, "learning_rate": 1.0960302235632316e-05, "loss": 3.399, "step": 183620 }, { "epoch": 0.0126208, "grad_norm": 0.7986762523651123, "learning_rate": 1.0959500675703807e-05, "loss": 2.8189, "step": 183630 }, { "epoch": 0.0126464, "grad_norm": 0.7565578818321228, "learning_rate": 1.0958699109553195e-05, "loss": 3.0103, "step": 183640 }, { "epoch": 0.012672, "grad_norm": 1.0800065994262695, "learning_rate": 1.0957897537185672e-05, "loss": 3.0246, "step": 183650 }, { "epoch": 0.0126976, "grad_norm": 1.0186277627944946, "learning_rate": 1.0957095958606436e-05, "loss": 2.8027, "step": 183660 }, { "epoch": 0.0127232, "grad_norm": 0.6888662576675415, "learning_rate": 1.095629437382069e-05, "loss": 3.0165, "step": 183670 }, { "epoch": 0.0127488, "grad_norm": 0.7830129265785217, "learning_rate": 1.0955492782833628e-05, "loss": 2.7868, "step": 183680 }, { "epoch": 0.0127744, "grad_norm": 0.7264889478683472, "learning_rate": 1.0954691185650448e-05, "loss": 2.8145, "step": 183690 }, { "epoch": 0.0128, "grad_norm": 2.7788891792297363, "learning_rate": 1.0953889582276352e-05, "loss": 2.8459, "step": 183700 }, { "epoch": 0.0128256, "grad_norm": 0.7666825652122498, "learning_rate": 1.0953087972716534e-05, "loss": 2.8269, "step": 183710 }, { "epoch": 0.0128512, "grad_norm": 1.0262436866760254, "learning_rate": 1.0952286356976195e-05, "loss": 2.7408, "step": 183720 }, { "epoch": 0.0128768, "grad_norm": 0.6849855184555054, "learning_rate": 1.095148473506053e-05, "loss": 2.8072, "step": 183730 }, { "epoch": 0.0129024, "grad_norm": 0.7874980568885803, "learning_rate": 1.095068310697474e-05, "loss": 2.8813, "step": 183740 }, { "epoch": 0.012928, "grad_norm": 0.7472628951072693, "learning_rate": 1.0949881472724024e-05, "loss": 3.0754, "step": 183750 }, { "epoch": 0.0129536, "grad_norm": 1.2307147979736328, "learning_rate": 1.0949079832313579e-05, "loss": 3.1334, "step": 183760 }, { "epoch": 0.0129792, "grad_norm": 0.8070753812789917, "learning_rate": 1.0948278185748602e-05, "loss": 2.7953, "step": 183770 }, { "epoch": 0.0130048, "grad_norm": 0.7003903985023499, "learning_rate": 1.094747653303429e-05, "loss": 2.9747, "step": 183780 }, { "epoch": 0.0130304, "grad_norm": 0.8941916227340698, "learning_rate": 1.0946674874175849e-05, "loss": 2.8443, "step": 183790 }, { "epoch": 0.013056, "grad_norm": 0.7981841564178467, "learning_rate": 1.0945873209178475e-05, "loss": 3.012, "step": 183800 }, { "epoch": 0.0130816, "grad_norm": 0.9356762766838074, "learning_rate": 1.094507153804736e-05, "loss": 2.826, "step": 183810 }, { "epoch": 0.0131072, "grad_norm": 0.9327558279037476, "learning_rate": 1.094426986078771e-05, "loss": 3.0316, "step": 183820 }, { "epoch": 0.0131328, "grad_norm": 0.7687799334526062, "learning_rate": 1.0943468177404719e-05, "loss": 3.2677, "step": 183830 }, { "epoch": 0.0131584, "grad_norm": 0.7643436193466187, "learning_rate": 1.094266648790359e-05, "loss": 3.2359, "step": 183840 }, { "epoch": 0.013184, "grad_norm": 10.145160675048828, "learning_rate": 1.0941864792289519e-05, "loss": 3.3172, "step": 183850 }, { "epoch": 0.0132096, "grad_norm": 2.0993502140045166, "learning_rate": 1.0941063090567703e-05, "loss": 2.8737, "step": 183860 }, { "epoch": 0.0132352, "grad_norm": 1.8616700172424316, "learning_rate": 1.0940261382743344e-05, "loss": 3.2632, "step": 183870 }, { "epoch": 0.0132608, "grad_norm": 1.2228370904922485, "learning_rate": 1.0939459668821639e-05, "loss": 2.9485, "step": 183880 }, { "epoch": 0.0132864, "grad_norm": 1.3793885707855225, "learning_rate": 1.093865794880779e-05, "loss": 2.8041, "step": 183890 }, { "epoch": 0.013312, "grad_norm": 0.7387499809265137, "learning_rate": 1.0937856222706994e-05, "loss": 3.0424, "step": 183900 }, { "epoch": 0.0133376, "grad_norm": 0.752390444278717, "learning_rate": 1.0937054490524445e-05, "loss": 2.8014, "step": 183910 }, { "epoch": 0.0133632, "grad_norm": 1.2534139156341553, "learning_rate": 1.093625275226535e-05, "loss": 3.1398, "step": 183920 }, { "epoch": 0.0133888, "grad_norm": 1.3666473627090454, "learning_rate": 1.0935451007934905e-05, "loss": 2.8952, "step": 183930 }, { "epoch": 0.0134144, "grad_norm": 0.8866696357727051, "learning_rate": 1.093464925753831e-05, "loss": 2.9278, "step": 183940 }, { "epoch": 0.01344, "grad_norm": 0.8438064455986023, "learning_rate": 1.0933847501080758e-05, "loss": 2.8766, "step": 183950 }, { "epoch": 0.0134656, "grad_norm": 0.8882771134376526, "learning_rate": 1.0933045738567458e-05, "loss": 2.7344, "step": 183960 }, { "epoch": 0.0134912, "grad_norm": 0.7417264580726624, "learning_rate": 1.0932243970003603e-05, "loss": 2.9059, "step": 183970 }, { "epoch": 0.0135168, "grad_norm": 0.8688225150108337, "learning_rate": 1.0931442195394392e-05, "loss": 2.9215, "step": 183980 }, { "epoch": 0.0135424, "grad_norm": 2.279038190841675, "learning_rate": 1.0930640414745027e-05, "loss": 3.1755, "step": 183990 }, { "epoch": 0.013568, "grad_norm": 0.77568119764328, "learning_rate": 1.0929838628060708e-05, "loss": 3.0211, "step": 184000 }, { "epoch": 0.0135936, "grad_norm": 0.6959554553031921, "learning_rate": 1.0929036835346632e-05, "loss": 2.8465, "step": 184010 }, { "epoch": 0.0136192, "grad_norm": 0.7007080912590027, "learning_rate": 1.0928235036607997e-05, "loss": 3.0135, "step": 184020 }, { "epoch": 0.0136448, "grad_norm": 0.7910307049751282, "learning_rate": 1.0927433231850004e-05, "loss": 3.0402, "step": 184030 }, { "epoch": 0.0136704, "grad_norm": 0.8333937525749207, "learning_rate": 1.0926631421077853e-05, "loss": 2.9621, "step": 184040 }, { "epoch": 0.013696, "grad_norm": 1.8278971910476685, "learning_rate": 1.0925829604296748e-05, "loss": 2.8094, "step": 184050 }, { "epoch": 0.0137216, "grad_norm": 0.8462402820587158, "learning_rate": 1.0925027781511881e-05, "loss": 3.2944, "step": 184060 }, { "epoch": 0.0137472, "grad_norm": 0.97621089220047, "learning_rate": 1.0924225952728454e-05, "loss": 3.0475, "step": 184070 }, { "epoch": 0.0137728, "grad_norm": 0.8354827761650085, "learning_rate": 1.0923424117951669e-05, "loss": 2.9225, "step": 184080 }, { "epoch": 0.0137984, "grad_norm": 0.8015969395637512, "learning_rate": 1.092262227718672e-05, "loss": 2.9108, "step": 184090 }, { "epoch": 0.013824, "grad_norm": 0.8210272789001465, "learning_rate": 1.0921820430438817e-05, "loss": 3.0288, "step": 184100 }, { "epoch": 0.0138496, "grad_norm": 0.84447181224823, "learning_rate": 1.092101857771315e-05, "loss": 2.9968, "step": 184110 }, { "epoch": 0.0138752, "grad_norm": 0.9895392060279846, "learning_rate": 1.0920216719014922e-05, "loss": 3.002, "step": 184120 }, { "epoch": 0.0139008, "grad_norm": 0.8037201166152954, "learning_rate": 1.0919414854349336e-05, "loss": 3.1319, "step": 184130 }, { "epoch": 0.0139264, "grad_norm": 0.7209386825561523, "learning_rate": 1.0918612983721583e-05, "loss": 2.9183, "step": 184140 }, { "epoch": 0.013952, "grad_norm": 0.8364415764808655, "learning_rate": 1.0917811107136873e-05, "loss": 2.8915, "step": 184150 }, { "epoch": 0.0139776, "grad_norm": 0.7658156156539917, "learning_rate": 1.0917009224600403e-05, "loss": 3.1453, "step": 184160 }, { "epoch": 0.0140032, "grad_norm": 0.7679816484451294, "learning_rate": 1.0916207336117371e-05, "loss": 3.0038, "step": 184170 }, { "epoch": 0.0140288, "grad_norm": 0.8935019969940186, "learning_rate": 1.0915405441692977e-05, "loss": 2.9293, "step": 184180 }, { "epoch": 0.0140544, "grad_norm": 0.7466182112693787, "learning_rate": 1.0914603541332423e-05, "loss": 2.993, "step": 184190 }, { "epoch": 0.01408, "grad_norm": 0.860224723815918, "learning_rate": 1.0913801635040909e-05, "loss": 3.0004, "step": 184200 }, { "epoch": 0.0141056, "grad_norm": 0.9568376541137695, "learning_rate": 1.091299972282363e-05, "loss": 3.0487, "step": 184210 }, { "epoch": 0.0141312, "grad_norm": 0.7976507544517517, "learning_rate": 1.0912197804685793e-05, "loss": 2.7487, "step": 184220 }, { "epoch": 0.0141568, "grad_norm": 1.4112268686294556, "learning_rate": 1.0911395880632598e-05, "loss": 2.8596, "step": 184230 }, { "epoch": 0.0141824, "grad_norm": 1.0406253337860107, "learning_rate": 1.0910593950669239e-05, "loss": 3.0491, "step": 184240 }, { "epoch": 0.014208, "grad_norm": 1.43271803855896, "learning_rate": 1.0909792014800921e-05, "loss": 3.339, "step": 184250 }, { "epoch": 0.0142336, "grad_norm": 0.7734593152999878, "learning_rate": 1.0908990073032849e-05, "loss": 2.7605, "step": 184260 }, { "epoch": 0.0142592, "grad_norm": 0.8316025137901306, "learning_rate": 1.0908188125370211e-05, "loss": 2.9525, "step": 184270 }, { "epoch": 0.0142848, "grad_norm": 0.7988935112953186, "learning_rate": 1.0907386171818218e-05, "loss": 3.0827, "step": 184280 }, { "epoch": 0.0143104, "grad_norm": 0.8021073937416077, "learning_rate": 1.0906584212382066e-05, "loss": 2.9878, "step": 184290 }, { "epoch": 0.014336, "grad_norm": 1.006219744682312, "learning_rate": 1.0905782247066957e-05, "loss": 3.1036, "step": 184300 }, { "epoch": 0.0143616, "grad_norm": 0.7666097283363342, "learning_rate": 1.0904980275878087e-05, "loss": 2.8125, "step": 184310 }, { "epoch": 0.0143872, "grad_norm": 0.8648911118507385, "learning_rate": 1.0904178298820665e-05, "loss": 2.944, "step": 184320 }, { "epoch": 0.0144128, "grad_norm": 0.769220769405365, "learning_rate": 1.0903376315899885e-05, "loss": 3.0173, "step": 184330 }, { "epoch": 0.0144384, "grad_norm": 0.7210599780082703, "learning_rate": 1.0902574327120951e-05, "loss": 2.9914, "step": 184340 }, { "epoch": 0.014464, "grad_norm": 0.9221017956733704, "learning_rate": 1.090177233248906e-05, "loss": 3.0861, "step": 184350 }, { "epoch": 0.0144896, "grad_norm": 0.7131431102752686, "learning_rate": 1.090097033200942e-05, "loss": 3.0392, "step": 184360 }, { "epoch": 0.0145152, "grad_norm": 0.8937090635299683, "learning_rate": 1.0900168325687221e-05, "loss": 2.8797, "step": 184370 }, { "epoch": 0.0145408, "grad_norm": 0.726690411567688, "learning_rate": 1.0899366313527673e-05, "loss": 2.9657, "step": 184380 }, { "epoch": 0.0145664, "grad_norm": 0.8466746807098389, "learning_rate": 1.0898564295535971e-05, "loss": 2.837, "step": 184390 }, { "epoch": 0.014592, "grad_norm": 0.8123103976249695, "learning_rate": 1.0897762271717321e-05, "loss": 2.9372, "step": 184400 }, { "epoch": 0.0146176, "grad_norm": 0.7918586730957031, "learning_rate": 1.0896960242076919e-05, "loss": 2.7834, "step": 184410 }, { "epoch": 0.0146432, "grad_norm": 1.0204120874404907, "learning_rate": 1.0896158206619968e-05, "loss": 2.9905, "step": 184420 }, { "epoch": 0.0146688, "grad_norm": 0.8206148147583008, "learning_rate": 1.0895356165351674e-05, "loss": 2.9475, "step": 184430 }, { "epoch": 0.0146944, "grad_norm": 0.7796416878700256, "learning_rate": 1.089455411827723e-05, "loss": 3.0222, "step": 184440 }, { "epoch": 0.01472, "grad_norm": 0.8180520534515381, "learning_rate": 1.0893752065401839e-05, "loss": 2.9558, "step": 184450 }, { "epoch": 0.0147456, "grad_norm": 0.8015095591545105, "learning_rate": 1.0892950006730707e-05, "loss": 3.0958, "step": 184460 }, { "epoch": 0.0147712, "grad_norm": 0.7654479742050171, "learning_rate": 1.089214794226903e-05, "loss": 3.0672, "step": 184470 }, { "epoch": 0.0147968, "grad_norm": 0.7757717967033386, "learning_rate": 1.0891345872022005e-05, "loss": 3.0469, "step": 184480 }, { "epoch": 0.0148224, "grad_norm": 0.7966974973678589, "learning_rate": 1.0890543795994846e-05, "loss": 2.9654, "step": 184490 }, { "epoch": 0.014848, "grad_norm": 0.7633460760116577, "learning_rate": 1.0889741714192744e-05, "loss": 2.8502, "step": 184500 }, { "epoch": 0.0148736, "grad_norm": 0.7064738273620605, "learning_rate": 1.0888939626620905e-05, "loss": 2.8918, "step": 184510 }, { "epoch": 0.0148992, "grad_norm": 0.8414672613143921, "learning_rate": 1.0888137533284526e-05, "loss": 2.9048, "step": 184520 }, { "epoch": 0.0149248, "grad_norm": 0.7485834956169128, "learning_rate": 1.0887335434188815e-05, "loss": 2.8761, "step": 184530 }, { "epoch": 0.0149504, "grad_norm": 1.0769832134246826, "learning_rate": 1.0886533329338968e-05, "loss": 2.8399, "step": 184540 }, { "epoch": 0.014976, "grad_norm": 0.8283976912498474, "learning_rate": 1.0885731218740188e-05, "loss": 3.0161, "step": 184550 }, { "epoch": 0.0150016, "grad_norm": 0.7711329460144043, "learning_rate": 1.0884929102397675e-05, "loss": 2.9646, "step": 184560 }, { "epoch": 0.0150272, "grad_norm": 0.7927440404891968, "learning_rate": 1.0884126980316632e-05, "loss": 2.633, "step": 184570 }, { "epoch": 0.0150528, "grad_norm": 0.7200552225112915, "learning_rate": 1.0883324852502261e-05, "loss": 2.9221, "step": 184580 }, { "epoch": 0.0150784, "grad_norm": 0.7109236121177673, "learning_rate": 1.0882522718959766e-05, "loss": 2.8191, "step": 184590 }, { "epoch": 0.015104, "grad_norm": 0.7978203892707825, "learning_rate": 1.0881720579694344e-05, "loss": 2.7873, "step": 184600 }, { "epoch": 0.0151296, "grad_norm": 0.7155009508132935, "learning_rate": 1.0880918434711197e-05, "loss": 2.9107, "step": 184610 }, { "epoch": 0.0151552, "grad_norm": 0.9054962396621704, "learning_rate": 1.0880116284015527e-05, "loss": 2.873, "step": 184620 }, { "epoch": 0.0151808, "grad_norm": 0.9354718327522278, "learning_rate": 1.0879314127612541e-05, "loss": 2.9072, "step": 184630 }, { "epoch": 0.0152064, "grad_norm": 0.90694659948349, "learning_rate": 1.0878511965507435e-05, "loss": 3.1004, "step": 184640 }, { "epoch": 0.015232, "grad_norm": 1.0942344665527344, "learning_rate": 1.0877709797705406e-05, "loss": 3.1628, "step": 184650 }, { "epoch": 0.0152576, "grad_norm": 0.7643879652023315, "learning_rate": 1.0876907624211672e-05, "loss": 3.0034, "step": 184660 }, { "epoch": 0.0152832, "grad_norm": 1.601778507232666, "learning_rate": 1.087610544503142e-05, "loss": 3.1061, "step": 184670 }, { "epoch": 0.0153088, "grad_norm": 0.9111954569816589, "learning_rate": 1.0875303260169858e-05, "loss": 2.9957, "step": 184680 }, { "epoch": 0.0153344, "grad_norm": 0.8371458053588867, "learning_rate": 1.0874501069632188e-05, "loss": 2.8715, "step": 184690 }, { "epoch": 0.01536, "grad_norm": 0.8151674270629883, "learning_rate": 1.087369887342361e-05, "loss": 2.9021, "step": 184700 }, { "epoch": 0.0153856, "grad_norm": 0.7121844291687012, "learning_rate": 1.0872896671549326e-05, "loss": 3.0033, "step": 184710 }, { "epoch": 0.0154112, "grad_norm": 0.7067834138870239, "learning_rate": 1.087209446401454e-05, "loss": 2.9955, "step": 184720 }, { "epoch": 0.0154368, "grad_norm": 0.8631871342658997, "learning_rate": 1.0871292250824455e-05, "loss": 2.7948, "step": 184730 }, { "epoch": 0.0154624, "grad_norm": 0.8028181791305542, "learning_rate": 1.0870490031984267e-05, "loss": 3.0741, "step": 184740 }, { "epoch": 0.015488, "grad_norm": 0.7499280571937561, "learning_rate": 1.0869687807499185e-05, "loss": 2.9145, "step": 184750 }, { "epoch": 0.0155136, "grad_norm": 0.7526406049728394, "learning_rate": 1.086888557737441e-05, "loss": 2.8957, "step": 184760 }, { "epoch": 0.0155392, "grad_norm": 0.8595507144927979, "learning_rate": 1.086808334161514e-05, "loss": 2.9347, "step": 184770 }, { "epoch": 0.0155648, "grad_norm": 1.0085703134536743, "learning_rate": 1.0867281100226583e-05, "loss": 2.8496, "step": 184780 }, { "epoch": 0.0155904, "grad_norm": 0.9524253010749817, "learning_rate": 1.086647885321394e-05, "loss": 2.8159, "step": 184790 }, { "epoch": 0.015616, "grad_norm": 0.8826138973236084, "learning_rate": 1.0865676600582409e-05, "loss": 2.7957, "step": 184800 }, { "epoch": 0.0156416, "grad_norm": 0.6954923272132874, "learning_rate": 1.0864874342337196e-05, "loss": 2.9063, "step": 184810 }, { "epoch": 0.0156672, "grad_norm": 0.9597102403640747, "learning_rate": 1.0864072078483501e-05, "loss": 3.0627, "step": 184820 }, { "epoch": 0.0156928, "grad_norm": 0.9366728067398071, "learning_rate": 1.0863269809026528e-05, "loss": 3.1044, "step": 184830 }, { "epoch": 0.0157184, "grad_norm": 0.8978691697120667, "learning_rate": 1.0862467533971482e-05, "loss": 2.8245, "step": 184840 }, { "epoch": 0.015744, "grad_norm": 0.7299543023109436, "learning_rate": 1.086166525332356e-05, "loss": 2.7894, "step": 184850 }, { "epoch": 0.0157696, "grad_norm": 0.8628016114234924, "learning_rate": 1.0860862967087976e-05, "loss": 3.0263, "step": 184860 }, { "epoch": 0.0157952, "grad_norm": 0.7254238724708557, "learning_rate": 1.0860060675269917e-05, "loss": 2.9159, "step": 184870 }, { "epoch": 0.0158208, "grad_norm": 0.7073713541030884, "learning_rate": 1.0859258377874592e-05, "loss": 2.9042, "step": 184880 }, { "epoch": 0.0158464, "grad_norm": 0.714709997177124, "learning_rate": 1.085845607490721e-05, "loss": 2.8916, "step": 184890 }, { "epoch": 0.015872, "grad_norm": 0.8310685753822327, "learning_rate": 1.0857653766372967e-05, "loss": 2.9719, "step": 184900 }, { "epoch": 0.0158976, "grad_norm": 0.7034101486206055, "learning_rate": 1.0856851452277064e-05, "loss": 2.8631, "step": 184910 }, { "epoch": 0.0159232, "grad_norm": 0.836487889289856, "learning_rate": 1.085604913262471e-05, "loss": 2.8536, "step": 184920 }, { "epoch": 0.0159488, "grad_norm": 0.8177892565727234, "learning_rate": 1.0855246807421104e-05, "loss": 3.1096, "step": 184930 }, { "epoch": 0.0159744, "grad_norm": 0.868632435798645, "learning_rate": 1.0854444476671451e-05, "loss": 2.8653, "step": 184940 }, { "epoch": 0.016, "grad_norm": 0.9025883674621582, "learning_rate": 1.0853642140380953e-05, "loss": 3.0121, "step": 184950 }, { "epoch": 0.0160256, "grad_norm": 1.208174467086792, "learning_rate": 1.0852839798554813e-05, "loss": 2.9318, "step": 184960 }, { "epoch": 0.0160512, "grad_norm": 0.7950413823127747, "learning_rate": 1.0852037451198232e-05, "loss": 2.7798, "step": 184970 }, { "epoch": 0.0160768, "grad_norm": 0.7518515586853027, "learning_rate": 1.0851235098316413e-05, "loss": 2.981, "step": 184980 }, { "epoch": 0.0161024, "grad_norm": 0.7556050419807434, "learning_rate": 1.0850432739914565e-05, "loss": 3.1522, "step": 184990 }, { "epoch": 0.016128, "grad_norm": 0.8162241578102112, "learning_rate": 1.0849630375997882e-05, "loss": 3.0858, "step": 185000 }, { "epoch": 0.0161536, "grad_norm": 0.8696116805076599, "learning_rate": 1.0848828006571573e-05, "loss": 2.9492, "step": 185010 }, { "epoch": 0.0161792, "grad_norm": 1.0487350225448608, "learning_rate": 1.0848025631640841e-05, "loss": 3.0225, "step": 185020 }, { "epoch": 0.0162048, "grad_norm": 0.8043749332427979, "learning_rate": 1.0847223251210888e-05, "loss": 2.9128, "step": 185030 }, { "epoch": 0.0162304, "grad_norm": 0.7800036072731018, "learning_rate": 1.0846420865286917e-05, "loss": 2.8142, "step": 185040 }, { "epoch": 0.016256, "grad_norm": 0.7682424187660217, "learning_rate": 1.0845618473874133e-05, "loss": 2.7658, "step": 185050 }, { "epoch": 0.0162816, "grad_norm": 0.7970525622367859, "learning_rate": 1.0844816076977738e-05, "loss": 3.0798, "step": 185060 }, { "epoch": 0.0163072, "grad_norm": 0.8978790044784546, "learning_rate": 1.0844013674602933e-05, "loss": 2.8386, "step": 185070 }, { "epoch": 0.0163328, "grad_norm": 0.8473784327507019, "learning_rate": 1.0843211266754926e-05, "loss": 2.7357, "step": 185080 }, { "epoch": 0.0163584, "grad_norm": 0.7088713645935059, "learning_rate": 1.0842408853438915e-05, "loss": 3.0873, "step": 185090 }, { "epoch": 0.016384, "grad_norm": 0.8400750756263733, "learning_rate": 1.0841606434660108e-05, "loss": 2.8882, "step": 185100 }, { "epoch": 0.0164096, "grad_norm": 0.7200151085853577, "learning_rate": 1.0840804010423707e-05, "loss": 3.1547, "step": 185110 }, { "epoch": 0.0164352, "grad_norm": 0.7468822598457336, "learning_rate": 1.0840001580734915e-05, "loss": 2.9274, "step": 185120 }, { "epoch": 0.0164608, "grad_norm": 0.7864710092544556, "learning_rate": 1.0839199145598938e-05, "loss": 3.1663, "step": 185130 }, { "epoch": 0.0164864, "grad_norm": 0.719144880771637, "learning_rate": 1.0838396705020976e-05, "loss": 3.0347, "step": 185140 }, { "epoch": 0.016512, "grad_norm": 0.75742107629776, "learning_rate": 1.0837594259006237e-05, "loss": 2.9265, "step": 185150 }, { "epoch": 0.0165376, "grad_norm": 0.739548921585083, "learning_rate": 1.0836791807559917e-05, "loss": 3.0138, "step": 185160 }, { "epoch": 0.0165632, "grad_norm": 0.7349414825439453, "learning_rate": 1.0835989350687223e-05, "loss": 2.9977, "step": 185170 }, { "epoch": 0.0165888, "grad_norm": 0.792788565158844, "learning_rate": 1.0835186888393367e-05, "loss": 2.8751, "step": 185180 }, { "epoch": 0.0166144, "grad_norm": 1.078357219696045, "learning_rate": 1.0834384420683544e-05, "loss": 2.9364, "step": 185190 }, { "epoch": 0.01664, "grad_norm": 1.129386067390442, "learning_rate": 1.0833581947562956e-05, "loss": 2.9669, "step": 185200 }, { "epoch": 0.0166656, "grad_norm": 0.8420944809913635, "learning_rate": 1.0832779469036814e-05, "loss": 3.207, "step": 185210 }, { "epoch": 0.0166912, "grad_norm": 0.7606034874916077, "learning_rate": 1.0831976985110319e-05, "loss": 3.0365, "step": 185220 }, { "epoch": 0.0167168, "grad_norm": 0.853888750076294, "learning_rate": 1.0831174495788675e-05, "loss": 3.0488, "step": 185230 }, { "epoch": 0.0167424, "grad_norm": 0.796186625957489, "learning_rate": 1.0830372001077081e-05, "loss": 2.9376, "step": 185240 }, { "epoch": 0.016768, "grad_norm": 1.0279313325881958, "learning_rate": 1.082956950098075e-05, "loss": 2.9715, "step": 185250 }, { "epoch": 0.0167936, "grad_norm": 1.2672488689422607, "learning_rate": 1.0828766995504876e-05, "loss": 2.7794, "step": 185260 }, { "epoch": 0.0168192, "grad_norm": 0.7296774983406067, "learning_rate": 1.0827964484654672e-05, "loss": 2.9672, "step": 185270 }, { "epoch": 0.0168448, "grad_norm": 0.9209886193275452, "learning_rate": 1.082716196843534e-05, "loss": 2.91, "step": 185280 }, { "epoch": 0.0168704, "grad_norm": 0.8058661818504333, "learning_rate": 1.0826359446852079e-05, "loss": 3.0127, "step": 185290 }, { "epoch": 0.016896, "grad_norm": 0.8039480447769165, "learning_rate": 1.0825556919910097e-05, "loss": 3.0462, "step": 185300 }, { "epoch": 0.0169216, "grad_norm": 1.0891610383987427, "learning_rate": 1.0824754387614599e-05, "loss": 3.2575, "step": 185310 }, { "epoch": 0.0169472, "grad_norm": 0.7709659337997437, "learning_rate": 1.0823951849970789e-05, "loss": 3.051, "step": 185320 }, { "epoch": 0.0169728, "grad_norm": 0.7554478645324707, "learning_rate": 1.0823149306983868e-05, "loss": 2.867, "step": 185330 }, { "epoch": 0.0169984, "grad_norm": 0.851015031337738, "learning_rate": 1.0822346758659044e-05, "loss": 2.6822, "step": 185340 }, { "epoch": 0.017024, "grad_norm": 0.8611158132553101, "learning_rate": 1.0821544205001518e-05, "loss": 2.9213, "step": 185350 }, { "epoch": 0.0170496, "grad_norm": 1.0125316381454468, "learning_rate": 1.0820741646016498e-05, "loss": 2.9458, "step": 185360 }, { "epoch": 0.0170752, "grad_norm": 0.8210718631744385, "learning_rate": 1.0819939081709184e-05, "loss": 2.9502, "step": 185370 }, { "epoch": 0.0171008, "grad_norm": 1.3960622549057007, "learning_rate": 1.0819136512084786e-05, "loss": 2.8173, "step": 185380 }, { "epoch": 0.0171264, "grad_norm": 0.8255804777145386, "learning_rate": 1.0818333937148504e-05, "loss": 2.8617, "step": 185390 }, { "epoch": 0.017152, "grad_norm": 0.83418208360672, "learning_rate": 1.0817531356905544e-05, "loss": 3.2473, "step": 185400 }, { "epoch": 0.0171776, "grad_norm": 0.986878514289856, "learning_rate": 1.0816728771361108e-05, "loss": 2.9421, "step": 185410 }, { "epoch": 0.0172032, "grad_norm": 0.8582480549812317, "learning_rate": 1.081592618052041e-05, "loss": 2.7481, "step": 185420 }, { "epoch": 0.0172288, "grad_norm": 1.0690810680389404, "learning_rate": 1.081512358438864e-05, "loss": 2.8632, "step": 185430 }, { "epoch": 0.0172544, "grad_norm": 1.4869383573532104, "learning_rate": 1.0814320982971012e-05, "loss": 3.3525, "step": 185440 }, { "epoch": 0.01728, "grad_norm": 1.053297758102417, "learning_rate": 1.0813518376272733e-05, "loss": 2.9226, "step": 185450 }, { "epoch": 0.0173056, "grad_norm": 0.7366692423820496, "learning_rate": 1.0812715764299e-05, "loss": 2.7903, "step": 185460 }, { "epoch": 0.0173312, "grad_norm": 0.7392699122428894, "learning_rate": 1.0811913147055023e-05, "loss": 3.1934, "step": 185470 }, { "epoch": 0.0173568, "grad_norm": 0.7727123498916626, "learning_rate": 1.0811110524546004e-05, "loss": 2.8334, "step": 185480 }, { "epoch": 0.0173824, "grad_norm": 0.8225556015968323, "learning_rate": 1.0810307896777147e-05, "loss": 3.0056, "step": 185490 }, { "epoch": 0.017408, "grad_norm": 0.861933708190918, "learning_rate": 1.0809505263753662e-05, "loss": 2.9152, "step": 185500 }, { "epoch": 0.0174336, "grad_norm": 0.9161790609359741, "learning_rate": 1.0808702625480747e-05, "loss": 3.0105, "step": 185510 }, { "epoch": 0.0174592, "grad_norm": 0.7798352837562561, "learning_rate": 1.0807899981963613e-05, "loss": 2.9702, "step": 185520 }, { "epoch": 0.0174848, "grad_norm": 0.7431838512420654, "learning_rate": 1.0807097333207461e-05, "loss": 3.0291, "step": 185530 }, { "epoch": 0.0175104, "grad_norm": 0.8735448718070984, "learning_rate": 1.0806294679217498e-05, "loss": 2.9294, "step": 185540 }, { "epoch": 0.017536, "grad_norm": 0.9182337522506714, "learning_rate": 1.0805492019998929e-05, "loss": 3.1341, "step": 185550 }, { "epoch": 0.0175616, "grad_norm": 0.987439751625061, "learning_rate": 1.0804689355556959e-05, "loss": 3.0775, "step": 185560 }, { "epoch": 0.0175872, "grad_norm": 0.7058978080749512, "learning_rate": 1.080388668589679e-05, "loss": 2.8439, "step": 185570 }, { "epoch": 0.0176128, "grad_norm": 1.1706095933914185, "learning_rate": 1.0803084011023632e-05, "loss": 3.0341, "step": 185580 }, { "epoch": 0.0176384, "grad_norm": 1.0047966241836548, "learning_rate": 1.0802281330942686e-05, "loss": 3.1597, "step": 185590 }, { "epoch": 0.017664, "grad_norm": 0.9806581735610962, "learning_rate": 1.0801478645659158e-05, "loss": 2.9693, "step": 185600 }, { "epoch": 0.0176896, "grad_norm": 1.344558596611023, "learning_rate": 1.0800675955178255e-05, "loss": 2.9528, "step": 185610 }, { "epoch": 0.0177152, "grad_norm": 0.8267179727554321, "learning_rate": 1.079987325950518e-05, "loss": 3.0166, "step": 185620 }, { "epoch": 0.0177408, "grad_norm": 0.779114305973053, "learning_rate": 1.079907055864514e-05, "loss": 3.0662, "step": 185630 }, { "epoch": 0.0177664, "grad_norm": 0.7192792296409607, "learning_rate": 1.0798267852603338e-05, "loss": 2.9999, "step": 185640 }, { "epoch": 0.017792, "grad_norm": 0.8996811509132385, "learning_rate": 1.0797465141384987e-05, "loss": 2.9163, "step": 185650 }, { "epoch": 0.0178176, "grad_norm": 0.773369312286377, "learning_rate": 1.0796662424995284e-05, "loss": 2.9537, "step": 185660 }, { "epoch": 0.0178432, "grad_norm": 0.9267645478248596, "learning_rate": 1.0795859703439433e-05, "loss": 2.8213, "step": 185670 }, { "epoch": 0.0178688, "grad_norm": 0.752470076084137, "learning_rate": 1.0795056976722649e-05, "loss": 2.7915, "step": 185680 }, { "epoch": 0.0178944, "grad_norm": 0.8387195467948914, "learning_rate": 1.0794254244850127e-05, "loss": 3.0017, "step": 185690 }, { "epoch": 0.01792, "grad_norm": 0.9397264122962952, "learning_rate": 1.0793451507827077e-05, "loss": 2.9463, "step": 185700 }, { "epoch": 0.0179456, "grad_norm": 1.0825406312942505, "learning_rate": 1.079264876565871e-05, "loss": 2.8056, "step": 185710 }, { "epoch": 0.0179712, "grad_norm": 0.7337681651115417, "learning_rate": 1.0791846018350223e-05, "loss": 3.1408, "step": 185720 }, { "epoch": 0.0179968, "grad_norm": 0.8051760792732239, "learning_rate": 1.0791043265906827e-05, "loss": 2.8059, "step": 185730 }, { "epoch": 0.0180224, "grad_norm": 0.7636560797691345, "learning_rate": 1.0790240508333723e-05, "loss": 2.8705, "step": 185740 }, { "epoch": 0.018048, "grad_norm": 0.8058044910430908, "learning_rate": 1.0789437745636122e-05, "loss": 3.0226, "step": 185750 }, { "epoch": 0.0180736, "grad_norm": 0.7063872814178467, "learning_rate": 1.0788634977819225e-05, "loss": 3.1241, "step": 185760 }, { "epoch": 0.0180992, "grad_norm": 0.8840232491493225, "learning_rate": 1.078783220488824e-05, "loss": 3.0661, "step": 185770 }, { "epoch": 0.0181248, "grad_norm": 0.851635754108429, "learning_rate": 1.0787029426848372e-05, "loss": 3.0554, "step": 185780 }, { "epoch": 0.0181504, "grad_norm": 0.873243510723114, "learning_rate": 1.078622664370483e-05, "loss": 3.026, "step": 185790 }, { "epoch": 0.018176, "grad_norm": 1.2204718589782715, "learning_rate": 1.0785423855462814e-05, "loss": 2.7383, "step": 185800 }, { "epoch": 0.0182016, "grad_norm": 0.7344863414764404, "learning_rate": 1.0784621062127536e-05, "loss": 2.9195, "step": 185810 }, { "epoch": 0.0182272, "grad_norm": 0.7842313051223755, "learning_rate": 1.0783818263704195e-05, "loss": 3.1221, "step": 185820 }, { "epoch": 0.0182528, "grad_norm": 0.7467857599258423, "learning_rate": 1.0783015460198005e-05, "loss": 3.0945, "step": 185830 }, { "epoch": 0.0182784, "grad_norm": 1.442676305770874, "learning_rate": 1.0782212651614165e-05, "loss": 3.1461, "step": 185840 }, { "epoch": 0.018304, "grad_norm": 1.2416954040527344, "learning_rate": 1.0781409837957888e-05, "loss": 2.9007, "step": 185850 }, { "epoch": 0.0183296, "grad_norm": 0.8343684673309326, "learning_rate": 1.078060701923437e-05, "loss": 2.9415, "step": 185860 }, { "epoch": 0.0183552, "grad_norm": 0.7559362053871155, "learning_rate": 1.0779804195448826e-05, "loss": 3.0623, "step": 185870 }, { "epoch": 0.0183808, "grad_norm": 0.730778157711029, "learning_rate": 1.0779001366606458e-05, "loss": 3.1257, "step": 185880 }, { "epoch": 0.0184064, "grad_norm": 1.3148144483566284, "learning_rate": 1.0778198532712474e-05, "loss": 2.9602, "step": 185890 }, { "epoch": 0.018432, "grad_norm": 0.8812267184257507, "learning_rate": 1.0777395693772079e-05, "loss": 3.0023, "step": 185900 }, { "epoch": 0.0184576, "grad_norm": 0.7066873908042908, "learning_rate": 1.0776592849790481e-05, "loss": 2.7383, "step": 185910 }, { "epoch": 0.0184832, "grad_norm": 0.7779138684272766, "learning_rate": 1.0775790000772884e-05, "loss": 2.9227, "step": 185920 }, { "epoch": 0.0185088, "grad_norm": 2.108520984649658, "learning_rate": 1.0774987146724494e-05, "loss": 2.8854, "step": 185930 }, { "epoch": 0.0185344, "grad_norm": 0.7840317487716675, "learning_rate": 1.0774184287650523e-05, "loss": 3.1629, "step": 185940 }, { "epoch": 0.01856, "grad_norm": 0.7429167628288269, "learning_rate": 1.0773381423556164e-05, "loss": 2.9975, "step": 185950 }, { "epoch": 0.0185856, "grad_norm": 0.8099053502082825, "learning_rate": 1.0772578554446636e-05, "loss": 2.7968, "step": 185960 }, { "epoch": 0.0186112, "grad_norm": 0.700048565864563, "learning_rate": 1.0771775680327143e-05, "loss": 2.8589, "step": 185970 }, { "epoch": 0.0186368, "grad_norm": 0.7056670784950256, "learning_rate": 1.077097280120289e-05, "loss": 2.8627, "step": 185980 }, { "epoch": 0.0186624, "grad_norm": 0.8149152994155884, "learning_rate": 1.0770169917079081e-05, "loss": 2.8824, "step": 185990 }, { "epoch": 0.018688, "grad_norm": 0.7668315768241882, "learning_rate": 1.0769367027960926e-05, "loss": 2.986, "step": 186000 }, { "epoch": 0.0187136, "grad_norm": 0.6957628130912781, "learning_rate": 1.0768564133853632e-05, "loss": 2.9292, "step": 186010 }, { "epoch": 0.0187392, "grad_norm": 0.7533958554267883, "learning_rate": 1.0767761234762402e-05, "loss": 3.0476, "step": 186020 }, { "epoch": 0.0187648, "grad_norm": 0.9516646862030029, "learning_rate": 1.0766958330692442e-05, "loss": 2.8519, "step": 186030 }, { "epoch": 0.0187904, "grad_norm": 0.8081334233283997, "learning_rate": 1.0766155421648965e-05, "loss": 3.112, "step": 186040 }, { "epoch": 0.018816, "grad_norm": 0.7227655053138733, "learning_rate": 1.0765352507637173e-05, "loss": 2.8828, "step": 186050 }, { "epoch": 0.0188416, "grad_norm": 1.4796361923217773, "learning_rate": 1.076454958866227e-05, "loss": 2.918, "step": 186060 }, { "epoch": 0.0188672, "grad_norm": 0.8163363337516785, "learning_rate": 1.076374666472947e-05, "loss": 2.8599, "step": 186070 }, { "epoch": 0.0188928, "grad_norm": 0.9270524978637695, "learning_rate": 1.0762943735843977e-05, "loss": 3.0094, "step": 186080 }, { "epoch": 0.0189184, "grad_norm": 0.7664945721626282, "learning_rate": 1.0762140802010992e-05, "loss": 3.1159, "step": 186090 }, { "epoch": 0.018944, "grad_norm": 0.9223145842552185, "learning_rate": 1.0761337863235728e-05, "loss": 3.0221, "step": 186100 }, { "epoch": 0.0189696, "grad_norm": 0.7740216851234436, "learning_rate": 1.0760534919523393e-05, "loss": 3.0081, "step": 186110 }, { "epoch": 0.0189952, "grad_norm": 1.1933236122131348, "learning_rate": 1.0759731970879187e-05, "loss": 2.8972, "step": 186120 }, { "epoch": 0.0190208, "grad_norm": 0.8448615670204163, "learning_rate": 1.0758929017308324e-05, "loss": 2.9091, "step": 186130 }, { "epoch": 0.0190464, "grad_norm": 0.7571372985839844, "learning_rate": 1.0758126058816007e-05, "loss": 2.887, "step": 186140 }, { "epoch": 0.019072, "grad_norm": 0.7894822359085083, "learning_rate": 1.0757323095407446e-05, "loss": 3.0584, "step": 186150 }, { "epoch": 0.0190976, "grad_norm": 1.879651427268982, "learning_rate": 1.0756520127087842e-05, "loss": 2.9021, "step": 186160 }, { "epoch": 0.0191232, "grad_norm": 0.7633472681045532, "learning_rate": 1.0755717153862411e-05, "loss": 2.9416, "step": 186170 }, { "epoch": 0.0191488, "grad_norm": 0.8057932257652283, "learning_rate": 1.075491417573635e-05, "loss": 3.0289, "step": 186180 }, { "epoch": 0.0191744, "grad_norm": 0.7285999059677124, "learning_rate": 1.0754111192714876e-05, "loss": 2.7939, "step": 186190 }, { "epoch": 0.0192, "grad_norm": 0.8340796828269958, "learning_rate": 1.0753308204803188e-05, "loss": 3.1442, "step": 186200 }, { "epoch": 0.0192256, "grad_norm": 0.7574070692062378, "learning_rate": 1.0752505212006498e-05, "loss": 3.1045, "step": 186210 }, { "epoch": 0.0192512, "grad_norm": 0.7778252363204956, "learning_rate": 1.0751702214330008e-05, "loss": 2.8985, "step": 186220 }, { "epoch": 0.0192768, "grad_norm": 0.800798237323761, "learning_rate": 1.0750899211778933e-05, "loss": 2.846, "step": 186230 }, { "epoch": 0.0193024, "grad_norm": 0.7592502236366272, "learning_rate": 1.0750096204358477e-05, "loss": 2.8887, "step": 186240 }, { "epoch": 0.019328, "grad_norm": 0.6839293241500854, "learning_rate": 1.0749293192073845e-05, "loss": 3.1176, "step": 186250 }, { "epoch": 0.0193536, "grad_norm": 0.8681856989860535, "learning_rate": 1.0748490174930244e-05, "loss": 2.8576, "step": 186260 }, { "epoch": 0.0193792, "grad_norm": 0.7760247588157654, "learning_rate": 1.0747687152932886e-05, "loss": 3.0569, "step": 186270 }, { "epoch": 0.0194048, "grad_norm": 1.1219370365142822, "learning_rate": 1.0746884126086973e-05, "loss": 2.1348, "step": 186280 }, { "epoch": 0.0194304, "grad_norm": 0.8950125575065613, "learning_rate": 1.0746081094397717e-05, "loss": 2.7927, "step": 186290 }, { "epoch": 0.019456, "grad_norm": 0.8048008680343628, "learning_rate": 1.0745278057870323e-05, "loss": 2.5735, "step": 186300 }, { "epoch": 0.0194816, "grad_norm": 0.7091920375823975, "learning_rate": 1.0744475016509997e-05, "loss": 2.9426, "step": 186310 }, { "epoch": 0.0195072, "grad_norm": 0.7278200387954712, "learning_rate": 1.074367197032195e-05, "loss": 2.7852, "step": 186320 }, { "epoch": 0.0195328, "grad_norm": 0.7288029789924622, "learning_rate": 1.0742868919311386e-05, "loss": 2.8534, "step": 186330 }, { "epoch": 0.0195584, "grad_norm": 0.7892138361930847, "learning_rate": 1.0742065863483519e-05, "loss": 2.8166, "step": 186340 }, { "epoch": 0.019584, "grad_norm": 0.7829885482788086, "learning_rate": 1.0741262802843548e-05, "loss": 2.9654, "step": 186350 }, { "epoch": 0.0196096, "grad_norm": 0.6888365149497986, "learning_rate": 1.0740459737396687e-05, "loss": 2.8374, "step": 186360 }, { "epoch": 0.0196352, "grad_norm": 0.7492437362670898, "learning_rate": 1.0739656667148141e-05, "loss": 2.8599, "step": 186370 }, { "epoch": 0.0196608, "grad_norm": 0.8881680369377136, "learning_rate": 1.0738853592103118e-05, "loss": 2.7526, "step": 186380 }, { "epoch": 0.0196864, "grad_norm": 0.8568975329399109, "learning_rate": 1.0738050512266822e-05, "loss": 2.7249, "step": 186390 }, { "epoch": 0.019712, "grad_norm": 0.7816208600997925, "learning_rate": 1.0737247427644468e-05, "loss": 2.4489, "step": 186400 }, { "epoch": 0.0197376, "grad_norm": 0.8841120004653931, "learning_rate": 1.073644433824126e-05, "loss": 2.9555, "step": 186410 }, { "epoch": 0.0197632, "grad_norm": 0.8965635895729065, "learning_rate": 1.0735641244062406e-05, "loss": 2.9676, "step": 186420 }, { "epoch": 0.0197888, "grad_norm": 0.7532212138175964, "learning_rate": 1.0734838145113115e-05, "loss": 2.7769, "step": 186430 }, { "epoch": 0.0198144, "grad_norm": 0.8368003368377686, "learning_rate": 1.0734035041398596e-05, "loss": 2.9676, "step": 186440 }, { "epoch": 0.01984, "grad_norm": 1.1374378204345703, "learning_rate": 1.073323193292405e-05, "loss": 2.8428, "step": 186450 }, { "epoch": 0.0198656, "grad_norm": 0.9346131086349487, "learning_rate": 1.0732428819694692e-05, "loss": 2.8216, "step": 186460 }, { "epoch": 0.0198912, "grad_norm": 0.7724200487136841, "learning_rate": 1.0731625701715732e-05, "loss": 2.834, "step": 186470 }, { "epoch": 0.0199168, "grad_norm": 0.770233154296875, "learning_rate": 1.0730822578992365e-05, "loss": 2.7225, "step": 186480 }, { "epoch": 0.0199424, "grad_norm": 0.8349646925926208, "learning_rate": 1.0730019451529812e-05, "loss": 2.5513, "step": 186490 }, { "epoch": 0.019968, "grad_norm": 0.8280562162399292, "learning_rate": 1.0729216319333277e-05, "loss": 2.7561, "step": 186500 }, { "epoch": 0.0199936, "grad_norm": 0.7924100756645203, "learning_rate": 1.0728413182407969e-05, "loss": 2.7828, "step": 186510 }, { "epoch": 0.0200192, "grad_norm": 0.8062343001365662, "learning_rate": 1.0727610040759093e-05, "loss": 2.7316, "step": 186520 }, { "epoch": 0.0200448, "grad_norm": 0.8811689019203186, "learning_rate": 1.072680689439186e-05, "loss": 2.7309, "step": 186530 }, { "epoch": 0.0200704, "grad_norm": 0.8671133518218994, "learning_rate": 1.0726003743311481e-05, "loss": 2.5846, "step": 186540 }, { "epoch": 0.020096, "grad_norm": 1.0474680662155151, "learning_rate": 1.0725200587523157e-05, "loss": 2.7715, "step": 186550 }, { "epoch": 0.0201216, "grad_norm": 0.7223705053329468, "learning_rate": 1.0724397427032101e-05, "loss": 2.8239, "step": 186560 }, { "epoch": 0.0201472, "grad_norm": 0.8572948575019836, "learning_rate": 1.0723594261843519e-05, "loss": 2.9807, "step": 186570 }, { "epoch": 0.0201728, "grad_norm": 0.7337327003479004, "learning_rate": 1.0722791091962622e-05, "loss": 2.7215, "step": 186580 }, { "epoch": 0.0201984, "grad_norm": 0.7211017608642578, "learning_rate": 1.0721987917394613e-05, "loss": 2.7283, "step": 186590 }, { "epoch": 0.020224, "grad_norm": 0.7055988907814026, "learning_rate": 1.072118473814471e-05, "loss": 2.9255, "step": 186600 }, { "epoch": 0.0202496, "grad_norm": 0.8230187892913818, "learning_rate": 1.0720381554218114e-05, "loss": 2.8155, "step": 186610 }, { "epoch": 0.0202752, "grad_norm": 0.853583812713623, "learning_rate": 1.0719578365620033e-05, "loss": 2.6772, "step": 186620 }, { "epoch": 0.0203008, "grad_norm": 0.7156698107719421, "learning_rate": 1.0718775172355678e-05, "loss": 2.3883, "step": 186630 }, { "epoch": 0.0203264, "grad_norm": 0.8694734573364258, "learning_rate": 1.0717971974430259e-05, "loss": 2.8424, "step": 186640 }, { "epoch": 0.020352, "grad_norm": 0.9696657061576843, "learning_rate": 1.071716877184898e-05, "loss": 3.0903, "step": 186650 }, { "epoch": 0.0203776, "grad_norm": 0.7113258838653564, "learning_rate": 1.0716365564617055e-05, "loss": 2.6996, "step": 186660 }, { "epoch": 0.0204032, "grad_norm": 0.717970609664917, "learning_rate": 1.0715562352739689e-05, "loss": 2.8602, "step": 186670 }, { "epoch": 0.0204288, "grad_norm": 0.8939131498336792, "learning_rate": 1.0714759136222089e-05, "loss": 2.564, "step": 186680 }, { "epoch": 0.0204544, "grad_norm": 0.9328873753547668, "learning_rate": 1.0713955915069466e-05, "loss": 2.7046, "step": 186690 }, { "epoch": 0.02048, "grad_norm": 0.7202339768409729, "learning_rate": 1.0713152689287032e-05, "loss": 2.8998, "step": 186700 }, { "epoch": 0.0205056, "grad_norm": 0.8809983730316162, "learning_rate": 1.0712349458879989e-05, "loss": 2.8215, "step": 186710 }, { "epoch": 0.0205312, "grad_norm": 0.8141489028930664, "learning_rate": 1.071154622385355e-05, "loss": 2.5714, "step": 186720 }, { "epoch": 0.0205568, "grad_norm": 0.9697741270065308, "learning_rate": 1.0710742984212924e-05, "loss": 2.867, "step": 186730 }, { "epoch": 0.0205824, "grad_norm": 0.7286829948425293, "learning_rate": 1.0709939739963318e-05, "loss": 2.7667, "step": 186740 }, { "epoch": 0.020608, "grad_norm": 0.7341431975364685, "learning_rate": 1.0709136491109939e-05, "loss": 2.9737, "step": 186750 }, { "epoch": 0.0206336, "grad_norm": 0.9780291318893433, "learning_rate": 1.0708333237658005e-05, "loss": 2.8231, "step": 186760 }, { "epoch": 0.0206592, "grad_norm": 0.6993506550788879, "learning_rate": 1.0707529979612712e-05, "loss": 2.6879, "step": 186770 }, { "epoch": 0.0206848, "grad_norm": 0.7963603734970093, "learning_rate": 1.0706726716979276e-05, "loss": 2.6812, "step": 186780 }, { "epoch": 0.0207104, "grad_norm": 0.7552944421768188, "learning_rate": 1.0705923449762907e-05, "loss": 2.7698, "step": 186790 }, { "epoch": 0.020736, "grad_norm": 0.8103477954864502, "learning_rate": 1.0705120177968813e-05, "loss": 2.4587, "step": 186800 }, { "epoch": 0.0207616, "grad_norm": 0.9111128449440002, "learning_rate": 1.0704316901602198e-05, "loss": 3.0358, "step": 186810 }, { "epoch": 0.0207872, "grad_norm": 0.7771928906440735, "learning_rate": 1.0703513620668275e-05, "loss": 2.735, "step": 186820 }, { "epoch": 0.0208128, "grad_norm": 0.7674885988235474, "learning_rate": 1.0702710335172258e-05, "loss": 2.8213, "step": 186830 }, { "epoch": 0.0208384, "grad_norm": 0.7664993405342102, "learning_rate": 1.0701907045119349e-05, "loss": 2.9084, "step": 186840 }, { "epoch": 0.020864, "grad_norm": 1.1713628768920898, "learning_rate": 1.0701103750514756e-05, "loss": 2.9933, "step": 186850 }, { "epoch": 0.0208896, "grad_norm": 0.7433863878250122, "learning_rate": 1.0700300451363696e-05, "loss": 2.7765, "step": 186860 }, { "epoch": 0.0209152, "grad_norm": 0.8497755527496338, "learning_rate": 1.0699497147671373e-05, "loss": 2.2281, "step": 186870 }, { "epoch": 0.0209408, "grad_norm": 0.8583799600601196, "learning_rate": 1.0698693839442996e-05, "loss": 2.7467, "step": 186880 }, { "epoch": 0.0209664, "grad_norm": 0.7447980642318726, "learning_rate": 1.0697890526683774e-05, "loss": 2.5955, "step": 186890 }, { "epoch": 0.020992, "grad_norm": 0.9406354427337646, "learning_rate": 1.0697087209398922e-05, "loss": 2.8678, "step": 186900 }, { "epoch": 2.56e-05, "grad_norm": 0.8512160181999207, "learning_rate": 1.0696283887593636e-05, "loss": 2.156, "step": 186910 }, { "epoch": 5.12e-05, "grad_norm": 0.8163312077522278, "learning_rate": 1.0695480561273141e-05, "loss": 2.8552, "step": 186920 }, { "epoch": 7.68e-05, "grad_norm": 1.079697608947754, "learning_rate": 1.0694677230442638e-05, "loss": 3.0999, "step": 186930 }, { "epoch": 0.0001024, "grad_norm": 0.7474485039710999, "learning_rate": 1.0693873895107338e-05, "loss": 2.7461, "step": 186940 }, { "epoch": 0.000128, "grad_norm": 0.8080676198005676, "learning_rate": 1.069307055527245e-05, "loss": 2.898, "step": 186950 }, { "epoch": 0.0001536, "grad_norm": 0.7837024927139282, "learning_rate": 1.0692267210943185e-05, "loss": 2.9109, "step": 186960 }, { "epoch": 0.0001792, "grad_norm": 0.7491540312767029, "learning_rate": 1.069146386212475e-05, "loss": 2.7344, "step": 186970 }, { "epoch": 0.0002048, "grad_norm": 0.812974214553833, "learning_rate": 1.0690660508822354e-05, "loss": 2.887, "step": 186980 }, { "epoch": 0.0002304, "grad_norm": 0.7806909084320068, "learning_rate": 1.0689857151041209e-05, "loss": 2.7197, "step": 186990 }, { "epoch": 0.000256, "grad_norm": 0.798923909664154, "learning_rate": 1.0689053788786524e-05, "loss": 2.8481, "step": 187000 }, { "epoch": 0.0002816, "grad_norm": 0.8040066957473755, "learning_rate": 1.0688250422063506e-05, "loss": 2.6167, "step": 187010 }, { "epoch": 0.0003072, "grad_norm": 0.7565546631813049, "learning_rate": 1.0687447050877372e-05, "loss": 2.8367, "step": 187020 }, { "epoch": 0.0003328, "grad_norm": 0.8744239211082458, "learning_rate": 1.0686643675233323e-05, "loss": 2.8958, "step": 187030 }, { "epoch": 0.0003584, "grad_norm": 0.8275086283683777, "learning_rate": 1.0685840295136573e-05, "loss": 3.0922, "step": 187040 }, { "epoch": 0.000384, "grad_norm": 0.7446326613426208, "learning_rate": 1.068503691059233e-05, "loss": 2.7186, "step": 187050 }, { "epoch": 0.0004096, "grad_norm": 0.8865787982940674, "learning_rate": 1.068423352160581e-05, "loss": 2.951, "step": 187060 }, { "epoch": 0.0004352, "grad_norm": 0.8024170994758606, "learning_rate": 1.0683430128182211e-05, "loss": 2.7452, "step": 187070 }, { "epoch": 0.0004608, "grad_norm": 0.8229860067367554, "learning_rate": 1.068262673032675e-05, "loss": 2.6433, "step": 187080 }, { "epoch": 0.0004864, "grad_norm": 0.9177296161651611, "learning_rate": 1.0681823328044639e-05, "loss": 2.9497, "step": 187090 }, { "epoch": 0.000512, "grad_norm": 0.7502473592758179, "learning_rate": 1.0681019921341083e-05, "loss": 2.839, "step": 187100 }, { "epoch": 0.0005376, "grad_norm": 0.7593597769737244, "learning_rate": 1.0680216510221292e-05, "loss": 2.6913, "step": 187110 }, { "epoch": 0.0005632, "grad_norm": 0.8401113152503967, "learning_rate": 1.0679413094690481e-05, "loss": 2.7198, "step": 187120 }, { "epoch": 0.0005888, "grad_norm": 0.8698801398277283, "learning_rate": 1.0678609674753857e-05, "loss": 2.6146, "step": 187130 }, { "epoch": 0.0006144, "grad_norm": 0.82856285572052, "learning_rate": 1.0677806250416627e-05, "loss": 2.6265, "step": 187140 }, { "epoch": 0.00064, "grad_norm": 0.8604505658149719, "learning_rate": 1.0677002821684006e-05, "loss": 2.9951, "step": 187150 }, { "epoch": 0.0006656, "grad_norm": 0.8112456798553467, "learning_rate": 1.0676199388561201e-05, "loss": 2.7398, "step": 187160 }, { "epoch": 0.0006912, "grad_norm": 0.7450129985809326, "learning_rate": 1.067539595105342e-05, "loss": 3.0447, "step": 187170 }, { "epoch": 0.0007168, "grad_norm": 0.7913231253623962, "learning_rate": 1.0674592509165878e-05, "loss": 2.7831, "step": 187180 }, { "epoch": 0.0007424, "grad_norm": 0.7046387195587158, "learning_rate": 1.0673789062903783e-05, "loss": 2.9101, "step": 187190 }, { "epoch": 0.000768, "grad_norm": 0.6904850602149963, "learning_rate": 1.0672985612272347e-05, "loss": 2.6515, "step": 187200 }, { "epoch": 0.0007936, "grad_norm": 0.7114004492759705, "learning_rate": 1.0672182157276774e-05, "loss": 2.551, "step": 187210 }, { "epoch": 0.0008192, "grad_norm": 0.7467836141586304, "learning_rate": 1.067137869792228e-05, "loss": 2.7029, "step": 187220 }, { "epoch": 0.0008448, "grad_norm": 0.7493768930435181, "learning_rate": 1.0670575234214075e-05, "loss": 2.8088, "step": 187230 }, { "epoch": 0.0008704, "grad_norm": 0.8266785144805908, "learning_rate": 1.0669771766157367e-05, "loss": 2.6542, "step": 187240 }, { "epoch": 0.000896, "grad_norm": 0.8436775207519531, "learning_rate": 1.0668968293757367e-05, "loss": 2.9003, "step": 187250 }, { "epoch": 0.0009216, "grad_norm": 0.7144865393638611, "learning_rate": 1.0668164817019284e-05, "loss": 2.4785, "step": 187260 }, { "epoch": 0.0009472, "grad_norm": 0.7808530926704407, "learning_rate": 1.066736133594833e-05, "loss": 2.9361, "step": 187270 }, { "epoch": 0.0009728, "grad_norm": 0.7097258567810059, "learning_rate": 1.0666557850549714e-05, "loss": 2.8982, "step": 187280 }, { "epoch": 0.0009984, "grad_norm": 0.9389522671699524, "learning_rate": 1.0665754360828652e-05, "loss": 2.8688, "step": 187290 }, { "epoch": 0.001024, "grad_norm": 0.9252989888191223, "learning_rate": 1.0664950866790348e-05, "loss": 2.6547, "step": 187300 }, { "epoch": 0.0010496, "grad_norm": 0.7552286386489868, "learning_rate": 1.0664147368440015e-05, "loss": 2.7995, "step": 187310 }, { "epoch": 0.0010752, "grad_norm": 0.7564489841461182, "learning_rate": 1.0663343865782861e-05, "loss": 2.6552, "step": 187320 }, { "epoch": 0.0011008, "grad_norm": 0.8427779674530029, "learning_rate": 1.0662540358824098e-05, "loss": 3.0647, "step": 187330 }, { "epoch": 0.0011264, "grad_norm": 0.7902590036392212, "learning_rate": 1.0661736847568938e-05, "loss": 2.6538, "step": 187340 }, { "epoch": 0.001152, "grad_norm": 0.8887258172035217, "learning_rate": 1.0660933332022589e-05, "loss": 3.0916, "step": 187350 }, { "epoch": 0.0011776, "grad_norm": 0.7467747330665588, "learning_rate": 1.0660129812190265e-05, "loss": 2.7889, "step": 187360 }, { "epoch": 0.0012032, "grad_norm": 0.770973265171051, "learning_rate": 1.0659326288077173e-05, "loss": 2.8282, "step": 187370 }, { "epoch": 0.0012288, "grad_norm": 0.9691235423088074, "learning_rate": 1.0658522759688526e-05, "loss": 2.8031, "step": 187380 }, { "epoch": 0.0012544, "grad_norm": 0.7355830073356628, "learning_rate": 1.0657719227029534e-05, "loss": 2.9152, "step": 187390 }, { "epoch": 0.00128, "grad_norm": 0.7887231111526489, "learning_rate": 1.065691569010541e-05, "loss": 2.7141, "step": 187400 }, { "epoch": 0.0013056, "grad_norm": 0.817051351070404, "learning_rate": 1.0656112148921357e-05, "loss": 2.6758, "step": 187410 }, { "epoch": 0.0013312, "grad_norm": 0.8414472937583923, "learning_rate": 1.0655308603482596e-05, "loss": 2.492, "step": 187420 }, { "epoch": 0.0013568, "grad_norm": 0.8986482620239258, "learning_rate": 1.0654505053794329e-05, "loss": 3.0145, "step": 187430 }, { "epoch": 0.0013824, "grad_norm": 0.8691920042037964, "learning_rate": 1.0653701499861773e-05, "loss": 2.83, "step": 187440 }, { "epoch": 0.001408, "grad_norm": 0.8190690279006958, "learning_rate": 1.0652897941690137e-05, "loss": 2.7826, "step": 187450 }, { "epoch": 0.0014336, "grad_norm": 0.9680182933807373, "learning_rate": 1.065209437928463e-05, "loss": 2.8739, "step": 187460 }, { "epoch": 0.0014592, "grad_norm": 0.7952733635902405, "learning_rate": 1.0651290812650466e-05, "loss": 2.9143, "step": 187470 }, { "epoch": 0.0014848, "grad_norm": 0.8563818335533142, "learning_rate": 1.0650487241792851e-05, "loss": 2.5438, "step": 187480 }, { "epoch": 0.0015104, "grad_norm": 0.8911154866218567, "learning_rate": 1.0649683666717003e-05, "loss": 2.43, "step": 187490 }, { "epoch": 0.001536, "grad_norm": 0.9293914437294006, "learning_rate": 1.0648880087428126e-05, "loss": 2.9202, "step": 187500 }, { "epoch": 2.56e-05, "grad_norm": 0.8814364671707153, "learning_rate": 1.0648076503931435e-05, "loss": 2.3767, "step": 187510 }, { "epoch": 5.12e-05, "grad_norm": 0.7740890383720398, "learning_rate": 1.0647272916232141e-05, "loss": 2.8215, "step": 187520 }, { "epoch": 7.68e-05, "grad_norm": 0.7911719679832458, "learning_rate": 1.0646469324335452e-05, "loss": 2.8138, "step": 187530 }, { "epoch": 0.0001024, "grad_norm": 0.9145867824554443, "learning_rate": 1.0645665728246581e-05, "loss": 2.7221, "step": 187540 }, { "epoch": 0.000128, "grad_norm": 1.0108970403671265, "learning_rate": 1.0644862127970744e-05, "loss": 2.7786, "step": 187550 }, { "epoch": 0.0001536, "grad_norm": 0.8506478071212769, "learning_rate": 1.0644058523513143e-05, "loss": 2.7057, "step": 187560 }, { "epoch": 0.0001792, "grad_norm": 0.8497013449668884, "learning_rate": 1.0643254914878995e-05, "loss": 2.7055, "step": 187570 }, { "epoch": 0.0002048, "grad_norm": 0.7114152312278748, "learning_rate": 1.064245130207351e-05, "loss": 2.7074, "step": 187580 }, { "epoch": 0.0002304, "grad_norm": 0.8793928027153015, "learning_rate": 1.0641647685101901e-05, "loss": 2.9027, "step": 187590 }, { "epoch": 0.000256, "grad_norm": 0.9196805953979492, "learning_rate": 1.0640844063969375e-05, "loss": 2.7658, "step": 187600 }, { "epoch": 0.0002816, "grad_norm": 0.7673410177230835, "learning_rate": 1.0640040438681143e-05, "loss": 2.7401, "step": 187610 }, { "epoch": 0.0003072, "grad_norm": 0.7234100699424744, "learning_rate": 1.0639236809242423e-05, "loss": 2.6002, "step": 187620 }, { "epoch": 0.0003328, "grad_norm": 0.9615457653999329, "learning_rate": 1.0638433175658421e-05, "loss": 3.1573, "step": 187630 }, { "epoch": 0.0003584, "grad_norm": 0.8546091318130493, "learning_rate": 1.0637629537934349e-05, "loss": 2.8525, "step": 187640 }, { "epoch": 0.000384, "grad_norm": 0.7185772657394409, "learning_rate": 1.063682589607542e-05, "loss": 2.6526, "step": 187650 }, { "epoch": 0.0004096, "grad_norm": 0.7569544315338135, "learning_rate": 1.0636022250086843e-05, "loss": 2.6589, "step": 187660 }, { "epoch": 0.0004352, "grad_norm": 0.7605822682380676, "learning_rate": 1.0635218599973832e-05, "loss": 2.6042, "step": 187670 }, { "epoch": 0.0004608, "grad_norm": 0.8306787610054016, "learning_rate": 1.0634414945741596e-05, "loss": 2.7921, "step": 187680 }, { "epoch": 0.0004864, "grad_norm": 0.7499761581420898, "learning_rate": 1.063361128739535e-05, "loss": 2.8343, "step": 187690 }, { "epoch": 0.000512, "grad_norm": 0.7122966051101685, "learning_rate": 1.06328076249403e-05, "loss": 2.692, "step": 187700 }, { "epoch": 0.0005376, "grad_norm": 0.7432339787483215, "learning_rate": 1.063200395838166e-05, "loss": 2.5786, "step": 187710 }, { "epoch": 0.0005632, "grad_norm": 0.7741906046867371, "learning_rate": 1.0631200287724648e-05, "loss": 2.6431, "step": 187720 }, { "epoch": 0.0005888, "grad_norm": 0.8637678623199463, "learning_rate": 1.0630396612974466e-05, "loss": 2.573, "step": 187730 }, { "epoch": 0.0006144, "grad_norm": 0.8345110416412354, "learning_rate": 1.0629592934136327e-05, "loss": 2.6624, "step": 187740 }, { "epoch": 0.00064, "grad_norm": 0.7531880736351013, "learning_rate": 1.0628789251215453e-05, "loss": 2.6596, "step": 187750 }, { "epoch": 0.0006656, "grad_norm": 0.8561967611312866, "learning_rate": 1.062798556421704e-05, "loss": 2.7578, "step": 187760 }, { "epoch": 0.0006912, "grad_norm": 0.9458214044570923, "learning_rate": 1.062718187314631e-05, "loss": 3.0519, "step": 187770 }, { "epoch": 0.0007168, "grad_norm": 0.7167134881019592, "learning_rate": 1.0626378178008472e-05, "loss": 2.7922, "step": 187780 }, { "epoch": 0.0007424, "grad_norm": 0.8808683156967163, "learning_rate": 1.0625574478808737e-05, "loss": 2.9465, "step": 187790 }, { "epoch": 0.000768, "grad_norm": 0.7718656063079834, "learning_rate": 1.062477077555232e-05, "loss": 2.8253, "step": 187800 }, { "epoch": 0.0007936, "grad_norm": 0.8590803742408752, "learning_rate": 1.062396706824443e-05, "loss": 2.5734, "step": 187810 }, { "epoch": 0.0008192, "grad_norm": 1.5727564096450806, "learning_rate": 1.0623163356890282e-05, "loss": 2.6218, "step": 187820 }, { "epoch": 0.0008448, "grad_norm": 0.8074990510940552, "learning_rate": 1.0622359641495081e-05, "loss": 2.5929, "step": 187830 }, { "epoch": 0.0008704, "grad_norm": 1.061776876449585, "learning_rate": 1.0621555922064044e-05, "loss": 2.9036, "step": 187840 }, { "epoch": 0.000896, "grad_norm": 0.7768974900245667, "learning_rate": 1.0620752198602386e-05, "loss": 2.6249, "step": 187850 }, { "epoch": 0.0009216, "grad_norm": 0.8258517384529114, "learning_rate": 1.0619948471115312e-05, "loss": 2.7435, "step": 187860 }, { "epoch": 0.0009472, "grad_norm": 0.6839719414710999, "learning_rate": 1.0619144739608031e-05, "loss": 2.6615, "step": 187870 }, { "epoch": 0.0009728, "grad_norm": 0.7945087552070618, "learning_rate": 1.061834100408577e-05, "loss": 2.5498, "step": 187880 }, { "epoch": 0.0009984, "grad_norm": 0.8492289185523987, "learning_rate": 1.0617537264553728e-05, "loss": 2.7101, "step": 187890 }, { "epoch": 0.001024, "grad_norm": 0.7893586158752441, "learning_rate": 1.0616733521017121e-05, "loss": 2.675, "step": 187900 }, { "epoch": 0.0010496, "grad_norm": 0.7094335556030273, "learning_rate": 1.0615929773481162e-05, "loss": 2.5833, "step": 187910 }, { "epoch": 0.0010752, "grad_norm": 0.7971278429031372, "learning_rate": 1.0615126021951063e-05, "loss": 2.8021, "step": 187920 }, { "epoch": 0.0011008, "grad_norm": 0.8100708723068237, "learning_rate": 1.0614322266432033e-05, "loss": 2.8915, "step": 187930 }, { "epoch": 0.0011264, "grad_norm": 0.9400140047073364, "learning_rate": 1.0613518506929285e-05, "loss": 2.88, "step": 187940 }, { "epoch": 0.001152, "grad_norm": 0.8932065963745117, "learning_rate": 1.0612714743448039e-05, "loss": 2.7248, "step": 187950 }, { "epoch": 0.0011776, "grad_norm": 0.7807867527008057, "learning_rate": 1.0611910975993493e-05, "loss": 2.6515, "step": 187960 }, { "epoch": 0.0012032, "grad_norm": 0.7770896553993225, "learning_rate": 1.0611107204570869e-05, "loss": 2.8844, "step": 187970 }, { "epoch": 0.0012288, "grad_norm": 0.8199320435523987, "learning_rate": 1.061030342918538e-05, "loss": 2.8124, "step": 187980 }, { "epoch": 0.0012544, "grad_norm": 0.7900694012641907, "learning_rate": 1.0609499649842235e-05, "loss": 3.0994, "step": 187990 }, { "epoch": 0.00128, "grad_norm": 0.7067587971687317, "learning_rate": 1.0608695866546645e-05, "loss": 2.6722, "step": 188000 }, { "epoch": 0.0013056, "grad_norm": 1.1017595529556274, "learning_rate": 1.0607892079303823e-05, "loss": 2.8219, "step": 188010 }, { "epoch": 0.0013312, "grad_norm": 0.7670609951019287, "learning_rate": 1.0607088288118988e-05, "loss": 2.7001, "step": 188020 }, { "epoch": 0.0013568, "grad_norm": 0.7673234343528748, "learning_rate": 1.0606284492997343e-05, "loss": 2.8395, "step": 188030 }, { "epoch": 0.0013824, "grad_norm": 0.8437913060188293, "learning_rate": 1.0605480693944101e-05, "loss": 2.6732, "step": 188040 }, { "epoch": 0.001408, "grad_norm": 0.7955170273780823, "learning_rate": 1.0604676890964482e-05, "loss": 2.6807, "step": 188050 }, { "epoch": 0.0014336, "grad_norm": 0.7131841778755188, "learning_rate": 1.0603873084063691e-05, "loss": 2.7197, "step": 188060 }, { "epoch": 0.0014592, "grad_norm": 0.7312015891075134, "learning_rate": 1.0603069273246946e-05, "loss": 2.7257, "step": 188070 }, { "epoch": 0.0014848, "grad_norm": 0.7307072281837463, "learning_rate": 1.0602265458519459e-05, "loss": 2.5496, "step": 188080 }, { "epoch": 0.0015104, "grad_norm": 0.7571739554405212, "learning_rate": 1.0601461639886437e-05, "loss": 2.5247, "step": 188090 }, { "epoch": 0.001536, "grad_norm": 0.8742241263389587, "learning_rate": 1.0600657817353096e-05, "loss": 2.3104, "step": 188100 }, { "epoch": 0.0015616, "grad_norm": 0.981002926826477, "learning_rate": 1.0599853990924651e-05, "loss": 2.5662, "step": 188110 }, { "epoch": 0.0015872, "grad_norm": 0.8420567512512207, "learning_rate": 1.0599050160606312e-05, "loss": 2.8155, "step": 188120 }, { "epoch": 0.0016128, "grad_norm": 0.8929325342178345, "learning_rate": 1.059824632640329e-05, "loss": 2.7185, "step": 188130 }, { "epoch": 0.0016384, "grad_norm": 0.8617829084396362, "learning_rate": 1.0597442488320802e-05, "loss": 2.6395, "step": 188140 }, { "epoch": 0.001664, "grad_norm": 0.902684211730957, "learning_rate": 1.0596638646364057e-05, "loss": 2.8497, "step": 188150 }, { "epoch": 0.0016896, "grad_norm": 0.7619622945785522, "learning_rate": 1.0595834800538268e-05, "loss": 2.9059, "step": 188160 }, { "epoch": 0.0017152, "grad_norm": 0.895946741104126, "learning_rate": 1.0595030950848647e-05, "loss": 2.7937, "step": 188170 }, { "epoch": 0.0017408, "grad_norm": 0.8179509043693542, "learning_rate": 1.0594227097300414e-05, "loss": 2.7982, "step": 188180 }, { "epoch": 0.0017664, "grad_norm": 0.7866499423980713, "learning_rate": 1.0593423239898772e-05, "loss": 2.902, "step": 188190 }, { "epoch": 0.001792, "grad_norm": 0.6988380551338196, "learning_rate": 1.0592619378648941e-05, "loss": 2.7573, "step": 188200 }, { "epoch": 2.56e-05, "grad_norm": 0.8504332900047302, "learning_rate": 1.0591815513556127e-05, "loss": 2.0384, "step": 188210 }, { "epoch": 5.12e-05, "grad_norm": 1.0253136157989502, "learning_rate": 1.0591011644625548e-05, "loss": 2.6348, "step": 188220 }, { "epoch": 7.68e-05, "grad_norm": 0.9299499988555908, "learning_rate": 1.0590207771862417e-05, "loss": 2.8427, "step": 188230 }, { "epoch": 0.0001024, "grad_norm": 0.737346887588501, "learning_rate": 1.0589403895271945e-05, "loss": 2.7124, "step": 188240 }, { "epoch": 0.000128, "grad_norm": 0.7771779298782349, "learning_rate": 1.0588600014859345e-05, "loss": 2.7067, "step": 188250 }, { "epoch": 0.0001536, "grad_norm": 0.7548636794090271, "learning_rate": 1.058779613062983e-05, "loss": 2.7694, "step": 188260 }, { "epoch": 0.0001792, "grad_norm": 0.7351616024971008, "learning_rate": 1.0586992242588613e-05, "loss": 2.7188, "step": 188270 }, { "epoch": 0.0002048, "grad_norm": 0.8756833672523499, "learning_rate": 1.0586188350740911e-05, "loss": 2.7065, "step": 188280 }, { "epoch": 0.0002304, "grad_norm": 0.7794528007507324, "learning_rate": 1.0585384455091929e-05, "loss": 2.7674, "step": 188290 }, { "epoch": 0.000256, "grad_norm": 0.8020036220550537, "learning_rate": 1.0584580555646883e-05, "loss": 2.9163, "step": 188300 }, { "epoch": 0.0002816, "grad_norm": 0.9302751421928406, "learning_rate": 1.0583776652410993e-05, "loss": 2.6267, "step": 188310 }, { "epoch": 0.0003072, "grad_norm": 0.7276760935783386, "learning_rate": 1.0582972745389463e-05, "loss": 2.7287, "step": 188320 }, { "epoch": 0.0003328, "grad_norm": 0.8810756802558899, "learning_rate": 1.058216883458751e-05, "loss": 2.5885, "step": 188330 }, { "epoch": 0.0003584, "grad_norm": 0.7936598062515259, "learning_rate": 1.0581364920010348e-05, "loss": 2.7431, "step": 188340 }, { "epoch": 0.000384, "grad_norm": 0.8410770893096924, "learning_rate": 1.058056100166319e-05, "loss": 2.7242, "step": 188350 }, { "epoch": 0.0004096, "grad_norm": 0.7150213122367859, "learning_rate": 1.0579757079551245e-05, "loss": 2.6649, "step": 188360 }, { "epoch": 0.0004352, "grad_norm": 0.8545235395431519, "learning_rate": 1.057895315367973e-05, "loss": 2.9105, "step": 188370 }, { "epoch": 0.0004608, "grad_norm": 0.7575117349624634, "learning_rate": 1.0578149224053862e-05, "loss": 2.7274, "step": 188380 }, { "epoch": 0.0004864, "grad_norm": 0.9124109148979187, "learning_rate": 1.0577345290678842e-05, "loss": 2.7062, "step": 188390 }, { "epoch": 0.000512, "grad_norm": 0.9181995987892151, "learning_rate": 1.0576541353559897e-05, "loss": 2.8337, "step": 188400 }, { "epoch": 0.0005376, "grad_norm": 0.7238244414329529, "learning_rate": 1.0575737412702234e-05, "loss": 2.5673, "step": 188410 }, { "epoch": 0.0005632, "grad_norm": 0.7360723614692688, "learning_rate": 1.0574933468111066e-05, "loss": 2.9386, "step": 188420 }, { "epoch": 0.0005888, "grad_norm": 0.8707374334335327, "learning_rate": 1.0574129519791607e-05, "loss": 2.8068, "step": 188430 }, { "epoch": 0.0006144, "grad_norm": 0.7343514561653137, "learning_rate": 1.0573325567749074e-05, "loss": 2.5002, "step": 188440 }, { "epoch": 0.00064, "grad_norm": 0.7361061573028564, "learning_rate": 1.0572521611988674e-05, "loss": 2.7036, "step": 188450 }, { "epoch": 0.0006656, "grad_norm": 0.8126599192619324, "learning_rate": 1.0571717652515621e-05, "loss": 2.795, "step": 188460 }, { "epoch": 0.0006912, "grad_norm": 0.8512844443321228, "learning_rate": 1.0570913689335134e-05, "loss": 3.2378, "step": 188470 }, { "epoch": 0.0007168, "grad_norm": 0.7252568006515503, "learning_rate": 1.0570109722452422e-05, "loss": 2.642, "step": 188480 }, { "epoch": 0.0007424, "grad_norm": 0.8630124926567078, "learning_rate": 1.05693057518727e-05, "loss": 2.7853, "step": 188490 }, { "epoch": 0.000768, "grad_norm": 0.7998701333999634, "learning_rate": 1.0568501777601184e-05, "loss": 2.764, "step": 188500 }, { "epoch": 0.0007936, "grad_norm": 1.0936791896820068, "learning_rate": 1.0567697799643084e-05, "loss": 2.6366, "step": 188510 }, { "epoch": 0.0008192, "grad_norm": 0.789006233215332, "learning_rate": 1.0566893818003613e-05, "loss": 2.8404, "step": 188520 }, { "epoch": 0.0008448, "grad_norm": 0.9009081721305847, "learning_rate": 1.0566089832687989e-05, "loss": 2.7743, "step": 188530 }, { "epoch": 0.0008704, "grad_norm": 0.8036825060844421, "learning_rate": 1.0565285843701422e-05, "loss": 2.6319, "step": 188540 }, { "epoch": 0.000896, "grad_norm": 0.8381128907203674, "learning_rate": 1.0564481851049127e-05, "loss": 2.6833, "step": 188550 }, { "epoch": 0.0009216, "grad_norm": 0.7227985858917236, "learning_rate": 1.0563677854736311e-05, "loss": 2.6817, "step": 188560 }, { "epoch": 0.0009472, "grad_norm": 0.8023351430892944, "learning_rate": 1.0562873854768202e-05, "loss": 3.4485, "step": 188570 }, { "epoch": 0.0009728, "grad_norm": 0.7190454006195068, "learning_rate": 1.0562069851150004e-05, "loss": 2.5513, "step": 188580 }, { "epoch": 0.0009984, "grad_norm": 0.740403950214386, "learning_rate": 1.056126584388693e-05, "loss": 2.6442, "step": 188590 }, { "epoch": 0.001024, "grad_norm": 0.7292060852050781, "learning_rate": 1.0560461832984198e-05, "loss": 2.7309, "step": 188600 }, { "epoch": 0.0010496, "grad_norm": 1.1678462028503418, "learning_rate": 1.055965781844702e-05, "loss": 2.9576, "step": 188610 }, { "epoch": 0.0010752, "grad_norm": 0.7857562899589539, "learning_rate": 1.055885380028061e-05, "loss": 2.8439, "step": 188620 }, { "epoch": 0.0011008, "grad_norm": 0.9116411805152893, "learning_rate": 1.0558049778490177e-05, "loss": 2.8308, "step": 188630 }, { "epoch": 0.0011264, "grad_norm": 0.9791483283042908, "learning_rate": 1.0557245753080946e-05, "loss": 2.8859, "step": 188640 }, { "epoch": 0.001152, "grad_norm": 0.8842620253562927, "learning_rate": 1.055644172405812e-05, "loss": 2.7733, "step": 188650 }, { "epoch": 0.0011776, "grad_norm": 0.978184700012207, "learning_rate": 1.0555637691426917e-05, "loss": 2.922, "step": 188660 }, { "epoch": 0.0012032, "grad_norm": 0.7126163840293884, "learning_rate": 1.0554833655192556e-05, "loss": 2.8512, "step": 188670 }, { "epoch": 0.0012288, "grad_norm": 0.7926096320152283, "learning_rate": 1.0554029615360241e-05, "loss": 3.0465, "step": 188680 }, { "epoch": 0.0012544, "grad_norm": 0.7588838934898376, "learning_rate": 1.0553225571935195e-05, "loss": 3.1245, "step": 188690 }, { "epoch": 0.00128, "grad_norm": 0.7646650671958923, "learning_rate": 1.0552421524922624e-05, "loss": 2.7084, "step": 188700 }, { "epoch": 0.0013056, "grad_norm": 0.7779067158699036, "learning_rate": 1.0551617474327751e-05, "loss": 2.6149, "step": 188710 }, { "epoch": 0.0013312, "grad_norm": 0.9015432596206665, "learning_rate": 1.0550813420155784e-05, "loss": 2.7111, "step": 188720 }, { "epoch": 0.0013568, "grad_norm": 0.9138110280036926, "learning_rate": 1.0550009362411937e-05, "loss": 2.9209, "step": 188730 }, { "epoch": 0.0013824, "grad_norm": 0.8868615627288818, "learning_rate": 1.0549205301101426e-05, "loss": 2.918, "step": 188740 }, { "epoch": 0.001408, "grad_norm": 0.690216600894928, "learning_rate": 1.0548401236229464e-05, "loss": 2.9229, "step": 188750 }, { "epoch": 0.0014336, "grad_norm": 0.8210095763206482, "learning_rate": 1.0547597167801264e-05, "loss": 2.8825, "step": 188760 }, { "epoch": 0.0014592, "grad_norm": 0.7659406661987305, "learning_rate": 1.0546793095822048e-05, "loss": 2.5874, "step": 188770 }, { "epoch": 0.0014848, "grad_norm": 0.7149298191070557, "learning_rate": 1.0545989020297019e-05, "loss": 2.5174, "step": 188780 }, { "epoch": 0.0015104, "grad_norm": 0.7579086422920227, "learning_rate": 1.0545184941231396e-05, "loss": 2.599, "step": 188790 }, { "epoch": 0.001536, "grad_norm": 0.7414752244949341, "learning_rate": 1.0544380858630396e-05, "loss": 2.4827, "step": 188800 }, { "epoch": 0.0015616, "grad_norm": 1.1027023792266846, "learning_rate": 1.0543576772499231e-05, "loss": 3.0616, "step": 188810 }, { "epoch": 0.0015872, "grad_norm": 0.7604153752326965, "learning_rate": 1.0542772682843114e-05, "loss": 2.7332, "step": 188820 }, { "epoch": 0.0016128, "grad_norm": 0.8013967275619507, "learning_rate": 1.0541968589667257e-05, "loss": 2.7222, "step": 188830 }, { "epoch": 0.0016384, "grad_norm": 0.7754907011985779, "learning_rate": 1.0541164492976885e-05, "loss": 2.8928, "step": 188840 }, { "epoch": 0.001664, "grad_norm": 0.8714032769203186, "learning_rate": 1.05403603927772e-05, "loss": 2.8983, "step": 188850 }, { "epoch": 0.0016896, "grad_norm": 0.8042787313461304, "learning_rate": 1.0539556289073423e-05, "loss": 3.0484, "step": 188860 }, { "epoch": 0.0017152, "grad_norm": 0.8288938403129578, "learning_rate": 1.0538752181870768e-05, "loss": 2.8224, "step": 188870 }, { "epoch": 0.0017408, "grad_norm": 0.8702886700630188, "learning_rate": 1.0537948071174446e-05, "loss": 2.7803, "step": 188880 }, { "epoch": 0.0017664, "grad_norm": 0.759424090385437, "learning_rate": 1.0537143956989674e-05, "loss": 2.936, "step": 188890 }, { "epoch": 0.001792, "grad_norm": 0.807808518409729, "learning_rate": 1.0536339839321667e-05, "loss": 2.624, "step": 188900 }, { "epoch": 0.0018176, "grad_norm": 0.7381454706192017, "learning_rate": 1.0535535718175638e-05, "loss": 2.6039, "step": 188910 }, { "epoch": 0.0018432, "grad_norm": 0.8154658079147339, "learning_rate": 1.0534731593556802e-05, "loss": 2.6752, "step": 188920 }, { "epoch": 0.0018688, "grad_norm": 0.7746941447257996, "learning_rate": 1.0533927465470374e-05, "loss": 2.6094, "step": 188930 }, { "epoch": 0.0018944, "grad_norm": 0.9155193567276001, "learning_rate": 1.053312333392157e-05, "loss": 3.03, "step": 188940 }, { "epoch": 0.00192, "grad_norm": 0.7779428958892822, "learning_rate": 1.0532319198915602e-05, "loss": 2.7271, "step": 188950 }, { "epoch": 0.0019456, "grad_norm": 0.811156153678894, "learning_rate": 1.0531515060457685e-05, "loss": 2.9093, "step": 188960 }, { "epoch": 0.0019712, "grad_norm": 0.8155035376548767, "learning_rate": 1.0530710918553036e-05, "loss": 2.8901, "step": 188970 }, { "epoch": 0.0019968, "grad_norm": 0.8135418891906738, "learning_rate": 1.0529906773206865e-05, "loss": 2.6267, "step": 188980 }, { "epoch": 0.0020224, "grad_norm": 0.748928964138031, "learning_rate": 1.052910262442439e-05, "loss": 3.04, "step": 188990 }, { "epoch": 0.002048, "grad_norm": 0.699117124080658, "learning_rate": 1.0528298472210826e-05, "loss": 2.7516, "step": 189000 }, { "epoch": 0.0020736, "grad_norm": 0.7166158556938171, "learning_rate": 1.0527494316571385e-05, "loss": 2.5092, "step": 189010 }, { "epoch": 0.0020992, "grad_norm": 0.9013195037841797, "learning_rate": 1.0526690157511284e-05, "loss": 2.9495, "step": 189020 }, { "epoch": 0.0021248, "grad_norm": 0.7811973690986633, "learning_rate": 1.0525885995035738e-05, "loss": 2.7939, "step": 189030 }, { "epoch": 0.0021504, "grad_norm": 0.8013681769371033, "learning_rate": 1.0525081829149964e-05, "loss": 3.0427, "step": 189040 }, { "epoch": 0.002176, "grad_norm": 0.990602433681488, "learning_rate": 1.052427765985917e-05, "loss": 3.0119, "step": 189050 }, { "epoch": 0.0022016, "grad_norm": 0.7531378865242004, "learning_rate": 1.0523473487168573e-05, "loss": 2.9255, "step": 189060 }, { "epoch": 0.0022272, "grad_norm": 0.8062220811843872, "learning_rate": 1.0522669311083396e-05, "loss": 3.0267, "step": 189070 }, { "epoch": 0.0022528, "grad_norm": 1.2529569864273071, "learning_rate": 1.0521865131608845e-05, "loss": 2.6777, "step": 189080 }, { "epoch": 0.0022784, "grad_norm": 0.7924194931983948, "learning_rate": 1.052106094875013e-05, "loss": 2.805, "step": 189090 }, { "epoch": 0.002304, "grad_norm": 0.7860348224639893, "learning_rate": 1.0520256762512482e-05, "loss": 2.5684, "step": 189100 }, { "epoch": 0.0023296, "grad_norm": 0.7777368426322937, "learning_rate": 1.0519452572901105e-05, "loss": 2.717, "step": 189110 }, { "epoch": 0.0023552, "grad_norm": 0.8329401016235352, "learning_rate": 1.0518648379921215e-05, "loss": 2.9204, "step": 189120 }, { "epoch": 0.0023808, "grad_norm": 0.7640992403030396, "learning_rate": 1.051784418357803e-05, "loss": 2.9061, "step": 189130 }, { "epoch": 0.0024064, "grad_norm": 0.7628771662712097, "learning_rate": 1.0517039983876762e-05, "loss": 2.8453, "step": 189140 }, { "epoch": 0.002432, "grad_norm": 0.889092206954956, "learning_rate": 1.0516235780822625e-05, "loss": 2.693, "step": 189150 }, { "epoch": 0.0024576, "grad_norm": 0.7110810279846191, "learning_rate": 1.0515431574420839e-05, "loss": 2.8257, "step": 189160 }, { "epoch": 0.0024832, "grad_norm": 0.7363725900650024, "learning_rate": 1.0514627364676615e-05, "loss": 2.8242, "step": 189170 }, { "epoch": 0.0025088, "grad_norm": 0.7393179535865784, "learning_rate": 1.0513823151595168e-05, "loss": 3.0413, "step": 189180 }, { "epoch": 0.0025344, "grad_norm": 0.8700782656669617, "learning_rate": 1.0513018935181716e-05, "loss": 2.8943, "step": 189190 }, { "epoch": 0.00256, "grad_norm": 0.7429696917533875, "learning_rate": 1.0512214715441474e-05, "loss": 2.8975, "step": 189200 }, { "epoch": 0.0025856, "grad_norm": 0.8571328520774841, "learning_rate": 1.0511410492379654e-05, "loss": 2.8704, "step": 189210 }, { "epoch": 0.0026112, "grad_norm": 1.0193085670471191, "learning_rate": 1.0510606266001473e-05, "loss": 3.1018, "step": 189220 }, { "epoch": 0.0026368, "grad_norm": 0.7833830714225769, "learning_rate": 1.0509802036312146e-05, "loss": 2.7528, "step": 189230 }, { "epoch": 0.0026624, "grad_norm": 0.7389554381370544, "learning_rate": 1.050899780331689e-05, "loss": 2.9194, "step": 189240 }, { "epoch": 0.002688, "grad_norm": 0.8351564407348633, "learning_rate": 1.0508193567020917e-05, "loss": 2.6498, "step": 189250 }, { "epoch": 0.0027136, "grad_norm": 0.7319939136505127, "learning_rate": 1.0507389327429443e-05, "loss": 2.9383, "step": 189260 }, { "epoch": 0.0027392, "grad_norm": 0.7768262028694153, "learning_rate": 1.0506585084547684e-05, "loss": 2.934, "step": 189270 }, { "epoch": 0.0027648, "grad_norm": 0.8445101380348206, "learning_rate": 1.0505780838380857e-05, "loss": 2.904, "step": 189280 }, { "epoch": 0.0027904, "grad_norm": 0.7581013441085815, "learning_rate": 1.0504976588934173e-05, "loss": 2.9894, "step": 189290 }, { "epoch": 0.002816, "grad_norm": 0.8196311593055725, "learning_rate": 1.0504172336212854e-05, "loss": 2.8476, "step": 189300 }, { "epoch": 0.0028416, "grad_norm": 0.7658888101577759, "learning_rate": 1.050336808022211e-05, "loss": 2.7911, "step": 189310 }, { "epoch": 0.0028672, "grad_norm": 0.7992104291915894, "learning_rate": 1.0502563820967155e-05, "loss": 3.0662, "step": 189320 }, { "epoch": 0.0028928, "grad_norm": 1.1578377485275269, "learning_rate": 1.0501759558453211e-05, "loss": 2.9856, "step": 189330 }, { "epoch": 0.0029184, "grad_norm": 0.8483161926269531, "learning_rate": 1.050095529268549e-05, "loss": 2.5693, "step": 189340 }, { "epoch": 0.002944, "grad_norm": 0.7191236615180969, "learning_rate": 1.05001510236692e-05, "loss": 2.7842, "step": 189350 }, { "epoch": 0.0029696, "grad_norm": 0.7303234338760376, "learning_rate": 1.0499346751409572e-05, "loss": 2.903, "step": 189360 }, { "epoch": 0.0029952, "grad_norm": 0.8992199301719666, "learning_rate": 1.0498542475911808e-05, "loss": 2.8809, "step": 189370 }, { "epoch": 0.0030208, "grad_norm": 0.7914143204689026, "learning_rate": 1.0497738197181129e-05, "loss": 3.0503, "step": 189380 }, { "epoch": 0.0030464, "grad_norm": 0.8083113431930542, "learning_rate": 1.0496933915222752e-05, "loss": 2.9193, "step": 189390 }, { "epoch": 0.003072, "grad_norm": 0.9626654386520386, "learning_rate": 1.0496129630041892e-05, "loss": 3.1912, "step": 189400 }, { "epoch": 0.0030976, "grad_norm": 0.7176828980445862, "learning_rate": 1.0495325341643759e-05, "loss": 3.1159, "step": 189410 }, { "epoch": 0.0031232, "grad_norm": 0.8457401394844055, "learning_rate": 1.0494521050033576e-05, "loss": 3.0774, "step": 189420 }, { "epoch": 0.0031488, "grad_norm": 0.7657620906829834, "learning_rate": 1.0493716755216552e-05, "loss": 2.9181, "step": 189430 }, { "epoch": 0.0031744, "grad_norm": 0.7950317859649658, "learning_rate": 1.0492912457197907e-05, "loss": 2.8464, "step": 189440 }, { "epoch": 0.0032, "grad_norm": 0.826543390750885, "learning_rate": 1.0492108155982856e-05, "loss": 2.7123, "step": 189450 }, { "epoch": 0.0032256, "grad_norm": 0.8338049054145813, "learning_rate": 1.0491303851576617e-05, "loss": 3.1151, "step": 189460 }, { "epoch": 0.0032512, "grad_norm": 0.8102612495422363, "learning_rate": 1.0490499543984403e-05, "loss": 2.9782, "step": 189470 }, { "epoch": 0.0032768, "grad_norm": 0.6567476391792297, "learning_rate": 1.0489695233211428e-05, "loss": 2.7927, "step": 189480 }, { "epoch": 0.0033024, "grad_norm": 0.8506841063499451, "learning_rate": 1.0488890919262909e-05, "loss": 2.9197, "step": 189490 }, { "epoch": 0.003328, "grad_norm": 0.7086018323898315, "learning_rate": 1.0488086602144066e-05, "loss": 2.7754, "step": 189500 }, { "epoch": 0.0033536, "grad_norm": 0.7546857595443726, "learning_rate": 1.0487282281860108e-05, "loss": 2.9789, "step": 189510 }, { "epoch": 0.0033792, "grad_norm": 0.7823271155357361, "learning_rate": 1.048647795841625e-05, "loss": 3.0402, "step": 189520 }, { "epoch": 0.0034048, "grad_norm": 0.8445858359336853, "learning_rate": 1.0485673631817716e-05, "loss": 2.7734, "step": 189530 }, { "epoch": 0.0034304, "grad_norm": 0.8596416115760803, "learning_rate": 1.0484869302069718e-05, "loss": 2.7972, "step": 189540 }, { "epoch": 0.003456, "grad_norm": 0.8386886119842529, "learning_rate": 1.048406496917747e-05, "loss": 2.9314, "step": 189550 }, { "epoch": 0.0034816, "grad_norm": 0.8738669157028198, "learning_rate": 1.0483260633146194e-05, "loss": 2.9175, "step": 189560 }, { "epoch": 0.0035072, "grad_norm": 0.7846934199333191, "learning_rate": 1.0482456293981095e-05, "loss": 2.8755, "step": 189570 }, { "epoch": 0.0035328, "grad_norm": 0.7859766483306885, "learning_rate": 1.0481651951687398e-05, "loss": 2.9756, "step": 189580 }, { "epoch": 0.0035584, "grad_norm": 1.2932120561599731, "learning_rate": 1.0480847606270315e-05, "loss": 2.7904, "step": 189590 }, { "epoch": 0.003584, "grad_norm": 1.1395539045333862, "learning_rate": 1.0480043257735067e-05, "loss": 2.9314, "step": 189600 }, { "epoch": 0.0036096, "grad_norm": 1.0365322828292847, "learning_rate": 1.047923890608686e-05, "loss": 3.0059, "step": 189610 }, { "epoch": 0.0036352, "grad_norm": 0.8674554228782654, "learning_rate": 1.0478434551330919e-05, "loss": 2.8832, "step": 189620 }, { "epoch": 0.0036608, "grad_norm": 0.7443742752075195, "learning_rate": 1.0477630193472457e-05, "loss": 2.7889, "step": 189630 }, { "epoch": 0.0036864, "grad_norm": 0.8551311492919922, "learning_rate": 1.047682583251669e-05, "loss": 2.8994, "step": 189640 }, { "epoch": 0.003712, "grad_norm": 0.7306486368179321, "learning_rate": 1.0476021468468834e-05, "loss": 2.9839, "step": 189650 }, { "epoch": 0.0037376, "grad_norm": 0.9747829437255859, "learning_rate": 1.0475217101334108e-05, "loss": 2.927, "step": 189660 }, { "epoch": 0.0037632, "grad_norm": 0.9265177845954895, "learning_rate": 1.0474412731117722e-05, "loss": 2.9213, "step": 189670 }, { "epoch": 0.0037888, "grad_norm": 0.7273955345153809, "learning_rate": 1.0473608357824895e-05, "loss": 3.0927, "step": 189680 }, { "epoch": 0.0038144, "grad_norm": 0.8076822757720947, "learning_rate": 1.0472803981460846e-05, "loss": 3.0547, "step": 189690 }, { "epoch": 0.00384, "grad_norm": 1.0815809965133667, "learning_rate": 1.0471999602030786e-05, "loss": 2.7938, "step": 189700 }, { "epoch": 0.0038656, "grad_norm": 0.7622167468070984, "learning_rate": 1.0471195219539936e-05, "loss": 3.0436, "step": 189710 }, { "epoch": 0.0038912, "grad_norm": 0.8173257112503052, "learning_rate": 1.0470390833993512e-05, "loss": 2.8644, "step": 189720 }, { "epoch": 0.0039168, "grad_norm": 0.8333418369293213, "learning_rate": 1.0469586445396726e-05, "loss": 3.0351, "step": 189730 }, { "epoch": 0.0039424, "grad_norm": 0.8853899240493774, "learning_rate": 1.0468782053754797e-05, "loss": 2.8909, "step": 189740 }, { "epoch": 0.003968, "grad_norm": 0.940217912197113, "learning_rate": 1.046797765907294e-05, "loss": 3.0108, "step": 189750 }, { "epoch": 0.0039936, "grad_norm": 0.8340900540351868, "learning_rate": 1.0467173261356378e-05, "loss": 2.9861, "step": 189760 }, { "epoch": 0.0040192, "grad_norm": 0.7147396802902222, "learning_rate": 1.0466368860610316e-05, "loss": 2.855, "step": 189770 }, { "epoch": 0.0040448, "grad_norm": 0.9059799909591675, "learning_rate": 1.0465564456839975e-05, "loss": 2.9579, "step": 189780 }, { "epoch": 0.0040704, "grad_norm": 1.221831202507019, "learning_rate": 1.0464760050050575e-05, "loss": 3.0865, "step": 189790 }, { "epoch": 0.004096, "grad_norm": 0.7207812070846558, "learning_rate": 1.046395564024733e-05, "loss": 2.8965, "step": 189800 }, { "epoch": 0.0041216, "grad_norm": 0.8472402095794678, "learning_rate": 1.0463151227435452e-05, "loss": 2.9004, "step": 189810 }, { "epoch": 0.0041472, "grad_norm": 0.7997706532478333, "learning_rate": 1.0462346811620166e-05, "loss": 3.0846, "step": 189820 }, { "epoch": 0.0041728, "grad_norm": 0.7982807159423828, "learning_rate": 1.0461542392806682e-05, "loss": 3.0509, "step": 189830 }, { "epoch": 0.0041984, "grad_norm": 0.7210063338279724, "learning_rate": 1.046073797100022e-05, "loss": 2.8048, "step": 189840 }, { "epoch": 0.004224, "grad_norm": 0.9828871488571167, "learning_rate": 1.045993354620599e-05, "loss": 2.7943, "step": 189850 }, { "epoch": 0.0042496, "grad_norm": 0.8066829442977905, "learning_rate": 1.0459129118429218e-05, "loss": 2.9168, "step": 189860 }, { "epoch": 0.0042752, "grad_norm": 0.8879944682121277, "learning_rate": 1.0458324687675111e-05, "loss": 3.0271, "step": 189870 }, { "epoch": 0.0043008, "grad_norm": 0.9401640295982361, "learning_rate": 1.0457520253948895e-05, "loss": 2.9011, "step": 189880 }, { "epoch": 0.0043264, "grad_norm": 0.754267156124115, "learning_rate": 1.045671581725578e-05, "loss": 2.9207, "step": 189890 }, { "epoch": 0.004352, "grad_norm": 0.6903390288352966, "learning_rate": 1.0455911377600985e-05, "loss": 2.9095, "step": 189900 }, { "epoch": 0.0043776, "grad_norm": 0.7596873044967651, "learning_rate": 1.0455106934989725e-05, "loss": 2.8802, "step": 189910 }, { "epoch": 0.0044032, "grad_norm": 0.7925681471824646, "learning_rate": 1.0454302489427217e-05, "loss": 3.0202, "step": 189920 }, { "epoch": 0.0044288, "grad_norm": 0.6592907309532166, "learning_rate": 1.045349804091868e-05, "loss": 3.0647, "step": 189930 }, { "epoch": 0.0044544, "grad_norm": 0.8072383999824524, "learning_rate": 1.0452693589469327e-05, "loss": 3.0245, "step": 189940 }, { "epoch": 0.00448, "grad_norm": 0.7949973344802856, "learning_rate": 1.0451889135084376e-05, "loss": 2.969, "step": 189950 }, { "epoch": 0.0045056, "grad_norm": 0.7666147947311401, "learning_rate": 1.0451084677769046e-05, "loss": 2.8017, "step": 189960 }, { "epoch": 0.0045312, "grad_norm": 0.7985819578170776, "learning_rate": 1.0450280217528548e-05, "loss": 3.0898, "step": 189970 }, { "epoch": 0.0045568, "grad_norm": 0.8286681175231934, "learning_rate": 1.0449475754368106e-05, "loss": 2.8355, "step": 189980 }, { "epoch": 0.0045824, "grad_norm": 0.7770845890045166, "learning_rate": 1.0448671288292933e-05, "loss": 2.9807, "step": 189990 }, { "epoch": 0.004608, "grad_norm": 0.8645066022872925, "learning_rate": 1.0447866819308245e-05, "loss": 2.9522, "step": 190000 }, { "epoch": 0.0046336, "grad_norm": 1.0242985486984253, "learning_rate": 1.0447062347419259e-05, "loss": 2.8345, "step": 190010 }, { "epoch": 0.0046592, "grad_norm": 0.9088624715805054, "learning_rate": 1.0446257872631194e-05, "loss": 2.9084, "step": 190020 }, { "epoch": 0.0046848, "grad_norm": 0.7357869744300842, "learning_rate": 1.0445453394949266e-05, "loss": 3.229, "step": 190030 }, { "epoch": 0.0047104, "grad_norm": 0.845777690410614, "learning_rate": 1.0444648914378685e-05, "loss": 3.0231, "step": 190040 }, { "epoch": 0.004736, "grad_norm": 0.736206591129303, "learning_rate": 1.0443844430924682e-05, "loss": 2.9091, "step": 190050 }, { "epoch": 0.0047616, "grad_norm": 0.756147027015686, "learning_rate": 1.0443039944592462e-05, "loss": 2.8704, "step": 190060 }, { "epoch": 0.0047872, "grad_norm": 1.0667884349822998, "learning_rate": 1.0442235455387246e-05, "loss": 3.0625, "step": 190070 }, { "epoch": 0.0048128, "grad_norm": 0.7797421813011169, "learning_rate": 1.044143096331425e-05, "loss": 2.9664, "step": 190080 }, { "epoch": 0.0048384, "grad_norm": 0.7489699721336365, "learning_rate": 1.0440626468378694e-05, "loss": 2.7576, "step": 190090 }, { "epoch": 0.004864, "grad_norm": 0.7464562654495239, "learning_rate": 1.0439821970585791e-05, "loss": 2.9968, "step": 190100 }, { "epoch": 0.0048896, "grad_norm": 0.9665318727493286, "learning_rate": 1.043901746994076e-05, "loss": 2.784, "step": 190110 }, { "epoch": 0.0049152, "grad_norm": 0.8683475852012634, "learning_rate": 1.0438212966448817e-05, "loss": 3.0247, "step": 190120 }, { "epoch": 0.0049408, "grad_norm": 0.9011716842651367, "learning_rate": 1.0437408460115178e-05, "loss": 2.7771, "step": 190130 }, { "epoch": 0.0049664, "grad_norm": 0.694195568561554, "learning_rate": 1.0436603950945061e-05, "loss": 2.7123, "step": 190140 }, { "epoch": 0.004992, "grad_norm": 0.8663414120674133, "learning_rate": 1.0435799438943686e-05, "loss": 2.7844, "step": 190150 }, { "epoch": 0.0050176, "grad_norm": 0.7868435978889465, "learning_rate": 1.0434994924116267e-05, "loss": 2.9105, "step": 190160 }, { "epoch": 0.0050432, "grad_norm": 0.750211238861084, "learning_rate": 1.0434190406468018e-05, "loss": 2.8651, "step": 190170 }, { "epoch": 0.0050688, "grad_norm": 0.8870049118995667, "learning_rate": 1.0433385886004164e-05, "loss": 2.7769, "step": 190180 }, { "epoch": 0.0050944, "grad_norm": 0.8456773161888123, "learning_rate": 1.0432581362729917e-05, "loss": 2.8579, "step": 190190 }, { "epoch": 0.00512, "grad_norm": 0.842397153377533, "learning_rate": 1.0431776836650493e-05, "loss": 2.902, "step": 190200 }, { "epoch": 0.0051456, "grad_norm": 0.7538192868232727, "learning_rate": 1.0430972307771113e-05, "loss": 2.8667, "step": 190210 }, { "epoch": 0.0051712, "grad_norm": 0.8301517963409424, "learning_rate": 1.043016777609699e-05, "loss": 2.7865, "step": 190220 }, { "epoch": 0.0051968, "grad_norm": 0.733814001083374, "learning_rate": 1.0429363241633344e-05, "loss": 2.6982, "step": 190230 }, { "epoch": 0.0052224, "grad_norm": 0.8134002089500427, "learning_rate": 1.042855870438539e-05, "loss": 2.8878, "step": 190240 }, { "epoch": 0.005248, "grad_norm": 0.8883689641952515, "learning_rate": 1.042775416435835e-05, "loss": 3.0855, "step": 190250 }, { "epoch": 0.0052736, "grad_norm": 0.7733215689659119, "learning_rate": 1.0426949621557436e-05, "loss": 3.0049, "step": 190260 }, { "epoch": 0.0052992, "grad_norm": 0.7977262139320374, "learning_rate": 1.0426145075987868e-05, "loss": 2.9591, "step": 190270 }, { "epoch": 0.0053248, "grad_norm": 0.8231276273727417, "learning_rate": 1.042534052765486e-05, "loss": 2.8857, "step": 190280 }, { "epoch": 0.0053504, "grad_norm": 0.8199840188026428, "learning_rate": 1.0424535976563637e-05, "loss": 2.947, "step": 190290 }, { "epoch": 0.005376, "grad_norm": 0.7752149105072021, "learning_rate": 1.0423731422719405e-05, "loss": 3.0179, "step": 190300 }, { "epoch": 0.0054016, "grad_norm": 0.6897476315498352, "learning_rate": 1.042292686612739e-05, "loss": 2.9701, "step": 190310 }, { "epoch": 0.0054272, "grad_norm": 0.8475947380065918, "learning_rate": 1.0422122306792808e-05, "loss": 2.9483, "step": 190320 }, { "epoch": 0.0054528, "grad_norm": 0.7335816621780396, "learning_rate": 1.0421317744720874e-05, "loss": 2.7768, "step": 190330 }, { "epoch": 0.0054784, "grad_norm": 0.7619728446006775, "learning_rate": 1.0420513179916807e-05, "loss": 2.9171, "step": 190340 }, { "epoch": 0.005504, "grad_norm": 0.7777618765830994, "learning_rate": 1.0419708612385823e-05, "loss": 3.0763, "step": 190350 }, { "epoch": 0.0055296, "grad_norm": 0.739362359046936, "learning_rate": 1.041890404213314e-05, "loss": 2.7565, "step": 190360 }, { "epoch": 0.0055552, "grad_norm": 0.8918215036392212, "learning_rate": 1.0418099469163975e-05, "loss": 2.8524, "step": 190370 }, { "epoch": 0.0055808, "grad_norm": 0.801939845085144, "learning_rate": 1.0417294893483548e-05, "loss": 3.007, "step": 190380 }, { "epoch": 0.0056064, "grad_norm": 0.7523275017738342, "learning_rate": 1.0416490315097072e-05, "loss": 2.9391, "step": 190390 }, { "epoch": 0.005632, "grad_norm": 0.7115250825881958, "learning_rate": 1.0415685734009768e-05, "loss": 2.9926, "step": 190400 }, { "epoch": 0.0056576, "grad_norm": 0.7749091386795044, "learning_rate": 1.0414881150226852e-05, "loss": 2.9078, "step": 190410 }, { "epoch": 0.0056832, "grad_norm": 0.8668674230575562, "learning_rate": 1.0414076563753546e-05, "loss": 2.7739, "step": 190420 }, { "epoch": 0.0057088, "grad_norm": 0.7177619934082031, "learning_rate": 1.0413271974595059e-05, "loss": 2.9485, "step": 190430 }, { "epoch": 0.0057344, "grad_norm": 0.8309880495071411, "learning_rate": 1.0412467382756615e-05, "loss": 2.96, "step": 190440 }, { "epoch": 0.00576, "grad_norm": 0.8902155756950378, "learning_rate": 1.0411662788243431e-05, "loss": 3.0351, "step": 190450 }, { "epoch": 0.0057856, "grad_norm": 0.7467007637023926, "learning_rate": 1.041085819106072e-05, "loss": 2.8794, "step": 190460 }, { "epoch": 0.0058112, "grad_norm": 0.710019052028656, "learning_rate": 1.0410053591213705e-05, "loss": 2.7782, "step": 190470 }, { "epoch": 0.0058368, "grad_norm": 0.9821754097938538, "learning_rate": 1.0409248988707598e-05, "loss": 2.8249, "step": 190480 }, { "epoch": 0.0058624, "grad_norm": 0.7382265329360962, "learning_rate": 1.0408444383547622e-05, "loss": 2.7596, "step": 190490 }, { "epoch": 0.005888, "grad_norm": 0.8488202095031738, "learning_rate": 1.0407639775738994e-05, "loss": 2.9774, "step": 190500 }, { "epoch": 0.0059136, "grad_norm": 0.7179195880889893, "learning_rate": 1.040683516528693e-05, "loss": 2.8715, "step": 190510 }, { "epoch": 0.0059392, "grad_norm": 0.7148680686950684, "learning_rate": 1.0406030552196649e-05, "loss": 2.7823, "step": 190520 }, { "epoch": 0.0059648, "grad_norm": 0.7695332765579224, "learning_rate": 1.0405225936473367e-05, "loss": 2.7707, "step": 190530 }, { "epoch": 0.0059904, "grad_norm": 0.8128268122673035, "learning_rate": 1.0404421318122303e-05, "loss": 3.0133, "step": 190540 }, { "epoch": 0.006016, "grad_norm": 0.8027509450912476, "learning_rate": 1.0403616697148673e-05, "loss": 3.2192, "step": 190550 }, { "epoch": 0.0060416, "grad_norm": 0.8451578617095947, "learning_rate": 1.0402812073557697e-05, "loss": 2.8167, "step": 190560 }, { "epoch": 0.0060672, "grad_norm": 0.7613236904144287, "learning_rate": 1.0402007447354587e-05, "loss": 2.8982, "step": 190570 }, { "epoch": 0.0060928, "grad_norm": 0.8084383606910706, "learning_rate": 1.0401202818544573e-05, "loss": 2.8103, "step": 190580 }, { "epoch": 0.0061184, "grad_norm": 0.8142582774162292, "learning_rate": 1.040039818713286e-05, "loss": 2.7513, "step": 190590 }, { "epoch": 0.006144, "grad_norm": 0.7558661699295044, "learning_rate": 1.0399593553124673e-05, "loss": 2.9295, "step": 190600 }, { "epoch": 0.0061696, "grad_norm": 0.6946911215782166, "learning_rate": 1.039878891652523e-05, "loss": 2.7597, "step": 190610 }, { "epoch": 0.0061952, "grad_norm": 0.7483018040657043, "learning_rate": 1.0397984277339745e-05, "loss": 2.8981, "step": 190620 }, { "epoch": 0.0062208, "grad_norm": 0.734413743019104, "learning_rate": 1.0397179635573437e-05, "loss": 2.7919, "step": 190630 }, { "epoch": 0.0062464, "grad_norm": 0.7994030117988586, "learning_rate": 1.0396374991231526e-05, "loss": 2.7166, "step": 190640 }, { "epoch": 0.006272, "grad_norm": 0.8153067827224731, "learning_rate": 1.0395570344319227e-05, "loss": 2.9178, "step": 190650 }, { "epoch": 0.0062976, "grad_norm": 0.7073977589607239, "learning_rate": 1.039476569484176e-05, "loss": 3.1704, "step": 190660 }, { "epoch": 0.0063232, "grad_norm": 0.7030051350593567, "learning_rate": 1.0393961042804343e-05, "loss": 2.9257, "step": 190670 }, { "epoch": 0.0063488, "grad_norm": 0.8489978313446045, "learning_rate": 1.0393156388212195e-05, "loss": 3.0305, "step": 190680 }, { "epoch": 0.0063744, "grad_norm": 0.7071354389190674, "learning_rate": 1.0392351731070528e-05, "loss": 2.8971, "step": 190690 }, { "epoch": 0.0064, "grad_norm": 0.7535512447357178, "learning_rate": 1.039154707138457e-05, "loss": 2.8177, "step": 190700 }, { "epoch": 0.0064256, "grad_norm": 0.8647302389144897, "learning_rate": 1.0390742409159526e-05, "loss": 3.0605, "step": 190710 }, { "epoch": 0.0064512, "grad_norm": 0.7244521379470825, "learning_rate": 1.038993774440063e-05, "loss": 2.9217, "step": 190720 }, { "epoch": 0.0064768, "grad_norm": 0.882881760597229, "learning_rate": 1.0389133077113085e-05, "loss": 2.8343, "step": 190730 }, { "epoch": 0.0065024, "grad_norm": 0.7423853278160095, "learning_rate": 1.0388328407302117e-05, "loss": 2.7844, "step": 190740 }, { "epoch": 0.006528, "grad_norm": 0.8799000978469849, "learning_rate": 1.0387523734972943e-05, "loss": 2.8231, "step": 190750 }, { "epoch": 0.0065536, "grad_norm": 0.7798517346382141, "learning_rate": 1.038671906013078e-05, "loss": 2.6985, "step": 190760 }, { "epoch": 0.0065792, "grad_norm": 0.7044264078140259, "learning_rate": 1.0385914382780846e-05, "loss": 2.8204, "step": 190770 }, { "epoch": 0.0066048, "grad_norm": 0.7550447583198547, "learning_rate": 1.0385109702928361e-05, "loss": 2.7935, "step": 190780 }, { "epoch": 0.0066304, "grad_norm": 0.8558501601219177, "learning_rate": 1.0384305020578542e-05, "loss": 2.8867, "step": 190790 }, { "epoch": 0.006656, "grad_norm": 0.8778136372566223, "learning_rate": 1.0383500335736608e-05, "loss": 2.9245, "step": 190800 }, { "epoch": 0.0066816, "grad_norm": 0.7541288733482361, "learning_rate": 1.0382695648407773e-05, "loss": 3.3411, "step": 190810 }, { "epoch": 0.0067072, "grad_norm": 0.7926966547966003, "learning_rate": 1.038189095859726e-05, "loss": 3.177, "step": 190820 }, { "epoch": 0.0067328, "grad_norm": 0.9154115915298462, "learning_rate": 1.0381086266310285e-05, "loss": 3.2327, "step": 190830 }, { "epoch": 0.0067584, "grad_norm": 0.7116553783416748, "learning_rate": 1.0380281571552069e-05, "loss": 2.8236, "step": 190840 }, { "epoch": 0.006784, "grad_norm": 0.9086008667945862, "learning_rate": 1.0379476874327827e-05, "loss": 2.7973, "step": 190850 }, { "epoch": 0.0068096, "grad_norm": 0.7507835626602173, "learning_rate": 1.0378672174642779e-05, "loss": 3.0612, "step": 190860 }, { "epoch": 0.0068352, "grad_norm": 0.753149151802063, "learning_rate": 1.0377867472502141e-05, "loss": 2.8922, "step": 190870 }, { "epoch": 0.0068608, "grad_norm": 0.8355460166931152, "learning_rate": 1.0377062767911134e-05, "loss": 2.9097, "step": 190880 }, { "epoch": 0.0068864, "grad_norm": 0.7934004068374634, "learning_rate": 1.0376258060874975e-05, "loss": 2.8822, "step": 190890 }, { "epoch": 0.006912, "grad_norm": 0.7127549052238464, "learning_rate": 1.0375453351398881e-05, "loss": 2.8166, "step": 190900 }, { "epoch": 0.0069376, "grad_norm": 0.8177149295806885, "learning_rate": 1.0374648639488074e-05, "loss": 2.8733, "step": 190910 }, { "epoch": 0.0069632, "grad_norm": 0.9171279668807983, "learning_rate": 1.0373843925147769e-05, "loss": 2.8921, "step": 190920 }, { "epoch": 0.0069888, "grad_norm": 0.6960368156433105, "learning_rate": 1.0373039208383184e-05, "loss": 3.2148, "step": 190930 }, { "epoch": 0.0070144, "grad_norm": 1.9401870965957642, "learning_rate": 1.0372234489199542e-05, "loss": 2.7925, "step": 190940 }, { "epoch": 0.00704, "grad_norm": 0.7998874187469482, "learning_rate": 1.0371429767602056e-05, "loss": 2.9477, "step": 190950 }, { "epoch": 0.0070656, "grad_norm": 0.7733396887779236, "learning_rate": 1.0370625043595948e-05, "loss": 2.9926, "step": 190960 }, { "epoch": 0.0070912, "grad_norm": 0.7734212875366211, "learning_rate": 1.0369820317186434e-05, "loss": 2.7632, "step": 190970 }, { "epoch": 0.0071168, "grad_norm": 0.7947595119476318, "learning_rate": 1.0369015588378738e-05, "loss": 2.9097, "step": 190980 }, { "epoch": 0.0071424, "grad_norm": 0.7470015287399292, "learning_rate": 1.036821085717807e-05, "loss": 2.8894, "step": 190990 }, { "epoch": 0.007168, "grad_norm": 0.7353874444961548, "learning_rate": 1.0367406123589649e-05, "loss": 2.9473, "step": 191000 }, { "epoch": 0.0071936, "grad_norm": 0.7828636765480042, "learning_rate": 1.0366601387618702e-05, "loss": 2.7599, "step": 191010 }, { "epoch": 0.0072192, "grad_norm": 1.0695985555648804, "learning_rate": 1.0365796649270441e-05, "loss": 3.1039, "step": 191020 }, { "epoch": 0.0072448, "grad_norm": 0.7823843955993652, "learning_rate": 1.0364991908550084e-05, "loss": 2.9873, "step": 191030 }, { "epoch": 0.0072704, "grad_norm": 0.8349603414535522, "learning_rate": 1.0364187165462856e-05, "loss": 2.987, "step": 191040 }, { "epoch": 0.007296, "grad_norm": 0.7686187028884888, "learning_rate": 1.0363382420013967e-05, "loss": 3.0184, "step": 191050 }, { "epoch": 0.0073216, "grad_norm": 0.7273340225219727, "learning_rate": 1.0362577672208642e-05, "loss": 2.7719, "step": 191060 }, { "epoch": 0.0073472, "grad_norm": 0.7707834243774414, "learning_rate": 1.0361772922052095e-05, "loss": 2.9525, "step": 191070 }, { "epoch": 0.0073728, "grad_norm": 0.7815587520599365, "learning_rate": 1.036096816954955e-05, "loss": 3.1005, "step": 191080 }, { "epoch": 0.0073984, "grad_norm": 0.9298770427703857, "learning_rate": 1.0360163414706216e-05, "loss": 2.6963, "step": 191090 }, { "epoch": 0.007424, "grad_norm": 1.031721830368042, "learning_rate": 1.0359358657527323e-05, "loss": 3.0843, "step": 191100 }, { "epoch": 0.0074496, "grad_norm": 0.6844889521598816, "learning_rate": 1.0358553898018087e-05, "loss": 2.8334, "step": 191110 }, { "epoch": 0.0074752, "grad_norm": 0.7818834185600281, "learning_rate": 1.035774913618372e-05, "loss": 3.1282, "step": 191120 }, { "epoch": 0.0075008, "grad_norm": 1.0775668621063232, "learning_rate": 1.0356944372029445e-05, "loss": 3.1433, "step": 191130 }, { "epoch": 0.0075264, "grad_norm": 0.7010362148284912, "learning_rate": 1.0356139605560485e-05, "loss": 2.8663, "step": 191140 }, { "epoch": 0.007552, "grad_norm": 2.0502893924713135, "learning_rate": 1.035533483678205e-05, "loss": 3.157, "step": 191150 }, { "epoch": 0.0075776, "grad_norm": 0.8503884077072144, "learning_rate": 1.0354530065699363e-05, "loss": 2.9957, "step": 191160 }, { "epoch": 0.0076032, "grad_norm": 0.7764557600021362, "learning_rate": 1.0353725292317642e-05, "loss": 3.1475, "step": 191170 }, { "epoch": 0.0076288, "grad_norm": 0.7512797713279724, "learning_rate": 1.0352920516642108e-05, "loss": 3.0034, "step": 191180 }, { "epoch": 0.0076544, "grad_norm": 0.735902726650238, "learning_rate": 1.0352115738677978e-05, "loss": 3.1024, "step": 191190 }, { "epoch": 0.00768, "grad_norm": 0.8383011817932129, "learning_rate": 1.0351310958430471e-05, "loss": 2.8455, "step": 191200 }, { "epoch": 0.0077056, "grad_norm": 0.9426651000976562, "learning_rate": 1.035050617590481e-05, "loss": 3.1006, "step": 191210 }, { "epoch": 0.0077312, "grad_norm": 0.9032086133956909, "learning_rate": 1.0349701391106204e-05, "loss": 2.7887, "step": 191220 }, { "epoch": 0.0077568, "grad_norm": 0.6727348566055298, "learning_rate": 1.034889660403988e-05, "loss": 3.013, "step": 191230 }, { "epoch": 0.0077824, "grad_norm": 1.3177495002746582, "learning_rate": 1.0348091814711055e-05, "loss": 2.859, "step": 191240 }, { "epoch": 0.007808, "grad_norm": 0.8057622313499451, "learning_rate": 1.0347287023124946e-05, "loss": 2.9464, "step": 191250 }, { "epoch": 0.0078336, "grad_norm": 0.9443507194519043, "learning_rate": 1.034648222928677e-05, "loss": 2.8862, "step": 191260 }, { "epoch": 0.0078592, "grad_norm": 0.7688590884208679, "learning_rate": 1.0345677433201754e-05, "loss": 2.9415, "step": 191270 }, { "epoch": 0.0078848, "grad_norm": 0.7372627854347229, "learning_rate": 1.0344872634875107e-05, "loss": 3.0781, "step": 191280 }, { "epoch": 0.0079104, "grad_norm": 0.6868742108345032, "learning_rate": 1.0344067834312055e-05, "loss": 2.8243, "step": 191290 }, { "epoch": 0.007936, "grad_norm": 0.8535084128379822, "learning_rate": 1.0343263031517812e-05, "loss": 2.7657, "step": 191300 }, { "epoch": 0.0079616, "grad_norm": 0.7988948225975037, "learning_rate": 1.0342458226497604e-05, "loss": 2.8039, "step": 191310 }, { "epoch": 0.0079872, "grad_norm": 0.7690409421920776, "learning_rate": 1.0341653419256643e-05, "loss": 3.0345, "step": 191320 }, { "epoch": 0.0080128, "grad_norm": 2.0791091918945312, "learning_rate": 1.034084860980015e-05, "loss": 2.9447, "step": 191330 }, { "epoch": 0.0080384, "grad_norm": 0.8321313858032227, "learning_rate": 1.0340043798133346e-05, "loss": 2.8342, "step": 191340 }, { "epoch": 0.008064, "grad_norm": 0.8100358843803406, "learning_rate": 1.0339238984261446e-05, "loss": 2.8353, "step": 191350 }, { "epoch": 0.0080896, "grad_norm": 0.846347987651825, "learning_rate": 1.0338434168189671e-05, "loss": 2.9694, "step": 191360 }, { "epoch": 0.0081152, "grad_norm": 0.8330489993095398, "learning_rate": 1.0337629349923244e-05, "loss": 2.9211, "step": 191370 }, { "epoch": 0.0081408, "grad_norm": 1.2051630020141602, "learning_rate": 1.0336824529467378e-05, "loss": 3.0532, "step": 191380 }, { "epoch": 0.0081664, "grad_norm": 0.724815845489502, "learning_rate": 1.0336019706827294e-05, "loss": 2.9707, "step": 191390 }, { "epoch": 0.008192, "grad_norm": 1.4534207582473755, "learning_rate": 1.0335214882008214e-05, "loss": 2.812, "step": 191400 }, { "epoch": 0.0082176, "grad_norm": 0.8136664628982544, "learning_rate": 1.0334410055015354e-05, "loss": 3.0646, "step": 191410 }, { "epoch": 0.0082432, "grad_norm": 0.8266032934188843, "learning_rate": 1.0333605225853931e-05, "loss": 2.9077, "step": 191420 }, { "epoch": 0.0082688, "grad_norm": 0.8033483624458313, "learning_rate": 1.0332800394529169e-05, "loss": 3.0354, "step": 191430 }, { "epoch": 0.0082944, "grad_norm": 0.7162717580795288, "learning_rate": 1.0331995561046283e-05, "loss": 2.8541, "step": 191440 }, { "epoch": 0.00832, "grad_norm": 0.7231002449989319, "learning_rate": 1.0331190725410493e-05, "loss": 2.779, "step": 191450 }, { "epoch": 0.0083456, "grad_norm": 0.7624704241752625, "learning_rate": 1.0330385887627024e-05, "loss": 2.8394, "step": 191460 }, { "epoch": 0.0083712, "grad_norm": 0.7842587828636169, "learning_rate": 1.0329581047701087e-05, "loss": 2.9789, "step": 191470 }, { "epoch": 0.0083968, "grad_norm": 0.8811713457107544, "learning_rate": 1.0328776205637907e-05, "loss": 3.0082, "step": 191480 }, { "epoch": 0.0084224, "grad_norm": 1.6513447761535645, "learning_rate": 1.0327971361442697e-05, "loss": 2.8817, "step": 191490 }, { "epoch": 0.008448, "grad_norm": 0.7792569398880005, "learning_rate": 1.0327166515120682e-05, "loss": 2.9939, "step": 191500 }, { "epoch": 0.0084736, "grad_norm": 0.7590814828872681, "learning_rate": 1.0326361666677081e-05, "loss": 2.8084, "step": 191510 }, { "epoch": 0.0084992, "grad_norm": 0.892952024936676, "learning_rate": 1.0325556816117106e-05, "loss": 2.9182, "step": 191520 }, { "epoch": 0.0085248, "grad_norm": 0.8341678380966187, "learning_rate": 1.0324751963445987e-05, "loss": 2.7362, "step": 191530 }, { "epoch": 0.0085504, "grad_norm": 0.689430296421051, "learning_rate": 1.0323947108668935e-05, "loss": 2.9072, "step": 191540 }, { "epoch": 0.008576, "grad_norm": 0.6988338232040405, "learning_rate": 1.0323142251791171e-05, "loss": 3.0285, "step": 191550 }, { "epoch": 0.0086016, "grad_norm": 0.7939666509628296, "learning_rate": 1.0322337392817918e-05, "loss": 2.9885, "step": 191560 }, { "epoch": 0.0086272, "grad_norm": 0.7402086853981018, "learning_rate": 1.0321532531754393e-05, "loss": 2.8665, "step": 191570 }, { "epoch": 0.0086528, "grad_norm": 0.6896170973777771, "learning_rate": 1.0320727668605814e-05, "loss": 2.8579, "step": 191580 }, { "epoch": 0.0086784, "grad_norm": 0.76689213514328, "learning_rate": 1.03199228033774e-05, "loss": 3.056, "step": 191590 }, { "epoch": 0.008704, "grad_norm": 0.7789192795753479, "learning_rate": 1.0319117936074373e-05, "loss": 2.8629, "step": 191600 }, { "epoch": 0.0087296, "grad_norm": 0.7684715986251831, "learning_rate": 1.031831306670195e-05, "loss": 2.9264, "step": 191610 }, { "epoch": 0.0087552, "grad_norm": 0.7299095392227173, "learning_rate": 1.0317508195265351e-05, "loss": 2.8262, "step": 191620 }, { "epoch": 0.0087808, "grad_norm": 0.8493964672088623, "learning_rate": 1.0316703321769799e-05, "loss": 2.8915, "step": 191630 }, { "epoch": 0.0088064, "grad_norm": 0.7464181780815125, "learning_rate": 1.0315898446220508e-05, "loss": 2.8778, "step": 191640 }, { "epoch": 0.008832, "grad_norm": 1.059226632118225, "learning_rate": 1.0315093568622699e-05, "loss": 3.0062, "step": 191650 }, { "epoch": 0.0088576, "grad_norm": 0.745122492313385, "learning_rate": 1.0314288688981591e-05, "loss": 3.1747, "step": 191660 }, { "epoch": 0.0088832, "grad_norm": 0.8849169015884399, "learning_rate": 1.0313483807302407e-05, "loss": 2.8602, "step": 191670 }, { "epoch": 0.0089088, "grad_norm": 0.8163596987724304, "learning_rate": 1.0312678923590362e-05, "loss": 2.9, "step": 191680 }, { "epoch": 0.0089344, "grad_norm": 0.7828478813171387, "learning_rate": 1.0311874037850678e-05, "loss": 2.9346, "step": 191690 }, { "epoch": 0.00896, "grad_norm": 0.9362538456916809, "learning_rate": 1.0311069150088572e-05, "loss": 2.9909, "step": 191700 }, { "epoch": 0.0089856, "grad_norm": 0.8907643556594849, "learning_rate": 1.031026426030927e-05, "loss": 2.964, "step": 191710 }, { "epoch": 0.0090112, "grad_norm": 0.7556707262992859, "learning_rate": 1.0309459368517982e-05, "loss": 3.1045, "step": 191720 }, { "epoch": 0.0090368, "grad_norm": 0.7559517621994019, "learning_rate": 1.0308654474719935e-05, "loss": 2.9017, "step": 191730 }, { "epoch": 0.0090624, "grad_norm": 0.7437829971313477, "learning_rate": 1.0307849578920344e-05, "loss": 2.9107, "step": 191740 }, { "epoch": 0.009088, "grad_norm": 0.6890770792961121, "learning_rate": 1.030704468112443e-05, "loss": 3.0017, "step": 191750 }, { "epoch": 0.0091136, "grad_norm": 0.8521563410758972, "learning_rate": 1.0306239781337415e-05, "loss": 3.2175, "step": 191760 }, { "epoch": 0.0091392, "grad_norm": 0.7616201639175415, "learning_rate": 1.0305434879564516e-05, "loss": 2.9348, "step": 191770 }, { "epoch": 0.0091648, "grad_norm": 0.7205955982208252, "learning_rate": 1.030462997581095e-05, "loss": 2.8292, "step": 191780 }, { "epoch": 0.0091904, "grad_norm": 0.8043673634529114, "learning_rate": 1.030382507008194e-05, "loss": 2.9586, "step": 191790 }, { "epoch": 0.009216, "grad_norm": 0.7503488659858704, "learning_rate": 1.0303020162382708e-05, "loss": 2.8236, "step": 191800 }, { "epoch": 0.0092416, "grad_norm": 0.8334885835647583, "learning_rate": 1.030221525271847e-05, "loss": 2.7489, "step": 191810 }, { "epoch": 0.0092672, "grad_norm": 0.7895580530166626, "learning_rate": 1.0301410341094447e-05, "loss": 2.9175, "step": 191820 }, { "epoch": 0.0092928, "grad_norm": 0.7339438796043396, "learning_rate": 1.0300605427515858e-05, "loss": 2.7755, "step": 191830 }, { "epoch": 0.0093184, "grad_norm": 0.828972339630127, "learning_rate": 1.029980051198792e-05, "loss": 2.9861, "step": 191840 }, { "epoch": 0.009344, "grad_norm": 0.7915014028549194, "learning_rate": 1.0298995594515857e-05, "loss": 2.9185, "step": 191850 }, { "epoch": 0.0093696, "grad_norm": 0.8184660077095032, "learning_rate": 1.0298190675104888e-05, "loss": 2.9694, "step": 191860 }, { "epoch": 0.0093952, "grad_norm": 0.7204355597496033, "learning_rate": 1.029738575376023e-05, "loss": 2.9307, "step": 191870 }, { "epoch": 0.0094208, "grad_norm": 0.6883009076118469, "learning_rate": 1.0296580830487107e-05, "loss": 3.0027, "step": 191880 }, { "epoch": 0.0094464, "grad_norm": 0.7845820784568787, "learning_rate": 1.0295775905290732e-05, "loss": 2.8761, "step": 191890 }, { "epoch": 0.009472, "grad_norm": 0.7739143371582031, "learning_rate": 1.0294970978176333e-05, "loss": 2.9138, "step": 191900 }, { "epoch": 0.0094976, "grad_norm": 0.803756833076477, "learning_rate": 1.0294166049149123e-05, "loss": 2.8029, "step": 191910 }, { "epoch": 0.0095232, "grad_norm": 0.8264036178588867, "learning_rate": 1.0293361118214324e-05, "loss": 3.1276, "step": 191920 }, { "epoch": 0.0095488, "grad_norm": 0.8709864020347595, "learning_rate": 1.029255618537716e-05, "loss": 2.9098, "step": 191930 }, { "epoch": 0.0095744, "grad_norm": 0.7277042269706726, "learning_rate": 1.0291751250642846e-05, "loss": 2.9707, "step": 191940 }, { "epoch": 0.0096, "grad_norm": 1.362764835357666, "learning_rate": 1.0290946314016599e-05, "loss": 3.1501, "step": 191950 }, { "epoch": 0.0096256, "grad_norm": 0.9653395414352417, "learning_rate": 1.0290141375503643e-05, "loss": 2.9004, "step": 191960 }, { "epoch": 0.0096512, "grad_norm": 0.8414797186851501, "learning_rate": 1.0289336435109199e-05, "loss": 2.9652, "step": 191970 }, { "epoch": 0.0096768, "grad_norm": 0.8088985681533813, "learning_rate": 1.0288531492838485e-05, "loss": 3.0473, "step": 191980 }, { "epoch": 0.0097024, "grad_norm": 0.8578226566314697, "learning_rate": 1.028772654869672e-05, "loss": 2.9702, "step": 191990 }, { "epoch": 0.009728, "grad_norm": 0.7319501638412476, "learning_rate": 1.0286921602689128e-05, "loss": 2.8626, "step": 192000 }, { "epoch": 0.0097536, "grad_norm": 0.7634103894233704, "learning_rate": 1.0286116654820924e-05, "loss": 2.7416, "step": 192010 }, { "epoch": 0.0097792, "grad_norm": 0.7873709797859192, "learning_rate": 1.0285311705097327e-05, "loss": 2.941, "step": 192020 }, { "epoch": 0.0098048, "grad_norm": 0.9544625878334045, "learning_rate": 1.0284506753523562e-05, "loss": 3.0368, "step": 192030 }, { "epoch": 0.0098304, "grad_norm": 0.7129387259483337, "learning_rate": 1.0283701800104844e-05, "loss": 2.7672, "step": 192040 }, { "epoch": 0.009856, "grad_norm": 0.8134719133377075, "learning_rate": 1.0282896844846395e-05, "loss": 3.1962, "step": 192050 }, { "epoch": 0.0098816, "grad_norm": 0.8345435857772827, "learning_rate": 1.0282091887753441e-05, "loss": 2.8065, "step": 192060 }, { "epoch": 0.0099072, "grad_norm": 0.788140058517456, "learning_rate": 1.0281286928831191e-05, "loss": 2.9725, "step": 192070 }, { "epoch": 0.0099328, "grad_norm": 0.7574419379234314, "learning_rate": 1.028048196808487e-05, "loss": 3.005, "step": 192080 }, { "epoch": 0.0099584, "grad_norm": 0.6772387027740479, "learning_rate": 1.02796770055197e-05, "loss": 2.9053, "step": 192090 }, { "epoch": 0.009984, "grad_norm": 0.75029057264328, "learning_rate": 1.0278872041140898e-05, "loss": 2.8898, "step": 192100 }, { "epoch": 0.0100096, "grad_norm": 0.8056616187095642, "learning_rate": 1.0278067074953683e-05, "loss": 2.8259, "step": 192110 }, { "epoch": 0.0100352, "grad_norm": 0.7921431660652161, "learning_rate": 1.027726210696328e-05, "loss": 2.6805, "step": 192120 }, { "epoch": 0.0100608, "grad_norm": 0.8598681688308716, "learning_rate": 1.0276457137174905e-05, "loss": 2.9761, "step": 192130 }, { "epoch": 0.0100864, "grad_norm": 0.7165738344192505, "learning_rate": 1.0275652165593777e-05, "loss": 2.75, "step": 192140 }, { "epoch": 0.010112, "grad_norm": 0.8485819697380066, "learning_rate": 1.0274847192225119e-05, "loss": 2.9403, "step": 192150 }, { "epoch": 0.0101376, "grad_norm": 0.7238340973854065, "learning_rate": 1.0274042217074151e-05, "loss": 3.0332, "step": 192160 }, { "epoch": 0.0101632, "grad_norm": 0.8954522609710693, "learning_rate": 1.0273237240146092e-05, "loss": 3.0656, "step": 192170 }, { "epoch": 0.0101888, "grad_norm": 0.7184801697731018, "learning_rate": 1.0272432261446161e-05, "loss": 3.0556, "step": 192180 }, { "epoch": 0.0102144, "grad_norm": 0.72600257396698, "learning_rate": 1.027162728097958e-05, "loss": 2.8615, "step": 192190 }, { "epoch": 0.01024, "grad_norm": 0.6874556541442871, "learning_rate": 1.027082229875157e-05, "loss": 2.7649, "step": 192200 }, { "epoch": 0.0102656, "grad_norm": 0.8094338774681091, "learning_rate": 1.0270017314767348e-05, "loss": 2.9269, "step": 192210 }, { "epoch": 0.0102912, "grad_norm": 0.8693355917930603, "learning_rate": 1.0269212329032131e-05, "loss": 3.0505, "step": 192220 }, { "epoch": 0.0103168, "grad_norm": 0.8595824837684631, "learning_rate": 1.026840734155115e-05, "loss": 3.1707, "step": 192230 }, { "epoch": 0.0103424, "grad_norm": 0.7696384191513062, "learning_rate": 1.0267602352329618e-05, "loss": 2.9476, "step": 192240 }, { "epoch": 0.010368, "grad_norm": 0.764105498790741, "learning_rate": 1.0266797361372753e-05, "loss": 3.054, "step": 192250 }, { "epoch": 0.0103936, "grad_norm": 0.8691015243530273, "learning_rate": 1.026599236868578e-05, "loss": 2.9544, "step": 192260 }, { "epoch": 0.0104192, "grad_norm": 0.8075529932975769, "learning_rate": 1.0265187374273917e-05, "loss": 2.869, "step": 192270 }, { "epoch": 0.0104448, "grad_norm": 1.194828748703003, "learning_rate": 1.0264382378142384e-05, "loss": 2.9622, "step": 192280 }, { "epoch": 0.0104704, "grad_norm": 1.282515048980713, "learning_rate": 1.02635773802964e-05, "loss": 2.8779, "step": 192290 }, { "epoch": 0.010496, "grad_norm": 2.859088897705078, "learning_rate": 1.026277238074119e-05, "loss": 2.8081, "step": 192300 }, { "epoch": 0.0105216, "grad_norm": 0.7865487933158875, "learning_rate": 1.0261967379481967e-05, "loss": 2.9627, "step": 192310 }, { "epoch": 0.0105472, "grad_norm": 0.7247976660728455, "learning_rate": 1.0261162376523958e-05, "loss": 3.2142, "step": 192320 }, { "epoch": 0.0105728, "grad_norm": 0.7608330249786377, "learning_rate": 1.026035737187238e-05, "loss": 3.2552, "step": 192330 }, { "epoch": 0.0105984, "grad_norm": 0.7256436347961426, "learning_rate": 1.0259552365532453e-05, "loss": 2.9193, "step": 192340 }, { "epoch": 0.010624, "grad_norm": 0.781309187412262, "learning_rate": 1.0258747357509397e-05, "loss": 2.9216, "step": 192350 }, { "epoch": 0.0106496, "grad_norm": 0.7594811916351318, "learning_rate": 1.0257942347808435e-05, "loss": 2.9493, "step": 192360 }, { "epoch": 0.0106752, "grad_norm": 0.7323684096336365, "learning_rate": 1.0257137336434787e-05, "loss": 2.7598, "step": 192370 }, { "epoch": 0.0107008, "grad_norm": 0.810799241065979, "learning_rate": 1.0256332323393667e-05, "loss": 2.8584, "step": 192380 }, { "epoch": 0.0107264, "grad_norm": 0.7769021391868591, "learning_rate": 1.0255527308690303e-05, "loss": 2.6596, "step": 192390 }, { "epoch": 0.010752, "grad_norm": 0.7374390363693237, "learning_rate": 1.025472229232991e-05, "loss": 2.9645, "step": 192400 }, { "epoch": 0.0107776, "grad_norm": 0.7349810600280762, "learning_rate": 1.0253917274317714e-05, "loss": 2.9866, "step": 192410 }, { "epoch": 0.0108032, "grad_norm": 0.7774659395217896, "learning_rate": 1.025311225465893e-05, "loss": 2.9657, "step": 192420 }, { "epoch": 0.0108288, "grad_norm": 0.7752719521522522, "learning_rate": 1.025230723335878e-05, "loss": 3.0191, "step": 192430 }, { "epoch": 0.0108544, "grad_norm": 0.6733680963516235, "learning_rate": 1.0251502210422487e-05, "loss": 2.9599, "step": 192440 }, { "epoch": 0.01088, "grad_norm": 0.7364615201950073, "learning_rate": 1.0250697185855265e-05, "loss": 2.874, "step": 192450 }, { "epoch": 0.0109056, "grad_norm": 0.7175659537315369, "learning_rate": 1.0249892159662342e-05, "loss": 2.9485, "step": 192460 }, { "epoch": 0.0109312, "grad_norm": 0.7266128063201904, "learning_rate": 1.0249087131848934e-05, "loss": 3.0362, "step": 192470 }, { "epoch": 0.0109568, "grad_norm": 0.8509339094161987, "learning_rate": 1.024828210242026e-05, "loss": 3.1083, "step": 192480 }, { "epoch": 0.0109824, "grad_norm": 0.8222407102584839, "learning_rate": 1.0247477071381544e-05, "loss": 3.1409, "step": 192490 }, { "epoch": 0.011008, "grad_norm": 0.9673833250999451, "learning_rate": 1.0246672038738006e-05, "loss": 2.9573, "step": 192500 }, { "epoch": 0.0110336, "grad_norm": 0.9158811569213867, "learning_rate": 1.0245867004494863e-05, "loss": 2.8746, "step": 192510 }, { "epoch": 0.0110592, "grad_norm": 0.6813980937004089, "learning_rate": 1.0245061968657342e-05, "loss": 3.0161, "step": 192520 }, { "epoch": 0.0110848, "grad_norm": 0.7356101870536804, "learning_rate": 1.0244256931230656e-05, "loss": 2.8521, "step": 192530 }, { "epoch": 0.0111104, "grad_norm": 0.9303914308547974, "learning_rate": 1.024345189222003e-05, "loss": 3.2113, "step": 192540 }, { "epoch": 0.011136, "grad_norm": 1.0604792833328247, "learning_rate": 1.024264685163068e-05, "loss": 2.9993, "step": 192550 }, { "epoch": 0.0111616, "grad_norm": 0.749144434928894, "learning_rate": 1.0241841809467834e-05, "loss": 3.0486, "step": 192560 }, { "epoch": 0.0111872, "grad_norm": 0.7430991530418396, "learning_rate": 1.0241036765736705e-05, "loss": 2.9461, "step": 192570 }, { "epoch": 0.0112128, "grad_norm": 0.9004727602005005, "learning_rate": 1.0240231720442517e-05, "loss": 3.0556, "step": 192580 }, { "epoch": 0.0112384, "grad_norm": 0.7475812435150146, "learning_rate": 1.0239426673590493e-05, "loss": 2.9597, "step": 192590 }, { "epoch": 0.011264, "grad_norm": 0.7931755185127258, "learning_rate": 1.0238621625185848e-05, "loss": 2.992, "step": 192600 }, { "epoch": 0.0112896, "grad_norm": 0.7601774334907532, "learning_rate": 1.0237816575233806e-05, "loss": 2.9716, "step": 192610 }, { "epoch": 0.0113152, "grad_norm": 0.7844532132148743, "learning_rate": 1.0237011523739588e-05, "loss": 2.9314, "step": 192620 }, { "epoch": 0.0113408, "grad_norm": 0.7636978626251221, "learning_rate": 1.0236206470708411e-05, "loss": 3.0461, "step": 192630 }, { "epoch": 0.0113664, "grad_norm": 0.7407160401344299, "learning_rate": 1.0235401416145498e-05, "loss": 2.9047, "step": 192640 }, { "epoch": 0.011392, "grad_norm": 1.4988919496536255, "learning_rate": 1.023459636005607e-05, "loss": 2.9398, "step": 192650 }, { "epoch": 0.0114176, "grad_norm": 0.9335858821868896, "learning_rate": 1.0233791302445346e-05, "loss": 2.7683, "step": 192660 }, { "epoch": 0.0114432, "grad_norm": 0.8771565556526184, "learning_rate": 1.0232986243318547e-05, "loss": 2.7802, "step": 192670 }, { "epoch": 0.0114688, "grad_norm": 0.8198371529579163, "learning_rate": 1.0232181182680897e-05, "loss": 2.8933, "step": 192680 }, { "epoch": 0.0114944, "grad_norm": 0.9335376620292664, "learning_rate": 1.0231376120537612e-05, "loss": 2.9203, "step": 192690 }, { "epoch": 0.01152, "grad_norm": 0.7512243390083313, "learning_rate": 1.0230571056893916e-05, "loss": 2.9531, "step": 192700 }, { "epoch": 0.0115456, "grad_norm": 1.8613865375518799, "learning_rate": 1.0229765991755027e-05, "loss": 3.1941, "step": 192710 }, { "epoch": 0.0115712, "grad_norm": 0.8714993596076965, "learning_rate": 1.0228960925126168e-05, "loss": 3.2031, "step": 192720 }, { "epoch": 0.0115968, "grad_norm": 0.8475188612937927, "learning_rate": 1.0228155857012556e-05, "loss": 3.1837, "step": 192730 }, { "epoch": 0.0116224, "grad_norm": 0.748102068901062, "learning_rate": 1.0227350787419412e-05, "loss": 2.8355, "step": 192740 }, { "epoch": 0.011648, "grad_norm": 0.9549334049224854, "learning_rate": 1.0226545716351963e-05, "loss": 3.1475, "step": 192750 }, { "epoch": 0.0116736, "grad_norm": 0.7743998169898987, "learning_rate": 1.0225740643815422e-05, "loss": 2.8963, "step": 192760 }, { "epoch": 0.0116992, "grad_norm": 0.9697620272636414, "learning_rate": 1.0224935569815013e-05, "loss": 2.89, "step": 192770 }, { "epoch": 0.0117248, "grad_norm": 2.4635047912597656, "learning_rate": 1.022413049435596e-05, "loss": 3.1079, "step": 192780 }, { "epoch": 0.0117504, "grad_norm": 1.038630723953247, "learning_rate": 1.0223325417443479e-05, "loss": 3.0484, "step": 192790 }, { "epoch": 0.011776, "grad_norm": 0.7618648409843445, "learning_rate": 1.0222520339082792e-05, "loss": 3.1106, "step": 192800 }, { "epoch": 0.0118016, "grad_norm": 0.825998067855835, "learning_rate": 1.0221715259279116e-05, "loss": 2.6213, "step": 192810 }, { "epoch": 0.0118272, "grad_norm": 0.7267927527427673, "learning_rate": 1.022091017803768e-05, "loss": 2.8304, "step": 192820 }, { "epoch": 0.0118528, "grad_norm": 0.9118193984031677, "learning_rate": 1.0220105095363697e-05, "loss": 3.0119, "step": 192830 }, { "epoch": 0.0118784, "grad_norm": 0.7567543983459473, "learning_rate": 1.0219300011262391e-05, "loss": 2.9347, "step": 192840 }, { "epoch": 0.011904, "grad_norm": 0.9752992391586304, "learning_rate": 1.0218494925738987e-05, "loss": 2.9638, "step": 192850 }, { "epoch": 0.0119296, "grad_norm": 0.7858240008354187, "learning_rate": 1.0217689838798696e-05, "loss": 2.6912, "step": 192860 }, { "epoch": 0.0119552, "grad_norm": 0.9739898443222046, "learning_rate": 1.0216884750446748e-05, "loss": 3.0619, "step": 192870 }, { "epoch": 0.0119808, "grad_norm": 0.8490264415740967, "learning_rate": 1.0216079660688358e-05, "loss": 2.6945, "step": 192880 }, { "epoch": 0.0120064, "grad_norm": 0.7393296360969543, "learning_rate": 1.0215274569528753e-05, "loss": 2.6233, "step": 192890 }, { "epoch": 0.012032, "grad_norm": 0.7120717763900757, "learning_rate": 1.0214469476973145e-05, "loss": 2.8696, "step": 192900 }, { "epoch": 2.56e-05, "grad_norm": 1.2446199655532837, "learning_rate": 1.0213664383026757e-05, "loss": 1.4476, "step": 192910 }, { "epoch": 5.12e-05, "grad_norm": 0.7916384339332581, "learning_rate": 1.0212859287694818e-05, "loss": 2.834, "step": 192920 }, { "epoch": 7.68e-05, "grad_norm": 0.8311423063278198, "learning_rate": 1.0212054190982544e-05, "loss": 2.9299, "step": 192930 }, { "epoch": 0.0001024, "grad_norm": 0.7941341996192932, "learning_rate": 1.021124909289515e-05, "loss": 2.9147, "step": 192940 }, { "epoch": 0.000128, "grad_norm": 0.7101889848709106, "learning_rate": 1.0210443993437866e-05, "loss": 2.8567, "step": 192950 }, { "epoch": 0.0001536, "grad_norm": 0.8131462335586548, "learning_rate": 1.0209638892615905e-05, "loss": 2.7439, "step": 192960 }, { "epoch": 0.0001792, "grad_norm": 0.9061316251754761, "learning_rate": 1.0208833790434494e-05, "loss": 2.755, "step": 192970 }, { "epoch": 0.0002048, "grad_norm": 0.8333636522293091, "learning_rate": 1.020802868689885e-05, "loss": 2.6631, "step": 192980 }, { "epoch": 0.0002304, "grad_norm": 0.6967452764511108, "learning_rate": 1.0207223582014196e-05, "loss": 2.6895, "step": 192990 }, { "epoch": 0.000256, "grad_norm": 0.8622738718986511, "learning_rate": 1.020641847578575e-05, "loss": 2.8244, "step": 193000 }, { "epoch": 0.0002816, "grad_norm": 0.7525705695152283, "learning_rate": 1.0205613368218736e-05, "loss": 2.7789, "step": 193010 }, { "epoch": 0.0003072, "grad_norm": 0.6946224570274353, "learning_rate": 1.0204808259318377e-05, "loss": 2.5539, "step": 193020 }, { "epoch": 0.0003328, "grad_norm": 0.8986850380897522, "learning_rate": 1.020400314908989e-05, "loss": 2.7267, "step": 193030 }, { "epoch": 0.0003584, "grad_norm": 0.7884198427200317, "learning_rate": 1.0203198037538493e-05, "loss": 2.9196, "step": 193040 }, { "epoch": 0.000384, "grad_norm": 0.7879482507705688, "learning_rate": 1.0202392924669416e-05, "loss": 2.9392, "step": 193050 }, { "epoch": 0.0004096, "grad_norm": 0.7162853479385376, "learning_rate": 1.0201587810487872e-05, "loss": 2.8157, "step": 193060 }, { "epoch": 0.0004352, "grad_norm": 0.9195162653923035, "learning_rate": 1.0200782694999083e-05, "loss": 2.5706, "step": 193070 }, { "epoch": 0.0004608, "grad_norm": 0.8008622527122498, "learning_rate": 1.0199977578208274e-05, "loss": 2.5885, "step": 193080 }, { "epoch": 0.0004864, "grad_norm": 0.8696520328521729, "learning_rate": 1.0199172460120661e-05, "loss": 2.7806, "step": 193090 }, { "epoch": 0.000512, "grad_norm": 0.7155457735061646, "learning_rate": 1.0198367340741468e-05, "loss": 2.9851, "step": 193100 }, { "epoch": 0.0005376, "grad_norm": 0.9846884608268738, "learning_rate": 1.0197562220075915e-05, "loss": 2.4434, "step": 193110 }, { "epoch": 0.0005632, "grad_norm": 0.8190251588821411, "learning_rate": 1.0196757098129229e-05, "loss": 2.6413, "step": 193120 }, { "epoch": 0.0005888, "grad_norm": 0.7591040134429932, "learning_rate": 1.019595197490662e-05, "loss": 2.8283, "step": 193130 }, { "epoch": 0.0006144, "grad_norm": 0.8014528155326843, "learning_rate": 1.0195146850413318e-05, "loss": 2.7098, "step": 193140 }, { "epoch": 0.00064, "grad_norm": 0.7282348275184631, "learning_rate": 1.0194341724654539e-05, "loss": 2.8665, "step": 193150 }, { "epoch": 0.0006656, "grad_norm": 0.6663546562194824, "learning_rate": 1.0193536597635506e-05, "loss": 2.942, "step": 193160 }, { "epoch": 0.0006912, "grad_norm": 0.7255133986473083, "learning_rate": 1.0192731469361439e-05, "loss": 2.7145, "step": 193170 }, { "epoch": 0.0007168, "grad_norm": 0.8396111130714417, "learning_rate": 1.019192633983756e-05, "loss": 2.6301, "step": 193180 }, { "epoch": 0.0007424, "grad_norm": 0.7395874857902527, "learning_rate": 1.0191121209069087e-05, "loss": 2.9537, "step": 193190 }, { "epoch": 0.000768, "grad_norm": 0.8103351593017578, "learning_rate": 1.0190316077061247e-05, "loss": 2.5187, "step": 193200 }, { "epoch": 0.0007936, "grad_norm": 0.7147185802459717, "learning_rate": 1.0189510943819254e-05, "loss": 2.8163, "step": 193210 }, { "epoch": 0.0008192, "grad_norm": 0.7756761312484741, "learning_rate": 1.0188705809348338e-05, "loss": 2.9112, "step": 193220 }, { "epoch": 0.0008448, "grad_norm": 0.7456347346305847, "learning_rate": 1.0187900673653712e-05, "loss": 2.8576, "step": 193230 }, { "epoch": 0.0008704, "grad_norm": 0.9242751598358154, "learning_rate": 1.0187095536740602e-05, "loss": 2.6776, "step": 193240 }, { "epoch": 0.000896, "grad_norm": 0.847589910030365, "learning_rate": 1.0186290398614227e-05, "loss": 2.8139, "step": 193250 }, { "epoch": 0.0009216, "grad_norm": 0.8073118329048157, "learning_rate": 1.0185485259279804e-05, "loss": 2.5244, "step": 193260 }, { "epoch": 0.0009472, "grad_norm": 0.7651981711387634, "learning_rate": 1.0184680118742562e-05, "loss": 2.5816, "step": 193270 }, { "epoch": 0.0009728, "grad_norm": 0.9166997671127319, "learning_rate": 1.018387497700772e-05, "loss": 2.8749, "step": 193280 }, { "epoch": 0.0009984, "grad_norm": 0.9517240524291992, "learning_rate": 1.0183069834080495e-05, "loss": 2.9805, "step": 193290 }, { "epoch": 0.001024, "grad_norm": 0.7155195474624634, "learning_rate": 1.018226468996611e-05, "loss": 2.7516, "step": 193300 }, { "epoch": 0.0010496, "grad_norm": 0.7264091372489929, "learning_rate": 1.0181459544669792e-05, "loss": 2.7453, "step": 193310 }, { "epoch": 0.0010752, "grad_norm": 0.7760669589042664, "learning_rate": 1.0180654398196753e-05, "loss": 2.9775, "step": 193320 }, { "epoch": 0.0011008, "grad_norm": 0.6853692531585693, "learning_rate": 1.0179849250552218e-05, "loss": 2.7078, "step": 193330 }, { "epoch": 0.0011264, "grad_norm": 0.7504889965057373, "learning_rate": 1.0179044101741409e-05, "loss": 2.9316, "step": 193340 }, { "epoch": 0.001152, "grad_norm": 0.9772563576698303, "learning_rate": 1.0178238951769546e-05, "loss": 2.893, "step": 193350 }, { "epoch": 0.0011776, "grad_norm": 0.9242625832557678, "learning_rate": 1.0177433800641851e-05, "loss": 2.8204, "step": 193360 }, { "epoch": 0.0012032, "grad_norm": 0.7675684690475464, "learning_rate": 1.0176628648363545e-05, "loss": 2.8089, "step": 193370 }, { "epoch": 0.0012288, "grad_norm": 0.7953625321388245, "learning_rate": 1.0175823494939852e-05, "loss": 2.8806, "step": 193380 }, { "epoch": 0.0012544, "grad_norm": 0.791385293006897, "learning_rate": 1.0175018340375988e-05, "loss": 2.7376, "step": 193390 }, { "epoch": 0.00128, "grad_norm": 0.9847152829170227, "learning_rate": 1.0174213184677176e-05, "loss": 2.7656, "step": 193400 }, { "epoch": 0.0013056, "grad_norm": 0.9045323133468628, "learning_rate": 1.017340802784864e-05, "loss": 2.9786, "step": 193410 }, { "epoch": 0.0013312, "grad_norm": 0.8744494318962097, "learning_rate": 1.0172602869895597e-05, "loss": 2.8366, "step": 193420 }, { "epoch": 0.0013568, "grad_norm": 0.9028922319412231, "learning_rate": 1.017179771082327e-05, "loss": 2.7584, "step": 193430 }, { "epoch": 0.0013824, "grad_norm": 0.8901078104972839, "learning_rate": 1.017099255063688e-05, "loss": 2.7403, "step": 193440 }, { "epoch": 0.001408, "grad_norm": 0.7284999489784241, "learning_rate": 1.0170187389341651e-05, "loss": 2.9252, "step": 193450 }, { "epoch": 0.0014336, "grad_norm": 0.7762867212295532, "learning_rate": 1.01693822269428e-05, "loss": 2.8049, "step": 193460 }, { "epoch": 0.0014592, "grad_norm": 0.852617621421814, "learning_rate": 1.0168577063445551e-05, "loss": 2.7677, "step": 193470 }, { "epoch": 0.0014848, "grad_norm": 0.7341381907463074, "learning_rate": 1.0167771898855126e-05, "loss": 2.7332, "step": 193480 }, { "epoch": 0.0015104, "grad_norm": 0.7654337882995605, "learning_rate": 1.0166966733176743e-05, "loss": 2.419, "step": 193490 }, { "epoch": 0.001536, "grad_norm": 0.8346768617630005, "learning_rate": 1.0166161566415627e-05, "loss": 2.2957, "step": 193500 }, { "epoch": 2.56e-05, "grad_norm": 1.2703402042388916, "learning_rate": 1.0165356398576997e-05, "loss": 1.4846, "step": 193510 }, { "epoch": 5.12e-05, "grad_norm": 0.7837549448013306, "learning_rate": 1.016455122966607e-05, "loss": 2.9419, "step": 193520 }, { "epoch": 7.68e-05, "grad_norm": 0.8322202563285828, "learning_rate": 1.0163746059688075e-05, "loss": 2.7903, "step": 193530 }, { "epoch": 0.0001024, "grad_norm": 0.7339732050895691, "learning_rate": 1.0162940888648233e-05, "loss": 2.9478, "step": 193540 }, { "epoch": 0.000128, "grad_norm": 0.9018951654434204, "learning_rate": 1.0162135716551762e-05, "loss": 2.6837, "step": 193550 }, { "epoch": 0.0001536, "grad_norm": 0.7956476211547852, "learning_rate": 1.0161330543403883e-05, "loss": 2.7749, "step": 193560 }, { "epoch": 0.0001792, "grad_norm": 0.9721594452857971, "learning_rate": 1.0160525369209817e-05, "loss": 2.4791, "step": 193570 }, { "epoch": 0.0002048, "grad_norm": 0.913395345211029, "learning_rate": 1.0159720193974789e-05, "loss": 2.9407, "step": 193580 }, { "epoch": 0.0002304, "grad_norm": 0.7260096073150635, "learning_rate": 1.0158915017704017e-05, "loss": 2.8169, "step": 193590 }, { "epoch": 0.000256, "grad_norm": 0.9788674712181091, "learning_rate": 1.015810984040272e-05, "loss": 2.6812, "step": 193600 }, { "epoch": 0.0002816, "grad_norm": 0.8579797148704529, "learning_rate": 1.0157304662076128e-05, "loss": 2.7795, "step": 193610 }, { "epoch": 0.0003072, "grad_norm": 0.6654996275901794, "learning_rate": 1.0156499482729454e-05, "loss": 2.5153, "step": 193620 }, { "epoch": 0.0003328, "grad_norm": 0.9011924862861633, "learning_rate": 1.0155694302367921e-05, "loss": 2.5509, "step": 193630 }, { "epoch": 0.0003584, "grad_norm": 0.789561927318573, "learning_rate": 1.0154889120996757e-05, "loss": 2.7159, "step": 193640 }, { "epoch": 0.000384, "grad_norm": 0.839187502861023, "learning_rate": 1.0154083938621175e-05, "loss": 2.8281, "step": 193650 }, { "epoch": 0.0004096, "grad_norm": 0.8386207818984985, "learning_rate": 1.01532787552464e-05, "loss": 2.9178, "step": 193660 }, { "epoch": 0.0004352, "grad_norm": 0.7382990717887878, "learning_rate": 1.0152473570877653e-05, "loss": 2.7698, "step": 193670 }, { "epoch": 0.0004608, "grad_norm": 1.0213390588760376, "learning_rate": 1.0151668385520157e-05, "loss": 2.8684, "step": 193680 }, { "epoch": 0.0004864, "grad_norm": 0.9354600310325623, "learning_rate": 1.015086319917913e-05, "loss": 2.7424, "step": 193690 }, { "epoch": 0.000512, "grad_norm": 0.7168666124343872, "learning_rate": 1.015005801185979e-05, "loss": 2.8376, "step": 193700 }, { "epoch": 0.0005376, "grad_norm": 0.8509896993637085, "learning_rate": 1.0149252823567373e-05, "loss": 2.4614, "step": 193710 }, { "epoch": 0.0005632, "grad_norm": 0.770260214805603, "learning_rate": 1.0148447634307088e-05, "loss": 2.6963, "step": 193720 }, { "epoch": 0.0005888, "grad_norm": 0.7882283329963684, "learning_rate": 1.0147642444084159e-05, "loss": 2.7234, "step": 193730 }, { "epoch": 0.0006144, "grad_norm": 1.1964375972747803, "learning_rate": 1.0146837252903808e-05, "loss": 2.4532, "step": 193740 }, { "epoch": 0.00064, "grad_norm": 0.8275858759880066, "learning_rate": 1.0146032060771256e-05, "loss": 2.9428, "step": 193750 }, { "epoch": 0.0006656, "grad_norm": 0.9132257103919983, "learning_rate": 1.0145226867691724e-05, "loss": 2.9204, "step": 193760 }, { "epoch": 0.0006912, "grad_norm": 0.7779669165611267, "learning_rate": 1.0144421673670434e-05, "loss": 2.6785, "step": 193770 }, { "epoch": 0.0007168, "grad_norm": 0.7980548739433289, "learning_rate": 1.0143616478712611e-05, "loss": 2.6936, "step": 193780 }, { "epoch": 0.0007424, "grad_norm": 0.6926672458648682, "learning_rate": 1.014281128282347e-05, "loss": 2.905, "step": 193790 }, { "epoch": 0.000768, "grad_norm": 0.834566593170166, "learning_rate": 1.0142006086008237e-05, "loss": 2.8204, "step": 193800 }, { "epoch": 0.0007936, "grad_norm": 0.9408842325210571, "learning_rate": 1.0141200888272136e-05, "loss": 2.9822, "step": 193810 }, { "epoch": 0.0008192, "grad_norm": 0.8615327477455139, "learning_rate": 1.014039568962038e-05, "loss": 2.9128, "step": 193820 }, { "epoch": 0.0008448, "grad_norm": 0.8460205793380737, "learning_rate": 1.0139590490058196e-05, "loss": 2.4193, "step": 193830 }, { "epoch": 0.0008704, "grad_norm": 0.7873665690422058, "learning_rate": 1.0138785289590806e-05, "loss": 2.765, "step": 193840 }, { "epoch": 0.000896, "grad_norm": 0.7393996119499207, "learning_rate": 1.0137980088223432e-05, "loss": 2.6966, "step": 193850 }, { "epoch": 0.0009216, "grad_norm": 0.7526448965072632, "learning_rate": 1.013717488596129e-05, "loss": 2.6996, "step": 193860 }, { "epoch": 0.0009472, "grad_norm": 0.8180150389671326, "learning_rate": 1.0136369682809605e-05, "loss": 2.7943, "step": 193870 }, { "epoch": 0.0009728, "grad_norm": 0.7661412954330444, "learning_rate": 1.0135564478773601e-05, "loss": 2.7928, "step": 193880 }, { "epoch": 0.0009984, "grad_norm": 0.7515110969543457, "learning_rate": 1.0134759273858496e-05, "loss": 2.6336, "step": 193890 }, { "epoch": 0.001024, "grad_norm": 0.8242220878601074, "learning_rate": 1.0133954068069513e-05, "loss": 2.8921, "step": 193900 }, { "epoch": 0.0010496, "grad_norm": 0.8276255130767822, "learning_rate": 1.0133148861411876e-05, "loss": 2.7224, "step": 193910 }, { "epoch": 0.0010752, "grad_norm": 1.116858959197998, "learning_rate": 1.0132343653890804e-05, "loss": 2.7468, "step": 193920 }, { "epoch": 0.0011008, "grad_norm": 0.7975620627403259, "learning_rate": 1.0131538445511517e-05, "loss": 2.9934, "step": 193930 }, { "epoch": 0.0011264, "grad_norm": 0.8675635457038879, "learning_rate": 1.013073323627924e-05, "loss": 2.8991, "step": 193940 }, { "epoch": 0.001152, "grad_norm": 0.9319199323654175, "learning_rate": 1.0129928026199189e-05, "loss": 2.7369, "step": 193950 }, { "epoch": 0.0011776, "grad_norm": 0.8175927400588989, "learning_rate": 1.0129122815276589e-05, "loss": 2.6762, "step": 193960 }, { "epoch": 0.0012032, "grad_norm": 0.7965586185455322, "learning_rate": 1.0128317603516666e-05, "loss": 2.8563, "step": 193970 }, { "epoch": 0.0012288, "grad_norm": 0.756384551525116, "learning_rate": 1.0127512390924635e-05, "loss": 2.7759, "step": 193980 }, { "epoch": 0.0012544, "grad_norm": 0.7655736207962036, "learning_rate": 1.0126707177505722e-05, "loss": 2.9962, "step": 193990 }, { "epoch": 0.00128, "grad_norm": 0.8536878824234009, "learning_rate": 1.0125901963265143e-05, "loss": 2.8116, "step": 194000 }, { "epoch": 0.0013056, "grad_norm": 0.8370334506034851, "learning_rate": 1.0125096748208127e-05, "loss": 2.8564, "step": 194010 }, { "epoch": 0.0013312, "grad_norm": 0.7312645316123962, "learning_rate": 1.012429153233989e-05, "loss": 2.8626, "step": 194020 }, { "epoch": 0.0013568, "grad_norm": 0.8986554145812988, "learning_rate": 1.0123486315665656e-05, "loss": 2.61, "step": 194030 }, { "epoch": 0.0013824, "grad_norm": 0.7456656098365784, "learning_rate": 1.0122681098190648e-05, "loss": 2.6265, "step": 194040 }, { "epoch": 0.001408, "grad_norm": 0.7581128478050232, "learning_rate": 1.012187587992008e-05, "loss": 2.5905, "step": 194050 }, { "epoch": 0.0014336, "grad_norm": 0.7103559374809265, "learning_rate": 1.0121070660859183e-05, "loss": 2.6741, "step": 194060 }, { "epoch": 0.0014592, "grad_norm": 0.7957890629768372, "learning_rate": 1.0120265441013175e-05, "loss": 2.7821, "step": 194070 }, { "epoch": 0.0014848, "grad_norm": 0.7375416159629822, "learning_rate": 1.0119460220387276e-05, "loss": 2.7598, "step": 194080 }, { "epoch": 0.0015104, "grad_norm": 0.7658825516700745, "learning_rate": 1.0118654998986712e-05, "loss": 2.6414, "step": 194090 }, { "epoch": 0.001536, "grad_norm": 0.7853516340255737, "learning_rate": 1.01178497768167e-05, "loss": 2.3278, "step": 194100 }, { "epoch": 0.0015616, "grad_norm": 0.8037070631980896, "learning_rate": 1.0117044553882464e-05, "loss": 2.7033, "step": 194110 }, { "epoch": 0.0015872, "grad_norm": 0.9150651693344116, "learning_rate": 1.0116239330189226e-05, "loss": 2.7404, "step": 194120 }, { "epoch": 0.0016128, "grad_norm": 0.7421777248382568, "learning_rate": 1.0115434105742205e-05, "loss": 3.0859, "step": 194130 }, { "epoch": 0.0016384, "grad_norm": 0.7566501498222351, "learning_rate": 1.0114628880546624e-05, "loss": 2.8529, "step": 194140 }, { "epoch": 0.001664, "grad_norm": 0.8494482040405273, "learning_rate": 1.0113823654607706e-05, "loss": 2.8064, "step": 194150 }, { "epoch": 0.0016896, "grad_norm": 0.8465354442596436, "learning_rate": 1.0113018427930673e-05, "loss": 2.8073, "step": 194160 }, { "epoch": 0.0017152, "grad_norm": 0.8197137117385864, "learning_rate": 1.0112213200520746e-05, "loss": 2.8064, "step": 194170 }, { "epoch": 0.0017408, "grad_norm": 0.9861582517623901, "learning_rate": 1.0111407972383144e-05, "loss": 2.8405, "step": 194180 }, { "epoch": 0.0017664, "grad_norm": 0.7878713607788086, "learning_rate": 1.0110602743523092e-05, "loss": 2.9516, "step": 194190 }, { "epoch": 0.001792, "grad_norm": 0.764108419418335, "learning_rate": 1.0109797513945812e-05, "loss": 2.5863, "step": 194200 }, { "epoch": 2.56e-05, "grad_norm": 1.2878761291503906, "learning_rate": 1.0108992283656525e-05, "loss": 2.0344, "step": 194210 }, { "epoch": 5.12e-05, "grad_norm": 1.0304453372955322, "learning_rate": 1.0108187052660445e-05, "loss": 2.557, "step": 194220 }, { "epoch": 7.68e-05, "grad_norm": 0.8023157119750977, "learning_rate": 1.010738182096281e-05, "loss": 2.7497, "step": 194230 }, { "epoch": 0.0001024, "grad_norm": 0.7030252814292908, "learning_rate": 1.0106576588568826e-05, "loss": 2.8469, "step": 194240 }, { "epoch": 0.000128, "grad_norm": 0.8410778641700745, "learning_rate": 1.0105771355483722e-05, "loss": 2.8198, "step": 194250 }, { "epoch": 0.0001536, "grad_norm": 0.8106781840324402, "learning_rate": 1.010496612171272e-05, "loss": 2.7515, "step": 194260 }, { "epoch": 0.0001792, "grad_norm": 0.8722884058952332, "learning_rate": 1.0104160887261044e-05, "loss": 2.6526, "step": 194270 }, { "epoch": 0.0002048, "grad_norm": 0.8761846423149109, "learning_rate": 1.0103355652133908e-05, "loss": 2.9401, "step": 194280 }, { "epoch": 0.0002304, "grad_norm": 0.8274577260017395, "learning_rate": 1.0102550416336538e-05, "loss": 2.6692, "step": 194290 }, { "epoch": 0.000256, "grad_norm": 0.9603460431098938, "learning_rate": 1.0101745179874158e-05, "loss": 2.6252, "step": 194300 }, { "epoch": 0.0002816, "grad_norm": 0.7696784138679504, "learning_rate": 1.0100939942751984e-05, "loss": 2.6647, "step": 194310 }, { "epoch": 0.0003072, "grad_norm": 0.7644719481468201, "learning_rate": 1.0100134704975244e-05, "loss": 2.743, "step": 194320 }, { "epoch": 0.0003328, "grad_norm": 0.7781533598899841, "learning_rate": 1.0099329466549159e-05, "loss": 2.8963, "step": 194330 }, { "epoch": 0.0003584, "grad_norm": 0.7498656511306763, "learning_rate": 1.0098524227478945e-05, "loss": 2.6072, "step": 194340 }, { "epoch": 0.000384, "grad_norm": 0.7790914177894592, "learning_rate": 1.0097718987769829e-05, "loss": 2.5482, "step": 194350 }, { "epoch": 0.0004096, "grad_norm": 0.7772948145866394, "learning_rate": 1.0096913747427033e-05, "loss": 2.7887, "step": 194360 }, { "epoch": 0.0004352, "grad_norm": 0.7359848022460938, "learning_rate": 1.0096108506455775e-05, "loss": 2.8381, "step": 194370 }, { "epoch": 0.0004608, "grad_norm": 0.9443811774253845, "learning_rate": 1.009530326486128e-05, "loss": 2.7029, "step": 194380 }, { "epoch": 0.0004864, "grad_norm": 0.8039208650588989, "learning_rate": 1.0094498022648766e-05, "loss": 2.8112, "step": 194390 }, { "epoch": 0.000512, "grad_norm": 0.7527207732200623, "learning_rate": 1.009369277982346e-05, "loss": 2.5529, "step": 194400 }, { "epoch": 0.0005376, "grad_norm": 0.7540146112442017, "learning_rate": 1.009288753639058e-05, "loss": 2.6587, "step": 194410 }, { "epoch": 0.0005632, "grad_norm": 0.8143194317817688, "learning_rate": 1.0092082292355352e-05, "loss": 2.6887, "step": 194420 }, { "epoch": 0.0005888, "grad_norm": 0.743382453918457, "learning_rate": 1.0091277047722993e-05, "loss": 2.7554, "step": 194430 }, { "epoch": 0.0006144, "grad_norm": 0.6959266066551208, "learning_rate": 1.0090471802498726e-05, "loss": 2.6722, "step": 194440 }, { "epoch": 0.00064, "grad_norm": 0.827103853225708, "learning_rate": 1.0089666556687774e-05, "loss": 2.8528, "step": 194450 }, { "epoch": 0.0006656, "grad_norm": 0.7504157423973083, "learning_rate": 1.0088861310295358e-05, "loss": 3.1792, "step": 194460 }, { "epoch": 0.0006912, "grad_norm": 0.668578565120697, "learning_rate": 1.0088056063326701e-05, "loss": 2.6906, "step": 194470 }, { "epoch": 0.0007168, "grad_norm": 0.7468105554580688, "learning_rate": 1.0087250815787019e-05, "loss": 2.4792, "step": 194480 }, { "epoch": 0.0007424, "grad_norm": 0.9775741100311279, "learning_rate": 1.0086445567681543e-05, "loss": 2.7467, "step": 194490 }, { "epoch": 0.000768, "grad_norm": 0.8252844214439392, "learning_rate": 1.0085640319015491e-05, "loss": 2.6551, "step": 194500 }, { "epoch": 0.0007936, "grad_norm": 0.7429177761077881, "learning_rate": 1.0084835069794083e-05, "loss": 2.7401, "step": 194510 }, { "epoch": 0.0008192, "grad_norm": 1.0515598058700562, "learning_rate": 1.008402982002254e-05, "loss": 2.7627, "step": 194520 }, { "epoch": 0.0008448, "grad_norm": 0.7516338229179382, "learning_rate": 1.0083224569706091e-05, "loss": 2.6505, "step": 194530 }, { "epoch": 0.0008704, "grad_norm": 0.814375102519989, "learning_rate": 1.008241931884995e-05, "loss": 2.6536, "step": 194540 }, { "epoch": 0.000896, "grad_norm": 0.7262110710144043, "learning_rate": 1.0081614067459342e-05, "loss": 2.5935, "step": 194550 }, { "epoch": 0.0009216, "grad_norm": 0.7634857892990112, "learning_rate": 1.0080808815539487e-05, "loss": 2.7063, "step": 194560 }, { "epoch": 0.0009472, "grad_norm": 0.9293407201766968, "learning_rate": 1.0080003563095608e-05, "loss": 2.616, "step": 194570 }, { "epoch": 0.0009728, "grad_norm": 0.716839075088501, "learning_rate": 1.0079198310132929e-05, "loss": 2.614, "step": 194580 }, { "epoch": 0.0009984, "grad_norm": 0.8984509110450745, "learning_rate": 1.0078393056656671e-05, "loss": 2.7778, "step": 194590 }, { "epoch": 0.001024, "grad_norm": 0.8074721693992615, "learning_rate": 1.0077587802672052e-05, "loss": 2.9699, "step": 194600 }, { "epoch": 0.0010496, "grad_norm": 0.8638820052146912, "learning_rate": 1.0076782548184298e-05, "loss": 2.7412, "step": 194610 }, { "epoch": 0.0010752, "grad_norm": 0.8288601040840149, "learning_rate": 1.007597729319863e-05, "loss": 2.7364, "step": 194620 }, { "epoch": 0.0011008, "grad_norm": 0.8492729067802429, "learning_rate": 1.0075172037720271e-05, "loss": 2.8595, "step": 194630 }, { "epoch": 0.0011264, "grad_norm": 0.7395238876342773, "learning_rate": 1.0074366781754439e-05, "loss": 2.6652, "step": 194640 }, { "epoch": 0.001152, "grad_norm": 0.8898137807846069, "learning_rate": 1.0073561525306357e-05, "loss": 2.647, "step": 194650 }, { "epoch": 0.0011776, "grad_norm": 0.7083278298377991, "learning_rate": 1.007275626838125e-05, "loss": 2.9466, "step": 194660 }, { "epoch": 0.0012032, "grad_norm": 0.9014190435409546, "learning_rate": 1.0071951010984337e-05, "loss": 2.8624, "step": 194670 }, { "epoch": 0.0012288, "grad_norm": 0.840323805809021, "learning_rate": 1.0071145753120841e-05, "loss": 3.0418, "step": 194680 }, { "epoch": 0.0012544, "grad_norm": 0.7921732664108276, "learning_rate": 1.0070340494795985e-05, "loss": 2.8069, "step": 194690 }, { "epoch": 0.00128, "grad_norm": 0.6622162461280823, "learning_rate": 1.0069535236014988e-05, "loss": 2.7461, "step": 194700 }, { "epoch": 0.0013056, "grad_norm": 0.8726994395256042, "learning_rate": 1.0068729976783075e-05, "loss": 2.5858, "step": 194710 }, { "epoch": 0.0013312, "grad_norm": 0.9830994606018066, "learning_rate": 1.0067924717105465e-05, "loss": 2.8749, "step": 194720 }, { "epoch": 0.0013568, "grad_norm": 0.8607262969017029, "learning_rate": 1.0067119456987385e-05, "loss": 2.35, "step": 194730 }, { "epoch": 0.0013824, "grad_norm": 1.3246252536773682, "learning_rate": 1.0066314196434046e-05, "loss": 2.9322, "step": 194740 }, { "epoch": 0.001408, "grad_norm": 0.979917585849762, "learning_rate": 1.006550893545068e-05, "loss": 2.8286, "step": 194750 }, { "epoch": 0.0014336, "grad_norm": 0.8007558584213257, "learning_rate": 1.006470367404251e-05, "loss": 2.6776, "step": 194760 }, { "epoch": 0.0014592, "grad_norm": 1.1292941570281982, "learning_rate": 1.006389841221475e-05, "loss": 2.7889, "step": 194770 }, { "epoch": 0.0014848, "grad_norm": 0.7420188188552856, "learning_rate": 1.0063093149972627e-05, "loss": 2.8045, "step": 194780 }, { "epoch": 0.0015104, "grad_norm": 0.8206437230110168, "learning_rate": 1.0062287887321361e-05, "loss": 2.6453, "step": 194790 }, { "epoch": 0.001536, "grad_norm": 0.8690692186355591, "learning_rate": 1.0061482624266177e-05, "loss": 2.3748, "step": 194800 }, { "epoch": 0.0015616, "grad_norm": 0.9020218849182129, "learning_rate": 1.0060677360812292e-05, "loss": 2.8002, "step": 194810 }, { "epoch": 0.0015872, "grad_norm": 0.7540686130523682, "learning_rate": 1.005987209696493e-05, "loss": 2.9744, "step": 194820 }, { "epoch": 0.0016128, "grad_norm": 0.8402876853942871, "learning_rate": 1.0059066832729314e-05, "loss": 2.8591, "step": 194830 }, { "epoch": 0.0016384, "grad_norm": 0.7622684836387634, "learning_rate": 1.0058261568110666e-05, "loss": 2.7429, "step": 194840 }, { "epoch": 0.001664, "grad_norm": 0.8781680464744568, "learning_rate": 1.0057456303114206e-05, "loss": 2.8075, "step": 194850 }, { "epoch": 0.0016896, "grad_norm": 0.8196223378181458, "learning_rate": 1.005665103774516e-05, "loss": 2.8411, "step": 194860 }, { "epoch": 0.0017152, "grad_norm": 0.8650708794593811, "learning_rate": 1.0055845772008743e-05, "loss": 2.7791, "step": 194870 }, { "epoch": 0.0017408, "grad_norm": 0.7013067007064819, "learning_rate": 1.0055040505910183e-05, "loss": 2.8693, "step": 194880 }, { "epoch": 0.0017664, "grad_norm": 0.7919566035270691, "learning_rate": 1.00542352394547e-05, "loss": 3.1091, "step": 194890 }, { "epoch": 0.001792, "grad_norm": 0.7766934633255005, "learning_rate": 1.0053429972647517e-05, "loss": 3.015, "step": 194900 }, { "epoch": 0.0018176, "grad_norm": 0.6967833042144775, "learning_rate": 1.0052624705493854e-05, "loss": 2.5971, "step": 194910 }, { "epoch": 0.0018432, "grad_norm": 0.8880152702331543, "learning_rate": 1.0051819437998932e-05, "loss": 2.4754, "step": 194920 }, { "epoch": 0.0018688, "grad_norm": 0.7082799673080444, "learning_rate": 1.0051014170167976e-05, "loss": 2.4209, "step": 194930 }, { "epoch": 0.0018944, "grad_norm": 0.911296010017395, "learning_rate": 1.0050208902006206e-05, "loss": 2.9725, "step": 194940 }, { "epoch": 0.00192, "grad_norm": 0.8568179607391357, "learning_rate": 1.0049403633518846e-05, "loss": 2.752, "step": 194950 }, { "epoch": 0.0019456, "grad_norm": 0.7474281787872314, "learning_rate": 1.0048598364711117e-05, "loss": 2.9444, "step": 194960 }, { "epoch": 0.0019712, "grad_norm": 0.7801693081855774, "learning_rate": 1.004779309558824e-05, "loss": 3.0056, "step": 194970 }, { "epoch": 0.0019968, "grad_norm": 0.8216150403022766, "learning_rate": 1.0046987826155437e-05, "loss": 2.9916, "step": 194980 }, { "epoch": 0.0020224, "grad_norm": 1.0367228984832764, "learning_rate": 1.004618255641793e-05, "loss": 3.0056, "step": 194990 }, { "epoch": 0.002048, "grad_norm": 0.7192697525024414, "learning_rate": 1.0045377286380942e-05, "loss": 2.7534, "step": 195000 }, { "epoch": 0.0020736, "grad_norm": 0.9381837248802185, "learning_rate": 1.0044572016049693e-05, "loss": 2.699, "step": 195010 }, { "epoch": 0.0020992, "grad_norm": 0.8338245749473572, "learning_rate": 1.0043766745429408e-05, "loss": 2.6447, "step": 195020 }, { "epoch": 0.0021248, "grad_norm": 0.7468827962875366, "learning_rate": 1.0042961474525308e-05, "loss": 2.7198, "step": 195030 }, { "epoch": 0.0021504, "grad_norm": 0.8584609031677246, "learning_rate": 1.0042156203342612e-05, "loss": 2.9779, "step": 195040 }, { "epoch": 0.002176, "grad_norm": 1.0213228464126587, "learning_rate": 1.0041350931886545e-05, "loss": 2.8252, "step": 195050 }, { "epoch": 0.0022016, "grad_norm": 0.7245917916297913, "learning_rate": 1.004054566016233e-05, "loss": 2.9448, "step": 195060 }, { "epoch": 0.0022272, "grad_norm": 0.7614426612854004, "learning_rate": 1.0039740388175185e-05, "loss": 2.9961, "step": 195070 }, { "epoch": 0.0022528, "grad_norm": 1.473052740097046, "learning_rate": 1.0038935115930335e-05, "loss": 2.5326, "step": 195080 }, { "epoch": 0.0022784, "grad_norm": 0.7633165121078491, "learning_rate": 1.0038129843433e-05, "loss": 2.816, "step": 195090 }, { "epoch": 0.002304, "grad_norm": 0.6933164596557617, "learning_rate": 1.0037324570688404e-05, "loss": 2.3644, "step": 195100 }, { "epoch": 0.0023296, "grad_norm": 0.7626866102218628, "learning_rate": 1.0036519297701767e-05, "loss": 2.775, "step": 195110 }, { "epoch": 0.0023552, "grad_norm": 0.7590758204460144, "learning_rate": 1.0035714024478316e-05, "loss": 2.7248, "step": 195120 }, { "epoch": 0.0023808, "grad_norm": 0.7832821011543274, "learning_rate": 1.0034908751023265e-05, "loss": 2.6062, "step": 195130 }, { "epoch": 0.0024064, "grad_norm": 0.8137063384056091, "learning_rate": 1.003410347734184e-05, "loss": 2.729, "step": 195140 }, { "epoch": 0.002432, "grad_norm": 0.7225062847137451, "learning_rate": 1.0033298203439266e-05, "loss": 2.7512, "step": 195150 }, { "epoch": 0.0024576, "grad_norm": 0.8470887541770935, "learning_rate": 1.003249292932076e-05, "loss": 2.8409, "step": 195160 }, { "epoch": 0.0024832, "grad_norm": 0.7481761574745178, "learning_rate": 1.0031687654991547e-05, "loss": 2.7525, "step": 195170 }, { "epoch": 0.0025088, "grad_norm": 0.8439481854438782, "learning_rate": 1.0030882380456844e-05, "loss": 3.0207, "step": 195180 }, { "epoch": 0.0025344, "grad_norm": 0.7536864876747131, "learning_rate": 1.0030077105721883e-05, "loss": 3.0394, "step": 195190 }, { "epoch": 0.00256, "grad_norm": 0.8710933923721313, "learning_rate": 1.0029271830791876e-05, "loss": 2.9014, "step": 195200 }, { "epoch": 0.0025856, "grad_norm": 0.7848872542381287, "learning_rate": 1.002846655567205e-05, "loss": 2.92, "step": 195210 }, { "epoch": 0.0026112, "grad_norm": 1.0752853155136108, "learning_rate": 1.0027661280367627e-05, "loss": 2.952, "step": 195220 }, { "epoch": 0.0026368, "grad_norm": 0.8120469450950623, "learning_rate": 1.0026856004883828e-05, "loss": 2.9421, "step": 195230 }, { "epoch": 0.0026624, "grad_norm": 0.7828359007835388, "learning_rate": 1.0026050729225874e-05, "loss": 2.8088, "step": 195240 }, { "epoch": 0.002688, "grad_norm": 0.7883274555206299, "learning_rate": 1.0025245453398986e-05, "loss": 2.8933, "step": 195250 }, { "epoch": 0.0027136, "grad_norm": 0.780799388885498, "learning_rate": 1.002444017740839e-05, "loss": 2.8642, "step": 195260 }, { "epoch": 0.0027392, "grad_norm": 0.7655680775642395, "learning_rate": 1.0023634901259305e-05, "loss": 2.8993, "step": 195270 }, { "epoch": 0.0027648, "grad_norm": 0.9185817837715149, "learning_rate": 1.0022829624956954e-05, "loss": 2.8134, "step": 195280 }, { "epoch": 0.0027904, "grad_norm": 0.7635614275932312, "learning_rate": 1.0022024348506562e-05, "loss": 2.9537, "step": 195290 }, { "epoch": 0.002816, "grad_norm": 0.7866108417510986, "learning_rate": 1.0021219071913345e-05, "loss": 2.9073, "step": 195300 }, { "epoch": 0.0028416, "grad_norm": 0.7110141515731812, "learning_rate": 1.0020413795182529e-05, "loss": 2.9871, "step": 195310 }, { "epoch": 0.0028672, "grad_norm": 0.7313554883003235, "learning_rate": 1.0019608518319335e-05, "loss": 2.8532, "step": 195320 }, { "epoch": 0.0028928, "grad_norm": 0.7487562298774719, "learning_rate": 1.0018803241328985e-05, "loss": 2.8791, "step": 195330 }, { "epoch": 0.0029184, "grad_norm": 0.7077341675758362, "learning_rate": 1.00179979642167e-05, "loss": 3.0102, "step": 195340 }, { "epoch": 0.002944, "grad_norm": 0.7523992657661438, "learning_rate": 1.0017192686987703e-05, "loss": 3.0584, "step": 195350 }, { "epoch": 0.0029696, "grad_norm": 0.7589243054389954, "learning_rate": 1.0016387409647214e-05, "loss": 3.1745, "step": 195360 }, { "epoch": 0.0029952, "grad_norm": 1.0085302591323853, "learning_rate": 1.001558213220046e-05, "loss": 2.9035, "step": 195370 }, { "epoch": 0.0030208, "grad_norm": 0.755281388759613, "learning_rate": 1.001477685465266e-05, "loss": 2.8395, "step": 195380 }, { "epoch": 0.0030464, "grad_norm": 0.7886793613433838, "learning_rate": 1.0013971577009037e-05, "loss": 2.9832, "step": 195390 }, { "epoch": 0.003072, "grad_norm": 0.721372127532959, "learning_rate": 1.001316629927481e-05, "loss": 2.8492, "step": 195400 }, { "epoch": 0.0030976, "grad_norm": 1.0005522966384888, "learning_rate": 1.0012361021455203e-05, "loss": 2.9772, "step": 195410 }, { "epoch": 0.0031232, "grad_norm": 0.741492509841919, "learning_rate": 1.0011555743555441e-05, "loss": 3.0416, "step": 195420 }, { "epoch": 0.0031488, "grad_norm": 0.7953940629959106, "learning_rate": 1.001075046558074e-05, "loss": 3.0514, "step": 195430 }, { "epoch": 0.0031744, "grad_norm": 0.6974450945854187, "learning_rate": 1.0009945187536324e-05, "loss": 2.9678, "step": 195440 }, { "epoch": 0.0032, "grad_norm": 0.8342363834381104, "learning_rate": 1.0009139909427419e-05, "loss": 2.9749, "step": 195450 }, { "epoch": 0.0032256, "grad_norm": 2.127346992492676, "learning_rate": 1.0008334631259242e-05, "loss": 2.9731, "step": 195460 }, { "epoch": 0.0032512, "grad_norm": 0.9036716818809509, "learning_rate": 1.0007529353037018e-05, "loss": 2.9011, "step": 195470 }, { "epoch": 0.0032768, "grad_norm": 1.013283610343933, "learning_rate": 1.0006724074765968e-05, "loss": 2.9991, "step": 195480 }, { "epoch": 0.0033024, "grad_norm": 0.8012641668319702, "learning_rate": 1.0005918796451317e-05, "loss": 3.0485, "step": 195490 }, { "epoch": 0.003328, "grad_norm": 0.8645431399345398, "learning_rate": 1.0005113518098282e-05, "loss": 2.8274, "step": 195500 }, { "epoch": 0.0033536, "grad_norm": 0.7840139865875244, "learning_rate": 1.0004308239712086e-05, "loss": 2.7911, "step": 195510 }, { "epoch": 0.0033792, "grad_norm": 0.7842152118682861, "learning_rate": 1.0003502961297953e-05, "loss": 2.8078, "step": 195520 }, { "epoch": 0.0034048, "grad_norm": 0.8206632137298584, "learning_rate": 1.0002697682861103e-05, "loss": 2.7358, "step": 195530 }, { "epoch": 0.0034304, "grad_norm": 0.7970391511917114, "learning_rate": 1.000189240440676e-05, "loss": 2.9655, "step": 195540 }, { "epoch": 0.003456, "grad_norm": 2.1902642250061035, "learning_rate": 1.0001087125940148e-05, "loss": 2.6427, "step": 195550 }, { "epoch": 0.0034816, "grad_norm": 0.6768858432769775, "learning_rate": 1.0000281847466482e-05, "loss": 2.7292, "step": 195560 }, { "epoch": 0.0035072, "grad_norm": 1.1707682609558105, "learning_rate": 9.999476568990992e-06, "loss": 3.0581, "step": 195570 }, { "epoch": 0.0035328, "grad_norm": 0.7524734139442444, "learning_rate": 9.998671290518893e-06, "loss": 2.8298, "step": 195580 }, { "epoch": 0.0035584, "grad_norm": 0.836155354976654, "learning_rate": 9.997866012055413e-06, "loss": 3.1236, "step": 195590 }, { "epoch": 0.003584, "grad_norm": 0.8752129077911377, "learning_rate": 9.997060733605771e-06, "loss": 2.8338, "step": 195600 }, { "epoch": 0.0036096, "grad_norm": 0.9364043474197388, "learning_rate": 9.996255455175189e-06, "loss": 3.2206, "step": 195610 }, { "epoch": 0.0036352, "grad_norm": 0.7352332472801208, "learning_rate": 9.99545017676889e-06, "loss": 3.0116, "step": 195620 }, { "epoch": 0.0036608, "grad_norm": 0.8960748314857483, "learning_rate": 9.994644898392093e-06, "loss": 2.9623, "step": 195630 }, { "epoch": 0.0036864, "grad_norm": 0.7203555107116699, "learning_rate": 9.993839620050024e-06, "loss": 2.9183, "step": 195640 }, { "epoch": 0.003712, "grad_norm": 0.8810323476791382, "learning_rate": 9.993034341747905e-06, "loss": 3.098, "step": 195650 }, { "epoch": 0.0037376, "grad_norm": 0.9261757135391235, "learning_rate": 9.99222906349095e-06, "loss": 2.9618, "step": 195660 }, { "epoch": 0.0037632, "grad_norm": 0.8543081283569336, "learning_rate": 9.991423785284391e-06, "loss": 3.052, "step": 195670 }, { "epoch": 0.0037888, "grad_norm": 0.7459955811500549, "learning_rate": 9.99061850713345e-06, "loss": 2.9775, "step": 195680 }, { "epoch": 0.0038144, "grad_norm": 0.7609755992889404, "learning_rate": 9.989813229043344e-06, "loss": 3.1259, "step": 195690 }, { "epoch": 0.00384, "grad_norm": 0.7727983593940735, "learning_rate": 9.989007951019296e-06, "loss": 2.9777, "step": 195700 }, { "epoch": 0.0038656, "grad_norm": 0.7761223912239075, "learning_rate": 9.98820267306653e-06, "loss": 2.9329, "step": 195710 }, { "epoch": 0.0038912, "grad_norm": 0.736716628074646, "learning_rate": 9.987397395190264e-06, "loss": 2.875, "step": 195720 }, { "epoch": 0.0039168, "grad_norm": 0.8211612701416016, "learning_rate": 9.986592117395722e-06, "loss": 2.8919, "step": 195730 }, { "epoch": 0.0039424, "grad_norm": 0.798500657081604, "learning_rate": 9.985786839688128e-06, "loss": 3.0027, "step": 195740 }, { "epoch": 0.003968, "grad_norm": 0.7876646518707275, "learning_rate": 9.984981562072705e-06, "loss": 3.0844, "step": 195750 }, { "epoch": 0.0039936, "grad_norm": 0.9689084887504578, "learning_rate": 9.984176284554669e-06, "loss": 2.7218, "step": 195760 }, { "epoch": 0.0040192, "grad_norm": 0.7988722324371338, "learning_rate": 9.983371007139245e-06, "loss": 3.0403, "step": 195770 }, { "epoch": 0.0040448, "grad_norm": 0.8056862354278564, "learning_rate": 9.982565729831657e-06, "loss": 3.0255, "step": 195780 }, { "epoch": 0.0040704, "grad_norm": 0.7566192746162415, "learning_rate": 9.981760452637128e-06, "loss": 2.8091, "step": 195790 }, { "epoch": 0.004096, "grad_norm": 0.989432692527771, "learning_rate": 9.980955175560875e-06, "loss": 2.8789, "step": 195800 }, { "epoch": 0.0041216, "grad_norm": 0.8143718838691711, "learning_rate": 9.980149898608123e-06, "loss": 2.8033, "step": 195810 }, { "epoch": 0.0041472, "grad_norm": 0.6959323287010193, "learning_rate": 9.979344621784094e-06, "loss": 2.9335, "step": 195820 }, { "epoch": 0.0041728, "grad_norm": 0.8739160299301147, "learning_rate": 9.978539345094006e-06, "loss": 2.9055, "step": 195830 }, { "epoch": 0.0041984, "grad_norm": 0.8245795965194702, "learning_rate": 9.97773406854309e-06, "loss": 2.9701, "step": 195840 }, { "epoch": 0.004224, "grad_norm": 0.7681260704994202, "learning_rate": 9.976928792136558e-06, "loss": 2.8376, "step": 195850 }, { "epoch": 0.0042496, "grad_norm": 0.8139556050300598, "learning_rate": 9.97612351587964e-06, "loss": 2.9061, "step": 195860 }, { "epoch": 0.0042752, "grad_norm": 0.7110723257064819, "learning_rate": 9.975318239777553e-06, "loss": 2.8133, "step": 195870 }, { "epoch": 0.0043008, "grad_norm": 0.786925196647644, "learning_rate": 9.974512963835523e-06, "loss": 3.0634, "step": 195880 }, { "epoch": 0.0043264, "grad_norm": 1.8982681035995483, "learning_rate": 9.973707688058768e-06, "loss": 2.7744, "step": 195890 }, { "epoch": 0.004352, "grad_norm": 0.7746964693069458, "learning_rate": 9.972902412452512e-06, "loss": 2.7995, "step": 195900 }, { "epoch": 0.0043776, "grad_norm": 0.7274700999259949, "learning_rate": 9.972097137021978e-06, "loss": 2.9372, "step": 195910 }, { "epoch": 0.0044032, "grad_norm": 0.918215811252594, "learning_rate": 9.971291861772385e-06, "loss": 2.936, "step": 195920 }, { "epoch": 0.0044288, "grad_norm": 0.7533648014068604, "learning_rate": 9.970486586708955e-06, "loss": 2.751, "step": 195930 }, { "epoch": 0.0044544, "grad_norm": 0.8015456199645996, "learning_rate": 9.969681311836914e-06, "loss": 3.0005, "step": 195940 }, { "epoch": 0.00448, "grad_norm": 1.2395485639572144, "learning_rate": 9.968876037161483e-06, "loss": 2.9284, "step": 195950 }, { "epoch": 0.0045056, "grad_norm": 0.8728375434875488, "learning_rate": 9.96807076268788e-06, "loss": 2.8277, "step": 195960 }, { "epoch": 0.0045312, "grad_norm": 0.7050085663795471, "learning_rate": 9.96726548842133e-06, "loss": 2.7511, "step": 195970 }, { "epoch": 0.0045568, "grad_norm": 0.7515767216682434, "learning_rate": 9.966460214367055e-06, "loss": 3.1668, "step": 195980 }, { "epoch": 0.0045824, "grad_norm": 0.8077532052993774, "learning_rate": 9.965654940530278e-06, "loss": 2.6165, "step": 195990 }, { "epoch": 0.004608, "grad_norm": 1.0198416709899902, "learning_rate": 9.964849666916213e-06, "loss": 2.7603, "step": 196000 }, { "epoch": 0.0046336, "grad_norm": 0.7924866080284119, "learning_rate": 9.964044393530096e-06, "loss": 2.857, "step": 196010 }, { "epoch": 0.0046592, "grad_norm": 0.8286534547805786, "learning_rate": 9.963239120377139e-06, "loss": 2.8256, "step": 196020 }, { "epoch": 0.0046848, "grad_norm": 1.0910521745681763, "learning_rate": 9.962433847462566e-06, "loss": 2.7947, "step": 196030 }, { "epoch": 0.0047104, "grad_norm": 0.9187127947807312, "learning_rate": 9.961628574791602e-06, "loss": 2.8915, "step": 196040 }, { "epoch": 0.004736, "grad_norm": 0.720908522605896, "learning_rate": 9.960823302369465e-06, "loss": 3.081, "step": 196050 }, { "epoch": 0.0047616, "grad_norm": 0.7971982955932617, "learning_rate": 9.960018030201377e-06, "loss": 2.9936, "step": 196060 }, { "epoch": 0.0047872, "grad_norm": 0.6945013403892517, "learning_rate": 9.959212758292562e-06, "loss": 2.9445, "step": 196070 }, { "epoch": 0.0048128, "grad_norm": 0.8380562663078308, "learning_rate": 9.958407486648243e-06, "loss": 2.8292, "step": 196080 }, { "epoch": 0.0048384, "grad_norm": 0.7129026055335999, "learning_rate": 9.95760221527364e-06, "loss": 2.9912, "step": 196090 }, { "epoch": 0.004864, "grad_norm": 0.7432771921157837, "learning_rate": 9.956796944173976e-06, "loss": 3.0056, "step": 196100 }, { "epoch": 0.0048896, "grad_norm": 0.8050894141197205, "learning_rate": 9.955991673354472e-06, "loss": 2.94, "step": 196110 }, { "epoch": 0.0049152, "grad_norm": 0.6792615652084351, "learning_rate": 9.95518640282035e-06, "loss": 2.8944, "step": 196120 }, { "epoch": 0.0049408, "grad_norm": 0.9113470315933228, "learning_rate": 9.95438113257683e-06, "loss": 2.9826, "step": 196130 }, { "epoch": 0.0049664, "grad_norm": 0.6858172416687012, "learning_rate": 9.953575862629139e-06, "loss": 2.9814, "step": 196140 }, { "epoch": 0.004992, "grad_norm": 1.0090932846069336, "learning_rate": 9.952770592982496e-06, "loss": 2.7361, "step": 196150 }, { "epoch": 0.0050176, "grad_norm": 0.7466955184936523, "learning_rate": 9.951965323642122e-06, "loss": 2.7942, "step": 196160 }, { "epoch": 0.0050432, "grad_norm": 0.8174957036972046, "learning_rate": 9.95116005461324e-06, "loss": 2.8173, "step": 196170 }, { "epoch": 0.0050688, "grad_norm": 0.7226668000221252, "learning_rate": 9.95035478590107e-06, "loss": 3.1636, "step": 196180 }, { "epoch": 0.0050944, "grad_norm": 0.872231125831604, "learning_rate": 9.949549517510838e-06, "loss": 3.0547, "step": 196190 }, { "epoch": 0.00512, "grad_norm": 0.953907310962677, "learning_rate": 9.948744249447765e-06, "loss": 2.7291, "step": 196200 }, { "epoch": 0.0051456, "grad_norm": 0.8618961572647095, "learning_rate": 9.947938981717072e-06, "loss": 2.999, "step": 196210 }, { "epoch": 0.0051712, "grad_norm": 0.7076606154441833, "learning_rate": 9.94713371432398e-06, "loss": 2.8108, "step": 196220 }, { "epoch": 0.0051968, "grad_norm": 0.8211856484413147, "learning_rate": 9.946328447273711e-06, "loss": 2.8323, "step": 196230 }, { "epoch": 0.0052224, "grad_norm": 0.9044139981269836, "learning_rate": 9.945523180571491e-06, "loss": 2.9195, "step": 196240 }, { "epoch": 0.005248, "grad_norm": 0.9663256406784058, "learning_rate": 9.944717914222536e-06, "loss": 2.8104, "step": 196250 }, { "epoch": 0.0052736, "grad_norm": 0.7854002118110657, "learning_rate": 9.943912648232071e-06, "loss": 2.9215, "step": 196260 }, { "epoch": 0.0052992, "grad_norm": 0.8044033646583557, "learning_rate": 9.943107382605318e-06, "loss": 2.9585, "step": 196270 }, { "epoch": 0.0053248, "grad_norm": 0.9198217988014221, "learning_rate": 9.942302117347499e-06, "loss": 3.0276, "step": 196280 }, { "epoch": 0.0053504, "grad_norm": 0.8564932942390442, "learning_rate": 9.941496852463835e-06, "loss": 2.9156, "step": 196290 }, { "epoch": 0.005376, "grad_norm": 0.7964183688163757, "learning_rate": 9.940691587959548e-06, "loss": 2.7423, "step": 196300 }, { "epoch": 0.0054016, "grad_norm": 0.739397406578064, "learning_rate": 9.939886323839862e-06, "loss": 2.978, "step": 196310 }, { "epoch": 0.0054272, "grad_norm": 0.6963285803794861, "learning_rate": 9.939081060109995e-06, "loss": 2.9182, "step": 196320 }, { "epoch": 0.0054528, "grad_norm": 1.0387874841690063, "learning_rate": 9.938275796775172e-06, "loss": 2.9598, "step": 196330 }, { "epoch": 0.0054784, "grad_norm": 0.6959365606307983, "learning_rate": 9.937470533840613e-06, "loss": 2.8713, "step": 196340 }, { "epoch": 0.005504, "grad_norm": 0.8836796879768372, "learning_rate": 9.936665271311538e-06, "loss": 2.9029, "step": 196350 }, { "epoch": 0.0055296, "grad_norm": 0.814874529838562, "learning_rate": 9.935860009193175e-06, "loss": 2.9095, "step": 196360 }, { "epoch": 0.0055552, "grad_norm": 0.7478077411651611, "learning_rate": 9.935054747490747e-06, "loss": 2.8417, "step": 196370 }, { "epoch": 0.0055808, "grad_norm": 0.7364027500152588, "learning_rate": 9.934249486209468e-06, "loss": 2.9969, "step": 196380 }, { "epoch": 0.0056064, "grad_norm": 0.7293922305107117, "learning_rate": 9.933444225354563e-06, "loss": 2.9438, "step": 196390 }, { "epoch": 0.005632, "grad_norm": 0.7434118390083313, "learning_rate": 9.932638964931254e-06, "loss": 2.9442, "step": 196400 }, { "epoch": 0.0056576, "grad_norm": 0.7352076172828674, "learning_rate": 9.931833704944767e-06, "loss": 2.9567, "step": 196410 }, { "epoch": 0.0056832, "grad_norm": 0.7610313892364502, "learning_rate": 9.931028445400317e-06, "loss": 3.1982, "step": 196420 }, { "epoch": 0.0057088, "grad_norm": 0.7841603755950928, "learning_rate": 9.93022318630313e-06, "loss": 2.9237, "step": 196430 }, { "epoch": 0.0057344, "grad_norm": 1.0044058561325073, "learning_rate": 9.929417927658429e-06, "loss": 2.7306, "step": 196440 }, { "epoch": 0.00576, "grad_norm": 0.8009125590324402, "learning_rate": 9.92861266947143e-06, "loss": 2.9058, "step": 196450 }, { "epoch": 0.0057856, "grad_norm": 0.7507212162017822, "learning_rate": 9.927807411747362e-06, "loss": 2.9061, "step": 196460 }, { "epoch": 0.0058112, "grad_norm": 0.7271414995193481, "learning_rate": 9.927002154491443e-06, "loss": 2.9127, "step": 196470 }, { "epoch": 0.0058368, "grad_norm": 0.7896319031715393, "learning_rate": 9.926196897708898e-06, "loss": 3.0843, "step": 196480 }, { "epoch": 0.0058624, "grad_norm": 0.8544557690620422, "learning_rate": 9.925391641404942e-06, "loss": 2.8624, "step": 196490 }, { "epoch": 0.005888, "grad_norm": 0.7898181080818176, "learning_rate": 9.924586385584803e-06, "loss": 2.7123, "step": 196500 }, { "epoch": 0.0059136, "grad_norm": 0.971541166305542, "learning_rate": 9.923781130253704e-06, "loss": 2.9422, "step": 196510 }, { "epoch": 0.0059392, "grad_norm": 0.8011928796768188, "learning_rate": 9.922975875416862e-06, "loss": 2.9161, "step": 196520 }, { "epoch": 0.0059648, "grad_norm": 0.8266571760177612, "learning_rate": 9.922170621079498e-06, "loss": 2.8231, "step": 196530 }, { "epoch": 0.0059904, "grad_norm": 0.7642956376075745, "learning_rate": 9.921365367246839e-06, "loss": 2.8627, "step": 196540 }, { "epoch": 0.006016, "grad_norm": 0.831967294216156, "learning_rate": 9.920560113924104e-06, "loss": 3.0385, "step": 196550 }, { "epoch": 0.0060416, "grad_norm": 0.7661146521568298, "learning_rate": 9.919754861116517e-06, "loss": 2.8682, "step": 196560 }, { "epoch": 0.0060672, "grad_norm": 1.0770803689956665, "learning_rate": 9.9189496088293e-06, "loss": 2.8087, "step": 196570 }, { "epoch": 0.0060928, "grad_norm": 0.8277595043182373, "learning_rate": 9.91814435706767e-06, "loss": 3.0185, "step": 196580 }, { "epoch": 0.0061184, "grad_norm": 0.8743789792060852, "learning_rate": 9.917339105836854e-06, "loss": 2.8677, "step": 196590 }, { "epoch": 0.006144, "grad_norm": 0.776174008846283, "learning_rate": 9.916533855142072e-06, "loss": 2.8344, "step": 196600 }, { "epoch": 0.0061696, "grad_norm": 0.8450120687484741, "learning_rate": 9.915728604988547e-06, "loss": 2.8355, "step": 196610 }, { "epoch": 0.0061952, "grad_norm": 0.7286408543586731, "learning_rate": 9.914923355381496e-06, "loss": 2.6189, "step": 196620 }, { "epoch": 0.0062208, "grad_norm": 0.8622093796730042, "learning_rate": 9.914118106326147e-06, "loss": 2.8582, "step": 196630 }, { "epoch": 0.0062464, "grad_norm": 0.8120010495185852, "learning_rate": 9.91331285782772e-06, "loss": 2.8575, "step": 196640 }, { "epoch": 0.006272, "grad_norm": 1.0370107889175415, "learning_rate": 9.912507609891433e-06, "loss": 2.8968, "step": 196650 }, { "epoch": 0.0062976, "grad_norm": 0.9161364436149597, "learning_rate": 9.911702362522513e-06, "loss": 2.5569, "step": 196660 }, { "epoch": 0.0063232, "grad_norm": 0.7825753688812256, "learning_rate": 9.910897115726178e-06, "loss": 3.1688, "step": 196670 }, { "epoch": 0.0063488, "grad_norm": 1.0119656324386597, "learning_rate": 9.910091869507654e-06, "loss": 2.8327, "step": 196680 }, { "epoch": 0.0063744, "grad_norm": 0.885275661945343, "learning_rate": 9.90928662387216e-06, "loss": 2.9587, "step": 196690 }, { "epoch": 0.0064, "grad_norm": 0.8064504861831665, "learning_rate": 9.908481378824913e-06, "loss": 2.832, "step": 196700 }, { "epoch": 0.0064256, "grad_norm": 0.7905579209327698, "learning_rate": 9.907676134371144e-06, "loss": 2.9294, "step": 196710 }, { "epoch": 0.0064512, "grad_norm": 0.9408935308456421, "learning_rate": 9.906870890516071e-06, "loss": 2.9435, "step": 196720 }, { "epoch": 0.0064768, "grad_norm": 0.8025221824645996, "learning_rate": 9.906065647264917e-06, "loss": 2.979, "step": 196730 }, { "epoch": 0.0065024, "grad_norm": 0.8669995069503784, "learning_rate": 9.905260404622901e-06, "loss": 2.9782, "step": 196740 }, { "epoch": 0.006528, "grad_norm": 0.7236800789833069, "learning_rate": 9.904455162595245e-06, "loss": 2.8821, "step": 196750 }, { "epoch": 0.0065536, "grad_norm": 0.777911901473999, "learning_rate": 9.903649921187172e-06, "loss": 3.0048, "step": 196760 }, { "epoch": 0.0065792, "grad_norm": 0.7911083698272705, "learning_rate": 9.902844680403906e-06, "loss": 2.9254, "step": 196770 }, { "epoch": 0.0066048, "grad_norm": 0.8049546480178833, "learning_rate": 9.902039440250664e-06, "loss": 2.9182, "step": 196780 }, { "epoch": 0.0066304, "grad_norm": 0.746060311794281, "learning_rate": 9.90123420073267e-06, "loss": 3.3223, "step": 196790 }, { "epoch": 0.006656, "grad_norm": 0.7671955823898315, "learning_rate": 9.90042896185515e-06, "loss": 2.7446, "step": 196800 }, { "epoch": 0.0066816, "grad_norm": 0.7315250039100647, "learning_rate": 9.899623723623317e-06, "loss": 2.8153, "step": 196810 }, { "epoch": 0.0067072, "grad_norm": 0.9165189862251282, "learning_rate": 9.8988184860424e-06, "loss": 3.0378, "step": 196820 }, { "epoch": 0.0067328, "grad_norm": 0.8908154368400574, "learning_rate": 9.898013249117616e-06, "loss": 3.0557, "step": 196830 }, { "epoch": 0.0067584, "grad_norm": 0.7074089646339417, "learning_rate": 9.897208012854192e-06, "loss": 2.8637, "step": 196840 }, { "epoch": 0.006784, "grad_norm": 0.8771126866340637, "learning_rate": 9.896402777257344e-06, "loss": 3.1776, "step": 196850 }, { "epoch": 0.0068096, "grad_norm": 0.7218257188796997, "learning_rate": 9.895597542332298e-06, "loss": 2.9738, "step": 196860 }, { "epoch": 0.0068352, "grad_norm": 0.763454258441925, "learning_rate": 9.89479230808427e-06, "loss": 2.8084, "step": 196870 }, { "epoch": 0.0068608, "grad_norm": 0.8688923120498657, "learning_rate": 9.89398707451849e-06, "loss": 2.9954, "step": 196880 }, { "epoch": 0.0068864, "grad_norm": 0.7095082998275757, "learning_rate": 9.893181841640175e-06, "loss": 3.0247, "step": 196890 }, { "epoch": 0.006912, "grad_norm": 0.9893020987510681, "learning_rate": 9.892376609454549e-06, "loss": 2.9322, "step": 196900 }, { "epoch": 0.0069376, "grad_norm": 0.7714157104492188, "learning_rate": 9.89157137796683e-06, "loss": 2.9582, "step": 196910 }, { "epoch": 0.0069632, "grad_norm": 0.7779009342193604, "learning_rate": 9.890766147182243e-06, "loss": 2.8679, "step": 196920 }, { "epoch": 0.0069888, "grad_norm": 0.7870602011680603, "learning_rate": 9.88996091710601e-06, "loss": 2.7947, "step": 196930 }, { "epoch": 0.0070144, "grad_norm": 0.7858654856681824, "learning_rate": 9.889155687743348e-06, "loss": 2.7489, "step": 196940 }, { "epoch": 0.00704, "grad_norm": 0.8250150680541992, "learning_rate": 9.888350459099482e-06, "loss": 2.8077, "step": 196950 }, { "epoch": 0.0070656, "grad_norm": 0.8640157580375671, "learning_rate": 9.887545231179635e-06, "loss": 2.9927, "step": 196960 }, { "epoch": 0.0070912, "grad_norm": 0.8506582975387573, "learning_rate": 9.886740003989028e-06, "loss": 2.9372, "step": 196970 }, { "epoch": 0.0071168, "grad_norm": 0.691016435623169, "learning_rate": 9.88593477753288e-06, "loss": 3.1625, "step": 196980 }, { "epoch": 0.0071424, "grad_norm": 0.8517560958862305, "learning_rate": 9.885129551816416e-06, "loss": 2.8669, "step": 196990 }, { "epoch": 0.007168, "grad_norm": 0.7660866379737854, "learning_rate": 9.884324326844858e-06, "loss": 2.8176, "step": 197000 }, { "epoch": 0.0071936, "grad_norm": 0.7617883086204529, "learning_rate": 9.883519102623423e-06, "loss": 2.8845, "step": 197010 }, { "epoch": 0.0072192, "grad_norm": 0.899026095867157, "learning_rate": 9.882713879157336e-06, "loss": 3.0601, "step": 197020 }, { "epoch": 0.0072448, "grad_norm": 0.7787336707115173, "learning_rate": 9.881908656451818e-06, "loss": 2.9403, "step": 197030 }, { "epoch": 0.0072704, "grad_norm": 0.8000504970550537, "learning_rate": 9.881103434512095e-06, "loss": 2.7233, "step": 197040 }, { "epoch": 0.007296, "grad_norm": 0.8312419056892395, "learning_rate": 9.880298213343376e-06, "loss": 2.8728, "step": 197050 }, { "epoch": 0.0073216, "grad_norm": 0.8003359436988831, "learning_rate": 9.879492992950899e-06, "loss": 3.1434, "step": 197060 }, { "epoch": 0.0073472, "grad_norm": 0.7849857211112976, "learning_rate": 9.878687773339875e-06, "loss": 2.9968, "step": 197070 }, { "epoch": 0.0073728, "grad_norm": 0.7540002465248108, "learning_rate": 9.87788255451553e-06, "loss": 3.0305, "step": 197080 }, { "epoch": 0.0073984, "grad_norm": 0.8792044520378113, "learning_rate": 9.877077336483082e-06, "loss": 2.8461, "step": 197090 }, { "epoch": 0.007424, "grad_norm": 0.8068345189094543, "learning_rate": 9.876272119247759e-06, "loss": 2.9402, "step": 197100 }, { "epoch": 0.0074496, "grad_norm": 0.7804010510444641, "learning_rate": 9.875466902814774e-06, "loss": 2.8333, "step": 197110 }, { "epoch": 0.0074752, "grad_norm": 0.7785695195198059, "learning_rate": 9.874661687189353e-06, "loss": 2.7665, "step": 197120 }, { "epoch": 0.0075008, "grad_norm": 0.7642216682434082, "learning_rate": 9.873856472376722e-06, "loss": 2.9584, "step": 197130 }, { "epoch": 0.0075264, "grad_norm": 0.9218984246253967, "learning_rate": 9.873051258382096e-06, "loss": 3.0923, "step": 197140 }, { "epoch": 0.007552, "grad_norm": 0.832014262676239, "learning_rate": 9.872246045210698e-06, "loss": 2.828, "step": 197150 }, { "epoch": 0.0075776, "grad_norm": 0.8143579959869385, "learning_rate": 9.87144083286775e-06, "loss": 2.9642, "step": 197160 }, { "epoch": 0.0076032, "grad_norm": 0.9114791750907898, "learning_rate": 9.870635621358478e-06, "loss": 2.8397, "step": 197170 }, { "epoch": 0.0076288, "grad_norm": 0.8082394003868103, "learning_rate": 9.869830410688096e-06, "loss": 3.0246, "step": 197180 }, { "epoch": 0.0076544, "grad_norm": 0.9235893487930298, "learning_rate": 9.869025200861829e-06, "loss": 3.0147, "step": 197190 }, { "epoch": 0.00768, "grad_norm": 0.87860107421875, "learning_rate": 9.8682199918849e-06, "loss": 2.9205, "step": 197200 }, { "epoch": 0.0077056, "grad_norm": 1.029285192489624, "learning_rate": 9.867414783762528e-06, "loss": 3.0365, "step": 197210 }, { "epoch": 0.0077312, "grad_norm": 0.7841971516609192, "learning_rate": 9.866609576499935e-06, "loss": 2.7614, "step": 197220 }, { "epoch": 0.0077568, "grad_norm": 0.7387968301773071, "learning_rate": 9.865804370102346e-06, "loss": 2.8978, "step": 197230 }, { "epoch": 0.0077824, "grad_norm": 0.798271656036377, "learning_rate": 9.864999164574978e-06, "loss": 2.9699, "step": 197240 }, { "epoch": 0.007808, "grad_norm": 0.8499426245689392, "learning_rate": 9.864193959923056e-06, "loss": 2.9245, "step": 197250 }, { "epoch": 0.0078336, "grad_norm": 0.8528083562850952, "learning_rate": 9.8633887561518e-06, "loss": 2.9558, "step": 197260 }, { "epoch": 0.0078592, "grad_norm": 0.7837406992912292, "learning_rate": 9.862583553266432e-06, "loss": 2.8632, "step": 197270 }, { "epoch": 0.0078848, "grad_norm": 0.7944239377975464, "learning_rate": 9.861778351272172e-06, "loss": 3.0081, "step": 197280 }, { "epoch": 0.0079104, "grad_norm": 0.7944080829620361, "learning_rate": 9.860973150174243e-06, "loss": 2.9342, "step": 197290 }, { "epoch": 0.007936, "grad_norm": 0.7825566530227661, "learning_rate": 9.860167949977868e-06, "loss": 2.7323, "step": 197300 }, { "epoch": 0.0079616, "grad_norm": 0.7394271492958069, "learning_rate": 9.859362750688264e-06, "loss": 2.9259, "step": 197310 }, { "epoch": 0.0079872, "grad_norm": 0.8047328591346741, "learning_rate": 9.858557552310655e-06, "loss": 2.9992, "step": 197320 }, { "epoch": 0.0080128, "grad_norm": 0.7421709299087524, "learning_rate": 9.857752354850267e-06, "loss": 2.9833, "step": 197330 }, { "epoch": 0.0080384, "grad_norm": 0.819723904132843, "learning_rate": 9.856947158312314e-06, "loss": 2.8113, "step": 197340 }, { "epoch": 0.008064, "grad_norm": 0.855335533618927, "learning_rate": 9.856141962702019e-06, "loss": 2.8062, "step": 197350 }, { "epoch": 0.0080896, "grad_norm": 0.9520034193992615, "learning_rate": 9.855336768024608e-06, "loss": 2.8285, "step": 197360 }, { "epoch": 0.0081152, "grad_norm": 0.8660393953323364, "learning_rate": 9.854531574285299e-06, "loss": 2.6802, "step": 197370 }, { "epoch": 0.0081408, "grad_norm": 0.7592372298240662, "learning_rate": 9.853726381489314e-06, "loss": 2.936, "step": 197380 }, { "epoch": 0.0081664, "grad_norm": 0.7478090524673462, "learning_rate": 9.852921189641874e-06, "loss": 3.0459, "step": 197390 }, { "epoch": 0.008192, "grad_norm": 1.0415065288543701, "learning_rate": 9.852115998748198e-06, "loss": 2.6421, "step": 197400 }, { "epoch": 0.0082176, "grad_norm": 0.7861488461494446, "learning_rate": 9.851310808813512e-06, "loss": 2.9866, "step": 197410 }, { "epoch": 0.0082432, "grad_norm": 0.7515959739685059, "learning_rate": 9.85050561984304e-06, "loss": 3.0501, "step": 197420 }, { "epoch": 0.0082688, "grad_norm": 0.8409316539764404, "learning_rate": 9.849700431841996e-06, "loss": 3.0374, "step": 197430 }, { "epoch": 0.0082944, "grad_norm": 0.7707468271255493, "learning_rate": 9.848895244815605e-06, "loss": 2.8011, "step": 197440 }, { "epoch": 0.00832, "grad_norm": 0.8195273280143738, "learning_rate": 9.848090058769088e-06, "loss": 3.1705, "step": 197450 }, { "epoch": 0.0083456, "grad_norm": 0.7831510901451111, "learning_rate": 9.847284873707669e-06, "loss": 3.0714, "step": 197460 }, { "epoch": 0.0083712, "grad_norm": 0.7802629470825195, "learning_rate": 9.846479689636566e-06, "loss": 2.9633, "step": 197470 }, { "epoch": 0.0083968, "grad_norm": 2.146521806716919, "learning_rate": 9.845674506561e-06, "loss": 2.9498, "step": 197480 }, { "epoch": 0.0084224, "grad_norm": 0.8772639036178589, "learning_rate": 9.844869324486196e-06, "loss": 3.1256, "step": 197490 }, { "epoch": 0.008448, "grad_norm": 0.7791375517845154, "learning_rate": 9.844064143417372e-06, "loss": 2.906, "step": 197500 }, { "epoch": 0.0084736, "grad_norm": 0.8335960507392883, "learning_rate": 9.843258963359749e-06, "loss": 2.7778, "step": 197510 }, { "epoch": 0.0084992, "grad_norm": 0.8258557915687561, "learning_rate": 9.84245378431855e-06, "loss": 2.6944, "step": 197520 }, { "epoch": 0.0085248, "grad_norm": 0.6830934882164001, "learning_rate": 9.841648606299e-06, "loss": 2.8863, "step": 197530 }, { "epoch": 0.0085504, "grad_norm": 0.7855957746505737, "learning_rate": 9.840843429306313e-06, "loss": 2.9448, "step": 197540 }, { "epoch": 0.008576, "grad_norm": 0.8021692633628845, "learning_rate": 9.840038253345716e-06, "loss": 2.9396, "step": 197550 }, { "epoch": 0.0086016, "grad_norm": 0.796389102935791, "learning_rate": 9.839233078422428e-06, "loss": 3.0693, "step": 197560 }, { "epoch": 0.0086272, "grad_norm": 1.6305630207061768, "learning_rate": 9.838427904541668e-06, "loss": 2.8667, "step": 197570 }, { "epoch": 0.0086528, "grad_norm": 0.8143199682235718, "learning_rate": 9.837622731708662e-06, "loss": 3.0296, "step": 197580 }, { "epoch": 0.0086784, "grad_norm": 0.8759089112281799, "learning_rate": 9.836817559928631e-06, "loss": 2.8989, "step": 197590 }, { "epoch": 0.008704, "grad_norm": 0.8075157999992371, "learning_rate": 9.836012389206794e-06, "loss": 2.8178, "step": 197600 }, { "epoch": 2.56e-05, "grad_norm": 0.817192018032074, "learning_rate": 9.835207219548373e-06, "loss": 2.3845, "step": 197610 }, { "epoch": 5.12e-05, "grad_norm": 0.7231879830360413, "learning_rate": 9.83440205095859e-06, "loss": 2.7891, "step": 197620 }, { "epoch": 7.68e-05, "grad_norm": 0.8414174914360046, "learning_rate": 9.833596883442666e-06, "loss": 2.8076, "step": 197630 }, { "epoch": 0.0001024, "grad_norm": 0.6605185866355896, "learning_rate": 9.83279171700582e-06, "loss": 2.8883, "step": 197640 }, { "epoch": 0.000128, "grad_norm": 0.7888829112052917, "learning_rate": 9.831986551653275e-06, "loss": 2.8582, "step": 197650 }, { "epoch": 0.0001536, "grad_norm": 1.2378754615783691, "learning_rate": 9.831181387390256e-06, "loss": 2.6869, "step": 197660 }, { "epoch": 0.0001792, "grad_norm": 0.9827542304992676, "learning_rate": 9.830376224221979e-06, "loss": 2.6695, "step": 197670 }, { "epoch": 0.0002048, "grad_norm": 0.7613958716392517, "learning_rate": 9.829571062153665e-06, "loss": 2.5635, "step": 197680 }, { "epoch": 0.0002304, "grad_norm": 0.7176100611686707, "learning_rate": 9.82876590119054e-06, "loss": 2.8058, "step": 197690 }, { "epoch": 0.000256, "grad_norm": 0.7856597900390625, "learning_rate": 9.827960741337823e-06, "loss": 2.9127, "step": 197700 }, { "epoch": 0.0002816, "grad_norm": 0.8141179084777832, "learning_rate": 9.827155582600733e-06, "loss": 2.7779, "step": 197710 }, { "epoch": 0.0003072, "grad_norm": 0.7382766604423523, "learning_rate": 9.826350424984493e-06, "loss": 2.5346, "step": 197720 }, { "epoch": 0.0003328, "grad_norm": 0.9102888703346252, "learning_rate": 9.825545268494327e-06, "loss": 2.9441, "step": 197730 }, { "epoch": 0.0003584, "grad_norm": 0.7735703587532043, "learning_rate": 9.824740113135447e-06, "loss": 3.0094, "step": 197740 }, { "epoch": 0.000384, "grad_norm": 0.7594502568244934, "learning_rate": 9.823934958913088e-06, "loss": 2.6359, "step": 197750 }, { "epoch": 0.0004096, "grad_norm": 0.8277096748352051, "learning_rate": 9.82312980583246e-06, "loss": 2.6048, "step": 197760 }, { "epoch": 0.0004352, "grad_norm": 0.783569872379303, "learning_rate": 9.82232465389879e-06, "loss": 2.9342, "step": 197770 }, { "epoch": 0.0004608, "grad_norm": 0.7208108901977539, "learning_rate": 9.821519503117295e-06, "loss": 2.6756, "step": 197780 }, { "epoch": 0.0004864, "grad_norm": 0.8305293321609497, "learning_rate": 9.820714353493202e-06, "loss": 2.8921, "step": 197790 }, { "epoch": 0.000512, "grad_norm": 0.7743400931358337, "learning_rate": 9.819909205031728e-06, "loss": 2.6404, "step": 197800 }, { "epoch": 0.0005376, "grad_norm": 1.0389906167984009, "learning_rate": 9.819104057738094e-06, "loss": 2.707, "step": 197810 }, { "epoch": 0.0005632, "grad_norm": 0.7515581846237183, "learning_rate": 9.818298911617524e-06, "loss": 2.622, "step": 197820 }, { "epoch": 0.0005888, "grad_norm": 0.891418993473053, "learning_rate": 9.817493766675236e-06, "loss": 2.6237, "step": 197830 }, { "epoch": 0.0006144, "grad_norm": 0.820554256439209, "learning_rate": 9.816688622916452e-06, "loss": 2.7597, "step": 197840 }, { "epoch": 0.00064, "grad_norm": 0.794388473033905, "learning_rate": 9.815883480346391e-06, "loss": 2.6268, "step": 197850 }, { "epoch": 0.0006656, "grad_norm": 0.9856070280075073, "learning_rate": 9.815078338970282e-06, "loss": 2.7147, "step": 197860 }, { "epoch": 0.0006912, "grad_norm": 0.7649460434913635, "learning_rate": 9.814273198793338e-06, "loss": 3.0996, "step": 197870 }, { "epoch": 0.0007168, "grad_norm": 0.8265355825424194, "learning_rate": 9.813468059820783e-06, "loss": 2.9929, "step": 197880 }, { "epoch": 0.0007424, "grad_norm": 1.1932448148727417, "learning_rate": 9.81266292205784e-06, "loss": 2.9408, "step": 197890 }, { "epoch": 0.000768, "grad_norm": 0.7805601358413696, "learning_rate": 9.811857785509726e-06, "loss": 2.5147, "step": 197900 }, { "epoch": 0.0007936, "grad_norm": 0.7796450853347778, "learning_rate": 9.811052650181664e-06, "loss": 2.7185, "step": 197910 }, { "epoch": 0.0008192, "grad_norm": 0.7486382126808167, "learning_rate": 9.810247516078874e-06, "loss": 2.6452, "step": 197920 }, { "epoch": 0.0008448, "grad_norm": 0.798564076423645, "learning_rate": 9.809442383206578e-06, "loss": 2.6911, "step": 197930 }, { "epoch": 0.0008704, "grad_norm": 0.7366716265678406, "learning_rate": 9.80863725157e-06, "loss": 2.7099, "step": 197940 }, { "epoch": 0.000896, "grad_norm": 0.915050208568573, "learning_rate": 9.80783212117436e-06, "loss": 2.5809, "step": 197950 }, { "epoch": 0.0009216, "grad_norm": 0.9122225642204285, "learning_rate": 9.807026992024876e-06, "loss": 2.8059, "step": 197960 }, { "epoch": 0.0009472, "grad_norm": 1.03326416015625, "learning_rate": 9.806221864126772e-06, "loss": 2.846, "step": 197970 }, { "epoch": 0.0009728, "grad_norm": 0.7487413883209229, "learning_rate": 9.805416737485265e-06, "loss": 2.7868, "step": 197980 }, { "epoch": 0.0009984, "grad_norm": 0.8387460112571716, "learning_rate": 9.804611612105582e-06, "loss": 2.7678, "step": 197990 }, { "epoch": 0.001024, "grad_norm": 0.7743194103240967, "learning_rate": 9.803806487992939e-06, "loss": 2.4251, "step": 198000 }, { "epoch": 0.0010496, "grad_norm": 0.7663047909736633, "learning_rate": 9.803001365152558e-06, "loss": 2.6942, "step": 198010 }, { "epoch": 0.0010752, "grad_norm": 0.8974171876907349, "learning_rate": 9.802196243589664e-06, "loss": 2.9521, "step": 198020 }, { "epoch": 0.0011008, "grad_norm": 0.9683269262313843, "learning_rate": 9.801391123309472e-06, "loss": 2.7211, "step": 198030 }, { "epoch": 0.0011264, "grad_norm": 0.6861109733581543, "learning_rate": 9.800586004317206e-06, "loss": 2.8909, "step": 198040 }, { "epoch": 0.001152, "grad_norm": 0.8117058277130127, "learning_rate": 9.799780886618086e-06, "loss": 2.6927, "step": 198050 }, { "epoch": 0.0011776, "grad_norm": 0.7854422330856323, "learning_rate": 9.798975770217336e-06, "loss": 2.9258, "step": 198060 }, { "epoch": 0.0012032, "grad_norm": 0.8929717540740967, "learning_rate": 9.798170655120172e-06, "loss": 2.8419, "step": 198070 }, { "epoch": 0.0012288, "grad_norm": 0.7480575442314148, "learning_rate": 9.797365541331819e-06, "loss": 2.8206, "step": 198080 }, { "epoch": 0.0012544, "grad_norm": 0.7919853925704956, "learning_rate": 9.796560428857495e-06, "loss": 2.8553, "step": 198090 }, { "epoch": 0.00128, "grad_norm": 0.664457380771637, "learning_rate": 9.795755317702421e-06, "loss": 2.6349, "step": 198100 }, { "epoch": 0.0013056, "grad_norm": 0.8104661107063293, "learning_rate": 9.794950207871826e-06, "loss": 2.7891, "step": 198110 }, { "epoch": 0.0013312, "grad_norm": 0.8731836676597595, "learning_rate": 9.79414509937092e-06, "loss": 2.5805, "step": 198120 }, { "epoch": 0.0013568, "grad_norm": 0.759494423866272, "learning_rate": 9.79333999220493e-06, "loss": 2.8441, "step": 198130 }, { "epoch": 0.0013824, "grad_norm": 0.8355826735496521, "learning_rate": 9.792534886379074e-06, "loss": 2.7165, "step": 198140 }, { "epoch": 0.001408, "grad_norm": 0.7949654459953308, "learning_rate": 9.791729781898578e-06, "loss": 2.699, "step": 198150 }, { "epoch": 0.0014336, "grad_norm": 0.7783393859863281, "learning_rate": 9.790924678768653e-06, "loss": 2.8048, "step": 198160 }, { "epoch": 0.0014592, "grad_norm": 0.9083520770072937, "learning_rate": 9.790119576994529e-06, "loss": 2.8726, "step": 198170 }, { "epoch": 0.0014848, "grad_norm": 0.7956222295761108, "learning_rate": 9.789314476581424e-06, "loss": 2.528, "step": 198180 }, { "epoch": 0.0015104, "grad_norm": 0.7867828011512756, "learning_rate": 9.788509377534559e-06, "loss": 2.4561, "step": 198190 }, { "epoch": 0.001536, "grad_norm": 0.8766977190971375, "learning_rate": 9.787704279859153e-06, "loss": 2.6508, "step": 198200 }, { "epoch": 0.0015616, "grad_norm": 0.8062736392021179, "learning_rate": 9.786899183560427e-06, "loss": 3.015, "step": 198210 }, { "epoch": 0.0015872, "grad_norm": 0.7777870297431946, "learning_rate": 9.786094088643607e-06, "loss": 2.9901, "step": 198220 }, { "epoch": 0.0016128, "grad_norm": 0.7546530961990356, "learning_rate": 9.785288995113908e-06, "loss": 2.7928, "step": 198230 }, { "epoch": 0.0016384, "grad_norm": 0.7989169955253601, "learning_rate": 9.784483902976552e-06, "loss": 2.7653, "step": 198240 }, { "epoch": 0.001664, "grad_norm": 0.7348253726959229, "learning_rate": 9.783678812236762e-06, "loss": 2.7615, "step": 198250 }, { "epoch": 0.0016896, "grad_norm": 0.9070345759391785, "learning_rate": 9.782873722899758e-06, "loss": 2.9407, "step": 198260 }, { "epoch": 0.0017152, "grad_norm": 0.6850849390029907, "learning_rate": 9.782068634970754e-06, "loss": 2.7446, "step": 198270 }, { "epoch": 0.0017408, "grad_norm": 0.691281795501709, "learning_rate": 9.781263548454985e-06, "loss": 2.7905, "step": 198280 }, { "epoch": 0.0017664, "grad_norm": 0.9924920201301575, "learning_rate": 9.78045846335766e-06, "loss": 2.9495, "step": 198290 }, { "epoch": 0.001792, "grad_norm": 0.7171319127082825, "learning_rate": 9.779653379684003e-06, "loss": 2.6378, "step": 198300 }, { "epoch": 0.0018176, "grad_norm": 0.7639375329017639, "learning_rate": 9.778848297439238e-06, "loss": 2.8812, "step": 198310 }, { "epoch": 0.0018432, "grad_norm": 0.6670176982879639, "learning_rate": 9.778043216628581e-06, "loss": 2.5652, "step": 198320 }, { "epoch": 0.0018688, "grad_norm": 0.7346078753471375, "learning_rate": 9.777238137257255e-06, "loss": 2.7443, "step": 198330 }, { "epoch": 0.0018944, "grad_norm": 0.7938486337661743, "learning_rate": 9.77643305933048e-06, "loss": 2.7972, "step": 198340 }, { "epoch": 0.00192, "grad_norm": 0.7673615217208862, "learning_rate": 9.77562798285348e-06, "loss": 3.2544, "step": 198350 }, { "epoch": 0.0019456, "grad_norm": 0.777906060218811, "learning_rate": 9.77482290783147e-06, "loss": 2.7586, "step": 198360 }, { "epoch": 0.0019712, "grad_norm": 0.6844608783721924, "learning_rate": 9.774017834269675e-06, "loss": 2.4034, "step": 198370 }, { "epoch": 0.0019968, "grad_norm": 0.7388970851898193, "learning_rate": 9.773212762173313e-06, "loss": 2.7004, "step": 198380 }, { "epoch": 0.0020224, "grad_norm": 0.7615099549293518, "learning_rate": 9.77240769154761e-06, "loss": 2.7862, "step": 198390 }, { "epoch": 0.002048, "grad_norm": 1.0200568437576294, "learning_rate": 9.771602622397777e-06, "loss": 2.7963, "step": 198400 }, { "epoch": 0.0020736, "grad_norm": 0.7977637648582458, "learning_rate": 9.770797554729043e-06, "loss": 2.829, "step": 198410 }, { "epoch": 0.0020992, "grad_norm": 1.0716300010681152, "learning_rate": 9.769992488546626e-06, "loss": 2.7081, "step": 198420 }, { "epoch": 0.0021248, "grad_norm": 0.7747970223426819, "learning_rate": 9.769187423855747e-06, "loss": 2.9064, "step": 198430 }, { "epoch": 0.0021504, "grad_norm": 0.862106442451477, "learning_rate": 9.768382360661622e-06, "loss": 2.8695, "step": 198440 }, { "epoch": 0.002176, "grad_norm": 1.049422025680542, "learning_rate": 9.767577298969479e-06, "loss": 2.985, "step": 198450 }, { "epoch": 0.0022016, "grad_norm": 0.7505191564559937, "learning_rate": 9.766772238784535e-06, "loss": 2.9532, "step": 198460 }, { "epoch": 0.0022272, "grad_norm": 0.7072507739067078, "learning_rate": 9.765967180112011e-06, "loss": 2.8672, "step": 198470 }, { "epoch": 0.0022528, "grad_norm": 0.7690965533256531, "learning_rate": 9.765162122957131e-06, "loss": 3.1511, "step": 198480 }, { "epoch": 0.0022784, "grad_norm": 0.797264575958252, "learning_rate": 9.76435706732511e-06, "loss": 2.908, "step": 198490 }, { "epoch": 0.002304, "grad_norm": 0.7878480553627014, "learning_rate": 9.763552013221169e-06, "loss": 2.8405, "step": 198500 }, { "epoch": 2.56e-05, "grad_norm": 1.3420252799987793, "learning_rate": 9.762746960650533e-06, "loss": 1.4183, "step": 198510 }, { "epoch": 5.12e-05, "grad_norm": 0.7670003175735474, "learning_rate": 9.761941909618419e-06, "loss": 2.7218, "step": 198520 }, { "epoch": 7.68e-05, "grad_norm": 0.7777345776557922, "learning_rate": 9.761136860130047e-06, "loss": 2.7906, "step": 198530 }, { "epoch": 0.0001024, "grad_norm": 0.7184532284736633, "learning_rate": 9.76033181219064e-06, "loss": 2.7868, "step": 198540 }, { "epoch": 0.000128, "grad_norm": 0.9882280826568604, "learning_rate": 9.75952676580542e-06, "loss": 2.8249, "step": 198550 }, { "epoch": 0.0001536, "grad_norm": 0.8261620402336121, "learning_rate": 9.7587217209796e-06, "loss": 2.7119, "step": 198560 }, { "epoch": 0.0001792, "grad_norm": 0.9708413481712341, "learning_rate": 9.757916677718409e-06, "loss": 2.7968, "step": 198570 }, { "epoch": 0.0002048, "grad_norm": 0.8511148691177368, "learning_rate": 9.757111636027064e-06, "loss": 2.7968, "step": 198580 }, { "epoch": 0.0002304, "grad_norm": 0.8160766959190369, "learning_rate": 9.756306595910785e-06, "loss": 2.767, "step": 198590 }, { "epoch": 0.000256, "grad_norm": 0.6948283314704895, "learning_rate": 9.755501557374792e-06, "loss": 2.7451, "step": 198600 }, { "epoch": 0.0002816, "grad_norm": 0.825191855430603, "learning_rate": 9.754696520424304e-06, "loss": 2.7455, "step": 198610 }, { "epoch": 0.0003072, "grad_norm": 0.7538373470306396, "learning_rate": 9.753891485064546e-06, "loss": 2.578, "step": 198620 }, { "epoch": 0.0003328, "grad_norm": 0.8025709390640259, "learning_rate": 9.753086451300736e-06, "loss": 2.8494, "step": 198630 }, { "epoch": 0.0003584, "grad_norm": 1.1761680841445923, "learning_rate": 9.752281419138099e-06, "loss": 2.6287, "step": 198640 }, { "epoch": 0.000384, "grad_norm": 0.7493432760238647, "learning_rate": 9.751476388581847e-06, "loss": 2.6229, "step": 198650 }, { "epoch": 0.0004096, "grad_norm": 0.8647481799125671, "learning_rate": 9.750671359637206e-06, "loss": 2.8715, "step": 198660 }, { "epoch": 0.0004352, "grad_norm": 0.7600035071372986, "learning_rate": 9.749866332309394e-06, "loss": 2.7701, "step": 198670 }, { "epoch": 0.0004608, "grad_norm": 0.9080729484558105, "learning_rate": 9.749061306603634e-06, "loss": 2.6116, "step": 198680 }, { "epoch": 0.0004864, "grad_norm": 0.8463791608810425, "learning_rate": 9.748256282525143e-06, "loss": 2.8582, "step": 198690 }, { "epoch": 0.000512, "grad_norm": 0.7244962453842163, "learning_rate": 9.747451260079144e-06, "loss": 2.5475, "step": 198700 }, { "epoch": 0.0005376, "grad_norm": 0.7886419892311096, "learning_rate": 9.746646239270858e-06, "loss": 2.7379, "step": 198710 }, { "epoch": 0.0005632, "grad_norm": 0.9097887277603149, "learning_rate": 9.745841220105502e-06, "loss": 2.6271, "step": 198720 }, { "epoch": 0.0005888, "grad_norm": 0.8670911192893982, "learning_rate": 9.745036202588298e-06, "loss": 2.69, "step": 198730 }, { "epoch": 0.0006144, "grad_norm": 0.7163254022598267, "learning_rate": 9.744231186724465e-06, "loss": 2.5684, "step": 198740 }, { "epoch": 0.00064, "grad_norm": 0.9201299548149109, "learning_rate": 9.743426172519229e-06, "loss": 2.695, "step": 198750 }, { "epoch": 0.0006656, "grad_norm": 0.8126393556594849, "learning_rate": 9.742621159977803e-06, "loss": 2.6713, "step": 198760 }, { "epoch": 0.0006912, "grad_norm": 0.7589896321296692, "learning_rate": 9.74181614910541e-06, "loss": 2.7529, "step": 198770 }, { "epoch": 0.0007168, "grad_norm": 0.8959205746650696, "learning_rate": 9.741011139907273e-06, "loss": 2.8093, "step": 198780 }, { "epoch": 0.0007424, "grad_norm": 0.9615406394004822, "learning_rate": 9.740206132388604e-06, "loss": 2.8817, "step": 198790 }, { "epoch": 0.000768, "grad_norm": 0.714184582233429, "learning_rate": 9.739401126554635e-06, "loss": 3.0003, "step": 198800 }, { "epoch": 0.0007936, "grad_norm": 0.7098860740661621, "learning_rate": 9.73859612241058e-06, "loss": 2.8287, "step": 198810 }, { "epoch": 0.0008192, "grad_norm": 0.7663411498069763, "learning_rate": 9.737791119961657e-06, "loss": 2.7568, "step": 198820 }, { "epoch": 0.0008448, "grad_norm": 0.8541398048400879, "learning_rate": 9.736986119213091e-06, "loss": 2.6771, "step": 198830 }, { "epoch": 0.0008704, "grad_norm": 0.7831572890281677, "learning_rate": 9.736181120170102e-06, "loss": 2.7137, "step": 198840 }, { "epoch": 0.000896, "grad_norm": 0.7527443170547485, "learning_rate": 9.735376122837905e-06, "loss": 2.6313, "step": 198850 }, { "epoch": 0.0009216, "grad_norm": 0.7399323582649231, "learning_rate": 9.734571127221723e-06, "loss": 3.3531, "step": 198860 }, { "epoch": 0.0009472, "grad_norm": 0.6774531602859497, "learning_rate": 9.733766133326779e-06, "loss": 2.4065, "step": 198870 }, { "epoch": 0.0009728, "grad_norm": 0.9468609094619751, "learning_rate": 9.732961141158291e-06, "loss": 2.8855, "step": 198880 }, { "epoch": 0.0009984, "grad_norm": 0.7719103693962097, "learning_rate": 9.732156150721479e-06, "loss": 2.9359, "step": 198890 }, { "epoch": 0.001024, "grad_norm": 1.068347454071045, "learning_rate": 9.731351162021562e-06, "loss": 3.041, "step": 198900 }, { "epoch": 0.0010496, "grad_norm": 0.8519234657287598, "learning_rate": 9.730546175063764e-06, "loss": 2.891, "step": 198910 }, { "epoch": 0.0010752, "grad_norm": 0.7622858881950378, "learning_rate": 9.729741189853299e-06, "loss": 2.6671, "step": 198920 }, { "epoch": 0.0011008, "grad_norm": 0.8404979109764099, "learning_rate": 9.728936206395391e-06, "loss": 2.9349, "step": 198930 }, { "epoch": 0.0011264, "grad_norm": 0.7847514748573303, "learning_rate": 9.72813122469526e-06, "loss": 2.7426, "step": 198940 }, { "epoch": 0.001152, "grad_norm": 0.8862506747245789, "learning_rate": 9.727326244758127e-06, "loss": 2.8298, "step": 198950 }, { "epoch": 0.0011776, "grad_norm": 0.7116496562957764, "learning_rate": 9.726521266589207e-06, "loss": 2.8364, "step": 198960 }, { "epoch": 0.0012032, "grad_norm": 0.779758632183075, "learning_rate": 9.725716290193728e-06, "loss": 2.8432, "step": 198970 }, { "epoch": 0.0012288, "grad_norm": 0.7558170557022095, "learning_rate": 9.724911315576904e-06, "loss": 3.1793, "step": 198980 }, { "epoch": 0.0012544, "grad_norm": 0.7150483131408691, "learning_rate": 9.724106342743958e-06, "loss": 2.6555, "step": 198990 }, { "epoch": 0.00128, "grad_norm": 0.8101390600204468, "learning_rate": 9.72330137170011e-06, "loss": 2.936, "step": 199000 }, { "epoch": 0.0013056, "grad_norm": 0.7793916463851929, "learning_rate": 9.722496402450577e-06, "loss": 2.7196, "step": 199010 }, { "epoch": 0.0013312, "grad_norm": 0.881943941116333, "learning_rate": 9.721691435000583e-06, "loss": 2.53, "step": 199020 }, { "epoch": 0.0013568, "grad_norm": 0.8019834756851196, "learning_rate": 9.720886469355344e-06, "loss": 2.8262, "step": 199030 }, { "epoch": 0.0013824, "grad_norm": 0.7881433963775635, "learning_rate": 9.720081505520085e-06, "loss": 2.8501, "step": 199040 }, { "epoch": 0.001408, "grad_norm": 0.7286394834518433, "learning_rate": 9.71927654350002e-06, "loss": 2.6287, "step": 199050 }, { "epoch": 0.0014336, "grad_norm": 0.8044498562812805, "learning_rate": 9.718471583300373e-06, "loss": 2.6844, "step": 199060 }, { "epoch": 0.0014592, "grad_norm": 0.8072863817214966, "learning_rate": 9.717666624926364e-06, "loss": 2.6121, "step": 199070 }, { "epoch": 0.0014848, "grad_norm": 0.934536874294281, "learning_rate": 9.716861668383212e-06, "loss": 2.5436, "step": 199080 }, { "epoch": 0.0015104, "grad_norm": 1.0243819952011108, "learning_rate": 9.716056713676136e-06, "loss": 2.2892, "step": 199090 }, { "epoch": 0.001536, "grad_norm": 0.8186686635017395, "learning_rate": 9.715251760810354e-06, "loss": 2.6163, "step": 199100 }, { "epoch": 0.0015616, "grad_norm": 0.7747260928153992, "learning_rate": 9.714446809791094e-06, "loss": 2.7896, "step": 199110 }, { "epoch": 0.0015872, "grad_norm": 0.6919050216674805, "learning_rate": 9.713641860623567e-06, "loss": 2.8516, "step": 199120 }, { "epoch": 0.0016128, "grad_norm": 0.7719065546989441, "learning_rate": 9.712836913312997e-06, "loss": 2.8344, "step": 199130 }, { "epoch": 0.0016384, "grad_norm": 0.9767478704452515, "learning_rate": 9.712031967864602e-06, "loss": 2.7234, "step": 199140 }, { "epoch": 0.001664, "grad_norm": 0.8182969093322754, "learning_rate": 9.711227024283604e-06, "loss": 2.6699, "step": 199150 }, { "epoch": 0.0016896, "grad_norm": 0.8716292381286621, "learning_rate": 9.710422082575222e-06, "loss": 2.8284, "step": 199160 }, { "epoch": 0.0017152, "grad_norm": 0.7566127777099609, "learning_rate": 9.709617142744678e-06, "loss": 2.8202, "step": 199170 }, { "epoch": 0.0017408, "grad_norm": 0.8076270222663879, "learning_rate": 9.708812204797188e-06, "loss": 2.9355, "step": 199180 }, { "epoch": 0.0017664, "grad_norm": 0.6845118999481201, "learning_rate": 9.708007268737974e-06, "loss": 2.7036, "step": 199190 }, { "epoch": 0.001792, "grad_norm": 0.8507477045059204, "learning_rate": 9.707202334572257e-06, "loss": 2.9644, "step": 199200 }, { "epoch": 2.56e-05, "grad_norm": 0.784572422504425, "learning_rate": 9.706397402305252e-06, "loss": 1.9946, "step": 199210 }, { "epoch": 5.12e-05, "grad_norm": 0.9131078720092773, "learning_rate": 9.705592471942183e-06, "loss": 2.9032, "step": 199220 }, { "epoch": 7.68e-05, "grad_norm": 0.7202063202857971, "learning_rate": 9.704787543488267e-06, "loss": 2.7141, "step": 199230 }, { "epoch": 0.0001024, "grad_norm": 0.7416545152664185, "learning_rate": 9.703982616948729e-06, "loss": 2.7285, "step": 199240 }, { "epoch": 0.000128, "grad_norm": 0.8665880560874939, "learning_rate": 9.70317769232878e-06, "loss": 2.713, "step": 199250 }, { "epoch": 0.0001536, "grad_norm": 1.0630601644515991, "learning_rate": 9.702372769633649e-06, "loss": 2.9046, "step": 199260 }, { "epoch": 0.0001792, "grad_norm": 0.8918479681015015, "learning_rate": 9.701567848868549e-06, "loss": 2.7589, "step": 199270 }, { "epoch": 0.0002048, "grad_norm": 0.8655666708946228, "learning_rate": 9.700762930038705e-06, "loss": 2.8767, "step": 199280 }, { "epoch": 0.0002304, "grad_norm": 1.0345244407653809, "learning_rate": 9.699958013149331e-06, "loss": 2.6992, "step": 199290 }, { "epoch": 0.000256, "grad_norm": 0.9097530245780945, "learning_rate": 9.699153098205652e-06, "loss": 2.6953, "step": 199300 }, { "epoch": 0.0002816, "grad_norm": 0.7553220987319946, "learning_rate": 9.69834818521288e-06, "loss": 2.6386, "step": 199310 }, { "epoch": 0.0003072, "grad_norm": 0.7239868640899658, "learning_rate": 9.697543274176243e-06, "loss": 2.7403, "step": 199320 }, { "epoch": 0.0003328, "grad_norm": 0.7936682105064392, "learning_rate": 9.69673836510096e-06, "loss": 2.6623, "step": 199330 }, { "epoch": 0.0003584, "grad_norm": 0.7288436889648438, "learning_rate": 9.695933457992246e-06, "loss": 2.8897, "step": 199340 }, { "epoch": 0.000384, "grad_norm": 0.8338145017623901, "learning_rate": 9.695128552855323e-06, "loss": 2.6541, "step": 199350 }, { "epoch": 0.0004096, "grad_norm": 0.7558872103691101, "learning_rate": 9.69432364969541e-06, "loss": 2.5748, "step": 199360 }, { "epoch": 0.0004352, "grad_norm": 0.8037660121917725, "learning_rate": 9.693518748517729e-06, "loss": 2.8204, "step": 199370 }, { "epoch": 0.0004608, "grad_norm": 0.9022569060325623, "learning_rate": 9.692713849327495e-06, "loss": 2.6167, "step": 199380 }, { "epoch": 0.0004864, "grad_norm": 0.9849441051483154, "learning_rate": 9.69190895212993e-06, "loss": 2.7888, "step": 199390 }, { "epoch": 0.000512, "grad_norm": 0.8118739724159241, "learning_rate": 9.691104056930255e-06, "loss": 2.6914, "step": 199400 }, { "epoch": 0.0005376, "grad_norm": 0.9460655450820923, "learning_rate": 9.690299163733688e-06, "loss": 2.6586, "step": 199410 }, { "epoch": 0.0005632, "grad_norm": 0.8165128231048584, "learning_rate": 9.689494272545447e-06, "loss": 2.6018, "step": 199420 }, { "epoch": 0.0005888, "grad_norm": 1.1054965257644653, "learning_rate": 9.688689383370753e-06, "loss": 2.5419, "step": 199430 }, { "epoch": 0.0006144, "grad_norm": 0.8717538118362427, "learning_rate": 9.68788449621483e-06, "loss": 2.7108, "step": 199440 }, { "epoch": 0.00064, "grad_norm": 0.7800902724266052, "learning_rate": 9.687079611082889e-06, "loss": 2.5085, "step": 199450 }, { "epoch": 0.0006656, "grad_norm": 0.8914211392402649, "learning_rate": 9.686274727980154e-06, "loss": 2.6375, "step": 199460 }, { "epoch": 0.0006912, "grad_norm": 1.0743253231048584, "learning_rate": 9.685469846911846e-06, "loss": 3.0533, "step": 199470 }, { "epoch": 0.0007168, "grad_norm": 0.8542785048484802, "learning_rate": 9.684664967883175e-06, "loss": 2.789, "step": 199480 }, { "epoch": 0.0007424, "grad_norm": 0.8960481882095337, "learning_rate": 9.683860090899374e-06, "loss": 2.7846, "step": 199490 }, { "epoch": 0.000768, "grad_norm": 0.8731821179389954, "learning_rate": 9.683055215965657e-06, "loss": 2.8064, "step": 199500 }, { "epoch": 0.0007936, "grad_norm": 0.8034944534301758, "learning_rate": 9.682250343087241e-06, "loss": 2.903, "step": 199510 }, { "epoch": 0.0008192, "grad_norm": 0.7808647751808167, "learning_rate": 9.681445472269349e-06, "loss": 2.8295, "step": 199520 }, { "epoch": 0.0008448, "grad_norm": 0.9205333590507507, "learning_rate": 9.680640603517198e-06, "loss": 2.5955, "step": 199530 }, { "epoch": 0.0008704, "grad_norm": 0.8483132719993591, "learning_rate": 9.679835736836006e-06, "loss": 2.8234, "step": 199540 }, { "epoch": 0.000896, "grad_norm": 0.7379748225212097, "learning_rate": 9.679030872230995e-06, "loss": 2.5532, "step": 199550 }, { "epoch": 0.0009216, "grad_norm": 0.7461762428283691, "learning_rate": 9.678226009707383e-06, "loss": 2.439, "step": 199560 }, { "epoch": 0.0009472, "grad_norm": 0.842664361000061, "learning_rate": 9.677421149270393e-06, "loss": 2.4298, "step": 199570 }, { "epoch": 0.0009728, "grad_norm": 0.7594050765037537, "learning_rate": 9.676616290925238e-06, "loss": 2.7691, "step": 199580 }, { "epoch": 0.0009984, "grad_norm": 0.7805612683296204, "learning_rate": 9.67581143467714e-06, "loss": 2.5962, "step": 199590 }, { "epoch": 0.001024, "grad_norm": 0.8042610287666321, "learning_rate": 9.675006580531323e-06, "loss": 2.7774, "step": 199600 }, { "epoch": 2.56e-05, "grad_norm": 0.805950403213501, "learning_rate": 9.674201728492996e-06, "loss": 2.0651, "step": 199610 }, { "epoch": 5.12e-05, "grad_norm": 0.8623476624488831, "learning_rate": 9.673396878567388e-06, "loss": 2.7625, "step": 199620 }, { "epoch": 7.68e-05, "grad_norm": 1.4944902658462524, "learning_rate": 9.672592030759713e-06, "loss": 3.0072, "step": 199630 }, { "epoch": 0.0001024, "grad_norm": 0.9248992800712585, "learning_rate": 9.671787185075194e-06, "loss": 2.5181, "step": 199640 }, { "epoch": 0.000128, "grad_norm": 0.8190217018127441, "learning_rate": 9.670982341519044e-06, "loss": 2.6899, "step": 199650 }, { "epoch": 0.0001536, "grad_norm": 0.7662463784217834, "learning_rate": 9.670177500096487e-06, "loss": 2.8309, "step": 199660 }, { "epoch": 0.0001792, "grad_norm": 0.9993248581886292, "learning_rate": 9.66937266081274e-06, "loss": 2.7042, "step": 199670 }, { "epoch": 0.0002048, "grad_norm": 0.7632424831390381, "learning_rate": 9.668567823673024e-06, "loss": 2.6459, "step": 199680 }, { "epoch": 0.0002304, "grad_norm": 0.8424177765846252, "learning_rate": 9.667762988682562e-06, "loss": 2.6862, "step": 199690 }, { "epoch": 0.000256, "grad_norm": 0.8056575655937195, "learning_rate": 9.666958155846565e-06, "loss": 2.6961, "step": 199700 }, { "epoch": 0.0002816, "grad_norm": 0.8041330575942993, "learning_rate": 9.666153325170257e-06, "loss": 2.571, "step": 199710 }, { "epoch": 0.0003072, "grad_norm": 0.8400055766105652, "learning_rate": 9.665348496658855e-06, "loss": 2.4671, "step": 199720 }, { "epoch": 0.0003328, "grad_norm": 0.8599428534507751, "learning_rate": 9.664543670317581e-06, "loss": 2.8359, "step": 199730 }, { "epoch": 0.0003584, "grad_norm": 0.8144948482513428, "learning_rate": 9.663738846151651e-06, "loss": 2.8907, "step": 199740 }, { "epoch": 0.000384, "grad_norm": 0.9271602034568787, "learning_rate": 9.662934024166285e-06, "loss": 2.7831, "step": 199750 }, { "epoch": 0.0004096, "grad_norm": 0.725073516368866, "learning_rate": 9.662129204366702e-06, "loss": 2.7102, "step": 199760 }, { "epoch": 0.0004352, "grad_norm": 0.8133559226989746, "learning_rate": 9.661324386758123e-06, "loss": 2.9014, "step": 199770 }, { "epoch": 0.0004608, "grad_norm": 0.9401199817657471, "learning_rate": 9.660519571345763e-06, "loss": 2.828, "step": 199780 }, { "epoch": 0.0004864, "grad_norm": 0.9464716911315918, "learning_rate": 9.659714758134844e-06, "loss": 2.9449, "step": 199790 }, { "epoch": 0.000512, "grad_norm": 0.8694116473197937, "learning_rate": 9.658909947130588e-06, "loss": 2.6438, "step": 199800 }, { "epoch": 0.0005376, "grad_norm": 0.8197281956672668, "learning_rate": 9.658105138338206e-06, "loss": 2.756, "step": 199810 }, { "epoch": 0.0005632, "grad_norm": 0.8761594295501709, "learning_rate": 9.65730033176292e-06, "loss": 2.7013, "step": 199820 }, { "epoch": 0.0005888, "grad_norm": 0.7738593220710754, "learning_rate": 9.656495527409951e-06, "loss": 2.4921, "step": 199830 }, { "epoch": 0.0006144, "grad_norm": 0.8968560695648193, "learning_rate": 9.655690725284518e-06, "loss": 2.4752, "step": 199840 }, { "epoch": 0.00064, "grad_norm": 0.7601984143257141, "learning_rate": 9.65488592539184e-06, "loss": 2.632, "step": 199850 }, { "epoch": 0.0006656, "grad_norm": 0.8074830174446106, "learning_rate": 9.654081127737137e-06, "loss": 2.83, "step": 199860 }, { "epoch": 0.0006912, "grad_norm": 0.8096455931663513, "learning_rate": 9.653276332325624e-06, "loss": 3.1071, "step": 199870 }, { "epoch": 0.0007168, "grad_norm": 0.8047357201576233, "learning_rate": 9.65247153916252e-06, "loss": 2.6987, "step": 199880 }, { "epoch": 0.0007424, "grad_norm": 0.7745975852012634, "learning_rate": 9.65166674825305e-06, "loss": 2.6132, "step": 199890 }, { "epoch": 0.000768, "grad_norm": 1.0325969457626343, "learning_rate": 9.650861959602426e-06, "loss": 3.2056, "step": 199900 }, { "epoch": 0.0007936, "grad_norm": 0.813312292098999, "learning_rate": 9.65005717321587e-06, "loss": 2.7572, "step": 199910 }, { "epoch": 0.0008192, "grad_norm": 0.7365385890007019, "learning_rate": 9.649252389098597e-06, "loss": 2.5498, "step": 199920 }, { "epoch": 0.0008448, "grad_norm": 0.8081306219100952, "learning_rate": 9.648447607255836e-06, "loss": 2.7571, "step": 199930 }, { "epoch": 0.0008704, "grad_norm": 0.8897267580032349, "learning_rate": 9.647642827692794e-06, "loss": 2.7581, "step": 199940 }, { "epoch": 0.000896, "grad_norm": 1.0255166292190552, "learning_rate": 9.646838050414694e-06, "loss": 2.8619, "step": 199950 }, { "epoch": 0.0009216, "grad_norm": 0.7501639723777771, "learning_rate": 9.646033275426757e-06, "loss": 2.5688, "step": 199960 }, { "epoch": 0.0009472, "grad_norm": 0.7012717127799988, "learning_rate": 9.6452285027342e-06, "loss": 2.5726, "step": 199970 }, { "epoch": 0.0009728, "grad_norm": 0.7891156673431396, "learning_rate": 9.644423732342242e-06, "loss": 2.7774, "step": 199980 }, { "epoch": 0.0009984, "grad_norm": 0.7432307600975037, "learning_rate": 9.643618964256101e-06, "loss": 2.7141, "step": 199990 }, { "epoch": 0.001024, "grad_norm": 0.7746476531028748, "learning_rate": 9.642814198480997e-06, "loss": 2.9044, "step": 200000 }, { "epoch": 0.0010496, "grad_norm": 0.8533769249916077, "learning_rate": 9.642009435022144e-06, "loss": 2.9723, "step": 200010 }, { "epoch": 0.0010752, "grad_norm": 0.8371717929840088, "learning_rate": 9.64120467388477e-06, "loss": 2.6387, "step": 200020 }, { "epoch": 0.0011008, "grad_norm": 0.8389933705329895, "learning_rate": 9.640399915074086e-06, "loss": 2.8698, "step": 200030 }, { "epoch": 0.0011264, "grad_norm": 0.7742375731468201, "learning_rate": 9.639595158595313e-06, "loss": 2.8466, "step": 200040 }, { "epoch": 0.001152, "grad_norm": 0.8270347714424133, "learning_rate": 9.63879040445367e-06, "loss": 2.5525, "step": 200050 }, { "epoch": 0.0011776, "grad_norm": 0.8517976999282837, "learning_rate": 9.637985652654375e-06, "loss": 2.7157, "step": 200060 }, { "epoch": 0.0012032, "grad_norm": 0.7651728391647339, "learning_rate": 9.637180903202648e-06, "loss": 2.7356, "step": 200070 }, { "epoch": 0.0012288, "grad_norm": 0.7628679871559143, "learning_rate": 9.636376156103706e-06, "loss": 2.9815, "step": 200080 }, { "epoch": 0.0012544, "grad_norm": 0.8486818075180054, "learning_rate": 9.63557141136277e-06, "loss": 2.9128, "step": 200090 }, { "epoch": 0.00128, "grad_norm": 0.8547489643096924, "learning_rate": 9.634766668985051e-06, "loss": 2.4995, "step": 200100 }, { "epoch": 0.0013056, "grad_norm": 0.7700185775756836, "learning_rate": 9.633961928975776e-06, "loss": 2.3985, "step": 200110 }, { "epoch": 0.0013312, "grad_norm": 1.0696213245391846, "learning_rate": 9.63315719134016e-06, "loss": 2.8672, "step": 200120 }, { "epoch": 0.0013568, "grad_norm": 0.8662105798721313, "learning_rate": 9.632352456083427e-06, "loss": 2.7409, "step": 200130 }, { "epoch": 0.0013824, "grad_norm": 0.8183448314666748, "learning_rate": 9.631547723210784e-06, "loss": 2.7271, "step": 200140 }, { "epoch": 0.001408, "grad_norm": 0.790101170539856, "learning_rate": 9.630742992727458e-06, "loss": 2.9736, "step": 200150 }, { "epoch": 0.0014336, "grad_norm": 0.8879699110984802, "learning_rate": 9.629938264638668e-06, "loss": 2.6966, "step": 200160 }, { "epoch": 0.0014592, "grad_norm": 0.7182961702346802, "learning_rate": 9.629133538949628e-06, "loss": 2.6362, "step": 200170 }, { "epoch": 0.0014848, "grad_norm": 0.7777477502822876, "learning_rate": 9.628328815665556e-06, "loss": 2.6086, "step": 200180 }, { "epoch": 0.0015104, "grad_norm": 0.8627166152000427, "learning_rate": 9.627524094791675e-06, "loss": 2.2941, "step": 200190 }, { "epoch": 0.001536, "grad_norm": 0.841729998588562, "learning_rate": 9.626719376333202e-06, "loss": 2.7905, "step": 200200 }, { "epoch": 0.0015616, "grad_norm": 0.8573154807090759, "learning_rate": 9.625914660295354e-06, "loss": 2.8013, "step": 200210 }, { "epoch": 0.0015872, "grad_norm": 0.8387582302093506, "learning_rate": 9.625109946683351e-06, "loss": 2.9968, "step": 200220 }, { "epoch": 0.0016128, "grad_norm": 0.9259862899780273, "learning_rate": 9.62430523550241e-06, "loss": 2.7646, "step": 200230 }, { "epoch": 0.0016384, "grad_norm": 0.8346534967422485, "learning_rate": 9.62350052675775e-06, "loss": 2.8974, "step": 200240 }, { "epoch": 0.001664, "grad_norm": 0.7527507543563843, "learning_rate": 9.622695820454588e-06, "loss": 2.6796, "step": 200250 }, { "epoch": 0.0016896, "grad_norm": 0.7985179424285889, "learning_rate": 9.621891116598145e-06, "loss": 2.64, "step": 200260 }, { "epoch": 0.0017152, "grad_norm": 1.5869020223617554, "learning_rate": 9.621086415193636e-06, "loss": 2.6204, "step": 200270 }, { "epoch": 0.0017408, "grad_norm": 0.8901395797729492, "learning_rate": 9.620281716246281e-06, "loss": 2.6975, "step": 200280 }, { "epoch": 0.0017664, "grad_norm": 0.819341778755188, "learning_rate": 9.619477019761301e-06, "loss": 2.974, "step": 200290 }, { "epoch": 0.001792, "grad_norm": 0.8768289685249329, "learning_rate": 9.618672325743908e-06, "loss": 2.7945, "step": 200300 }, { "epoch": 0.0018176, "grad_norm": 0.9375269412994385, "learning_rate": 9.617867634199324e-06, "loss": 2.6099, "step": 200310 }, { "epoch": 0.0018432, "grad_norm": 0.7445455193519592, "learning_rate": 9.617062945132767e-06, "loss": 2.5622, "step": 200320 }, { "epoch": 0.0018688, "grad_norm": 0.782004714012146, "learning_rate": 9.616258258549456e-06, "loss": 2.5246, "step": 200330 }, { "epoch": 0.0018944, "grad_norm": 0.7606692314147949, "learning_rate": 9.615453574454607e-06, "loss": 3.0159, "step": 200340 }, { "epoch": 0.00192, "grad_norm": 0.8762985467910767, "learning_rate": 9.614648892853438e-06, "loss": 2.8586, "step": 200350 }, { "epoch": 0.0019456, "grad_norm": 0.7287071347236633, "learning_rate": 9.61384421375117e-06, "loss": 2.8229, "step": 200360 }, { "epoch": 0.0019712, "grad_norm": 0.7031945586204529, "learning_rate": 9.613039537153019e-06, "loss": 2.7309, "step": 200370 }, { "epoch": 0.0019968, "grad_norm": 0.8371948003768921, "learning_rate": 9.612234863064205e-06, "loss": 2.8244, "step": 200380 }, { "epoch": 0.0020224, "grad_norm": 0.7586254477500916, "learning_rate": 9.611430191489946e-06, "loss": 2.83, "step": 200390 }, { "epoch": 0.002048, "grad_norm": 0.8953272104263306, "learning_rate": 9.610625522435457e-06, "loss": 2.8358, "step": 200400 }, { "epoch": 0.0020736, "grad_norm": 0.8762088418006897, "learning_rate": 9.609820855905958e-06, "loss": 2.5771, "step": 200410 }, { "epoch": 0.0020992, "grad_norm": 0.7679941058158875, "learning_rate": 9.60901619190667e-06, "loss": 2.6869, "step": 200420 }, { "epoch": 0.0021248, "grad_norm": 0.845038890838623, "learning_rate": 9.608211530442807e-06, "loss": 2.6676, "step": 200430 }, { "epoch": 0.0021504, "grad_norm": 0.7862495183944702, "learning_rate": 9.607406871519584e-06, "loss": 2.7393, "step": 200440 }, { "epoch": 0.002176, "grad_norm": 0.8260658383369446, "learning_rate": 9.606602215142226e-06, "loss": 2.9835, "step": 200450 }, { "epoch": 0.0022016, "grad_norm": 0.7847174406051636, "learning_rate": 9.60579756131595e-06, "loss": 3.0712, "step": 200460 }, { "epoch": 0.0022272, "grad_norm": 0.8519521951675415, "learning_rate": 9.604992910045971e-06, "loss": 2.925, "step": 200470 }, { "epoch": 0.0022528, "grad_norm": 1.061997890472412, "learning_rate": 9.604188261337505e-06, "loss": 2.9328, "step": 200480 }, { "epoch": 0.0022784, "grad_norm": 0.7668279409408569, "learning_rate": 9.603383615195778e-06, "loss": 2.7591, "step": 200490 }, { "epoch": 0.002304, "grad_norm": 0.8549255132675171, "learning_rate": 9.602578971625998e-06, "loss": 3.0791, "step": 200500 }, { "epoch": 0.0023296, "grad_norm": 0.7913409471511841, "learning_rate": 9.60177433063339e-06, "loss": 2.6542, "step": 200510 }, { "epoch": 0.0023552, "grad_norm": 0.8169540166854858, "learning_rate": 9.600969692223168e-06, "loss": 2.8716, "step": 200520 }, { "epoch": 0.0023808, "grad_norm": 0.8052738308906555, "learning_rate": 9.60016505640055e-06, "loss": 2.9205, "step": 200530 }, { "epoch": 0.0024064, "grad_norm": 0.7720394134521484, "learning_rate": 9.599360423170758e-06, "loss": 2.5848, "step": 200540 }, { "epoch": 0.002432, "grad_norm": 1.057883620262146, "learning_rate": 9.59855579253901e-06, "loss": 2.713, "step": 200550 }, { "epoch": 0.0024576, "grad_norm": 1.3184270858764648, "learning_rate": 9.597751164510518e-06, "loss": 2.7888, "step": 200560 }, { "epoch": 0.0024832, "grad_norm": 0.7881327271461487, "learning_rate": 9.596946539090503e-06, "loss": 2.9041, "step": 200570 }, { "epoch": 0.0025088, "grad_norm": 0.820358395576477, "learning_rate": 9.596141916284184e-06, "loss": 2.6505, "step": 200580 }, { "epoch": 0.0025344, "grad_norm": 1.2064533233642578, "learning_rate": 9.595337296096774e-06, "loss": 3.0434, "step": 200590 }, { "epoch": 0.00256, "grad_norm": 0.8773373365402222, "learning_rate": 9.594532678533496e-06, "loss": 3.0451, "step": 200600 }, { "epoch": 0.0025856, "grad_norm": 0.7551149725914001, "learning_rate": 9.593728063599566e-06, "loss": 2.7736, "step": 200610 }, { "epoch": 0.0026112, "grad_norm": 0.7361243367195129, "learning_rate": 9.592923451300203e-06, "loss": 2.9034, "step": 200620 }, { "epoch": 0.0026368, "grad_norm": 0.7733967900276184, "learning_rate": 9.592118841640622e-06, "loss": 2.7947, "step": 200630 }, { "epoch": 0.0026624, "grad_norm": 0.796054482460022, "learning_rate": 9.591314234626042e-06, "loss": 2.8087, "step": 200640 }, { "epoch": 0.002688, "grad_norm": 0.9331441521644592, "learning_rate": 9.590509630261678e-06, "loss": 2.8985, "step": 200650 }, { "epoch": 0.0027136, "grad_norm": 0.864383339881897, "learning_rate": 9.589705028552755e-06, "loss": 3.1139, "step": 200660 }, { "epoch": 0.0027392, "grad_norm": 0.9214456081390381, "learning_rate": 9.588900429504482e-06, "loss": 2.6708, "step": 200670 }, { "epoch": 0.0027648, "grad_norm": 0.8394685983657837, "learning_rate": 9.588095833122081e-06, "loss": 3.0517, "step": 200680 }, { "epoch": 0.0027904, "grad_norm": 0.8379393815994263, "learning_rate": 9.587291239410772e-06, "loss": 2.964, "step": 200690 }, { "epoch": 0.002816, "grad_norm": 0.7603936195373535, "learning_rate": 9.586486648375763e-06, "loss": 3.0977, "step": 200700 }, { "epoch": 0.0028416, "grad_norm": 0.7696166634559631, "learning_rate": 9.585682060022284e-06, "loss": 2.9094, "step": 200710 }, { "epoch": 0.0028672, "grad_norm": 0.7422009706497192, "learning_rate": 9.584877474355544e-06, "loss": 2.8052, "step": 200720 }, { "epoch": 0.0028928, "grad_norm": 0.8396299481391907, "learning_rate": 9.584072891380765e-06, "loss": 2.8538, "step": 200730 }, { "epoch": 0.0029184, "grad_norm": 0.9045770168304443, "learning_rate": 9.583268311103161e-06, "loss": 2.8295, "step": 200740 }, { "epoch": 0.002944, "grad_norm": 0.8841625452041626, "learning_rate": 9.582463733527955e-06, "loss": 3.0712, "step": 200750 }, { "epoch": 0.0029696, "grad_norm": 0.8026592135429382, "learning_rate": 9.581659158660358e-06, "loss": 2.8313, "step": 200760 }, { "epoch": 0.0029952, "grad_norm": 0.7213956713676453, "learning_rate": 9.580854586505588e-06, "loss": 2.7823, "step": 200770 }, { "epoch": 0.0030208, "grad_norm": 0.8380191326141357, "learning_rate": 9.58005001706887e-06, "loss": 2.9003, "step": 200780 }, { "epoch": 0.0030464, "grad_norm": 0.7586811780929565, "learning_rate": 9.579245450355413e-06, "loss": 2.7759, "step": 200790 }, { "epoch": 0.003072, "grad_norm": 0.8173037171363831, "learning_rate": 9.578440886370437e-06, "loss": 2.9534, "step": 200800 }, { "epoch": 0.0030976, "grad_norm": 1.019277811050415, "learning_rate": 9.57763632511916e-06, "loss": 3.1078, "step": 200810 }, { "epoch": 0.0031232, "grad_norm": 0.9771445989608765, "learning_rate": 9.5768317666068e-06, "loss": 2.825, "step": 200820 }, { "epoch": 0.0031488, "grad_norm": 1.001278042793274, "learning_rate": 9.576027210838572e-06, "loss": 2.7685, "step": 200830 }, { "epoch": 0.0031744, "grad_norm": 0.9498536586761475, "learning_rate": 9.575222657819696e-06, "loss": 3.0963, "step": 200840 }, { "epoch": 0.0032, "grad_norm": 0.7308784127235413, "learning_rate": 9.574418107555388e-06, "loss": 2.9465, "step": 200850 }, { "epoch": 0.0032256, "grad_norm": 0.7849959135055542, "learning_rate": 9.573613560050868e-06, "loss": 2.8284, "step": 200860 }, { "epoch": 0.0032512, "grad_norm": 0.790610134601593, "learning_rate": 9.572809015311348e-06, "loss": 3.0362, "step": 200870 }, { "epoch": 0.0032768, "grad_norm": 0.7627864480018616, "learning_rate": 9.572004473342045e-06, "loss": 2.9632, "step": 200880 }, { "epoch": 0.0033024, "grad_norm": 0.830177366733551, "learning_rate": 9.571199934148182e-06, "loss": 2.8419, "step": 200890 }, { "epoch": 0.003328, "grad_norm": 0.7742394208908081, "learning_rate": 9.570395397734975e-06, "loss": 2.6731, "step": 200900 }, { "epoch": 0.0033536, "grad_norm": 0.775205671787262, "learning_rate": 9.569590864107639e-06, "loss": 2.8414, "step": 200910 }, { "epoch": 0.0033792, "grad_norm": 0.7902529835700989, "learning_rate": 9.568786333271392e-06, "loss": 2.923, "step": 200920 }, { "epoch": 0.0034048, "grad_norm": 0.7372641563415527, "learning_rate": 9.567981805231448e-06, "loss": 2.8898, "step": 200930 }, { "epoch": 0.0034304, "grad_norm": 0.8254261016845703, "learning_rate": 9.56717727999303e-06, "loss": 2.8289, "step": 200940 }, { "epoch": 0.003456, "grad_norm": 0.7600114345550537, "learning_rate": 9.566372757561354e-06, "loss": 2.7352, "step": 200950 }, { "epoch": 0.0034816, "grad_norm": 0.6975481510162354, "learning_rate": 9.565568237941634e-06, "loss": 2.9028, "step": 200960 }, { "epoch": 0.0035072, "grad_norm": 0.8708170056343079, "learning_rate": 9.564763721139087e-06, "loss": 2.6138, "step": 200970 }, { "epoch": 0.0035328, "grad_norm": 0.7080447673797607, "learning_rate": 9.563959207158934e-06, "loss": 2.8885, "step": 200980 }, { "epoch": 0.0035584, "grad_norm": 0.8422085642814636, "learning_rate": 9.563154696006388e-06, "loss": 2.6974, "step": 200990 }, { "epoch": 0.003584, "grad_norm": 0.9367926120758057, "learning_rate": 9.562350187686668e-06, "loss": 2.9675, "step": 201000 }, { "epoch": 0.0036096, "grad_norm": 0.8324150443077087, "learning_rate": 9.561545682204989e-06, "loss": 2.8617, "step": 201010 }, { "epoch": 0.0036352, "grad_norm": 0.798962414264679, "learning_rate": 9.560741179566574e-06, "loss": 3.0518, "step": 201020 }, { "epoch": 0.0036608, "grad_norm": 0.8877171277999878, "learning_rate": 9.559936679776632e-06, "loss": 3.0922, "step": 201030 }, { "epoch": 0.0036864, "grad_norm": 0.8440646529197693, "learning_rate": 9.559132182840385e-06, "loss": 2.9397, "step": 201040 }, { "epoch": 0.003712, "grad_norm": 0.7985784411430359, "learning_rate": 9.558327688763045e-06, "loss": 3.0138, "step": 201050 }, { "epoch": 0.0037376, "grad_norm": 0.7642530202865601, "learning_rate": 9.557523197549834e-06, "loss": 2.8737, "step": 201060 }, { "epoch": 0.0037632, "grad_norm": 0.8481995463371277, "learning_rate": 9.556718709205969e-06, "loss": 2.808, "step": 201070 }, { "epoch": 0.0037888, "grad_norm": 0.7587720155715942, "learning_rate": 9.555914223736668e-06, "loss": 2.9212, "step": 201080 }, { "epoch": 0.0038144, "grad_norm": 0.772927463054657, "learning_rate": 9.555109741147143e-06, "loss": 2.8391, "step": 201090 }, { "epoch": 0.00384, "grad_norm": 0.8184411525726318, "learning_rate": 9.554305261442614e-06, "loss": 3.1147, "step": 201100 }, { "epoch": 0.0038656, "grad_norm": 1.1057511568069458, "learning_rate": 9.553500784628297e-06, "loss": 2.7872, "step": 201110 }, { "epoch": 0.0038912, "grad_norm": 0.858406126499176, "learning_rate": 9.552696310709407e-06, "loss": 2.7825, "step": 201120 }, { "epoch": 0.0039168, "grad_norm": 0.9233435392379761, "learning_rate": 9.551891839691163e-06, "loss": 2.7982, "step": 201130 }, { "epoch": 0.0039424, "grad_norm": 0.7552922368049622, "learning_rate": 9.551087371578782e-06, "loss": 2.9976, "step": 201140 }, { "epoch": 0.003968, "grad_norm": 0.8201253414154053, "learning_rate": 9.55028290637748e-06, "loss": 2.9799, "step": 201150 }, { "epoch": 0.0039936, "grad_norm": 0.714002788066864, "learning_rate": 9.549478444092473e-06, "loss": 2.8434, "step": 201160 }, { "epoch": 0.0040192, "grad_norm": 1.005210280418396, "learning_rate": 9.54867398472898e-06, "loss": 2.8626, "step": 201170 }, { "epoch": 0.0040448, "grad_norm": 0.7460026144981384, "learning_rate": 9.547869528292215e-06, "loss": 2.8641, "step": 201180 }, { "epoch": 0.0040704, "grad_norm": 0.7719852328300476, "learning_rate": 9.547065074787397e-06, "loss": 2.947, "step": 201190 }, { "epoch": 0.004096, "grad_norm": 0.7687134742736816, "learning_rate": 9.54626062421974e-06, "loss": 2.9581, "step": 201200 }, { "epoch": 0.0041216, "grad_norm": 0.7321064472198486, "learning_rate": 9.545456176594464e-06, "loss": 3.0609, "step": 201210 }, { "epoch": 0.0041472, "grad_norm": 0.7375218868255615, "learning_rate": 9.54465173191678e-06, "loss": 2.7984, "step": 201220 }, { "epoch": 0.0041728, "grad_norm": 1.1731457710266113, "learning_rate": 9.54384729019191e-06, "loss": 2.9053, "step": 201230 }, { "epoch": 0.0041984, "grad_norm": 0.8604990243911743, "learning_rate": 9.543042851425072e-06, "loss": 2.9198, "step": 201240 }, { "epoch": 0.004224, "grad_norm": 0.6996168494224548, "learning_rate": 9.542238415621478e-06, "loss": 2.896, "step": 201250 }, { "epoch": 0.0042496, "grad_norm": 0.8137400150299072, "learning_rate": 9.541433982786347e-06, "loss": 3.0247, "step": 201260 }, { "epoch": 0.0042752, "grad_norm": 0.7388460040092468, "learning_rate": 9.540629552924894e-06, "loss": 3.2025, "step": 201270 }, { "epoch": 0.0043008, "grad_norm": 0.7521500587463379, "learning_rate": 9.539825126042338e-06, "loss": 3.0271, "step": 201280 }, { "epoch": 0.0043264, "grad_norm": 0.7901250720024109, "learning_rate": 9.539020702143892e-06, "loss": 2.6079, "step": 201290 }, { "epoch": 0.004352, "grad_norm": 0.7560886740684509, "learning_rate": 9.538216281234774e-06, "loss": 3.003, "step": 201300 }, { "epoch": 0.0043776, "grad_norm": 0.8145491480827332, "learning_rate": 9.537411863320203e-06, "loss": 3.143, "step": 201310 }, { "epoch": 0.0044032, "grad_norm": 0.7438097596168518, "learning_rate": 9.536607448405392e-06, "loss": 2.9194, "step": 201320 }, { "epoch": 0.0044288, "grad_norm": 0.869806170463562, "learning_rate": 9.535803036495557e-06, "loss": 2.9641, "step": 201330 }, { "epoch": 0.0044544, "grad_norm": 0.7633569836616516, "learning_rate": 9.534998627595917e-06, "loss": 2.7564, "step": 201340 }, { "epoch": 0.00448, "grad_norm": 0.747553288936615, "learning_rate": 9.534194221711688e-06, "loss": 3.1667, "step": 201350 }, { "epoch": 0.0045056, "grad_norm": 0.7528709769248962, "learning_rate": 9.533389818848084e-06, "loss": 2.7924, "step": 201360 }, { "epoch": 0.0045312, "grad_norm": 0.7580112814903259, "learning_rate": 9.532585419010324e-06, "loss": 3.0272, "step": 201370 }, { "epoch": 0.0045568, "grad_norm": 0.7795208692550659, "learning_rate": 9.531781022203624e-06, "loss": 2.9035, "step": 201380 }, { "epoch": 0.0045824, "grad_norm": 0.7559414505958557, "learning_rate": 9.5309766284332e-06, "loss": 2.7788, "step": 201390 }, { "epoch": 0.004608, "grad_norm": 0.8690165281295776, "learning_rate": 9.530172237704263e-06, "loss": 3.278, "step": 201400 }, { "epoch": 0.0046336, "grad_norm": 0.7268831729888916, "learning_rate": 9.529367850022037e-06, "loss": 2.8934, "step": 201410 }, { "epoch": 0.0046592, "grad_norm": 0.7768687605857849, "learning_rate": 9.528563465391735e-06, "loss": 3.0193, "step": 201420 }, { "epoch": 0.0046848, "grad_norm": 0.7656417489051819, "learning_rate": 9.527759083818576e-06, "loss": 2.8534, "step": 201430 }, { "epoch": 0.0047104, "grad_norm": 0.8502042889595032, "learning_rate": 9.526954705307774e-06, "loss": 2.9146, "step": 201440 }, { "epoch": 0.004736, "grad_norm": 0.7711755633354187, "learning_rate": 9.526150329864544e-06, "loss": 3.0753, "step": 201450 }, { "epoch": 0.0047616, "grad_norm": 0.7663801312446594, "learning_rate": 9.525345957494102e-06, "loss": 3.0932, "step": 201460 }, { "epoch": 0.0047872, "grad_norm": 0.9521984457969666, "learning_rate": 9.524541588201666e-06, "loss": 2.92, "step": 201470 }, { "epoch": 0.0048128, "grad_norm": 0.7786270976066589, "learning_rate": 9.523737221992453e-06, "loss": 2.8237, "step": 201480 }, { "epoch": 0.0048384, "grad_norm": 0.7557692527770996, "learning_rate": 9.522932858871677e-06, "loss": 2.9084, "step": 201490 }, { "epoch": 0.004864, "grad_norm": 0.751311719417572, "learning_rate": 9.522128498844553e-06, "loss": 2.8441, "step": 201500 }, { "epoch": 0.0048896, "grad_norm": 0.8139442801475525, "learning_rate": 9.521324141916302e-06, "loss": 3.0266, "step": 201510 }, { "epoch": 0.0049152, "grad_norm": 0.8194706439971924, "learning_rate": 9.520519788092133e-06, "loss": 2.8567, "step": 201520 }, { "epoch": 0.0049408, "grad_norm": 0.7923263907432556, "learning_rate": 9.519715437377266e-06, "loss": 2.8735, "step": 201530 }, { "epoch": 0.0049664, "grad_norm": 0.7251827120780945, "learning_rate": 9.518911089776918e-06, "loss": 2.8563, "step": 201540 }, { "epoch": 0.004992, "grad_norm": 0.704957127571106, "learning_rate": 9.518106745296305e-06, "loss": 2.8667, "step": 201550 }, { "epoch": 0.0050176, "grad_norm": 0.7458284497261047, "learning_rate": 9.517302403940641e-06, "loss": 3.0562, "step": 201560 }, { "epoch": 0.0050432, "grad_norm": 0.7925475239753723, "learning_rate": 9.51649806571514e-06, "loss": 2.9521, "step": 201570 }, { "epoch": 0.0050688, "grad_norm": 1.6787160634994507, "learning_rate": 9.51569373062502e-06, "loss": 2.8026, "step": 201580 }, { "epoch": 0.0050944, "grad_norm": 0.7749454379081726, "learning_rate": 9.514889398675499e-06, "loss": 2.6229, "step": 201590 }, { "epoch": 0.00512, "grad_norm": 0.7247548699378967, "learning_rate": 9.514085069871795e-06, "loss": 2.7017, "step": 201600 }, { "epoch": 0.0051456, "grad_norm": 0.8416243195533752, "learning_rate": 9.513280744219115e-06, "loss": 2.8668, "step": 201610 }, { "epoch": 0.0051712, "grad_norm": 0.8027507662773132, "learning_rate": 9.512476421722683e-06, "loss": 3.0855, "step": 201620 }, { "epoch": 0.0051968, "grad_norm": 0.7796766757965088, "learning_rate": 9.51167210238771e-06, "loss": 2.8925, "step": 201630 }, { "epoch": 0.0052224, "grad_norm": 0.9134702086448669, "learning_rate": 9.510867786219416e-06, "loss": 2.8275, "step": 201640 }, { "epoch": 0.005248, "grad_norm": 0.7653339505195618, "learning_rate": 9.510063473223014e-06, "loss": 2.9886, "step": 201650 }, { "epoch": 0.0052736, "grad_norm": 0.7185186147689819, "learning_rate": 9.50925916340372e-06, "loss": 2.9352, "step": 201660 }, { "epoch": 0.0052992, "grad_norm": 0.8227011561393738, "learning_rate": 9.50845485676675e-06, "loss": 2.8267, "step": 201670 }, { "epoch": 0.0053248, "grad_norm": 0.9025833010673523, "learning_rate": 9.507650553317317e-06, "loss": 2.9021, "step": 201680 }, { "epoch": 0.0053504, "grad_norm": 0.838712751865387, "learning_rate": 9.506846253060642e-06, "loss": 2.7618, "step": 201690 }, { "epoch": 0.005376, "grad_norm": 1.0042693614959717, "learning_rate": 9.506041956001937e-06, "loss": 2.9576, "step": 201700 }, { "epoch": 0.0054016, "grad_norm": 0.8719683289527893, "learning_rate": 9.50523766214642e-06, "loss": 2.6238, "step": 201710 }, { "epoch": 0.0054272, "grad_norm": 1.1411263942718506, "learning_rate": 9.504433371499305e-06, "loss": 3.062, "step": 201720 }, { "epoch": 0.0054528, "grad_norm": 0.784099280834198, "learning_rate": 9.503629084065805e-06, "loss": 2.7038, "step": 201730 }, { "epoch": 0.0054784, "grad_norm": 0.8496609926223755, "learning_rate": 9.50282479985114e-06, "loss": 3.004, "step": 201740 }, { "epoch": 0.005504, "grad_norm": 0.8401645421981812, "learning_rate": 9.502020518860523e-06, "loss": 2.9736, "step": 201750 }, { "epoch": 0.0055296, "grad_norm": 0.7408748865127563, "learning_rate": 9.50121624109917e-06, "loss": 2.689, "step": 201760 }, { "epoch": 0.0055552, "grad_norm": 0.8628934621810913, "learning_rate": 9.500411966572301e-06, "loss": 2.7705, "step": 201770 }, { "epoch": 0.0055808, "grad_norm": 0.8310173749923706, "learning_rate": 9.499607695285125e-06, "loss": 2.8445, "step": 201780 }, { "epoch": 0.0056064, "grad_norm": 0.7745563387870789, "learning_rate": 9.498803427242862e-06, "loss": 2.8644, "step": 201790 }, { "epoch": 0.005632, "grad_norm": 0.760672390460968, "learning_rate": 9.497999162450726e-06, "loss": 3.0161, "step": 201800 }, { "epoch": 0.0056576, "grad_norm": 0.7007029056549072, "learning_rate": 9.497194900913929e-06, "loss": 2.9437, "step": 201810 }, { "epoch": 0.0056832, "grad_norm": 0.8247358202934265, "learning_rate": 9.496390642637692e-06, "loss": 2.8608, "step": 201820 }, { "epoch": 0.0057088, "grad_norm": 0.9919453859329224, "learning_rate": 9.495586387627226e-06, "loss": 2.7679, "step": 201830 }, { "epoch": 0.0057344, "grad_norm": 0.81596440076828, "learning_rate": 9.494782135887751e-06, "loss": 2.9798, "step": 201840 }, { "epoch": 0.00576, "grad_norm": 0.7750299572944641, "learning_rate": 9.493977887424477e-06, "loss": 2.8729, "step": 201850 }, { "epoch": 0.0057856, "grad_norm": 0.8825880885124207, "learning_rate": 9.49317364224262e-06, "loss": 2.9754, "step": 201860 }, { "epoch": 0.0058112, "grad_norm": 0.8680994510650635, "learning_rate": 9.492369400347403e-06, "loss": 2.7279, "step": 201870 }, { "epoch": 0.0058368, "grad_norm": 0.7664499878883362, "learning_rate": 9.491565161744031e-06, "loss": 2.8258, "step": 201880 }, { "epoch": 0.0058624, "grad_norm": 0.7864626049995422, "learning_rate": 9.490760926437725e-06, "loss": 3.001, "step": 201890 }, { "epoch": 0.005888, "grad_norm": 0.7578887343406677, "learning_rate": 9.489956694433699e-06, "loss": 2.6748, "step": 201900 }, { "epoch": 0.0059136, "grad_norm": 0.7439091205596924, "learning_rate": 9.48915246573717e-06, "loss": 2.7856, "step": 201910 }, { "epoch": 0.0059392, "grad_norm": 0.7279726266860962, "learning_rate": 9.488348240353346e-06, "loss": 2.9503, "step": 201920 }, { "epoch": 0.0059648, "grad_norm": 0.746731698513031, "learning_rate": 9.487544018287452e-06, "loss": 2.8297, "step": 201930 }, { "epoch": 0.0059904, "grad_norm": 0.762376606464386, "learning_rate": 9.486739799544699e-06, "loss": 2.7265, "step": 201940 }, { "epoch": 0.006016, "grad_norm": 0.8585491180419922, "learning_rate": 9.4859355841303e-06, "loss": 3.0463, "step": 201950 }, { "epoch": 0.0060416, "grad_norm": 2.314096450805664, "learning_rate": 9.485131372049473e-06, "loss": 2.772, "step": 201960 }, { "epoch": 0.0060672, "grad_norm": 0.7656092047691345, "learning_rate": 9.484327163307434e-06, "loss": 3.1648, "step": 201970 }, { "epoch": 0.0060928, "grad_norm": 0.8772436380386353, "learning_rate": 9.483522957909396e-06, "loss": 2.7963, "step": 201980 }, { "epoch": 0.0061184, "grad_norm": 0.7966966032981873, "learning_rate": 9.482718755860573e-06, "loss": 2.7764, "step": 201990 }, { "epoch": 0.006144, "grad_norm": 0.8783000707626343, "learning_rate": 9.481914557166184e-06, "loss": 2.8585, "step": 202000 }, { "epoch": 0.0061696, "grad_norm": 0.9676735401153564, "learning_rate": 9.481110361831439e-06, "loss": 2.8515, "step": 202010 }, { "epoch": 0.0061952, "grad_norm": 0.7731080651283264, "learning_rate": 9.480306169861554e-06, "loss": 2.9143, "step": 202020 }, { "epoch": 0.0062208, "grad_norm": 0.745768129825592, "learning_rate": 9.479501981261746e-06, "loss": 2.8281, "step": 202030 }, { "epoch": 0.0062464, "grad_norm": 0.7055412530899048, "learning_rate": 9.478697796037233e-06, "loss": 2.8472, "step": 202040 }, { "epoch": 0.006272, "grad_norm": 0.7343994975090027, "learning_rate": 9.477893614193223e-06, "loss": 2.8692, "step": 202050 }, { "epoch": 0.0062976, "grad_norm": 0.8245214819908142, "learning_rate": 9.477089435734933e-06, "loss": 3.0583, "step": 202060 }, { "epoch": 0.0063232, "grad_norm": 0.7212872505187988, "learning_rate": 9.476285260667584e-06, "loss": 2.8584, "step": 202070 }, { "epoch": 0.0063488, "grad_norm": 0.8254389762878418, "learning_rate": 9.47548108899638e-06, "loss": 2.9891, "step": 202080 }, { "epoch": 0.0063744, "grad_norm": 0.8556188941001892, "learning_rate": 9.474676920726544e-06, "loss": 2.9307, "step": 202090 }, { "epoch": 0.0064, "grad_norm": 0.8144597411155701, "learning_rate": 9.473872755863287e-06, "loss": 3.0671, "step": 202100 }, { "epoch": 0.0064256, "grad_norm": 0.7393926382064819, "learning_rate": 9.473068594411827e-06, "loss": 2.9769, "step": 202110 }, { "epoch": 0.0064512, "grad_norm": 1.1080608367919922, "learning_rate": 9.472264436377377e-06, "loss": 2.9365, "step": 202120 }, { "epoch": 0.0064768, "grad_norm": 0.789829432964325, "learning_rate": 9.471460281765155e-06, "loss": 2.8846, "step": 202130 }, { "epoch": 0.0065024, "grad_norm": 0.7725123763084412, "learning_rate": 9.47065613058037e-06, "loss": 2.7853, "step": 202140 }, { "epoch": 0.006528, "grad_norm": 0.7625093460083008, "learning_rate": 9.469851982828238e-06, "loss": 2.8556, "step": 202150 }, { "epoch": 0.0065536, "grad_norm": 0.8598356246948242, "learning_rate": 9.469047838513975e-06, "loss": 2.9247, "step": 202160 }, { "epoch": 0.0065792, "grad_norm": 0.7518532276153564, "learning_rate": 9.468243697642798e-06, "loss": 2.9153, "step": 202170 }, { "epoch": 0.0066048, "grad_norm": 0.7452237606048584, "learning_rate": 9.467439560219919e-06, "loss": 2.8239, "step": 202180 }, { "epoch": 0.0066304, "grad_norm": 0.7524687051773071, "learning_rate": 9.466635426250551e-06, "loss": 2.8732, "step": 202190 }, { "epoch": 0.006656, "grad_norm": 0.9922015070915222, "learning_rate": 9.465831295739914e-06, "loss": 2.8106, "step": 202200 }, { "epoch": 0.0066816, "grad_norm": 0.7830831408500671, "learning_rate": 9.465027168693215e-06, "loss": 2.8087, "step": 202210 }, { "epoch": 0.0067072, "grad_norm": 0.8897582292556763, "learning_rate": 9.464223045115673e-06, "loss": 2.9312, "step": 202220 }, { "epoch": 0.0067328, "grad_norm": 0.8142715692520142, "learning_rate": 9.463418925012504e-06, "loss": 2.855, "step": 202230 }, { "epoch": 0.0067584, "grad_norm": 0.7050673961639404, "learning_rate": 9.462614808388922e-06, "loss": 2.8568, "step": 202240 }, { "epoch": 0.006784, "grad_norm": 0.821658730506897, "learning_rate": 9.461810695250137e-06, "loss": 2.8375, "step": 202250 }, { "epoch": 0.0068096, "grad_norm": 0.8378446102142334, "learning_rate": 9.461006585601367e-06, "loss": 3.1843, "step": 202260 }, { "epoch": 0.0068352, "grad_norm": 0.8217183947563171, "learning_rate": 9.460202479447824e-06, "loss": 2.8435, "step": 202270 }, { "epoch": 0.0068608, "grad_norm": 0.8775679469108582, "learning_rate": 9.459398376794728e-06, "loss": 2.9216, "step": 202280 }, { "epoch": 0.0068864, "grad_norm": 0.786481499671936, "learning_rate": 9.45859427764729e-06, "loss": 2.9283, "step": 202290 }, { "epoch": 0.006912, "grad_norm": 1.1651853322982788, "learning_rate": 9.457790182010725e-06, "loss": 2.7113, "step": 202300 }, { "epoch": 0.0069376, "grad_norm": 0.7724609375, "learning_rate": 9.456986089890243e-06, "loss": 2.951, "step": 202310 }, { "epoch": 0.0069632, "grad_norm": 0.8181072473526001, "learning_rate": 9.456182001291065e-06, "loss": 2.9917, "step": 202320 }, { "epoch": 0.0069888, "grad_norm": 0.8347690105438232, "learning_rate": 9.455377916218402e-06, "loss": 2.8491, "step": 202330 }, { "epoch": 0.0070144, "grad_norm": 0.8332646489143372, "learning_rate": 9.454573834677466e-06, "loss": 2.9534, "step": 202340 }, { "epoch": 0.00704, "grad_norm": 0.8817780613899231, "learning_rate": 9.453769756673476e-06, "loss": 2.8154, "step": 202350 }, { "epoch": 0.0070656, "grad_norm": 0.7834857106208801, "learning_rate": 9.452965682211642e-06, "loss": 2.8618, "step": 202360 }, { "epoch": 0.0070912, "grad_norm": 0.7639154195785522, "learning_rate": 9.452161611297184e-06, "loss": 2.8269, "step": 202370 }, { "epoch": 0.0071168, "grad_norm": 0.9198042750358582, "learning_rate": 9.451357543935309e-06, "loss": 3.1033, "step": 202380 }, { "epoch": 0.0071424, "grad_norm": 0.7764999866485596, "learning_rate": 9.450553480131235e-06, "loss": 2.8291, "step": 202390 }, { "epoch": 0.007168, "grad_norm": 0.8036410212516785, "learning_rate": 9.449749419890178e-06, "loss": 3.1636, "step": 202400 }, { "epoch": 0.0071936, "grad_norm": 1.3357309103012085, "learning_rate": 9.448945363217347e-06, "loss": 3.1932, "step": 202410 }, { "epoch": 0.0072192, "grad_norm": 0.8016025424003601, "learning_rate": 9.448141310117958e-06, "loss": 2.9784, "step": 202420 }, { "epoch": 0.0072448, "grad_norm": 1.228304386138916, "learning_rate": 9.447337260597228e-06, "loss": 2.7663, "step": 202430 }, { "epoch": 0.0072704, "grad_norm": 0.8059925436973572, "learning_rate": 9.446533214660365e-06, "loss": 2.9366, "step": 202440 }, { "epoch": 0.007296, "grad_norm": 0.7828405499458313, "learning_rate": 9.44572917231259e-06, "loss": 3.1788, "step": 202450 }, { "epoch": 0.0073216, "grad_norm": 0.7679480910301208, "learning_rate": 9.444925133559117e-06, "loss": 2.8248, "step": 202460 }, { "epoch": 0.0073472, "grad_norm": 0.9332424998283386, "learning_rate": 9.444121098405153e-06, "loss": 2.8772, "step": 202470 }, { "epoch": 0.0073728, "grad_norm": 0.740145742893219, "learning_rate": 9.443317066855917e-06, "loss": 2.9334, "step": 202480 }, { "epoch": 0.0073984, "grad_norm": 0.7127499580383301, "learning_rate": 9.442513038916624e-06, "loss": 2.9285, "step": 202490 }, { "epoch": 0.007424, "grad_norm": 1.271674394607544, "learning_rate": 9.441709014592483e-06, "loss": 3.024, "step": 202500 }, { "epoch": 0.0074496, "grad_norm": 0.713335394859314, "learning_rate": 9.440904993888712e-06, "loss": 2.737, "step": 202510 }, { "epoch": 0.0074752, "grad_norm": 0.7562575340270996, "learning_rate": 9.440100976810524e-06, "loss": 2.8936, "step": 202520 }, { "epoch": 0.0075008, "grad_norm": 0.7244033813476562, "learning_rate": 9.439296963363133e-06, "loss": 3.2256, "step": 202530 }, { "epoch": 0.0075264, "grad_norm": 0.7789117693901062, "learning_rate": 9.43849295355175e-06, "loss": 3.1108, "step": 202540 }, { "epoch": 0.007552, "grad_norm": 0.7480767369270325, "learning_rate": 9.437688947381591e-06, "loss": 2.8516, "step": 202550 }, { "epoch": 0.0075776, "grad_norm": 0.7978320717811584, "learning_rate": 9.436884944857872e-06, "loss": 2.92, "step": 202560 }, { "epoch": 0.0076032, "grad_norm": 0.8113477230072021, "learning_rate": 9.436080945985802e-06, "loss": 2.9661, "step": 202570 }, { "epoch": 0.0076288, "grad_norm": 0.7441399693489075, "learning_rate": 9.435276950770599e-06, "loss": 2.9716, "step": 202580 }, { "epoch": 0.0076544, "grad_norm": 0.718515157699585, "learning_rate": 9.434472959217473e-06, "loss": 2.8729, "step": 202590 }, { "epoch": 0.00768, "grad_norm": 0.767629861831665, "learning_rate": 9.433668971331642e-06, "loss": 2.9106, "step": 202600 }, { "epoch": 0.0077056, "grad_norm": 0.7534518837928772, "learning_rate": 9.432864987118316e-06, "loss": 2.8489, "step": 202610 }, { "epoch": 0.0077312, "grad_norm": 0.7685655355453491, "learning_rate": 9.432061006582707e-06, "loss": 3.0369, "step": 202620 }, { "epoch": 0.0077568, "grad_norm": 0.80987948179245, "learning_rate": 9.431257029730033e-06, "loss": 2.8572, "step": 202630 }, { "epoch": 0.0077824, "grad_norm": 0.983260452747345, "learning_rate": 9.430453056565508e-06, "loss": 2.9332, "step": 202640 }, { "epoch": 0.007808, "grad_norm": 0.8269244432449341, "learning_rate": 9.429649087094341e-06, "loss": 2.9292, "step": 202650 }, { "epoch": 0.0078336, "grad_norm": 0.8422827124595642, "learning_rate": 9.428845121321752e-06, "loss": 3.1861, "step": 202660 }, { "epoch": 0.0078592, "grad_norm": 0.9236971735954285, "learning_rate": 9.428041159252947e-06, "loss": 2.8681, "step": 202670 }, { "epoch": 0.0078848, "grad_norm": 0.7519313097000122, "learning_rate": 9.427237200893145e-06, "loss": 2.8468, "step": 202680 }, { "epoch": 0.0079104, "grad_norm": 0.6773302555084229, "learning_rate": 9.426433246247559e-06, "loss": 2.6125, "step": 202690 }, { "epoch": 0.007936, "grad_norm": 0.6893882155418396, "learning_rate": 9.425629295321397e-06, "loss": 2.7184, "step": 202700 }, { "epoch": 0.0079616, "grad_norm": 0.7358335256576538, "learning_rate": 9.424825348119879e-06, "loss": 2.8857, "step": 202710 }, { "epoch": 0.0079872, "grad_norm": 0.8682985901832581, "learning_rate": 9.424021404648214e-06, "loss": 2.9645, "step": 202720 }, { "epoch": 0.0080128, "grad_norm": 0.7492583394050598, "learning_rate": 9.42321746491162e-06, "loss": 3.131, "step": 202730 }, { "epoch": 0.0080384, "grad_norm": 0.884854793548584, "learning_rate": 9.422413528915304e-06, "loss": 3.1012, "step": 202740 }, { "epoch": 0.008064, "grad_norm": 0.7535346746444702, "learning_rate": 9.421609596664485e-06, "loss": 2.9341, "step": 202750 }, { "epoch": 0.0080896, "grad_norm": 0.7775455117225647, "learning_rate": 9.420805668164376e-06, "loss": 2.7814, "step": 202760 }, { "epoch": 0.0081152, "grad_norm": 0.9263045787811279, "learning_rate": 9.420001743420185e-06, "loss": 2.7926, "step": 202770 }, { "epoch": 0.0081408, "grad_norm": 0.7733561396598816, "learning_rate": 9.419197822437128e-06, "loss": 2.8278, "step": 202780 }, { "epoch": 0.0081664, "grad_norm": 0.9112483859062195, "learning_rate": 9.418393905220418e-06, "loss": 3.0326, "step": 202790 }, { "epoch": 0.008192, "grad_norm": 0.7518881559371948, "learning_rate": 9.41758999177527e-06, "loss": 2.8769, "step": 202800 }, { "epoch": 0.0082176, "grad_norm": 0.9427737593650818, "learning_rate": 9.416786082106898e-06, "loss": 2.9321, "step": 202810 }, { "epoch": 0.0082432, "grad_norm": 0.7654945254325867, "learning_rate": 9.415982176220512e-06, "loss": 2.8576, "step": 202820 }, { "epoch": 0.0082688, "grad_norm": 0.9658382534980774, "learning_rate": 9.415178274121326e-06, "loss": 3.0523, "step": 202830 }, { "epoch": 0.0082944, "grad_norm": 2.0498545169830322, "learning_rate": 9.414374375814554e-06, "loss": 2.9194, "step": 202840 }, { "epoch": 0.00832, "grad_norm": 0.7877857089042664, "learning_rate": 9.41357048130541e-06, "loss": 2.8806, "step": 202850 }, { "epoch": 0.0083456, "grad_norm": 0.8468320369720459, "learning_rate": 9.412766590599106e-06, "loss": 2.9489, "step": 202860 }, { "epoch": 0.0083712, "grad_norm": 0.766937792301178, "learning_rate": 9.411962703700853e-06, "loss": 2.9294, "step": 202870 }, { "epoch": 0.0083968, "grad_norm": 0.8683730959892273, "learning_rate": 9.411158820615864e-06, "loss": 2.9598, "step": 202880 }, { "epoch": 0.0084224, "grad_norm": 0.8586782813072205, "learning_rate": 9.410354941349356e-06, "loss": 2.7769, "step": 202890 }, { "epoch": 0.008448, "grad_norm": 0.7880978584289551, "learning_rate": 9.409551065906537e-06, "loss": 2.8737, "step": 202900 }, { "epoch": 0.0084736, "grad_norm": 0.9855575561523438, "learning_rate": 9.408747194292625e-06, "loss": 2.8108, "step": 202910 }, { "epoch": 0.0084992, "grad_norm": 1.2076027393341064, "learning_rate": 9.407943326512828e-06, "loss": 2.939, "step": 202920 }, { "epoch": 0.0085248, "grad_norm": 0.709125280380249, "learning_rate": 9.407139462572365e-06, "loss": 2.8709, "step": 202930 }, { "epoch": 0.0085504, "grad_norm": 0.9908324480056763, "learning_rate": 9.406335602476442e-06, "loss": 2.9904, "step": 202940 }, { "epoch": 0.008576, "grad_norm": 0.7854001522064209, "learning_rate": 9.405531746230275e-06, "loss": 2.9516, "step": 202950 }, { "epoch": 0.0086016, "grad_norm": 0.8038404583930969, "learning_rate": 9.404727893839074e-06, "loss": 2.8866, "step": 202960 }, { "epoch": 0.0086272, "grad_norm": 0.7941174507141113, "learning_rate": 9.403924045308057e-06, "loss": 2.9241, "step": 202970 }, { "epoch": 0.0086528, "grad_norm": 0.8059280514717102, "learning_rate": 9.403120200642437e-06, "loss": 3.0488, "step": 202980 }, { "epoch": 0.0086784, "grad_norm": 0.8170306086540222, "learning_rate": 9.40231635984742e-06, "loss": 3.025, "step": 202990 }, { "epoch": 0.008704, "grad_norm": 0.9352321624755859, "learning_rate": 9.401512522928224e-06, "loss": 2.9714, "step": 203000 }, { "epoch": 0.0087296, "grad_norm": 0.8090307116508484, "learning_rate": 9.40070868989006e-06, "loss": 2.8386, "step": 203010 }, { "epoch": 0.0087552, "grad_norm": 0.8542028665542603, "learning_rate": 9.39990486073814e-06, "loss": 3.0101, "step": 203020 }, { "epoch": 0.0087808, "grad_norm": 0.753264844417572, "learning_rate": 9.39910103547768e-06, "loss": 2.9017, "step": 203030 }, { "epoch": 0.0088064, "grad_norm": 0.7753426432609558, "learning_rate": 9.39829721411389e-06, "loss": 2.8311, "step": 203040 }, { "epoch": 0.008832, "grad_norm": 0.8134520053863525, "learning_rate": 9.397493396651979e-06, "loss": 2.908, "step": 203050 }, { "epoch": 0.0088576, "grad_norm": 0.7396135330200195, "learning_rate": 9.396689583097167e-06, "loss": 2.9255, "step": 203060 }, { "epoch": 0.0088832, "grad_norm": 0.9163760542869568, "learning_rate": 9.395885773454661e-06, "loss": 2.7736, "step": 203070 }, { "epoch": 0.0089088, "grad_norm": 0.8354882597923279, "learning_rate": 9.395081967729674e-06, "loss": 2.9344, "step": 203080 }, { "epoch": 0.0089344, "grad_norm": 0.8606446385383606, "learning_rate": 9.394278165927423e-06, "loss": 2.5573, "step": 203090 }, { "epoch": 0.00896, "grad_norm": 0.8023850917816162, "learning_rate": 9.393474368053115e-06, "loss": 3.0276, "step": 203100 }, { "epoch": 0.0089856, "grad_norm": 0.7318295836448669, "learning_rate": 9.392670574111964e-06, "loss": 3.1487, "step": 203110 }, { "epoch": 0.0090112, "grad_norm": 0.7946649193763733, "learning_rate": 9.391866784109183e-06, "loss": 2.6045, "step": 203120 }, { "epoch": 0.0090368, "grad_norm": 0.822217583656311, "learning_rate": 9.391062998049987e-06, "loss": 2.875, "step": 203130 }, { "epoch": 0.0090624, "grad_norm": 0.868829071521759, "learning_rate": 9.39025921593958e-06, "loss": 2.9296, "step": 203140 }, { "epoch": 0.009088, "grad_norm": 1.2217172384262085, "learning_rate": 9.389455437783184e-06, "loss": 2.8297, "step": 203150 }, { "epoch": 0.0091136, "grad_norm": 0.8295103907585144, "learning_rate": 9.388651663586005e-06, "loss": 3.0286, "step": 203160 }, { "epoch": 0.0091392, "grad_norm": 0.7254271507263184, "learning_rate": 9.38784789335326e-06, "loss": 2.9022, "step": 203170 }, { "epoch": 0.0091648, "grad_norm": 0.784457802772522, "learning_rate": 9.387044127090159e-06, "loss": 2.8721, "step": 203180 }, { "epoch": 0.0091904, "grad_norm": 0.7823969125747681, "learning_rate": 9.386240364801913e-06, "loss": 2.8672, "step": 203190 }, { "epoch": 0.009216, "grad_norm": 0.7169515490531921, "learning_rate": 9.385436606493734e-06, "loss": 2.9215, "step": 203200 }, { "epoch": 0.0092416, "grad_norm": 1.230830192565918, "learning_rate": 9.384632852170835e-06, "loss": 2.862, "step": 203210 }, { "epoch": 0.0092672, "grad_norm": 0.8036385178565979, "learning_rate": 9.383829101838432e-06, "loss": 3.0164, "step": 203220 }, { "epoch": 0.0092928, "grad_norm": 0.8340041637420654, "learning_rate": 9.38302535550173e-06, "loss": 2.9936, "step": 203230 }, { "epoch": 0.0093184, "grad_norm": 0.7865079641342163, "learning_rate": 9.382221613165947e-06, "loss": 2.6411, "step": 203240 }, { "epoch": 0.009344, "grad_norm": 1.244512915611267, "learning_rate": 9.38141787483629e-06, "loss": 2.5085, "step": 203250 }, { "epoch": 0.0093696, "grad_norm": 0.8480522632598877, "learning_rate": 9.380614140517976e-06, "loss": 2.9839, "step": 203260 }, { "epoch": 0.0093952, "grad_norm": 0.8955258131027222, "learning_rate": 9.379810410216213e-06, "loss": 2.9722, "step": 203270 }, { "epoch": 0.0094208, "grad_norm": 0.704116940498352, "learning_rate": 9.379006683936215e-06, "loss": 2.7337, "step": 203280 }, { "epoch": 0.0094464, "grad_norm": 0.8890746831893921, "learning_rate": 9.378202961683197e-06, "loss": 2.9647, "step": 203290 }, { "epoch": 0.009472, "grad_norm": 1.0135657787322998, "learning_rate": 9.377399243462363e-06, "loss": 2.9536, "step": 203300 }, { "epoch": 0.0094976, "grad_norm": 0.750969409942627, "learning_rate": 9.376595529278929e-06, "loss": 3.0898, "step": 203310 }, { "epoch": 0.0095232, "grad_norm": 0.7022930979728699, "learning_rate": 9.375791819138109e-06, "loss": 2.7432, "step": 203320 }, { "epoch": 0.0095488, "grad_norm": 0.7861766815185547, "learning_rate": 9.374988113045113e-06, "loss": 3.1812, "step": 203330 }, { "epoch": 0.0095744, "grad_norm": 0.7551541328430176, "learning_rate": 9.374184411005152e-06, "loss": 3.0038, "step": 203340 }, { "epoch": 0.0096, "grad_norm": 0.7505406737327576, "learning_rate": 9.373380713023442e-06, "loss": 2.9665, "step": 203350 }, { "epoch": 0.0096256, "grad_norm": 0.7683351635932922, "learning_rate": 9.37257701910519e-06, "loss": 3.1148, "step": 203360 }, { "epoch": 0.0096512, "grad_norm": 0.8269187211990356, "learning_rate": 9.371773329255608e-06, "loss": 2.9787, "step": 203370 }, { "epoch": 0.0096768, "grad_norm": 0.8819183707237244, "learning_rate": 9.370969643479913e-06, "loss": 2.9995, "step": 203380 }, { "epoch": 0.0097024, "grad_norm": 0.8433733582496643, "learning_rate": 9.370165961783309e-06, "loss": 2.9108, "step": 203390 }, { "epoch": 0.009728, "grad_norm": 0.8759458065032959, "learning_rate": 9.36936228417101e-06, "loss": 2.8245, "step": 203400 }, { "epoch": 0.0097536, "grad_norm": 0.9475337266921997, "learning_rate": 9.368558610648233e-06, "loss": 2.863, "step": 203410 }, { "epoch": 0.0097792, "grad_norm": 0.8088938593864441, "learning_rate": 9.367754941220185e-06, "loss": 2.9333, "step": 203420 }, { "epoch": 0.0098048, "grad_norm": 0.9088096022605896, "learning_rate": 9.366951275892076e-06, "loss": 2.8117, "step": 203430 }, { "epoch": 0.0098304, "grad_norm": 0.8034725785255432, "learning_rate": 9.366147614669118e-06, "loss": 2.9875, "step": 203440 }, { "epoch": 0.009856, "grad_norm": 0.7626102566719055, "learning_rate": 9.365343957556528e-06, "loss": 2.9787, "step": 203450 }, { "epoch": 0.0098816, "grad_norm": 0.851870059967041, "learning_rate": 9.364540304559515e-06, "loss": 2.8561, "step": 203460 }, { "epoch": 0.0099072, "grad_norm": 0.8563599586486816, "learning_rate": 9.363736655683285e-06, "loss": 3.03, "step": 203470 }, { "epoch": 0.0099328, "grad_norm": 1.04728102684021, "learning_rate": 9.362933010933056e-06, "loss": 2.9534, "step": 203480 }, { "epoch": 0.0099584, "grad_norm": 0.834470808506012, "learning_rate": 9.362129370314033e-06, "loss": 3.0071, "step": 203490 }, { "epoch": 0.009984, "grad_norm": 0.7600909471511841, "learning_rate": 9.361325733831434e-06, "loss": 2.8461, "step": 203500 }, { "epoch": 0.0100096, "grad_norm": 0.7514165639877319, "learning_rate": 9.360522101490471e-06, "loss": 2.9186, "step": 203510 }, { "epoch": 0.0100352, "grad_norm": 0.8338351845741272, "learning_rate": 9.35971847329635e-06, "loss": 2.6981, "step": 203520 }, { "epoch": 0.0100608, "grad_norm": 0.7719930410385132, "learning_rate": 9.358914849254285e-06, "loss": 2.7932, "step": 203530 }, { "epoch": 0.0100864, "grad_norm": 0.8781935572624207, "learning_rate": 9.358111229369486e-06, "loss": 3.0384, "step": 203540 }, { "epoch": 0.010112, "grad_norm": 0.7583976984024048, "learning_rate": 9.357307613647168e-06, "loss": 2.9321, "step": 203550 }, { "epoch": 0.0101376, "grad_norm": 0.8380942344665527, "learning_rate": 9.356504002092536e-06, "loss": 2.8219, "step": 203560 }, { "epoch": 0.0101632, "grad_norm": 0.8089441657066345, "learning_rate": 9.355700394710805e-06, "loss": 2.8081, "step": 203570 }, { "epoch": 0.0101888, "grad_norm": 0.8495820760726929, "learning_rate": 9.354896791507188e-06, "loss": 3.0726, "step": 203580 }, { "epoch": 0.0102144, "grad_norm": 0.7836809754371643, "learning_rate": 9.35409319248689e-06, "loss": 3.0695, "step": 203590 }, { "epoch": 0.01024, "grad_norm": 0.8392677903175354, "learning_rate": 9.353289597655128e-06, "loss": 2.8863, "step": 203600 }, { "epoch": 0.0102656, "grad_norm": 0.7109405994415283, "learning_rate": 9.352486007017111e-06, "loss": 2.7376, "step": 203610 }, { "epoch": 0.0102912, "grad_norm": 1.2522448301315308, "learning_rate": 9.351682420578052e-06, "loss": 2.9739, "step": 203620 }, { "epoch": 0.0103168, "grad_norm": 0.7425383925437927, "learning_rate": 9.350878838343159e-06, "loss": 2.9577, "step": 203630 }, { "epoch": 0.0103424, "grad_norm": 0.7819759249687195, "learning_rate": 9.350075260317643e-06, "loss": 2.9652, "step": 203640 }, { "epoch": 0.010368, "grad_norm": 0.7541211843490601, "learning_rate": 9.349271686506718e-06, "loss": 3.0764, "step": 203650 }, { "epoch": 0.0103936, "grad_norm": 0.74814373254776, "learning_rate": 9.348468116915588e-06, "loss": 2.9635, "step": 203660 }, { "epoch": 0.0104192, "grad_norm": 0.7891543507575989, "learning_rate": 9.347664551549473e-06, "loss": 3.0298, "step": 203670 }, { "epoch": 0.0104448, "grad_norm": 1.3467453718185425, "learning_rate": 9.346860990413581e-06, "loss": 2.9618, "step": 203680 }, { "epoch": 0.0104704, "grad_norm": 1.156760811805725, "learning_rate": 9.34605743351312e-06, "loss": 3.2138, "step": 203690 }, { "epoch": 0.010496, "grad_norm": 0.8040039539337158, "learning_rate": 9.345253880853304e-06, "loss": 2.7316, "step": 203700 }, { "epoch": 0.0105216, "grad_norm": 0.6930514574050903, "learning_rate": 9.344450332439344e-06, "loss": 3.0027, "step": 203710 }, { "epoch": 0.0105472, "grad_norm": 0.7141439318656921, "learning_rate": 9.343646788276448e-06, "loss": 3.1243, "step": 203720 }, { "epoch": 0.0105728, "grad_norm": 0.7294965386390686, "learning_rate": 9.342843248369827e-06, "loss": 2.9059, "step": 203730 }, { "epoch": 0.0105984, "grad_norm": 0.7395519018173218, "learning_rate": 9.342039712724692e-06, "loss": 3.1182, "step": 203740 }, { "epoch": 0.010624, "grad_norm": 1.0174353122711182, "learning_rate": 9.341236181346259e-06, "loss": 3.0664, "step": 203750 }, { "epoch": 0.0106496, "grad_norm": 0.8397086262702942, "learning_rate": 9.340432654239731e-06, "loss": 2.9847, "step": 203760 }, { "epoch": 0.0106752, "grad_norm": 0.8854563236236572, "learning_rate": 9.339629131410323e-06, "loss": 3.0369, "step": 203770 }, { "epoch": 0.0107008, "grad_norm": 0.7584291100502014, "learning_rate": 9.338825612863246e-06, "loss": 2.8432, "step": 203780 }, { "epoch": 0.0107264, "grad_norm": 0.8132201433181763, "learning_rate": 9.338022098603706e-06, "loss": 2.8856, "step": 203790 }, { "epoch": 0.010752, "grad_norm": 0.8620318174362183, "learning_rate": 9.337218588636919e-06, "loss": 2.9531, "step": 203800 }, { "epoch": 0.0107776, "grad_norm": 0.7013142108917236, "learning_rate": 9.33641508296809e-06, "loss": 2.9495, "step": 203810 }, { "epoch": 0.0108032, "grad_norm": 0.7262905836105347, "learning_rate": 9.335611581602438e-06, "loss": 2.8356, "step": 203820 }, { "epoch": 0.0108288, "grad_norm": 0.7303122878074646, "learning_rate": 9.33480808454516e-06, "loss": 2.9763, "step": 203830 }, { "epoch": 0.0108544, "grad_norm": 0.9889902472496033, "learning_rate": 9.334004591801483e-06, "loss": 2.7186, "step": 203840 }, { "epoch": 0.01088, "grad_norm": 0.7929050326347351, "learning_rate": 9.333201103376605e-06, "loss": 2.9238, "step": 203850 }, { "epoch": 0.0109056, "grad_norm": 0.8773273229598999, "learning_rate": 9.332397619275743e-06, "loss": 2.8792, "step": 203860 }, { "epoch": 0.0109312, "grad_norm": 0.7568305134773254, "learning_rate": 9.331594139504105e-06, "loss": 2.872, "step": 203870 }, { "epoch": 0.0109568, "grad_norm": 0.7612236738204956, "learning_rate": 9.330790664066902e-06, "loss": 2.8968, "step": 203880 }, { "epoch": 0.0109824, "grad_norm": 0.911867618560791, "learning_rate": 9.329987192969342e-06, "loss": 2.866, "step": 203890 }, { "epoch": 0.011008, "grad_norm": 2.5202419757843018, "learning_rate": 9.329183726216635e-06, "loss": 2.9195, "step": 203900 }, { "epoch": 0.0110336, "grad_norm": 0.8208234906196594, "learning_rate": 9.328380263813999e-06, "loss": 2.9927, "step": 203910 }, { "epoch": 0.0110592, "grad_norm": 0.8226832151412964, "learning_rate": 9.327576805766633e-06, "loss": 2.9037, "step": 203920 }, { "epoch": 0.0110848, "grad_norm": 0.7744237780570984, "learning_rate": 9.326773352079754e-06, "loss": 3.0293, "step": 203930 }, { "epoch": 0.0111104, "grad_norm": 0.787834882736206, "learning_rate": 9.325969902758571e-06, "loss": 2.8038, "step": 203940 }, { "epoch": 0.011136, "grad_norm": 0.9449793100357056, "learning_rate": 9.325166457808298e-06, "loss": 3.0984, "step": 203950 }, { "epoch": 0.0111616, "grad_norm": 0.8026636242866516, "learning_rate": 9.324363017234138e-06, "loss": 2.8887, "step": 203960 }, { "epoch": 0.0111872, "grad_norm": 0.9927797913551331, "learning_rate": 9.323559581041303e-06, "loss": 2.9654, "step": 203970 }, { "epoch": 0.0112128, "grad_norm": 0.7648754119873047, "learning_rate": 9.322756149235008e-06, "loss": 2.8143, "step": 203980 }, { "epoch": 0.0112384, "grad_norm": 0.9824708104133606, "learning_rate": 9.321952721820455e-06, "loss": 2.8502, "step": 203990 }, { "epoch": 0.011264, "grad_norm": 1.2520633935928345, "learning_rate": 9.32114929880286e-06, "loss": 3.0467, "step": 204000 }, { "epoch": 0.0112896, "grad_norm": 0.7519276738166809, "learning_rate": 9.320345880187429e-06, "loss": 2.8001, "step": 204010 }, { "epoch": 0.0113152, "grad_norm": 0.8694254755973816, "learning_rate": 9.319542465979376e-06, "loss": 2.9565, "step": 204020 }, { "epoch": 0.0113408, "grad_norm": 0.8236536383628845, "learning_rate": 9.31873905618391e-06, "loss": 2.9163, "step": 204030 }, { "epoch": 0.0113664, "grad_norm": 0.7914469242095947, "learning_rate": 9.317935650806242e-06, "loss": 2.7901, "step": 204040 }, { "epoch": 0.011392, "grad_norm": 0.983726441860199, "learning_rate": 9.317132249851579e-06, "loss": 3.0205, "step": 204050 }, { "epoch": 0.0114176, "grad_norm": 0.8833231329917908, "learning_rate": 9.316328853325129e-06, "loss": 2.944, "step": 204060 }, { "epoch": 0.0114432, "grad_norm": 0.7593960762023926, "learning_rate": 9.31552546123211e-06, "loss": 2.8132, "step": 204070 }, { "epoch": 0.0114688, "grad_norm": 1.0508390665054321, "learning_rate": 9.314722073577723e-06, "loss": 2.8041, "step": 204080 }, { "epoch": 0.0114944, "grad_norm": 0.6838204264640808, "learning_rate": 9.31391869036718e-06, "loss": 2.6364, "step": 204090 }, { "epoch": 0.01152, "grad_norm": 0.7178101539611816, "learning_rate": 9.313115311605695e-06, "loss": 2.7547, "step": 204100 }, { "epoch": 0.0115456, "grad_norm": 0.9310528039932251, "learning_rate": 9.312311937298474e-06, "loss": 2.7248, "step": 204110 }, { "epoch": 0.0115712, "grad_norm": 2.1748740673065186, "learning_rate": 9.311508567450727e-06, "loss": 3.0004, "step": 204120 }, { "epoch": 0.0115968, "grad_norm": 0.7988707423210144, "learning_rate": 9.310705202067662e-06, "loss": 3.0818, "step": 204130 }, { "epoch": 0.0116224, "grad_norm": 0.8041377663612366, "learning_rate": 9.309901841154491e-06, "loss": 2.7893, "step": 204140 }, { "epoch": 0.011648, "grad_norm": 0.827399730682373, "learning_rate": 9.309098484716427e-06, "loss": 3.1037, "step": 204150 }, { "epoch": 0.0116736, "grad_norm": 0.7542697787284851, "learning_rate": 9.308295132758672e-06, "loss": 2.8119, "step": 204160 }, { "epoch": 0.0116992, "grad_norm": 0.8583202958106995, "learning_rate": 9.30749178528644e-06, "loss": 2.8373, "step": 204170 }, { "epoch": 0.0117248, "grad_norm": 0.7971251606941223, "learning_rate": 9.306688442304937e-06, "loss": 3.0366, "step": 204180 }, { "epoch": 0.0117504, "grad_norm": 0.8226079940795898, "learning_rate": 9.305885103819379e-06, "loss": 2.7199, "step": 204190 }, { "epoch": 0.011776, "grad_norm": 0.807599663734436, "learning_rate": 9.305081769834972e-06, "loss": 3.0529, "step": 204200 }, { "epoch": 0.0118016, "grad_norm": 0.8156586289405823, "learning_rate": 9.304278440356923e-06, "loss": 3.0916, "step": 204210 }, { "epoch": 0.0118272, "grad_norm": 0.7442139983177185, "learning_rate": 9.303475115390443e-06, "loss": 2.6438, "step": 204220 }, { "epoch": 0.0118528, "grad_norm": 0.9058976173400879, "learning_rate": 9.302671794940743e-06, "loss": 2.8768, "step": 204230 }, { "epoch": 0.0118784, "grad_norm": 0.9002186059951782, "learning_rate": 9.301868479013033e-06, "loss": 3.1009, "step": 204240 }, { "epoch": 0.011904, "grad_norm": 1.1022483110427856, "learning_rate": 9.301065167612518e-06, "loss": 2.9279, "step": 204250 }, { "epoch": 0.0119296, "grad_norm": 0.9063594341278076, "learning_rate": 9.30026186074441e-06, "loss": 2.6874, "step": 204260 }, { "epoch": 0.0119552, "grad_norm": 1.1042451858520508, "learning_rate": 9.299458558413919e-06, "loss": 2.9332, "step": 204270 }, { "epoch": 0.0119808, "grad_norm": 0.8238090872764587, "learning_rate": 9.298655260626252e-06, "loss": 2.8258, "step": 204280 }, { "epoch": 0.0120064, "grad_norm": 0.7163366675376892, "learning_rate": 9.29785196738662e-06, "loss": 3.0542, "step": 204290 }, { "epoch": 0.012032, "grad_norm": 0.8807970285415649, "learning_rate": 9.297048678700228e-06, "loss": 3.0038, "step": 204300 }, { "epoch": 0.0120576, "grad_norm": 0.8531003594398499, "learning_rate": 9.296245394572295e-06, "loss": 2.9079, "step": 204310 }, { "epoch": 0.0120832, "grad_norm": 1.4248265027999878, "learning_rate": 9.295442115008018e-06, "loss": 2.9104, "step": 204320 }, { "epoch": 0.0121088, "grad_norm": 0.8371783494949341, "learning_rate": 9.294638840012613e-06, "loss": 2.9012, "step": 204330 }, { "epoch": 0.0121344, "grad_norm": 0.8374404311180115, "learning_rate": 9.293835569591288e-06, "loss": 3.1021, "step": 204340 }, { "epoch": 0.01216, "grad_norm": 0.8227185010910034, "learning_rate": 9.293032303749254e-06, "loss": 3.0083, "step": 204350 }, { "epoch": 0.0121856, "grad_norm": 0.710069477558136, "learning_rate": 9.29222904249171e-06, "loss": 2.9926, "step": 204360 }, { "epoch": 0.0122112, "grad_norm": 1.1883399486541748, "learning_rate": 9.291425785823879e-06, "loss": 3.1636, "step": 204370 }, { "epoch": 0.0122368, "grad_norm": 0.7747491002082825, "learning_rate": 9.290622533750962e-06, "loss": 2.8466, "step": 204380 }, { "epoch": 0.0122624, "grad_norm": 0.7804075479507446, "learning_rate": 9.289819286278168e-06, "loss": 2.8965, "step": 204390 }, { "epoch": 0.012288, "grad_norm": 0.8153262138366699, "learning_rate": 9.289016043410712e-06, "loss": 3.1063, "step": 204400 }, { "epoch": 0.0123136, "grad_norm": 0.7968229055404663, "learning_rate": 9.288212805153793e-06, "loss": 3.1061, "step": 204410 }, { "epoch": 0.0123392, "grad_norm": 0.8083781003952026, "learning_rate": 9.287409571512626e-06, "loss": 2.9229, "step": 204420 }, { "epoch": 0.0123648, "grad_norm": 0.7541447281837463, "learning_rate": 9.286606342492419e-06, "loss": 2.8553, "step": 204430 }, { "epoch": 0.0123904, "grad_norm": 2.6431291103363037, "learning_rate": 9.285803118098382e-06, "loss": 2.8336, "step": 204440 }, { "epoch": 0.012416, "grad_norm": 0.7091284990310669, "learning_rate": 9.284999898335721e-06, "loss": 3.0196, "step": 204450 }, { "epoch": 0.0124416, "grad_norm": 0.8604588508605957, "learning_rate": 9.284196683209644e-06, "loss": 2.9574, "step": 204460 }, { "epoch": 0.0124672, "grad_norm": 0.9171689748764038, "learning_rate": 9.283393472725363e-06, "loss": 3.0763, "step": 204470 }, { "epoch": 0.0124928, "grad_norm": 0.7502530217170715, "learning_rate": 9.282590266888085e-06, "loss": 3.1432, "step": 204480 }, { "epoch": 0.0125184, "grad_norm": 0.7605189681053162, "learning_rate": 9.281787065703017e-06, "loss": 3.3938, "step": 204490 }, { "epoch": 0.012544, "grad_norm": 2.3206279277801514, "learning_rate": 9.28098386917537e-06, "loss": 2.9114, "step": 204500 }, { "epoch": 0.0125696, "grad_norm": 0.7550814747810364, "learning_rate": 9.280180677310352e-06, "loss": 2.9755, "step": 204510 }, { "epoch": 0.0125952, "grad_norm": 0.8420106172561646, "learning_rate": 9.27937749011317e-06, "loss": 2.8688, "step": 204520 }, { "epoch": 0.0126208, "grad_norm": 0.7764305472373962, "learning_rate": 9.27857430758903e-06, "loss": 2.9484, "step": 204530 }, { "epoch": 0.0126464, "grad_norm": 1.2674447298049927, "learning_rate": 9.277771129743147e-06, "loss": 2.8075, "step": 204540 }, { "epoch": 0.012672, "grad_norm": 0.8737602233886719, "learning_rate": 9.276967956580727e-06, "loss": 2.7635, "step": 204550 }, { "epoch": 0.0126976, "grad_norm": 0.7933999300003052, "learning_rate": 9.276164788106977e-06, "loss": 3.0826, "step": 204560 }, { "epoch": 0.0127232, "grad_norm": 0.7949358820915222, "learning_rate": 9.275361624327109e-06, "loss": 2.8867, "step": 204570 }, { "epoch": 0.0127488, "grad_norm": 0.9551724195480347, "learning_rate": 9.274558465246324e-06, "loss": 3.0173, "step": 204580 }, { "epoch": 0.0127744, "grad_norm": 1.0332508087158203, "learning_rate": 9.273755310869836e-06, "loss": 2.9821, "step": 204590 }, { "epoch": 0.0128, "grad_norm": 0.7603543996810913, "learning_rate": 9.272952161202855e-06, "loss": 2.9999, "step": 204600 }, { "epoch": 0.0128256, "grad_norm": 0.8386486172676086, "learning_rate": 9.27214901625058e-06, "loss": 2.8998, "step": 204610 }, { "epoch": 0.0128512, "grad_norm": 0.8287830352783203, "learning_rate": 9.27134587601823e-06, "loss": 3.0471, "step": 204620 }, { "epoch": 0.0128768, "grad_norm": 0.7283585071563721, "learning_rate": 9.270542740511006e-06, "loss": 3.0279, "step": 204630 }, { "epoch": 0.0129024, "grad_norm": 0.8172305226325989, "learning_rate": 9.26973960973412e-06, "loss": 3.06, "step": 204640 }, { "epoch": 0.012928, "grad_norm": 0.7758508920669556, "learning_rate": 9.268936483692778e-06, "loss": 2.9432, "step": 204650 }, { "epoch": 0.0129536, "grad_norm": 0.9199916124343872, "learning_rate": 9.268133362392186e-06, "loss": 2.8902, "step": 204660 }, { "epoch": 0.0129792, "grad_norm": 0.7199216485023499, "learning_rate": 9.26733024583756e-06, "loss": 3.0185, "step": 204670 }, { "epoch": 0.0130048, "grad_norm": 0.7072306275367737, "learning_rate": 9.266527134034098e-06, "loss": 3.3838, "step": 204680 }, { "epoch": 0.0130304, "grad_norm": 0.7246655821800232, "learning_rate": 9.265724026987014e-06, "loss": 3.0498, "step": 204690 }, { "epoch": 0.013056, "grad_norm": 0.8046821355819702, "learning_rate": 9.264920924701513e-06, "loss": 3.0278, "step": 204700 }, { "epoch": 0.0130816, "grad_norm": 1.2849873304367065, "learning_rate": 9.264117827182804e-06, "loss": 2.8797, "step": 204710 }, { "epoch": 0.0131072, "grad_norm": 1.0846619606018066, "learning_rate": 9.263314734436097e-06, "loss": 3.0037, "step": 204720 }, { "epoch": 0.0131328, "grad_norm": 0.7434141635894775, "learning_rate": 9.2625116464666e-06, "loss": 2.9074, "step": 204730 }, { "epoch": 0.0131584, "grad_norm": 0.8676333427429199, "learning_rate": 9.261708563279517e-06, "loss": 2.9497, "step": 204740 }, { "epoch": 0.013184, "grad_norm": 1.5073236227035522, "learning_rate": 9.260905484880058e-06, "loss": 2.9101, "step": 204750 }, { "epoch": 0.0132096, "grad_norm": 0.7086648941040039, "learning_rate": 9.260102411273432e-06, "loss": 2.957, "step": 204760 }, { "epoch": 0.0132352, "grad_norm": 0.7748920917510986, "learning_rate": 9.259299342464845e-06, "loss": 3.0899, "step": 204770 }, { "epoch": 0.0132608, "grad_norm": 1.146548867225647, "learning_rate": 9.258496278459503e-06, "loss": 3.2058, "step": 204780 }, { "epoch": 0.0132864, "grad_norm": 0.9305663108825684, "learning_rate": 9.257693219262615e-06, "loss": 2.8259, "step": 204790 }, { "epoch": 0.013312, "grad_norm": 0.8156698346138, "learning_rate": 9.256890164879394e-06, "loss": 2.7821, "step": 204800 }, { "epoch": 0.0133376, "grad_norm": 2.334768533706665, "learning_rate": 9.256087115315037e-06, "loss": 3.8105, "step": 204810 }, { "epoch": 0.0133632, "grad_norm": 0.803459107875824, "learning_rate": 9.25528407057476e-06, "loss": 2.8122, "step": 204820 }, { "epoch": 0.0133888, "grad_norm": 0.9319550395011902, "learning_rate": 9.254481030663767e-06, "loss": 2.9697, "step": 204830 }, { "epoch": 0.0134144, "grad_norm": 0.9066783785820007, "learning_rate": 9.253677995587269e-06, "loss": 3.0516, "step": 204840 }, { "epoch": 0.01344, "grad_norm": 0.8049957156181335, "learning_rate": 9.25287496535047e-06, "loss": 2.8383, "step": 204850 }, { "epoch": 0.0134656, "grad_norm": 0.7917327880859375, "learning_rate": 9.252071939958575e-06, "loss": 2.9786, "step": 204860 }, { "epoch": 0.0134912, "grad_norm": 0.6928421854972839, "learning_rate": 9.2512689194168e-06, "loss": 3.1232, "step": 204870 }, { "epoch": 0.0135168, "grad_norm": 0.7512383460998535, "learning_rate": 9.250465903730338e-06, "loss": 2.8109, "step": 204880 }, { "epoch": 0.0135424, "grad_norm": 0.744546115398407, "learning_rate": 9.249662892904413e-06, "loss": 3.0044, "step": 204890 }, { "epoch": 0.013568, "grad_norm": 0.8039630651473999, "learning_rate": 9.248859886944223e-06, "loss": 2.879, "step": 204900 }, { "epoch": 0.0135936, "grad_norm": 0.7450879216194153, "learning_rate": 9.248056885854976e-06, "loss": 2.9974, "step": 204910 }, { "epoch": 0.0136192, "grad_norm": 0.7776108384132385, "learning_rate": 9.247253889641881e-06, "loss": 2.7543, "step": 204920 }, { "epoch": 0.0136448, "grad_norm": 0.7337761521339417, "learning_rate": 9.246450898310147e-06, "loss": 3.0162, "step": 204930 }, { "epoch": 0.0136704, "grad_norm": 0.7589050531387329, "learning_rate": 9.245647911864976e-06, "loss": 3.1607, "step": 204940 }, { "epoch": 0.013696, "grad_norm": 1.7924078702926636, "learning_rate": 9.244844930311577e-06, "loss": 3.4473, "step": 204950 }, { "epoch": 0.0137216, "grad_norm": 0.7053468227386475, "learning_rate": 9.244041953655161e-06, "loss": 2.8491, "step": 204960 }, { "epoch": 0.0137472, "grad_norm": 0.770986795425415, "learning_rate": 9.24323898190093e-06, "loss": 2.8857, "step": 204970 }, { "epoch": 0.0137728, "grad_norm": 0.7257314324378967, "learning_rate": 9.242436015054093e-06, "loss": 3.0067, "step": 204980 }, { "epoch": 0.0137984, "grad_norm": 0.7823812365531921, "learning_rate": 9.241633053119857e-06, "loss": 2.9076, "step": 204990 }, { "epoch": 0.013824, "grad_norm": 0.7687897682189941, "learning_rate": 9.240830096103432e-06, "loss": 2.9405, "step": 205000 }, { "epoch": 0.0138496, "grad_norm": 0.8195009231567383, "learning_rate": 9.24002714401002e-06, "loss": 2.9233, "step": 205010 }, { "epoch": 0.0138752, "grad_norm": 0.8472775220870972, "learning_rate": 9.239224196844828e-06, "loss": 2.9451, "step": 205020 }, { "epoch": 0.0139008, "grad_norm": 0.7736210227012634, "learning_rate": 9.238421254613066e-06, "loss": 2.8931, "step": 205030 }, { "epoch": 0.0139264, "grad_norm": 0.7639421820640564, "learning_rate": 9.237618317319942e-06, "loss": 3.0118, "step": 205040 }, { "epoch": 0.013952, "grad_norm": 0.7927337288856506, "learning_rate": 9.236815384970652e-06, "loss": 2.6401, "step": 205050 }, { "epoch": 0.0139776, "grad_norm": 0.7922711372375488, "learning_rate": 9.23601245757042e-06, "loss": 2.9664, "step": 205060 }, { "epoch": 0.0140032, "grad_norm": 0.7667341828346252, "learning_rate": 9.235209535124442e-06, "loss": 3.1468, "step": 205070 }, { "epoch": 0.0140288, "grad_norm": 0.7455588579177856, "learning_rate": 9.234406617637928e-06, "loss": 2.9648, "step": 205080 }, { "epoch": 0.0140544, "grad_norm": 0.873620867729187, "learning_rate": 9.233603705116084e-06, "loss": 2.8291, "step": 205090 }, { "epoch": 0.01408, "grad_norm": 0.720570981502533, "learning_rate": 9.232800797564115e-06, "loss": 2.9632, "step": 205100 }, { "epoch": 0.0141056, "grad_norm": 0.7358753681182861, "learning_rate": 9.231997894987229e-06, "loss": 2.8797, "step": 205110 }, { "epoch": 0.0141312, "grad_norm": 0.864628791809082, "learning_rate": 9.231194997390632e-06, "loss": 2.9651, "step": 205120 }, { "epoch": 0.0141568, "grad_norm": 0.8519470691680908, "learning_rate": 9.230392104779533e-06, "loss": 2.9722, "step": 205130 }, { "epoch": 0.0141824, "grad_norm": 0.7973008155822754, "learning_rate": 9.229589217159136e-06, "loss": 2.8509, "step": 205140 }, { "epoch": 0.014208, "grad_norm": 0.9905625581741333, "learning_rate": 9.228786334534647e-06, "loss": 3.1907, "step": 205150 }, { "epoch": 0.0142336, "grad_norm": 0.9284759759902954, "learning_rate": 9.227983456911275e-06, "loss": 3.1163, "step": 205160 }, { "epoch": 0.0142592, "grad_norm": 0.8861280083656311, "learning_rate": 9.227180584294224e-06, "loss": 2.7314, "step": 205170 }, { "epoch": 0.0142848, "grad_norm": 0.7292516827583313, "learning_rate": 9.226377716688703e-06, "loss": 3.0497, "step": 205180 }, { "epoch": 0.0143104, "grad_norm": 0.7105713486671448, "learning_rate": 9.225574854099915e-06, "loss": 3.0347, "step": 205190 }, { "epoch": 0.014336, "grad_norm": 0.7690906524658203, "learning_rate": 9.22477199653307e-06, "loss": 3.0336, "step": 205200 }, { "epoch": 0.0143616, "grad_norm": 0.8845694065093994, "learning_rate": 9.22396914399337e-06, "loss": 2.7553, "step": 205210 }, { "epoch": 0.0143872, "grad_norm": 1.2627424001693726, "learning_rate": 9.223166296486026e-06, "loss": 2.8956, "step": 205220 }, { "epoch": 0.0144128, "grad_norm": 1.0396803617477417, "learning_rate": 9.222363454016238e-06, "loss": 3.0162, "step": 205230 }, { "epoch": 0.0144384, "grad_norm": 0.8208797574043274, "learning_rate": 9.22156061658922e-06, "loss": 2.9307, "step": 205240 }, { "epoch": 0.014464, "grad_norm": 1.1201742887496948, "learning_rate": 9.220757784210174e-06, "loss": 2.9962, "step": 205250 }, { "epoch": 0.0144896, "grad_norm": 0.7013896107673645, "learning_rate": 9.219954956884309e-06, "loss": 2.9501, "step": 205260 }, { "epoch": 0.0145152, "grad_norm": 0.7163876295089722, "learning_rate": 9.219152134616827e-06, "loss": 2.9587, "step": 205270 }, { "epoch": 0.0145408, "grad_norm": 0.8045738935470581, "learning_rate": 9.218349317412936e-06, "loss": 2.8123, "step": 205280 }, { "epoch": 0.0145664, "grad_norm": 0.8077322840690613, "learning_rate": 9.217546505277842e-06, "loss": 3.2488, "step": 205290 }, { "epoch": 0.014592, "grad_norm": 0.7617099285125732, "learning_rate": 9.216743698216752e-06, "loss": 2.8892, "step": 205300 }, { "epoch": 0.0146176, "grad_norm": 0.8108129501342773, "learning_rate": 9.215940896234869e-06, "loss": 2.8718, "step": 205310 }, { "epoch": 0.0146432, "grad_norm": 0.7498009204864502, "learning_rate": 9.215138099337402e-06, "loss": 2.9024, "step": 205320 }, { "epoch": 0.0146688, "grad_norm": 0.8520826697349548, "learning_rate": 9.214335307529555e-06, "loss": 2.8491, "step": 205330 }, { "epoch": 0.0146944, "grad_norm": 0.7909996509552002, "learning_rate": 9.213532520816536e-06, "loss": 2.8735, "step": 205340 }, { "epoch": 0.01472, "grad_norm": 0.7446692585945129, "learning_rate": 9.212729739203547e-06, "loss": 2.8541, "step": 205350 }, { "epoch": 0.0147456, "grad_norm": 0.7036274671554565, "learning_rate": 9.2119269626958e-06, "loss": 2.7598, "step": 205360 }, { "epoch": 0.0147712, "grad_norm": 0.8572657108306885, "learning_rate": 9.211124191298494e-06, "loss": 3.0173, "step": 205370 }, { "epoch": 0.0147968, "grad_norm": 0.858508825302124, "learning_rate": 9.21032142501684e-06, "loss": 2.8956, "step": 205380 }, { "epoch": 0.0148224, "grad_norm": 0.7680807709693909, "learning_rate": 9.20951866385604e-06, "loss": 2.9051, "step": 205390 }, { "epoch": 0.014848, "grad_norm": 0.7107660174369812, "learning_rate": 9.2087159078213e-06, "loss": 2.8782, "step": 205400 }, { "epoch": 0.0148736, "grad_norm": 0.734760046005249, "learning_rate": 9.20791315691783e-06, "loss": 2.8591, "step": 205410 }, { "epoch": 0.0148992, "grad_norm": 0.7372239828109741, "learning_rate": 9.207110411150832e-06, "loss": 2.892, "step": 205420 }, { "epoch": 0.0149248, "grad_norm": 0.9208599925041199, "learning_rate": 9.206307670525512e-06, "loss": 3.0418, "step": 205430 }, { "epoch": 0.0149504, "grad_norm": 0.773535966873169, "learning_rate": 9.205504935047076e-06, "loss": 2.9844, "step": 205440 }, { "epoch": 0.014976, "grad_norm": 0.8100717663764954, "learning_rate": 9.204702204720729e-06, "loss": 3.2567, "step": 205450 }, { "epoch": 0.0150016, "grad_norm": 1.25120210647583, "learning_rate": 9.20389947955168e-06, "loss": 2.8321, "step": 205460 }, { "epoch": 0.0150272, "grad_norm": 0.8201779723167419, "learning_rate": 9.203096759545127e-06, "loss": 2.8724, "step": 205470 }, { "epoch": 0.0150528, "grad_norm": 0.7748227119445801, "learning_rate": 9.20229404470628e-06, "loss": 2.9029, "step": 205480 }, { "epoch": 0.0150784, "grad_norm": 0.9367896318435669, "learning_rate": 9.201491335040347e-06, "loss": 2.913, "step": 205490 }, { "epoch": 0.015104, "grad_norm": 0.788055956363678, "learning_rate": 9.20068863055253e-06, "loss": 2.9796, "step": 205500 }, { "epoch": 0.0151296, "grad_norm": 0.8274694085121155, "learning_rate": 9.199885931248033e-06, "loss": 2.9714, "step": 205510 }, { "epoch": 0.0151552, "grad_norm": 1.0248647928237915, "learning_rate": 9.199083237132062e-06, "loss": 2.7833, "step": 205520 }, { "epoch": 0.0151808, "grad_norm": 0.7487801909446716, "learning_rate": 9.198280548209829e-06, "loss": 2.8788, "step": 205530 }, { "epoch": 0.0152064, "grad_norm": 0.880756676197052, "learning_rate": 9.197477864486529e-06, "loss": 2.9444, "step": 205540 }, { "epoch": 0.015232, "grad_norm": 0.8293062448501587, "learning_rate": 9.19667518596737e-06, "loss": 2.8844, "step": 205550 }, { "epoch": 0.0152576, "grad_norm": 0.7377788424491882, "learning_rate": 9.195872512657565e-06, "loss": 2.9857, "step": 205560 }, { "epoch": 0.0152832, "grad_norm": 0.8716056942939758, "learning_rate": 9.195069844562304e-06, "loss": 2.8114, "step": 205570 }, { "epoch": 0.0153088, "grad_norm": 0.7671197652816772, "learning_rate": 9.194267181686808e-06, "loss": 2.768, "step": 205580 }, { "epoch": 0.0153344, "grad_norm": 0.7620956897735596, "learning_rate": 9.193464524036273e-06, "loss": 2.8825, "step": 205590 }, { "epoch": 0.01536, "grad_norm": 0.7673085927963257, "learning_rate": 9.192661871615906e-06, "loss": 2.8085, "step": 205600 }, { "epoch": 0.0153856, "grad_norm": 0.7234835624694824, "learning_rate": 9.191859224430914e-06, "loss": 3.1668, "step": 205610 }, { "epoch": 0.0154112, "grad_norm": 0.9277253746986389, "learning_rate": 9.191056582486501e-06, "loss": 2.7606, "step": 205620 }, { "epoch": 0.0154368, "grad_norm": 0.9055207967758179, "learning_rate": 9.190253945787869e-06, "loss": 2.7498, "step": 205630 }, { "epoch": 0.0154624, "grad_norm": 0.8103088736534119, "learning_rate": 9.189451314340224e-06, "loss": 2.8364, "step": 205640 }, { "epoch": 0.015488, "grad_norm": 0.7024251818656921, "learning_rate": 9.188648688148774e-06, "loss": 2.9817, "step": 205650 }, { "epoch": 0.0155136, "grad_norm": 0.833440899848938, "learning_rate": 9.187846067218722e-06, "loss": 3.0557, "step": 205660 }, { "epoch": 0.0155392, "grad_norm": 1.0132769346237183, "learning_rate": 9.187043451555271e-06, "loss": 3.1171, "step": 205670 }, { "epoch": 0.0155648, "grad_norm": 0.8586598038673401, "learning_rate": 9.186240841163627e-06, "loss": 2.9124, "step": 205680 }, { "epoch": 0.0155904, "grad_norm": 0.7839391827583313, "learning_rate": 9.185438236048996e-06, "loss": 2.8214, "step": 205690 }, { "epoch": 0.015616, "grad_norm": 1.0939760208129883, "learning_rate": 9.184635636216581e-06, "loss": 2.9104, "step": 205700 }, { "epoch": 0.0156416, "grad_norm": 0.8412904739379883, "learning_rate": 9.183833041671585e-06, "loss": 2.8314, "step": 205710 }, { "epoch": 0.0156672, "grad_norm": 0.8880844116210938, "learning_rate": 9.183030452419217e-06, "loss": 2.8948, "step": 205720 }, { "epoch": 0.0156928, "grad_norm": 0.7635259628295898, "learning_rate": 9.182227868464682e-06, "loss": 2.9768, "step": 205730 }, { "epoch": 0.0157184, "grad_norm": 0.9714073538780212, "learning_rate": 9.181425289813178e-06, "loss": 2.9675, "step": 205740 }, { "epoch": 0.015744, "grad_norm": 1.5396615266799927, "learning_rate": 9.180622716469913e-06, "loss": 3.2715, "step": 205750 }, { "epoch": 0.0157696, "grad_norm": 0.7546941041946411, "learning_rate": 9.179820148440092e-06, "loss": 2.8502, "step": 205760 }, { "epoch": 0.0157952, "grad_norm": 1.0665546655654907, "learning_rate": 9.17901758572892e-06, "loss": 3.0355, "step": 205770 }, { "epoch": 0.0158208, "grad_norm": 0.9630105495452881, "learning_rate": 9.178215028341605e-06, "loss": 2.9098, "step": 205780 }, { "epoch": 0.0158464, "grad_norm": 0.7992235422134399, "learning_rate": 9.177412476283342e-06, "loss": 3.0121, "step": 205790 }, { "epoch": 0.015872, "grad_norm": 0.7733979821205139, "learning_rate": 9.176609929559341e-06, "loss": 2.893, "step": 205800 }, { "epoch": 0.0158976, "grad_norm": 0.7613145709037781, "learning_rate": 9.175807388174809e-06, "loss": 2.953, "step": 205810 }, { "epoch": 0.0159232, "grad_norm": 0.9016727805137634, "learning_rate": 9.175004852134946e-06, "loss": 3.0413, "step": 205820 }, { "epoch": 0.0159488, "grad_norm": 0.9184519648551941, "learning_rate": 9.174202321444954e-06, "loss": 2.7247, "step": 205830 }, { "epoch": 0.0159744, "grad_norm": 0.7381618022918701, "learning_rate": 9.173399796110043e-06, "loss": 2.7678, "step": 205840 }, { "epoch": 0.016, "grad_norm": 0.7016613483428955, "learning_rate": 9.172597276135417e-06, "loss": 3.0437, "step": 205850 }, { "epoch": 0.0160256, "grad_norm": 0.9013299345970154, "learning_rate": 9.171794761526273e-06, "loss": 2.904, "step": 205860 }, { "epoch": 0.0160512, "grad_norm": 0.9176357388496399, "learning_rate": 9.170992252287822e-06, "loss": 3.0001, "step": 205870 }, { "epoch": 0.0160768, "grad_norm": 0.7700033783912659, "learning_rate": 9.170189748425267e-06, "loss": 2.8527, "step": 205880 }, { "epoch": 0.0161024, "grad_norm": 0.9818748831748962, "learning_rate": 9.16938724994381e-06, "loss": 2.8229, "step": 205890 }, { "epoch": 0.016128, "grad_norm": 0.7624607086181641, "learning_rate": 9.168584756848657e-06, "loss": 2.9502, "step": 205900 }, { "epoch": 0.0161536, "grad_norm": 0.8565415143966675, "learning_rate": 9.167782269145008e-06, "loss": 2.7638, "step": 205910 }, { "epoch": 0.0161792, "grad_norm": 0.7413187026977539, "learning_rate": 9.16697978683807e-06, "loss": 2.9237, "step": 205920 }, { "epoch": 0.0162048, "grad_norm": 0.8037371039390564, "learning_rate": 9.166177309933048e-06, "loss": 2.8311, "step": 205930 }, { "epoch": 0.0162304, "grad_norm": 0.8293194770812988, "learning_rate": 9.165374838435144e-06, "loss": 2.9632, "step": 205940 }, { "epoch": 0.016256, "grad_norm": 3.942760705947876, "learning_rate": 9.164572372349565e-06, "loss": 3.4562, "step": 205950 }, { "epoch": 0.0162816, "grad_norm": 0.7266656160354614, "learning_rate": 9.16376991168151e-06, "loss": 2.9587, "step": 205960 }, { "epoch": 0.0163072, "grad_norm": 0.7602135539054871, "learning_rate": 9.162967456436184e-06, "loss": 2.8008, "step": 205970 }, { "epoch": 0.0163328, "grad_norm": 0.741424024105072, "learning_rate": 9.162165006618794e-06, "loss": 2.8275, "step": 205980 }, { "epoch": 0.0163584, "grad_norm": 1.516093373298645, "learning_rate": 9.16136256223454e-06, "loss": 2.8477, "step": 205990 }, { "epoch": 0.016384, "grad_norm": 0.8045639395713806, "learning_rate": 9.160560123288626e-06, "loss": 2.8462, "step": 206000 }, { "epoch": 0.0164096, "grad_norm": 0.9041446447372437, "learning_rate": 9.159757689786257e-06, "loss": 2.8506, "step": 206010 }, { "epoch": 0.0164352, "grad_norm": 0.8286184072494507, "learning_rate": 9.158955261732639e-06, "loss": 2.8859, "step": 206020 }, { "epoch": 0.0164608, "grad_norm": 0.8015335202217102, "learning_rate": 9.15815283913297e-06, "loss": 3.1869, "step": 206030 }, { "epoch": 0.0164864, "grad_norm": 0.8812963366508484, "learning_rate": 9.157350421992454e-06, "loss": 3.0203, "step": 206040 }, { "epoch": 0.016512, "grad_norm": 0.7356315851211548, "learning_rate": 9.156548010316301e-06, "loss": 3.0226, "step": 206050 }, { "epoch": 0.0165376, "grad_norm": 1.0075603723526, "learning_rate": 9.155745604109708e-06, "loss": 2.8433, "step": 206060 }, { "epoch": 0.0165632, "grad_norm": 0.7359568476676941, "learning_rate": 9.15494320337788e-06, "loss": 2.9072, "step": 206070 }, { "epoch": 0.0165888, "grad_norm": 0.7688089609146118, "learning_rate": 9.154140808126019e-06, "loss": 2.8115, "step": 206080 }, { "epoch": 0.0166144, "grad_norm": 0.8626114130020142, "learning_rate": 9.153338418359334e-06, "loss": 2.9846, "step": 206090 }, { "epoch": 0.01664, "grad_norm": 0.8417251706123352, "learning_rate": 9.152536034083018e-06, "loss": 2.9313, "step": 206100 }, { "epoch": 0.0166656, "grad_norm": 0.7325869202613831, "learning_rate": 9.151733655302286e-06, "loss": 3.0061, "step": 206110 }, { "epoch": 0.0166912, "grad_norm": 0.7956803441047668, "learning_rate": 9.150931282022334e-06, "loss": 3.0009, "step": 206120 }, { "epoch": 0.0167168, "grad_norm": 0.9044244289398193, "learning_rate": 9.150128914248368e-06, "loss": 2.7579, "step": 206130 }, { "epoch": 0.0167424, "grad_norm": 0.7840790748596191, "learning_rate": 9.149326551985588e-06, "loss": 3.2433, "step": 206140 }, { "epoch": 0.016768, "grad_norm": 0.8191894888877869, "learning_rate": 9.148524195239203e-06, "loss": 2.9323, "step": 206150 }, { "epoch": 0.0167936, "grad_norm": 0.9455527663230896, "learning_rate": 9.147721844014411e-06, "loss": 3.2476, "step": 206160 }, { "epoch": 0.0168192, "grad_norm": 0.875374972820282, "learning_rate": 9.146919498316416e-06, "loss": 2.9097, "step": 206170 }, { "epoch": 0.0168448, "grad_norm": 0.7916672229766846, "learning_rate": 9.146117158150421e-06, "loss": 2.9058, "step": 206180 }, { "epoch": 0.0168704, "grad_norm": 0.820372998714447, "learning_rate": 9.145314823521629e-06, "loss": 2.9493, "step": 206190 }, { "epoch": 0.016896, "grad_norm": 0.8565426468849182, "learning_rate": 9.144512494435244e-06, "loss": 2.8678, "step": 206200 }, { "epoch": 0.0169216, "grad_norm": 0.7490829825401306, "learning_rate": 9.143710170896467e-06, "loss": 2.8669, "step": 206210 }, { "epoch": 0.0169472, "grad_norm": 0.7578275799751282, "learning_rate": 9.142907852910505e-06, "loss": 2.7469, "step": 206220 }, { "epoch": 0.0169728, "grad_norm": 1.2081416845321655, "learning_rate": 9.142105540482552e-06, "loss": 2.9053, "step": 206230 }, { "epoch": 0.0169984, "grad_norm": 0.9895362854003906, "learning_rate": 9.141303233617821e-06, "loss": 2.8797, "step": 206240 }, { "epoch": 0.017024, "grad_norm": 1.2031657695770264, "learning_rate": 9.140500932321511e-06, "loss": 2.8137, "step": 206250 }, { "epoch": 0.0170496, "grad_norm": 1.0134646892547607, "learning_rate": 9.139698636598822e-06, "loss": 3.0879, "step": 206260 }, { "epoch": 0.0170752, "grad_norm": 0.9081059694290161, "learning_rate": 9.138896346454957e-06, "loss": 3.0117, "step": 206270 }, { "epoch": 0.0171008, "grad_norm": 0.8951759934425354, "learning_rate": 9.13809406189512e-06, "loss": 2.9993, "step": 206280 }, { "epoch": 0.0171264, "grad_norm": 0.8264755606651306, "learning_rate": 9.137291782924517e-06, "loss": 2.8401, "step": 206290 }, { "epoch": 0.017152, "grad_norm": 0.821401834487915, "learning_rate": 9.136489509548346e-06, "loss": 2.7628, "step": 206300 }, { "epoch": 0.0171776, "grad_norm": 0.7482059597969055, "learning_rate": 9.135687241771813e-06, "loss": 3.0513, "step": 206310 }, { "epoch": 0.0172032, "grad_norm": 0.8791103363037109, "learning_rate": 9.134884979600116e-06, "loss": 2.8225, "step": 206320 }, { "epoch": 0.0172288, "grad_norm": 0.7211155891418457, "learning_rate": 9.134082723038463e-06, "loss": 2.9424, "step": 206330 }, { "epoch": 0.0172544, "grad_norm": 0.8448596596717834, "learning_rate": 9.13328047209205e-06, "loss": 3.0993, "step": 206340 }, { "epoch": 0.01728, "grad_norm": 0.875514805316925, "learning_rate": 9.132478226766086e-06, "loss": 3.0514, "step": 206350 }, { "epoch": 0.0173056, "grad_norm": 0.83383709192276, "learning_rate": 9.131675987065768e-06, "loss": 3.1005, "step": 206360 }, { "epoch": 0.0173312, "grad_norm": 0.738400399684906, "learning_rate": 9.1308737529963e-06, "loss": 2.8352, "step": 206370 }, { "epoch": 0.0173568, "grad_norm": 0.7155601382255554, "learning_rate": 9.13007152456289e-06, "loss": 2.9727, "step": 206380 }, { "epoch": 0.0173824, "grad_norm": 0.7672757506370544, "learning_rate": 9.129269301770728e-06, "loss": 3.0286, "step": 206390 }, { "epoch": 0.017408, "grad_norm": 0.8318163752555847, "learning_rate": 9.128467084625026e-06, "loss": 2.8305, "step": 206400 }, { "epoch": 0.0174336, "grad_norm": 0.7515308856964111, "learning_rate": 9.127664873130983e-06, "loss": 2.9242, "step": 206410 }, { "epoch": 0.0174592, "grad_norm": 0.7967491745948792, "learning_rate": 9.126862667293804e-06, "loss": 2.971, "step": 206420 }, { "epoch": 0.0174848, "grad_norm": 0.7513344883918762, "learning_rate": 9.126060467118687e-06, "loss": 2.8261, "step": 206430 }, { "epoch": 0.0175104, "grad_norm": 0.8698447942733765, "learning_rate": 9.125258272610831e-06, "loss": 3.1155, "step": 206440 }, { "epoch": 0.017536, "grad_norm": 0.7573950886726379, "learning_rate": 9.124456083775446e-06, "loss": 2.9983, "step": 206450 }, { "epoch": 0.0175616, "grad_norm": 0.8959690928459167, "learning_rate": 9.12365390061773e-06, "loss": 2.9904, "step": 206460 }, { "epoch": 0.0175872, "grad_norm": 0.9061518311500549, "learning_rate": 9.12285172314289e-06, "loss": 2.9812, "step": 206470 }, { "epoch": 0.0176128, "grad_norm": 0.6829664707183838, "learning_rate": 9.12204955135612e-06, "loss": 2.8986, "step": 206480 }, { "epoch": 0.0176384, "grad_norm": 0.7678834199905396, "learning_rate": 9.121247385262625e-06, "loss": 2.8432, "step": 206490 }, { "epoch": 0.017664, "grad_norm": 0.8339932560920715, "learning_rate": 9.120445224867606e-06, "loss": 2.9429, "step": 206500 }, { "epoch": 0.0176896, "grad_norm": 0.7955848574638367, "learning_rate": 9.119643070176269e-06, "loss": 3.0527, "step": 206510 }, { "epoch": 0.0177152, "grad_norm": 0.8058307766914368, "learning_rate": 9.118840921193812e-06, "loss": 2.926, "step": 206520 }, { "epoch": 0.0177408, "grad_norm": 0.7298773527145386, "learning_rate": 9.118038777925435e-06, "loss": 2.8906, "step": 206530 }, { "epoch": 0.0177664, "grad_norm": 0.8197038173675537, "learning_rate": 9.117236640376344e-06, "loss": 2.9144, "step": 206540 }, { "epoch": 0.017792, "grad_norm": 0.8954117894172668, "learning_rate": 9.11643450855174e-06, "loss": 2.8513, "step": 206550 }, { "epoch": 0.0178176, "grad_norm": 0.7808060646057129, "learning_rate": 9.115632382456821e-06, "loss": 3.4037, "step": 206560 }, { "epoch": 0.0178432, "grad_norm": 0.7691532373428345, "learning_rate": 9.114830262096792e-06, "loss": 2.8096, "step": 206570 }, { "epoch": 0.0178688, "grad_norm": 0.7805320024490356, "learning_rate": 9.114028147476855e-06, "loss": 2.9566, "step": 206580 }, { "epoch": 0.0178944, "grad_norm": 0.7708964943885803, "learning_rate": 9.113226038602209e-06, "loss": 3.084, "step": 206590 }, { "epoch": 0.01792, "grad_norm": 0.7293590903282166, "learning_rate": 9.112423935478055e-06, "loss": 2.8366, "step": 206600 }, { "epoch": 0.0179456, "grad_norm": 0.9648771286010742, "learning_rate": 9.111621838109596e-06, "loss": 2.8962, "step": 206610 }, { "epoch": 0.0179712, "grad_norm": 1.0470072031021118, "learning_rate": 9.110819746502032e-06, "loss": 2.9472, "step": 206620 }, { "epoch": 0.0179968, "grad_norm": 0.7071294784545898, "learning_rate": 9.110017660660568e-06, "loss": 2.8685, "step": 206630 }, { "epoch": 0.0180224, "grad_norm": 0.7971479892730713, "learning_rate": 9.109215580590404e-06, "loss": 2.973, "step": 206640 }, { "epoch": 0.018048, "grad_norm": 0.9005448818206787, "learning_rate": 9.108413506296739e-06, "loss": 2.979, "step": 206650 }, { "epoch": 0.0180736, "grad_norm": 0.7623934745788574, "learning_rate": 9.107611437784773e-06, "loss": 2.9457, "step": 206660 }, { "epoch": 0.0180992, "grad_norm": 0.8784456253051758, "learning_rate": 9.106809375059715e-06, "loss": 3.1834, "step": 206670 }, { "epoch": 0.0181248, "grad_norm": 0.6849877834320068, "learning_rate": 9.106007318126757e-06, "loss": 2.9294, "step": 206680 }, { "epoch": 0.0181504, "grad_norm": 0.7225426435470581, "learning_rate": 9.105205266991105e-06, "loss": 2.968, "step": 206690 }, { "epoch": 0.018176, "grad_norm": 0.8753663301467896, "learning_rate": 9.104403221657957e-06, "loss": 2.95, "step": 206700 }, { "epoch": 0.0182016, "grad_norm": 0.7779686450958252, "learning_rate": 9.10360118213252e-06, "loss": 2.9323, "step": 206710 }, { "epoch": 0.0182272, "grad_norm": 0.9141167998313904, "learning_rate": 9.102799148419989e-06, "loss": 2.9154, "step": 206720 }, { "epoch": 0.0182528, "grad_norm": 0.9452885985374451, "learning_rate": 9.101997120525567e-06, "loss": 2.9986, "step": 206730 }, { "epoch": 0.0182784, "grad_norm": 0.8141237497329712, "learning_rate": 9.101195098454456e-06, "loss": 2.6997, "step": 206740 }, { "epoch": 0.018304, "grad_norm": 0.7131535410881042, "learning_rate": 9.100393082211855e-06, "loss": 2.9586, "step": 206750 }, { "epoch": 0.0183296, "grad_norm": 0.9225637316703796, "learning_rate": 9.099591071802965e-06, "loss": 3.0234, "step": 206760 }, { "epoch": 0.0183552, "grad_norm": 4.7299909591674805, "learning_rate": 9.098789067232988e-06, "loss": 2.8208, "step": 206770 }, { "epoch": 0.0183808, "grad_norm": 1.0693106651306152, "learning_rate": 9.097987068507126e-06, "loss": 2.9874, "step": 206780 }, { "epoch": 0.0184064, "grad_norm": 0.9201074242591858, "learning_rate": 9.097185075630573e-06, "loss": 3.085, "step": 206790 }, { "epoch": 0.018432, "grad_norm": 0.8852073550224304, "learning_rate": 9.09638308860854e-06, "loss": 2.9394, "step": 206800 }, { "epoch": 0.0184576, "grad_norm": 0.7279281616210938, "learning_rate": 9.09558110744622e-06, "loss": 2.967, "step": 206810 }, { "epoch": 0.0184832, "grad_norm": 0.783089280128479, "learning_rate": 9.094779132148818e-06, "loss": 2.8861, "step": 206820 }, { "epoch": 0.0185088, "grad_norm": 0.7921633720397949, "learning_rate": 9.093977162721531e-06, "loss": 3.0144, "step": 206830 }, { "epoch": 0.0185344, "grad_norm": 0.7041509747505188, "learning_rate": 9.093175199169565e-06, "loss": 2.6556, "step": 206840 }, { "epoch": 0.01856, "grad_norm": 0.7325218319892883, "learning_rate": 9.092373241498113e-06, "loss": 2.9774, "step": 206850 }, { "epoch": 0.0185856, "grad_norm": 0.7389519810676575, "learning_rate": 9.091571289712382e-06, "loss": 2.8586, "step": 206860 }, { "epoch": 0.0186112, "grad_norm": 0.8294035792350769, "learning_rate": 9.090769343817568e-06, "loss": 2.9376, "step": 206870 }, { "epoch": 0.0186368, "grad_norm": 0.9180010557174683, "learning_rate": 9.089967403818874e-06, "loss": 2.9542, "step": 206880 }, { "epoch": 0.0186624, "grad_norm": 0.7735179662704468, "learning_rate": 9.089165469721498e-06, "loss": 2.9827, "step": 206890 }, { "epoch": 0.018688, "grad_norm": 0.9884973168373108, "learning_rate": 9.088363541530643e-06, "loss": 3.0123, "step": 206900 }, { "epoch": 0.0187136, "grad_norm": 0.7663178443908691, "learning_rate": 9.08756161925151e-06, "loss": 2.9578, "step": 206910 }, { "epoch": 0.0187392, "grad_norm": 0.7707507610321045, "learning_rate": 9.086759702889294e-06, "loss": 3.0145, "step": 206920 }, { "epoch": 0.0187648, "grad_norm": 0.8908954858779907, "learning_rate": 9.0859577924492e-06, "loss": 2.9123, "step": 206930 }, { "epoch": 0.0187904, "grad_norm": 0.923905074596405, "learning_rate": 9.085155887936428e-06, "loss": 2.9643, "step": 206940 }, { "epoch": 0.018816, "grad_norm": 0.8009641766548157, "learning_rate": 9.084353989356175e-06, "loss": 3.0612, "step": 206950 }, { "epoch": 0.0188416, "grad_norm": 0.7640849351882935, "learning_rate": 9.083552096713645e-06, "loss": 2.9195, "step": 206960 }, { "epoch": 0.0188672, "grad_norm": 0.9349434971809387, "learning_rate": 9.082750210014032e-06, "loss": 3.0294, "step": 206970 }, { "epoch": 0.0188928, "grad_norm": 0.7387866973876953, "learning_rate": 9.081948329262542e-06, "loss": 2.8921, "step": 206980 }, { "epoch": 0.0189184, "grad_norm": 0.7373398542404175, "learning_rate": 9.081146454464373e-06, "loss": 2.8989, "step": 206990 }, { "epoch": 0.018944, "grad_norm": 0.9580647349357605, "learning_rate": 9.080344585624728e-06, "loss": 2.908, "step": 207000 }, { "epoch": 0.0189696, "grad_norm": 0.9828168749809265, "learning_rate": 9.079542722748802e-06, "loss": 2.914, "step": 207010 }, { "epoch": 0.0189952, "grad_norm": 0.787567675113678, "learning_rate": 9.078740865841798e-06, "loss": 2.8788, "step": 207020 }, { "epoch": 0.0190208, "grad_norm": 0.6835291981697083, "learning_rate": 9.077939014908911e-06, "loss": 2.9607, "step": 207030 }, { "epoch": 0.0190464, "grad_norm": 0.8427354693412781, "learning_rate": 9.07713716995535e-06, "loss": 2.9068, "step": 207040 }, { "epoch": 0.019072, "grad_norm": 1.0625922679901123, "learning_rate": 9.076335330986305e-06, "loss": 2.9459, "step": 207050 }, { "epoch": 0.0190976, "grad_norm": 0.7715465426445007, "learning_rate": 9.07553349800698e-06, "loss": 2.9214, "step": 207060 }, { "epoch": 0.0191232, "grad_norm": 0.6776921153068542, "learning_rate": 9.074731671022578e-06, "loss": 2.9319, "step": 207070 }, { "epoch": 0.0191488, "grad_norm": 0.7440983057022095, "learning_rate": 9.073929850038294e-06, "loss": 2.9958, "step": 207080 }, { "epoch": 0.0191744, "grad_norm": 0.7656397819519043, "learning_rate": 9.073128035059327e-06, "loss": 2.9009, "step": 207090 }, { "epoch": 0.0192, "grad_norm": 0.7584489583969116, "learning_rate": 9.072326226090877e-06, "loss": 2.64, "step": 207100 }, { "epoch": 0.0192256, "grad_norm": 0.7935684323310852, "learning_rate": 9.07152442313815e-06, "loss": 3.005, "step": 207110 }, { "epoch": 0.0192512, "grad_norm": 1.506089210510254, "learning_rate": 9.070722626206336e-06, "loss": 3.0505, "step": 207120 }, { "epoch": 0.0192768, "grad_norm": 0.8079202175140381, "learning_rate": 9.06992083530064e-06, "loss": 2.7983, "step": 207130 }, { "epoch": 0.0193024, "grad_norm": 0.7310112714767456, "learning_rate": 9.069119050426258e-06, "loss": 3.1102, "step": 207140 }, { "epoch": 0.019328, "grad_norm": 0.7543558478355408, "learning_rate": 9.068317271588393e-06, "loss": 2.8623, "step": 207150 }, { "epoch": 0.0193536, "grad_norm": 1.0372143983840942, "learning_rate": 9.067515498792246e-06, "loss": 2.7338, "step": 207160 }, { "epoch": 0.0193792, "grad_norm": 0.7278075218200684, "learning_rate": 9.06671373204301e-06, "loss": 2.676, "step": 207170 }, { "epoch": 0.0194048, "grad_norm": 0.8931963443756104, "learning_rate": 9.065911971345887e-06, "loss": 2.7686, "step": 207180 }, { "epoch": 0.0194304, "grad_norm": 0.7730277180671692, "learning_rate": 9.065110216706078e-06, "loss": 2.6939, "step": 207190 }, { "epoch": 0.019456, "grad_norm": 0.776332676410675, "learning_rate": 9.064308468128782e-06, "loss": 2.7361, "step": 207200 }, { "epoch": 0.0194816, "grad_norm": 0.8667188286781311, "learning_rate": 9.063506725619196e-06, "loss": 2.8444, "step": 207210 }, { "epoch": 0.0195072, "grad_norm": 1.0175246000289917, "learning_rate": 9.062704989182518e-06, "loss": 2.4673, "step": 207220 }, { "epoch": 0.0195328, "grad_norm": 0.7664919495582581, "learning_rate": 9.061903258823951e-06, "loss": 2.8319, "step": 207230 }, { "epoch": 0.0195584, "grad_norm": 0.9657641649246216, "learning_rate": 9.061101534548693e-06, "loss": 2.8295, "step": 207240 }, { "epoch": 0.019584, "grad_norm": 1.0835014581680298, "learning_rate": 9.060299816361941e-06, "loss": 2.785, "step": 207250 }, { "epoch": 0.0196096, "grad_norm": 0.7840412855148315, "learning_rate": 9.059498104268893e-06, "loss": 2.6191, "step": 207260 }, { "epoch": 0.0196352, "grad_norm": 0.7946926355361938, "learning_rate": 9.058696398274754e-06, "loss": 2.6229, "step": 207270 }, { "epoch": 0.0196608, "grad_norm": 0.749812126159668, "learning_rate": 9.057894698384715e-06, "loss": 2.5568, "step": 207280 }, { "epoch": 0.0196864, "grad_norm": 0.9435539841651917, "learning_rate": 9.057093004603979e-06, "loss": 2.8815, "step": 207290 }, { "epoch": 0.019712, "grad_norm": 0.8359376192092896, "learning_rate": 9.056291316937745e-06, "loss": 2.8429, "step": 207300 }, { "epoch": 0.0197376, "grad_norm": 0.7452563047409058, "learning_rate": 9.055489635391208e-06, "loss": 2.7517, "step": 207310 }, { "epoch": 0.0197632, "grad_norm": 0.995966911315918, "learning_rate": 9.054687959969572e-06, "loss": 2.6217, "step": 207320 }, { "epoch": 0.0197888, "grad_norm": 0.8191102743148804, "learning_rate": 9.053886290678036e-06, "loss": 2.7354, "step": 207330 }, { "epoch": 0.0198144, "grad_norm": 0.7139374613761902, "learning_rate": 9.053084627521793e-06, "loss": 2.8161, "step": 207340 }, { "epoch": 0.01984, "grad_norm": 0.8924617767333984, "learning_rate": 9.052282970506045e-06, "loss": 2.7712, "step": 207350 }, { "epoch": 0.0198656, "grad_norm": 0.718863844871521, "learning_rate": 9.051481319635993e-06, "loss": 2.7108, "step": 207360 }, { "epoch": 0.0198912, "grad_norm": 0.7530993819236755, "learning_rate": 9.050679674916829e-06, "loss": 2.5887, "step": 207370 }, { "epoch": 0.0199168, "grad_norm": 0.8352065086364746, "learning_rate": 9.049878036353756e-06, "loss": 2.6696, "step": 207380 }, { "epoch": 0.0199424, "grad_norm": 0.8061779141426086, "learning_rate": 9.049076403951971e-06, "loss": 2.4769, "step": 207390 }, { "epoch": 0.019968, "grad_norm": 0.7845072150230408, "learning_rate": 9.048274777716675e-06, "loss": 2.715, "step": 207400 }, { "epoch": 0.0199936, "grad_norm": 0.8233450055122375, "learning_rate": 9.047473157653063e-06, "loss": 2.9813, "step": 207410 }, { "epoch": 0.0200192, "grad_norm": 0.7783143520355225, "learning_rate": 9.046671543766333e-06, "loss": 2.7533, "step": 207420 }, { "epoch": 0.0200448, "grad_norm": 0.7538989782333374, "learning_rate": 9.045869936061685e-06, "loss": 3.0144, "step": 207430 }, { "epoch": 0.0200704, "grad_norm": 0.8051595091819763, "learning_rate": 9.04506833454432e-06, "loss": 2.7932, "step": 207440 }, { "epoch": 0.020096, "grad_norm": 0.7257870435714722, "learning_rate": 9.044266739219429e-06, "loss": 2.7256, "step": 207450 }, { "epoch": 0.0201216, "grad_norm": 0.7318617105484009, "learning_rate": 9.043465150092216e-06, "loss": 3.1127, "step": 207460 }, { "epoch": 0.0201472, "grad_norm": 0.7428960204124451, "learning_rate": 9.042663567167877e-06, "loss": 2.6424, "step": 207470 }, { "epoch": 0.0201728, "grad_norm": 0.8943918347358704, "learning_rate": 9.041861990451611e-06, "loss": 2.8942, "step": 207480 }, { "epoch": 0.0201984, "grad_norm": 0.7176865935325623, "learning_rate": 9.041060419948611e-06, "loss": 2.8675, "step": 207490 }, { "epoch": 0.020224, "grad_norm": 0.7340482473373413, "learning_rate": 9.040258855664083e-06, "loss": 2.7834, "step": 207500 }, { "epoch": 0.0202496, "grad_norm": 0.9786879420280457, "learning_rate": 9.039457297603221e-06, "loss": 2.7081, "step": 207510 }, { "epoch": 0.0202752, "grad_norm": 0.8015834093093872, "learning_rate": 9.038655745771223e-06, "loss": 2.6338, "step": 207520 }, { "epoch": 0.0203008, "grad_norm": 0.8951922655105591, "learning_rate": 9.037854200173289e-06, "loss": 2.598, "step": 207530 }, { "epoch": 0.0203264, "grad_norm": 0.884161114692688, "learning_rate": 9.037052660814613e-06, "loss": 3.0062, "step": 207540 }, { "epoch": 0.020352, "grad_norm": 0.8684247732162476, "learning_rate": 9.036251127700394e-06, "loss": 2.8643, "step": 207550 }, { "epoch": 0.0203776, "grad_norm": 0.7232090830802917, "learning_rate": 9.035449600835832e-06, "loss": 2.6006, "step": 207560 }, { "epoch": 0.0204032, "grad_norm": 0.7194027304649353, "learning_rate": 9.034648080226122e-06, "loss": 2.5338, "step": 207570 }, { "epoch": 0.0204288, "grad_norm": 0.7934588193893433, "learning_rate": 9.033846565876463e-06, "loss": 2.5328, "step": 207580 }, { "epoch": 0.0204544, "grad_norm": 0.8140222430229187, "learning_rate": 9.033045057792051e-06, "loss": 2.9474, "step": 207590 }, { "epoch": 0.02048, "grad_norm": 0.7384459972381592, "learning_rate": 9.032243555978087e-06, "loss": 2.8193, "step": 207600 }, { "epoch": 0.0205056, "grad_norm": 0.9160641431808472, "learning_rate": 9.031442060439764e-06, "loss": 2.6469, "step": 207610 }, { "epoch": 0.0205312, "grad_norm": 0.770088255405426, "learning_rate": 9.030640571182283e-06, "loss": 2.6237, "step": 207620 }, { "epoch": 0.0205568, "grad_norm": 0.8304487466812134, "learning_rate": 9.02983908821084e-06, "loss": 2.7677, "step": 207630 }, { "epoch": 0.0205824, "grad_norm": 0.8445963859558105, "learning_rate": 9.029037611530634e-06, "loss": 2.7526, "step": 207640 }, { "epoch": 0.020608, "grad_norm": 0.897142767906189, "learning_rate": 9.02823614114686e-06, "loss": 2.4568, "step": 207650 }, { "epoch": 0.0206336, "grad_norm": 0.8751212954521179, "learning_rate": 9.027434677064713e-06, "loss": 2.6094, "step": 207660 }, { "epoch": 0.0206592, "grad_norm": 0.7188673615455627, "learning_rate": 9.026633219289396e-06, "loss": 2.765, "step": 207670 }, { "epoch": 0.0206848, "grad_norm": 0.8886620998382568, "learning_rate": 9.025831767826106e-06, "loss": 2.9005, "step": 207680 }, { "epoch": 0.0207104, "grad_norm": 0.7449609637260437, "learning_rate": 9.025030322680038e-06, "loss": 2.6076, "step": 207690 }, { "epoch": 0.020736, "grad_norm": 0.8125739693641663, "learning_rate": 9.024228883856387e-06, "loss": 2.4815, "step": 207700 }, { "epoch": 0.0207616, "grad_norm": 0.7413489818572998, "learning_rate": 9.023427451360354e-06, "loss": 2.7051, "step": 207710 }, { "epoch": 0.0207872, "grad_norm": 0.8333553075790405, "learning_rate": 9.022626025197133e-06, "loss": 3.0277, "step": 207720 }, { "epoch": 0.0208128, "grad_norm": 0.7453364729881287, "learning_rate": 9.021824605371925e-06, "loss": 2.6722, "step": 207730 }, { "epoch": 0.0208384, "grad_norm": 1.0270804166793823, "learning_rate": 9.021023191889922e-06, "loss": 2.6726, "step": 207740 }, { "epoch": 0.020864, "grad_norm": 0.960541307926178, "learning_rate": 9.020221784756325e-06, "loss": 2.5782, "step": 207750 }, { "epoch": 0.0208896, "grad_norm": 0.7184701561927795, "learning_rate": 9.019420383976329e-06, "loss": 2.6876, "step": 207760 }, { "epoch": 0.0209152, "grad_norm": 1.0041874647140503, "learning_rate": 9.01861898955513e-06, "loss": 2.6465, "step": 207770 }, { "epoch": 0.0209408, "grad_norm": 0.8306108117103577, "learning_rate": 9.017817601497927e-06, "loss": 2.738, "step": 207780 }, { "epoch": 0.0209664, "grad_norm": 0.6922736167907715, "learning_rate": 9.017016219809916e-06, "loss": 2.8403, "step": 207790 }, { "epoch": 0.020992, "grad_norm": 0.7686352133750916, "learning_rate": 9.016214844496294e-06, "loss": 2.94, "step": 207800 }, { "epoch": 2.56e-05, "grad_norm": 0.7899758219718933, "learning_rate": 9.015413475562256e-06, "loss": 2.1614, "step": 207810 }, { "epoch": 5.12e-05, "grad_norm": 0.8450926542282104, "learning_rate": 9.014612113013e-06, "loss": 2.7808, "step": 207820 }, { "epoch": 7.68e-05, "grad_norm": 0.8114476203918457, "learning_rate": 9.013810756853724e-06, "loss": 2.7733, "step": 207830 }, { "epoch": 0.0001024, "grad_norm": 0.7946872711181641, "learning_rate": 9.013009407089618e-06, "loss": 2.6489, "step": 207840 }, { "epoch": 0.000128, "grad_norm": 0.8792074918746948, "learning_rate": 9.012208063725889e-06, "loss": 2.8238, "step": 207850 }, { "epoch": 0.0001536, "grad_norm": 0.938625693321228, "learning_rate": 9.011406726767728e-06, "loss": 2.6609, "step": 207860 }, { "epoch": 0.0001792, "grad_norm": 0.7451978325843811, "learning_rate": 9.01060539622033e-06, "loss": 2.9257, "step": 207870 }, { "epoch": 0.0002048, "grad_norm": 0.7602325081825256, "learning_rate": 9.009804072088893e-06, "loss": 2.7543, "step": 207880 }, { "epoch": 0.0002304, "grad_norm": 0.8815080523490906, "learning_rate": 9.009002754378615e-06, "loss": 2.8376, "step": 207890 }, { "epoch": 0.000256, "grad_norm": 0.9994405508041382, "learning_rate": 9.008201443094689e-06, "loss": 2.8302, "step": 207900 }, { "epoch": 0.0002816, "grad_norm": 0.7824707627296448, "learning_rate": 9.007400138242316e-06, "loss": 2.6481, "step": 207910 }, { "epoch": 0.0003072, "grad_norm": 0.7767515182495117, "learning_rate": 9.006598839826686e-06, "loss": 2.5886, "step": 207920 }, { "epoch": 0.0003328, "grad_norm": 0.8230604529380798, "learning_rate": 9.005797547853002e-06, "loss": 2.6377, "step": 207930 }, { "epoch": 0.0003584, "grad_norm": 0.8889525532722473, "learning_rate": 9.004996262326453e-06, "loss": 3.0514, "step": 207940 }, { "epoch": 0.000384, "grad_norm": 0.7954071760177612, "learning_rate": 9.00419498325224e-06, "loss": 2.6941, "step": 207950 }, { "epoch": 0.0004096, "grad_norm": 0.8364259004592896, "learning_rate": 9.003393710635559e-06, "loss": 2.6795, "step": 207960 }, { "epoch": 0.0004352, "grad_norm": 0.936177670955658, "learning_rate": 9.002592444481604e-06, "loss": 2.763, "step": 207970 }, { "epoch": 0.0004608, "grad_norm": 0.7102601528167725, "learning_rate": 9.001791184795572e-06, "loss": 2.5316, "step": 207980 }, { "epoch": 0.0004864, "grad_norm": 0.9888921976089478, "learning_rate": 9.000989931582657e-06, "loss": 2.6545, "step": 207990 }, { "epoch": 0.000512, "grad_norm": 0.7932806611061096, "learning_rate": 9.000188684848062e-06, "loss": 2.7311, "step": 208000 }, { "epoch": 0.0005376, "grad_norm": 0.9406062364578247, "learning_rate": 8.999387444596969e-06, "loss": 2.7077, "step": 208010 }, { "epoch": 0.0005632, "grad_norm": 0.7880719900131226, "learning_rate": 8.998586210834589e-06, "loss": 2.5783, "step": 208020 }, { "epoch": 0.0005888, "grad_norm": 0.8659006357192993, "learning_rate": 8.997784983566108e-06, "loss": 2.6744, "step": 208030 }, { "epoch": 0.0006144, "grad_norm": 0.7643877863883972, "learning_rate": 8.996983762796727e-06, "loss": 2.6897, "step": 208040 }, { "epoch": 0.00064, "grad_norm": 0.8022593855857849, "learning_rate": 8.996182548531643e-06, "loss": 2.5805, "step": 208050 }, { "epoch": 0.0006656, "grad_norm": 0.8500314354896545, "learning_rate": 8.995381340776044e-06, "loss": 2.8736, "step": 208060 }, { "epoch": 0.0006912, "grad_norm": 0.8005205988883972, "learning_rate": 8.994580139535131e-06, "loss": 3.1277, "step": 208070 }, { "epoch": 0.0007168, "grad_norm": 0.7634875178337097, "learning_rate": 8.993778944814097e-06, "loss": 2.6593, "step": 208080 }, { "epoch": 0.0007424, "grad_norm": 0.9030922055244446, "learning_rate": 8.992977756618144e-06, "loss": 2.7556, "step": 208090 }, { "epoch": 0.000768, "grad_norm": 0.9615602493286133, "learning_rate": 8.992176574952459e-06, "loss": 2.7965, "step": 208100 }, { "epoch": 0.0007936, "grad_norm": 0.7346752285957336, "learning_rate": 8.991375399822241e-06, "loss": 2.677, "step": 208110 }, { "epoch": 0.0008192, "grad_norm": 0.8125182390213013, "learning_rate": 8.990574231232685e-06, "loss": 2.7428, "step": 208120 }, { "epoch": 0.0008448, "grad_norm": 0.7434933185577393, "learning_rate": 8.989773069188991e-06, "loss": 2.698, "step": 208130 }, { "epoch": 0.0008704, "grad_norm": 0.8134037852287292, "learning_rate": 8.988971913696347e-06, "loss": 2.6275, "step": 208140 }, { "epoch": 0.000896, "grad_norm": 0.7389684915542603, "learning_rate": 8.98817076475995e-06, "loss": 2.388, "step": 208150 }, { "epoch": 0.0009216, "grad_norm": 1.7462880611419678, "learning_rate": 8.987369622385e-06, "loss": 2.8674, "step": 208160 }, { "epoch": 0.0009472, "grad_norm": 0.8995219469070435, "learning_rate": 8.986568486576686e-06, "loss": 2.7166, "step": 208170 }, { "epoch": 0.0009728, "grad_norm": 0.7910851836204529, "learning_rate": 8.985767357340205e-06, "loss": 2.602, "step": 208180 }, { "epoch": 0.0009984, "grad_norm": 0.757780909538269, "learning_rate": 8.984966234680754e-06, "loss": 2.6701, "step": 208190 }, { "epoch": 0.001024, "grad_norm": 0.7596404552459717, "learning_rate": 8.984165118603529e-06, "loss": 2.4355, "step": 208200 }, { "epoch": 0.0010496, "grad_norm": 0.9112117886543274, "learning_rate": 8.983364009113721e-06, "loss": 2.6251, "step": 208210 }, { "epoch": 0.0010752, "grad_norm": 0.9453021287918091, "learning_rate": 8.98256290621653e-06, "loss": 2.9668, "step": 208220 }, { "epoch": 0.0011008, "grad_norm": 0.7648867964744568, "learning_rate": 8.981761809917147e-06, "loss": 2.6575, "step": 208230 }, { "epoch": 0.0011264, "grad_norm": 0.8630638122558594, "learning_rate": 8.980960720220768e-06, "loss": 2.9977, "step": 208240 }, { "epoch": 0.001152, "grad_norm": 0.7711654305458069, "learning_rate": 8.980159637132591e-06, "loss": 2.5268, "step": 208250 }, { "epoch": 0.0011776, "grad_norm": 0.7695065140724182, "learning_rate": 8.979358560657802e-06, "loss": 2.7581, "step": 208260 }, { "epoch": 0.0012032, "grad_norm": 0.7502961754798889, "learning_rate": 8.978557490801605e-06, "loss": 2.8232, "step": 208270 }, { "epoch": 0.0012288, "grad_norm": 0.729205846786499, "learning_rate": 8.97775642756919e-06, "loss": 2.457, "step": 208280 }, { "epoch": 0.0012544, "grad_norm": 0.7207658290863037, "learning_rate": 8.976955370965754e-06, "loss": 2.5703, "step": 208290 }, { "epoch": 0.00128, "grad_norm": 0.8262843489646912, "learning_rate": 8.97615432099649e-06, "loss": 2.5944, "step": 208300 }, { "epoch": 0.0013056, "grad_norm": 0.7624902725219727, "learning_rate": 8.975353277666592e-06, "loss": 2.8541, "step": 208310 }, { "epoch": 0.0013312, "grad_norm": 1.0062881708145142, "learning_rate": 8.974552240981257e-06, "loss": 2.8113, "step": 208320 }, { "epoch": 0.0013568, "grad_norm": 1.2802107334136963, "learning_rate": 8.97375121094568e-06, "loss": 2.7583, "step": 208330 }, { "epoch": 0.0013824, "grad_norm": 0.8083350658416748, "learning_rate": 8.97295018756505e-06, "loss": 2.8717, "step": 208340 }, { "epoch": 0.001408, "grad_norm": 0.8265560865402222, "learning_rate": 8.972149170844566e-06, "loss": 2.8539, "step": 208350 }, { "epoch": 0.0014336, "grad_norm": 0.7641417384147644, "learning_rate": 8.97134816078942e-06, "loss": 2.7085, "step": 208360 }, { "epoch": 0.0014592, "grad_norm": 0.8393718600273132, "learning_rate": 8.970547157404809e-06, "loss": 2.8426, "step": 208370 }, { "epoch": 0.0014848, "grad_norm": 0.7792239189147949, "learning_rate": 8.969746160695929e-06, "loss": 2.4505, "step": 208380 }, { "epoch": 0.0015104, "grad_norm": 1.027329921722412, "learning_rate": 8.968945170667967e-06, "loss": 2.7102, "step": 208390 }, { "epoch": 0.001536, "grad_norm": 1.1132380962371826, "learning_rate": 8.968144187326124e-06, "loss": 2.5685, "step": 208400 }, { "epoch": 0.0015616, "grad_norm": 0.7971861362457275, "learning_rate": 8.96734321067559e-06, "loss": 3.1258, "step": 208410 }, { "epoch": 0.0015872, "grad_norm": 0.9536266326904297, "learning_rate": 8.966542240721566e-06, "loss": 2.6533, "step": 208420 }, { "epoch": 0.0016128, "grad_norm": 0.8095105886459351, "learning_rate": 8.965741277469235e-06, "loss": 2.8286, "step": 208430 }, { "epoch": 0.0016384, "grad_norm": 0.8206236958503723, "learning_rate": 8.964940320923798e-06, "loss": 2.9888, "step": 208440 }, { "epoch": 0.001664, "grad_norm": 0.989061176776886, "learning_rate": 8.964139371090453e-06, "loss": 2.721, "step": 208450 }, { "epoch": 0.0016896, "grad_norm": 0.8214403390884399, "learning_rate": 8.963338427974383e-06, "loss": 2.7092, "step": 208460 }, { "epoch": 0.0017152, "grad_norm": 1.2450151443481445, "learning_rate": 8.96253749158079e-06, "loss": 2.5488, "step": 208470 }, { "epoch": 0.0017408, "grad_norm": 0.7798155546188354, "learning_rate": 8.961736561914865e-06, "loss": 2.7469, "step": 208480 }, { "epoch": 0.0017664, "grad_norm": 0.8026940226554871, "learning_rate": 8.960935638981804e-06, "loss": 2.7983, "step": 208490 }, { "epoch": 0.001792, "grad_norm": 0.7400136590003967, "learning_rate": 8.960134722786797e-06, "loss": 2.6971, "step": 208500 }, { "epoch": 0.0018176, "grad_norm": 0.763418972492218, "learning_rate": 8.95933381333504e-06, "loss": 2.5869, "step": 208510 }, { "epoch": 0.0018432, "grad_norm": 0.8057984113693237, "learning_rate": 8.958532910631728e-06, "loss": 2.8602, "step": 208520 }, { "epoch": 0.0018688, "grad_norm": 0.8762497901916504, "learning_rate": 8.95773201468205e-06, "loss": 2.6246, "step": 208530 }, { "epoch": 0.0018944, "grad_norm": 0.8019502758979797, "learning_rate": 8.956931125491206e-06, "loss": 2.9563, "step": 208540 }, { "epoch": 0.00192, "grad_norm": 0.6839445233345032, "learning_rate": 8.956130243064387e-06, "loss": 2.8093, "step": 208550 }, { "epoch": 0.0019456, "grad_norm": 0.7055720090866089, "learning_rate": 8.955329367406785e-06, "loss": 2.8749, "step": 208560 }, { "epoch": 0.0019712, "grad_norm": 0.8384780287742615, "learning_rate": 8.954528498523594e-06, "loss": 2.9327, "step": 208570 }, { "epoch": 0.0019968, "grad_norm": 0.7051168084144592, "learning_rate": 8.953727636420012e-06, "loss": 2.9271, "step": 208580 }, { "epoch": 0.0020224, "grad_norm": 0.8863044381141663, "learning_rate": 8.952926781101224e-06, "loss": 2.9556, "step": 208590 }, { "epoch": 0.002048, "grad_norm": 0.7366214394569397, "learning_rate": 8.952125932572426e-06, "loss": 2.6834, "step": 208600 }, { "epoch": 0.0020736, "grad_norm": 0.712492823600769, "learning_rate": 8.951325090838815e-06, "loss": 2.5359, "step": 208610 }, { "epoch": 0.0020992, "grad_norm": 0.8180164098739624, "learning_rate": 8.950524255905585e-06, "loss": 2.8207, "step": 208620 }, { "epoch": 0.0021248, "grad_norm": 1.0395500659942627, "learning_rate": 8.949723427777921e-06, "loss": 3.0382, "step": 208630 }, { "epoch": 0.0021504, "grad_norm": 0.8469575047492981, "learning_rate": 8.948922606461023e-06, "loss": 3.0099, "step": 208640 }, { "epoch": 0.002176, "grad_norm": 0.7184958457946777, "learning_rate": 8.948121791960085e-06, "loss": 2.6961, "step": 208650 }, { "epoch": 0.0022016, "grad_norm": 0.9816743731498718, "learning_rate": 8.947320984280297e-06, "loss": 2.9954, "step": 208660 }, { "epoch": 0.0022272, "grad_norm": 0.7845896482467651, "learning_rate": 8.946520183426851e-06, "loss": 2.7721, "step": 208670 }, { "epoch": 0.0022528, "grad_norm": 0.8963308334350586, "learning_rate": 8.94571938940494e-06, "loss": 2.9248, "step": 208680 }, { "epoch": 0.0022784, "grad_norm": 0.7914292216300964, "learning_rate": 8.944918602219764e-06, "loss": 2.8873, "step": 208690 }, { "epoch": 0.002304, "grad_norm": 0.7561102509498596, "learning_rate": 8.944117821876507e-06, "loss": 2.6602, "step": 208700 }, { "epoch": 2.56e-05, "grad_norm": 0.885704517364502, "learning_rate": 8.943317048380363e-06, "loss": 1.9897, "step": 208710 }, { "epoch": 5.12e-05, "grad_norm": 0.7954381704330444, "learning_rate": 8.94251628173653e-06, "loss": 2.8586, "step": 208720 }, { "epoch": 7.68e-05, "grad_norm": 1.0856735706329346, "learning_rate": 8.941715521950197e-06, "loss": 2.9106, "step": 208730 }, { "epoch": 0.0001024, "grad_norm": 0.7867459654808044, "learning_rate": 8.940914769026557e-06, "loss": 2.4738, "step": 208740 }, { "epoch": 0.000128, "grad_norm": 0.8210570216178894, "learning_rate": 8.940114022970806e-06, "loss": 2.7037, "step": 208750 }, { "epoch": 0.0001536, "grad_norm": 0.8606745600700378, "learning_rate": 8.939313283788131e-06, "loss": 2.7984, "step": 208760 }, { "epoch": 0.0001792, "grad_norm": 0.8489003777503967, "learning_rate": 8.938512551483728e-06, "loss": 2.5713, "step": 208770 }, { "epoch": 0.0002048, "grad_norm": 1.2473726272583008, "learning_rate": 8.937711826062793e-06, "loss": 2.6946, "step": 208780 }, { "epoch": 0.0002304, "grad_norm": 0.9966732859611511, "learning_rate": 8.93691110753051e-06, "loss": 2.8109, "step": 208790 }, { "epoch": 0.000256, "grad_norm": 0.7482894062995911, "learning_rate": 8.936110395892078e-06, "loss": 2.7884, "step": 208800 }, { "epoch": 0.0002816, "grad_norm": 0.7058998942375183, "learning_rate": 8.935309691152685e-06, "loss": 2.6157, "step": 208810 }, { "epoch": 0.0003072, "grad_norm": 0.8415139317512512, "learning_rate": 8.934508993317529e-06, "loss": 2.3429, "step": 208820 }, { "epoch": 0.0003328, "grad_norm": 0.8677098155021667, "learning_rate": 8.933708302391796e-06, "loss": 2.5434, "step": 208830 }, { "epoch": 0.0003584, "grad_norm": 0.8096327185630798, "learning_rate": 8.932907618380684e-06, "loss": 2.8291, "step": 208840 }, { "epoch": 0.000384, "grad_norm": 1.0685968399047852, "learning_rate": 8.932106941289381e-06, "loss": 2.6598, "step": 208850 }, { "epoch": 0.0004096, "grad_norm": 0.7672200202941895, "learning_rate": 8.93130627112308e-06, "loss": 2.6927, "step": 208860 }, { "epoch": 0.0004352, "grad_norm": 0.8032088279724121, "learning_rate": 8.930505607886975e-06, "loss": 2.9476, "step": 208870 }, { "epoch": 0.0004608, "grad_norm": 0.8506495356559753, "learning_rate": 8.929704951586254e-06, "loss": 2.6612, "step": 208880 }, { "epoch": 0.0004864, "grad_norm": 0.7936574816703796, "learning_rate": 8.928904302226114e-06, "loss": 2.7382, "step": 208890 }, { "epoch": 0.000512, "grad_norm": 0.6898098587989807, "learning_rate": 8.928103659811744e-06, "loss": 2.5878, "step": 208900 }, { "epoch": 0.0005376, "grad_norm": 0.7073619365692139, "learning_rate": 8.92730302434834e-06, "loss": 2.6329, "step": 208910 }, { "epoch": 0.0005632, "grad_norm": 0.7834053039550781, "learning_rate": 8.926502395841087e-06, "loss": 2.8076, "step": 208920 }, { "epoch": 0.0005888, "grad_norm": 0.7698588371276855, "learning_rate": 8.925701774295183e-06, "loss": 2.8663, "step": 208930 }, { "epoch": 0.0006144, "grad_norm": 0.8601493835449219, "learning_rate": 8.924901159715819e-06, "loss": 2.7132, "step": 208940 }, { "epoch": 0.00064, "grad_norm": 0.7400060296058655, "learning_rate": 8.924100552108182e-06, "loss": 2.6781, "step": 208950 }, { "epoch": 0.0006656, "grad_norm": 0.8128757476806641, "learning_rate": 8.923299951477468e-06, "loss": 2.6892, "step": 208960 }, { "epoch": 0.0006912, "grad_norm": 0.8610486388206482, "learning_rate": 8.922499357828867e-06, "loss": 3.1583, "step": 208970 }, { "epoch": 0.0007168, "grad_norm": 0.9435805678367615, "learning_rate": 8.921698771167575e-06, "loss": 2.8508, "step": 208980 }, { "epoch": 0.0007424, "grad_norm": 0.763589084148407, "learning_rate": 8.920898191498777e-06, "loss": 2.6924, "step": 208990 }, { "epoch": 0.000768, "grad_norm": 1.125862956047058, "learning_rate": 8.920097618827666e-06, "loss": 2.6706, "step": 209000 }, { "epoch": 0.0007936, "grad_norm": 0.7432852387428284, "learning_rate": 8.919297053159437e-06, "loss": 2.7462, "step": 209010 }, { "epoch": 0.0008192, "grad_norm": 0.8666630387306213, "learning_rate": 8.91849649449928e-06, "loss": 2.3249, "step": 209020 }, { "epoch": 0.0008448, "grad_norm": 0.8403299450874329, "learning_rate": 8.917695942852385e-06, "loss": 2.8706, "step": 209030 }, { "epoch": 0.0008704, "grad_norm": 0.9855852723121643, "learning_rate": 8.916895398223943e-06, "loss": 2.6617, "step": 209040 }, { "epoch": 0.000896, "grad_norm": 0.7876486778259277, "learning_rate": 8.916094860619145e-06, "loss": 2.614, "step": 209050 }, { "epoch": 0.0009216, "grad_norm": 0.7360567450523376, "learning_rate": 8.915294330043186e-06, "loss": 2.4877, "step": 209060 }, { "epoch": 0.0009472, "grad_norm": 0.8026654720306396, "learning_rate": 8.914493806501257e-06, "loss": 2.8409, "step": 209070 }, { "epoch": 0.0009728, "grad_norm": 0.9011322855949402, "learning_rate": 8.913693289998545e-06, "loss": 2.3249, "step": 209080 }, { "epoch": 0.0009984, "grad_norm": 0.9576182961463928, "learning_rate": 8.912892780540241e-06, "loss": 2.6897, "step": 209090 }, { "epoch": 0.001024, "grad_norm": 0.8151909708976746, "learning_rate": 8.91209227813154e-06, "loss": 2.8145, "step": 209100 }, { "epoch": 0.0010496, "grad_norm": 0.7820362448692322, "learning_rate": 8.911291782777635e-06, "loss": 2.5232, "step": 209110 }, { "epoch": 0.0010752, "grad_norm": 0.7612929344177246, "learning_rate": 8.910491294483712e-06, "loss": 2.6978, "step": 209120 }, { "epoch": 0.0011008, "grad_norm": 1.218329906463623, "learning_rate": 8.90969081325496e-06, "loss": 3.0469, "step": 209130 }, { "epoch": 0.0011264, "grad_norm": 0.779592752456665, "learning_rate": 8.908890339096579e-06, "loss": 2.7345, "step": 209140 }, { "epoch": 0.001152, "grad_norm": 0.8095972537994385, "learning_rate": 8.90808987201375e-06, "loss": 2.6298, "step": 209150 }, { "epoch": 0.0011776, "grad_norm": 1.0030418634414673, "learning_rate": 8.907289412011667e-06, "loss": 2.8106, "step": 209160 }, { "epoch": 0.0012032, "grad_norm": 0.8486787676811218, "learning_rate": 8.906488959095524e-06, "loss": 2.8671, "step": 209170 }, { "epoch": 0.0012288, "grad_norm": 0.8690713047981262, "learning_rate": 8.905688513270511e-06, "loss": 2.8413, "step": 209180 }, { "epoch": 0.0012544, "grad_norm": 0.7891060709953308, "learning_rate": 8.904888074541814e-06, "loss": 3.0494, "step": 209190 }, { "epoch": 0.00128, "grad_norm": 0.7826346755027771, "learning_rate": 8.904087642914627e-06, "loss": 2.6197, "step": 209200 }, { "epoch": 0.0013056, "grad_norm": 0.9016532301902771, "learning_rate": 8.903287218394141e-06, "loss": 2.8672, "step": 209210 }, { "epoch": 0.0013312, "grad_norm": 0.7459434866905212, "learning_rate": 8.902486800985548e-06, "loss": 2.8319, "step": 209220 }, { "epoch": 0.0013568, "grad_norm": 0.8780128359794617, "learning_rate": 8.90168639069403e-06, "loss": 2.7871, "step": 209230 }, { "epoch": 0.0013824, "grad_norm": 0.9804460406303406, "learning_rate": 8.90088598752479e-06, "loss": 2.6864, "step": 209240 }, { "epoch": 0.001408, "grad_norm": 0.894512951374054, "learning_rate": 8.900085591483011e-06, "loss": 2.8801, "step": 209250 }, { "epoch": 0.0014336, "grad_norm": 0.7809577584266663, "learning_rate": 8.899285202573884e-06, "loss": 2.6122, "step": 209260 }, { "epoch": 0.0014592, "grad_norm": 0.8745792508125305, "learning_rate": 8.898484820802603e-06, "loss": 2.629, "step": 209270 }, { "epoch": 0.0014848, "grad_norm": 1.1371742486953735, "learning_rate": 8.89768444617435e-06, "loss": 2.7668, "step": 209280 }, { "epoch": 0.0015104, "grad_norm": 0.7934445738792419, "learning_rate": 8.896884078694323e-06, "loss": 2.7731, "step": 209290 }, { "epoch": 0.001536, "grad_norm": 0.8267709612846375, "learning_rate": 8.89608371836771e-06, "loss": 2.4198, "step": 209300 }, { "epoch": 0.0015616, "grad_norm": 0.8813862204551697, "learning_rate": 8.8952833651997e-06, "loss": 2.7889, "step": 209310 }, { "epoch": 0.0015872, "grad_norm": 0.8120932579040527, "learning_rate": 8.894483019195484e-06, "loss": 2.8897, "step": 209320 }, { "epoch": 0.0016128, "grad_norm": 0.8883657455444336, "learning_rate": 8.893682680360252e-06, "loss": 2.9492, "step": 209330 }, { "epoch": 0.0016384, "grad_norm": 0.8916484713554382, "learning_rate": 8.892882348699195e-06, "loss": 2.9609, "step": 209340 }, { "epoch": 0.001664, "grad_norm": 0.7839609384536743, "learning_rate": 8.8920820242175e-06, "loss": 2.6765, "step": 209350 }, { "epoch": 0.0016896, "grad_norm": 0.8052304983139038, "learning_rate": 8.891281706920359e-06, "loss": 2.5657, "step": 209360 }, { "epoch": 0.0017152, "grad_norm": 0.9309009909629822, "learning_rate": 8.89048139681296e-06, "loss": 2.6846, "step": 209370 }, { "epoch": 0.0017408, "grad_norm": 0.8514214754104614, "learning_rate": 8.889681093900497e-06, "loss": 2.9518, "step": 209380 }, { "epoch": 0.0017664, "grad_norm": 0.8773918747901917, "learning_rate": 8.888880798188155e-06, "loss": 2.9184, "step": 209390 }, { "epoch": 0.001792, "grad_norm": 0.7885432243347168, "learning_rate": 8.888080509681125e-06, "loss": 2.5692, "step": 209400 }, { "epoch": 0.0018176, "grad_norm": 1.0683892965316772, "learning_rate": 8.887280228384598e-06, "loss": 2.9521, "step": 209410 }, { "epoch": 0.0018432, "grad_norm": 0.6988425850868225, "learning_rate": 8.886479954303762e-06, "loss": 2.4959, "step": 209420 }, { "epoch": 0.0018688, "grad_norm": 1.5686284303665161, "learning_rate": 8.88567968744381e-06, "loss": 2.9997, "step": 209430 }, { "epoch": 0.0018944, "grad_norm": 0.7601734399795532, "learning_rate": 8.88487942780993e-06, "loss": 2.6807, "step": 209440 }, { "epoch": 0.00192, "grad_norm": 0.7565600275993347, "learning_rate": 8.884079175407309e-06, "loss": 2.97, "step": 209450 }, { "epoch": 0.0019456, "grad_norm": 0.7971969246864319, "learning_rate": 8.883278930241137e-06, "loss": 2.7359, "step": 209460 }, { "epoch": 0.0019712, "grad_norm": 0.672481119632721, "learning_rate": 8.882478692316608e-06, "loss": 2.5069, "step": 209470 }, { "epoch": 0.0019968, "grad_norm": 0.7859963178634644, "learning_rate": 8.881678461638902e-06, "loss": 2.7258, "step": 209480 }, { "epoch": 0.0020224, "grad_norm": 0.8432694673538208, "learning_rate": 8.880878238213217e-06, "loss": 2.6191, "step": 209490 }, { "epoch": 0.002048, "grad_norm": 0.8420726656913757, "learning_rate": 8.880078022044739e-06, "loss": 2.6311, "step": 209500 }, { "epoch": 0.0020736, "grad_norm": 0.7767423391342163, "learning_rate": 8.87927781313866e-06, "loss": 2.5716, "step": 209510 }, { "epoch": 0.0020992, "grad_norm": 0.9606332182884216, "learning_rate": 8.878477611500163e-06, "loss": 2.8597, "step": 209520 }, { "epoch": 0.0021248, "grad_norm": 0.7853345274925232, "learning_rate": 8.877677417134441e-06, "loss": 2.7624, "step": 209530 }, { "epoch": 0.0021504, "grad_norm": 0.8128888010978699, "learning_rate": 8.876877230046686e-06, "loss": 2.7824, "step": 209540 }, { "epoch": 0.002176, "grad_norm": 0.8792323470115662, "learning_rate": 8.87607705024208e-06, "loss": 2.8739, "step": 209550 }, { "epoch": 0.0022016, "grad_norm": 0.8122170567512512, "learning_rate": 8.875276877725815e-06, "loss": 2.8824, "step": 209560 }, { "epoch": 0.0022272, "grad_norm": 0.7967479228973389, "learning_rate": 8.874476712503082e-06, "loss": 2.5513, "step": 209570 }, { "epoch": 0.0022528, "grad_norm": 0.7382898926734924, "learning_rate": 8.873676554579068e-06, "loss": 2.6381, "step": 209580 }, { "epoch": 0.0022784, "grad_norm": 1.4386361837387085, "learning_rate": 8.872876403958961e-06, "loss": 2.593, "step": 209590 }, { "epoch": 0.002304, "grad_norm": 0.7898109555244446, "learning_rate": 8.872076260647954e-06, "loss": 2.5917, "step": 209600 }, { "epoch": 0.0023296, "grad_norm": 0.9297420382499695, "learning_rate": 8.87127612465123e-06, "loss": 2.8854, "step": 209610 }, { "epoch": 0.0023552, "grad_norm": 0.8110248446464539, "learning_rate": 8.870475995973983e-06, "loss": 2.8524, "step": 209620 }, { "epoch": 0.0023808, "grad_norm": 0.8870580196380615, "learning_rate": 8.869675874621396e-06, "loss": 2.6914, "step": 209630 }, { "epoch": 0.0024064, "grad_norm": 0.9027256965637207, "learning_rate": 8.868875760598663e-06, "loss": 2.6944, "step": 209640 }, { "epoch": 0.002432, "grad_norm": 0.8625361323356628, "learning_rate": 8.86807565391097e-06, "loss": 2.9971, "step": 209650 }, { "epoch": 0.0024576, "grad_norm": 0.7639431357383728, "learning_rate": 8.867275554563505e-06, "loss": 2.814, "step": 209660 }, { "epoch": 0.0024832, "grad_norm": 0.8337306976318359, "learning_rate": 8.866475462561459e-06, "loss": 2.7062, "step": 209670 }, { "epoch": 0.0025088, "grad_norm": 0.8229225873947144, "learning_rate": 8.865675377910015e-06, "loss": 2.979, "step": 209680 }, { "epoch": 0.0025344, "grad_norm": 0.7965254187583923, "learning_rate": 8.864875300614367e-06, "loss": 2.8329, "step": 209690 }, { "epoch": 0.00256, "grad_norm": 0.8964807391166687, "learning_rate": 8.864075230679698e-06, "loss": 2.7394, "step": 209700 }, { "epoch": 0.0025856, "grad_norm": 0.7843095660209656, "learning_rate": 8.863275168111204e-06, "loss": 2.8348, "step": 209710 }, { "epoch": 0.0026112, "grad_norm": 0.7877594232559204, "learning_rate": 8.862475112914064e-06, "loss": 2.8325, "step": 209720 }, { "epoch": 0.0026368, "grad_norm": 0.8977944850921631, "learning_rate": 8.861675065093472e-06, "loss": 3.0673, "step": 209730 }, { "epoch": 0.0026624, "grad_norm": 0.7792796492576599, "learning_rate": 8.860875024654615e-06, "loss": 2.8764, "step": 209740 }, { "epoch": 0.002688, "grad_norm": 0.8361102938652039, "learning_rate": 8.860074991602678e-06, "loss": 2.8185, "step": 209750 }, { "epoch": 0.0027136, "grad_norm": 0.9173125624656677, "learning_rate": 8.859274965942855e-06, "loss": 2.9743, "step": 209760 }, { "epoch": 0.0027392, "grad_norm": 0.7398263812065125, "learning_rate": 8.85847494768033e-06, "loss": 2.9142, "step": 209770 }, { "epoch": 0.0027648, "grad_norm": 0.8082517385482788, "learning_rate": 8.85767493682029e-06, "loss": 2.9284, "step": 209780 }, { "epoch": 0.0027904, "grad_norm": 0.8655384182929993, "learning_rate": 8.856874933367926e-06, "loss": 2.9736, "step": 209790 }, { "epoch": 0.002816, "grad_norm": 1.105886697769165, "learning_rate": 8.856074937328427e-06, "loss": 2.7309, "step": 209800 }, { "epoch": 0.0028416, "grad_norm": 0.8182551860809326, "learning_rate": 8.855274948706974e-06, "loss": 2.9677, "step": 209810 }, { "epoch": 0.0028672, "grad_norm": 1.774935007095337, "learning_rate": 8.85447496750876e-06, "loss": 2.7908, "step": 209820 }, { "epoch": 0.0028928, "grad_norm": 1.0818322896957397, "learning_rate": 8.853674993738971e-06, "loss": 2.8463, "step": 209830 }, { "epoch": 0.0029184, "grad_norm": 0.7716238498687744, "learning_rate": 8.852875027402796e-06, "loss": 2.9712, "step": 209840 }, { "epoch": 0.002944, "grad_norm": 0.7374482750892639, "learning_rate": 8.852075068505422e-06, "loss": 2.7924, "step": 209850 }, { "epoch": 0.0029696, "grad_norm": 0.7273393869400024, "learning_rate": 8.851275117052035e-06, "loss": 2.9511, "step": 209860 }, { "epoch": 0.0029952, "grad_norm": 0.7704911828041077, "learning_rate": 8.850475173047824e-06, "loss": 2.8641, "step": 209870 }, { "epoch": 0.0030208, "grad_norm": 0.7518275380134583, "learning_rate": 8.849675236497975e-06, "loss": 2.8937, "step": 209880 }, { "epoch": 0.0030464, "grad_norm": 0.7961088418960571, "learning_rate": 8.848875307407677e-06, "loss": 2.978, "step": 209890 }, { "epoch": 0.003072, "grad_norm": 0.8080861568450928, "learning_rate": 8.848075385782117e-06, "loss": 2.935, "step": 209900 }, { "epoch": 0.0030976, "grad_norm": 0.7579876184463501, "learning_rate": 8.847275471626484e-06, "loss": 3.1463, "step": 209910 }, { "epoch": 0.0031232, "grad_norm": 0.7214590907096863, "learning_rate": 8.846475564945955e-06, "loss": 2.8338, "step": 209920 }, { "epoch": 0.0031488, "grad_norm": 0.8782939910888672, "learning_rate": 8.845675665745734e-06, "loss": 2.8634, "step": 209930 }, { "epoch": 0.0031744, "grad_norm": 0.7047213912010193, "learning_rate": 8.844875774030995e-06, "loss": 2.682, "step": 209940 }, { "epoch": 0.0032, "grad_norm": 0.8903359174728394, "learning_rate": 8.844075889806931e-06, "loss": 3.0131, "step": 209950 }, { "epoch": 0.0032256, "grad_norm": 1.06458580493927, "learning_rate": 8.84327601307873e-06, "loss": 3.1122, "step": 209960 }, { "epoch": 0.0032512, "grad_norm": 1.108688235282898, "learning_rate": 8.842476143851573e-06, "loss": 2.8642, "step": 209970 }, { "epoch": 0.0032768, "grad_norm": 0.7713078260421753, "learning_rate": 8.841676282130654e-06, "loss": 2.8969, "step": 209980 }, { "epoch": 0.0033024, "grad_norm": 0.7474251389503479, "learning_rate": 8.840876427921153e-06, "loss": 2.9581, "step": 209990 }, { "epoch": 0.003328, "grad_norm": 0.7393157482147217, "learning_rate": 8.840076581228263e-06, "loss": 2.9799, "step": 210000 }, { "epoch": 0.0033536, "grad_norm": 0.7711453437805176, "learning_rate": 8.839276742057169e-06, "loss": 3.0912, "step": 210010 }, { "epoch": 0.0033792, "grad_norm": 0.8678669929504395, "learning_rate": 8.838476910413053e-06, "loss": 2.7516, "step": 210020 }, { "epoch": 0.0034048, "grad_norm": 0.8983787298202515, "learning_rate": 8.83767708630111e-06, "loss": 2.7974, "step": 210030 }, { "epoch": 0.0034304, "grad_norm": 0.8420687317848206, "learning_rate": 8.83687726972652e-06, "loss": 3.1183, "step": 210040 }, { "epoch": 0.003456, "grad_norm": 0.7540594339370728, "learning_rate": 8.83607746069447e-06, "loss": 2.9881, "step": 210050 }, { "epoch": 0.0034816, "grad_norm": 0.7635976672172546, "learning_rate": 8.83527765921015e-06, "loss": 2.8681, "step": 210060 }, { "epoch": 0.0035072, "grad_norm": 0.832817018032074, "learning_rate": 8.834477865278745e-06, "loss": 2.7841, "step": 210070 }, { "epoch": 0.0035328, "grad_norm": 0.7615604400634766, "learning_rate": 8.83367807890544e-06, "loss": 2.7562, "step": 210080 }, { "epoch": 0.0035584, "grad_norm": 0.8646098971366882, "learning_rate": 8.832878300095424e-06, "loss": 2.8623, "step": 210090 }, { "epoch": 0.003584, "grad_norm": 0.8253235220909119, "learning_rate": 8.832078528853878e-06, "loss": 2.9262, "step": 210100 }, { "epoch": 0.0036096, "grad_norm": 0.7315114140510559, "learning_rate": 8.831278765185996e-06, "loss": 2.9662, "step": 210110 }, { "epoch": 0.0036352, "grad_norm": 0.8170020580291748, "learning_rate": 8.830479009096962e-06, "loss": 3.1676, "step": 210120 }, { "epoch": 0.0036608, "grad_norm": 0.8038083910942078, "learning_rate": 8.82967926059196e-06, "loss": 2.6736, "step": 210130 }, { "epoch": 0.0036864, "grad_norm": 0.7818100452423096, "learning_rate": 8.828879519676177e-06, "loss": 3.0599, "step": 210140 }, { "epoch": 0.003712, "grad_norm": 0.8089538216590881, "learning_rate": 8.828079786354797e-06, "loss": 2.9119, "step": 210150 }, { "epoch": 0.0037376, "grad_norm": 0.805147111415863, "learning_rate": 8.827280060633012e-06, "loss": 2.8962, "step": 210160 }, { "epoch": 0.0037632, "grad_norm": 0.7739282250404358, "learning_rate": 8.826480342516e-06, "loss": 2.9317, "step": 210170 }, { "epoch": 0.0037888, "grad_norm": 0.9118303656578064, "learning_rate": 8.825680632008952e-06, "loss": 3.0369, "step": 210180 }, { "epoch": 0.0038144, "grad_norm": 0.7600557208061218, "learning_rate": 8.824880929117055e-06, "loss": 2.857, "step": 210190 }, { "epoch": 0.00384, "grad_norm": 0.7903892993927002, "learning_rate": 8.824081233845492e-06, "loss": 2.7545, "step": 210200 }, { "epoch": 0.0038656, "grad_norm": 0.8181195855140686, "learning_rate": 8.823281546199449e-06, "loss": 3.057, "step": 210210 }, { "epoch": 0.0038912, "grad_norm": 0.7927067279815674, "learning_rate": 8.822481866184114e-06, "loss": 2.7276, "step": 210220 }, { "epoch": 0.0039168, "grad_norm": 0.7688618898391724, "learning_rate": 8.82168219380467e-06, "loss": 2.8675, "step": 210230 }, { "epoch": 0.0039424, "grad_norm": 1.0643064975738525, "learning_rate": 8.820882529066304e-06, "loss": 2.9552, "step": 210240 }, { "epoch": 0.003968, "grad_norm": 0.7049685716629028, "learning_rate": 8.820082871974201e-06, "loss": 2.8953, "step": 210250 }, { "epoch": 0.0039936, "grad_norm": 0.8140230774879456, "learning_rate": 8.819283222533547e-06, "loss": 2.9637, "step": 210260 }, { "epoch": 0.0040192, "grad_norm": 0.873725414276123, "learning_rate": 8.818483580749525e-06, "loss": 2.8614, "step": 210270 }, { "epoch": 0.0040448, "grad_norm": 0.8643167614936829, "learning_rate": 8.817683946627324e-06, "loss": 2.9924, "step": 210280 }, { "epoch": 0.0040704, "grad_norm": 0.7840427160263062, "learning_rate": 8.816884320172132e-06, "loss": 2.9923, "step": 210290 }, { "epoch": 0.004096, "grad_norm": 1.681199550628662, "learning_rate": 8.81608470138913e-06, "loss": 2.9607, "step": 210300 }, { "epoch": 0.0041216, "grad_norm": 0.7632778286933899, "learning_rate": 8.815285090283501e-06, "loss": 2.7462, "step": 210310 }, { "epoch": 0.0041472, "grad_norm": 1.1393928527832031, "learning_rate": 8.814485486860434e-06, "loss": 2.8411, "step": 210320 }, { "epoch": 0.0041728, "grad_norm": 0.8541383743286133, "learning_rate": 8.813685891125118e-06, "loss": 2.7969, "step": 210330 }, { "epoch": 0.0041984, "grad_norm": 0.7469300031661987, "learning_rate": 8.812886303082729e-06, "loss": 2.854, "step": 210340 }, { "epoch": 0.004224, "grad_norm": 0.853962779045105, "learning_rate": 8.812086722738458e-06, "loss": 2.8947, "step": 210350 }, { "epoch": 0.0042496, "grad_norm": 0.8454592227935791, "learning_rate": 8.811287150097492e-06, "loss": 2.982, "step": 210360 }, { "epoch": 0.0042752, "grad_norm": 1.1476026773452759, "learning_rate": 8.810487585165008e-06, "loss": 2.9718, "step": 210370 }, { "epoch": 0.0043008, "grad_norm": 0.7887254357337952, "learning_rate": 8.809688027946199e-06, "loss": 3.1146, "step": 210380 }, { "epoch": 0.0043264, "grad_norm": 1.8441544771194458, "learning_rate": 8.808888478446245e-06, "loss": 2.9843, "step": 210390 }, { "epoch": 0.004352, "grad_norm": 0.7399590015411377, "learning_rate": 8.808088936670334e-06, "loss": 3.0138, "step": 210400 }, { "epoch": 0.0043776, "grad_norm": 0.7739291787147522, "learning_rate": 8.807289402623648e-06, "loss": 2.756, "step": 210410 }, { "epoch": 0.0044032, "grad_norm": 0.8572872281074524, "learning_rate": 8.806489876311376e-06, "loss": 3.1296, "step": 210420 }, { "epoch": 0.0044288, "grad_norm": 0.8756565451622009, "learning_rate": 8.805690357738699e-06, "loss": 2.7898, "step": 210430 }, { "epoch": 0.0044544, "grad_norm": 0.7867383360862732, "learning_rate": 8.8048908469108e-06, "loss": 2.9483, "step": 210440 }, { "epoch": 0.00448, "grad_norm": 0.8905184268951416, "learning_rate": 8.804091343832865e-06, "loss": 2.6509, "step": 210450 }, { "epoch": 0.0045056, "grad_norm": 0.9919703602790833, "learning_rate": 8.803291848510084e-06, "loss": 2.8132, "step": 210460 }, { "epoch": 0.0045312, "grad_norm": 0.7812279462814331, "learning_rate": 8.802492360947636e-06, "loss": 3.1117, "step": 210470 }, { "epoch": 0.0045568, "grad_norm": 0.8951085805892944, "learning_rate": 8.801692881150705e-06, "loss": 2.8782, "step": 210480 }, { "epoch": 0.0045824, "grad_norm": 0.7926578521728516, "learning_rate": 8.80089340912448e-06, "loss": 2.8254, "step": 210490 }, { "epoch": 0.004608, "grad_norm": 0.7651793956756592, "learning_rate": 8.800093944874141e-06, "loss": 2.9712, "step": 210500 }, { "epoch": 0.0046336, "grad_norm": 0.9558988809585571, "learning_rate": 8.799294488404875e-06, "loss": 2.995, "step": 210510 }, { "epoch": 0.0046592, "grad_norm": 0.7115241289138794, "learning_rate": 8.798495039721864e-06, "loss": 2.8873, "step": 210520 }, { "epoch": 0.0046848, "grad_norm": 1.0474740266799927, "learning_rate": 8.797695598830296e-06, "loss": 3.0769, "step": 210530 }, { "epoch": 0.0047104, "grad_norm": 0.87917560338974, "learning_rate": 8.79689616573535e-06, "loss": 2.9294, "step": 210540 }, { "epoch": 0.004736, "grad_norm": 0.9375991225242615, "learning_rate": 8.796096740442212e-06, "loss": 2.7576, "step": 210550 }, { "epoch": 0.0047616, "grad_norm": 0.8228734135627747, "learning_rate": 8.79529732295607e-06, "loss": 2.8572, "step": 210560 }, { "epoch": 0.0047872, "grad_norm": 0.9732128977775574, "learning_rate": 8.794497913282102e-06, "loss": 3.3336, "step": 210570 }, { "epoch": 0.0048128, "grad_norm": 0.7606515288352966, "learning_rate": 8.793698511425495e-06, "loss": 2.8997, "step": 210580 }, { "epoch": 0.0048384, "grad_norm": 0.8425851464271545, "learning_rate": 8.792899117391432e-06, "loss": 3.0313, "step": 210590 }, { "epoch": 0.004864, "grad_norm": 0.7514724731445312, "learning_rate": 8.7920997311851e-06, "loss": 2.8496, "step": 210600 }, { "epoch": 0.0048896, "grad_norm": 0.7766870260238647, "learning_rate": 8.791300352811677e-06, "loss": 2.9261, "step": 210610 }, { "epoch": 0.0049152, "grad_norm": 0.848305881023407, "learning_rate": 8.790500982276348e-06, "loss": 2.7301, "step": 210620 }, { "epoch": 0.0049408, "grad_norm": 0.8292567133903503, "learning_rate": 8.789701619584302e-06, "loss": 2.7858, "step": 210630 }, { "epoch": 0.0049664, "grad_norm": 0.8276904225349426, "learning_rate": 8.788902264740719e-06, "loss": 2.8161, "step": 210640 }, { "epoch": 0.004992, "grad_norm": 0.7743882536888123, "learning_rate": 8.788102917750785e-06, "loss": 2.8047, "step": 210650 }, { "epoch": 0.0050176, "grad_norm": 0.8044727444648743, "learning_rate": 8.78730357861968e-06, "loss": 2.9683, "step": 210660 }, { "epoch": 0.0050432, "grad_norm": 0.7481581568717957, "learning_rate": 8.786504247352588e-06, "loss": 2.9196, "step": 210670 }, { "epoch": 0.0050688, "grad_norm": 0.9673894047737122, "learning_rate": 8.785704923954693e-06, "loss": 2.881, "step": 210680 }, { "epoch": 0.0050944, "grad_norm": 1.1047966480255127, "learning_rate": 8.784905608431183e-06, "loss": 2.851, "step": 210690 }, { "epoch": 0.00512, "grad_norm": 0.8224273920059204, "learning_rate": 8.784106300787233e-06, "loss": 2.6612, "step": 210700 }, { "epoch": 0.0051456, "grad_norm": 1.0013372898101807, "learning_rate": 8.78330700102803e-06, "loss": 3.0234, "step": 210710 }, { "epoch": 0.0051712, "grad_norm": 0.9015184044837952, "learning_rate": 8.78250770915876e-06, "loss": 2.97, "step": 210720 }, { "epoch": 0.0051968, "grad_norm": 0.7612518072128296, "learning_rate": 8.781708425184605e-06, "loss": 2.9268, "step": 210730 }, { "epoch": 0.0052224, "grad_norm": 0.7936068177223206, "learning_rate": 8.780909149110744e-06, "loss": 2.9142, "step": 210740 }, { "epoch": 0.005248, "grad_norm": 0.8593422174453735, "learning_rate": 8.780109880942365e-06, "loss": 2.912, "step": 210750 }, { "epoch": 0.0052736, "grad_norm": 0.7997118234634399, "learning_rate": 8.77931062068465e-06, "loss": 2.9567, "step": 210760 }, { "epoch": 0.0052992, "grad_norm": 0.9658674597740173, "learning_rate": 8.77851136834278e-06, "loss": 2.8804, "step": 210770 }, { "epoch": 0.0053248, "grad_norm": 0.7994981408119202, "learning_rate": 8.777712123921938e-06, "loss": 2.9299, "step": 210780 }, { "epoch": 0.0053504, "grad_norm": 0.7737393975257874, "learning_rate": 8.776912887427307e-06, "loss": 3.0308, "step": 210790 }, { "epoch": 0.005376, "grad_norm": 0.8619610667228699, "learning_rate": 8.776113658864073e-06, "loss": 2.9928, "step": 210800 }, { "epoch": 0.0054016, "grad_norm": 0.7396864891052246, "learning_rate": 8.775314438237417e-06, "loss": 2.8727, "step": 210810 }, { "epoch": 0.0054272, "grad_norm": 0.8626465797424316, "learning_rate": 8.774515225552522e-06, "loss": 2.7992, "step": 210820 }, { "epoch": 0.0054528, "grad_norm": 0.8567948937416077, "learning_rate": 8.773716020814568e-06, "loss": 2.8275, "step": 210830 }, { "epoch": 0.0054784, "grad_norm": 0.7800024151802063, "learning_rate": 8.772916824028741e-06, "loss": 2.8664, "step": 210840 }, { "epoch": 0.005504, "grad_norm": 0.7611941695213318, "learning_rate": 8.772117635200224e-06, "loss": 2.7583, "step": 210850 }, { "epoch": 0.0055296, "grad_norm": 0.7778162360191345, "learning_rate": 8.771318454334196e-06, "loss": 2.9525, "step": 210860 }, { "epoch": 0.0055552, "grad_norm": 0.7347140908241272, "learning_rate": 8.77051928143584e-06, "loss": 2.8405, "step": 210870 }, { "epoch": 0.0055808, "grad_norm": 0.9224341511726379, "learning_rate": 8.76972011651034e-06, "loss": 2.8446, "step": 210880 }, { "epoch": 0.0056064, "grad_norm": 0.732715368270874, "learning_rate": 8.768920959562882e-06, "loss": 2.8799, "step": 210890 }, { "epoch": 0.005632, "grad_norm": 0.8420145511627197, "learning_rate": 8.76812181059864e-06, "loss": 3.0315, "step": 210900 }, { "epoch": 0.0056576, "grad_norm": 0.8434765934944153, "learning_rate": 8.767322669622801e-06, "loss": 2.8504, "step": 210910 }, { "epoch": 0.0056832, "grad_norm": 0.9919111728668213, "learning_rate": 8.76652353664055e-06, "loss": 2.5857, "step": 210920 }, { "epoch": 0.0057088, "grad_norm": 0.7867773175239563, "learning_rate": 8.765724411657064e-06, "loss": 2.9699, "step": 210930 }, { "epoch": 0.0057344, "grad_norm": 0.7287809252738953, "learning_rate": 8.764925294677526e-06, "loss": 2.9099, "step": 210940 }, { "epoch": 0.00576, "grad_norm": 0.8457443118095398, "learning_rate": 8.764126185707121e-06, "loss": 2.9035, "step": 210950 }, { "epoch": 0.0057856, "grad_norm": 0.7572841048240662, "learning_rate": 8.76332708475103e-06, "loss": 2.8006, "step": 210960 }, { "epoch": 0.0058112, "grad_norm": 0.686055600643158, "learning_rate": 8.762527991814427e-06, "loss": 2.924, "step": 210970 }, { "epoch": 0.0058368, "grad_norm": 0.8061091303825378, "learning_rate": 8.761728906902511e-06, "loss": 2.8189, "step": 210980 }, { "epoch": 0.0058624, "grad_norm": 0.9416834115982056, "learning_rate": 8.760929830020448e-06, "loss": 2.8882, "step": 210990 }, { "epoch": 0.005888, "grad_norm": 0.7531444430351257, "learning_rate": 8.760130761173427e-06, "loss": 2.8307, "step": 211000 }, { "epoch": 0.0059136, "grad_norm": 1.0306917428970337, "learning_rate": 8.759331700366628e-06, "loss": 3.0794, "step": 211010 }, { "epoch": 0.0059392, "grad_norm": 0.8048553466796875, "learning_rate": 8.758532647605237e-06, "loss": 2.6226, "step": 211020 }, { "epoch": 0.0059648, "grad_norm": 2.7778987884521484, "learning_rate": 8.757733602894428e-06, "loss": 3.1881, "step": 211030 }, { "epoch": 0.0059904, "grad_norm": 0.8792504668235779, "learning_rate": 8.756934566239388e-06, "loss": 3.0573, "step": 211040 }, { "epoch": 0.006016, "grad_norm": 0.9410239458084106, "learning_rate": 8.756135537645298e-06, "loss": 2.929, "step": 211050 }, { "epoch": 0.0060416, "grad_norm": 0.8096909523010254, "learning_rate": 8.755336517117337e-06, "loss": 2.7634, "step": 211060 }, { "epoch": 0.0060672, "grad_norm": 1.8190909624099731, "learning_rate": 8.754537504660687e-06, "loss": 2.7838, "step": 211070 }, { "epoch": 0.0060928, "grad_norm": 1.660557508468628, "learning_rate": 8.753738500280531e-06, "loss": 3.0552, "step": 211080 }, { "epoch": 0.0061184, "grad_norm": 0.7938266396522522, "learning_rate": 8.752939503982051e-06, "loss": 2.9335, "step": 211090 }, { "epoch": 0.006144, "grad_norm": 0.8554903864860535, "learning_rate": 8.752140515770425e-06, "loss": 2.7281, "step": 211100 }, { "epoch": 0.0061696, "grad_norm": 0.9743415713310242, "learning_rate": 8.751341535650836e-06, "loss": 2.846, "step": 211110 }, { "epoch": 0.0061952, "grad_norm": 0.7680554389953613, "learning_rate": 8.750542563628468e-06, "loss": 3.0071, "step": 211120 }, { "epoch": 0.0062208, "grad_norm": 0.7761518955230713, "learning_rate": 8.749743599708496e-06, "loss": 2.9173, "step": 211130 }, { "epoch": 0.0062464, "grad_norm": 1.0009783506393433, "learning_rate": 8.748944643896104e-06, "loss": 2.7602, "step": 211140 }, { "epoch": 0.006272, "grad_norm": 0.6812371611595154, "learning_rate": 8.748145696196474e-06, "loss": 2.8892, "step": 211150 }, { "epoch": 0.0062976, "grad_norm": 1.0212960243225098, "learning_rate": 8.747346756614788e-06, "loss": 2.8813, "step": 211160 }, { "epoch": 0.0063232, "grad_norm": 0.7049535512924194, "learning_rate": 8.746547825156224e-06, "loss": 2.8641, "step": 211170 }, { "epoch": 0.0063488, "grad_norm": 0.8608631491661072, "learning_rate": 8.745748901825968e-06, "loss": 2.9464, "step": 211180 }, { "epoch": 0.0063744, "grad_norm": 0.8270500302314758, "learning_rate": 8.744949986629194e-06, "loss": 2.7012, "step": 211190 }, { "epoch": 0.0064, "grad_norm": 0.9829452633857727, "learning_rate": 8.744151079571086e-06, "loss": 2.8946, "step": 211200 }, { "epoch": 0.0064256, "grad_norm": 0.9880341291427612, "learning_rate": 8.743352180656824e-06, "loss": 2.7924, "step": 211210 }, { "epoch": 0.0064512, "grad_norm": 0.7728831768035889, "learning_rate": 8.742553289891591e-06, "loss": 3.0309, "step": 211220 }, { "epoch": 0.0064768, "grad_norm": 0.7969833612442017, "learning_rate": 8.741754407280565e-06, "loss": 2.9166, "step": 211230 }, { "epoch": 0.0065024, "grad_norm": 0.7887051105499268, "learning_rate": 8.740955532828928e-06, "loss": 2.9311, "step": 211240 }, { "epoch": 0.006528, "grad_norm": 0.7284262180328369, "learning_rate": 8.740156666541859e-06, "loss": 2.9696, "step": 211250 }, { "epoch": 0.0065536, "grad_norm": 0.8264285922050476, "learning_rate": 8.739357808424539e-06, "loss": 2.7204, "step": 211260 }, { "epoch": 0.0065792, "grad_norm": 0.7659797072410583, "learning_rate": 8.738558958482149e-06, "loss": 2.9201, "step": 211270 }, { "epoch": 0.0066048, "grad_norm": 0.7631392478942871, "learning_rate": 8.737760116719866e-06, "loss": 2.792, "step": 211280 }, { "epoch": 0.0066304, "grad_norm": 1.338452935218811, "learning_rate": 8.736961283142878e-06, "loss": 2.9, "step": 211290 }, { "epoch": 0.006656, "grad_norm": 0.7889070510864258, "learning_rate": 8.736162457756358e-06, "loss": 2.7146, "step": 211300 }, { "epoch": 0.0066816, "grad_norm": 2.9855756759643555, "learning_rate": 8.735363640565488e-06, "loss": 3.3928, "step": 211310 }, { "epoch": 0.0067072, "grad_norm": 0.7437244653701782, "learning_rate": 8.734564831575446e-06, "loss": 2.9239, "step": 211320 }, { "epoch": 0.0067328, "grad_norm": 0.8910791873931885, "learning_rate": 8.733766030791418e-06, "loss": 2.889, "step": 211330 }, { "epoch": 0.0067584, "grad_norm": 0.7491727471351624, "learning_rate": 8.732967238218581e-06, "loss": 2.8865, "step": 211340 }, { "epoch": 0.006784, "grad_norm": 0.7293985486030579, "learning_rate": 8.732168453862115e-06, "loss": 2.8333, "step": 211350 }, { "epoch": 0.0068096, "grad_norm": 0.8206905722618103, "learning_rate": 8.731369677727198e-06, "loss": 2.8852, "step": 211360 }, { "epoch": 0.0068352, "grad_norm": 0.725385308265686, "learning_rate": 8.730570909819012e-06, "loss": 2.9483, "step": 211370 }, { "epoch": 0.0068608, "grad_norm": 1.0906445980072021, "learning_rate": 8.729772150142738e-06, "loss": 2.7299, "step": 211380 }, { "epoch": 0.0068864, "grad_norm": 0.8877891898155212, "learning_rate": 8.728973398703553e-06, "loss": 2.8811, "step": 211390 }, { "epoch": 0.006912, "grad_norm": 0.7880367040634155, "learning_rate": 8.728174655506637e-06, "loss": 2.933, "step": 211400 }, { "epoch": 0.0069376, "grad_norm": 0.7986512780189514, "learning_rate": 8.727375920557171e-06, "loss": 2.9434, "step": 211410 }, { "epoch": 0.0069632, "grad_norm": 0.9080222249031067, "learning_rate": 8.726577193860335e-06, "loss": 3.0221, "step": 211420 }, { "epoch": 0.0069888, "grad_norm": 0.7055436968803406, "learning_rate": 8.725778475421306e-06, "loss": 2.7635, "step": 211430 }, { "epoch": 0.0070144, "grad_norm": 0.7973359227180481, "learning_rate": 8.724979765245265e-06, "loss": 2.6696, "step": 211440 }, { "epoch": 0.00704, "grad_norm": 2.1921794414520264, "learning_rate": 8.724181063337393e-06, "loss": 2.7402, "step": 211450 }, { "epoch": 0.0070656, "grad_norm": 0.7321155071258545, "learning_rate": 8.723382369702867e-06, "loss": 2.7739, "step": 211460 }, { "epoch": 0.0070912, "grad_norm": 0.7757271528244019, "learning_rate": 8.722583684346865e-06, "loss": 2.9582, "step": 211470 }, { "epoch": 0.0071168, "grad_norm": 0.7780730724334717, "learning_rate": 8.721785007274569e-06, "loss": 2.7485, "step": 211480 }, { "epoch": 0.0071424, "grad_norm": 0.8236942291259766, "learning_rate": 8.720986338491156e-06, "loss": 2.8499, "step": 211490 }, { "epoch": 0.007168, "grad_norm": 0.9369782209396362, "learning_rate": 8.720187678001808e-06, "loss": 2.8216, "step": 211500 }, { "epoch": 0.0071936, "grad_norm": 0.821841299533844, "learning_rate": 8.719389025811706e-06, "loss": 2.9858, "step": 211510 }, { "epoch": 0.0072192, "grad_norm": 0.8503934144973755, "learning_rate": 8.718590381926022e-06, "loss": 2.7762, "step": 211520 }, { "epoch": 0.0072448, "grad_norm": 0.7169243097305298, "learning_rate": 8.717791746349939e-06, "loss": 2.9935, "step": 211530 }, { "epoch": 0.0072704, "grad_norm": 0.868732750415802, "learning_rate": 8.716993119088638e-06, "loss": 3.1056, "step": 211540 }, { "epoch": 0.007296, "grad_norm": 0.7749574184417725, "learning_rate": 8.716194500147295e-06, "loss": 2.8829, "step": 211550 }, { "epoch": 0.0073216, "grad_norm": 0.8162074685096741, "learning_rate": 8.71539588953109e-06, "loss": 2.7425, "step": 211560 }, { "epoch": 0.0073472, "grad_norm": 0.707055926322937, "learning_rate": 8.7145972872452e-06, "loss": 2.9601, "step": 211570 }, { "epoch": 0.0073728, "grad_norm": 0.7432527542114258, "learning_rate": 8.713798693294806e-06, "loss": 2.847, "step": 211580 }, { "epoch": 0.0073984, "grad_norm": 0.9709039926528931, "learning_rate": 8.713000107685085e-06, "loss": 2.9375, "step": 211590 }, { "epoch": 0.007424, "grad_norm": 0.7977606654167175, "learning_rate": 8.712201530421216e-06, "loss": 2.894, "step": 211600 }, { "epoch": 0.0074496, "grad_norm": 0.6831883192062378, "learning_rate": 8.711402961508378e-06, "loss": 2.7426, "step": 211610 }, { "epoch": 0.0074752, "grad_norm": 0.8108482360839844, "learning_rate": 8.71060440095175e-06, "loss": 3.0089, "step": 211620 }, { "epoch": 0.0075008, "grad_norm": 0.7965016961097717, "learning_rate": 8.70980584875651e-06, "loss": 2.969, "step": 211630 }, { "epoch": 0.0075264, "grad_norm": 0.7861230969429016, "learning_rate": 8.709007304927834e-06, "loss": 2.9684, "step": 211640 }, { "epoch": 0.007552, "grad_norm": 0.8952104449272156, "learning_rate": 8.708208769470905e-06, "loss": 2.912, "step": 211650 }, { "epoch": 0.0075776, "grad_norm": 0.8021107316017151, "learning_rate": 8.707410242390893e-06, "loss": 2.8949, "step": 211660 }, { "epoch": 0.0076032, "grad_norm": 0.8994021415710449, "learning_rate": 8.706611723692989e-06, "loss": 2.8192, "step": 211670 }, { "epoch": 0.0076288, "grad_norm": 0.9255643486976624, "learning_rate": 8.705813213382361e-06, "loss": 2.9366, "step": 211680 }, { "epoch": 0.0076544, "grad_norm": 0.7444255948066711, "learning_rate": 8.705014711464191e-06, "loss": 3.0485, "step": 211690 }, { "epoch": 0.00768, "grad_norm": 0.8267530798912048, "learning_rate": 8.704216217943657e-06, "loss": 2.7719, "step": 211700 }, { "epoch": 0.0077056, "grad_norm": 0.7583630681037903, "learning_rate": 8.703417732825937e-06, "loss": 2.9439, "step": 211710 }, { "epoch": 0.0077312, "grad_norm": 0.7211284041404724, "learning_rate": 8.702619256116208e-06, "loss": 2.6939, "step": 211720 }, { "epoch": 0.0077568, "grad_norm": 0.7853361964225769, "learning_rate": 8.701820787819647e-06, "loss": 2.8407, "step": 211730 }, { "epoch": 0.0077824, "grad_norm": 2.558685541152954, "learning_rate": 8.701022327941437e-06, "loss": 3.27, "step": 211740 }, { "epoch": 0.007808, "grad_norm": 1.875403642654419, "learning_rate": 8.700223876486747e-06, "loss": 2.7464, "step": 211750 }, { "epoch": 0.0078336, "grad_norm": 0.7949110865592957, "learning_rate": 8.699425433460763e-06, "loss": 2.8774, "step": 211760 }, { "epoch": 0.0078592, "grad_norm": 0.8071862459182739, "learning_rate": 8.698626998868656e-06, "loss": 3.0023, "step": 211770 }, { "epoch": 0.0078848, "grad_norm": 0.7731406688690186, "learning_rate": 8.697828572715611e-06, "loss": 2.8738, "step": 211780 }, { "epoch": 0.0079104, "grad_norm": 0.7066157460212708, "learning_rate": 8.697030155006801e-06, "loss": 2.6626, "step": 211790 }, { "epoch": 0.007936, "grad_norm": 0.8107637166976929, "learning_rate": 8.696231745747402e-06, "loss": 2.8422, "step": 211800 }, { "epoch": 0.0079616, "grad_norm": 0.808540940284729, "learning_rate": 8.695433344942594e-06, "loss": 3.1069, "step": 211810 }, { "epoch": 0.0079872, "grad_norm": 0.7384865283966064, "learning_rate": 8.694634952597557e-06, "loss": 2.9215, "step": 211820 }, { "epoch": 0.0080128, "grad_norm": 0.8755768537521362, "learning_rate": 8.693836568717462e-06, "loss": 2.9875, "step": 211830 }, { "epoch": 0.0080384, "grad_norm": 0.7208448052406311, "learning_rate": 8.693038193307488e-06, "loss": 2.9244, "step": 211840 }, { "epoch": 0.008064, "grad_norm": 0.7341676950454712, "learning_rate": 8.692239826372817e-06, "loss": 2.6975, "step": 211850 }, { "epoch": 0.0080896, "grad_norm": 0.8341747522354126, "learning_rate": 8.691441467918622e-06, "loss": 3.0713, "step": 211860 }, { "epoch": 0.0081152, "grad_norm": 1.2779011726379395, "learning_rate": 8.690643117950084e-06, "loss": 2.9163, "step": 211870 }, { "epoch": 0.0081408, "grad_norm": 0.7767733335494995, "learning_rate": 8.689844776472377e-06, "loss": 2.8611, "step": 211880 }, { "epoch": 0.0081664, "grad_norm": 0.755425214767456, "learning_rate": 8.689046443490675e-06, "loss": 2.842, "step": 211890 }, { "epoch": 0.008192, "grad_norm": 0.7358736991882324, "learning_rate": 8.688248119010161e-06, "loss": 2.9775, "step": 211900 }, { "epoch": 0.0082176, "grad_norm": 0.8929316401481628, "learning_rate": 8.68744980303601e-06, "loss": 3.049, "step": 211910 }, { "epoch": 0.0082432, "grad_norm": 0.7199826836585999, "learning_rate": 8.686651495573397e-06, "loss": 2.8819, "step": 211920 }, { "epoch": 0.0082688, "grad_norm": 0.763916552066803, "learning_rate": 8.6858531966275e-06, "loss": 2.85, "step": 211930 }, { "epoch": 0.0082944, "grad_norm": 0.8380950689315796, "learning_rate": 8.685054906203496e-06, "loss": 3.0356, "step": 211940 }, { "epoch": 0.00832, "grad_norm": 0.7999187111854553, "learning_rate": 8.684256624306563e-06, "loss": 2.9605, "step": 211950 }, { "epoch": 0.0083456, "grad_norm": 0.7828765511512756, "learning_rate": 8.683458350941872e-06, "loss": 2.9689, "step": 211960 }, { "epoch": 0.0083712, "grad_norm": 0.8250520825386047, "learning_rate": 8.682660086114605e-06, "loss": 2.7304, "step": 211970 }, { "epoch": 0.0083968, "grad_norm": 0.8770881295204163, "learning_rate": 8.68186182982994e-06, "loss": 2.9015, "step": 211980 }, { "epoch": 0.0084224, "grad_norm": 0.7514860033988953, "learning_rate": 8.68106358209305e-06, "loss": 2.7909, "step": 211990 }, { "epoch": 0.008448, "grad_norm": 0.9133517742156982, "learning_rate": 8.680265342909112e-06, "loss": 2.7643, "step": 212000 }, { "epoch": 0.0084736, "grad_norm": 0.7572235465049744, "learning_rate": 8.679467112283298e-06, "loss": 2.9064, "step": 212010 }, { "epoch": 0.0084992, "grad_norm": 0.7580674290657043, "learning_rate": 8.678668890220792e-06, "loss": 3.0731, "step": 212020 }, { "epoch": 0.0085248, "grad_norm": 0.7099090814590454, "learning_rate": 8.677870676726766e-06, "loss": 2.9329, "step": 212030 }, { "epoch": 0.0085504, "grad_norm": 0.9270772933959961, "learning_rate": 8.677072471806402e-06, "loss": 2.9308, "step": 212040 }, { "epoch": 0.008576, "grad_norm": 0.8740893602371216, "learning_rate": 8.676274275464866e-06, "loss": 2.8747, "step": 212050 }, { "epoch": 0.0086016, "grad_norm": 1.1054868698120117, "learning_rate": 8.675476087707342e-06, "loss": 2.9202, "step": 212060 }, { "epoch": 0.0086272, "grad_norm": 0.8139952421188354, "learning_rate": 8.674677908539005e-06, "loss": 2.9228, "step": 212070 }, { "epoch": 0.0086528, "grad_norm": 0.7217222452163696, "learning_rate": 8.673879737965027e-06, "loss": 2.9327, "step": 212080 }, { "epoch": 0.0086784, "grad_norm": 0.7785961627960205, "learning_rate": 8.673081575990588e-06, "loss": 2.9168, "step": 212090 }, { "epoch": 0.008704, "grad_norm": 0.7734640836715698, "learning_rate": 8.67228342262086e-06, "loss": 2.7889, "step": 212100 }, { "epoch": 0.0087296, "grad_norm": 1.0331530570983887, "learning_rate": 8.671485277861026e-06, "loss": 2.8886, "step": 212110 }, { "epoch": 0.0087552, "grad_norm": 0.9592775106430054, "learning_rate": 8.670687141716253e-06, "loss": 2.805, "step": 212120 }, { "epoch": 0.0087808, "grad_norm": 0.7567878365516663, "learning_rate": 8.66988901419172e-06, "loss": 2.9238, "step": 212130 }, { "epoch": 0.0088064, "grad_norm": 0.8610472083091736, "learning_rate": 8.669090895292606e-06, "loss": 2.942, "step": 212140 }, { "epoch": 0.008832, "grad_norm": 0.8049466013908386, "learning_rate": 8.668292785024082e-06, "loss": 2.8653, "step": 212150 }, { "epoch": 0.0088576, "grad_norm": 0.7213836312294006, "learning_rate": 8.667494683391327e-06, "loss": 2.9528, "step": 212160 }, { "epoch": 0.0088832, "grad_norm": 0.7642658948898315, "learning_rate": 8.666696590399513e-06, "loss": 2.7987, "step": 212170 }, { "epoch": 0.0089088, "grad_norm": 0.8452383279800415, "learning_rate": 8.66589850605382e-06, "loss": 2.8897, "step": 212180 }, { "epoch": 0.0089344, "grad_norm": 0.857989490032196, "learning_rate": 8.665100430359412e-06, "loss": 2.9063, "step": 212190 }, { "epoch": 0.00896, "grad_norm": 0.7861829996109009, "learning_rate": 8.664302363321482e-06, "loss": 2.9639, "step": 212200 }, { "epoch": 0.0089856, "grad_norm": 0.7715269327163696, "learning_rate": 8.663504304945194e-06, "loss": 3.0782, "step": 212210 }, { "epoch": 0.0090112, "grad_norm": 0.7647062540054321, "learning_rate": 8.662706255235724e-06, "loss": 2.9269, "step": 212220 }, { "epoch": 0.0090368, "grad_norm": 1.0964045524597168, "learning_rate": 8.66190821419825e-06, "loss": 2.8891, "step": 212230 }, { "epoch": 0.0090624, "grad_norm": 1.157030701637268, "learning_rate": 8.661110181837945e-06, "loss": 2.8151, "step": 212240 }, { "epoch": 0.009088, "grad_norm": 0.8093422651290894, "learning_rate": 8.660312158159984e-06, "loss": 2.9394, "step": 212250 }, { "epoch": 0.0091136, "grad_norm": 0.93422532081604, "learning_rate": 8.659514143169543e-06, "loss": 2.9277, "step": 212260 }, { "epoch": 0.0091392, "grad_norm": 0.8212112188339233, "learning_rate": 8.6587161368718e-06, "loss": 2.9102, "step": 212270 }, { "epoch": 0.0091648, "grad_norm": 0.8041401505470276, "learning_rate": 8.657918139271922e-06, "loss": 2.8698, "step": 212280 }, { "epoch": 0.0091904, "grad_norm": 0.7922601699829102, "learning_rate": 8.65712015037509e-06, "loss": 2.8467, "step": 212290 }, { "epoch": 0.009216, "grad_norm": 1.1020070314407349, "learning_rate": 8.656322170186475e-06, "loss": 2.8929, "step": 212300 }, { "epoch": 0.0092416, "grad_norm": 0.6992572546005249, "learning_rate": 8.655524198711256e-06, "loss": 2.7959, "step": 212310 }, { "epoch": 0.0092672, "grad_norm": 0.7688683867454529, "learning_rate": 8.654726235954605e-06, "loss": 2.8386, "step": 212320 }, { "epoch": 0.0092928, "grad_norm": 0.7502524256706238, "learning_rate": 8.653928281921695e-06, "loss": 2.8079, "step": 212330 }, { "epoch": 0.0093184, "grad_norm": 0.727428674697876, "learning_rate": 8.653130336617706e-06, "loss": 2.7808, "step": 212340 }, { "epoch": 0.009344, "grad_norm": 0.780292272567749, "learning_rate": 8.652332400047806e-06, "loss": 3.079, "step": 212350 }, { "epoch": 0.0093696, "grad_norm": 0.7715849876403809, "learning_rate": 8.65153447221717e-06, "loss": 3.035, "step": 212360 }, { "epoch": 0.0093952, "grad_norm": 0.8486058712005615, "learning_rate": 8.650736553130975e-06, "loss": 2.7849, "step": 212370 }, { "epoch": 0.0094208, "grad_norm": 0.7476438879966736, "learning_rate": 8.649938642794397e-06, "loss": 3.0311, "step": 212380 }, { "epoch": 0.0094464, "grad_norm": 0.8323671817779541, "learning_rate": 8.649140741212607e-06, "loss": 2.8833, "step": 212390 }, { "epoch": 0.009472, "grad_norm": 0.7269765138626099, "learning_rate": 8.648342848390783e-06, "loss": 2.9071, "step": 212400 }, { "epoch": 0.0094976, "grad_norm": 0.6888360381126404, "learning_rate": 8.647544964334095e-06, "loss": 2.7416, "step": 212410 }, { "epoch": 0.0095232, "grad_norm": 0.7578864693641663, "learning_rate": 8.646747089047717e-06, "loss": 2.9372, "step": 212420 }, { "epoch": 0.0095488, "grad_norm": 0.7410658597946167, "learning_rate": 8.645949222536826e-06, "loss": 2.9566, "step": 212430 }, { "epoch": 0.0095744, "grad_norm": 0.8837472200393677, "learning_rate": 8.645151364806595e-06, "loss": 2.7773, "step": 212440 }, { "epoch": 0.0096, "grad_norm": 0.8716523051261902, "learning_rate": 8.644353515862195e-06, "loss": 3.072, "step": 212450 }, { "epoch": 0.0096256, "grad_norm": 0.8642404079437256, "learning_rate": 8.643555675708804e-06, "loss": 2.8238, "step": 212460 }, { "epoch": 0.0096512, "grad_norm": 0.746936559677124, "learning_rate": 8.642757844351595e-06, "loss": 2.996, "step": 212470 }, { "epoch": 0.0096768, "grad_norm": 0.7765989899635315, "learning_rate": 8.641960021795738e-06, "loss": 2.7426, "step": 212480 }, { "epoch": 0.0097024, "grad_norm": 0.7306638360023499, "learning_rate": 8.64116220804641e-06, "loss": 2.8396, "step": 212490 }, { "epoch": 0.009728, "grad_norm": 0.815329909324646, "learning_rate": 8.640364403108784e-06, "loss": 2.9821, "step": 212500 }, { "epoch": 0.0097536, "grad_norm": 1.004672646522522, "learning_rate": 8.639566606988036e-06, "loss": 2.9983, "step": 212510 }, { "epoch": 0.0097792, "grad_norm": 0.865272581577301, "learning_rate": 8.638768819689335e-06, "loss": 2.9866, "step": 212520 }, { "epoch": 0.0098048, "grad_norm": 0.7986000180244446, "learning_rate": 8.637971041217855e-06, "loss": 2.7741, "step": 212530 }, { "epoch": 0.0098304, "grad_norm": 0.9661323428153992, "learning_rate": 8.63717327157877e-06, "loss": 3.0643, "step": 212540 }, { "epoch": 0.009856, "grad_norm": 0.8360763192176819, "learning_rate": 8.636375510777256e-06, "loss": 2.7835, "step": 212550 }, { "epoch": 0.0098816, "grad_norm": 0.809637188911438, "learning_rate": 8.635577758818487e-06, "loss": 2.9881, "step": 212560 }, { "epoch": 0.0099072, "grad_norm": 0.737308919429779, "learning_rate": 8.63478001570763e-06, "loss": 2.8536, "step": 212570 }, { "epoch": 0.0099328, "grad_norm": 0.7254703640937805, "learning_rate": 8.633982281449863e-06, "loss": 2.9274, "step": 212580 }, { "epoch": 0.0099584, "grad_norm": 0.7336729764938354, "learning_rate": 8.633184556050359e-06, "loss": 2.9191, "step": 212590 }, { "epoch": 0.009984, "grad_norm": 1.0867220163345337, "learning_rate": 8.63238683951429e-06, "loss": 2.9012, "step": 212600 }, { "epoch": 0.0100096, "grad_norm": 0.9436990022659302, "learning_rate": 8.631589131846826e-06, "loss": 2.8837, "step": 212610 }, { "epoch": 0.0100352, "grad_norm": 0.8432207703590393, "learning_rate": 8.630791433053146e-06, "loss": 2.9422, "step": 212620 }, { "epoch": 0.0100608, "grad_norm": 0.726723849773407, "learning_rate": 8.629993743138418e-06, "loss": 3.1081, "step": 212630 }, { "epoch": 0.0100864, "grad_norm": 0.7023378610610962, "learning_rate": 8.629196062107816e-06, "loss": 2.9886, "step": 212640 }, { "epoch": 0.010112, "grad_norm": 0.8422847986221313, "learning_rate": 8.628398389966515e-06, "loss": 3.0879, "step": 212650 }, { "epoch": 0.0101376, "grad_norm": 0.7433765530586243, "learning_rate": 8.627600726719683e-06, "loss": 2.9372, "step": 212660 }, { "epoch": 0.0101632, "grad_norm": 0.7288565039634705, "learning_rate": 8.626803072372499e-06, "loss": 2.9537, "step": 212670 }, { "epoch": 0.0101888, "grad_norm": 0.836365818977356, "learning_rate": 8.62600542693013e-06, "loss": 3.0896, "step": 212680 }, { "epoch": 0.0102144, "grad_norm": 0.8358953595161438, "learning_rate": 8.625207790397752e-06, "loss": 2.7378, "step": 212690 }, { "epoch": 0.01024, "grad_norm": 1.0966572761535645, "learning_rate": 8.624410162780534e-06, "loss": 2.7663, "step": 212700 }, { "epoch": 0.0102656, "grad_norm": 0.7848294377326965, "learning_rate": 8.623612544083648e-06, "loss": 2.8558, "step": 212710 }, { "epoch": 0.0102912, "grad_norm": 0.8305509686470032, "learning_rate": 8.622814934312274e-06, "loss": 2.798, "step": 212720 }, { "epoch": 0.0103168, "grad_norm": 0.7468937635421753, "learning_rate": 8.622017333471579e-06, "loss": 2.8857, "step": 212730 }, { "epoch": 0.0103424, "grad_norm": 1.1811184883117676, "learning_rate": 8.621219741566734e-06, "loss": 2.8008, "step": 212740 }, { "epoch": 0.010368, "grad_norm": 0.8853616118431091, "learning_rate": 8.620422158602911e-06, "loss": 2.9967, "step": 212750 }, { "epoch": 0.0103936, "grad_norm": 0.7959702610969543, "learning_rate": 8.619624584585286e-06, "loss": 2.9554, "step": 212760 }, { "epoch": 0.0104192, "grad_norm": 0.8084654211997986, "learning_rate": 8.618827019519029e-06, "loss": 2.9009, "step": 212770 }, { "epoch": 0.0104448, "grad_norm": 0.720280110836029, "learning_rate": 8.61802946340931e-06, "loss": 3.1361, "step": 212780 }, { "epoch": 0.0104704, "grad_norm": 0.7499275803565979, "learning_rate": 8.617231916261303e-06, "loss": 3.1844, "step": 212790 }, { "epoch": 0.010496, "grad_norm": 0.793320894241333, "learning_rate": 8.61643437808018e-06, "loss": 2.9605, "step": 212800 }, { "epoch": 0.0105216, "grad_norm": 1.0520097017288208, "learning_rate": 8.615636848871113e-06, "loss": 2.8787, "step": 212810 }, { "epoch": 0.0105472, "grad_norm": 0.7660418748855591, "learning_rate": 8.614839328639272e-06, "loss": 2.8609, "step": 212820 }, { "epoch": 0.0105728, "grad_norm": 0.8498163819313049, "learning_rate": 8.614041817389833e-06, "loss": 2.8395, "step": 212830 }, { "epoch": 0.0105984, "grad_norm": 0.8139067888259888, "learning_rate": 8.61324431512796e-06, "loss": 2.8564, "step": 212840 }, { "epoch": 0.010624, "grad_norm": 0.7497142553329468, "learning_rate": 8.612446821858832e-06, "loss": 2.8699, "step": 212850 }, { "epoch": 0.0106496, "grad_norm": 0.7556332349777222, "learning_rate": 8.611649337587613e-06, "loss": 2.8751, "step": 212860 }, { "epoch": 0.0106752, "grad_norm": 0.7902863621711731, "learning_rate": 8.610851862319485e-06, "loss": 3.0301, "step": 212870 }, { "epoch": 0.0107008, "grad_norm": 0.7676848769187927, "learning_rate": 8.610054396059606e-06, "loss": 3.0851, "step": 212880 }, { "epoch": 0.0107264, "grad_norm": 1.0213630199432373, "learning_rate": 8.609256938813162e-06, "loss": 2.9198, "step": 212890 }, { "epoch": 0.010752, "grad_norm": 0.7702681422233582, "learning_rate": 8.608459490585313e-06, "loss": 2.8271, "step": 212900 }, { "epoch": 0.0107776, "grad_norm": 0.8528329730033875, "learning_rate": 8.607662051381235e-06, "loss": 2.9833, "step": 212910 }, { "epoch": 0.0108032, "grad_norm": 0.9924198389053345, "learning_rate": 8.606864621206102e-06, "loss": 2.7147, "step": 212920 }, { "epoch": 0.0108288, "grad_norm": 1.0938211679458618, "learning_rate": 8.606067200065081e-06, "loss": 2.7653, "step": 212930 }, { "epoch": 0.0108544, "grad_norm": 0.832708477973938, "learning_rate": 8.60526978796334e-06, "loss": 2.9725, "step": 212940 }, { "epoch": 0.01088, "grad_norm": 0.7799808979034424, "learning_rate": 8.604472384906056e-06, "loss": 2.9252, "step": 212950 }, { "epoch": 0.0109056, "grad_norm": 0.7014152407646179, "learning_rate": 8.603674990898399e-06, "loss": 2.7535, "step": 212960 }, { "epoch": 0.0109312, "grad_norm": 0.908877968788147, "learning_rate": 8.602877605945536e-06, "loss": 3.099, "step": 212970 }, { "epoch": 0.0109568, "grad_norm": 0.7416722774505615, "learning_rate": 8.602080230052641e-06, "loss": 2.9195, "step": 212980 }, { "epoch": 0.0109824, "grad_norm": 0.7785053849220276, "learning_rate": 8.601282863224886e-06, "loss": 2.9857, "step": 212990 }, { "epoch": 0.011008, "grad_norm": 0.9108628034591675, "learning_rate": 8.600485505467439e-06, "loss": 3.0135, "step": 213000 }, { "epoch": 0.0110336, "grad_norm": 0.7596656084060669, "learning_rate": 8.599688156785472e-06, "loss": 2.6773, "step": 213010 }, { "epoch": 0.0110592, "grad_norm": 1.034067988395691, "learning_rate": 8.598890817184153e-06, "loss": 3.1803, "step": 213020 }, { "epoch": 0.0110848, "grad_norm": 0.8089556694030762, "learning_rate": 8.598093486668659e-06, "loss": 2.8391, "step": 213030 }, { "epoch": 0.0111104, "grad_norm": 0.8412977457046509, "learning_rate": 8.597296165244152e-06, "loss": 2.833, "step": 213040 }, { "epoch": 0.011136, "grad_norm": 0.9397544860839844, "learning_rate": 8.596498852915807e-06, "loss": 2.9076, "step": 213050 }, { "epoch": 0.0111616, "grad_norm": 0.7371543049812317, "learning_rate": 8.595701549688793e-06, "loss": 2.7563, "step": 213060 }, { "epoch": 0.0111872, "grad_norm": 0.9803540706634521, "learning_rate": 8.594904255568282e-06, "loss": 3.0014, "step": 213070 }, { "epoch": 0.0112128, "grad_norm": 1.1120166778564453, "learning_rate": 8.594106970559444e-06, "loss": 3.2021, "step": 213080 }, { "epoch": 0.0112384, "grad_norm": 0.7367624640464783, "learning_rate": 8.59330969466745e-06, "loss": 3.0848, "step": 213090 }, { "epoch": 0.011264, "grad_norm": 0.7640525102615356, "learning_rate": 8.592512427897468e-06, "loss": 3.0242, "step": 213100 }, { "epoch": 0.0112896, "grad_norm": 0.7272746562957764, "learning_rate": 8.591715170254668e-06, "loss": 2.8636, "step": 213110 }, { "epoch": 0.0113152, "grad_norm": 0.7128954529762268, "learning_rate": 8.590917921744224e-06, "loss": 2.9258, "step": 213120 }, { "epoch": 0.0113408, "grad_norm": 0.9514352083206177, "learning_rate": 8.5901206823713e-06, "loss": 3.1602, "step": 213130 }, { "epoch": 0.0113664, "grad_norm": 0.8950555324554443, "learning_rate": 8.589323452141068e-06, "loss": 2.8142, "step": 213140 }, { "epoch": 0.011392, "grad_norm": 0.7764764428138733, "learning_rate": 8.5885262310587e-06, "loss": 2.8136, "step": 213150 }, { "epoch": 0.0114176, "grad_norm": 0.7548501491546631, "learning_rate": 8.587729019129365e-06, "loss": 2.8744, "step": 213160 }, { "epoch": 0.0114432, "grad_norm": 1.2277858257293701, "learning_rate": 8.58693181635823e-06, "loss": 2.833, "step": 213170 }, { "epoch": 0.0114688, "grad_norm": 0.8269773721694946, "learning_rate": 8.586134622750466e-06, "loss": 2.9784, "step": 213180 }, { "epoch": 0.0114944, "grad_norm": 0.8233281970024109, "learning_rate": 8.585337438311246e-06, "loss": 3.0262, "step": 213190 }, { "epoch": 0.01152, "grad_norm": 0.7739484906196594, "learning_rate": 8.584540263045734e-06, "loss": 3.0026, "step": 213200 }, { "epoch": 0.0115456, "grad_norm": 0.9144868850708008, "learning_rate": 8.583743096959106e-06, "loss": 3.0358, "step": 213210 }, { "epoch": 0.0115712, "grad_norm": 0.904945433139801, "learning_rate": 8.582945940056525e-06, "loss": 2.8418, "step": 213220 }, { "epoch": 0.0115968, "grad_norm": 0.9041976928710938, "learning_rate": 8.582148792343159e-06, "loss": 2.889, "step": 213230 }, { "epoch": 0.0116224, "grad_norm": 0.8689512014389038, "learning_rate": 8.581351653824185e-06, "loss": 2.8568, "step": 213240 }, { "epoch": 0.011648, "grad_norm": 0.8282896876335144, "learning_rate": 8.580554524504771e-06, "loss": 2.9397, "step": 213250 }, { "epoch": 0.0116736, "grad_norm": 0.749634861946106, "learning_rate": 8.579757404390081e-06, "loss": 2.8127, "step": 213260 }, { "epoch": 0.0116992, "grad_norm": 0.8786984086036682, "learning_rate": 8.578960293485287e-06, "loss": 2.8625, "step": 213270 }, { "epoch": 0.0117248, "grad_norm": 0.8550015687942505, "learning_rate": 8.578163191795558e-06, "loss": 2.9413, "step": 213280 }, { "epoch": 0.0117504, "grad_norm": 0.8983431458473206, "learning_rate": 8.577366099326064e-06, "loss": 2.7817, "step": 213290 }, { "epoch": 0.011776, "grad_norm": 0.767422080039978, "learning_rate": 8.576569016081971e-06, "loss": 2.94, "step": 213300 }, { "epoch": 0.0118016, "grad_norm": 0.8002700209617615, "learning_rate": 8.57577194206845e-06, "loss": 2.7394, "step": 213310 }, { "epoch": 0.0118272, "grad_norm": 0.7511500120162964, "learning_rate": 8.574974877290672e-06, "loss": 2.8192, "step": 213320 }, { "epoch": 0.0118528, "grad_norm": 0.7969254851341248, "learning_rate": 8.5741778217538e-06, "loss": 3.1737, "step": 213330 }, { "epoch": 0.0118784, "grad_norm": 1.0962415933609009, "learning_rate": 8.573380775463007e-06, "loss": 3.0191, "step": 213340 }, { "epoch": 0.011904, "grad_norm": 0.7294135689735413, "learning_rate": 8.572583738423459e-06, "loss": 2.6897, "step": 213350 }, { "epoch": 0.0119296, "grad_norm": 0.676345705986023, "learning_rate": 8.571786710640329e-06, "loss": 2.8713, "step": 213360 }, { "epoch": 0.0119552, "grad_norm": 0.9721921682357788, "learning_rate": 8.570989692118781e-06, "loss": 3.0911, "step": 213370 }, { "epoch": 0.0119808, "grad_norm": 0.8784854412078857, "learning_rate": 8.570192682863986e-06, "loss": 2.8084, "step": 213380 }, { "epoch": 0.0120064, "grad_norm": 0.7184792757034302, "learning_rate": 8.56939568288111e-06, "loss": 3.0118, "step": 213390 }, { "epoch": 0.012032, "grad_norm": 0.7482955455780029, "learning_rate": 8.56859869217532e-06, "loss": 2.9233, "step": 213400 }, { "epoch": 0.0120576, "grad_norm": 0.7423181533813477, "learning_rate": 8.56780171075179e-06, "loss": 2.6545, "step": 213410 }, { "epoch": 0.0120832, "grad_norm": 0.7866873741149902, "learning_rate": 8.567004738615687e-06, "loss": 2.8118, "step": 213420 }, { "epoch": 0.0121088, "grad_norm": 0.7795780897140503, "learning_rate": 8.566207775772176e-06, "loss": 2.9474, "step": 213430 }, { "epoch": 0.0121344, "grad_norm": 0.7571890950202942, "learning_rate": 8.565410822226427e-06, "loss": 3.2113, "step": 213440 }, { "epoch": 0.01216, "grad_norm": 0.9888190031051636, "learning_rate": 8.564613877983609e-06, "loss": 3.0865, "step": 213450 }, { "epoch": 0.0121856, "grad_norm": 0.7976964116096497, "learning_rate": 8.563816943048887e-06, "loss": 2.8996, "step": 213460 }, { "epoch": 0.0122112, "grad_norm": 0.7054387927055359, "learning_rate": 8.563020017427427e-06, "loss": 2.9448, "step": 213470 }, { "epoch": 0.0122368, "grad_norm": 0.9425429701805115, "learning_rate": 8.562223101124404e-06, "loss": 3.0234, "step": 213480 }, { "epoch": 0.0122624, "grad_norm": 1.3072373867034912, "learning_rate": 8.561426194144983e-06, "loss": 2.8245, "step": 213490 }, { "epoch": 0.012288, "grad_norm": 0.7943924069404602, "learning_rate": 8.560629296494329e-06, "loss": 2.9352, "step": 213500 }, { "epoch": 0.0123136, "grad_norm": 0.8327152729034424, "learning_rate": 8.559832408177611e-06, "loss": 2.9729, "step": 213510 }, { "epoch": 0.0123392, "grad_norm": 0.7815876007080078, "learning_rate": 8.5590355292e-06, "loss": 2.882, "step": 213520 }, { "epoch": 0.0123648, "grad_norm": 0.7913270592689514, "learning_rate": 8.55823865956666e-06, "loss": 3.1832, "step": 213530 }, { "epoch": 0.0123904, "grad_norm": 0.7444201111793518, "learning_rate": 8.557441799282755e-06, "loss": 2.9071, "step": 213540 }, { "epoch": 0.012416, "grad_norm": 0.8925700783729553, "learning_rate": 8.55664494835346e-06, "loss": 3.4155, "step": 213550 }, { "epoch": 0.0124416, "grad_norm": 0.9123128056526184, "learning_rate": 8.55584810678394e-06, "loss": 2.9251, "step": 213560 }, { "epoch": 0.0124672, "grad_norm": 1.0579100847244263, "learning_rate": 8.555051274579358e-06, "loss": 3.0459, "step": 213570 }, { "epoch": 0.0124928, "grad_norm": 0.7631088495254517, "learning_rate": 8.554254451744883e-06, "loss": 2.7316, "step": 213580 }, { "epoch": 0.0125184, "grad_norm": 0.7717832922935486, "learning_rate": 8.553457638285687e-06, "loss": 2.8244, "step": 213590 }, { "epoch": 0.012544, "grad_norm": 0.7934092283248901, "learning_rate": 8.552660834206931e-06, "loss": 2.9769, "step": 213600 }, { "epoch": 0.0125696, "grad_norm": 1.2459886074066162, "learning_rate": 8.551864039513787e-06, "loss": 2.853, "step": 213610 }, { "epoch": 0.0125952, "grad_norm": 0.8270199298858643, "learning_rate": 8.551067254211423e-06, "loss": 2.8054, "step": 213620 }, { "epoch": 0.0126208, "grad_norm": 0.7968934178352356, "learning_rate": 8.550270478304999e-06, "loss": 2.9459, "step": 213630 }, { "epoch": 0.0126464, "grad_norm": 0.9783633351325989, "learning_rate": 8.549473711799685e-06, "loss": 3.0048, "step": 213640 }, { "epoch": 0.012672, "grad_norm": 1.0047597885131836, "learning_rate": 8.548676954700651e-06, "loss": 3.0169, "step": 213650 }, { "epoch": 0.0126976, "grad_norm": 0.8487242460250854, "learning_rate": 8.54788020701306e-06, "loss": 3.0146, "step": 213660 }, { "epoch": 0.0127232, "grad_norm": 0.7718701362609863, "learning_rate": 8.54708346874208e-06, "loss": 3.0271, "step": 213670 }, { "epoch": 0.0127488, "grad_norm": 0.9360507130622864, "learning_rate": 8.546286739892878e-06, "loss": 3.4278, "step": 213680 }, { "epoch": 0.0127744, "grad_norm": 0.7517455220222473, "learning_rate": 8.54549002047062e-06, "loss": 3.2365, "step": 213690 }, { "epoch": 0.0128, "grad_norm": 0.9521962404251099, "learning_rate": 8.544693310480471e-06, "loss": 2.9326, "step": 213700 }, { "epoch": 0.0128256, "grad_norm": 0.7631519436836243, "learning_rate": 8.5438966099276e-06, "loss": 2.9847, "step": 213710 }, { "epoch": 0.0128512, "grad_norm": 0.8219574689865112, "learning_rate": 8.543099918817175e-06, "loss": 2.7408, "step": 213720 }, { "epoch": 0.0128768, "grad_norm": 0.8240377306938171, "learning_rate": 8.542303237154357e-06, "loss": 3.0104, "step": 213730 }, { "epoch": 0.0129024, "grad_norm": 0.8266006112098694, "learning_rate": 8.541506564944315e-06, "loss": 2.9804, "step": 213740 }, { "epoch": 0.012928, "grad_norm": 0.8063459396362305, "learning_rate": 8.540709902192215e-06, "loss": 2.9721, "step": 213750 }, { "epoch": 0.0129536, "grad_norm": 0.7873693704605103, "learning_rate": 8.539913248903225e-06, "loss": 2.9258, "step": 213760 }, { "epoch": 0.0129792, "grad_norm": 0.7113468647003174, "learning_rate": 8.539116605082507e-06, "loss": 2.9137, "step": 213770 }, { "epoch": 0.0130048, "grad_norm": 0.7980536222457886, "learning_rate": 8.538319970735234e-06, "loss": 2.8794, "step": 213780 }, { "epoch": 0.0130304, "grad_norm": 0.7891770005226135, "learning_rate": 8.537523345866565e-06, "loss": 2.9021, "step": 213790 }, { "epoch": 0.013056, "grad_norm": 0.8675798773765564, "learning_rate": 8.536726730481668e-06, "loss": 2.8477, "step": 213800 }, { "epoch": 0.0130816, "grad_norm": 0.676400363445282, "learning_rate": 8.53593012458571e-06, "loss": 2.8159, "step": 213810 }, { "epoch": 0.0131072, "grad_norm": 0.8156598806381226, "learning_rate": 8.535133528183856e-06, "loss": 2.8967, "step": 213820 }, { "epoch": 0.0131328, "grad_norm": 0.7568632364273071, "learning_rate": 8.534336941281272e-06, "loss": 2.7555, "step": 213830 }, { "epoch": 0.0131584, "grad_norm": 0.7554410696029663, "learning_rate": 8.533540363883123e-06, "loss": 2.9364, "step": 213840 }, { "epoch": 0.013184, "grad_norm": 0.8258329629898071, "learning_rate": 8.532743795994577e-06, "loss": 2.9337, "step": 213850 }, { "epoch": 0.0132096, "grad_norm": 0.7429772615432739, "learning_rate": 8.531947237620794e-06, "loss": 2.8833, "step": 213860 }, { "epoch": 0.0132352, "grad_norm": 0.7038571238517761, "learning_rate": 8.531150688766946e-06, "loss": 2.9188, "step": 213870 }, { "epoch": 0.0132608, "grad_norm": 0.7781675457954407, "learning_rate": 8.530354149438192e-06, "loss": 2.6504, "step": 213880 }, { "epoch": 0.0132864, "grad_norm": 0.8815502524375916, "learning_rate": 8.529557619639705e-06, "loss": 2.8786, "step": 213890 }, { "epoch": 0.013312, "grad_norm": 1.2628415822982788, "learning_rate": 8.528761099376643e-06, "loss": 2.6123, "step": 213900 }, { "epoch": 2.56e-05, "grad_norm": 0.7456233501434326, "learning_rate": 8.527964588654175e-06, "loss": 2.3565, "step": 213910 }, { "epoch": 5.12e-05, "grad_norm": 0.7933441400527954, "learning_rate": 8.527168087477468e-06, "loss": 2.7512, "step": 213920 }, { "epoch": 7.68e-05, "grad_norm": 0.9206775426864624, "learning_rate": 8.526371595851677e-06, "loss": 2.6892, "step": 213930 }, { "epoch": 0.0001024, "grad_norm": 0.7119245529174805, "learning_rate": 8.525575113781982e-06, "loss": 2.6507, "step": 213940 }, { "epoch": 0.000128, "grad_norm": 0.7869321703910828, "learning_rate": 8.524778641273537e-06, "loss": 2.8453, "step": 213950 }, { "epoch": 0.0001536, "grad_norm": 0.7800251245498657, "learning_rate": 8.523982178331512e-06, "loss": 3.0398, "step": 213960 }, { "epoch": 0.0001792, "grad_norm": 0.7366837859153748, "learning_rate": 8.523185724961068e-06, "loss": 2.7686, "step": 213970 }, { "epoch": 0.0002048, "grad_norm": 0.7908276915550232, "learning_rate": 8.522389281167376e-06, "loss": 2.6243, "step": 213980 }, { "epoch": 0.0002304, "grad_norm": 0.800573468208313, "learning_rate": 8.521592846955594e-06, "loss": 2.6029, "step": 213990 }, { "epoch": 0.000256, "grad_norm": 0.7816543579101562, "learning_rate": 8.52079642233089e-06, "loss": 2.7008, "step": 214000 }, { "epoch": 0.0002816, "grad_norm": 0.8317422270774841, "learning_rate": 8.520000007298429e-06, "loss": 2.6653, "step": 214010 }, { "epoch": 0.0003072, "grad_norm": 0.759190022945404, "learning_rate": 8.519203601863375e-06, "loss": 2.5691, "step": 214020 }, { "epoch": 0.0003328, "grad_norm": 0.7866836190223694, "learning_rate": 8.51840720603089e-06, "loss": 2.6719, "step": 214030 }, { "epoch": 0.0003584, "grad_norm": 0.8080089092254639, "learning_rate": 8.51761081980614e-06, "loss": 2.9242, "step": 214040 }, { "epoch": 0.000384, "grad_norm": 0.8190345764160156, "learning_rate": 8.516814443194295e-06, "loss": 2.843, "step": 214050 }, { "epoch": 0.0004096, "grad_norm": 0.8659216165542603, "learning_rate": 8.516018076200509e-06, "loss": 2.8539, "step": 214060 }, { "epoch": 0.0004352, "grad_norm": 0.7688527703285217, "learning_rate": 8.515221718829954e-06, "loss": 2.7373, "step": 214070 }, { "epoch": 0.0004608, "grad_norm": 0.7134596705436707, "learning_rate": 8.514425371087788e-06, "loss": 2.7559, "step": 214080 }, { "epoch": 0.0004864, "grad_norm": 0.890807032585144, "learning_rate": 8.513629032979184e-06, "loss": 2.8018, "step": 214090 }, { "epoch": 0.000512, "grad_norm": 0.6686593890190125, "learning_rate": 8.512832704509293e-06, "loss": 2.4272, "step": 214100 }, { "epoch": 0.0005376, "grad_norm": 0.8100127577781677, "learning_rate": 8.512036385683294e-06, "loss": 2.7688, "step": 214110 }, { "epoch": 0.0005632, "grad_norm": 0.8388323783874512, "learning_rate": 8.51124007650634e-06, "loss": 2.4792, "step": 214120 }, { "epoch": 0.0005888, "grad_norm": 0.8335398435592651, "learning_rate": 8.510443776983599e-06, "loss": 2.8548, "step": 214130 }, { "epoch": 0.0006144, "grad_norm": 0.7626023292541504, "learning_rate": 8.509647487120236e-06, "loss": 2.6186, "step": 214140 }, { "epoch": 0.00064, "grad_norm": 0.7795376777648926, "learning_rate": 8.508851206921411e-06, "loss": 2.681, "step": 214150 }, { "epoch": 0.0006656, "grad_norm": 0.8570836782455444, "learning_rate": 8.50805493639229e-06, "loss": 3.0934, "step": 214160 }, { "epoch": 0.0006912, "grad_norm": 0.7723334431648254, "learning_rate": 8.507258675538036e-06, "loss": 2.9094, "step": 214170 }, { "epoch": 0.0007168, "grad_norm": 0.7697914838790894, "learning_rate": 8.506462424363815e-06, "loss": 2.5423, "step": 214180 }, { "epoch": 0.0007424, "grad_norm": 0.8906450271606445, "learning_rate": 8.505666182874787e-06, "loss": 2.7111, "step": 214190 }, { "epoch": 0.000768, "grad_norm": 0.8387356400489807, "learning_rate": 8.504869951076116e-06, "loss": 2.7298, "step": 214200 }, { "epoch": 0.0007936, "grad_norm": 0.7241302132606506, "learning_rate": 8.504073728972966e-06, "loss": 2.8065, "step": 214210 }, { "epoch": 0.0008192, "grad_norm": 0.7915300726890564, "learning_rate": 8.503277516570501e-06, "loss": 2.8331, "step": 214220 }, { "epoch": 0.0008448, "grad_norm": 1.0433863401412964, "learning_rate": 8.502481313873883e-06, "loss": 2.6576, "step": 214230 }, { "epoch": 0.0008704, "grad_norm": 0.7380679845809937, "learning_rate": 8.501685120888274e-06, "loss": 2.5596, "step": 214240 }, { "epoch": 0.000896, "grad_norm": 0.8963937759399414, "learning_rate": 8.500888937618842e-06, "loss": 2.7036, "step": 214250 }, { "epoch": 0.0009216, "grad_norm": 0.8953298330307007, "learning_rate": 8.500092764070746e-06, "loss": 2.7858, "step": 214260 }, { "epoch": 0.0009472, "grad_norm": 0.8693639039993286, "learning_rate": 8.499296600249147e-06, "loss": 2.5999, "step": 214270 }, { "epoch": 0.0009728, "grad_norm": 0.7691906094551086, "learning_rate": 8.498500446159213e-06, "loss": 2.527, "step": 214280 }, { "epoch": 0.0009984, "grad_norm": 0.8787195086479187, "learning_rate": 8.497704301806104e-06, "loss": 2.7475, "step": 214290 }, { "epoch": 0.001024, "grad_norm": 0.8660697937011719, "learning_rate": 8.496908167194983e-06, "loss": 2.914, "step": 214300 }, { "epoch": 0.0010496, "grad_norm": 0.8963997960090637, "learning_rate": 8.496112042331015e-06, "loss": 2.6968, "step": 214310 }, { "epoch": 0.0010752, "grad_norm": 0.6908331513404846, "learning_rate": 8.495315927219358e-06, "loss": 2.712, "step": 214320 }, { "epoch": 0.0011008, "grad_norm": 0.9526765942573547, "learning_rate": 8.49451982186518e-06, "loss": 2.7261, "step": 214330 }, { "epoch": 0.0011264, "grad_norm": 0.7754822373390198, "learning_rate": 8.493723726273641e-06, "loss": 2.6421, "step": 214340 }, { "epoch": 0.001152, "grad_norm": 0.754737913608551, "learning_rate": 8.492927640449902e-06, "loss": 2.937, "step": 214350 }, { "epoch": 0.0011776, "grad_norm": 0.7753884196281433, "learning_rate": 8.492131564399125e-06, "loss": 2.9189, "step": 214360 }, { "epoch": 0.0012032, "grad_norm": 0.8414368033409119, "learning_rate": 8.491335498126477e-06, "loss": 2.8978, "step": 214370 }, { "epoch": 0.0012288, "grad_norm": 0.7731813788414001, "learning_rate": 8.490539441637117e-06, "loss": 2.8938, "step": 214380 }, { "epoch": 0.0012544, "grad_norm": 0.6972397565841675, "learning_rate": 8.489743394936207e-06, "loss": 2.5886, "step": 214390 }, { "epoch": 0.00128, "grad_norm": 1.0014628171920776, "learning_rate": 8.488947358028907e-06, "loss": 2.9095, "step": 214400 }, { "epoch": 0.0013056, "grad_norm": 0.7919074892997742, "learning_rate": 8.488151330920387e-06, "loss": 2.8353, "step": 214410 }, { "epoch": 0.0013312, "grad_norm": 0.7727381587028503, "learning_rate": 8.4873553136158e-06, "loss": 2.4738, "step": 214420 }, { "epoch": 0.0013568, "grad_norm": 0.8860692381858826, "learning_rate": 8.486559306120311e-06, "loss": 2.7738, "step": 214430 }, { "epoch": 0.0013824, "grad_norm": 0.858277440071106, "learning_rate": 8.485763308439085e-06, "loss": 2.7011, "step": 214440 }, { "epoch": 0.001408, "grad_norm": 0.8247984051704407, "learning_rate": 8.484967320577278e-06, "loss": 2.6819, "step": 214450 }, { "epoch": 0.0014336, "grad_norm": 0.8307798504829407, "learning_rate": 8.484171342540058e-06, "loss": 2.6913, "step": 214460 }, { "epoch": 0.0014592, "grad_norm": 0.8659791350364685, "learning_rate": 8.483375374332586e-06, "loss": 2.9, "step": 214470 }, { "epoch": 0.0014848, "grad_norm": 0.8292589783668518, "learning_rate": 8.48257941596002e-06, "loss": 2.9192, "step": 214480 }, { "epoch": 0.0015104, "grad_norm": 0.8328129053115845, "learning_rate": 8.481783467427523e-06, "loss": 2.3561, "step": 214490 }, { "epoch": 0.001536, "grad_norm": 0.8240500092506409, "learning_rate": 8.480987528740256e-06, "loss": 2.7068, "step": 214500 }, { "epoch": 0.0015616, "grad_norm": 0.8261181712150574, "learning_rate": 8.480191599903383e-06, "loss": 2.7165, "step": 214510 }, { "epoch": 0.0015872, "grad_norm": 0.89216548204422, "learning_rate": 8.479395680922062e-06, "loss": 2.7339, "step": 214520 }, { "epoch": 0.0016128, "grad_norm": 0.8150882124900818, "learning_rate": 8.478599771801457e-06, "loss": 2.9211, "step": 214530 }, { "epoch": 0.0016384, "grad_norm": 0.857133150100708, "learning_rate": 8.477803872546728e-06, "loss": 2.7814, "step": 214540 }, { "epoch": 0.001664, "grad_norm": 0.7747033834457397, "learning_rate": 8.477007983163035e-06, "loss": 2.7648, "step": 214550 }, { "epoch": 0.0016896, "grad_norm": 0.9812487959861755, "learning_rate": 8.476212103655541e-06, "loss": 2.8224, "step": 214560 }, { "epoch": 0.0017152, "grad_norm": 0.8173041343688965, "learning_rate": 8.475416234029407e-06, "loss": 2.6053, "step": 214570 }, { "epoch": 0.0017408, "grad_norm": 0.7788585424423218, "learning_rate": 8.474620374289795e-06, "loss": 2.8507, "step": 214580 }, { "epoch": 0.0017664, "grad_norm": 0.7639211416244507, "learning_rate": 8.473824524441863e-06, "loss": 2.8751, "step": 214590 }, { "epoch": 0.001792, "grad_norm": 0.8590350151062012, "learning_rate": 8.473028684490772e-06, "loss": 2.6749, "step": 214600 }, { "epoch": 0.0018176, "grad_norm": 0.7586527466773987, "learning_rate": 8.472232854441688e-06, "loss": 2.8337, "step": 214610 }, { "epoch": 0.0018432, "grad_norm": 0.7398071885108948, "learning_rate": 8.471437034299762e-06, "loss": 2.5717, "step": 214620 }, { "epoch": 0.0018688, "grad_norm": 0.7911516427993774, "learning_rate": 8.470641224070164e-06, "loss": 2.9204, "step": 214630 }, { "epoch": 0.0018944, "grad_norm": 0.7701271772384644, "learning_rate": 8.46984542375805e-06, "loss": 2.917, "step": 214640 }, { "epoch": 0.00192, "grad_norm": 1.2921850681304932, "learning_rate": 8.469049633368584e-06, "loss": 2.8157, "step": 214650 }, { "epoch": 0.0019456, "grad_norm": 0.753717839717865, "learning_rate": 8.468253852906922e-06, "loss": 2.797, "step": 214660 }, { "epoch": 0.0019712, "grad_norm": 1.0107256174087524, "learning_rate": 8.46745808237823e-06, "loss": 2.8389, "step": 214670 }, { "epoch": 0.0019968, "grad_norm": 0.8930476307868958, "learning_rate": 8.466662321787662e-06, "loss": 2.9894, "step": 214680 }, { "epoch": 0.0020224, "grad_norm": 0.7594685554504395, "learning_rate": 8.465866571140381e-06, "loss": 2.6009, "step": 214690 }, { "epoch": 0.002048, "grad_norm": 0.813424825668335, "learning_rate": 8.465070830441548e-06, "loss": 2.6109, "step": 214700 }, { "epoch": 0.0020736, "grad_norm": 0.8803795576095581, "learning_rate": 8.464275099696325e-06, "loss": 2.4857, "step": 214710 }, { "epoch": 0.0020992, "grad_norm": 0.7622498869895935, "learning_rate": 8.463479378909869e-06, "loss": 2.7423, "step": 214720 }, { "epoch": 0.0021248, "grad_norm": 0.7539752721786499, "learning_rate": 8.46268366808734e-06, "loss": 2.7524, "step": 214730 }, { "epoch": 0.0021504, "grad_norm": 0.7957976460456848, "learning_rate": 8.4618879672339e-06, "loss": 2.8279, "step": 214740 }, { "epoch": 0.002176, "grad_norm": 0.7577748894691467, "learning_rate": 8.461092276354705e-06, "loss": 2.8793, "step": 214750 }, { "epoch": 0.0022016, "grad_norm": 0.8124725222587585, "learning_rate": 8.460296595454919e-06, "loss": 2.9336, "step": 214760 }, { "epoch": 0.0022272, "grad_norm": 0.7687097787857056, "learning_rate": 8.4595009245397e-06, "loss": 2.9608, "step": 214770 }, { "epoch": 0.0022528, "grad_norm": 0.7867189645767212, "learning_rate": 8.458705263614212e-06, "loss": 2.5506, "step": 214780 }, { "epoch": 0.0022784, "grad_norm": 1.2288473844528198, "learning_rate": 8.457909612683607e-06, "loss": 2.7544, "step": 214790 }, { "epoch": 0.002304, "grad_norm": 0.7257153987884521, "learning_rate": 8.457113971753046e-06, "loss": 2.8778, "step": 214800 }, { "epoch": 0.0023296, "grad_norm": 0.7706202864646912, "learning_rate": 8.456318340827693e-06, "loss": 2.6035, "step": 214810 }, { "epoch": 0.0023552, "grad_norm": 0.7640217542648315, "learning_rate": 8.455522719912705e-06, "loss": 2.6886, "step": 214820 }, { "epoch": 0.0023808, "grad_norm": 0.7647495865821838, "learning_rate": 8.454727109013245e-06, "loss": 2.5125, "step": 214830 }, { "epoch": 0.0024064, "grad_norm": 0.7850562930107117, "learning_rate": 8.453931508134466e-06, "loss": 2.6939, "step": 214840 }, { "epoch": 0.002432, "grad_norm": 0.8145049810409546, "learning_rate": 8.453135917281531e-06, "loss": 2.9829, "step": 214850 }, { "epoch": 0.0024576, "grad_norm": 0.7379438877105713, "learning_rate": 8.452340336459598e-06, "loss": 2.6979, "step": 214860 }, { "epoch": 0.0024832, "grad_norm": 0.7440460324287415, "learning_rate": 8.451544765673829e-06, "loss": 2.8884, "step": 214870 }, { "epoch": 0.0025088, "grad_norm": 0.9654611945152283, "learning_rate": 8.450749204929379e-06, "loss": 2.9986, "step": 214880 }, { "epoch": 0.0025344, "grad_norm": 0.7517687082290649, "learning_rate": 8.449953654231406e-06, "loss": 2.7175, "step": 214890 }, { "epoch": 0.00256, "grad_norm": 0.7990628480911255, "learning_rate": 8.449158113585074e-06, "loss": 3.018, "step": 214900 }, { "epoch": 0.0025856, "grad_norm": 0.8421434760093689, "learning_rate": 8.448362582995542e-06, "loss": 2.8245, "step": 214910 }, { "epoch": 0.0026112, "grad_norm": 0.8120920658111572, "learning_rate": 8.447567062467963e-06, "loss": 2.9664, "step": 214920 }, { "epoch": 0.0026368, "grad_norm": 0.8518388867378235, "learning_rate": 8.446771552007498e-06, "loss": 2.8952, "step": 214930 }, { "epoch": 0.0026624, "grad_norm": 1.2673510313034058, "learning_rate": 8.44597605161931e-06, "loss": 2.9927, "step": 214940 }, { "epoch": 0.002688, "grad_norm": 0.7682651877403259, "learning_rate": 8.445180561308553e-06, "loss": 3.0098, "step": 214950 }, { "epoch": 0.0027136, "grad_norm": 0.8585212230682373, "learning_rate": 8.444385081080386e-06, "loss": 2.6011, "step": 214960 }, { "epoch": 0.0027392, "grad_norm": 0.7269401550292969, "learning_rate": 8.443589610939967e-06, "loss": 2.8862, "step": 214970 }, { "epoch": 0.0027648, "grad_norm": 0.7720443606376648, "learning_rate": 8.442794150892457e-06, "loss": 2.8729, "step": 214980 }, { "epoch": 0.0027904, "grad_norm": 0.8541314005851746, "learning_rate": 8.441998700943013e-06, "loss": 2.8592, "step": 214990 }, { "epoch": 0.002816, "grad_norm": 1.2194766998291016, "learning_rate": 8.441203261096796e-06, "loss": 3.0063, "step": 215000 }, { "epoch": 0.0028416, "grad_norm": 0.7574167251586914, "learning_rate": 8.440407831358959e-06, "loss": 2.8967, "step": 215010 }, { "epoch": 0.0028672, "grad_norm": 0.8317840099334717, "learning_rate": 8.439612411734663e-06, "loss": 2.8185, "step": 215020 }, { "epoch": 0.0028928, "grad_norm": 1.0108853578567505, "learning_rate": 8.438817002229067e-06, "loss": 2.7297, "step": 215030 }, { "epoch": 0.0029184, "grad_norm": 0.7542571425437927, "learning_rate": 8.438021602847327e-06, "loss": 2.801, "step": 215040 }, { "epoch": 0.002944, "grad_norm": 0.7841405272483826, "learning_rate": 8.437226213594601e-06, "loss": 2.6188, "step": 215050 }, { "epoch": 0.0029696, "grad_norm": 0.7374236583709717, "learning_rate": 8.436430834476048e-06, "loss": 2.9839, "step": 215060 }, { "epoch": 0.0029952, "grad_norm": 0.737595796585083, "learning_rate": 8.435635465496828e-06, "loss": 2.9861, "step": 215070 }, { "epoch": 0.0030208, "grad_norm": 0.8259321451187134, "learning_rate": 8.434840106662094e-06, "loss": 2.8076, "step": 215080 }, { "epoch": 0.0030464, "grad_norm": 0.770903468132019, "learning_rate": 8.434044757977004e-06, "loss": 2.817, "step": 215090 }, { "epoch": 0.003072, "grad_norm": 0.869245707988739, "learning_rate": 8.433249419446722e-06, "loss": 2.8124, "step": 215100 }, { "epoch": 0.0030976, "grad_norm": 0.7890912890434265, "learning_rate": 8.432454091076398e-06, "loss": 2.8683, "step": 215110 }, { "epoch": 0.0031232, "grad_norm": 0.740443229675293, "learning_rate": 8.431658772871194e-06, "loss": 2.6205, "step": 215120 }, { "epoch": 0.0031488, "grad_norm": 0.8788890838623047, "learning_rate": 8.430863464836263e-06, "loss": 2.859, "step": 215130 }, { "epoch": 0.0031744, "grad_norm": 0.7427490949630737, "learning_rate": 8.430068166976767e-06, "loss": 3.1616, "step": 215140 }, { "epoch": 0.0032, "grad_norm": 0.7886714339256287, "learning_rate": 8.429272879297861e-06, "loss": 2.8005, "step": 215150 }, { "epoch": 0.0032256, "grad_norm": 0.7570083141326904, "learning_rate": 8.428477601804706e-06, "loss": 2.7966, "step": 215160 }, { "epoch": 0.0032512, "grad_norm": 0.7404485940933228, "learning_rate": 8.427682334502455e-06, "loss": 2.9974, "step": 215170 }, { "epoch": 0.0032768, "grad_norm": 0.7481997013092041, "learning_rate": 8.426887077396265e-06, "loss": 2.9385, "step": 215180 }, { "epoch": 0.0033024, "grad_norm": 0.8223326206207275, "learning_rate": 8.426091830491294e-06, "loss": 2.9399, "step": 215190 }, { "epoch": 0.003328, "grad_norm": 0.7860541939735413, "learning_rate": 8.425296593792703e-06, "loss": 2.898, "step": 215200 }, { "epoch": 0.0033536, "grad_norm": 0.7652743458747864, "learning_rate": 8.424501367305642e-06, "loss": 2.8815, "step": 215210 }, { "epoch": 0.0033792, "grad_norm": 0.8615493178367615, "learning_rate": 8.42370615103527e-06, "loss": 2.7206, "step": 215220 }, { "epoch": 0.0034048, "grad_norm": 0.8849385976791382, "learning_rate": 8.422910944986747e-06, "loss": 2.7188, "step": 215230 }, { "epoch": 0.0034304, "grad_norm": 0.8153460621833801, "learning_rate": 8.422115749165225e-06, "loss": 3.1022, "step": 215240 }, { "epoch": 0.003456, "grad_norm": 0.8475494384765625, "learning_rate": 8.421320563575865e-06, "loss": 2.5924, "step": 215250 }, { "epoch": 0.0034816, "grad_norm": 0.7459132671356201, "learning_rate": 8.42052538822382e-06, "loss": 2.8818, "step": 215260 }, { "epoch": 0.0035072, "grad_norm": 0.721982479095459, "learning_rate": 8.41973022311425e-06, "loss": 2.7787, "step": 215270 }, { "epoch": 0.0035328, "grad_norm": 0.7270991206169128, "learning_rate": 8.41893506825231e-06, "loss": 2.6748, "step": 215280 }, { "epoch": 0.0035584, "grad_norm": 0.7334159016609192, "learning_rate": 8.418139923643152e-06, "loss": 2.9197, "step": 215290 }, { "epoch": 0.003584, "grad_norm": 0.7539730668067932, "learning_rate": 8.417344789291941e-06, "loss": 2.9228, "step": 215300 }, { "epoch": 0.0036096, "grad_norm": 1.1032233238220215, "learning_rate": 8.416549665203826e-06, "loss": 3.1488, "step": 215310 }, { "epoch": 0.0036352, "grad_norm": 0.7522763609886169, "learning_rate": 8.415754551383962e-06, "loss": 2.7991, "step": 215320 }, { "epoch": 0.0036608, "grad_norm": 0.7833921909332275, "learning_rate": 8.414959447837511e-06, "loss": 3.0024, "step": 215330 }, { "epoch": 0.0036864, "grad_norm": 0.8235538005828857, "learning_rate": 8.414164354569628e-06, "loss": 2.9365, "step": 215340 }, { "epoch": 0.003712, "grad_norm": 0.7851873636245728, "learning_rate": 8.413369271585468e-06, "loss": 2.9317, "step": 215350 }, { "epoch": 0.0037376, "grad_norm": 0.7392165660858154, "learning_rate": 8.412574198890187e-06, "loss": 3.0111, "step": 215360 }, { "epoch": 0.0037632, "grad_norm": 0.8016875386238098, "learning_rate": 8.41177913648894e-06, "loss": 2.8181, "step": 215370 }, { "epoch": 0.0037888, "grad_norm": 0.8334395289421082, "learning_rate": 8.410984084386881e-06, "loss": 3.0415, "step": 215380 }, { "epoch": 0.0038144, "grad_norm": 0.799584150314331, "learning_rate": 8.41018904258917e-06, "loss": 2.9974, "step": 215390 }, { "epoch": 0.00384, "grad_norm": 0.8782963156700134, "learning_rate": 8.409394011100963e-06, "loss": 2.9379, "step": 215400 }, { "epoch": 0.0038656, "grad_norm": 0.9749144315719604, "learning_rate": 8.40859898992741e-06, "loss": 2.7256, "step": 215410 }, { "epoch": 0.0038912, "grad_norm": 0.7924063205718994, "learning_rate": 8.407803979073668e-06, "loss": 2.9483, "step": 215420 }, { "epoch": 0.0039168, "grad_norm": 0.8879562616348267, "learning_rate": 8.407008978544898e-06, "loss": 2.8086, "step": 215430 }, { "epoch": 0.0039424, "grad_norm": 0.8999969959259033, "learning_rate": 8.40621398834625e-06, "loss": 2.7025, "step": 215440 }, { "epoch": 0.003968, "grad_norm": 0.7967795729637146, "learning_rate": 8.405419008482879e-06, "loss": 2.9422, "step": 215450 }, { "epoch": 0.0039936, "grad_norm": 0.8255517482757568, "learning_rate": 8.404624038959943e-06, "loss": 2.756, "step": 215460 }, { "epoch": 0.0040192, "grad_norm": 0.7693129181861877, "learning_rate": 8.403829079782597e-06, "loss": 2.8617, "step": 215470 }, { "epoch": 0.0040448, "grad_norm": 0.8189471960067749, "learning_rate": 8.403034130955995e-06, "loss": 2.8791, "step": 215480 }, { "epoch": 0.0040704, "grad_norm": 0.7549679279327393, "learning_rate": 8.402239192485289e-06, "loss": 3.0623, "step": 215490 }, { "epoch": 0.004096, "grad_norm": 0.7148168087005615, "learning_rate": 8.401444264375638e-06, "loss": 2.7198, "step": 215500 }, { "epoch": 0.0041216, "grad_norm": 0.8143709897994995, "learning_rate": 8.400649346632199e-06, "loss": 3.1172, "step": 215510 }, { "epoch": 0.0041472, "grad_norm": 0.7949382066726685, "learning_rate": 8.399854439260124e-06, "loss": 2.7054, "step": 215520 }, { "epoch": 0.0041728, "grad_norm": 0.7841269373893738, "learning_rate": 8.399059542264565e-06, "loss": 2.8242, "step": 215530 }, { "epoch": 0.0041984, "grad_norm": 0.774340808391571, "learning_rate": 8.398264655650681e-06, "loss": 2.9422, "step": 215540 }, { "epoch": 0.004224, "grad_norm": 0.7507271766662598, "learning_rate": 8.397469779423625e-06, "loss": 3.0001, "step": 215550 }, { "epoch": 0.0042496, "grad_norm": 0.8108255863189697, "learning_rate": 8.396674913588554e-06, "loss": 2.9345, "step": 215560 }, { "epoch": 0.0042752, "grad_norm": 0.9445663690567017, "learning_rate": 8.395880058150617e-06, "loss": 2.7224, "step": 215570 }, { "epoch": 0.0043008, "grad_norm": 0.8749269843101501, "learning_rate": 8.395085213114971e-06, "loss": 2.7679, "step": 215580 }, { "epoch": 0.0043264, "grad_norm": 0.7794943451881409, "learning_rate": 8.394290378486772e-06, "loss": 2.9626, "step": 215590 }, { "epoch": 0.004352, "grad_norm": 0.877456784248352, "learning_rate": 8.393495554271174e-06, "loss": 2.9336, "step": 215600 }, { "epoch": 0.0043776, "grad_norm": 0.8272773623466492, "learning_rate": 8.39270074047333e-06, "loss": 2.8294, "step": 215610 }, { "epoch": 0.0044032, "grad_norm": 0.8342248797416687, "learning_rate": 8.391905937098393e-06, "loss": 3.0028, "step": 215620 }, { "epoch": 0.0044288, "grad_norm": 0.981437623500824, "learning_rate": 8.391111144151521e-06, "loss": 3.0126, "step": 215630 }, { "epoch": 0.0044544, "grad_norm": 0.7688474655151367, "learning_rate": 8.390316361637864e-06, "loss": 2.8983, "step": 215640 }, { "epoch": 0.00448, "grad_norm": 0.8567092418670654, "learning_rate": 8.389521589562577e-06, "loss": 3.0254, "step": 215650 }, { "epoch": 0.0045056, "grad_norm": 0.8012016415596008, "learning_rate": 8.388726827930815e-06, "loss": 2.9371, "step": 215660 }, { "epoch": 0.0045312, "grad_norm": 0.7649691104888916, "learning_rate": 8.38793207674773e-06, "loss": 2.8746, "step": 215670 }, { "epoch": 0.0045568, "grad_norm": 0.7768503427505493, "learning_rate": 8.387137336018478e-06, "loss": 2.7669, "step": 215680 }, { "epoch": 0.0045824, "grad_norm": 0.7545779943466187, "learning_rate": 8.386342605748214e-06, "loss": 2.9249, "step": 215690 }, { "epoch": 0.004608, "grad_norm": 0.9008504152297974, "learning_rate": 8.385547885942087e-06, "loss": 2.9396, "step": 215700 }, { "epoch": 0.0046336, "grad_norm": 0.809136688709259, "learning_rate": 8.384753176605254e-06, "loss": 2.9911, "step": 215710 }, { "epoch": 0.0046592, "grad_norm": 0.8383516669273376, "learning_rate": 8.383958477742869e-06, "loss": 3.0586, "step": 215720 }, { "epoch": 0.0046848, "grad_norm": 0.812902569770813, "learning_rate": 8.38316378936008e-06, "loss": 3.0422, "step": 215730 }, { "epoch": 0.0047104, "grad_norm": 0.9132821559906006, "learning_rate": 8.382369111462046e-06, "loss": 3.0558, "step": 215740 }, { "epoch": 0.004736, "grad_norm": 0.8088604211807251, "learning_rate": 8.381574444053917e-06, "loss": 2.7747, "step": 215750 }, { "epoch": 0.0047616, "grad_norm": 0.8461294770240784, "learning_rate": 8.380779787140852e-06, "loss": 2.6756, "step": 215760 }, { "epoch": 0.0047872, "grad_norm": 0.770899772644043, "learning_rate": 8.379985140727995e-06, "loss": 2.7788, "step": 215770 }, { "epoch": 0.0048128, "grad_norm": 0.7616757154464722, "learning_rate": 8.379190504820505e-06, "loss": 2.643, "step": 215780 }, { "epoch": 0.0048384, "grad_norm": 0.7944172620773315, "learning_rate": 8.378395879423535e-06, "loss": 2.9242, "step": 215790 }, { "epoch": 0.004864, "grad_norm": 0.713240385055542, "learning_rate": 8.377601264542237e-06, "loss": 2.814, "step": 215800 }, { "epoch": 0.0048896, "grad_norm": 0.6973483562469482, "learning_rate": 8.376806660181764e-06, "loss": 2.7781, "step": 215810 }, { "epoch": 0.0049152, "grad_norm": 0.8344720005989075, "learning_rate": 8.376012066347266e-06, "loss": 2.6912, "step": 215820 }, { "epoch": 0.0049408, "grad_norm": 0.8285285830497742, "learning_rate": 8.3752174830439e-06, "loss": 2.8146, "step": 215830 }, { "epoch": 0.0049664, "grad_norm": 0.7388558387756348, "learning_rate": 8.374422910276812e-06, "loss": 2.7571, "step": 215840 }, { "epoch": 0.004992, "grad_norm": 0.8512907028198242, "learning_rate": 8.373628348051165e-06, "loss": 2.8563, "step": 215850 }, { "epoch": 0.0050176, "grad_norm": 0.7468264698982239, "learning_rate": 8.372833796372104e-06, "loss": 2.981, "step": 215860 }, { "epoch": 0.0050432, "grad_norm": 0.8511261940002441, "learning_rate": 8.372039255244783e-06, "loss": 2.8916, "step": 215870 }, { "epoch": 0.0050688, "grad_norm": 0.8122633695602417, "learning_rate": 8.371244724674355e-06, "loss": 2.7076, "step": 215880 }, { "epoch": 0.0050944, "grad_norm": 1.5199908018112183, "learning_rate": 8.370450204665975e-06, "loss": 2.8151, "step": 215890 }, { "epoch": 0.00512, "grad_norm": 0.8881699442863464, "learning_rate": 8.36965569522479e-06, "loss": 2.7572, "step": 215900 }, { "epoch": 0.0051456, "grad_norm": 0.7765741348266602, "learning_rate": 8.368861196355953e-06, "loss": 2.8928, "step": 215910 }, { "epoch": 0.0051712, "grad_norm": 0.7757739424705505, "learning_rate": 8.368066708064621e-06, "loss": 2.9106, "step": 215920 }, { "epoch": 0.0051968, "grad_norm": 0.7347384691238403, "learning_rate": 8.36727223035594e-06, "loss": 2.8944, "step": 215930 }, { "epoch": 0.0052224, "grad_norm": 0.8002515435218811, "learning_rate": 8.366477763235065e-06, "loss": 2.5985, "step": 215940 }, { "epoch": 0.005248, "grad_norm": 0.7295383810997009, "learning_rate": 8.365683306707148e-06, "loss": 2.8696, "step": 215950 }, { "epoch": 0.0052736, "grad_norm": 0.7949422597885132, "learning_rate": 8.364888860777342e-06, "loss": 2.9459, "step": 215960 }, { "epoch": 0.0052992, "grad_norm": 0.8537551760673523, "learning_rate": 8.364094425450796e-06, "loss": 2.8893, "step": 215970 }, { "epoch": 0.0053248, "grad_norm": 0.9238329529762268, "learning_rate": 8.363300000732662e-06, "loss": 2.5352, "step": 215980 }, { "epoch": 0.0053504, "grad_norm": 0.7690852880477905, "learning_rate": 8.36250558662809e-06, "loss": 3.0616, "step": 215990 }, { "epoch": 0.005376, "grad_norm": 0.7326582670211792, "learning_rate": 8.36171118314224e-06, "loss": 2.9382, "step": 216000 }, { "epoch": 0.0054016, "grad_norm": 0.715112030506134, "learning_rate": 8.360916790280251e-06, "loss": 2.8012, "step": 216010 }, { "epoch": 0.0054272, "grad_norm": 0.8124692440032959, "learning_rate": 8.360122408047285e-06, "loss": 2.9068, "step": 216020 }, { "epoch": 0.0054528, "grad_norm": 0.780846357345581, "learning_rate": 8.359328036448488e-06, "loss": 2.7805, "step": 216030 }, { "epoch": 0.0054784, "grad_norm": 0.7192072868347168, "learning_rate": 8.358533675489014e-06, "loss": 2.8794, "step": 216040 }, { "epoch": 0.005504, "grad_norm": 0.748824417591095, "learning_rate": 8.357739325174015e-06, "loss": 2.7553, "step": 216050 }, { "epoch": 0.0055296, "grad_norm": 0.7573769688606262, "learning_rate": 8.356944985508636e-06, "loss": 2.7636, "step": 216060 }, { "epoch": 0.0055552, "grad_norm": 0.7550779581069946, "learning_rate": 8.356150656498034e-06, "loss": 2.8961, "step": 216070 }, { "epoch": 0.0055808, "grad_norm": 0.8833766579627991, "learning_rate": 8.355356338147356e-06, "loss": 2.9617, "step": 216080 }, { "epoch": 0.0056064, "grad_norm": 0.8637098670005798, "learning_rate": 8.354562030461758e-06, "loss": 3.017, "step": 216090 }, { "epoch": 0.005632, "grad_norm": 1.0503016710281372, "learning_rate": 8.353767733446387e-06, "loss": 2.9033, "step": 216100 }, { "epoch": 0.0056576, "grad_norm": 0.813678503036499, "learning_rate": 8.352973447106394e-06, "loss": 2.6647, "step": 216110 }, { "epoch": 0.0056832, "grad_norm": 0.7454414963722229, "learning_rate": 8.352179171446933e-06, "loss": 2.7565, "step": 216120 }, { "epoch": 0.0057088, "grad_norm": 0.7901340126991272, "learning_rate": 8.35138490647315e-06, "loss": 2.8084, "step": 216130 }, { "epoch": 0.0057344, "grad_norm": 0.7432326674461365, "learning_rate": 8.350590652190198e-06, "loss": 2.6294, "step": 216140 }, { "epoch": 0.00576, "grad_norm": 0.8212890028953552, "learning_rate": 8.349796408603225e-06, "loss": 2.7245, "step": 216150 }, { "epoch": 0.0057856, "grad_norm": 0.7167782187461853, "learning_rate": 8.349002175717388e-06, "loss": 2.767, "step": 216160 }, { "epoch": 0.0058112, "grad_norm": 0.7629572153091431, "learning_rate": 8.34820795353783e-06, "loss": 2.8236, "step": 216170 }, { "epoch": 0.0058368, "grad_norm": 0.8469465374946594, "learning_rate": 8.347413742069705e-06, "loss": 2.8064, "step": 216180 }, { "epoch": 0.0058624, "grad_norm": 1.3005088567733765, "learning_rate": 8.346619541318161e-06, "loss": 2.7013, "step": 216190 }, { "epoch": 0.005888, "grad_norm": 0.8821097612380981, "learning_rate": 8.345825351288352e-06, "loss": 2.9522, "step": 216200 }, { "epoch": 0.0059136, "grad_norm": 0.9282352924346924, "learning_rate": 8.345031171985426e-06, "loss": 3.109, "step": 216210 }, { "epoch": 0.0059392, "grad_norm": 0.75807785987854, "learning_rate": 8.344237003414532e-06, "loss": 2.8456, "step": 216220 }, { "epoch": 0.0059648, "grad_norm": 0.7527384161949158, "learning_rate": 8.343442845580823e-06, "loss": 3.0069, "step": 216230 }, { "epoch": 0.0059904, "grad_norm": 0.8424018621444702, "learning_rate": 8.342648698489443e-06, "loss": 2.5535, "step": 216240 }, { "epoch": 0.006016, "grad_norm": 2.3860299587249756, "learning_rate": 8.341854562145551e-06, "loss": 2.8655, "step": 216250 }, { "epoch": 0.0060416, "grad_norm": 0.780433177947998, "learning_rate": 8.341060436554289e-06, "loss": 2.8996, "step": 216260 }, { "epoch": 0.0060672, "grad_norm": 0.7289023995399475, "learning_rate": 8.340266321720809e-06, "loss": 2.9714, "step": 216270 }, { "epoch": 0.0060928, "grad_norm": 1.1839221715927124, "learning_rate": 8.33947221765026e-06, "loss": 2.8318, "step": 216280 }, { "epoch": 0.0061184, "grad_norm": 0.8559364080429077, "learning_rate": 8.338678124347794e-06, "loss": 2.7882, "step": 216290 }, { "epoch": 0.006144, "grad_norm": 0.7645866870880127, "learning_rate": 8.337884041818559e-06, "loss": 2.6172, "step": 216300 }, { "epoch": 0.0061696, "grad_norm": 0.8255678415298462, "learning_rate": 8.337089970067702e-06, "loss": 2.8121, "step": 216310 }, { "epoch": 0.0061952, "grad_norm": 0.9671597480773926, "learning_rate": 8.336295909100379e-06, "loss": 2.8169, "step": 216320 }, { "epoch": 0.0062208, "grad_norm": 0.7359659671783447, "learning_rate": 8.33550185892173e-06, "loss": 3.0102, "step": 216330 }, { "epoch": 0.0062464, "grad_norm": 0.7483954429626465, "learning_rate": 8.334707819536912e-06, "loss": 2.6522, "step": 216340 }, { "epoch": 0.006272, "grad_norm": 0.753512442111969, "learning_rate": 8.333913790951067e-06, "loss": 2.916, "step": 216350 }, { "epoch": 0.0062976, "grad_norm": 0.8410946130752563, "learning_rate": 8.33311977316935e-06, "loss": 2.9481, "step": 216360 }, { "epoch": 0.0063232, "grad_norm": 0.8620734214782715, "learning_rate": 8.33232576619691e-06, "loss": 3.8276, "step": 216370 }, { "epoch": 0.0063488, "grad_norm": 0.7897917628288269, "learning_rate": 8.331531770038894e-06, "loss": 2.9017, "step": 216380 }, { "epoch": 0.0063744, "grad_norm": 0.7230649590492249, "learning_rate": 8.330737784700451e-06, "loss": 3.1596, "step": 216390 }, { "epoch": 0.0064, "grad_norm": 0.7658762335777283, "learning_rate": 8.329943810186728e-06, "loss": 2.9771, "step": 216400 }, { "epoch": 0.0064256, "grad_norm": 0.6956505179405212, "learning_rate": 8.32914984650288e-06, "loss": 2.9181, "step": 216410 }, { "epoch": 0.0064512, "grad_norm": 0.7471106648445129, "learning_rate": 8.328355893654047e-06, "loss": 2.8536, "step": 216420 }, { "epoch": 0.0064768, "grad_norm": 0.960076630115509, "learning_rate": 8.327561951645383e-06, "loss": 2.8965, "step": 216430 }, { "epoch": 0.0065024, "grad_norm": 0.9001531600952148, "learning_rate": 8.326768020482036e-06, "loss": 3.0128, "step": 216440 }, { "epoch": 0.006528, "grad_norm": 0.8348401188850403, "learning_rate": 8.325974100169154e-06, "loss": 2.7009, "step": 216450 }, { "epoch": 0.0065536, "grad_norm": 0.7753674387931824, "learning_rate": 8.325180190711884e-06, "loss": 2.9144, "step": 216460 }, { "epoch": 0.0065792, "grad_norm": 0.8639554381370544, "learning_rate": 8.324386292115374e-06, "loss": 2.8425, "step": 216470 }, { "epoch": 0.0066048, "grad_norm": 1.1044642925262451, "learning_rate": 8.323592404384777e-06, "loss": 2.6359, "step": 216480 }, { "epoch": 0.0066304, "grad_norm": 0.9087499380111694, "learning_rate": 8.322798527525237e-06, "loss": 2.9425, "step": 216490 }, { "epoch": 0.006656, "grad_norm": 0.7850306630134583, "learning_rate": 8.3220046615419e-06, "loss": 3.0273, "step": 216500 }, { "epoch": 0.0066816, "grad_norm": 0.763998806476593, "learning_rate": 8.321210806439918e-06, "loss": 2.8871, "step": 216510 }, { "epoch": 0.0067072, "grad_norm": 0.7567317485809326, "learning_rate": 8.32041696222444e-06, "loss": 2.814, "step": 216520 }, { "epoch": 0.0067328, "grad_norm": 3.178800582885742, "learning_rate": 8.319623128900609e-06, "loss": 2.823, "step": 216530 }, { "epoch": 0.0067584, "grad_norm": 0.8505974411964417, "learning_rate": 8.318829306473575e-06, "loss": 2.9312, "step": 216540 }, { "epoch": 0.006784, "grad_norm": 0.736963152885437, "learning_rate": 8.318035494948486e-06, "loss": 2.8363, "step": 216550 }, { "epoch": 0.0068096, "grad_norm": 0.8842220902442932, "learning_rate": 8.31724169433049e-06, "loss": 2.7549, "step": 216560 }, { "epoch": 0.0068352, "grad_norm": 1.0413658618927002, "learning_rate": 8.316447904624736e-06, "loss": 2.7432, "step": 216570 }, { "epoch": 0.0068608, "grad_norm": 0.9101493954658508, "learning_rate": 8.315654125836372e-06, "loss": 3.0264, "step": 216580 }, { "epoch": 0.0068864, "grad_norm": 0.845587968826294, "learning_rate": 8.31486035797054e-06, "loss": 2.8805, "step": 216590 }, { "epoch": 0.006912, "grad_norm": 3.6710495948791504, "learning_rate": 8.31406660103239e-06, "loss": 2.9269, "step": 216600 }, { "epoch": 0.0069376, "grad_norm": 0.7605859637260437, "learning_rate": 8.313272855027074e-06, "loss": 2.6925, "step": 216610 }, { "epoch": 0.0069632, "grad_norm": 0.7918967008590698, "learning_rate": 8.312479119959731e-06, "loss": 2.7816, "step": 216620 }, { "epoch": 0.0069888, "grad_norm": 0.9264959096908569, "learning_rate": 8.311685395835514e-06, "loss": 2.9739, "step": 216630 }, { "epoch": 0.0070144, "grad_norm": 0.7967023253440857, "learning_rate": 8.310891682659569e-06, "loss": 3.0495, "step": 216640 }, { "epoch": 0.00704, "grad_norm": 0.858770489692688, "learning_rate": 8.310097980437044e-06, "loss": 2.927, "step": 216650 }, { "epoch": 0.0070656, "grad_norm": 0.8652154803276062, "learning_rate": 8.309304289173082e-06, "loss": 2.7911, "step": 216660 }, { "epoch": 0.0070912, "grad_norm": 0.7701793313026428, "learning_rate": 8.308510608872833e-06, "loss": 2.7382, "step": 216670 }, { "epoch": 0.0071168, "grad_norm": 0.770660400390625, "learning_rate": 8.307716939541443e-06, "loss": 2.8871, "step": 216680 }, { "epoch": 0.0071424, "grad_norm": 0.8779729008674622, "learning_rate": 8.30692328118406e-06, "loss": 3.1298, "step": 216690 }, { "epoch": 0.007168, "grad_norm": 0.8707393407821655, "learning_rate": 8.306129633805828e-06, "loss": 2.8947, "step": 216700 }, { "epoch": 0.0071936, "grad_norm": 0.9135432839393616, "learning_rate": 8.305335997411894e-06, "loss": 3.0338, "step": 216710 }, { "epoch": 0.0072192, "grad_norm": 0.7373940348625183, "learning_rate": 8.304542372007407e-06, "loss": 2.9728, "step": 216720 }, { "epoch": 0.0072448, "grad_norm": 0.7915639877319336, "learning_rate": 8.303748757597511e-06, "loss": 3.1032, "step": 216730 }, { "epoch": 0.0072704, "grad_norm": 1.1146838665008545, "learning_rate": 8.302955154187357e-06, "loss": 2.8954, "step": 216740 }, { "epoch": 0.007296, "grad_norm": 0.7591655254364014, "learning_rate": 8.302161561782086e-06, "loss": 3.0046, "step": 216750 }, { "epoch": 0.0073216, "grad_norm": 0.808245837688446, "learning_rate": 8.301367980386846e-06, "loss": 2.8743, "step": 216760 }, { "epoch": 0.0073472, "grad_norm": 0.7666369676589966, "learning_rate": 8.300574410006783e-06, "loss": 2.7974, "step": 216770 }, { "epoch": 0.0073728, "grad_norm": 0.7019927501678467, "learning_rate": 8.299780850647044e-06, "loss": 2.8989, "step": 216780 }, { "epoch": 0.0073984, "grad_norm": 0.8024898767471313, "learning_rate": 8.298987302312775e-06, "loss": 2.7647, "step": 216790 }, { "epoch": 0.007424, "grad_norm": 0.8144740462303162, "learning_rate": 8.29819376500912e-06, "loss": 3.0108, "step": 216800 }, { "epoch": 0.0074496, "grad_norm": 0.7331655621528625, "learning_rate": 8.297400238741226e-06, "loss": 2.9051, "step": 216810 }, { "epoch": 0.0074752, "grad_norm": 0.7405144572257996, "learning_rate": 8.29660672351424e-06, "loss": 2.964, "step": 216820 }, { "epoch": 0.0075008, "grad_norm": 0.7441715598106384, "learning_rate": 8.295813219333303e-06, "loss": 2.9554, "step": 216830 }, { "epoch": 0.0075264, "grad_norm": 0.8149384260177612, "learning_rate": 8.295019726203568e-06, "loss": 2.7237, "step": 216840 }, { "epoch": 0.007552, "grad_norm": 0.8493891358375549, "learning_rate": 8.29422624413018e-06, "loss": 2.8371, "step": 216850 }, { "epoch": 0.0075776, "grad_norm": 0.8319283723831177, "learning_rate": 8.293432773118276e-06, "loss": 2.8112, "step": 216860 }, { "epoch": 0.0076032, "grad_norm": 0.7419518232345581, "learning_rate": 8.292639313173008e-06, "loss": 3.0091, "step": 216870 }, { "epoch": 0.0076288, "grad_norm": 0.7938895225524902, "learning_rate": 8.291845864299518e-06, "loss": 2.9714, "step": 216880 }, { "epoch": 0.0076544, "grad_norm": 0.7656148672103882, "learning_rate": 8.291052426502956e-06, "loss": 2.878, "step": 216890 }, { "epoch": 0.00768, "grad_norm": 0.7852144837379456, "learning_rate": 8.290258999788464e-06, "loss": 3.0371, "step": 216900 }, { "epoch": 0.0077056, "grad_norm": 0.9524335265159607, "learning_rate": 8.289465584161192e-06, "loss": 2.6635, "step": 216910 }, { "epoch": 0.0077312, "grad_norm": 0.7580116391181946, "learning_rate": 8.288672179626277e-06, "loss": 3.0687, "step": 216920 }, { "epoch": 0.0077568, "grad_norm": 0.9372535943984985, "learning_rate": 8.28787878618887e-06, "loss": 2.8748, "step": 216930 }, { "epoch": 0.0077824, "grad_norm": 0.8442044854164124, "learning_rate": 8.287085403854114e-06, "loss": 3.1431, "step": 216940 }, { "epoch": 0.007808, "grad_norm": 0.8021499514579773, "learning_rate": 8.286292032627154e-06, "loss": 2.998, "step": 216950 }, { "epoch": 0.0078336, "grad_norm": 0.7810829281806946, "learning_rate": 8.285498672513133e-06, "loss": 2.7528, "step": 216960 }, { "epoch": 0.0078592, "grad_norm": 0.7318596243858337, "learning_rate": 8.284705323517198e-06, "loss": 2.9279, "step": 216970 }, { "epoch": 0.0078848, "grad_norm": 0.8137549757957458, "learning_rate": 8.283911985644496e-06, "loss": 2.9411, "step": 216980 }, { "epoch": 0.0079104, "grad_norm": 0.7838531136512756, "learning_rate": 8.283118658900165e-06, "loss": 2.6778, "step": 216990 }, { "epoch": 0.007936, "grad_norm": 0.9135526418685913, "learning_rate": 8.282325343289355e-06, "loss": 2.9929, "step": 217000 }, { "epoch": 0.0079616, "grad_norm": 0.7698044776916504, "learning_rate": 8.281532038817209e-06, "loss": 3.0794, "step": 217010 }, { "epoch": 0.0079872, "grad_norm": 0.920093834400177, "learning_rate": 8.28073874548887e-06, "loss": 3.0351, "step": 217020 }, { "epoch": 0.0080128, "grad_norm": 0.8182392120361328, "learning_rate": 8.279945463309483e-06, "loss": 2.7579, "step": 217030 }, { "epoch": 0.0080384, "grad_norm": 0.8319531679153442, "learning_rate": 8.279152192284193e-06, "loss": 2.8939, "step": 217040 }, { "epoch": 0.008064, "grad_norm": 0.7235521674156189, "learning_rate": 8.278358932418146e-06, "loss": 2.8309, "step": 217050 }, { "epoch": 0.0080896, "grad_norm": 0.7723724246025085, "learning_rate": 8.277565683716475e-06, "loss": 2.9275, "step": 217060 }, { "epoch": 0.0081152, "grad_norm": 0.8556366562843323, "learning_rate": 8.276772446184343e-06, "loss": 2.9351, "step": 217070 }, { "epoch": 0.0081408, "grad_norm": 0.7403267621994019, "learning_rate": 8.275979219826879e-06, "loss": 2.799, "step": 217080 }, { "epoch": 0.0081664, "grad_norm": 1.005760908126831, "learning_rate": 8.275186004649232e-06, "loss": 2.8898, "step": 217090 }, { "epoch": 0.008192, "grad_norm": 0.7436572313308716, "learning_rate": 8.274392800656546e-06, "loss": 2.9883, "step": 217100 }, { "epoch": 0.0082176, "grad_norm": 0.8740007281303406, "learning_rate": 8.273599607853964e-06, "loss": 2.9554, "step": 217110 }, { "epoch": 0.0082432, "grad_norm": 0.7462175488471985, "learning_rate": 8.27280642624663e-06, "loss": 2.8272, "step": 217120 }, { "epoch": 0.0082688, "grad_norm": 0.7493792772293091, "learning_rate": 8.272013255839687e-06, "loss": 2.7758, "step": 217130 }, { "epoch": 0.0082944, "grad_norm": 0.9808236956596375, "learning_rate": 8.27122009663828e-06, "loss": 3.1186, "step": 217140 }, { "epoch": 0.00832, "grad_norm": 0.7322142720222473, "learning_rate": 8.27042694864755e-06, "loss": 2.7552, "step": 217150 }, { "epoch": 0.0083456, "grad_norm": 0.8016778826713562, "learning_rate": 8.26963381187264e-06, "loss": 2.8873, "step": 217160 }, { "epoch": 0.0083712, "grad_norm": 0.9016302227973938, "learning_rate": 8.268840686318697e-06, "loss": 2.7831, "step": 217170 }, { "epoch": 0.0083968, "grad_norm": 0.817600667476654, "learning_rate": 8.268047571990863e-06, "loss": 2.9767, "step": 217180 }, { "epoch": 0.0084224, "grad_norm": 0.8147240877151489, "learning_rate": 8.267254468894279e-06, "loss": 2.7379, "step": 217190 }, { "epoch": 0.008448, "grad_norm": 0.7317540645599365, "learning_rate": 8.266461377034087e-06, "loss": 2.8395, "step": 217200 }, { "epoch": 0.0084736, "grad_norm": 0.8342564702033997, "learning_rate": 8.265668296415436e-06, "loss": 2.7717, "step": 217210 }, { "epoch": 0.0084992, "grad_norm": 0.8083041310310364, "learning_rate": 8.264875227043464e-06, "loss": 2.9682, "step": 217220 }, { "epoch": 0.0085248, "grad_norm": 0.7693819999694824, "learning_rate": 8.264082168923312e-06, "loss": 2.9773, "step": 217230 }, { "epoch": 0.0085504, "grad_norm": 0.8553653955459595, "learning_rate": 8.263289122060127e-06, "loss": 2.829, "step": 217240 }, { "epoch": 0.008576, "grad_norm": 0.8825093507766724, "learning_rate": 8.262496086459052e-06, "loss": 2.8869, "step": 217250 }, { "epoch": 0.0086016, "grad_norm": 0.956484854221344, "learning_rate": 8.261703062125227e-06, "loss": 2.8763, "step": 217260 }, { "epoch": 0.0086272, "grad_norm": 0.7809504866600037, "learning_rate": 8.260910049063797e-06, "loss": 2.8902, "step": 217270 }, { "epoch": 0.0086528, "grad_norm": 0.749428391456604, "learning_rate": 8.260117047279902e-06, "loss": 2.9055, "step": 217280 }, { "epoch": 0.0086784, "grad_norm": 0.7977467179298401, "learning_rate": 8.259324056778687e-06, "loss": 3.0098, "step": 217290 }, { "epoch": 0.008704, "grad_norm": 0.7921832203865051, "learning_rate": 8.258531077565293e-06, "loss": 2.9315, "step": 217300 }, { "epoch": 0.0087296, "grad_norm": 0.7553516030311584, "learning_rate": 8.25773810964486e-06, "loss": 2.9688, "step": 217310 }, { "epoch": 0.0087552, "grad_norm": 0.7402666807174683, "learning_rate": 8.256945153022535e-06, "loss": 2.8666, "step": 217320 }, { "epoch": 0.0087808, "grad_norm": 0.7727240920066833, "learning_rate": 8.256152207703454e-06, "loss": 2.8699, "step": 217330 }, { "epoch": 0.0088064, "grad_norm": 0.9164043068885803, "learning_rate": 8.255359273692766e-06, "loss": 2.9087, "step": 217340 }, { "epoch": 0.008832, "grad_norm": 0.8786606788635254, "learning_rate": 8.254566350995607e-06, "loss": 2.8128, "step": 217350 }, { "epoch": 0.0088576, "grad_norm": 0.7443834543228149, "learning_rate": 8.253773439617121e-06, "loss": 2.92, "step": 217360 }, { "epoch": 0.0088832, "grad_norm": 0.749321460723877, "learning_rate": 8.252980539562451e-06, "loss": 2.7623, "step": 217370 }, { "epoch": 0.0089088, "grad_norm": 0.7688674926757812, "learning_rate": 8.25218765083674e-06, "loss": 2.7556, "step": 217380 }, { "epoch": 0.0089344, "grad_norm": 0.7718021273612976, "learning_rate": 8.251394773445126e-06, "loss": 2.7996, "step": 217390 }, { "epoch": 0.00896, "grad_norm": 0.7940022349357605, "learning_rate": 8.250601907392752e-06, "loss": 2.9626, "step": 217400 }, { "epoch": 0.0089856, "grad_norm": 0.8832836151123047, "learning_rate": 8.249809052684756e-06, "loss": 3.0748, "step": 217410 }, { "epoch": 0.0090112, "grad_norm": 0.7387914657592773, "learning_rate": 8.249016209326287e-06, "loss": 2.8961, "step": 217420 }, { "epoch": 0.0090368, "grad_norm": 0.7689379453659058, "learning_rate": 8.248223377322485e-06, "loss": 2.9116, "step": 217430 }, { "epoch": 0.0090624, "grad_norm": 1.769026756286621, "learning_rate": 8.247430556678485e-06, "loss": 2.6826, "step": 217440 }, { "epoch": 0.009088, "grad_norm": 0.8361427783966064, "learning_rate": 8.246637747399433e-06, "loss": 2.7288, "step": 217450 }, { "epoch": 0.0091136, "grad_norm": 0.7829599976539612, "learning_rate": 8.245844949490468e-06, "loss": 2.8098, "step": 217460 }, { "epoch": 0.0091392, "grad_norm": 1.9312118291854858, "learning_rate": 8.245052162956736e-06, "loss": 3.0412, "step": 217470 }, { "epoch": 0.0091648, "grad_norm": 0.8163774609565735, "learning_rate": 8.244259387803372e-06, "loss": 2.9211, "step": 217480 }, { "epoch": 0.0091904, "grad_norm": 0.887183427810669, "learning_rate": 8.243466624035519e-06, "loss": 2.9028, "step": 217490 }, { "epoch": 0.009216, "grad_norm": 0.8271515965461731, "learning_rate": 8.242673871658319e-06, "loss": 3.066, "step": 217500 }, { "epoch": 0.0092416, "grad_norm": 0.8730633854866028, "learning_rate": 8.24188113067691e-06, "loss": 2.8494, "step": 217510 }, { "epoch": 0.0092672, "grad_norm": 0.7832195162773132, "learning_rate": 8.241088401096437e-06, "loss": 2.9457, "step": 217520 }, { "epoch": 0.0092928, "grad_norm": 0.930368959903717, "learning_rate": 8.240295682922036e-06, "loss": 2.8426, "step": 217530 }, { "epoch": 0.0093184, "grad_norm": 0.790823757648468, "learning_rate": 8.239502976158853e-06, "loss": 2.9522, "step": 217540 }, { "epoch": 0.009344, "grad_norm": 0.747687816619873, "learning_rate": 8.238710280812022e-06, "loss": 2.8257, "step": 217550 }, { "epoch": 0.0093696, "grad_norm": 0.9719208478927612, "learning_rate": 8.237917596886689e-06, "loss": 2.8072, "step": 217560 }, { "epoch": 0.0093952, "grad_norm": 0.8590615391731262, "learning_rate": 8.23712492438799e-06, "loss": 2.9742, "step": 217570 }, { "epoch": 0.0094208, "grad_norm": 0.8283329010009766, "learning_rate": 8.236332263321065e-06, "loss": 2.8155, "step": 217580 }, { "epoch": 0.0094464, "grad_norm": 0.8250512480735779, "learning_rate": 8.23553961369106e-06, "loss": 2.9217, "step": 217590 }, { "epoch": 0.009472, "grad_norm": 0.8298590183258057, "learning_rate": 8.234746975503114e-06, "loss": 2.9041, "step": 217600 }, { "epoch": 0.0094976, "grad_norm": 0.9087252020835876, "learning_rate": 8.233954348762362e-06, "loss": 2.9946, "step": 217610 }, { "epoch": 0.0095232, "grad_norm": 0.7894208431243896, "learning_rate": 8.233161733473946e-06, "loss": 2.9361, "step": 217620 }, { "epoch": 0.0095488, "grad_norm": 0.7001471519470215, "learning_rate": 8.23236912964301e-06, "loss": 2.8149, "step": 217630 }, { "epoch": 0.0095744, "grad_norm": 0.8472777009010315, "learning_rate": 8.231576537274688e-06, "loss": 2.8266, "step": 217640 }, { "epoch": 0.0096, "grad_norm": 0.8589779138565063, "learning_rate": 8.23078395637412e-06, "loss": 2.8797, "step": 217650 }, { "epoch": 0.0096256, "grad_norm": 0.7554994225502014, "learning_rate": 8.229991386946449e-06, "loss": 2.9267, "step": 217660 }, { "epoch": 0.0096512, "grad_norm": 0.9310508370399475, "learning_rate": 8.229198828996817e-06, "loss": 3.0794, "step": 217670 }, { "epoch": 0.0096768, "grad_norm": 0.8356700539588928, "learning_rate": 8.228406282530357e-06, "loss": 2.8733, "step": 217680 }, { "epoch": 0.0097024, "grad_norm": 0.9308610558509827, "learning_rate": 8.227613747552212e-06, "loss": 2.9582, "step": 217690 }, { "epoch": 0.009728, "grad_norm": 0.9345495700836182, "learning_rate": 8.226821224067523e-06, "loss": 3.0183, "step": 217700 }, { "epoch": 0.0097536, "grad_norm": 1.0389646291732788, "learning_rate": 8.226028712081424e-06, "loss": 2.8267, "step": 217710 }, { "epoch": 0.0097792, "grad_norm": 0.8361062407493591, "learning_rate": 8.225236211599058e-06, "loss": 2.9679, "step": 217720 }, { "epoch": 0.0098048, "grad_norm": 0.731678307056427, "learning_rate": 8.224443722625564e-06, "loss": 2.7922, "step": 217730 }, { "epoch": 0.0098304, "grad_norm": 1.2502886056900024, "learning_rate": 8.223651245166082e-06, "loss": 3.0989, "step": 217740 }, { "epoch": 0.009856, "grad_norm": 0.7452460527420044, "learning_rate": 8.222858779225741e-06, "loss": 2.7891, "step": 217750 }, { "epoch": 0.0098816, "grad_norm": 0.8476534485816956, "learning_rate": 8.222066324809697e-06, "loss": 2.9964, "step": 217760 }, { "epoch": 0.0099072, "grad_norm": 0.7842895984649658, "learning_rate": 8.22127388192308e-06, "loss": 3.0005, "step": 217770 }, { "epoch": 0.0099328, "grad_norm": 0.7952612638473511, "learning_rate": 8.220481450571026e-06, "loss": 2.8321, "step": 217780 }, { "epoch": 0.0099584, "grad_norm": 0.8794410228729248, "learning_rate": 8.219689030758679e-06, "loss": 3.0199, "step": 217790 }, { "epoch": 0.009984, "grad_norm": 0.8577861189842224, "learning_rate": 8.218896622491177e-06, "loss": 3.4429, "step": 217800 }, { "epoch": 0.0100096, "grad_norm": 0.8323978185653687, "learning_rate": 8.218104225773655e-06, "loss": 2.7844, "step": 217810 }, { "epoch": 0.0100352, "grad_norm": 0.7184524536132812, "learning_rate": 8.217311840611254e-06, "loss": 2.9149, "step": 217820 }, { "epoch": 0.0100608, "grad_norm": 0.8411396145820618, "learning_rate": 8.216519467009112e-06, "loss": 2.7737, "step": 217830 }, { "epoch": 0.0100864, "grad_norm": 0.8073025941848755, "learning_rate": 8.215727104972367e-06, "loss": 2.9575, "step": 217840 }, { "epoch": 0.010112, "grad_norm": 0.7476686835289001, "learning_rate": 8.214934754506156e-06, "loss": 2.9005, "step": 217850 }, { "epoch": 0.0101376, "grad_norm": 0.7762824296951294, "learning_rate": 8.21414241561562e-06, "loss": 2.9782, "step": 217860 }, { "epoch": 0.0101632, "grad_norm": 0.7489705681800842, "learning_rate": 8.213350088305899e-06, "loss": 2.8858, "step": 217870 }, { "epoch": 0.0101888, "grad_norm": 0.7784137725830078, "learning_rate": 8.212557772582123e-06, "loss": 2.7593, "step": 217880 }, { "epoch": 0.0102144, "grad_norm": 0.7556471228599548, "learning_rate": 8.211765468449436e-06, "loss": 2.8625, "step": 217890 }, { "epoch": 0.01024, "grad_norm": 0.8198294043540955, "learning_rate": 8.210973175912977e-06, "loss": 3.1099, "step": 217900 }, { "epoch": 0.0102656, "grad_norm": 0.8106407523155212, "learning_rate": 8.21018089497788e-06, "loss": 2.7425, "step": 217910 }, { "epoch": 0.0102912, "grad_norm": 0.9266706109046936, "learning_rate": 8.209388625649284e-06, "loss": 2.823, "step": 217920 }, { "epoch": 0.0103168, "grad_norm": 0.7946645021438599, "learning_rate": 8.208596367932324e-06, "loss": 2.9391, "step": 217930 }, { "epoch": 0.0103424, "grad_norm": 0.8257967233657837, "learning_rate": 8.207804121832143e-06, "loss": 3.021, "step": 217940 }, { "epoch": 0.010368, "grad_norm": 0.9805594682693481, "learning_rate": 8.207011887353875e-06, "loss": 3.0703, "step": 217950 }, { "epoch": 0.0103936, "grad_norm": 0.8563771843910217, "learning_rate": 8.206219664502661e-06, "loss": 2.9446, "step": 217960 }, { "epoch": 0.0104192, "grad_norm": 0.9295702576637268, "learning_rate": 8.205427453283634e-06, "loss": 2.7844, "step": 217970 }, { "epoch": 0.0104448, "grad_norm": 0.7791233062744141, "learning_rate": 8.204635253701933e-06, "loss": 3.1435, "step": 217980 }, { "epoch": 0.0104704, "grad_norm": 0.8584176898002625, "learning_rate": 8.203843065762695e-06, "loss": 2.8362, "step": 217990 }, { "epoch": 0.010496, "grad_norm": 0.8292179703712463, "learning_rate": 8.20305088947106e-06, "loss": 3.1005, "step": 218000 }, { "epoch": 0.0105216, "grad_norm": 0.984631359577179, "learning_rate": 8.20225872483216e-06, "loss": 2.9018, "step": 218010 }, { "epoch": 0.0105472, "grad_norm": 0.878298819065094, "learning_rate": 8.201466571851136e-06, "loss": 3.0774, "step": 218020 }, { "epoch": 0.0105728, "grad_norm": 0.790287971496582, "learning_rate": 8.200674430533123e-06, "loss": 2.8003, "step": 218030 }, { "epoch": 0.0105984, "grad_norm": 0.7842586040496826, "learning_rate": 8.199882300883257e-06, "loss": 3.0974, "step": 218040 }, { "epoch": 0.010624, "grad_norm": 0.7901291847229004, "learning_rate": 8.199090182906675e-06, "loss": 2.978, "step": 218050 }, { "epoch": 0.0106496, "grad_norm": 1.019837737083435, "learning_rate": 8.198298076608515e-06, "loss": 2.8734, "step": 218060 }, { "epoch": 0.0106752, "grad_norm": 0.87307208776474, "learning_rate": 8.197505981993916e-06, "loss": 3.0, "step": 218070 }, { "epoch": 0.0107008, "grad_norm": 0.8344899415969849, "learning_rate": 8.19671389906801e-06, "loss": 2.863, "step": 218080 }, { "epoch": 0.0107264, "grad_norm": 0.8284819722175598, "learning_rate": 8.195921827835933e-06, "loss": 3.3563, "step": 218090 }, { "epoch": 0.010752, "grad_norm": 0.9435453414916992, "learning_rate": 8.195129768302824e-06, "loss": 2.8134, "step": 218100 }, { "epoch": 0.0107776, "grad_norm": 0.8210684061050415, "learning_rate": 8.194337720473818e-06, "loss": 2.9215, "step": 218110 }, { "epoch": 0.0108032, "grad_norm": 0.7173504829406738, "learning_rate": 8.193545684354054e-06, "loss": 3.017, "step": 218120 }, { "epoch": 0.0108288, "grad_norm": 0.7890676856040955, "learning_rate": 8.192753659948667e-06, "loss": 2.831, "step": 218130 }, { "epoch": 0.0108544, "grad_norm": 0.8365398645401001, "learning_rate": 8.19196164726279e-06, "loss": 2.9009, "step": 218140 }, { "epoch": 0.01088, "grad_norm": 0.7817155718803406, "learning_rate": 8.191169646301562e-06, "loss": 2.8871, "step": 218150 }, { "epoch": 0.0109056, "grad_norm": 0.9793848991394043, "learning_rate": 8.190377657070121e-06, "loss": 2.926, "step": 218160 }, { "epoch": 0.0109312, "grad_norm": 1.098870038986206, "learning_rate": 8.189585679573596e-06, "loss": 2.8596, "step": 218170 }, { "epoch": 0.0109568, "grad_norm": 1.0523709058761597, "learning_rate": 8.188793713817127e-06, "loss": 3.2228, "step": 218180 }, { "epoch": 0.0109824, "grad_norm": 0.7461323738098145, "learning_rate": 8.188001759805851e-06, "loss": 2.7823, "step": 218190 }, { "epoch": 0.011008, "grad_norm": 0.9281839728355408, "learning_rate": 8.187209817544903e-06, "loss": 2.8644, "step": 218200 }, { "epoch": 0.0110336, "grad_norm": 0.9742804169654846, "learning_rate": 8.186417887039416e-06, "loss": 2.8347, "step": 218210 }, { "epoch": 0.0110592, "grad_norm": 0.7129731178283691, "learning_rate": 8.185625968294528e-06, "loss": 2.7387, "step": 218220 }, { "epoch": 0.0110848, "grad_norm": 0.9158539175987244, "learning_rate": 8.184834061315373e-06, "loss": 2.7918, "step": 218230 }, { "epoch": 0.0111104, "grad_norm": 0.9052668809890747, "learning_rate": 8.184042166107086e-06, "loss": 2.7183, "step": 218240 }, { "epoch": 0.011136, "grad_norm": 0.7583763003349304, "learning_rate": 8.183250282674803e-06, "loss": 3.0284, "step": 218250 }, { "epoch": 0.0111616, "grad_norm": 0.8880191445350647, "learning_rate": 8.182458411023658e-06, "loss": 2.8974, "step": 218260 }, { "epoch": 0.0111872, "grad_norm": 0.7612489461898804, "learning_rate": 8.18166655115879e-06, "loss": 2.956, "step": 218270 }, { "epoch": 0.0112128, "grad_norm": 0.6877922415733337, "learning_rate": 8.180874703085326e-06, "loss": 2.6496, "step": 218280 }, { "epoch": 0.0112384, "grad_norm": 0.9598662853240967, "learning_rate": 8.180082866808411e-06, "loss": 2.8949, "step": 218290 }, { "epoch": 0.011264, "grad_norm": 0.7058702707290649, "learning_rate": 8.179291042333172e-06, "loss": 2.9481, "step": 218300 }, { "epoch": 0.0112896, "grad_norm": 0.9976160526275635, "learning_rate": 8.178499229664747e-06, "loss": 3.0827, "step": 218310 }, { "epoch": 0.0113152, "grad_norm": 0.9791445136070251, "learning_rate": 8.177707428808275e-06, "loss": 3.033, "step": 218320 }, { "epoch": 0.0113408, "grad_norm": 0.792060136795044, "learning_rate": 8.176915639768882e-06, "loss": 2.9228, "step": 218330 }, { "epoch": 0.0113664, "grad_norm": 0.8606634736061096, "learning_rate": 8.176123862551706e-06, "loss": 2.8843, "step": 218340 }, { "epoch": 0.011392, "grad_norm": 0.8081191778182983, "learning_rate": 8.175332097161883e-06, "loss": 2.9791, "step": 218350 }, { "epoch": 0.0114176, "grad_norm": 0.7950722575187683, "learning_rate": 8.174540343604548e-06, "loss": 2.8779, "step": 218360 }, { "epoch": 0.0114432, "grad_norm": 0.9116170406341553, "learning_rate": 8.173748601884832e-06, "loss": 3.0305, "step": 218370 }, { "epoch": 0.0114688, "grad_norm": 1.3067147731781006, "learning_rate": 8.17295687200787e-06, "loss": 2.9721, "step": 218380 }, { "epoch": 0.0114944, "grad_norm": 0.788001537322998, "learning_rate": 8.172165153978801e-06, "loss": 2.9292, "step": 218390 }, { "epoch": 0.01152, "grad_norm": 0.8768718242645264, "learning_rate": 8.171373447802754e-06, "loss": 2.7307, "step": 218400 }, { "epoch": 0.0115456, "grad_norm": 0.7498801350593567, "learning_rate": 8.170581753484862e-06, "loss": 2.8839, "step": 218410 }, { "epoch": 0.0115712, "grad_norm": 0.8444442749023438, "learning_rate": 8.169790071030262e-06, "loss": 3.0811, "step": 218420 }, { "epoch": 0.0115968, "grad_norm": 0.836551308631897, "learning_rate": 8.168998400444088e-06, "loss": 2.9243, "step": 218430 }, { "epoch": 0.0116224, "grad_norm": 0.7662800550460815, "learning_rate": 8.168206741731472e-06, "loss": 3.0999, "step": 218440 }, { "epoch": 0.011648, "grad_norm": 0.8920270800590515, "learning_rate": 8.167415094897545e-06, "loss": 3.1517, "step": 218450 }, { "epoch": 0.0116736, "grad_norm": 0.7353922724723816, "learning_rate": 8.166623459947447e-06, "loss": 2.8619, "step": 218460 }, { "epoch": 0.0116992, "grad_norm": 0.7889896035194397, "learning_rate": 8.165831836886308e-06, "loss": 2.8453, "step": 218470 }, { "epoch": 0.0117248, "grad_norm": 0.7801373600959778, "learning_rate": 8.165040225719263e-06, "loss": 2.6622, "step": 218480 }, { "epoch": 0.0117504, "grad_norm": 0.8881158232688904, "learning_rate": 8.164248626451445e-06, "loss": 2.9727, "step": 218490 }, { "epoch": 0.011776, "grad_norm": 1.3182326555252075, "learning_rate": 8.163457039087987e-06, "loss": 3.0956, "step": 218500 }, { "epoch": 0.0118016, "grad_norm": 0.7657951712608337, "learning_rate": 8.162665463634019e-06, "loss": 2.9751, "step": 218510 }, { "epoch": 0.0118272, "grad_norm": 0.7308034896850586, "learning_rate": 8.16187390009468e-06, "loss": 2.9379, "step": 218520 }, { "epoch": 0.0118528, "grad_norm": 0.8915238976478577, "learning_rate": 8.1610823484751e-06, "loss": 2.894, "step": 218530 }, { "epoch": 0.0118784, "grad_norm": 0.8574779629707336, "learning_rate": 8.160290808780412e-06, "loss": 2.8377, "step": 218540 }, { "epoch": 0.011904, "grad_norm": 0.7585484385490417, "learning_rate": 8.159499281015748e-06, "loss": 2.8185, "step": 218550 }, { "epoch": 0.0119296, "grad_norm": 0.7833702564239502, "learning_rate": 8.158707765186245e-06, "loss": 2.8455, "step": 218560 }, { "epoch": 0.0119552, "grad_norm": 0.7353063225746155, "learning_rate": 8.15791626129703e-06, "loss": 2.7259, "step": 218570 }, { "epoch": 0.0119808, "grad_norm": 0.7757417559623718, "learning_rate": 8.15712476935324e-06, "loss": 2.9959, "step": 218580 }, { "epoch": 0.0120064, "grad_norm": 0.7986888885498047, "learning_rate": 8.156333289360006e-06, "loss": 2.8903, "step": 218590 }, { "epoch": 0.012032, "grad_norm": 0.803796648979187, "learning_rate": 8.15554182132246e-06, "loss": 2.7989, "step": 218600 }, { "epoch": 0.0120576, "grad_norm": 0.795356273651123, "learning_rate": 8.154750365245735e-06, "loss": 2.5824, "step": 218610 }, { "epoch": 0.0120832, "grad_norm": 0.772516131401062, "learning_rate": 8.153958921134961e-06, "loss": 2.9096, "step": 218620 }, { "epoch": 0.0121088, "grad_norm": 0.7211371660232544, "learning_rate": 8.153167488995275e-06, "loss": 2.8854, "step": 218630 }, { "epoch": 0.0121344, "grad_norm": 0.7893436551094055, "learning_rate": 8.152376068831806e-06, "loss": 2.7415, "step": 218640 }, { "epoch": 0.01216, "grad_norm": 1.3551157712936401, "learning_rate": 8.151584660649691e-06, "loss": 2.8632, "step": 218650 }, { "epoch": 0.0121856, "grad_norm": 0.7117769718170166, "learning_rate": 8.150793264454056e-06, "loss": 2.9726, "step": 218660 }, { "epoch": 0.0122112, "grad_norm": 0.7462221384048462, "learning_rate": 8.150001880250033e-06, "loss": 2.835, "step": 218670 }, { "epoch": 0.0122368, "grad_norm": 0.761435329914093, "learning_rate": 8.149210508042758e-06, "loss": 3.1507, "step": 218680 }, { "epoch": 0.0122624, "grad_norm": 0.7520955801010132, "learning_rate": 8.148419147837363e-06, "loss": 2.7818, "step": 218690 }, { "epoch": 0.012288, "grad_norm": 0.7512128353118896, "learning_rate": 8.147627799638976e-06, "loss": 2.8314, "step": 218700 }, { "epoch": 0.0123136, "grad_norm": 0.7761387228965759, "learning_rate": 8.146836463452728e-06, "loss": 3.0175, "step": 218710 }, { "epoch": 0.0123392, "grad_norm": 0.9292716979980469, "learning_rate": 8.146045139283758e-06, "loss": 2.8933, "step": 218720 }, { "epoch": 0.0123648, "grad_norm": 0.9295843243598938, "learning_rate": 8.14525382713719e-06, "loss": 2.822, "step": 218730 }, { "epoch": 0.0123904, "grad_norm": 0.7637511491775513, "learning_rate": 8.144462527018156e-06, "loss": 2.9828, "step": 218740 }, { "epoch": 0.012416, "grad_norm": 0.7981075644493103, "learning_rate": 8.143671238931791e-06, "loss": 2.8555, "step": 218750 }, { "epoch": 0.0124416, "grad_norm": 0.8540627956390381, "learning_rate": 8.142879962883226e-06, "loss": 2.9941, "step": 218760 }, { "epoch": 0.0124672, "grad_norm": 1.0199673175811768, "learning_rate": 8.14208869887759e-06, "loss": 2.7361, "step": 218770 }, { "epoch": 0.0124928, "grad_norm": 0.8109627366065979, "learning_rate": 8.141297446920014e-06, "loss": 3.0737, "step": 218780 }, { "epoch": 0.0125184, "grad_norm": 0.8638134002685547, "learning_rate": 8.140506207015632e-06, "loss": 2.9885, "step": 218790 }, { "epoch": 0.012544, "grad_norm": 0.8740606307983398, "learning_rate": 8.139714979169568e-06, "loss": 2.7468, "step": 218800 }, { "epoch": 0.0125696, "grad_norm": 0.7384420037269592, "learning_rate": 8.138923763386964e-06, "loss": 2.9732, "step": 218810 }, { "epoch": 0.0125952, "grad_norm": 0.7262102961540222, "learning_rate": 8.13813255967294e-06, "loss": 3.0199, "step": 218820 }, { "epoch": 0.0126208, "grad_norm": 0.8784823417663574, "learning_rate": 8.137341368032634e-06, "loss": 3.0536, "step": 218830 }, { "epoch": 0.0126464, "grad_norm": 0.8326631188392639, "learning_rate": 8.136550188471174e-06, "loss": 3.0156, "step": 218840 }, { "epoch": 0.012672, "grad_norm": 1.327592372894287, "learning_rate": 8.135759020993693e-06, "loss": 2.8454, "step": 218850 }, { "epoch": 0.0126976, "grad_norm": 0.8036624789237976, "learning_rate": 8.134967865605316e-06, "loss": 2.8942, "step": 218860 }, { "epoch": 0.0127232, "grad_norm": 0.7947967648506165, "learning_rate": 8.13417672231118e-06, "loss": 3.0531, "step": 218870 }, { "epoch": 0.0127488, "grad_norm": 0.730155348777771, "learning_rate": 8.133385591116408e-06, "loss": 2.9208, "step": 218880 }, { "epoch": 0.0127744, "grad_norm": 0.7119919061660767, "learning_rate": 8.132594472026138e-06, "loss": 2.8952, "step": 218890 }, { "epoch": 0.0128, "grad_norm": 0.8043543696403503, "learning_rate": 8.131803365045495e-06, "loss": 3.0435, "step": 218900 }, { "epoch": 0.0128256, "grad_norm": 0.9871075749397278, "learning_rate": 8.13101227017961e-06, "loss": 2.997, "step": 218910 }, { "epoch": 0.0128512, "grad_norm": 1.057724952697754, "learning_rate": 8.130221187433617e-06, "loss": 2.8261, "step": 218920 }, { "epoch": 0.0128768, "grad_norm": 0.920061469078064, "learning_rate": 8.12943011681264e-06, "loss": 2.8798, "step": 218930 }, { "epoch": 0.0129024, "grad_norm": 0.7646991610527039, "learning_rate": 8.12863905832181e-06, "loss": 2.8588, "step": 218940 }, { "epoch": 0.012928, "grad_norm": 0.8794153332710266, "learning_rate": 8.12784801196626e-06, "loss": 2.8114, "step": 218950 }, { "epoch": 0.0129536, "grad_norm": 0.8146997094154358, "learning_rate": 8.12705697775112e-06, "loss": 2.9495, "step": 218960 }, { "epoch": 0.0129792, "grad_norm": 1.0125778913497925, "learning_rate": 8.12626595568151e-06, "loss": 3.0094, "step": 218970 }, { "epoch": 0.0130048, "grad_norm": 0.8089586496353149, "learning_rate": 8.125474945762577e-06, "loss": 2.8879, "step": 218980 }, { "epoch": 0.0130304, "grad_norm": 0.7628533244132996, "learning_rate": 8.124683947999436e-06, "loss": 2.7274, "step": 218990 }, { "epoch": 0.013056, "grad_norm": 0.9686856865882874, "learning_rate": 8.12389296239722e-06, "loss": 2.8282, "step": 219000 }, { "epoch": 0.0130816, "grad_norm": 0.6852037310600281, "learning_rate": 8.123101988961064e-06, "loss": 2.8654, "step": 219010 }, { "epoch": 0.0131072, "grad_norm": 0.730365514755249, "learning_rate": 8.12231102769609e-06, "loss": 2.9301, "step": 219020 }, { "epoch": 0.0131328, "grad_norm": 0.7841706275939941, "learning_rate": 8.12152007860743e-06, "loss": 2.8487, "step": 219030 }, { "epoch": 0.0131584, "grad_norm": 0.7805949449539185, "learning_rate": 8.120729141700214e-06, "loss": 2.7707, "step": 219040 }, { "epoch": 0.013184, "grad_norm": 0.8186023235321045, "learning_rate": 8.11993821697957e-06, "loss": 2.8676, "step": 219050 }, { "epoch": 0.0132096, "grad_norm": 0.8779324889183044, "learning_rate": 8.119147304450626e-06, "loss": 2.8371, "step": 219060 }, { "epoch": 0.0132352, "grad_norm": 0.9332177639007568, "learning_rate": 8.118356404118513e-06, "loss": 3.1618, "step": 219070 }, { "epoch": 0.0132608, "grad_norm": 0.839987576007843, "learning_rate": 8.117565515988357e-06, "loss": 2.9436, "step": 219080 }, { "epoch": 0.0132864, "grad_norm": 0.8310104012489319, "learning_rate": 8.11677464006529e-06, "loss": 2.9206, "step": 219090 }, { "epoch": 0.013312, "grad_norm": 0.7170895338058472, "learning_rate": 8.11598377635444e-06, "loss": 2.6825, "step": 219100 }, { "epoch": 0.0133376, "grad_norm": 0.8249255418777466, "learning_rate": 8.115192924860931e-06, "loss": 2.9918, "step": 219110 }, { "epoch": 0.0133632, "grad_norm": 0.9011591076850891, "learning_rate": 8.114402085589898e-06, "loss": 3.0504, "step": 219120 }, { "epoch": 0.0133888, "grad_norm": 0.7688161730766296, "learning_rate": 8.113611258546465e-06, "loss": 2.8147, "step": 219130 }, { "epoch": 0.0134144, "grad_norm": 0.741934597492218, "learning_rate": 8.112820443735761e-06, "loss": 2.776, "step": 219140 }, { "epoch": 0.01344, "grad_norm": 0.8799918293952942, "learning_rate": 8.112029641162913e-06, "loss": 2.9732, "step": 219150 }, { "epoch": 0.0134656, "grad_norm": 0.6825637817382812, "learning_rate": 8.111238850833054e-06, "loss": 2.9584, "step": 219160 }, { "epoch": 0.0134912, "grad_norm": 1.0725525617599487, "learning_rate": 8.110448072751308e-06, "loss": 2.8369, "step": 219170 }, { "epoch": 0.0135168, "grad_norm": 0.8582862615585327, "learning_rate": 8.109657306922807e-06, "loss": 3.0077, "step": 219180 }, { "epoch": 0.0135424, "grad_norm": 0.7992868423461914, "learning_rate": 8.108866553352672e-06, "loss": 2.9667, "step": 219190 }, { "epoch": 0.013568, "grad_norm": 0.778588056564331, "learning_rate": 8.108075812046037e-06, "loss": 2.9576, "step": 219200 }, { "epoch": 0.0135936, "grad_norm": 0.8420400023460388, "learning_rate": 8.107285083008028e-06, "loss": 2.9577, "step": 219210 }, { "epoch": 0.0136192, "grad_norm": 0.8312826156616211, "learning_rate": 8.106494366243771e-06, "loss": 2.9345, "step": 219220 }, { "epoch": 0.0136448, "grad_norm": 1.020593285560608, "learning_rate": 8.105703661758396e-06, "loss": 3.0839, "step": 219230 }, { "epoch": 0.0136704, "grad_norm": 1.0177009105682373, "learning_rate": 8.104912969557028e-06, "loss": 2.5524, "step": 219240 }, { "epoch": 0.013696, "grad_norm": 0.7693725824356079, "learning_rate": 8.104122289644796e-06, "loss": 2.9165, "step": 219250 }, { "epoch": 0.0137216, "grad_norm": 0.8667006492614746, "learning_rate": 8.103331622026828e-06, "loss": 2.7216, "step": 219260 }, { "epoch": 0.0137472, "grad_norm": 0.7381107807159424, "learning_rate": 8.102540966708249e-06, "loss": 2.8126, "step": 219270 }, { "epoch": 0.0137728, "grad_norm": 0.7473783493041992, "learning_rate": 8.10175032369419e-06, "loss": 3.183, "step": 219280 }, { "epoch": 0.0137984, "grad_norm": 0.7507482171058655, "learning_rate": 8.100959692989772e-06, "loss": 2.9837, "step": 219290 }, { "epoch": 0.013824, "grad_norm": 0.7302364706993103, "learning_rate": 8.100169074600129e-06, "loss": 2.9787, "step": 219300 }, { "epoch": 0.0138496, "grad_norm": 0.7914021015167236, "learning_rate": 8.099378468530382e-06, "loss": 2.8064, "step": 219310 }, { "epoch": 0.0138752, "grad_norm": 0.8953066468238831, "learning_rate": 8.098587874785658e-06, "loss": 2.9837, "step": 219320 }, { "epoch": 0.0139008, "grad_norm": 0.8360332250595093, "learning_rate": 8.09779729337109e-06, "loss": 2.7364, "step": 219330 }, { "epoch": 0.0139264, "grad_norm": 0.8942590951919556, "learning_rate": 8.097006724291802e-06, "loss": 2.9528, "step": 219340 }, { "epoch": 0.013952, "grad_norm": 1.003909707069397, "learning_rate": 8.09621616755292e-06, "loss": 2.96, "step": 219350 }, { "epoch": 0.0139776, "grad_norm": 1.5714102983474731, "learning_rate": 8.095425623159568e-06, "loss": 3.0905, "step": 219360 }, { "epoch": 0.0140032, "grad_norm": 0.9028148651123047, "learning_rate": 8.094635091116873e-06, "loss": 2.7413, "step": 219370 }, { "epoch": 0.0140288, "grad_norm": 0.766806423664093, "learning_rate": 8.093844571429969e-06, "loss": 2.9407, "step": 219380 }, { "epoch": 0.0140544, "grad_norm": 0.9119616150856018, "learning_rate": 8.093054064103971e-06, "loss": 2.609, "step": 219390 }, { "epoch": 0.01408, "grad_norm": 0.8086591958999634, "learning_rate": 8.092263569144012e-06, "loss": 2.9292, "step": 219400 }, { "epoch": 0.0141056, "grad_norm": 0.7334283590316772, "learning_rate": 8.091473086555219e-06, "loss": 2.7699, "step": 219410 }, { "epoch": 0.0141312, "grad_norm": 0.8273009061813354, "learning_rate": 8.090682616342714e-06, "loss": 2.8156, "step": 219420 }, { "epoch": 0.0141568, "grad_norm": 0.730739951133728, "learning_rate": 8.089892158511625e-06, "loss": 3.0348, "step": 219430 }, { "epoch": 0.0141824, "grad_norm": 0.8620910048484802, "learning_rate": 8.089101713067076e-06, "loss": 2.8653, "step": 219440 }, { "epoch": 0.014208, "grad_norm": 1.2760272026062012, "learning_rate": 8.088311280014198e-06, "loss": 2.8985, "step": 219450 }, { "epoch": 0.0142336, "grad_norm": 0.9460715055465698, "learning_rate": 8.087520859358112e-06, "loss": 2.945, "step": 219460 }, { "epoch": 0.0142592, "grad_norm": 1.059523105621338, "learning_rate": 8.086730451103942e-06, "loss": 2.8993, "step": 219470 }, { "epoch": 0.0142848, "grad_norm": 1.1245765686035156, "learning_rate": 8.08594005525682e-06, "loss": 3.041, "step": 219480 }, { "epoch": 0.0143104, "grad_norm": 0.8009091019630432, "learning_rate": 8.085149671821864e-06, "loss": 2.8932, "step": 219490 }, { "epoch": 0.014336, "grad_norm": 0.8830633163452148, "learning_rate": 8.084359300804208e-06, "loss": 2.9495, "step": 219500 }, { "epoch": 0.0143616, "grad_norm": 0.8265579342842102, "learning_rate": 8.08356894220897e-06, "loss": 2.7383, "step": 219510 }, { "epoch": 0.0143872, "grad_norm": 0.8380465507507324, "learning_rate": 8.082778596041279e-06, "loss": 2.962, "step": 219520 }, { "epoch": 0.0144128, "grad_norm": 0.7891857028007507, "learning_rate": 8.081988262306258e-06, "loss": 2.9039, "step": 219530 }, { "epoch": 0.0144384, "grad_norm": 0.9380042552947998, "learning_rate": 8.081197941009037e-06, "loss": 2.9102, "step": 219540 }, { "epoch": 0.014464, "grad_norm": 0.9021577835083008, "learning_rate": 8.080407632154735e-06, "loss": 2.6738, "step": 219550 }, { "epoch": 0.0144896, "grad_norm": 0.7594534158706665, "learning_rate": 8.07961733574848e-06, "loss": 2.8976, "step": 219560 }, { "epoch": 0.0145152, "grad_norm": 0.7829083800315857, "learning_rate": 8.078827051795394e-06, "loss": 2.9344, "step": 219570 }, { "epoch": 0.0145408, "grad_norm": 0.7788241505622864, "learning_rate": 8.078036780300608e-06, "loss": 2.9871, "step": 219580 }, { "epoch": 0.0145664, "grad_norm": 0.7702704668045044, "learning_rate": 8.07724652126924e-06, "loss": 3.201, "step": 219590 }, { "epoch": 0.014592, "grad_norm": 0.8579767346382141, "learning_rate": 8.076456274706416e-06, "loss": 2.6442, "step": 219600 }, { "epoch": 0.0146176, "grad_norm": 0.8083486557006836, "learning_rate": 8.075666040617266e-06, "loss": 2.8173, "step": 219610 }, { "epoch": 0.0146432, "grad_norm": 0.7609859704971313, "learning_rate": 8.074875819006908e-06, "loss": 3.0509, "step": 219620 }, { "epoch": 0.0146688, "grad_norm": 0.8013820052146912, "learning_rate": 8.074085609880468e-06, "loss": 2.8943, "step": 219630 }, { "epoch": 0.0146944, "grad_norm": 0.9434479475021362, "learning_rate": 8.073295413243071e-06, "loss": 2.693, "step": 219640 }, { "epoch": 0.01472, "grad_norm": 0.8484447598457336, "learning_rate": 8.072505229099844e-06, "loss": 3.2454, "step": 219650 }, { "epoch": 0.0147456, "grad_norm": 0.851312518119812, "learning_rate": 8.071715057455904e-06, "loss": 2.7804, "step": 219660 }, { "epoch": 0.0147712, "grad_norm": 1.0504515171051025, "learning_rate": 8.070924898316379e-06, "loss": 3.0087, "step": 219670 }, { "epoch": 0.0147968, "grad_norm": 1.23940908908844, "learning_rate": 8.070134751686396e-06, "loss": 2.9287, "step": 219680 }, { "epoch": 0.0148224, "grad_norm": 0.7736712694168091, "learning_rate": 8.069344617571076e-06, "loss": 2.9687, "step": 219690 }, { "epoch": 0.014848, "grad_norm": 0.6860988736152649, "learning_rate": 8.068554495975545e-06, "loss": 2.7295, "step": 219700 }, { "epoch": 0.0148736, "grad_norm": 0.853640079498291, "learning_rate": 8.067764386904922e-06, "loss": 2.9079, "step": 219710 }, { "epoch": 0.0148992, "grad_norm": 0.7750726938247681, "learning_rate": 8.066974290364334e-06, "loss": 3.0875, "step": 219720 }, { "epoch": 0.0149248, "grad_norm": 0.7757344841957092, "learning_rate": 8.066184206358904e-06, "loss": 2.8506, "step": 219730 }, { "epoch": 0.0149504, "grad_norm": 0.6979496479034424, "learning_rate": 8.065394134893757e-06, "loss": 3.0403, "step": 219740 }, { "epoch": 0.014976, "grad_norm": 0.8375146389007568, "learning_rate": 8.064604075974014e-06, "loss": 2.881, "step": 219750 }, { "epoch": 0.0150016, "grad_norm": 0.7435652017593384, "learning_rate": 8.063814029604797e-06, "loss": 2.7865, "step": 219760 }, { "epoch": 0.0150272, "grad_norm": 0.7362397313117981, "learning_rate": 8.063023995791233e-06, "loss": 2.9111, "step": 219770 }, { "epoch": 0.0150528, "grad_norm": 0.7427371740341187, "learning_rate": 8.062233974538446e-06, "loss": 2.8566, "step": 219780 }, { "epoch": 0.0150784, "grad_norm": 0.8029153347015381, "learning_rate": 8.061443965851554e-06, "loss": 2.9596, "step": 219790 }, { "epoch": 0.015104, "grad_norm": 0.9435024857521057, "learning_rate": 8.060653969735682e-06, "loss": 2.9489, "step": 219800 }, { "epoch": 0.0151296, "grad_norm": 0.8157020211219788, "learning_rate": 8.059863986195957e-06, "loss": 2.6753, "step": 219810 }, { "epoch": 0.0151552, "grad_norm": 0.8156512975692749, "learning_rate": 8.059074015237495e-06, "loss": 2.9314, "step": 219820 }, { "epoch": 0.0151808, "grad_norm": 1.1181482076644897, "learning_rate": 8.058284056865425e-06, "loss": 2.9643, "step": 219830 }, { "epoch": 0.0152064, "grad_norm": 0.8419070839881897, "learning_rate": 8.057494111084862e-06, "loss": 2.983, "step": 219840 }, { "epoch": 0.015232, "grad_norm": 0.9418067932128906, "learning_rate": 8.056704177900935e-06, "loss": 3.0173, "step": 219850 }, { "epoch": 0.0152576, "grad_norm": 0.8353896737098694, "learning_rate": 8.055914257318767e-06, "loss": 3.2128, "step": 219860 }, { "epoch": 0.0152832, "grad_norm": 0.7691283226013184, "learning_rate": 8.05512434934348e-06, "loss": 2.8459, "step": 219870 }, { "epoch": 0.0153088, "grad_norm": 0.9781494736671448, "learning_rate": 8.054334453980191e-06, "loss": 2.9561, "step": 219880 }, { "epoch": 0.0153344, "grad_norm": 0.736181914806366, "learning_rate": 8.053544571234028e-06, "loss": 2.7641, "step": 219890 }, { "epoch": 0.01536, "grad_norm": 0.7891079187393188, "learning_rate": 8.052754701110112e-06, "loss": 2.9299, "step": 219900 }, { "epoch": 0.0153856, "grad_norm": 0.8468713760375977, "learning_rate": 8.051964843613562e-06, "loss": 2.958, "step": 219910 }, { "epoch": 0.0154112, "grad_norm": 0.7938398718833923, "learning_rate": 8.051174998749503e-06, "loss": 2.7749, "step": 219920 }, { "epoch": 0.0154368, "grad_norm": 0.8715662956237793, "learning_rate": 8.050385166523056e-06, "loss": 2.8502, "step": 219930 }, { "epoch": 0.0154624, "grad_norm": 0.8323684930801392, "learning_rate": 8.049595346939347e-06, "loss": 2.7296, "step": 219940 }, { "epoch": 0.015488, "grad_norm": 1.2294247150421143, "learning_rate": 8.048805540003487e-06, "loss": 2.8828, "step": 219950 }, { "epoch": 0.0155136, "grad_norm": 0.7786037921905518, "learning_rate": 8.04801574572061e-06, "loss": 2.9199, "step": 219960 }, { "epoch": 0.0155392, "grad_norm": 0.7710546255111694, "learning_rate": 8.047225964095828e-06, "loss": 2.9305, "step": 219970 }, { "epoch": 0.0155648, "grad_norm": 1.4733433723449707, "learning_rate": 8.04643619513427e-06, "loss": 2.8835, "step": 219980 }, { "epoch": 0.0155904, "grad_norm": 0.9568831324577332, "learning_rate": 8.045646438841052e-06, "loss": 2.7992, "step": 219990 }, { "epoch": 0.015616, "grad_norm": 0.6996495127677917, "learning_rate": 8.0448566952213e-06, "loss": 2.8511, "step": 220000 }, { "epoch": 0.0156416, "grad_norm": 0.8925891518592834, "learning_rate": 8.044066964280132e-06, "loss": 2.9433, "step": 220010 }, { "epoch": 0.0156672, "grad_norm": 0.739810049533844, "learning_rate": 8.043277246022664e-06, "loss": 2.7514, "step": 220020 }, { "epoch": 0.0156928, "grad_norm": 0.9719194173812866, "learning_rate": 8.04248754045403e-06, "loss": 2.9928, "step": 220030 }, { "epoch": 0.0157184, "grad_norm": 0.8646379113197327, "learning_rate": 8.04169784757934e-06, "loss": 2.8334, "step": 220040 }, { "epoch": 0.015744, "grad_norm": 0.866020679473877, "learning_rate": 8.04090816740372e-06, "loss": 2.9456, "step": 220050 }, { "epoch": 0.0157696, "grad_norm": 0.7902090549468994, "learning_rate": 8.040118499932292e-06, "loss": 3.1105, "step": 220060 }, { "epoch": 0.0157952, "grad_norm": 0.7638059854507446, "learning_rate": 8.039328845170174e-06, "loss": 2.9277, "step": 220070 }, { "epoch": 0.0158208, "grad_norm": 0.8451544046401978, "learning_rate": 8.038539203122486e-06, "loss": 2.8855, "step": 220080 }, { "epoch": 0.0158464, "grad_norm": 0.8009201884269714, "learning_rate": 8.03774957379435e-06, "loss": 2.7854, "step": 220090 }, { "epoch": 0.015872, "grad_norm": 0.7657690644264221, "learning_rate": 8.036959957190889e-06, "loss": 2.6503, "step": 220100 }, { "epoch": 0.0158976, "grad_norm": 0.7492641806602478, "learning_rate": 8.03617035331722e-06, "loss": 2.8916, "step": 220110 }, { "epoch": 0.0159232, "grad_norm": 0.8657733201980591, "learning_rate": 8.035380762178462e-06, "loss": 2.8875, "step": 220120 }, { "epoch": 0.0159488, "grad_norm": 0.7613677382469177, "learning_rate": 8.034591183779739e-06, "loss": 2.9168, "step": 220130 }, { "epoch": 0.0159744, "grad_norm": 0.7211799621582031, "learning_rate": 8.033801618126171e-06, "loss": 2.7926, "step": 220140 }, { "epoch": 0.016, "grad_norm": 0.8595815300941467, "learning_rate": 8.033012065222876e-06, "loss": 2.8982, "step": 220150 }, { "epoch": 0.0160256, "grad_norm": 0.9415726661682129, "learning_rate": 8.032222525074975e-06, "loss": 3.0185, "step": 220160 }, { "epoch": 0.0160512, "grad_norm": 0.744714081287384, "learning_rate": 8.031432997687588e-06, "loss": 2.9519, "step": 220170 }, { "epoch": 0.0160768, "grad_norm": 1.0972373485565186, "learning_rate": 8.030643483065836e-06, "loss": 2.9012, "step": 220180 }, { "epoch": 0.0161024, "grad_norm": 0.8143154382705688, "learning_rate": 8.029853981214832e-06, "loss": 3.0214, "step": 220190 }, { "epoch": 0.016128, "grad_norm": 0.8335825204849243, "learning_rate": 8.029064492139707e-06, "loss": 3.1359, "step": 220200 }, { "epoch": 0.0161536, "grad_norm": 0.9453213810920715, "learning_rate": 8.028275015845573e-06, "loss": 3.0357, "step": 220210 }, { "epoch": 0.0161792, "grad_norm": 0.8972650170326233, "learning_rate": 8.027485552337552e-06, "loss": 2.866, "step": 220220 }, { "epoch": 0.0162048, "grad_norm": 0.7608214616775513, "learning_rate": 8.026696101620765e-06, "loss": 2.8263, "step": 220230 }, { "epoch": 0.0162304, "grad_norm": 0.8362168669700623, "learning_rate": 8.025906663700325e-06, "loss": 2.9211, "step": 220240 }, { "epoch": 0.016256, "grad_norm": 0.9044292569160461, "learning_rate": 8.02511723858136e-06, "loss": 2.7763, "step": 220250 }, { "epoch": 0.0162816, "grad_norm": 0.8712329268455505, "learning_rate": 8.02432782626898e-06, "loss": 3.0398, "step": 220260 }, { "epoch": 0.0163072, "grad_norm": 0.8079741597175598, "learning_rate": 8.023538426768315e-06, "loss": 3.0159, "step": 220270 }, { "epoch": 0.0163328, "grad_norm": 0.7851377129554749, "learning_rate": 8.022749040084475e-06, "loss": 2.8769, "step": 220280 }, { "epoch": 0.0163584, "grad_norm": 0.9281992316246033, "learning_rate": 8.02195966622258e-06, "loss": 2.9764, "step": 220290 }, { "epoch": 0.016384, "grad_norm": 0.7451213002204895, "learning_rate": 8.021170305187755e-06, "loss": 2.9417, "step": 220300 }, { "epoch": 0.0164096, "grad_norm": 0.9827214479446411, "learning_rate": 8.020380956985112e-06, "loss": 2.7135, "step": 220310 }, { "epoch": 0.0164352, "grad_norm": 0.789469301700592, "learning_rate": 8.019591621619773e-06, "loss": 2.772, "step": 220320 }, { "epoch": 0.0164608, "grad_norm": 0.7429611086845398, "learning_rate": 8.018802299096855e-06, "loss": 3.0077, "step": 220330 }, { "epoch": 0.0164864, "grad_norm": 1.0383042097091675, "learning_rate": 8.018012989421479e-06, "loss": 3.0122, "step": 220340 }, { "epoch": 0.016512, "grad_norm": 1.1024724245071411, "learning_rate": 8.017223692598761e-06, "loss": 3.0142, "step": 220350 }, { "epoch": 0.0165376, "grad_norm": 0.9120734333992004, "learning_rate": 8.016434408633818e-06, "loss": 2.8336, "step": 220360 }, { "epoch": 0.0165632, "grad_norm": 0.8867793679237366, "learning_rate": 8.015645137531774e-06, "loss": 2.7844, "step": 220370 }, { "epoch": 0.0165888, "grad_norm": 1.3866331577301025, "learning_rate": 8.014855879297742e-06, "loss": 2.8498, "step": 220380 }, { "epoch": 0.0166144, "grad_norm": 0.7496854066848755, "learning_rate": 8.014066633936843e-06, "loss": 2.7412, "step": 220390 }, { "epoch": 0.01664, "grad_norm": 0.764617383480072, "learning_rate": 8.013277401454197e-06, "loss": 3.0415, "step": 220400 }, { "epoch": 0.0166656, "grad_norm": 0.8833651542663574, "learning_rate": 8.012488181854916e-06, "loss": 2.9322, "step": 220410 }, { "epoch": 0.0166912, "grad_norm": 0.8022500872612, "learning_rate": 8.01169897514412e-06, "loss": 2.792, "step": 220420 }, { "epoch": 0.0167168, "grad_norm": 0.8767418265342712, "learning_rate": 8.010909781326932e-06, "loss": 2.8459, "step": 220430 }, { "epoch": 0.0167424, "grad_norm": 1.380353331565857, "learning_rate": 8.010120600408462e-06, "loss": 2.9486, "step": 220440 }, { "epoch": 0.016768, "grad_norm": 0.8300257325172424, "learning_rate": 8.00933143239383e-06, "loss": 3.0097, "step": 220450 }, { "epoch": 0.0167936, "grad_norm": 0.7397971153259277, "learning_rate": 8.008542277288156e-06, "loss": 2.7328, "step": 220460 }, { "epoch": 0.0168192, "grad_norm": 0.9709165692329407, "learning_rate": 8.007753135096559e-06, "loss": 3.0931, "step": 220470 }, { "epoch": 0.0168448, "grad_norm": 1.1194791793823242, "learning_rate": 8.006964005824149e-06, "loss": 2.7756, "step": 220480 }, { "epoch": 0.0168704, "grad_norm": 0.7787452340126038, "learning_rate": 8.00617488947605e-06, "loss": 2.9966, "step": 220490 }, { "epoch": 0.016896, "grad_norm": 0.7564992904663086, "learning_rate": 8.005385786057378e-06, "loss": 2.8279, "step": 220500 }, { "epoch": 0.0169216, "grad_norm": 0.8145260810852051, "learning_rate": 8.004596695573246e-06, "loss": 2.8925, "step": 220510 }, { "epoch": 0.0169472, "grad_norm": 0.7901429533958435, "learning_rate": 8.003807618028776e-06, "loss": 2.8828, "step": 220520 }, { "epoch": 0.0169728, "grad_norm": 2.7403481006622314, "learning_rate": 8.003018553429083e-06, "loss": 2.9584, "step": 220530 }, { "epoch": 0.0169984, "grad_norm": 0.8276891708374023, "learning_rate": 8.002229501779281e-06, "loss": 2.9418, "step": 220540 }, { "epoch": 0.017024, "grad_norm": 0.8335863351821899, "learning_rate": 8.001440463084492e-06, "loss": 2.8332, "step": 220550 }, { "epoch": 0.0170496, "grad_norm": 0.8194169998168945, "learning_rate": 8.000651437349832e-06, "loss": 3.1984, "step": 220560 }, { "epoch": 0.0170752, "grad_norm": 0.9167952537536621, "learning_rate": 7.999862424580415e-06, "loss": 2.7541, "step": 220570 }, { "epoch": 0.0171008, "grad_norm": 1.1321197748184204, "learning_rate": 7.999073424781357e-06, "loss": 2.9588, "step": 220580 }, { "epoch": 0.0171264, "grad_norm": 0.8478376269340515, "learning_rate": 7.99828443795778e-06, "loss": 2.8553, "step": 220590 }, { "epoch": 0.017152, "grad_norm": 0.8465337157249451, "learning_rate": 7.997495464114792e-06, "loss": 2.7543, "step": 220600 }, { "epoch": 0.0171776, "grad_norm": 0.9096953868865967, "learning_rate": 7.996706503257516e-06, "loss": 2.6864, "step": 220610 }, { "epoch": 0.0172032, "grad_norm": 0.7645022869110107, "learning_rate": 7.995917555391066e-06, "loss": 2.7241, "step": 220620 }, { "epoch": 0.0172288, "grad_norm": 0.8641278743743896, "learning_rate": 7.99512862052056e-06, "loss": 3.2513, "step": 220630 }, { "epoch": 0.0172544, "grad_norm": 0.7865971326828003, "learning_rate": 7.994339698651107e-06, "loss": 3.0299, "step": 220640 }, { "epoch": 0.01728, "grad_norm": 0.8540371060371399, "learning_rate": 7.993550789787832e-06, "loss": 2.6959, "step": 220650 }, { "epoch": 0.0173056, "grad_norm": 0.748484194278717, "learning_rate": 7.992761893935844e-06, "loss": 2.9941, "step": 220660 }, { "epoch": 0.0173312, "grad_norm": 0.8773643970489502, "learning_rate": 7.991973011100265e-06, "loss": 2.8104, "step": 220670 }, { "epoch": 0.0173568, "grad_norm": 0.8000795841217041, "learning_rate": 7.991184141286205e-06, "loss": 2.9021, "step": 220680 }, { "epoch": 0.0173824, "grad_norm": 0.7551770210266113, "learning_rate": 7.990395284498784e-06, "loss": 2.9654, "step": 220690 }, { "epoch": 0.017408, "grad_norm": 0.889508843421936, "learning_rate": 7.989606440743113e-06, "loss": 2.994, "step": 220700 }, { "epoch": 0.0174336, "grad_norm": 0.7287730574607849, "learning_rate": 7.988817610024308e-06, "loss": 3.0109, "step": 220710 }, { "epoch": 0.0174592, "grad_norm": 0.8769452571868896, "learning_rate": 7.988028792347491e-06, "loss": 2.9262, "step": 220720 }, { "epoch": 0.0174848, "grad_norm": 0.8371119499206543, "learning_rate": 7.98723998771777e-06, "loss": 2.9584, "step": 220730 }, { "epoch": 0.0175104, "grad_norm": 0.7687571048736572, "learning_rate": 7.986451196140264e-06, "loss": 2.9699, "step": 220740 }, { "epoch": 0.017536, "grad_norm": 0.8909671306610107, "learning_rate": 7.985662417620086e-06, "loss": 2.9038, "step": 220750 }, { "epoch": 0.0175616, "grad_norm": 0.732632040977478, "learning_rate": 7.984873652162353e-06, "loss": 3.1584, "step": 220760 }, { "epoch": 0.0175872, "grad_norm": 0.8238404989242554, "learning_rate": 7.984084899772178e-06, "loss": 2.9691, "step": 220770 }, { "epoch": 0.0176128, "grad_norm": 0.7093174457550049, "learning_rate": 7.983296160454677e-06, "loss": 2.9507, "step": 220780 }, { "epoch": 0.0176384, "grad_norm": 0.7824722528457642, "learning_rate": 7.982507434214965e-06, "loss": 3.0667, "step": 220790 }, { "epoch": 0.017664, "grad_norm": 0.8854503631591797, "learning_rate": 7.981718721058155e-06, "loss": 2.9397, "step": 220800 }, { "epoch": 0.0176896, "grad_norm": 0.7537582516670227, "learning_rate": 7.980930020989362e-06, "loss": 2.9646, "step": 220810 }, { "epoch": 0.0177152, "grad_norm": 0.6817665100097656, "learning_rate": 7.980141334013703e-06, "loss": 2.9565, "step": 220820 }, { "epoch": 0.0177408, "grad_norm": 0.924831211566925, "learning_rate": 7.979352660136292e-06, "loss": 2.8661, "step": 220830 }, { "epoch": 0.0177664, "grad_norm": 0.8362655639648438, "learning_rate": 7.978563999362238e-06, "loss": 2.983, "step": 220840 }, { "epoch": 0.017792, "grad_norm": 0.745495080947876, "learning_rate": 7.977775351696662e-06, "loss": 3.0228, "step": 220850 }, { "epoch": 0.0178176, "grad_norm": 1.2613688707351685, "learning_rate": 7.976986717144673e-06, "loss": 2.6192, "step": 220860 }, { "epoch": 0.0178432, "grad_norm": 0.8780039548873901, "learning_rate": 7.97619809571139e-06, "loss": 2.8, "step": 220870 }, { "epoch": 0.0178688, "grad_norm": 0.7136077284812927, "learning_rate": 7.975409487401924e-06, "loss": 2.8955, "step": 220880 }, { "epoch": 0.0178944, "grad_norm": 0.9337124228477478, "learning_rate": 7.974620892221387e-06, "loss": 2.8456, "step": 220890 }, { "epoch": 0.01792, "grad_norm": 0.7346137166023254, "learning_rate": 7.973832310174896e-06, "loss": 2.9291, "step": 220900 }, { "epoch": 0.0179456, "grad_norm": 0.853057861328125, "learning_rate": 7.973043741267564e-06, "loss": 2.9603, "step": 220910 }, { "epoch": 0.0179712, "grad_norm": 0.7846654057502747, "learning_rate": 7.972255185504507e-06, "loss": 2.9268, "step": 220920 }, { "epoch": 0.0179968, "grad_norm": 1.1813161373138428, "learning_rate": 7.971466642890834e-06, "loss": 2.9344, "step": 220930 }, { "epoch": 0.0180224, "grad_norm": 1.1909828186035156, "learning_rate": 7.970678113431662e-06, "loss": 2.9772, "step": 220940 }, { "epoch": 0.018048, "grad_norm": 0.7309576272964478, "learning_rate": 7.969889597132102e-06, "loss": 2.9305, "step": 220950 }, { "epoch": 0.0180736, "grad_norm": 0.7906273007392883, "learning_rate": 7.969101093997271e-06, "loss": 2.9334, "step": 220960 }, { "epoch": 0.0180992, "grad_norm": 0.7311593294143677, "learning_rate": 7.968312604032277e-06, "loss": 2.9062, "step": 220970 }, { "epoch": 0.0181248, "grad_norm": 0.7266827821731567, "learning_rate": 7.967524127242237e-06, "loss": 2.9227, "step": 220980 }, { "epoch": 0.0181504, "grad_norm": 0.7844982743263245, "learning_rate": 7.966735663632264e-06, "loss": 2.9337, "step": 220990 }, { "epoch": 0.018176, "grad_norm": 0.9756381511688232, "learning_rate": 7.965947213207467e-06, "loss": 2.8214, "step": 221000 }, { "epoch": 0.0182016, "grad_norm": 0.7906094193458557, "learning_rate": 7.965158775972963e-06, "loss": 2.9609, "step": 221010 }, { "epoch": 0.0182272, "grad_norm": 0.7432315349578857, "learning_rate": 7.964370351933864e-06, "loss": 2.8703, "step": 221020 }, { "epoch": 0.0182528, "grad_norm": 1.654206395149231, "learning_rate": 7.963581941095283e-06, "loss": 3.1363, "step": 221030 }, { "epoch": 0.0182784, "grad_norm": 0.7644127607345581, "learning_rate": 7.962793543462331e-06, "loss": 2.957, "step": 221040 }, { "epoch": 0.018304, "grad_norm": 0.905565619468689, "learning_rate": 7.96200515904012e-06, "loss": 3.1664, "step": 221050 }, { "epoch": 0.0183296, "grad_norm": 0.8400794863700867, "learning_rate": 7.961216787833763e-06, "loss": 3.0585, "step": 221060 }, { "epoch": 0.0183552, "grad_norm": 0.7937585115432739, "learning_rate": 7.960428429848374e-06, "loss": 3.0831, "step": 221070 }, { "epoch": 0.0183808, "grad_norm": 0.7623277902603149, "learning_rate": 7.959640085089066e-06, "loss": 3.0275, "step": 221080 }, { "epoch": 0.0184064, "grad_norm": 0.7732846736907959, "learning_rate": 7.958851753560953e-06, "loss": 2.8807, "step": 221090 }, { "epoch": 0.018432, "grad_norm": 0.8450660109519958, "learning_rate": 7.958063435269138e-06, "loss": 2.9317, "step": 221100 }, { "epoch": 0.0184576, "grad_norm": 0.8048205971717834, "learning_rate": 7.957275130218743e-06, "loss": 2.9161, "step": 221110 }, { "epoch": 0.0184832, "grad_norm": 0.7848640084266663, "learning_rate": 7.956486838414876e-06, "loss": 2.7832, "step": 221120 }, { "epoch": 0.0185088, "grad_norm": 0.8817443251609802, "learning_rate": 7.955698559862646e-06, "loss": 2.7053, "step": 221130 }, { "epoch": 0.0185344, "grad_norm": 0.7000821828842163, "learning_rate": 7.95491029456717e-06, "loss": 2.5872, "step": 221140 }, { "epoch": 0.01856, "grad_norm": 0.8102647066116333, "learning_rate": 7.954122042533554e-06, "loss": 2.9275, "step": 221150 }, { "epoch": 0.0185856, "grad_norm": 0.8237581253051758, "learning_rate": 7.953333803766917e-06, "loss": 2.9194, "step": 221160 }, { "epoch": 0.0186112, "grad_norm": 0.762505054473877, "learning_rate": 7.952545578272363e-06, "loss": 3.149, "step": 221170 }, { "epoch": 0.0186368, "grad_norm": 0.8300415873527527, "learning_rate": 7.951757366055007e-06, "loss": 2.893, "step": 221180 }, { "epoch": 0.0186624, "grad_norm": 0.738457441329956, "learning_rate": 7.950969167119964e-06, "loss": 2.9828, "step": 221190 }, { "epoch": 0.018688, "grad_norm": 0.7174066305160522, "learning_rate": 7.950180981472337e-06, "loss": 2.9792, "step": 221200 }, { "epoch": 0.0187136, "grad_norm": 0.7338463664054871, "learning_rate": 7.949392809117242e-06, "loss": 2.9488, "step": 221210 }, { "epoch": 0.0187392, "grad_norm": 0.8128381967544556, "learning_rate": 7.94860465005979e-06, "loss": 3.1357, "step": 221220 }, { "epoch": 0.0187648, "grad_norm": 1.0065995454788208, "learning_rate": 7.94781650430509e-06, "loss": 2.7852, "step": 221230 }, { "epoch": 0.0187904, "grad_norm": 0.7545990943908691, "learning_rate": 7.947028371858257e-06, "loss": 3.0058, "step": 221240 }, { "epoch": 0.018816, "grad_norm": 0.8213661909103394, "learning_rate": 7.9462402527244e-06, "loss": 2.9516, "step": 221250 }, { "epoch": 0.0188416, "grad_norm": 1.0148597955703735, "learning_rate": 7.94545214690863e-06, "loss": 2.8538, "step": 221260 }, { "epoch": 0.0188672, "grad_norm": 0.7054936289787292, "learning_rate": 7.944664054416053e-06, "loss": 2.8703, "step": 221270 }, { "epoch": 0.0188928, "grad_norm": 0.7484840750694275, "learning_rate": 7.943875975251786e-06, "loss": 2.956, "step": 221280 }, { "epoch": 0.0189184, "grad_norm": 0.7929757237434387, "learning_rate": 7.943087909420938e-06, "loss": 2.7865, "step": 221290 }, { "epoch": 0.018944, "grad_norm": 0.8110424876213074, "learning_rate": 7.942299856928617e-06, "loss": 3.0216, "step": 221300 }, { "epoch": 0.0189696, "grad_norm": 0.8785050511360168, "learning_rate": 7.941511817779933e-06, "loss": 2.9338, "step": 221310 }, { "epoch": 0.0189952, "grad_norm": 0.8921856880187988, "learning_rate": 7.94072379198e-06, "loss": 2.9149, "step": 221320 }, { "epoch": 0.0190208, "grad_norm": 0.8885855674743652, "learning_rate": 7.939935779533924e-06, "loss": 2.9272, "step": 221330 }, { "epoch": 0.0190464, "grad_norm": 0.8557040691375732, "learning_rate": 7.939147780446818e-06, "loss": 2.8167, "step": 221340 }, { "epoch": 0.019072, "grad_norm": 0.7942785620689392, "learning_rate": 7.938359794723792e-06, "loss": 2.8883, "step": 221350 }, { "epoch": 0.0190976, "grad_norm": 0.7673454284667969, "learning_rate": 7.937571822369954e-06, "loss": 2.8634, "step": 221360 }, { "epoch": 0.0191232, "grad_norm": 0.7496093511581421, "learning_rate": 7.936783863390414e-06, "loss": 2.8585, "step": 221370 }, { "epoch": 0.0191488, "grad_norm": 0.8951230645179749, "learning_rate": 7.935995917790283e-06, "loss": 2.7533, "step": 221380 }, { "epoch": 0.0191744, "grad_norm": 0.914581298828125, "learning_rate": 7.93520798557467e-06, "loss": 2.9188, "step": 221390 }, { "epoch": 0.0192, "grad_norm": 0.9576253294944763, "learning_rate": 7.934420066748684e-06, "loss": 2.7775, "step": 221400 }, { "epoch": 0.0192256, "grad_norm": 0.7642064690589905, "learning_rate": 7.933632161317432e-06, "loss": 2.697, "step": 221410 }, { "epoch": 0.0192512, "grad_norm": 0.9758651852607727, "learning_rate": 7.932844269286028e-06, "loss": 2.811, "step": 221420 }, { "epoch": 0.0192768, "grad_norm": 0.884868860244751, "learning_rate": 7.93205639065958e-06, "loss": 2.774, "step": 221430 }, { "epoch": 0.0193024, "grad_norm": 0.8076619505882263, "learning_rate": 7.931268525443198e-06, "loss": 3.1369, "step": 221440 }, { "epoch": 0.019328, "grad_norm": 0.7795596122741699, "learning_rate": 7.930480673641991e-06, "loss": 2.979, "step": 221450 }, { "epoch": 0.0193536, "grad_norm": 0.8250590562820435, "learning_rate": 7.929692835261064e-06, "loss": 2.9827, "step": 221460 }, { "epoch": 0.0193792, "grad_norm": 0.7771921157836914, "learning_rate": 7.92890501030553e-06, "loss": 2.9372, "step": 221470 }, { "epoch": 0.0194048, "grad_norm": 0.9516621232032776, "learning_rate": 7.928117198780497e-06, "loss": 2.6345, "step": 221480 }, { "epoch": 0.0194304, "grad_norm": 0.7383352518081665, "learning_rate": 7.927329400691072e-06, "loss": 2.6731, "step": 221490 }, { "epoch": 0.019456, "grad_norm": 0.8249710202217102, "learning_rate": 7.926541616042366e-06, "loss": 2.6394, "step": 221500 }, { "epoch": 0.0194816, "grad_norm": 0.9349868893623352, "learning_rate": 7.925753844839485e-06, "loss": 2.6485, "step": 221510 }, { "epoch": 0.0195072, "grad_norm": 0.8505820035934448, "learning_rate": 7.924966087087542e-06, "loss": 2.6127, "step": 221520 }, { "epoch": 0.0195328, "grad_norm": 0.7083699703216553, "learning_rate": 7.92417834279164e-06, "loss": 2.6764, "step": 221530 }, { "epoch": 0.0195584, "grad_norm": 1.343182921409607, "learning_rate": 7.92339061195689e-06, "loss": 2.6553, "step": 221540 }, { "epoch": 0.019584, "grad_norm": 0.7619633078575134, "learning_rate": 7.9226028945884e-06, "loss": 2.6071, "step": 221550 }, { "epoch": 0.0196096, "grad_norm": 0.7801735401153564, "learning_rate": 7.92181519069128e-06, "loss": 2.7106, "step": 221560 }, { "epoch": 0.0196352, "grad_norm": 0.8090248107910156, "learning_rate": 7.921027500270635e-06, "loss": 2.5714, "step": 221570 }, { "epoch": 0.0196608, "grad_norm": 0.7277584671974182, "learning_rate": 7.92023982333157e-06, "loss": 2.6036, "step": 221580 }, { "epoch": 0.0196864, "grad_norm": 0.7717956900596619, "learning_rate": 7.919452159879202e-06, "loss": 2.635, "step": 221590 }, { "epoch": 0.019712, "grad_norm": 0.7800621390342712, "learning_rate": 7.918664509918631e-06, "loss": 2.8144, "step": 221600 }, { "epoch": 0.0197376, "grad_norm": 0.8258799910545349, "learning_rate": 7.91787687345497e-06, "loss": 2.6254, "step": 221610 }, { "epoch": 0.0197632, "grad_norm": 0.8748639822006226, "learning_rate": 7.917089250493324e-06, "loss": 2.8056, "step": 221620 }, { "epoch": 0.0197888, "grad_norm": 0.8247752785682678, "learning_rate": 7.916301641038799e-06, "loss": 3.0321, "step": 221630 }, { "epoch": 0.0198144, "grad_norm": 0.8508428335189819, "learning_rate": 7.915514045096506e-06, "loss": 2.7762, "step": 221640 }, { "epoch": 0.01984, "grad_norm": 0.9093145132064819, "learning_rate": 7.914726462671551e-06, "loss": 2.742, "step": 221650 }, { "epoch": 0.0198656, "grad_norm": 0.7622162103652954, "learning_rate": 7.913938893769038e-06, "loss": 2.8901, "step": 221660 }, { "epoch": 0.0198912, "grad_norm": 1.103139877319336, "learning_rate": 7.913151338394077e-06, "loss": 2.4996, "step": 221670 }, { "epoch": 0.0199168, "grad_norm": 0.7916914820671082, "learning_rate": 7.912363796551777e-06, "loss": 2.5399, "step": 221680 }, { "epoch": 0.0199424, "grad_norm": 0.773281455039978, "learning_rate": 7.911576268247242e-06, "loss": 2.8112, "step": 221690 }, { "epoch": 0.019968, "grad_norm": 0.7974185943603516, "learning_rate": 7.91078875348558e-06, "loss": 2.6834, "step": 221700 }, { "epoch": 0.0199936, "grad_norm": 0.9477707147598267, "learning_rate": 7.910001252271895e-06, "loss": 3.0326, "step": 221710 }, { "epoch": 0.0200192, "grad_norm": 0.928239107131958, "learning_rate": 7.909213764611303e-06, "loss": 3.02, "step": 221720 }, { "epoch": 0.0200448, "grad_norm": 0.737207293510437, "learning_rate": 7.908426290508899e-06, "loss": 3.0352, "step": 221730 }, { "epoch": 0.0200704, "grad_norm": 1.3111464977264404, "learning_rate": 7.907638829969795e-06, "loss": 2.771, "step": 221740 }, { "epoch": 0.020096, "grad_norm": 0.7517765760421753, "learning_rate": 7.906851382999096e-06, "loss": 2.7111, "step": 221750 }, { "epoch": 0.0201216, "grad_norm": 1.0844449996948242, "learning_rate": 7.90606394960191e-06, "loss": 2.7113, "step": 221760 }, { "epoch": 0.0201472, "grad_norm": 0.8361253142356873, "learning_rate": 7.905276529783343e-06, "loss": 2.7132, "step": 221770 }, { "epoch": 0.0201728, "grad_norm": 0.7550765872001648, "learning_rate": 7.904489123548504e-06, "loss": 2.6073, "step": 221780 }, { "epoch": 0.0201984, "grad_norm": 0.7657467722892761, "learning_rate": 7.903701730902494e-06, "loss": 2.7795, "step": 221790 }, { "epoch": 0.020224, "grad_norm": 0.8821398019790649, "learning_rate": 7.902914351850419e-06, "loss": 2.7999, "step": 221800 }, { "epoch": 0.0202496, "grad_norm": 0.8066459894180298, "learning_rate": 7.902126986397392e-06, "loss": 2.5361, "step": 221810 }, { "epoch": 0.0202752, "grad_norm": 0.8474935293197632, "learning_rate": 7.90133963454851e-06, "loss": 2.4944, "step": 221820 }, { "epoch": 0.0203008, "grad_norm": 0.694094717502594, "learning_rate": 7.900552296308883e-06, "loss": 2.5663, "step": 221830 }, { "epoch": 0.0203264, "grad_norm": 0.8005003929138184, "learning_rate": 7.899764971683617e-06, "loss": 2.7731, "step": 221840 }, { "epoch": 0.020352, "grad_norm": 0.7923669219017029, "learning_rate": 7.898977660677819e-06, "loss": 2.7733, "step": 221850 }, { "epoch": 0.0203776, "grad_norm": 0.7767762541770935, "learning_rate": 7.898190363296591e-06, "loss": 2.8123, "step": 221860 }, { "epoch": 0.0204032, "grad_norm": 0.9161626100540161, "learning_rate": 7.897403079545038e-06, "loss": 2.4613, "step": 221870 }, { "epoch": 0.0204288, "grad_norm": 0.8273918032646179, "learning_rate": 7.896615809428272e-06, "loss": 2.895, "step": 221880 }, { "epoch": 0.0204544, "grad_norm": 0.7662951946258545, "learning_rate": 7.895828552951388e-06, "loss": 2.6288, "step": 221890 }, { "epoch": 0.02048, "grad_norm": 0.7395727634429932, "learning_rate": 7.8950413101195e-06, "loss": 2.7097, "step": 221900 }, { "epoch": 0.0205056, "grad_norm": 0.7900673747062683, "learning_rate": 7.894254080937707e-06, "loss": 2.6879, "step": 221910 }, { "epoch": 0.0205312, "grad_norm": 0.8228249549865723, "learning_rate": 7.893466865411121e-06, "loss": 2.7493, "step": 221920 }, { "epoch": 0.0205568, "grad_norm": 0.7559446692466736, "learning_rate": 7.892679663544836e-06, "loss": 2.6159, "step": 221930 }, { "epoch": 0.0205824, "grad_norm": 0.8724362254142761, "learning_rate": 7.891892475343968e-06, "loss": 2.4851, "step": 221940 }, { "epoch": 0.020608, "grad_norm": 0.816333532333374, "learning_rate": 7.891105300813616e-06, "loss": 2.9146, "step": 221950 }, { "epoch": 0.0206336, "grad_norm": 0.8332330584526062, "learning_rate": 7.890318139958885e-06, "loss": 2.6495, "step": 221960 }, { "epoch": 0.0206592, "grad_norm": 0.804242730140686, "learning_rate": 7.889530992784881e-06, "loss": 2.5467, "step": 221970 }, { "epoch": 0.0206848, "grad_norm": 0.7922579646110535, "learning_rate": 7.88874385929671e-06, "loss": 2.9518, "step": 221980 }, { "epoch": 0.0207104, "grad_norm": 0.9561498165130615, "learning_rate": 7.88795673949947e-06, "loss": 2.7463, "step": 221990 }, { "epoch": 0.020736, "grad_norm": 0.7891426682472229, "learning_rate": 7.887169633398271e-06, "loss": 2.7649, "step": 222000 }, { "epoch": 0.0207616, "grad_norm": 0.8284196257591248, "learning_rate": 7.886382540998216e-06, "loss": 2.5367, "step": 222010 }, { "epoch": 0.0207872, "grad_norm": 0.7200817465782166, "learning_rate": 7.885595462304407e-06, "loss": 2.5372, "step": 222020 }, { "epoch": 0.0208128, "grad_norm": 0.9357978701591492, "learning_rate": 7.88480839732195e-06, "loss": 2.623, "step": 222030 }, { "epoch": 0.0208384, "grad_norm": 0.7436351776123047, "learning_rate": 7.884021346055948e-06, "loss": 2.5653, "step": 222040 }, { "epoch": 0.020864, "grad_norm": 0.8094027638435364, "learning_rate": 7.883234308511508e-06, "loss": 2.7653, "step": 222050 }, { "epoch": 0.0208896, "grad_norm": 0.8664580583572388, "learning_rate": 7.882447284693728e-06, "loss": 2.1717, "step": 222060 }, { "epoch": 0.0209152, "grad_norm": 1.055889368057251, "learning_rate": 7.881660274607716e-06, "loss": 2.5639, "step": 222070 }, { "epoch": 0.0209408, "grad_norm": 0.8444912433624268, "learning_rate": 7.880873278258574e-06, "loss": 2.8459, "step": 222080 }, { "epoch": 0.0209664, "grad_norm": 0.9935893416404724, "learning_rate": 7.880086295651405e-06, "loss": 2.8035, "step": 222090 }, { "epoch": 0.020992, "grad_norm": 0.790205717086792, "learning_rate": 7.87929932679131e-06, "loss": 2.718, "step": 222100 }, { "epoch": 2.56e-05, "grad_norm": 0.8549762964248657, "learning_rate": 7.878512371683397e-06, "loss": 2.1331, "step": 222110 }, { "epoch": 5.12e-05, "grad_norm": 0.7730506062507629, "learning_rate": 7.877725430332769e-06, "loss": 2.5609, "step": 222120 }, { "epoch": 7.68e-05, "grad_norm": 0.7926088571548462, "learning_rate": 7.876938502744527e-06, "loss": 2.7032, "step": 222130 }, { "epoch": 0.0001024, "grad_norm": 0.8730273246765137, "learning_rate": 7.876151588923777e-06, "loss": 2.7286, "step": 222140 }, { "epoch": 0.000128, "grad_norm": 0.7947468757629395, "learning_rate": 7.875364688875617e-06, "loss": 2.6391, "step": 222150 }, { "epoch": 0.0001536, "grad_norm": 0.842170238494873, "learning_rate": 7.87457780260515e-06, "loss": 2.8119, "step": 222160 }, { "epoch": 0.0001792, "grad_norm": 0.9070627093315125, "learning_rate": 7.873790930117484e-06, "loss": 2.6052, "step": 222170 }, { "epoch": 0.0002048, "grad_norm": 0.7928528189659119, "learning_rate": 7.87300407141772e-06, "loss": 2.7241, "step": 222180 }, { "epoch": 0.0002304, "grad_norm": 0.9697148203849792, "learning_rate": 7.87221722651096e-06, "loss": 2.8288, "step": 222190 }, { "epoch": 0.000256, "grad_norm": 0.7815393805503845, "learning_rate": 7.871430395402303e-06, "loss": 2.7211, "step": 222200 }, { "epoch": 0.0002816, "grad_norm": 0.7971534729003906, "learning_rate": 7.870643578096856e-06, "loss": 2.8299, "step": 222210 }, { "epoch": 0.0003072, "grad_norm": 0.9522470831871033, "learning_rate": 7.869856774599716e-06, "loss": 2.7146, "step": 222220 }, { "epoch": 0.0003328, "grad_norm": 0.8187173008918762, "learning_rate": 7.869069984915992e-06, "loss": 2.7709, "step": 222230 }, { "epoch": 0.0003584, "grad_norm": 0.8655644655227661, "learning_rate": 7.86828320905078e-06, "loss": 2.8926, "step": 222240 }, { "epoch": 0.000384, "grad_norm": 0.8268070816993713, "learning_rate": 7.867496447009188e-06, "loss": 2.6786, "step": 222250 }, { "epoch": 0.0004096, "grad_norm": 0.7140868902206421, "learning_rate": 7.866709698796313e-06, "loss": 2.5655, "step": 222260 }, { "epoch": 0.0004352, "grad_norm": 0.795754611492157, "learning_rate": 7.865922964417259e-06, "loss": 2.7109, "step": 222270 }, { "epoch": 0.0004608, "grad_norm": 0.8646671175956726, "learning_rate": 7.865136243877125e-06, "loss": 2.6541, "step": 222280 }, { "epoch": 0.0004864, "grad_norm": 0.7947627305984497, "learning_rate": 7.864349537181017e-06, "loss": 2.6836, "step": 222290 }, { "epoch": 0.000512, "grad_norm": 0.834888756275177, "learning_rate": 7.863562844334036e-06, "loss": 2.6444, "step": 222300 }, { "epoch": 0.0005376, "grad_norm": 0.8105637431144714, "learning_rate": 7.862776165341282e-06, "loss": 2.5737, "step": 222310 }, { "epoch": 0.0005632, "grad_norm": 0.8033678531646729, "learning_rate": 7.861989500207855e-06, "loss": 2.5475, "step": 222320 }, { "epoch": 0.0005888, "grad_norm": 0.8575671315193176, "learning_rate": 7.861202848938858e-06, "loss": 2.6179, "step": 222330 }, { "epoch": 0.0006144, "grad_norm": 0.9733109474182129, "learning_rate": 7.860416211539395e-06, "loss": 2.8918, "step": 222340 }, { "epoch": 0.00064, "grad_norm": 0.7955946326255798, "learning_rate": 7.85962958801456e-06, "loss": 2.7773, "step": 222350 }, { "epoch": 0.0006656, "grad_norm": 0.8505295515060425, "learning_rate": 7.858842978369461e-06, "loss": 3.1417, "step": 222360 }, { "epoch": 0.0006912, "grad_norm": 0.8903306126594543, "learning_rate": 7.858056382609197e-06, "loss": 2.7619, "step": 222370 }, { "epoch": 0.0007168, "grad_norm": 0.7535945773124695, "learning_rate": 7.857269800738867e-06, "loss": 2.7244, "step": 222380 }, { "epoch": 0.0007424, "grad_norm": 1.0656720399856567, "learning_rate": 7.856483232763572e-06, "loss": 2.6971, "step": 222390 }, { "epoch": 0.000768, "grad_norm": 0.8596741557121277, "learning_rate": 7.855696678688414e-06, "loss": 2.7325, "step": 222400 }, { "epoch": 0.0007936, "grad_norm": 0.9510244727134705, "learning_rate": 7.854910138518495e-06, "loss": 2.5169, "step": 222410 }, { "epoch": 0.0008192, "grad_norm": 0.7889576554298401, "learning_rate": 7.854123612258912e-06, "loss": 2.8523, "step": 222420 }, { "epoch": 0.0008448, "grad_norm": 0.7753627300262451, "learning_rate": 7.853337099914766e-06, "loss": 2.7218, "step": 222430 }, { "epoch": 0.0008704, "grad_norm": 0.8142701983451843, "learning_rate": 7.852550601491159e-06, "loss": 2.6673, "step": 222440 }, { "epoch": 0.000896, "grad_norm": 0.7915888428688049, "learning_rate": 7.851764116993188e-06, "loss": 2.4813, "step": 222450 }, { "epoch": 0.0009216, "grad_norm": 0.7172534465789795, "learning_rate": 7.850977646425959e-06, "loss": 2.7286, "step": 222460 }, { "epoch": 0.0009472, "grad_norm": 1.001006841659546, "learning_rate": 7.85019118979457e-06, "loss": 2.7682, "step": 222470 }, { "epoch": 0.0009728, "grad_norm": 0.9138456583023071, "learning_rate": 7.849404747104117e-06, "loss": 2.6539, "step": 222480 }, { "epoch": 0.0009984, "grad_norm": 0.9170445203781128, "learning_rate": 7.848618318359704e-06, "loss": 2.7268, "step": 222490 }, { "epoch": 0.001024, "grad_norm": 0.8882749080657959, "learning_rate": 7.847831903566431e-06, "loss": 2.6228, "step": 222500 }, { "epoch": 0.0010496, "grad_norm": 0.8817253708839417, "learning_rate": 7.847045502729395e-06, "loss": 2.8421, "step": 222510 }, { "epoch": 0.0010752, "grad_norm": 0.8467176556587219, "learning_rate": 7.846259115853696e-06, "loss": 2.696, "step": 222520 }, { "epoch": 0.0011008, "grad_norm": 0.8123210668563843, "learning_rate": 7.845472742944433e-06, "loss": 2.5637, "step": 222530 }, { "epoch": 0.0011264, "grad_norm": 0.8596957325935364, "learning_rate": 7.844686384006711e-06, "loss": 2.8572, "step": 222540 }, { "epoch": 0.001152, "grad_norm": 0.9733255505561829, "learning_rate": 7.843900039045621e-06, "loss": 2.9006, "step": 222550 }, { "epoch": 0.0011776, "grad_norm": 0.7541818618774414, "learning_rate": 7.843113708066267e-06, "loss": 2.7058, "step": 222560 }, { "epoch": 0.0012032, "grad_norm": 0.8306242227554321, "learning_rate": 7.84232739107375e-06, "loss": 2.957, "step": 222570 }, { "epoch": 0.0012288, "grad_norm": 0.7468985319137573, "learning_rate": 7.841541088073163e-06, "loss": 2.7023, "step": 222580 }, { "epoch": 0.0012544, "grad_norm": 0.7786072492599487, "learning_rate": 7.84075479906961e-06, "loss": 2.8622, "step": 222590 }, { "epoch": 0.00128, "grad_norm": 0.8236788511276245, "learning_rate": 7.839968524068187e-06, "loss": 2.8209, "step": 222600 }, { "epoch": 0.0013056, "grad_norm": 1.0643340349197388, "learning_rate": 7.839182263073997e-06, "loss": 2.671, "step": 222610 }, { "epoch": 0.0013312, "grad_norm": 2.522200345993042, "learning_rate": 7.838396016092134e-06, "loss": 2.6858, "step": 222620 }, { "epoch": 0.0013568, "grad_norm": 0.7896426320075989, "learning_rate": 7.837609783127695e-06, "loss": 2.6829, "step": 222630 }, { "epoch": 0.0013824, "grad_norm": 1.03038489818573, "learning_rate": 7.836823564185785e-06, "loss": 2.8022, "step": 222640 }, { "epoch": 0.001408, "grad_norm": 0.8774058222770691, "learning_rate": 7.836037359271498e-06, "loss": 2.7964, "step": 222650 }, { "epoch": 0.0014336, "grad_norm": 0.8374051451683044, "learning_rate": 7.835251168389935e-06, "loss": 2.6706, "step": 222660 }, { "epoch": 0.0014592, "grad_norm": 0.9524774551391602, "learning_rate": 7.834464991546193e-06, "loss": 2.5112, "step": 222670 }, { "epoch": 0.0014848, "grad_norm": 0.927131175994873, "learning_rate": 7.83367882874537e-06, "loss": 2.579, "step": 222680 }, { "epoch": 0.0015104, "grad_norm": 0.8078533411026001, "learning_rate": 7.832892679992564e-06, "loss": 2.6748, "step": 222690 }, { "epoch": 0.001536, "grad_norm": 0.7545738816261292, "learning_rate": 7.832106545292873e-06, "loss": 2.7536, "step": 222700 }, { "epoch": 0.0015616, "grad_norm": 0.8462046980857849, "learning_rate": 7.831320424651395e-06, "loss": 2.9575, "step": 222710 }, { "epoch": 0.0015872, "grad_norm": 0.777372419834137, "learning_rate": 7.830534318073227e-06, "loss": 2.6883, "step": 222720 }, { "epoch": 0.0016128, "grad_norm": 0.9091501832008362, "learning_rate": 7.829748225563466e-06, "loss": 2.7058, "step": 222730 }, { "epoch": 0.0016384, "grad_norm": 1.160391926765442, "learning_rate": 7.828962147127216e-06, "loss": 2.8143, "step": 222740 }, { "epoch": 0.001664, "grad_norm": 0.7897612452507019, "learning_rate": 7.828176082769566e-06, "loss": 2.7557, "step": 222750 }, { "epoch": 0.0016896, "grad_norm": 0.7460994720458984, "learning_rate": 7.827390032495616e-06, "loss": 2.7639, "step": 222760 }, { "epoch": 0.0017152, "grad_norm": 0.8703435063362122, "learning_rate": 7.826603996310467e-06, "loss": 2.7708, "step": 222770 }, { "epoch": 0.0017408, "grad_norm": 0.8389007449150085, "learning_rate": 7.82581797421921e-06, "loss": 2.6876, "step": 222780 }, { "epoch": 0.0017664, "grad_norm": 0.9419337511062622, "learning_rate": 7.825031966226948e-06, "loss": 2.5991, "step": 222790 }, { "epoch": 0.001792, "grad_norm": 0.7850248217582703, "learning_rate": 7.824245972338773e-06, "loss": 2.5822, "step": 222800 }, { "epoch": 0.0018176, "grad_norm": 0.7902411818504333, "learning_rate": 7.823459992559786e-06, "loss": 2.6723, "step": 222810 }, { "epoch": 0.0018432, "grad_norm": 0.9222451448440552, "learning_rate": 7.822674026895083e-06, "loss": 2.6286, "step": 222820 }, { "epoch": 0.0018688, "grad_norm": 0.7863141298294067, "learning_rate": 7.821888075349762e-06, "loss": 2.8433, "step": 222830 }, { "epoch": 0.0018944, "grad_norm": 0.9351059198379517, "learning_rate": 7.821102137928915e-06, "loss": 2.9525, "step": 222840 }, { "epoch": 0.00192, "grad_norm": 0.9714484214782715, "learning_rate": 7.820316214637643e-06, "loss": 2.6242, "step": 222850 }, { "epoch": 0.0019456, "grad_norm": 0.740886926651001, "learning_rate": 7.81953030548104e-06, "loss": 2.7948, "step": 222860 }, { "epoch": 0.0019712, "grad_norm": 0.7192357778549194, "learning_rate": 7.818744410464206e-06, "loss": 2.3567, "step": 222870 }, { "epoch": 0.0019968, "grad_norm": 0.9027471542358398, "learning_rate": 7.817958529592233e-06, "loss": 2.8912, "step": 222880 }, { "epoch": 0.0020224, "grad_norm": 0.9359740018844604, "learning_rate": 7.81717266287022e-06, "loss": 2.8154, "step": 222890 }, { "epoch": 0.002048, "grad_norm": 0.8521834015846252, "learning_rate": 7.816386810303263e-06, "loss": 2.5465, "step": 222900 }, { "epoch": 0.0020736, "grad_norm": 0.8683561682701111, "learning_rate": 7.815600971896454e-06, "loss": 2.7311, "step": 222910 }, { "epoch": 0.0020992, "grad_norm": 0.7769398093223572, "learning_rate": 7.814815147654895e-06, "loss": 2.5661, "step": 222920 }, { "epoch": 0.0021248, "grad_norm": 0.8309851884841919, "learning_rate": 7.814029337583676e-06, "loss": 2.6528, "step": 222930 }, { "epoch": 0.0021504, "grad_norm": 0.96876060962677, "learning_rate": 7.8132435416879e-06, "loss": 2.6162, "step": 222940 }, { "epoch": 0.002176, "grad_norm": 0.7860285043716431, "learning_rate": 7.812457759972655e-06, "loss": 2.8063, "step": 222950 }, { "epoch": 0.0022016, "grad_norm": 0.7642021179199219, "learning_rate": 7.81167199244304e-06, "loss": 2.7459, "step": 222960 }, { "epoch": 0.0022272, "grad_norm": 0.9150305390357971, "learning_rate": 7.81088623910415e-06, "loss": 2.9316, "step": 222970 }, { "epoch": 0.0022528, "grad_norm": 0.8806827068328857, "learning_rate": 7.810100499961082e-06, "loss": 2.5549, "step": 222980 }, { "epoch": 0.0022784, "grad_norm": 0.975161612033844, "learning_rate": 7.809314775018933e-06, "loss": 2.709, "step": 222990 }, { "epoch": 0.002304, "grad_norm": 0.7912049889564514, "learning_rate": 7.808529064282793e-06, "loss": 2.8905, "step": 223000 }, { "epoch": 0.0023296, "grad_norm": 0.870470404624939, "learning_rate": 7.80774336775776e-06, "loss": 2.3357, "step": 223010 }, { "epoch": 0.0023552, "grad_norm": 0.7947487235069275, "learning_rate": 7.806957685448926e-06, "loss": 2.5638, "step": 223020 }, { "epoch": 0.0023808, "grad_norm": 0.788760244846344, "learning_rate": 7.806172017361394e-06, "loss": 2.8591, "step": 223030 }, { "epoch": 0.0024064, "grad_norm": 0.7941039204597473, "learning_rate": 7.805386363500249e-06, "loss": 2.6852, "step": 223040 }, { "epoch": 0.002432, "grad_norm": 0.747180163860321, "learning_rate": 7.804600723870592e-06, "loss": 2.6568, "step": 223050 }, { "epoch": 0.0024576, "grad_norm": 0.9635581970214844, "learning_rate": 7.803815098477515e-06, "loss": 2.7487, "step": 223060 }, { "epoch": 0.0024832, "grad_norm": 1.6030220985412598, "learning_rate": 7.803029487326114e-06, "loss": 3.0224, "step": 223070 }, { "epoch": 0.0025088, "grad_norm": 0.8443633913993835, "learning_rate": 7.802243890421484e-06, "loss": 2.9466, "step": 223080 }, { "epoch": 0.0025344, "grad_norm": 0.8811989426612854, "learning_rate": 7.801458307768717e-06, "loss": 2.9933, "step": 223090 }, { "epoch": 0.00256, "grad_norm": 0.7967018485069275, "learning_rate": 7.80067273937291e-06, "loss": 2.8657, "step": 223100 }, { "epoch": 0.0025856, "grad_norm": 0.8243731260299683, "learning_rate": 7.799887185239154e-06, "loss": 2.763, "step": 223110 }, { "epoch": 0.0026112, "grad_norm": 0.9075321555137634, "learning_rate": 7.799101645372545e-06, "loss": 2.9714, "step": 223120 }, { "epoch": 0.0026368, "grad_norm": 0.7416611909866333, "learning_rate": 7.798316119778177e-06, "loss": 2.7559, "step": 223130 }, { "epoch": 0.0026624, "grad_norm": 0.913298487663269, "learning_rate": 7.797530608461145e-06, "loss": 2.9119, "step": 223140 }, { "epoch": 0.002688, "grad_norm": 0.8844023942947388, "learning_rate": 7.796745111426536e-06, "loss": 2.9351, "step": 223150 }, { "epoch": 0.0027136, "grad_norm": 0.8712520003318787, "learning_rate": 7.795959628679455e-06, "loss": 2.7835, "step": 223160 }, { "epoch": 0.0027392, "grad_norm": 0.843043327331543, "learning_rate": 7.79517416022499e-06, "loss": 2.8957, "step": 223170 }, { "epoch": 0.0027648, "grad_norm": 0.7356094717979431, "learning_rate": 7.794388706068232e-06, "loss": 2.9978, "step": 223180 }, { "epoch": 0.0027904, "grad_norm": 0.940481424331665, "learning_rate": 7.79360326621428e-06, "loss": 2.9274, "step": 223190 }, { "epoch": 0.002816, "grad_norm": 0.7476873397827148, "learning_rate": 7.792817840668224e-06, "loss": 2.6504, "step": 223200 }, { "epoch": 0.0028416, "grad_norm": 0.7521158456802368, "learning_rate": 7.792032429435157e-06, "loss": 2.8133, "step": 223210 }, { "epoch": 0.0028672, "grad_norm": 0.8667768836021423, "learning_rate": 7.791247032520171e-06, "loss": 2.9819, "step": 223220 }, { "epoch": 0.0028928, "grad_norm": 0.8527265787124634, "learning_rate": 7.790461649928365e-06, "loss": 2.8731, "step": 223230 }, { "epoch": 0.0029184, "grad_norm": 0.7506497502326965, "learning_rate": 7.789676281664827e-06, "loss": 2.8635, "step": 223240 }, { "epoch": 0.002944, "grad_norm": 0.7405410408973694, "learning_rate": 7.788890927734649e-06, "loss": 2.8704, "step": 223250 }, { "epoch": 0.0029696, "grad_norm": 0.8442150354385376, "learning_rate": 7.788105588142925e-06, "loss": 2.9159, "step": 223260 }, { "epoch": 0.0029952, "grad_norm": 0.7603812217712402, "learning_rate": 7.787320262894751e-06, "loss": 2.9174, "step": 223270 }, { "epoch": 0.0030208, "grad_norm": 0.9427676200866699, "learning_rate": 7.786534951995216e-06, "loss": 2.8184, "step": 223280 }, { "epoch": 0.0030464, "grad_norm": 0.721085250377655, "learning_rate": 7.785749655449412e-06, "loss": 2.9201, "step": 223290 }, { "epoch": 0.003072, "grad_norm": 0.9136326909065247, "learning_rate": 7.784964373262436e-06, "loss": 2.9714, "step": 223300 }, { "epoch": 0.0030976, "grad_norm": 0.7535620927810669, "learning_rate": 7.784179105439376e-06, "loss": 2.8619, "step": 223310 }, { "epoch": 0.0031232, "grad_norm": 0.9174911975860596, "learning_rate": 7.783393851985322e-06, "loss": 2.6587, "step": 223320 }, { "epoch": 0.0031488, "grad_norm": 0.8746079802513123, "learning_rate": 7.782608612905374e-06, "loss": 3.0739, "step": 223330 }, { "epoch": 0.0031744, "grad_norm": 0.7426153421401978, "learning_rate": 7.781823388204617e-06, "loss": 2.9118, "step": 223340 }, { "epoch": 0.0032, "grad_norm": 0.9331109523773193, "learning_rate": 7.781038177888148e-06, "loss": 2.8187, "step": 223350 }, { "epoch": 0.0032256, "grad_norm": 0.8177664875984192, "learning_rate": 7.780252981961058e-06, "loss": 2.9562, "step": 223360 }, { "epoch": 0.0032512, "grad_norm": 0.7496710419654846, "learning_rate": 7.779467800428436e-06, "loss": 2.8829, "step": 223370 }, { "epoch": 0.0032768, "grad_norm": 0.7729038596153259, "learning_rate": 7.778682633295374e-06, "loss": 2.7595, "step": 223380 }, { "epoch": 0.0033024, "grad_norm": 0.845610499382019, "learning_rate": 7.777897480566967e-06, "loss": 2.8522, "step": 223390 }, { "epoch": 0.003328, "grad_norm": 0.8821427226066589, "learning_rate": 7.777112342248302e-06, "loss": 2.9409, "step": 223400 }, { "epoch": 0.0033536, "grad_norm": 0.7588126063346863, "learning_rate": 7.776327218344475e-06, "loss": 2.7356, "step": 223410 }, { "epoch": 0.0033792, "grad_norm": 0.7448330521583557, "learning_rate": 7.775542108860573e-06, "loss": 3.051, "step": 223420 }, { "epoch": 0.0034048, "grad_norm": 0.7200319766998291, "learning_rate": 7.774757013801692e-06, "loss": 2.8262, "step": 223430 }, { "epoch": 0.0034304, "grad_norm": 0.7772827744483948, "learning_rate": 7.773971933172918e-06, "loss": 2.7968, "step": 223440 }, { "epoch": 0.003456, "grad_norm": 0.7819352149963379, "learning_rate": 7.773186866979344e-06, "loss": 2.9944, "step": 223450 }, { "epoch": 0.0034816, "grad_norm": 0.7930442690849304, "learning_rate": 7.772401815226064e-06, "loss": 2.8542, "step": 223460 }, { "epoch": 0.0035072, "grad_norm": 0.8106480836868286, "learning_rate": 7.771616777918165e-06, "loss": 2.8854, "step": 223470 }, { "epoch": 0.0035328, "grad_norm": 1.1745606660842896, "learning_rate": 7.770831755060738e-06, "loss": 2.7773, "step": 223480 }, { "epoch": 0.0035584, "grad_norm": 0.7650108337402344, "learning_rate": 7.770046746658873e-06, "loss": 2.8376, "step": 223490 }, { "epoch": 0.003584, "grad_norm": 0.8025096654891968, "learning_rate": 7.769261752717662e-06, "loss": 3.1447, "step": 223500 }, { "epoch": 0.0036096, "grad_norm": 1.4696840047836304, "learning_rate": 7.768476773242198e-06, "loss": 2.8987, "step": 223510 }, { "epoch": 0.0036352, "grad_norm": 0.8574017882347107, "learning_rate": 7.76769180823757e-06, "loss": 3.0261, "step": 223520 }, { "epoch": 0.0036608, "grad_norm": 0.7525452375411987, "learning_rate": 7.766906857708866e-06, "loss": 2.8325, "step": 223530 }, { "epoch": 0.0036864, "grad_norm": 0.7845581769943237, "learning_rate": 7.766121921661176e-06, "loss": 2.9014, "step": 223540 }, { "epoch": 0.003712, "grad_norm": 0.8035506010055542, "learning_rate": 7.765337000099594e-06, "loss": 2.9931, "step": 223550 }, { "epoch": 0.0037376, "grad_norm": 0.784777045249939, "learning_rate": 7.764552093029209e-06, "loss": 2.9173, "step": 223560 }, { "epoch": 0.0037632, "grad_norm": 0.7702402472496033, "learning_rate": 7.763767200455107e-06, "loss": 2.7622, "step": 223570 }, { "epoch": 0.0037888, "grad_norm": 0.7797010540962219, "learning_rate": 7.76298232238238e-06, "loss": 3.1416, "step": 223580 }, { "epoch": 0.0038144, "grad_norm": 0.8812538385391235, "learning_rate": 7.762197458816121e-06, "loss": 3.1054, "step": 223590 }, { "epoch": 0.00384, "grad_norm": 0.7064584493637085, "learning_rate": 7.761412609761413e-06, "loss": 2.8862, "step": 223600 }, { "epoch": 0.0038656, "grad_norm": 1.0303254127502441, "learning_rate": 7.760627775223353e-06, "loss": 2.8863, "step": 223610 }, { "epoch": 0.0038912, "grad_norm": 0.8706523776054382, "learning_rate": 7.759842955207023e-06, "loss": 3.0015, "step": 223620 }, { "epoch": 0.0039168, "grad_norm": 0.737785279750824, "learning_rate": 7.75905814971752e-06, "loss": 2.787, "step": 223630 }, { "epoch": 0.0039424, "grad_norm": 0.8483245372772217, "learning_rate": 7.758273358759927e-06, "loss": 2.7661, "step": 223640 }, { "epoch": 0.003968, "grad_norm": 0.774157702922821, "learning_rate": 7.757488582339336e-06, "loss": 3.0233, "step": 223650 }, { "epoch": 0.0039936, "grad_norm": 0.8186348080635071, "learning_rate": 7.756703820460836e-06, "loss": 2.8076, "step": 223660 }, { "epoch": 0.0040192, "grad_norm": 0.9197737574577332, "learning_rate": 7.75591907312951e-06, "loss": 2.9371, "step": 223670 }, { "epoch": 0.0040448, "grad_norm": 0.8415409326553345, "learning_rate": 7.75513434035046e-06, "loss": 2.9667, "step": 223680 }, { "epoch": 0.0040704, "grad_norm": 0.9231383800506592, "learning_rate": 7.754349622128763e-06, "loss": 2.8184, "step": 223690 }, { "epoch": 0.004096, "grad_norm": 0.704800009727478, "learning_rate": 7.753564918469514e-06, "loss": 2.8947, "step": 223700 }, { "epoch": 0.0041216, "grad_norm": 0.826592743396759, "learning_rate": 7.752780229377798e-06, "loss": 2.8853, "step": 223710 }, { "epoch": 0.0041472, "grad_norm": 0.7947266697883606, "learning_rate": 7.751995554858706e-06, "loss": 2.7457, "step": 223720 }, { "epoch": 0.0041728, "grad_norm": 0.7167239189147949, "learning_rate": 7.751210894917326e-06, "loss": 2.7919, "step": 223730 }, { "epoch": 0.0041984, "grad_norm": 0.9453202486038208, "learning_rate": 7.750426249558745e-06, "loss": 2.7881, "step": 223740 }, { "epoch": 0.004224, "grad_norm": 0.8145944476127625, "learning_rate": 7.749641618788051e-06, "loss": 2.9154, "step": 223750 }, { "epoch": 0.0042496, "grad_norm": 0.7962920665740967, "learning_rate": 7.748857002610336e-06, "loss": 2.992, "step": 223760 }, { "epoch": 0.0042752, "grad_norm": 0.8840938210487366, "learning_rate": 7.748072401030683e-06, "loss": 2.7734, "step": 223770 }, { "epoch": 0.0043008, "grad_norm": 0.905554473400116, "learning_rate": 7.74728781405418e-06, "loss": 2.933, "step": 223780 }, { "epoch": 0.0043264, "grad_norm": 0.8237380385398865, "learning_rate": 7.74650324168592e-06, "loss": 2.8338, "step": 223790 }, { "epoch": 0.004352, "grad_norm": 0.6828517317771912, "learning_rate": 7.745718683930985e-06, "loss": 2.8203, "step": 223800 }, { "epoch": 0.0043776, "grad_norm": 0.7559202313423157, "learning_rate": 7.744934140794466e-06, "loss": 2.9365, "step": 223810 }, { "epoch": 0.0044032, "grad_norm": 0.7338179349899292, "learning_rate": 7.74414961228145e-06, "loss": 2.9737, "step": 223820 }, { "epoch": 0.0044288, "grad_norm": 0.7092872262001038, "learning_rate": 7.743365098397026e-06, "loss": 2.9577, "step": 223830 }, { "epoch": 0.0044544, "grad_norm": 0.7452249526977539, "learning_rate": 7.742580599146275e-06, "loss": 2.6654, "step": 223840 }, { "epoch": 0.00448, "grad_norm": 0.7596805691719055, "learning_rate": 7.741796114534293e-06, "loss": 2.87, "step": 223850 }, { "epoch": 0.0045056, "grad_norm": 0.6985857486724854, "learning_rate": 7.741011644566161e-06, "loss": 2.8162, "step": 223860 }, { "epoch": 0.0045312, "grad_norm": 0.8030250668525696, "learning_rate": 7.740227189246969e-06, "loss": 2.7426, "step": 223870 }, { "epoch": 0.0045568, "grad_norm": 0.7865399122238159, "learning_rate": 7.739442748581805e-06, "loss": 2.947, "step": 223880 }, { "epoch": 0.0045824, "grad_norm": 0.7772747874259949, "learning_rate": 7.738658322575752e-06, "loss": 3.0245, "step": 223890 }, { "epoch": 0.004608, "grad_norm": 0.8688088655471802, "learning_rate": 7.737873911233898e-06, "loss": 2.8557, "step": 223900 }, { "epoch": 0.0046336, "grad_norm": 0.7375752329826355, "learning_rate": 7.73708951456133e-06, "loss": 2.9675, "step": 223910 }, { "epoch": 0.0046592, "grad_norm": 1.0492072105407715, "learning_rate": 7.73630513256314e-06, "loss": 2.8334, "step": 223920 }, { "epoch": 0.0046848, "grad_norm": 0.8462607860565186, "learning_rate": 7.735520765244403e-06, "loss": 3.1536, "step": 223930 }, { "epoch": 0.0047104, "grad_norm": 0.8338688015937805, "learning_rate": 7.734736412610215e-06, "loss": 2.982, "step": 223940 }, { "epoch": 0.004736, "grad_norm": 0.936669111251831, "learning_rate": 7.73395207466566e-06, "loss": 2.7831, "step": 223950 }, { "epoch": 0.0047616, "grad_norm": 1.0505518913269043, "learning_rate": 7.733167751415825e-06, "loss": 2.9358, "step": 223960 }, { "epoch": 0.0047872, "grad_norm": 0.7269933223724365, "learning_rate": 7.732383442865792e-06, "loss": 2.8627, "step": 223970 }, { "epoch": 0.0048128, "grad_norm": 0.7564818263053894, "learning_rate": 7.73159914902065e-06, "loss": 2.7912, "step": 223980 }, { "epoch": 0.0048384, "grad_norm": 0.7900147438049316, "learning_rate": 7.730814869885487e-06, "loss": 2.9006, "step": 223990 }, { "epoch": 0.004864, "grad_norm": 0.9548514485359192, "learning_rate": 7.730030605465384e-06, "loss": 2.8098, "step": 224000 }, { "epoch": 0.0048896, "grad_norm": 0.8054944276809692, "learning_rate": 7.72924635576543e-06, "loss": 2.9338, "step": 224010 }, { "epoch": 0.0049152, "grad_norm": 0.8758978843688965, "learning_rate": 7.728462120790707e-06, "loss": 2.7329, "step": 224020 }, { "epoch": 0.0049408, "grad_norm": 0.6869204640388489, "learning_rate": 7.727677900546305e-06, "loss": 2.7643, "step": 224030 }, { "epoch": 0.0049664, "grad_norm": 0.829255223274231, "learning_rate": 7.726893695037308e-06, "loss": 2.7314, "step": 224040 }, { "epoch": 0.004992, "grad_norm": 0.7454425096511841, "learning_rate": 7.726109504268804e-06, "loss": 2.9213, "step": 224050 }, { "epoch": 0.0050176, "grad_norm": 0.7481867074966431, "learning_rate": 7.725325328245874e-06, "loss": 2.8491, "step": 224060 }, { "epoch": 0.0050432, "grad_norm": 0.9269742965698242, "learning_rate": 7.724541166973603e-06, "loss": 3.0736, "step": 224070 }, { "epoch": 0.0050688, "grad_norm": 0.727698802947998, "learning_rate": 7.723757020457082e-06, "loss": 2.9123, "step": 224080 }, { "epoch": 0.0050944, "grad_norm": 0.9136136174201965, "learning_rate": 7.722972888701389e-06, "loss": 2.869, "step": 224090 }, { "epoch": 0.00512, "grad_norm": 0.7877634167671204, "learning_rate": 7.72218877171161e-06, "loss": 3.0317, "step": 224100 }, { "epoch": 0.0051456, "grad_norm": 0.7425364255905151, "learning_rate": 7.721404669492834e-06, "loss": 2.8076, "step": 224110 }, { "epoch": 0.0051712, "grad_norm": 1.0134286880493164, "learning_rate": 7.720620582050143e-06, "loss": 2.8903, "step": 224120 }, { "epoch": 0.0051968, "grad_norm": 0.791404664516449, "learning_rate": 7.71983650938862e-06, "loss": 2.7863, "step": 224130 }, { "epoch": 0.0052224, "grad_norm": 0.7426941394805908, "learning_rate": 7.719052451513353e-06, "loss": 2.9536, "step": 224140 }, { "epoch": 0.005248, "grad_norm": 0.773307740688324, "learning_rate": 7.718268408429424e-06, "loss": 2.8249, "step": 224150 }, { "epoch": 0.0052736, "grad_norm": 0.8645437955856323, "learning_rate": 7.717484380141919e-06, "loss": 3.0195, "step": 224160 }, { "epoch": 0.0052992, "grad_norm": 0.7886078357696533, "learning_rate": 7.71670036665592e-06, "loss": 2.8252, "step": 224170 }, { "epoch": 0.0053248, "grad_norm": 0.90793776512146, "learning_rate": 7.715916367976512e-06, "loss": 2.9868, "step": 224180 }, { "epoch": 0.0053504, "grad_norm": 0.7261369228363037, "learning_rate": 7.715132384108777e-06, "loss": 2.9427, "step": 224190 }, { "epoch": 0.005376, "grad_norm": 0.8445484042167664, "learning_rate": 7.714348415057802e-06, "loss": 2.8969, "step": 224200 }, { "epoch": 0.0054016, "grad_norm": 0.8883426785469055, "learning_rate": 7.713564460828675e-06, "loss": 2.8186, "step": 224210 }, { "epoch": 0.0054272, "grad_norm": 0.7701151371002197, "learning_rate": 7.712780521426471e-06, "loss": 2.6771, "step": 224220 }, { "epoch": 0.0054528, "grad_norm": 0.7732457518577576, "learning_rate": 7.711996596856277e-06, "loss": 2.8441, "step": 224230 }, { "epoch": 0.0054784, "grad_norm": 0.929219663143158, "learning_rate": 7.711212687123179e-06, "loss": 2.6507, "step": 224240 }, { "epoch": 0.005504, "grad_norm": 0.7414196133613586, "learning_rate": 7.710428792232258e-06, "loss": 2.6897, "step": 224250 }, { "epoch": 0.0055296, "grad_norm": 0.8255957365036011, "learning_rate": 7.709644912188599e-06, "loss": 2.838, "step": 224260 }, { "epoch": 0.0055552, "grad_norm": 0.7673500180244446, "learning_rate": 7.708861046997281e-06, "loss": 2.863, "step": 224270 }, { "epoch": 0.0055808, "grad_norm": 0.7468115091323853, "learning_rate": 7.708077196663393e-06, "loss": 2.9384, "step": 224280 }, { "epoch": 0.0056064, "grad_norm": 0.9067320227622986, "learning_rate": 7.707293361192014e-06, "loss": 2.9256, "step": 224290 }, { "epoch": 0.005632, "grad_norm": 0.7999630570411682, "learning_rate": 7.706509540588227e-06, "loss": 2.84, "step": 224300 }, { "epoch": 0.0056576, "grad_norm": 0.7502970695495605, "learning_rate": 7.705725734857118e-06, "loss": 2.7095, "step": 224310 }, { "epoch": 0.0056832, "grad_norm": 0.7215370535850525, "learning_rate": 7.70494194400377e-06, "loss": 2.966, "step": 224320 }, { "epoch": 0.0057088, "grad_norm": 0.8348323106765747, "learning_rate": 7.70415816803326e-06, "loss": 2.74, "step": 224330 }, { "epoch": 0.0057344, "grad_norm": 0.8324591517448425, "learning_rate": 7.703374406950673e-06, "loss": 2.8786, "step": 224340 }, { "epoch": 0.00576, "grad_norm": 0.905005693435669, "learning_rate": 7.702590660761094e-06, "loss": 3.0086, "step": 224350 }, { "epoch": 0.0057856, "grad_norm": 0.7557154893875122, "learning_rate": 7.701806929469607e-06, "loss": 2.8619, "step": 224360 }, { "epoch": 0.0058112, "grad_norm": 0.9174853563308716, "learning_rate": 7.701023213081283e-06, "loss": 2.8156, "step": 224370 }, { "epoch": 0.0058368, "grad_norm": 0.7293236255645752, "learning_rate": 7.70023951160122e-06, "loss": 2.7939, "step": 224380 }, { "epoch": 0.0058624, "grad_norm": 0.8183560371398926, "learning_rate": 7.69945582503449e-06, "loss": 2.8872, "step": 224390 }, { "epoch": 0.005888, "grad_norm": 0.7380894422531128, "learning_rate": 7.698672153386177e-06, "loss": 2.7159, "step": 224400 }, { "epoch": 0.0059136, "grad_norm": 0.80124831199646, "learning_rate": 7.697888496661367e-06, "loss": 2.9147, "step": 224410 }, { "epoch": 0.0059392, "grad_norm": 0.7691766023635864, "learning_rate": 7.697104854865136e-06, "loss": 2.6851, "step": 224420 }, { "epoch": 0.0059648, "grad_norm": 0.8678250312805176, "learning_rate": 7.696321228002567e-06, "loss": 2.7882, "step": 224430 }, { "epoch": 0.0059904, "grad_norm": 0.7820208072662354, "learning_rate": 7.695537616078744e-06, "loss": 2.6016, "step": 224440 }, { "epoch": 0.006016, "grad_norm": 0.9047303199768066, "learning_rate": 7.694754019098749e-06, "loss": 2.8804, "step": 224450 }, { "epoch": 0.0060416, "grad_norm": 0.7820132970809937, "learning_rate": 7.693970437067658e-06, "loss": 2.766, "step": 224460 }, { "epoch": 0.0060672, "grad_norm": 0.7290043830871582, "learning_rate": 7.693186869990557e-06, "loss": 2.7259, "step": 224470 }, { "epoch": 0.0060928, "grad_norm": 0.8625798225402832, "learning_rate": 7.692403317872528e-06, "loss": 2.7823, "step": 224480 }, { "epoch": 0.0061184, "grad_norm": 0.8551818132400513, "learning_rate": 7.691619780718646e-06, "loss": 2.6912, "step": 224490 }, { "epoch": 0.006144, "grad_norm": 0.7769790887832642, "learning_rate": 7.690836258534e-06, "loss": 2.7901, "step": 224500 }, { "epoch": 0.0061696, "grad_norm": 0.8749818205833435, "learning_rate": 7.690052751323664e-06, "loss": 2.7905, "step": 224510 }, { "epoch": 0.0061952, "grad_norm": 0.927340567111969, "learning_rate": 7.689269259092726e-06, "loss": 2.8072, "step": 224520 }, { "epoch": 0.0062208, "grad_norm": 0.7144753336906433, "learning_rate": 7.68848578184626e-06, "loss": 2.8063, "step": 224530 }, { "epoch": 0.0062464, "grad_norm": 0.9367243051528931, "learning_rate": 7.687702319589347e-06, "loss": 2.6406, "step": 224540 }, { "epoch": 0.006272, "grad_norm": 0.7967907190322876, "learning_rate": 7.686918872327072e-06, "loss": 2.979, "step": 224550 }, { "epoch": 0.0062976, "grad_norm": 0.7839783430099487, "learning_rate": 7.686135440064514e-06, "loss": 3.0584, "step": 224560 }, { "epoch": 0.0063232, "grad_norm": 2.4263739585876465, "learning_rate": 7.685352022806753e-06, "loss": 3.4607, "step": 224570 }, { "epoch": 0.0063488, "grad_norm": 0.8073060512542725, "learning_rate": 7.68456862055887e-06, "loss": 2.8293, "step": 224580 }, { "epoch": 0.0063744, "grad_norm": 0.7354047894477844, "learning_rate": 7.683785233325942e-06, "loss": 2.6521, "step": 224590 }, { "epoch": 0.0064, "grad_norm": 0.7345486879348755, "learning_rate": 7.683001861113052e-06, "loss": 2.6921, "step": 224600 }, { "epoch": 0.0064256, "grad_norm": 0.7587950229644775, "learning_rate": 7.682218503925282e-06, "loss": 2.837, "step": 224610 }, { "epoch": 0.0064512, "grad_norm": 0.8573896288871765, "learning_rate": 7.681435161767706e-06, "loss": 2.8837, "step": 224620 }, { "epoch": 0.0064768, "grad_norm": 0.7923617959022522, "learning_rate": 7.680651834645406e-06, "loss": 2.8059, "step": 224630 }, { "epoch": 0.0065024, "grad_norm": 0.7438244223594666, "learning_rate": 7.679868522563463e-06, "loss": 2.75, "step": 224640 }, { "epoch": 0.006528, "grad_norm": 0.9110978245735168, "learning_rate": 7.679085225526957e-06, "loss": 2.9314, "step": 224650 }, { "epoch": 0.0065536, "grad_norm": 0.7724844813346863, "learning_rate": 7.678301943540968e-06, "loss": 2.7565, "step": 224660 }, { "epoch": 0.0065792, "grad_norm": 0.9657053351402283, "learning_rate": 7.67751867661057e-06, "loss": 3.0967, "step": 224670 }, { "epoch": 0.0066048, "grad_norm": 0.7263516187667847, "learning_rate": 7.67673542474085e-06, "loss": 2.9653, "step": 224680 }, { "epoch": 0.0066304, "grad_norm": 1.6194393634796143, "learning_rate": 7.67595218793688e-06, "loss": 2.8525, "step": 224690 }, { "epoch": 0.006656, "grad_norm": 0.7861118316650391, "learning_rate": 7.675168966203744e-06, "loss": 2.7346, "step": 224700 }, { "epoch": 0.0066816, "grad_norm": 0.7348057627677917, "learning_rate": 7.674385759546517e-06, "loss": 2.7713, "step": 224710 }, { "epoch": 0.0067072, "grad_norm": 0.8162330389022827, "learning_rate": 7.673602567970282e-06, "loss": 2.816, "step": 224720 }, { "epoch": 0.0067328, "grad_norm": 0.8290749192237854, "learning_rate": 7.672819391480115e-06, "loss": 2.8408, "step": 224730 }, { "epoch": 0.0067584, "grad_norm": 0.7450800538063049, "learning_rate": 7.6720362300811e-06, "loss": 2.8947, "step": 224740 }, { "epoch": 0.006784, "grad_norm": 0.9453580975532532, "learning_rate": 7.671253083778308e-06, "loss": 2.8146, "step": 224750 }, { "epoch": 0.0068096, "grad_norm": 0.7669623494148254, "learning_rate": 7.67046995257682e-06, "loss": 2.89, "step": 224760 }, { "epoch": 0.0068352, "grad_norm": 0.772329568862915, "learning_rate": 7.66968683648172e-06, "loss": 2.8654, "step": 224770 }, { "epoch": 0.0068608, "grad_norm": 0.8584880828857422, "learning_rate": 7.668903735498077e-06, "loss": 2.8947, "step": 224780 }, { "epoch": 0.0068864, "grad_norm": 0.7517973780632019, "learning_rate": 7.668120649630975e-06, "loss": 2.9455, "step": 224790 }, { "epoch": 0.006912, "grad_norm": 0.8141652941703796, "learning_rate": 7.667337578885489e-06, "loss": 2.9381, "step": 224800 }, { "epoch": 0.0069376, "grad_norm": 0.8923761248588562, "learning_rate": 7.666554523266703e-06, "loss": 2.8351, "step": 224810 }, { "epoch": 0.0069632, "grad_norm": 4.203935623168945, "learning_rate": 7.665771482779687e-06, "loss": 2.7309, "step": 224820 }, { "epoch": 0.0069888, "grad_norm": 0.75910484790802, "learning_rate": 7.664988457429524e-06, "loss": 2.7204, "step": 224830 }, { "epoch": 0.0070144, "grad_norm": 0.8838079571723938, "learning_rate": 7.664205447221291e-06, "loss": 2.6632, "step": 224840 }, { "epoch": 0.00704, "grad_norm": 0.9220057129859924, "learning_rate": 7.663422452160065e-06, "loss": 2.983, "step": 224850 }, { "epoch": 0.0070656, "grad_norm": 0.9733765125274658, "learning_rate": 7.662639472250923e-06, "loss": 2.7437, "step": 224860 }, { "epoch": 0.0070912, "grad_norm": 0.8162880539894104, "learning_rate": 7.661856507498942e-06, "loss": 2.7332, "step": 224870 }, { "epoch": 0.0071168, "grad_norm": 0.7291954755783081, "learning_rate": 7.6610735579092e-06, "loss": 2.5912, "step": 224880 }, { "epoch": 0.0071424, "grad_norm": 0.7380422949790955, "learning_rate": 7.660290623486771e-06, "loss": 2.7356, "step": 224890 }, { "epoch": 0.007168, "grad_norm": 1.2878975868225098, "learning_rate": 7.659507704236742e-06, "loss": 2.8362, "step": 224900 }, { "epoch": 0.0071936, "grad_norm": 0.7951452732086182, "learning_rate": 7.65872480016418e-06, "loss": 2.9974, "step": 224910 }, { "epoch": 0.0072192, "grad_norm": 0.7844328880310059, "learning_rate": 7.657941911274166e-06, "loss": 2.9002, "step": 224920 }, { "epoch": 0.0072448, "grad_norm": 0.8509463667869568, "learning_rate": 7.657159037571776e-06, "loss": 2.86, "step": 224930 }, { "epoch": 0.0072704, "grad_norm": 0.8081140518188477, "learning_rate": 7.65637617906209e-06, "loss": 2.8875, "step": 224940 }, { "epoch": 0.007296, "grad_norm": 0.7995471954345703, "learning_rate": 7.655593335750178e-06, "loss": 3.0489, "step": 224950 }, { "epoch": 0.0073216, "grad_norm": 0.8017489910125732, "learning_rate": 7.65481050764112e-06, "loss": 2.7706, "step": 224960 }, { "epoch": 0.0073472, "grad_norm": 0.9523279070854187, "learning_rate": 7.654027694739997e-06, "loss": 2.9345, "step": 224970 }, { "epoch": 0.0073728, "grad_norm": 0.7999243140220642, "learning_rate": 7.653244897051876e-06, "loss": 2.8696, "step": 224980 }, { "epoch": 0.0073984, "grad_norm": 0.7804606556892395, "learning_rate": 7.652462114581841e-06, "loss": 2.8723, "step": 224990 }, { "epoch": 0.007424, "grad_norm": 0.7983022332191467, "learning_rate": 7.651679347334963e-06, "loss": 2.7065, "step": 225000 }, { "epoch": 0.0074496, "grad_norm": 0.8239557147026062, "learning_rate": 7.650896595316323e-06, "loss": 3.068, "step": 225010 }, { "epoch": 0.0074752, "grad_norm": 0.7875622510910034, "learning_rate": 7.650113858530993e-06, "loss": 2.9949, "step": 225020 }, { "epoch": 0.0075008, "grad_norm": 0.8452543616294861, "learning_rate": 7.64933113698405e-06, "loss": 2.9866, "step": 225030 }, { "epoch": 0.0075264, "grad_norm": 1.0420818328857422, "learning_rate": 7.64854843068057e-06, "loss": 3.0251, "step": 225040 }, { "epoch": 0.007552, "grad_norm": 0.7565840482711792, "learning_rate": 7.647765739625628e-06, "loss": 2.7978, "step": 225050 }, { "epoch": 0.0075776, "grad_norm": 0.7632127404212952, "learning_rate": 7.646983063824295e-06, "loss": 2.9225, "step": 225060 }, { "epoch": 0.0076032, "grad_norm": 0.7683786749839783, "learning_rate": 7.64620040328166e-06, "loss": 2.7809, "step": 225070 }, { "epoch": 0.0076288, "grad_norm": 0.8118898272514343, "learning_rate": 7.645417758002786e-06, "loss": 3.0111, "step": 225080 }, { "epoch": 0.0076544, "grad_norm": 0.899219274520874, "learning_rate": 7.644635127992752e-06, "loss": 2.9718, "step": 225090 }, { "epoch": 0.00768, "grad_norm": 0.7158111929893494, "learning_rate": 7.643852513256635e-06, "loss": 2.8526, "step": 225100 }, { "epoch": 0.0077056, "grad_norm": 0.7894721627235413, "learning_rate": 7.643069913799508e-06, "loss": 2.865, "step": 225110 }, { "epoch": 0.0077312, "grad_norm": 0.824394166469574, "learning_rate": 7.642287329626445e-06, "loss": 2.9207, "step": 225120 }, { "epoch": 0.0077568, "grad_norm": 0.8297189474105835, "learning_rate": 7.641504760742522e-06, "loss": 2.8719, "step": 225130 }, { "epoch": 0.0077824, "grad_norm": 0.8224695920944214, "learning_rate": 7.640722207152815e-06, "loss": 2.915, "step": 225140 }, { "epoch": 0.007808, "grad_norm": 0.8161991834640503, "learning_rate": 7.639939668862397e-06, "loss": 3.0837, "step": 225150 }, { "epoch": 0.0078336, "grad_norm": 0.7273837327957153, "learning_rate": 7.639157145876341e-06, "loss": 2.9593, "step": 225160 }, { "epoch": 0.0078592, "grad_norm": 0.7558660507202148, "learning_rate": 7.638374638199729e-06, "loss": 2.744, "step": 225170 }, { "epoch": 0.0078848, "grad_norm": 0.7743884921073914, "learning_rate": 7.637592145837624e-06, "loss": 2.9232, "step": 225180 }, { "epoch": 0.0079104, "grad_norm": 0.7898077964782715, "learning_rate": 7.636809668795108e-06, "loss": 2.8292, "step": 225190 }, { "epoch": 0.007936, "grad_norm": 0.7352606654167175, "learning_rate": 7.636027207077253e-06, "loss": 2.8696, "step": 225200 }, { "epoch": 0.0079616, "grad_norm": 0.8361998796463013, "learning_rate": 7.635244760689136e-06, "loss": 3.0147, "step": 225210 }, { "epoch": 0.0079872, "grad_norm": 0.8149480819702148, "learning_rate": 7.634462329635825e-06, "loss": 2.9714, "step": 225220 }, { "epoch": 0.0080128, "grad_norm": 0.7283406257629395, "learning_rate": 7.633679913922397e-06, "loss": 2.8183, "step": 225230 }, { "epoch": 0.0080384, "grad_norm": 0.7360242009162903, "learning_rate": 7.632897513553926e-06, "loss": 2.7753, "step": 225240 }, { "epoch": 0.008064, "grad_norm": 0.8613230586051941, "learning_rate": 7.632115128535485e-06, "loss": 2.8882, "step": 225250 }, { "epoch": 0.0080896, "grad_norm": 0.8034119009971619, "learning_rate": 7.631332758872149e-06, "loss": 3.0107, "step": 225260 }, { "epoch": 0.0081152, "grad_norm": 0.7241524457931519, "learning_rate": 7.630550404568993e-06, "loss": 2.8234, "step": 225270 }, { "epoch": 0.0081408, "grad_norm": 0.7865414619445801, "learning_rate": 7.629768065631084e-06, "loss": 2.9751, "step": 225280 }, { "epoch": 0.0081664, "grad_norm": 0.7902804017066956, "learning_rate": 7.6289857420635015e-06, "loss": 2.8565, "step": 225290 }, { "epoch": 0.008192, "grad_norm": 0.7839124202728271, "learning_rate": 7.6282034338713175e-06, "loss": 2.8174, "step": 225300 }, { "epoch": 0.0082176, "grad_norm": 0.7917754054069519, "learning_rate": 7.627421141059602e-06, "loss": 2.8522, "step": 225310 }, { "epoch": 0.0082432, "grad_norm": 0.8272274732589722, "learning_rate": 7.62663886363343e-06, "loss": 3.0172, "step": 225320 }, { "epoch": 0.0082688, "grad_norm": 0.7669395804405212, "learning_rate": 7.6258566015978745e-06, "loss": 2.6974, "step": 225330 }, { "epoch": 0.0082944, "grad_norm": 0.9450449347496033, "learning_rate": 7.625074354958009e-06, "loss": 2.8969, "step": 225340 }, { "epoch": 0.00832, "grad_norm": 0.7401437163352966, "learning_rate": 7.624292123718903e-06, "loss": 2.8911, "step": 225350 }, { "epoch": 0.0083456, "grad_norm": 0.6849464774131775, "learning_rate": 7.623509907885631e-06, "loss": 2.9457, "step": 225360 }, { "epoch": 0.0083712, "grad_norm": 0.7371878623962402, "learning_rate": 7.622727707463269e-06, "loss": 2.9226, "step": 225370 }, { "epoch": 0.0083968, "grad_norm": 0.7992516160011292, "learning_rate": 7.621945522456884e-06, "loss": 2.7915, "step": 225380 }, { "epoch": 0.0084224, "grad_norm": 0.7706938982009888, "learning_rate": 7.621163352871549e-06, "loss": 3.0318, "step": 225390 }, { "epoch": 0.008448, "grad_norm": 1.006635308265686, "learning_rate": 7.6203811987123375e-06, "loss": 2.9361, "step": 225400 }, { "epoch": 0.0084736, "grad_norm": 0.7344557046890259, "learning_rate": 7.619599059984321e-06, "loss": 2.9093, "step": 225410 }, { "epoch": 0.0084992, "grad_norm": 0.9451607465744019, "learning_rate": 7.6188169366925725e-06, "loss": 2.9591, "step": 225420 }, { "epoch": 0.0085248, "grad_norm": 0.7968072295188904, "learning_rate": 7.618034828842165e-06, "loss": 2.9849, "step": 225430 }, { "epoch": 0.0085504, "grad_norm": 0.9220768213272095, "learning_rate": 7.617252736438169e-06, "loss": 3.0987, "step": 225440 }, { "epoch": 0.008576, "grad_norm": 0.8322171568870544, "learning_rate": 7.616470659485653e-06, "loss": 2.7557, "step": 225450 }, { "epoch": 0.0086016, "grad_norm": 0.8083220720291138, "learning_rate": 7.6156885979896934e-06, "loss": 2.9276, "step": 225460 }, { "epoch": 0.0086272, "grad_norm": 0.8281475901603699, "learning_rate": 7.614906551955361e-06, "loss": 3.0165, "step": 225470 }, { "epoch": 0.0086528, "grad_norm": 0.7298689484596252, "learning_rate": 7.6141245213877225e-06, "loss": 2.8626, "step": 225480 }, { "epoch": 0.0086784, "grad_norm": 0.7773233652114868, "learning_rate": 7.613342506291854e-06, "loss": 3.007, "step": 225490 }, { "epoch": 0.008704, "grad_norm": 0.8533651232719421, "learning_rate": 7.612560506672826e-06, "loss": 2.9208, "step": 225500 }, { "epoch": 0.0087296, "grad_norm": 0.7573497891426086, "learning_rate": 7.611778522535706e-06, "loss": 2.7557, "step": 225510 }, { "epoch": 0.0087552, "grad_norm": 0.9174337983131409, "learning_rate": 7.61099655388557e-06, "loss": 2.9282, "step": 225520 }, { "epoch": 0.0087808, "grad_norm": 0.8395511507987976, "learning_rate": 7.6102146007274844e-06, "loss": 2.7041, "step": 225530 }, { "epoch": 0.0088064, "grad_norm": 0.7768842577934265, "learning_rate": 7.609432663066525e-06, "loss": 2.8968, "step": 225540 }, { "epoch": 0.008832, "grad_norm": 0.8163425326347351, "learning_rate": 7.608650740907757e-06, "loss": 2.9506, "step": 225550 }, { "epoch": 0.0088576, "grad_norm": 0.8474366664886475, "learning_rate": 7.6078688342562536e-06, "loss": 2.9466, "step": 225560 }, { "epoch": 0.0088832, "grad_norm": 0.8043003678321838, "learning_rate": 7.6070869431170855e-06, "loss": 2.8983, "step": 225570 }, { "epoch": 0.0089088, "grad_norm": 1.0548772811889648, "learning_rate": 7.606305067495319e-06, "loss": 3.1315, "step": 225580 }, { "epoch": 0.0089344, "grad_norm": 0.8168244361877441, "learning_rate": 7.605523207396033e-06, "loss": 2.8558, "step": 225590 }, { "epoch": 0.00896, "grad_norm": 0.7060512900352478, "learning_rate": 7.60474136282429e-06, "loss": 2.8968, "step": 225600 }, { "epoch": 0.0089856, "grad_norm": 0.7003252506256104, "learning_rate": 7.603959533785163e-06, "loss": 2.8518, "step": 225610 }, { "epoch": 0.0090112, "grad_norm": 0.7770854830741882, "learning_rate": 7.603177720283722e-06, "loss": 2.749, "step": 225620 }, { "epoch": 0.0090368, "grad_norm": 0.9509363770484924, "learning_rate": 7.602395922325038e-06, "loss": 2.8484, "step": 225630 }, { "epoch": 0.0090624, "grad_norm": 0.954262912273407, "learning_rate": 7.601614139914177e-06, "loss": 3.1239, "step": 225640 }, { "epoch": 0.009088, "grad_norm": 0.7350845336914062, "learning_rate": 7.600832373056213e-06, "loss": 2.9584, "step": 225650 }, { "epoch": 0.0091136, "grad_norm": 1.0110881328582764, "learning_rate": 7.600050621756213e-06, "loss": 2.9416, "step": 225660 }, { "epoch": 0.0091392, "grad_norm": 0.778174102306366, "learning_rate": 7.5992688860192465e-06, "loss": 3.1133, "step": 225670 }, { "epoch": 0.0091648, "grad_norm": 0.8475741147994995, "learning_rate": 7.598487165850382e-06, "loss": 2.7526, "step": 225680 }, { "epoch": 0.0091904, "grad_norm": 0.829048216342926, "learning_rate": 7.5977054612546904e-06, "loss": 2.7192, "step": 225690 }, { "epoch": 0.009216, "grad_norm": 0.7210890650749207, "learning_rate": 7.596923772237243e-06, "loss": 2.9406, "step": 225700 }, { "epoch": 0.0092416, "grad_norm": 0.7992998361587524, "learning_rate": 7.596142098803103e-06, "loss": 2.8476, "step": 225710 }, { "epoch": 0.0092672, "grad_norm": 0.7821301817893982, "learning_rate": 7.595360440957344e-06, "loss": 3.2026, "step": 225720 }, { "epoch": 0.0092928, "grad_norm": 0.7863298654556274, "learning_rate": 7.594578798705033e-06, "loss": 2.9349, "step": 225730 }, { "epoch": 0.0093184, "grad_norm": 0.7913373708724976, "learning_rate": 7.593797172051242e-06, "loss": 2.8592, "step": 225740 }, { "epoch": 0.009344, "grad_norm": 0.7663036584854126, "learning_rate": 7.593015561001033e-06, "loss": 2.8563, "step": 225750 }, { "epoch": 0.0093696, "grad_norm": 0.7042489051818848, "learning_rate": 7.5922339655594765e-06, "loss": 2.8152, "step": 225760 }, { "epoch": 0.0093952, "grad_norm": 0.7775182723999023, "learning_rate": 7.5914523857316455e-06, "loss": 2.8303, "step": 225770 }, { "epoch": 0.0094208, "grad_norm": 0.8576377630233765, "learning_rate": 7.5906708215226066e-06, "loss": 2.6561, "step": 225780 }, { "epoch": 0.0094464, "grad_norm": 0.8001193404197693, "learning_rate": 7.589889272937428e-06, "loss": 2.9296, "step": 225790 }, { "epoch": 0.009472, "grad_norm": 0.7839401960372925, "learning_rate": 7.589107739981176e-06, "loss": 2.7412, "step": 225800 }, { "epoch": 0.0094976, "grad_norm": 0.9131547212600708, "learning_rate": 7.588326222658918e-06, "loss": 2.7845, "step": 225810 }, { "epoch": 0.0095232, "grad_norm": 0.7604316473007202, "learning_rate": 7.5875447209757246e-06, "loss": 2.9277, "step": 225820 }, { "epoch": 0.0095488, "grad_norm": 0.8366090059280396, "learning_rate": 7.586763234936665e-06, "loss": 2.9184, "step": 225830 }, { "epoch": 0.0095744, "grad_norm": 0.7959209084510803, "learning_rate": 7.5859817645468015e-06, "loss": 3.0302, "step": 225840 }, { "epoch": 0.0096, "grad_norm": 0.7035276293754578, "learning_rate": 7.585200309811205e-06, "loss": 3.1406, "step": 225850 }, { "epoch": 0.0096256, "grad_norm": 1.0299752950668335, "learning_rate": 7.584418870734944e-06, "loss": 2.9599, "step": 225860 }, { "epoch": 0.0096512, "grad_norm": 0.7485557198524475, "learning_rate": 7.5836374473230835e-06, "loss": 2.8873, "step": 225870 }, { "epoch": 0.0096768, "grad_norm": 0.875673234462738, "learning_rate": 7.582856039580691e-06, "loss": 2.7525, "step": 225880 }, { "epoch": 0.0097024, "grad_norm": 0.8248760104179382, "learning_rate": 7.582074647512836e-06, "loss": 2.8597, "step": 225890 }, { "epoch": 0.009728, "grad_norm": 0.802676260471344, "learning_rate": 7.581293271124585e-06, "loss": 3.0161, "step": 225900 }, { "epoch": 0.0097536, "grad_norm": 0.9185221791267395, "learning_rate": 7.580511910421005e-06, "loss": 3.0151, "step": 225910 }, { "epoch": 0.0097792, "grad_norm": 1.0377280712127686, "learning_rate": 7.57973056540716e-06, "loss": 2.9756, "step": 225920 }, { "epoch": 0.0098048, "grad_norm": 0.7778937816619873, "learning_rate": 7.578949236088117e-06, "loss": 2.8064, "step": 225930 }, { "epoch": 0.0098304, "grad_norm": 0.7639576196670532, "learning_rate": 7.578167922468946e-06, "loss": 2.9727, "step": 225940 }, { "epoch": 0.009856, "grad_norm": 0.7322406768798828, "learning_rate": 7.5773866245547145e-06, "loss": 2.9476, "step": 225950 }, { "epoch": 0.0098816, "grad_norm": 0.7431378364562988, "learning_rate": 7.576605342350487e-06, "loss": 2.5719, "step": 225960 }, { "epoch": 0.0099072, "grad_norm": 0.931770384311676, "learning_rate": 7.575824075861329e-06, "loss": 3.2557, "step": 225970 }, { "epoch": 0.0099328, "grad_norm": 0.7933812141418457, "learning_rate": 7.575042825092307e-06, "loss": 2.8996, "step": 225980 }, { "epoch": 0.0099584, "grad_norm": 1.0906201601028442, "learning_rate": 7.57426159004849e-06, "loss": 2.975, "step": 225990 }, { "epoch": 0.009984, "grad_norm": 0.8124772906303406, "learning_rate": 7.57348037073494e-06, "loss": 2.8611, "step": 226000 }, { "epoch": 0.0100096, "grad_norm": 0.8700548410415649, "learning_rate": 7.572699167156725e-06, "loss": 3.0798, "step": 226010 }, { "epoch": 0.0100352, "grad_norm": 0.7557306885719299, "learning_rate": 7.57191797931891e-06, "loss": 2.8691, "step": 226020 }, { "epoch": 0.0100608, "grad_norm": 0.9085885286331177, "learning_rate": 7.571136807226565e-06, "loss": 2.8444, "step": 226030 }, { "epoch": 0.0100864, "grad_norm": 0.9210959076881409, "learning_rate": 7.57035565088475e-06, "loss": 2.9621, "step": 226040 }, { "epoch": 0.010112, "grad_norm": 0.7866324782371521, "learning_rate": 7.569574510298532e-06, "loss": 3.2208, "step": 226050 }, { "epoch": 0.0101376, "grad_norm": 0.7296047806739807, "learning_rate": 7.56879338547298e-06, "loss": 2.8984, "step": 226060 }, { "epoch": 0.0101632, "grad_norm": 0.7304167747497559, "learning_rate": 7.568012276413155e-06, "loss": 2.8843, "step": 226070 }, { "epoch": 0.0101888, "grad_norm": 1.2067861557006836, "learning_rate": 7.567231183124123e-06, "loss": 2.7119, "step": 226080 }, { "epoch": 0.0102144, "grad_norm": 0.7799324989318848, "learning_rate": 7.5664501056109505e-06, "loss": 2.957, "step": 226090 }, { "epoch": 0.01024, "grad_norm": 0.8368343114852905, "learning_rate": 7.565669043878705e-06, "loss": 2.9682, "step": 226100 }, { "epoch": 0.0102656, "grad_norm": 0.8049653172492981, "learning_rate": 7.5648879979324416e-06, "loss": 3.0517, "step": 226110 }, { "epoch": 0.0102912, "grad_norm": 0.8096044063568115, "learning_rate": 7.5641069677772405e-06, "loss": 2.9506, "step": 226120 }, { "epoch": 0.0103168, "grad_norm": 1.0337234735488892, "learning_rate": 7.5633259534181544e-06, "loss": 3.2847, "step": 226130 }, { "epoch": 0.0103424, "grad_norm": 0.8239107131958008, "learning_rate": 7.562544954860253e-06, "loss": 3.067, "step": 226140 }, { "epoch": 0.010368, "grad_norm": 0.8062755465507507, "learning_rate": 7.561763972108599e-06, "loss": 2.8415, "step": 226150 }, { "epoch": 0.0103936, "grad_norm": 0.7257677316665649, "learning_rate": 7.56098300516826e-06, "loss": 2.8628, "step": 226160 }, { "epoch": 0.0104192, "grad_norm": 0.944346010684967, "learning_rate": 7.560202054044296e-06, "loss": 2.6892, "step": 226170 }, { "epoch": 0.0104448, "grad_norm": 0.7335224747657776, "learning_rate": 7.559421118741772e-06, "loss": 2.9306, "step": 226180 }, { "epoch": 0.0104704, "grad_norm": 0.787739634513855, "learning_rate": 7.558640199265758e-06, "loss": 2.8652, "step": 226190 }, { "epoch": 0.010496, "grad_norm": 0.905517041683197, "learning_rate": 7.557859295621309e-06, "loss": 2.7043, "step": 226200 }, { "epoch": 0.0105216, "grad_norm": 0.7995606660842896, "learning_rate": 7.557078407813495e-06, "loss": 2.8061, "step": 226210 }, { "epoch": 0.0105472, "grad_norm": 1.0020370483398438, "learning_rate": 7.556297535847378e-06, "loss": 2.8147, "step": 226220 }, { "epoch": 0.0105728, "grad_norm": 0.7721514701843262, "learning_rate": 7.555516679728023e-06, "loss": 2.8199, "step": 226230 }, { "epoch": 0.0105984, "grad_norm": 0.7889450788497925, "learning_rate": 7.554735839460492e-06, "loss": 2.9749, "step": 226240 }, { "epoch": 0.010624, "grad_norm": 0.8687942624092102, "learning_rate": 7.553955015049849e-06, "loss": 3.0084, "step": 226250 }, { "epoch": 0.0106496, "grad_norm": 1.4476341009140015, "learning_rate": 7.5531742065011595e-06, "loss": 3.4681, "step": 226260 }, { "epoch": 0.0106752, "grad_norm": 0.7575473189353943, "learning_rate": 7.552393413819482e-06, "loss": 2.6968, "step": 226270 }, { "epoch": 0.0107008, "grad_norm": 0.7771850228309631, "learning_rate": 7.551612637009882e-06, "loss": 2.9716, "step": 226280 }, { "epoch": 0.0107264, "grad_norm": 1.1024795770645142, "learning_rate": 7.5508318760774245e-06, "loss": 2.7085, "step": 226290 }, { "epoch": 0.010752, "grad_norm": 0.8406946659088135, "learning_rate": 7.550051131027172e-06, "loss": 2.8051, "step": 226300 }, { "epoch": 0.0107776, "grad_norm": 0.7503443360328674, "learning_rate": 7.549270401864186e-06, "loss": 2.7871, "step": 226310 }, { "epoch": 0.0108032, "grad_norm": 0.8006724119186401, "learning_rate": 7.548489688593532e-06, "loss": 2.8822, "step": 226320 }, { "epoch": 0.0108288, "grad_norm": 0.8354693651199341, "learning_rate": 7.547708991220271e-06, "loss": 2.7966, "step": 226330 }, { "epoch": 0.0108544, "grad_norm": 0.7943509817123413, "learning_rate": 7.546928309749463e-06, "loss": 3.1971, "step": 226340 }, { "epoch": 0.01088, "grad_norm": 0.7658198475837708, "learning_rate": 7.546147644186174e-06, "loss": 2.9952, "step": 226350 }, { "epoch": 0.0109056, "grad_norm": 0.7592348456382751, "learning_rate": 7.545366994535467e-06, "loss": 2.7294, "step": 226360 }, { "epoch": 0.0109312, "grad_norm": 0.8227871656417847, "learning_rate": 7.544586360802401e-06, "loss": 2.932, "step": 226370 }, { "epoch": 0.0109568, "grad_norm": 0.9095953106880188, "learning_rate": 7.5438057429920394e-06, "loss": 2.8776, "step": 226380 }, { "epoch": 0.0109824, "grad_norm": 0.8112788200378418, "learning_rate": 7.543025141109448e-06, "loss": 2.7365, "step": 226390 }, { "epoch": 0.011008, "grad_norm": 1.0449328422546387, "learning_rate": 7.542244555159682e-06, "loss": 2.9642, "step": 226400 }, { "epoch": 0.0110336, "grad_norm": 0.8369930386543274, "learning_rate": 7.541463985147808e-06, "loss": 2.8103, "step": 226410 }, { "epoch": 0.0110592, "grad_norm": 0.9100437164306641, "learning_rate": 7.540683431078886e-06, "loss": 2.981, "step": 226420 }, { "epoch": 0.0110848, "grad_norm": 0.7796320915222168, "learning_rate": 7.539902892957981e-06, "loss": 2.8183, "step": 226430 }, { "epoch": 0.0111104, "grad_norm": 0.7864221930503845, "learning_rate": 7.53912237079015e-06, "loss": 2.9174, "step": 226440 }, { "epoch": 0.011136, "grad_norm": 0.7694200873374939, "learning_rate": 7.5383418645804544e-06, "loss": 2.9749, "step": 226450 }, { "epoch": 0.0111616, "grad_norm": 0.9409664273262024, "learning_rate": 7.537561374333958e-06, "loss": 2.9438, "step": 226460 }, { "epoch": 0.0111872, "grad_norm": 1.0667976140975952, "learning_rate": 7.536780900055722e-06, "loss": 2.8341, "step": 226470 }, { "epoch": 0.0112128, "grad_norm": 1.4910492897033691, "learning_rate": 7.536000441750811e-06, "loss": 2.8912, "step": 226480 }, { "epoch": 0.0112384, "grad_norm": 0.9133172035217285, "learning_rate": 7.535219999424279e-06, "loss": 2.7357, "step": 226490 }, { "epoch": 0.011264, "grad_norm": 0.819423258304596, "learning_rate": 7.53443957308119e-06, "loss": 2.9551, "step": 226500 }, { "epoch": 0.0112896, "grad_norm": 1.0150954723358154, "learning_rate": 7.533659162726606e-06, "loss": 3.0529, "step": 226510 }, { "epoch": 0.0113152, "grad_norm": 0.8460549712181091, "learning_rate": 7.532878768365589e-06, "loss": 3.1044, "step": 226520 }, { "epoch": 0.0113408, "grad_norm": 0.8780903816223145, "learning_rate": 7.532098390003196e-06, "loss": 3.0402, "step": 226530 }, { "epoch": 0.0113664, "grad_norm": 0.8493303060531616, "learning_rate": 7.531318027644487e-06, "loss": 2.7675, "step": 226540 }, { "epoch": 0.011392, "grad_norm": 0.88905268907547, "learning_rate": 7.530537681294525e-06, "loss": 2.9327, "step": 226550 }, { "epoch": 0.0114176, "grad_norm": 0.756653904914856, "learning_rate": 7.529757350958373e-06, "loss": 2.912, "step": 226560 }, { "epoch": 0.0114432, "grad_norm": 0.8717381954193115, "learning_rate": 7.528977036641085e-06, "loss": 3.0289, "step": 226570 }, { "epoch": 0.0114688, "grad_norm": 0.8404857516288757, "learning_rate": 7.528196738347725e-06, "loss": 2.8934, "step": 226580 }, { "epoch": 0.0114944, "grad_norm": 0.8161652684211731, "learning_rate": 7.527416456083354e-06, "loss": 3.0016, "step": 226590 }, { "epoch": 0.01152, "grad_norm": 1.0477162599563599, "learning_rate": 7.526636189853027e-06, "loss": 2.8222, "step": 226600 }, { "epoch": 0.0115456, "grad_norm": 0.832905650138855, "learning_rate": 7.525855939661808e-06, "loss": 2.8314, "step": 226610 }, { "epoch": 0.0115712, "grad_norm": 0.7635290026664734, "learning_rate": 7.525075705514756e-06, "loss": 2.6711, "step": 226620 }, { "epoch": 0.0115968, "grad_norm": 1.1021329164505005, "learning_rate": 7.524295487416929e-06, "loss": 2.986, "step": 226630 }, { "epoch": 0.0116224, "grad_norm": 1.0898579359054565, "learning_rate": 7.523515285373387e-06, "loss": 2.9853, "step": 226640 }, { "epoch": 0.011648, "grad_norm": 0.7730545401573181, "learning_rate": 7.522735099389194e-06, "loss": 3.044, "step": 226650 }, { "epoch": 0.0116736, "grad_norm": 0.7490707039833069, "learning_rate": 7.521954929469404e-06, "loss": 2.8597, "step": 226660 }, { "epoch": 0.0116992, "grad_norm": 0.8153200149536133, "learning_rate": 7.5211747756190775e-06, "loss": 2.8038, "step": 226670 }, { "epoch": 0.0117248, "grad_norm": 0.7583196759223938, "learning_rate": 7.5203946378432755e-06, "loss": 2.8814, "step": 226680 }, { "epoch": 0.0117504, "grad_norm": 0.8961819410324097, "learning_rate": 7.519614516147054e-06, "loss": 2.8183, "step": 226690 }, { "epoch": 0.011776, "grad_norm": 0.8395124077796936, "learning_rate": 7.518834410535471e-06, "loss": 2.9728, "step": 226700 }, { "epoch": 0.0118016, "grad_norm": 0.8339666724205017, "learning_rate": 7.51805432101359e-06, "loss": 2.9133, "step": 226710 }, { "epoch": 0.0118272, "grad_norm": 0.9332476854324341, "learning_rate": 7.517274247586466e-06, "loss": 2.819, "step": 226720 }, { "epoch": 0.0118528, "grad_norm": 0.7818000912666321, "learning_rate": 7.516494190259159e-06, "loss": 2.91, "step": 226730 }, { "epoch": 0.0118784, "grad_norm": 1.0488375425338745, "learning_rate": 7.515714149036727e-06, "loss": 3.0546, "step": 226740 }, { "epoch": 0.011904, "grad_norm": 1.5403817892074585, "learning_rate": 7.514934123924232e-06, "loss": 2.9783, "step": 226750 }, { "epoch": 0.0119296, "grad_norm": 0.7544465065002441, "learning_rate": 7.514154114926724e-06, "loss": 2.7636, "step": 226760 }, { "epoch": 0.0119552, "grad_norm": 0.7887693643569946, "learning_rate": 7.513374122049266e-06, "loss": 2.9092, "step": 226770 }, { "epoch": 0.0119808, "grad_norm": 0.8306525945663452, "learning_rate": 7.512594145296917e-06, "loss": 2.8224, "step": 226780 }, { "epoch": 0.0120064, "grad_norm": 0.7914901375770569, "learning_rate": 7.511814184674736e-06, "loss": 2.9176, "step": 226790 }, { "epoch": 0.012032, "grad_norm": 0.7629109621047974, "learning_rate": 7.511034240187772e-06, "loss": 3.0947, "step": 226800 }, { "epoch": 0.0120576, "grad_norm": 0.8663826584815979, "learning_rate": 7.510254311841097e-06, "loss": 2.8482, "step": 226810 }, { "epoch": 0.0120832, "grad_norm": 0.7607318162918091, "learning_rate": 7.509474399639758e-06, "loss": 3.0703, "step": 226820 }, { "epoch": 0.0121088, "grad_norm": 0.7265464067459106, "learning_rate": 7.508694503588815e-06, "loss": 2.8673, "step": 226830 }, { "epoch": 0.0121344, "grad_norm": 0.9021744728088379, "learning_rate": 7.507914623693326e-06, "loss": 2.8373, "step": 226840 }, { "epoch": 0.01216, "grad_norm": 2.9746932983398438, "learning_rate": 7.507134759958352e-06, "loss": 2.9008, "step": 226850 }, { "epoch": 0.0121856, "grad_norm": 0.7856020331382751, "learning_rate": 7.506354912388944e-06, "loss": 3.012, "step": 226860 }, { "epoch": 0.0122112, "grad_norm": 0.8292780518531799, "learning_rate": 7.50557508099016e-06, "loss": 3.0168, "step": 226870 }, { "epoch": 0.0122368, "grad_norm": 0.8816992044448853, "learning_rate": 7.504795265767062e-06, "loss": 2.8851, "step": 226880 }, { "epoch": 0.0122624, "grad_norm": 0.6975011825561523, "learning_rate": 7.5040154667247015e-06, "loss": 3.0564, "step": 226890 }, { "epoch": 0.012288, "grad_norm": 0.8748137950897217, "learning_rate": 7.503235683868137e-06, "loss": 2.8891, "step": 226900 }, { "epoch": 0.0123136, "grad_norm": 1.0646578073501587, "learning_rate": 7.502455917202427e-06, "loss": 2.9679, "step": 226910 }, { "epoch": 0.0123392, "grad_norm": 0.7154342532157898, "learning_rate": 7.501676166732627e-06, "loss": 2.844, "step": 226920 }, { "epoch": 0.0123648, "grad_norm": 0.8244153261184692, "learning_rate": 7.5008964324637914e-06, "loss": 3.0118, "step": 226930 }, { "epoch": 0.0123904, "grad_norm": 0.9501535296440125, "learning_rate": 7.500116714400978e-06, "loss": 2.8672, "step": 226940 }, { "epoch": 0.012416, "grad_norm": 0.8923887014389038, "learning_rate": 7.499337012549246e-06, "loss": 3.4203, "step": 226950 }, { "epoch": 0.0124416, "grad_norm": 0.8868750929832458, "learning_rate": 7.498557326913647e-06, "loss": 2.8892, "step": 226960 }, { "epoch": 0.0124672, "grad_norm": 0.8268434405326843, "learning_rate": 7.497777657499238e-06, "loss": 2.9416, "step": 226970 }, { "epoch": 0.0124928, "grad_norm": 0.75315260887146, "learning_rate": 7.496998004311073e-06, "loss": 2.5666, "step": 226980 }, { "epoch": 0.0125184, "grad_norm": 0.7995443940162659, "learning_rate": 7.496218367354215e-06, "loss": 2.9042, "step": 226990 }, { "epoch": 0.012544, "grad_norm": 0.8015064597129822, "learning_rate": 7.4954387466337145e-06, "loss": 2.9627, "step": 227000 }, { "epoch": 0.0125696, "grad_norm": 0.7530717849731445, "learning_rate": 7.49465914215463e-06, "loss": 2.8607, "step": 227010 }, { "epoch": 0.0125952, "grad_norm": 0.7923406958580017, "learning_rate": 7.493879553922014e-06, "loss": 2.8062, "step": 227020 }, { "epoch": 0.0126208, "grad_norm": 0.80167156457901, "learning_rate": 7.4930999819409224e-06, "loss": 2.716, "step": 227030 }, { "epoch": 0.0126464, "grad_norm": 0.7785676121711731, "learning_rate": 7.492320426216411e-06, "loss": 2.8395, "step": 227040 }, { "epoch": 0.012672, "grad_norm": 0.7436546087265015, "learning_rate": 7.491540886753536e-06, "loss": 2.8195, "step": 227050 }, { "epoch": 0.0126976, "grad_norm": 0.7614055275917053, "learning_rate": 7.490761363557353e-06, "loss": 2.9321, "step": 227060 }, { "epoch": 0.0127232, "grad_norm": 0.8122978806495667, "learning_rate": 7.489981856632913e-06, "loss": 3.0341, "step": 227070 }, { "epoch": 0.0127488, "grad_norm": 1.1120191812515259, "learning_rate": 7.489202365985275e-06, "loss": 2.8023, "step": 227080 }, { "epoch": 0.0127744, "grad_norm": 0.937224805355072, "learning_rate": 7.488422891619492e-06, "loss": 3.1636, "step": 227090 }, { "epoch": 0.0128, "grad_norm": 0.9201505780220032, "learning_rate": 7.487643433540619e-06, "loss": 2.8516, "step": 227100 }, { "epoch": 0.0128256, "grad_norm": 0.8560100793838501, "learning_rate": 7.48686399175371e-06, "loss": 2.9334, "step": 227110 }, { "epoch": 0.0128512, "grad_norm": 0.7737686634063721, "learning_rate": 7.486084566263822e-06, "loss": 2.894, "step": 227120 }, { "epoch": 0.0128768, "grad_norm": 0.8804389834403992, "learning_rate": 7.485305157076006e-06, "loss": 2.8321, "step": 227130 }, { "epoch": 0.0129024, "grad_norm": 0.7697224617004395, "learning_rate": 7.484525764195318e-06, "loss": 3.09, "step": 227140 }, { "epoch": 0.012928, "grad_norm": 8.314922332763672, "learning_rate": 7.483746387626809e-06, "loss": 2.7749, "step": 227150 }, { "epoch": 0.0129536, "grad_norm": 1.0475199222564697, "learning_rate": 7.482967027375538e-06, "loss": 2.7439, "step": 227160 }, { "epoch": 0.0129792, "grad_norm": 1.2255029678344727, "learning_rate": 7.4821876834465604e-06, "loss": 3.0734, "step": 227170 }, { "epoch": 0.0130048, "grad_norm": 0.8013322949409485, "learning_rate": 7.481408355844924e-06, "loss": 2.8773, "step": 227180 }, { "epoch": 0.0130304, "grad_norm": 0.929044783115387, "learning_rate": 7.480629044575683e-06, "loss": 2.8542, "step": 227190 }, { "epoch": 0.013056, "grad_norm": 0.7984309792518616, "learning_rate": 7.479849749643895e-06, "loss": 2.8872, "step": 227200 }, { "epoch": 0.0130816, "grad_norm": 0.6974067687988281, "learning_rate": 7.479070471054613e-06, "loss": 2.8427, "step": 227210 }, { "epoch": 0.0131072, "grad_norm": 0.8482993841171265, "learning_rate": 7.478291208812887e-06, "loss": 2.8386, "step": 227220 }, { "epoch": 0.0131328, "grad_norm": 0.8495494723320007, "learning_rate": 7.477511962923772e-06, "loss": 2.8949, "step": 227230 }, { "epoch": 0.0131584, "grad_norm": 0.9809382557868958, "learning_rate": 7.476732733392322e-06, "loss": 2.7877, "step": 227240 }, { "epoch": 0.013184, "grad_norm": 0.8229444622993469, "learning_rate": 7.475953520223592e-06, "loss": 2.73, "step": 227250 }, { "epoch": 0.0132096, "grad_norm": 0.7735115885734558, "learning_rate": 7.475174323422631e-06, "loss": 2.7974, "step": 227260 }, { "epoch": 0.0132352, "grad_norm": 0.9665094614028931, "learning_rate": 7.4743951429944916e-06, "loss": 3.1709, "step": 227270 }, { "epoch": 0.0132608, "grad_norm": 1.902188777923584, "learning_rate": 7.473615978944232e-06, "loss": 2.8799, "step": 227280 }, { "epoch": 0.0132864, "grad_norm": 1.6058084964752197, "learning_rate": 7.4728368312768994e-06, "loss": 2.9365, "step": 227290 }, { "epoch": 0.013312, "grad_norm": 0.8908938765525818, "learning_rate": 7.472057699997547e-06, "loss": 3.1389, "step": 227300 }, { "epoch": 0.0133376, "grad_norm": 1.8595993518829346, "learning_rate": 7.47127858511123e-06, "loss": 3.0002, "step": 227310 }, { "epoch": 0.0133632, "grad_norm": 1.1285021305084229, "learning_rate": 7.470499486622996e-06, "loss": 3.0835, "step": 227320 }, { "epoch": 0.0133888, "grad_norm": 1.758978247642517, "learning_rate": 7.469720404537904e-06, "loss": 2.9785, "step": 227330 }, { "epoch": 0.0134144, "grad_norm": 1.0683314800262451, "learning_rate": 7.4689413388610035e-06, "loss": 2.6177, "step": 227340 }, { "epoch": 0.01344, "grad_norm": 0.7249557375907898, "learning_rate": 7.4681622895973435e-06, "loss": 2.8419, "step": 227350 }, { "epoch": 0.0134656, "grad_norm": 0.8215747475624084, "learning_rate": 7.467383256751979e-06, "loss": 2.9459, "step": 227360 }, { "epoch": 0.0134912, "grad_norm": 0.7943587899208069, "learning_rate": 7.466604240329963e-06, "loss": 2.9713, "step": 227370 }, { "epoch": 0.0135168, "grad_norm": 0.7894514203071594, "learning_rate": 7.465825240336344e-06, "loss": 2.8551, "step": 227380 }, { "epoch": 0.0135424, "grad_norm": 0.8016325831413269, "learning_rate": 7.465046256776174e-06, "loss": 2.9782, "step": 227390 }, { "epoch": 0.013568, "grad_norm": 0.7543492317199707, "learning_rate": 7.464267289654504e-06, "loss": 2.8251, "step": 227400 }, { "epoch": 0.0135936, "grad_norm": 1.0655550956726074, "learning_rate": 7.46348833897639e-06, "loss": 2.8433, "step": 227410 }, { "epoch": 0.0136192, "grad_norm": 0.7044832110404968, "learning_rate": 7.4627094047468784e-06, "loss": 2.8726, "step": 227420 }, { "epoch": 0.0136448, "grad_norm": 0.8077667355537415, "learning_rate": 7.461930486971021e-06, "loss": 2.9898, "step": 227430 }, { "epoch": 0.0136704, "grad_norm": 0.8139181733131409, "learning_rate": 7.461151585653871e-06, "loss": 2.9994, "step": 227440 }, { "epoch": 0.013696, "grad_norm": 0.7116496562957764, "learning_rate": 7.46037270080048e-06, "loss": 2.9569, "step": 227450 }, { "epoch": 0.0137216, "grad_norm": 0.8491995334625244, "learning_rate": 7.459593832415894e-06, "loss": 2.944, "step": 227460 }, { "epoch": 0.0137472, "grad_norm": 0.864926815032959, "learning_rate": 7.458814980505167e-06, "loss": 3.0318, "step": 227470 }, { "epoch": 0.0137728, "grad_norm": 1.7701878547668457, "learning_rate": 7.458036145073353e-06, "loss": 2.9452, "step": 227480 }, { "epoch": 0.0137984, "grad_norm": 0.8937796354293823, "learning_rate": 7.457257326125496e-06, "loss": 2.9951, "step": 227490 }, { "epoch": 0.013824, "grad_norm": 0.8566206097602844, "learning_rate": 7.456478523666647e-06, "loss": 3.003, "step": 227500 }, { "epoch": 0.0138496, "grad_norm": 0.7596827745437622, "learning_rate": 7.4556997377018615e-06, "loss": 2.892, "step": 227510 }, { "epoch": 0.0138752, "grad_norm": 0.7126838564872742, "learning_rate": 7.454920968236185e-06, "loss": 2.7274, "step": 227520 }, { "epoch": 0.0139008, "grad_norm": 0.7071728706359863, "learning_rate": 7.454142215274673e-06, "loss": 2.7744, "step": 227530 }, { "epoch": 0.0139264, "grad_norm": 0.8777564167976379, "learning_rate": 7.453363478822373e-06, "loss": 3.0185, "step": 227540 }, { "epoch": 0.013952, "grad_norm": 0.9207074046134949, "learning_rate": 7.452584758884332e-06, "loss": 2.9126, "step": 227550 }, { "epoch": 0.0139776, "grad_norm": 0.8154584765434265, "learning_rate": 7.4518060554656005e-06, "loss": 3.1026, "step": 227560 }, { "epoch": 0.0140032, "grad_norm": 0.8044978380203247, "learning_rate": 7.451027368571232e-06, "loss": 2.8871, "step": 227570 }, { "epoch": 0.0140288, "grad_norm": 0.8882056474685669, "learning_rate": 7.4502486982062725e-06, "loss": 2.6506, "step": 227580 }, { "epoch": 0.0140544, "grad_norm": 0.9518028497695923, "learning_rate": 7.4494700443757715e-06, "loss": 3.1632, "step": 227590 }, { "epoch": 0.01408, "grad_norm": 0.7834179997444153, "learning_rate": 7.448691407084781e-06, "loss": 3.0005, "step": 227600 }, { "epoch": 0.0141056, "grad_norm": 0.927428662776947, "learning_rate": 7.447912786338348e-06, "loss": 2.6402, "step": 227610 }, { "epoch": 0.0141312, "grad_norm": 0.917767345905304, "learning_rate": 7.447134182141522e-06, "loss": 2.8493, "step": 227620 }, { "epoch": 0.0141568, "grad_norm": 0.7895434498786926, "learning_rate": 7.446355594499352e-06, "loss": 2.8137, "step": 227630 }, { "epoch": 0.0141824, "grad_norm": 0.8418500423431396, "learning_rate": 7.44557702341689e-06, "loss": 2.9359, "step": 227640 }, { "epoch": 0.014208, "grad_norm": 0.8146124482154846, "learning_rate": 7.444798468899179e-06, "loss": 2.8554, "step": 227650 }, { "epoch": 0.0142336, "grad_norm": 0.7810711860656738, "learning_rate": 7.444019930951272e-06, "loss": 3.0092, "step": 227660 }, { "epoch": 0.0142592, "grad_norm": 0.9472891688346863, "learning_rate": 7.443241409578214e-06, "loss": 2.8351, "step": 227670 }, { "epoch": 0.0142848, "grad_norm": 0.8337320685386658, "learning_rate": 7.442462904785057e-06, "loss": 2.7222, "step": 227680 }, { "epoch": 0.0143104, "grad_norm": 1.2037023305892944, "learning_rate": 7.441684416576849e-06, "loss": 2.8584, "step": 227690 }, { "epoch": 0.014336, "grad_norm": 1.0034960508346558, "learning_rate": 7.440905944958641e-06, "loss": 2.9021, "step": 227700 }, { "epoch": 0.0143616, "grad_norm": 0.8608083128929138, "learning_rate": 7.440127489935474e-06, "loss": 2.7192, "step": 227710 }, { "epoch": 0.0143872, "grad_norm": 0.7962020039558411, "learning_rate": 7.439349051512401e-06, "loss": 3.1411, "step": 227720 }, { "epoch": 0.0144128, "grad_norm": 0.7209932804107666, "learning_rate": 7.438570629694468e-06, "loss": 2.6752, "step": 227730 }, { "epoch": 0.0144384, "grad_norm": 0.8311460614204407, "learning_rate": 7.437792224486726e-06, "loss": 2.8752, "step": 227740 }, { "epoch": 0.014464, "grad_norm": 1.3523311614990234, "learning_rate": 7.43701383589422e-06, "loss": 2.994, "step": 227750 }, { "epoch": 0.0144896, "grad_norm": 0.9114249348640442, "learning_rate": 7.436235463921997e-06, "loss": 3.0139, "step": 227760 }, { "epoch": 0.0145152, "grad_norm": 0.7951630353927612, "learning_rate": 7.4354571085751075e-06, "loss": 3.0434, "step": 227770 }, { "epoch": 0.0145408, "grad_norm": 0.7344707250595093, "learning_rate": 7.434678769858595e-06, "loss": 2.8044, "step": 227780 }, { "epoch": 0.0145664, "grad_norm": 0.8143326044082642, "learning_rate": 7.433900447777508e-06, "loss": 2.8899, "step": 227790 }, { "epoch": 0.014592, "grad_norm": 0.7947902083396912, "learning_rate": 7.433122142336895e-06, "loss": 2.8832, "step": 227800 }, { "epoch": 0.0146176, "grad_norm": 0.948718786239624, "learning_rate": 7.432343853541806e-06, "loss": 2.8758, "step": 227810 }, { "epoch": 0.0146432, "grad_norm": 0.684836208820343, "learning_rate": 7.431565581397282e-06, "loss": 2.843, "step": 227820 }, { "epoch": 0.0146688, "grad_norm": 0.7263327836990356, "learning_rate": 7.430787325908373e-06, "loss": 2.8517, "step": 227830 }, { "epoch": 0.0146944, "grad_norm": 0.9249165058135986, "learning_rate": 7.4300090870801256e-06, "loss": 3.0349, "step": 227840 }, { "epoch": 0.01472, "grad_norm": 1.0036135911941528, "learning_rate": 7.4292308649175824e-06, "loss": 3.137, "step": 227850 }, { "epoch": 0.0147456, "grad_norm": 0.8950598239898682, "learning_rate": 7.428452659425798e-06, "loss": 2.6794, "step": 227860 }, { "epoch": 0.0147712, "grad_norm": 0.7487627267837524, "learning_rate": 7.427674470609814e-06, "loss": 2.9049, "step": 227870 }, { "epoch": 0.0147968, "grad_norm": 0.9192511439323425, "learning_rate": 7.426896298474677e-06, "loss": 2.806, "step": 227880 }, { "epoch": 0.0148224, "grad_norm": 0.9049121737480164, "learning_rate": 7.426118143025434e-06, "loss": 2.6249, "step": 227890 }, { "epoch": 0.014848, "grad_norm": 0.9245825409889221, "learning_rate": 7.425340004267132e-06, "loss": 2.992, "step": 227900 }, { "epoch": 0.0148736, "grad_norm": 1.5813037157058716, "learning_rate": 7.424561882204815e-06, "loss": 2.864, "step": 227910 }, { "epoch": 0.0148992, "grad_norm": 0.8632029294967651, "learning_rate": 7.423783776843528e-06, "loss": 3.0595, "step": 227920 }, { "epoch": 0.0149248, "grad_norm": 0.7682383060455322, "learning_rate": 7.42300568818832e-06, "loss": 2.7841, "step": 227930 }, { "epoch": 0.0149504, "grad_norm": 0.9098476767539978, "learning_rate": 7.422227616244238e-06, "loss": 2.8199, "step": 227940 }, { "epoch": 0.014976, "grad_norm": 1.9076924324035645, "learning_rate": 7.421449561016321e-06, "loss": 2.7941, "step": 227950 }, { "epoch": 0.0150016, "grad_norm": 0.8267217874526978, "learning_rate": 7.420671522509618e-06, "loss": 2.74, "step": 227960 }, { "epoch": 0.0150272, "grad_norm": 0.7923847436904907, "learning_rate": 7.419893500729178e-06, "loss": 2.9235, "step": 227970 }, { "epoch": 0.0150528, "grad_norm": 0.8933171033859253, "learning_rate": 7.41911549568004e-06, "loss": 2.9173, "step": 227980 }, { "epoch": 0.0150784, "grad_norm": 0.7995720505714417, "learning_rate": 7.418337507367253e-06, "loss": 2.6027, "step": 227990 }, { "epoch": 0.015104, "grad_norm": 0.7155951857566833, "learning_rate": 7.4175595357958605e-06, "loss": 2.8212, "step": 228000 }, { "epoch": 0.0151296, "grad_norm": 0.7399287223815918, "learning_rate": 7.416781580970911e-06, "loss": 2.9324, "step": 228010 }, { "epoch": 0.0151552, "grad_norm": 0.7527096271514893, "learning_rate": 7.416003642897439e-06, "loss": 2.9017, "step": 228020 }, { "epoch": 0.0151808, "grad_norm": 0.8978663086891174, "learning_rate": 7.415225721580503e-06, "loss": 2.979, "step": 228030 }, { "epoch": 0.0152064, "grad_norm": 0.863756000995636, "learning_rate": 7.41444781702514e-06, "loss": 2.9768, "step": 228040 }, { "epoch": 0.015232, "grad_norm": 0.8188830018043518, "learning_rate": 7.413669929236395e-06, "loss": 2.9286, "step": 228050 }, { "epoch": 0.0152576, "grad_norm": 0.9076647162437439, "learning_rate": 7.412892058219317e-06, "loss": 2.7936, "step": 228060 }, { "epoch": 0.0152832, "grad_norm": 0.7557579278945923, "learning_rate": 7.412114203978942e-06, "loss": 2.8862, "step": 228070 }, { "epoch": 0.0153088, "grad_norm": 0.804697573184967, "learning_rate": 7.4113363665203184e-06, "loss": 2.8612, "step": 228080 }, { "epoch": 0.0153344, "grad_norm": 0.9616523385047913, "learning_rate": 7.410558545848492e-06, "loss": 2.8959, "step": 228090 }, { "epoch": 0.01536, "grad_norm": 0.81722092628479, "learning_rate": 7.409780741968508e-06, "loss": 3.0572, "step": 228100 }, { "epoch": 0.0153856, "grad_norm": 0.7945312857627869, "learning_rate": 7.409002954885404e-06, "loss": 2.7659, "step": 228110 }, { "epoch": 0.0154112, "grad_norm": 0.7759965062141418, "learning_rate": 7.408225184604227e-06, "loss": 3.0365, "step": 228120 }, { "epoch": 0.0154368, "grad_norm": 0.775428831577301, "learning_rate": 7.4074474311300215e-06, "loss": 2.8667, "step": 228130 }, { "epoch": 0.0154624, "grad_norm": 1.1298491954803467, "learning_rate": 7.4066696944678325e-06, "loss": 3.1111, "step": 228140 }, { "epoch": 0.015488, "grad_norm": 1.0493123531341553, "learning_rate": 7.405891974622698e-06, "loss": 2.6225, "step": 228150 }, { "epoch": 0.0155136, "grad_norm": 0.9013817310333252, "learning_rate": 7.405114271599667e-06, "loss": 2.7348, "step": 228160 }, { "epoch": 0.0155392, "grad_norm": 0.7571948766708374, "learning_rate": 7.404336585403782e-06, "loss": 3.042, "step": 228170 }, { "epoch": 0.0155648, "grad_norm": 0.9958547353744507, "learning_rate": 7.403558916040081e-06, "loss": 3.0721, "step": 228180 }, { "epoch": 0.0155904, "grad_norm": 0.6851269602775574, "learning_rate": 7.402781263513608e-06, "loss": 2.669, "step": 228190 }, { "epoch": 0.015616, "grad_norm": 0.7389848232269287, "learning_rate": 7.4020036278294114e-06, "loss": 2.9013, "step": 228200 }, { "epoch": 0.0156416, "grad_norm": 0.7494125962257385, "learning_rate": 7.40122600899253e-06, "loss": 2.8048, "step": 228210 }, { "epoch": 0.0156672, "grad_norm": 0.8979671597480774, "learning_rate": 7.400448407008008e-06, "loss": 2.7717, "step": 228220 }, { "epoch": 0.0156928, "grad_norm": 0.7441792488098145, "learning_rate": 7.399670821880888e-06, "loss": 2.8006, "step": 228230 }, { "epoch": 0.0157184, "grad_norm": 0.8318825960159302, "learning_rate": 7.398893253616212e-06, "loss": 3.017, "step": 228240 }, { "epoch": 0.015744, "grad_norm": 0.8575557470321655, "learning_rate": 7.39811570221902e-06, "loss": 2.9657, "step": 228250 }, { "epoch": 0.0157696, "grad_norm": 0.7982114553451538, "learning_rate": 7.397338167694359e-06, "loss": 2.7695, "step": 228260 }, { "epoch": 0.0157952, "grad_norm": 0.9172466397285461, "learning_rate": 7.396560650047267e-06, "loss": 2.9369, "step": 228270 }, { "epoch": 0.0158208, "grad_norm": 0.8976559042930603, "learning_rate": 7.395783149282785e-06, "loss": 2.7793, "step": 228280 }, { "epoch": 0.0158464, "grad_norm": 0.7687751650810242, "learning_rate": 7.395005665405958e-06, "loss": 2.8223, "step": 228290 }, { "epoch": 0.015872, "grad_norm": 0.8413302302360535, "learning_rate": 7.394228198421831e-06, "loss": 2.9821, "step": 228300 }, { "epoch": 0.0158976, "grad_norm": 1.0550105571746826, "learning_rate": 7.393450748335438e-06, "loss": 2.6589, "step": 228310 }, { "epoch": 0.0159232, "grad_norm": 0.7825183272361755, "learning_rate": 7.392673315151824e-06, "loss": 3.0158, "step": 228320 }, { "epoch": 0.0159488, "grad_norm": 0.6899970769882202, "learning_rate": 7.39189589887603e-06, "loss": 2.8465, "step": 228330 }, { "epoch": 0.0159744, "grad_norm": 0.7751114964485168, "learning_rate": 7.391118499513101e-06, "loss": 2.9505, "step": 228340 }, { "epoch": 0.016, "grad_norm": 0.7899521589279175, "learning_rate": 7.390341117068073e-06, "loss": 2.9042, "step": 228350 }, { "epoch": 0.0160256, "grad_norm": 0.8104633092880249, "learning_rate": 7.38956375154599e-06, "loss": 2.9534, "step": 228360 }, { "epoch": 0.0160512, "grad_norm": 0.8828237652778625, "learning_rate": 7.388786402951889e-06, "loss": 2.7961, "step": 228370 }, { "epoch": 0.0160768, "grad_norm": 0.7371370196342468, "learning_rate": 7.388009071290816e-06, "loss": 2.844, "step": 228380 }, { "epoch": 0.0161024, "grad_norm": 0.7509877681732178, "learning_rate": 7.387231756567812e-06, "loss": 2.8942, "step": 228390 }, { "epoch": 0.016128, "grad_norm": 0.8241857290267944, "learning_rate": 7.386454458787915e-06, "loss": 2.9548, "step": 228400 }, { "epoch": 0.0161536, "grad_norm": 2.6960391998291016, "learning_rate": 7.3856771779561655e-06, "loss": 3.1611, "step": 228410 }, { "epoch": 0.0161792, "grad_norm": 0.8241780996322632, "learning_rate": 7.384899914077605e-06, "loss": 2.889, "step": 228420 }, { "epoch": 0.0162048, "grad_norm": 0.8903581500053406, "learning_rate": 7.384122667157274e-06, "loss": 2.7438, "step": 228430 }, { "epoch": 0.0162304, "grad_norm": 0.7428287863731384, "learning_rate": 7.383345437200212e-06, "loss": 2.9348, "step": 228440 }, { "epoch": 0.016256, "grad_norm": 0.8622326254844666, "learning_rate": 7.3825682242114594e-06, "loss": 2.7875, "step": 228450 }, { "epoch": 0.0162816, "grad_norm": 0.928043782711029, "learning_rate": 7.381791028196058e-06, "loss": 2.8288, "step": 228460 }, { "epoch": 0.0163072, "grad_norm": 0.7893290519714355, "learning_rate": 7.381013849159045e-06, "loss": 2.962, "step": 228470 }, { "epoch": 0.0163328, "grad_norm": 0.7139514088630676, "learning_rate": 7.380236687105459e-06, "loss": 2.9834, "step": 228480 }, { "epoch": 0.0163584, "grad_norm": 0.7676225900650024, "learning_rate": 7.379459542040343e-06, "loss": 2.7601, "step": 228490 }, { "epoch": 0.016384, "grad_norm": 0.9780951738357544, "learning_rate": 7.378682413968736e-06, "loss": 2.8177, "step": 228500 }, { "epoch": 0.0164096, "grad_norm": 0.8045657277107239, "learning_rate": 7.377905302895676e-06, "loss": 2.7299, "step": 228510 }, { "epoch": 0.0164352, "grad_norm": 0.7292447090148926, "learning_rate": 7.3771282088262035e-06, "loss": 2.995, "step": 228520 }, { "epoch": 0.0164608, "grad_norm": 1.4343743324279785, "learning_rate": 7.376351131765358e-06, "loss": 2.6297, "step": 228530 }, { "epoch": 0.0164864, "grad_norm": 0.8654559850692749, "learning_rate": 7.3755740717181745e-06, "loss": 2.8764, "step": 228540 }, { "epoch": 0.016512, "grad_norm": 0.9056954383850098, "learning_rate": 7.374797028689698e-06, "loss": 2.8875, "step": 228550 }, { "epoch": 0.0165376, "grad_norm": 0.7216144800186157, "learning_rate": 7.374020002684969e-06, "loss": 2.8332, "step": 228560 }, { "epoch": 0.0165632, "grad_norm": 0.799103856086731, "learning_rate": 7.373242993709019e-06, "loss": 2.9773, "step": 228570 }, { "epoch": 0.0165888, "grad_norm": 0.856468141078949, "learning_rate": 7.372466001766889e-06, "loss": 2.8624, "step": 228580 }, { "epoch": 0.0166144, "grad_norm": 0.8614787459373474, "learning_rate": 7.371689026863622e-06, "loss": 2.9117, "step": 228590 }, { "epoch": 0.01664, "grad_norm": 0.8697903156280518, "learning_rate": 7.370912069004251e-06, "loss": 3.1508, "step": 228600 }, { "epoch": 0.0166656, "grad_norm": 0.762244701385498, "learning_rate": 7.370135128193816e-06, "loss": 3.1101, "step": 228610 }, { "epoch": 0.0166912, "grad_norm": 0.8048393130302429, "learning_rate": 7.369358204437357e-06, "loss": 2.8477, "step": 228620 }, { "epoch": 0.0167168, "grad_norm": 0.737295389175415, "learning_rate": 7.3685812977399115e-06, "loss": 3.2428, "step": 228630 }, { "epoch": 0.0167424, "grad_norm": 1.1324963569641113, "learning_rate": 7.367804408106516e-06, "loss": 2.9844, "step": 228640 }, { "epoch": 0.016768, "grad_norm": 0.8058040142059326, "learning_rate": 7.367027535542208e-06, "loss": 2.8897, "step": 228650 }, { "epoch": 0.0167936, "grad_norm": 0.7179549932479858, "learning_rate": 7.366250680052031e-06, "loss": 3.0773, "step": 228660 }, { "epoch": 0.0168192, "grad_norm": 0.7881062626838684, "learning_rate": 7.3654738416410145e-06, "loss": 2.8009, "step": 228670 }, { "epoch": 0.0168448, "grad_norm": 0.8794838786125183, "learning_rate": 7.3646970203142e-06, "loss": 2.8828, "step": 228680 }, { "epoch": 0.0168704, "grad_norm": 2.3313918113708496, "learning_rate": 7.3639202160766255e-06, "loss": 2.7744, "step": 228690 }, { "epoch": 0.016896, "grad_norm": 0.7656843066215515, "learning_rate": 7.363143428933331e-06, "loss": 3.0061, "step": 228700 }, { "epoch": 0.0169216, "grad_norm": 0.7869149446487427, "learning_rate": 7.3623666588893465e-06, "loss": 2.9334, "step": 228710 }, { "epoch": 0.0169472, "grad_norm": 0.8805742859840393, "learning_rate": 7.361589905949711e-06, "loss": 2.8049, "step": 228720 }, { "epoch": 0.0169728, "grad_norm": 0.9506463408470154, "learning_rate": 7.360813170119467e-06, "loss": 2.956, "step": 228730 }, { "epoch": 0.0169984, "grad_norm": 0.7532191276550293, "learning_rate": 7.360036451403647e-06, "loss": 2.7514, "step": 228740 }, { "epoch": 0.017024, "grad_norm": 0.9261078238487244, "learning_rate": 7.359259749807291e-06, "loss": 2.9736, "step": 228750 }, { "epoch": 0.0170496, "grad_norm": 0.7929372787475586, "learning_rate": 7.358483065335433e-06, "loss": 2.8927, "step": 228760 }, { "epoch": 0.0170752, "grad_norm": 0.8751689791679382, "learning_rate": 7.357706397993111e-06, "loss": 2.8366, "step": 228770 }, { "epoch": 0.0171008, "grad_norm": 0.9040488600730896, "learning_rate": 7.35692974778536e-06, "loss": 2.7528, "step": 228780 }, { "epoch": 0.0171264, "grad_norm": 0.8106488585472107, "learning_rate": 7.356153114717218e-06, "loss": 2.7222, "step": 228790 }, { "epoch": 0.017152, "grad_norm": 0.7719770073890686, "learning_rate": 7.355376498793719e-06, "loss": 2.7805, "step": 228800 }, { "epoch": 0.0171776, "grad_norm": 1.1966663599014282, "learning_rate": 7.354599900019901e-06, "loss": 2.9329, "step": 228810 }, { "epoch": 0.0172032, "grad_norm": 1.1729927062988281, "learning_rate": 7.353823318400799e-06, "loss": 3.0013, "step": 228820 }, { "epoch": 0.0172288, "grad_norm": 0.8311567306518555, "learning_rate": 7.353046753941453e-06, "loss": 2.9121, "step": 228830 }, { "epoch": 0.0172544, "grad_norm": 0.8614813089370728, "learning_rate": 7.352270206646892e-06, "loss": 3.0571, "step": 228840 }, { "epoch": 0.01728, "grad_norm": 0.857474684715271, "learning_rate": 7.3514936765221545e-06, "loss": 2.8055, "step": 228850 }, { "epoch": 0.0173056, "grad_norm": 0.7779080271720886, "learning_rate": 7.3507171635722795e-06, "loss": 3.0822, "step": 228860 }, { "epoch": 0.0173312, "grad_norm": 0.8226438760757446, "learning_rate": 7.349940667802298e-06, "loss": 2.9731, "step": 228870 }, { "epoch": 0.0173568, "grad_norm": 0.7394111156463623, "learning_rate": 7.3491641892172465e-06, "loss": 2.8726, "step": 228880 }, { "epoch": 0.0173824, "grad_norm": 0.9029639363288879, "learning_rate": 7.348387727822157e-06, "loss": 3.0327, "step": 228890 }, { "epoch": 0.017408, "grad_norm": 0.897900402545929, "learning_rate": 7.347611283622072e-06, "loss": 2.7934, "step": 228900 }, { "epoch": 0.0174336, "grad_norm": 0.7260076999664307, "learning_rate": 7.346834856622023e-06, "loss": 2.8649, "step": 228910 }, { "epoch": 0.0174592, "grad_norm": 1.3424947261810303, "learning_rate": 7.3460584468270465e-06, "loss": 3.0325, "step": 228920 }, { "epoch": 0.0174848, "grad_norm": 0.7743948698043823, "learning_rate": 7.3452820542421745e-06, "loss": 2.9356, "step": 228930 }, { "epoch": 0.0175104, "grad_norm": 0.8161030411720276, "learning_rate": 7.3445056788724425e-06, "loss": 3.0521, "step": 228940 }, { "epoch": 0.017536, "grad_norm": 0.9184212684631348, "learning_rate": 7.343729320722887e-06, "loss": 2.9251, "step": 228950 }, { "epoch": 0.0175616, "grad_norm": 0.7889397740364075, "learning_rate": 7.342952979798539e-06, "loss": 2.7684, "step": 228960 }, { "epoch": 0.0175872, "grad_norm": 0.8165489435195923, "learning_rate": 7.342176656104434e-06, "loss": 2.9628, "step": 228970 }, { "epoch": 0.0176128, "grad_norm": 0.6771039962768555, "learning_rate": 7.341400349645608e-06, "loss": 2.898, "step": 228980 }, { "epoch": 0.0176384, "grad_norm": 1.0288591384887695, "learning_rate": 7.340624060427095e-06, "loss": 2.8587, "step": 228990 }, { "epoch": 0.017664, "grad_norm": 0.776294469833374, "learning_rate": 7.339847788453928e-06, "loss": 2.962, "step": 229000 }, { "epoch": 0.0176896, "grad_norm": 0.7124413251876831, "learning_rate": 7.33907153373114e-06, "loss": 2.9129, "step": 229010 }, { "epoch": 0.0177152, "grad_norm": 0.7067509293556213, "learning_rate": 7.3382952962637665e-06, "loss": 2.9046, "step": 229020 }, { "epoch": 0.0177408, "grad_norm": 0.7547208666801453, "learning_rate": 7.3375190760568425e-06, "loss": 2.9663, "step": 229030 }, { "epoch": 0.0177664, "grad_norm": 0.7704535722732544, "learning_rate": 7.336742873115397e-06, "loss": 2.7216, "step": 229040 }, { "epoch": 0.017792, "grad_norm": 0.8244872689247131, "learning_rate": 7.335966687444467e-06, "loss": 3.0072, "step": 229050 }, { "epoch": 0.0178176, "grad_norm": 0.8708889484405518, "learning_rate": 7.335190519049083e-06, "loss": 2.8678, "step": 229060 }, { "epoch": 0.0178432, "grad_norm": 0.8970882296562195, "learning_rate": 7.3344143679342805e-06, "loss": 2.9881, "step": 229070 }, { "epoch": 0.0178688, "grad_norm": 0.7991591691970825, "learning_rate": 7.333638234105096e-06, "loss": 2.7915, "step": 229080 }, { "epoch": 0.0178944, "grad_norm": 1.2478082180023193, "learning_rate": 7.332862117566557e-06, "loss": 2.9348, "step": 229090 }, { "epoch": 0.01792, "grad_norm": 0.8163549304008484, "learning_rate": 7.332086018323697e-06, "loss": 2.8114, "step": 229100 }, { "epoch": 0.0179456, "grad_norm": 0.7593935132026672, "learning_rate": 7.331309936381552e-06, "loss": 2.9869, "step": 229110 }, { "epoch": 0.0179712, "grad_norm": 0.7760517597198486, "learning_rate": 7.330533871745154e-06, "loss": 2.9909, "step": 229120 }, { "epoch": 0.0179968, "grad_norm": 1.1424181461334229, "learning_rate": 7.329757824419532e-06, "loss": 2.7563, "step": 229130 }, { "epoch": 0.0180224, "grad_norm": 0.746931254863739, "learning_rate": 7.328981794409722e-06, "loss": 2.9042, "step": 229140 }, { "epoch": 0.018048, "grad_norm": 0.8848418593406677, "learning_rate": 7.3282057817207565e-06, "loss": 2.8974, "step": 229150 }, { "epoch": 0.0180736, "grad_norm": 0.8261959552764893, "learning_rate": 7.327429786357664e-06, "loss": 3.0417, "step": 229160 }, { "epoch": 0.0180992, "grad_norm": 0.804690420627594, "learning_rate": 7.326653808325479e-06, "loss": 2.9144, "step": 229170 }, { "epoch": 0.0181248, "grad_norm": 0.7243397235870361, "learning_rate": 7.325877847629234e-06, "loss": 3.0465, "step": 229180 }, { "epoch": 0.0181504, "grad_norm": 0.750307559967041, "learning_rate": 7.325101904273962e-06, "loss": 2.9016, "step": 229190 }, { "epoch": 0.018176, "grad_norm": 0.8792596459388733, "learning_rate": 7.324325978264692e-06, "loss": 3.0396, "step": 229200 }, { "epoch": 0.0182016, "grad_norm": 0.9874143600463867, "learning_rate": 7.323550069606457e-06, "loss": 3.1073, "step": 229210 }, { "epoch": 0.0182272, "grad_norm": 0.8566623330116272, "learning_rate": 7.322774178304287e-06, "loss": 3.0464, "step": 229220 }, { "epoch": 0.0182528, "grad_norm": 0.7504807114601135, "learning_rate": 7.321998304363218e-06, "loss": 2.9454, "step": 229230 }, { "epoch": 0.0182784, "grad_norm": 0.9339033365249634, "learning_rate": 7.321222447788272e-06, "loss": 2.911, "step": 229240 }, { "epoch": 0.018304, "grad_norm": 0.7934624552726746, "learning_rate": 7.3204466085844925e-06, "loss": 2.8517, "step": 229250 }, { "epoch": 0.0183296, "grad_norm": 1.6297571659088135, "learning_rate": 7.319670786756903e-06, "loss": 2.8273, "step": 229260 }, { "epoch": 0.0183552, "grad_norm": 0.7542644143104553, "learning_rate": 7.318894982310536e-06, "loss": 2.7332, "step": 229270 }, { "epoch": 0.0183808, "grad_norm": 0.7884509563446045, "learning_rate": 7.318119195250425e-06, "loss": 2.9324, "step": 229280 }, { "epoch": 0.0184064, "grad_norm": 0.8387172818183899, "learning_rate": 7.317343425581595e-06, "loss": 3.0143, "step": 229290 }, { "epoch": 0.018432, "grad_norm": 0.929293692111969, "learning_rate": 7.316567673309081e-06, "loss": 2.9431, "step": 229300 }, { "epoch": 0.0184576, "grad_norm": 0.7958167791366577, "learning_rate": 7.3157919384379115e-06, "loss": 2.8137, "step": 229310 }, { "epoch": 0.0184832, "grad_norm": 0.7360419034957886, "learning_rate": 7.315016220973122e-06, "loss": 2.9421, "step": 229320 }, { "epoch": 0.0185088, "grad_norm": 1.1388156414031982, "learning_rate": 7.3142405209197344e-06, "loss": 2.9641, "step": 229330 }, { "epoch": 0.0185344, "grad_norm": 0.7387695908546448, "learning_rate": 7.3134648382827846e-06, "loss": 2.9683, "step": 229340 }, { "epoch": 0.01856, "grad_norm": 0.7692110538482666, "learning_rate": 7.312689173067303e-06, "loss": 3.15, "step": 229350 }, { "epoch": 0.0185856, "grad_norm": 0.9752076268196106, "learning_rate": 7.311913525278315e-06, "loss": 2.9359, "step": 229360 }, { "epoch": 0.0186112, "grad_norm": 0.8087831139564514, "learning_rate": 7.311137894920853e-06, "loss": 2.9833, "step": 229370 }, { "epoch": 0.0186368, "grad_norm": 0.9136424660682678, "learning_rate": 7.310362281999949e-06, "loss": 2.8758, "step": 229380 }, { "epoch": 0.0186624, "grad_norm": 0.8780825138092041, "learning_rate": 7.309586686520631e-06, "loss": 2.7443, "step": 229390 }, { "epoch": 0.018688, "grad_norm": 0.799857497215271, "learning_rate": 7.308811108487928e-06, "loss": 2.8754, "step": 229400 }, { "epoch": 0.0187136, "grad_norm": 1.1245570182800293, "learning_rate": 7.308035547906866e-06, "loss": 2.9599, "step": 229410 }, { "epoch": 0.0187392, "grad_norm": 1.0364131927490234, "learning_rate": 7.307260004782482e-06, "loss": 2.8989, "step": 229420 }, { "epoch": 0.0187648, "grad_norm": 0.9762852787971497, "learning_rate": 7.3064844791198e-06, "loss": 3.0544, "step": 229430 }, { "epoch": 0.0187904, "grad_norm": 0.9655800461769104, "learning_rate": 7.305708970923851e-06, "loss": 2.8565, "step": 229440 }, { "epoch": 0.018816, "grad_norm": 0.8080943822860718, "learning_rate": 7.304933480199666e-06, "loss": 2.8076, "step": 229450 }, { "epoch": 0.0188416, "grad_norm": 0.7440446019172668, "learning_rate": 7.3041580069522664e-06, "loss": 2.9954, "step": 229460 }, { "epoch": 0.0188672, "grad_norm": 0.7899914383888245, "learning_rate": 7.303382551186687e-06, "loss": 3.1202, "step": 229470 }, { "epoch": 0.0188928, "grad_norm": 0.9484474062919617, "learning_rate": 7.302607112907958e-06, "loss": 2.9519, "step": 229480 }, { "epoch": 0.0189184, "grad_norm": 0.7483934164047241, "learning_rate": 7.3018316921211015e-06, "loss": 2.9809, "step": 229490 }, { "epoch": 0.018944, "grad_norm": 0.8917219638824463, "learning_rate": 7.3010562888311495e-06, "loss": 3.0437, "step": 229500 }, { "epoch": 0.0189696, "grad_norm": 0.6964004039764404, "learning_rate": 7.300280903043131e-06, "loss": 2.877, "step": 229510 }, { "epoch": 0.0189952, "grad_norm": 0.885942280292511, "learning_rate": 7.299505534762076e-06, "loss": 2.8621, "step": 229520 }, { "epoch": 0.0190208, "grad_norm": 0.8620332479476929, "learning_rate": 7.298730183993007e-06, "loss": 3.0497, "step": 229530 }, { "epoch": 0.0190464, "grad_norm": 0.8897623419761658, "learning_rate": 7.2979548507409545e-06, "loss": 2.9501, "step": 229540 }, { "epoch": 0.019072, "grad_norm": 0.7378594279289246, "learning_rate": 7.29717953501095e-06, "loss": 2.9351, "step": 229550 }, { "epoch": 0.0190976, "grad_norm": 0.7443456649780273, "learning_rate": 7.296404236808014e-06, "loss": 2.8225, "step": 229560 }, { "epoch": 0.0191232, "grad_norm": 0.8238505125045776, "learning_rate": 7.295628956137179e-06, "loss": 3.0153, "step": 229570 }, { "epoch": 0.0191488, "grad_norm": 0.8877707719802856, "learning_rate": 7.294853693003472e-06, "loss": 2.8415, "step": 229580 }, { "epoch": 0.0191744, "grad_norm": 0.738874614238739, "learning_rate": 7.294078447411917e-06, "loss": 2.7769, "step": 229590 }, { "epoch": 0.0192, "grad_norm": 0.761959969997406, "learning_rate": 7.293303219367547e-06, "loss": 2.9432, "step": 229600 }, { "epoch": 0.0192256, "grad_norm": 0.9899263381958008, "learning_rate": 7.292528008875388e-06, "loss": 2.9978, "step": 229610 }, { "epoch": 0.0192512, "grad_norm": 0.8243181705474854, "learning_rate": 7.291752815940463e-06, "loss": 3.1203, "step": 229620 }, { "epoch": 0.0192768, "grad_norm": 0.8822950720787048, "learning_rate": 7.2909776405678004e-06, "loss": 2.6255, "step": 229630 }, { "epoch": 0.0193024, "grad_norm": 0.698168933391571, "learning_rate": 7.290202482762428e-06, "loss": 2.7119, "step": 229640 }, { "epoch": 0.019328, "grad_norm": 0.7465658187866211, "learning_rate": 7.289427342529376e-06, "loss": 2.7918, "step": 229650 }, { "epoch": 0.0193536, "grad_norm": 0.6802142262458801, "learning_rate": 7.288652219873664e-06, "loss": 2.7922, "step": 229660 }, { "epoch": 0.0193792, "grad_norm": 0.7666388750076294, "learning_rate": 7.2878771148003215e-06, "loss": 2.7312, "step": 229670 }, { "epoch": 0.0194048, "grad_norm": 0.6778931021690369, "learning_rate": 7.287102027314379e-06, "loss": 2.7235, "step": 229680 }, { "epoch": 0.0194304, "grad_norm": 0.7957576513290405, "learning_rate": 7.286326957420855e-06, "loss": 2.9211, "step": 229690 }, { "epoch": 0.019456, "grad_norm": 0.7227603197097778, "learning_rate": 7.2855519051247795e-06, "loss": 2.6754, "step": 229700 }, { "epoch": 0.0194816, "grad_norm": 0.8883431553840637, "learning_rate": 7.284776870431179e-06, "loss": 2.6147, "step": 229710 }, { "epoch": 0.0195072, "grad_norm": 0.7233338356018066, "learning_rate": 7.284001853345082e-06, "loss": 2.5325, "step": 229720 }, { "epoch": 0.0195328, "grad_norm": 0.7583380341529846, "learning_rate": 7.283226853871508e-06, "loss": 2.5402, "step": 229730 }, { "epoch": 0.0195584, "grad_norm": 1.3501025438308716, "learning_rate": 7.282451872015485e-06, "loss": 2.8886, "step": 229740 }, { "epoch": 0.019584, "grad_norm": 0.8754446506500244, "learning_rate": 7.281676907782043e-06, "loss": 2.9741, "step": 229750 }, { "epoch": 0.0196096, "grad_norm": 0.744961678981781, "learning_rate": 7.280901961176196e-06, "loss": 2.6439, "step": 229760 }, { "epoch": 0.0196352, "grad_norm": 0.8643264770507812, "learning_rate": 7.280127032202984e-06, "loss": 2.5894, "step": 229770 }, { "epoch": 0.0196608, "grad_norm": 0.8170880675315857, "learning_rate": 7.279352120867424e-06, "loss": 2.6552, "step": 229780 }, { "epoch": 0.0196864, "grad_norm": 0.7898263335227966, "learning_rate": 7.278577227174541e-06, "loss": 2.6309, "step": 229790 }, { "epoch": 0.019712, "grad_norm": 0.9753612279891968, "learning_rate": 7.277802351129363e-06, "loss": 2.9168, "step": 229800 }, { "epoch": 0.0197376, "grad_norm": 0.7492408752441406, "learning_rate": 7.277027492736915e-06, "loss": 2.5615, "step": 229810 }, { "epoch": 0.0197632, "grad_norm": 1.009777545928955, "learning_rate": 7.2762526520022166e-06, "loss": 2.6943, "step": 229820 }, { "epoch": 0.0197888, "grad_norm": 0.7988572716712952, "learning_rate": 7.275477828930298e-06, "loss": 2.7698, "step": 229830 }, { "epoch": 0.0198144, "grad_norm": 1.025312900543213, "learning_rate": 7.274703023526182e-06, "loss": 2.5737, "step": 229840 }, { "epoch": 0.01984, "grad_norm": 0.7487760186195374, "learning_rate": 7.273928235794891e-06, "loss": 2.6032, "step": 229850 }, { "epoch": 0.0198656, "grad_norm": 0.9865720272064209, "learning_rate": 7.273153465741451e-06, "loss": 2.5661, "step": 229860 }, { "epoch": 0.0198912, "grad_norm": 0.9220285415649414, "learning_rate": 7.272378713370887e-06, "loss": 2.7694, "step": 229870 }, { "epoch": 0.0199168, "grad_norm": 0.8991823792457581, "learning_rate": 7.271603978688222e-06, "loss": 2.6575, "step": 229880 }, { "epoch": 0.0199424, "grad_norm": 0.868608295917511, "learning_rate": 7.2708292616984795e-06, "loss": 2.7441, "step": 229890 }, { "epoch": 0.019968, "grad_norm": 0.7659127712249756, "learning_rate": 7.270054562406684e-06, "loss": 2.7791, "step": 229900 }, { "epoch": 0.0199936, "grad_norm": 0.7610802054405212, "learning_rate": 7.2692798808178585e-06, "loss": 2.5277, "step": 229910 }, { "epoch": 0.0200192, "grad_norm": 0.8165189623832703, "learning_rate": 7.26850521693703e-06, "loss": 2.7743, "step": 229920 }, { "epoch": 0.0200448, "grad_norm": 0.7440963387489319, "learning_rate": 7.267730570769211e-06, "loss": 2.5538, "step": 229930 }, { "epoch": 0.0200704, "grad_norm": 0.9131391644477844, "learning_rate": 7.266955942319442e-06, "loss": 2.7107, "step": 229940 }, { "epoch": 0.020096, "grad_norm": 0.8839331865310669, "learning_rate": 7.266181331592734e-06, "loss": 2.6428, "step": 229950 }, { "epoch": 0.0201216, "grad_norm": 0.7673640251159668, "learning_rate": 7.2654067385941144e-06, "loss": 2.2629, "step": 229960 }, { "epoch": 0.0201472, "grad_norm": 0.8267011046409607, "learning_rate": 7.264632163328606e-06, "loss": 2.8007, "step": 229970 }, { "epoch": 0.0201728, "grad_norm": 0.779476523399353, "learning_rate": 7.26385760580123e-06, "loss": 2.677, "step": 229980 }, { "epoch": 0.0201984, "grad_norm": 1.0322327613830566, "learning_rate": 7.2630830660170094e-06, "loss": 2.7042, "step": 229990 }, { "epoch": 0.020224, "grad_norm": 0.9615327715873718, "learning_rate": 7.2623085439809684e-06, "loss": 2.7545, "step": 230000 }, { "epoch": 0.0202496, "grad_norm": 0.7405501008033752, "learning_rate": 7.261534039698131e-06, "loss": 2.5314, "step": 230010 }, { "epoch": 0.0202752, "grad_norm": 0.7854935526847839, "learning_rate": 7.260759553173516e-06, "loss": 2.6943, "step": 230020 }, { "epoch": 0.0203008, "grad_norm": 0.7550786733627319, "learning_rate": 7.2599850844121455e-06, "loss": 2.5912, "step": 230030 }, { "epoch": 0.0203264, "grad_norm": 0.8032732605934143, "learning_rate": 7.259210633419047e-06, "loss": 2.9459, "step": 230040 }, { "epoch": 0.020352, "grad_norm": 0.8752873539924622, "learning_rate": 7.258436200199237e-06, "loss": 2.5969, "step": 230050 }, { "epoch": 0.0203776, "grad_norm": 0.771593451499939, "learning_rate": 7.257661784757739e-06, "loss": 2.5326, "step": 230060 }, { "epoch": 0.0204032, "grad_norm": 1.00259268283844, "learning_rate": 7.256887387099576e-06, "loss": 3.5078, "step": 230070 }, { "epoch": 0.0204288, "grad_norm": 0.8206866979598999, "learning_rate": 7.256113007229771e-06, "loss": 2.8838, "step": 230080 }, { "epoch": 0.0204544, "grad_norm": 0.7464618682861328, "learning_rate": 7.255338645153343e-06, "loss": 2.9582, "step": 230090 }, { "epoch": 0.02048, "grad_norm": 0.8009199500083923, "learning_rate": 7.254564300875313e-06, "loss": 2.837, "step": 230100 }, { "epoch": 0.0205056, "grad_norm": 0.808277428150177, "learning_rate": 7.253789974400703e-06, "loss": 2.8104, "step": 230110 }, { "epoch": 0.0205312, "grad_norm": 0.8560925126075745, "learning_rate": 7.253015665734535e-06, "loss": 2.6874, "step": 230120 }, { "epoch": 0.0205568, "grad_norm": 0.9743252396583557, "learning_rate": 7.252241374881832e-06, "loss": 2.7411, "step": 230130 }, { "epoch": 0.0205824, "grad_norm": 1.4250730276107788, "learning_rate": 7.251467101847616e-06, "loss": 2.9156, "step": 230140 }, { "epoch": 0.020608, "grad_norm": 0.8070447444915771, "learning_rate": 7.250692846636902e-06, "loss": 2.6052, "step": 230150 }, { "epoch": 0.0206336, "grad_norm": 1.0414437055587769, "learning_rate": 7.249918609254715e-06, "loss": 2.6261, "step": 230160 }, { "epoch": 0.0206592, "grad_norm": 0.7629665732383728, "learning_rate": 7.249144389706078e-06, "loss": 2.6583, "step": 230170 }, { "epoch": 0.0206848, "grad_norm": 0.8029323220252991, "learning_rate": 7.248370187996004e-06, "loss": 2.5208, "step": 230180 }, { "epoch": 0.0207104, "grad_norm": 0.9155033230781555, "learning_rate": 7.24759600412952e-06, "loss": 2.3347, "step": 230190 }, { "epoch": 0.020736, "grad_norm": 0.834494948387146, "learning_rate": 7.246821838111644e-06, "loss": 2.6883, "step": 230200 }, { "epoch": 0.0207616, "grad_norm": 0.7282970547676086, "learning_rate": 7.246047689947397e-06, "loss": 2.7351, "step": 230210 }, { "epoch": 0.0207872, "grad_norm": 0.8325924873352051, "learning_rate": 7.245273559641799e-06, "loss": 2.7372, "step": 230220 }, { "epoch": 0.0208128, "grad_norm": 0.7995005249977112, "learning_rate": 7.2444994471998685e-06, "loss": 2.8433, "step": 230230 }, { "epoch": 0.0208384, "grad_norm": 0.7953996062278748, "learning_rate": 7.24372535262663e-06, "loss": 2.6861, "step": 230240 }, { "epoch": 0.020864, "grad_norm": 0.8912025690078735, "learning_rate": 7.242951275927097e-06, "loss": 2.7291, "step": 230250 }, { "epoch": 0.0208896, "grad_norm": 0.8286469578742981, "learning_rate": 7.242177217106293e-06, "loss": 2.7572, "step": 230260 }, { "epoch": 0.0209152, "grad_norm": 1.0771892070770264, "learning_rate": 7.2414031761692375e-06, "loss": 2.8502, "step": 230270 }, { "epoch": 0.0209408, "grad_norm": 0.7913247346878052, "learning_rate": 7.240629153120946e-06, "loss": 2.8262, "step": 230280 }, { "epoch": 0.0209664, "grad_norm": 0.7529346942901611, "learning_rate": 7.2398551479664425e-06, "loss": 2.7907, "step": 230290 }, { "epoch": 0.020992, "grad_norm": 0.7693561911582947, "learning_rate": 7.239081160710748e-06, "loss": 2.6366, "step": 230300 }, { "epoch": 2.56e-05, "grad_norm": 0.7388443350791931, "learning_rate": 7.238307191358875e-06, "loss": 2.2501, "step": 230310 }, { "epoch": 5.12e-05, "grad_norm": 0.8457381725311279, "learning_rate": 7.237533239915848e-06, "loss": 2.6059, "step": 230320 }, { "epoch": 7.68e-05, "grad_norm": 0.7686026692390442, "learning_rate": 7.236759306386683e-06, "loss": 2.7509, "step": 230330 }, { "epoch": 0.0001024, "grad_norm": 1.304432988166809, "learning_rate": 7.235985390776401e-06, "loss": 2.5821, "step": 230340 }, { "epoch": 0.000128, "grad_norm": 0.7736934423446655, "learning_rate": 7.2352114930900174e-06, "loss": 2.8408, "step": 230350 }, { "epoch": 0.0001536, "grad_norm": 0.8355465531349182, "learning_rate": 7.2344376133325524e-06, "loss": 2.7302, "step": 230360 }, { "epoch": 0.0001792, "grad_norm": 0.7937565445899963, "learning_rate": 7.2336637515090255e-06, "loss": 2.6326, "step": 230370 }, { "epoch": 0.0002048, "grad_norm": 0.7774593830108643, "learning_rate": 7.2328899076244544e-06, "loss": 2.87, "step": 230380 }, { "epoch": 0.0002304, "grad_norm": 0.8131508827209473, "learning_rate": 7.232116081683855e-06, "loss": 2.7279, "step": 230390 }, { "epoch": 0.000256, "grad_norm": 0.7700799107551575, "learning_rate": 7.231342273692248e-06, "loss": 2.6753, "step": 230400 }, { "epoch": 0.0002816, "grad_norm": 0.817956805229187, "learning_rate": 7.230568483654652e-06, "loss": 2.6433, "step": 230410 }, { "epoch": 0.0003072, "grad_norm": 0.8137132525444031, "learning_rate": 7.229794711576081e-06, "loss": 2.6386, "step": 230420 }, { "epoch": 0.0003328, "grad_norm": 0.8058715462684631, "learning_rate": 7.229020957461557e-06, "loss": 2.608, "step": 230430 }, { "epoch": 0.0003584, "grad_norm": 0.8369467854499817, "learning_rate": 7.2282472213160955e-06, "loss": 2.679, "step": 230440 }, { "epoch": 0.000384, "grad_norm": 0.8455249071121216, "learning_rate": 7.227473503144713e-06, "loss": 2.6265, "step": 230450 }, { "epoch": 0.0004096, "grad_norm": 1.7101192474365234, "learning_rate": 7.226699802952426e-06, "loss": 3.0542, "step": 230460 }, { "epoch": 0.0004352, "grad_norm": 0.7872796058654785, "learning_rate": 7.225926120744255e-06, "loss": 2.7772, "step": 230470 }, { "epoch": 0.0004608, "grad_norm": 0.7252117991447449, "learning_rate": 7.2251524565252165e-06, "loss": 2.673, "step": 230480 }, { "epoch": 0.0004864, "grad_norm": 0.8842104077339172, "learning_rate": 7.224378810300326e-06, "loss": 2.6821, "step": 230490 }, { "epoch": 0.000512, "grad_norm": 0.8658633232116699, "learning_rate": 7.223605182074604e-06, "loss": 2.6379, "step": 230500 }, { "epoch": 0.0005376, "grad_norm": 0.7772956490516663, "learning_rate": 7.222831571853063e-06, "loss": 2.7321, "step": 230510 }, { "epoch": 0.0005632, "grad_norm": 0.8575543761253357, "learning_rate": 7.222057979640721e-06, "loss": 2.7676, "step": 230520 }, { "epoch": 0.0005888, "grad_norm": 1.0599820613861084, "learning_rate": 7.221284405442595e-06, "loss": 2.5692, "step": 230530 }, { "epoch": 0.0006144, "grad_norm": 0.8601192831993103, "learning_rate": 7.220510849263702e-06, "loss": 2.7147, "step": 230540 }, { "epoch": 0.00064, "grad_norm": 0.8059431314468384, "learning_rate": 7.219737311109056e-06, "loss": 2.5316, "step": 230550 }, { "epoch": 0.0006656, "grad_norm": 1.0073803663253784, "learning_rate": 7.218963790983674e-06, "loss": 2.6723, "step": 230560 }, { "epoch": 0.0006912, "grad_norm": 0.7782091498374939, "learning_rate": 7.2181902888925755e-06, "loss": 3.0428, "step": 230570 }, { "epoch": 0.0007168, "grad_norm": 0.7914935350418091, "learning_rate": 7.217416804840771e-06, "loss": 2.4796, "step": 230580 }, { "epoch": 0.0007424, "grad_norm": 0.9011908173561096, "learning_rate": 7.2166433388332805e-06, "loss": 2.6436, "step": 230590 }, { "epoch": 0.000768, "grad_norm": 0.8114417195320129, "learning_rate": 7.215869890875117e-06, "loss": 2.7657, "step": 230600 }, { "epoch": 0.0007936, "grad_norm": 0.7764653563499451, "learning_rate": 7.2150964609713e-06, "loss": 2.6622, "step": 230610 }, { "epoch": 0.0008192, "grad_norm": 0.7391345500946045, "learning_rate": 7.214323049126839e-06, "loss": 2.4386, "step": 230620 }, { "epoch": 0.0008448, "grad_norm": 0.914460301399231, "learning_rate": 7.213549655346752e-06, "loss": 2.6889, "step": 230630 }, { "epoch": 0.0008704, "grad_norm": 0.8830307722091675, "learning_rate": 7.2127762796360555e-06, "loss": 2.7557, "step": 230640 }, { "epoch": 0.000896, "grad_norm": 0.8710086345672607, "learning_rate": 7.212002921999765e-06, "loss": 2.6609, "step": 230650 }, { "epoch": 0.0009216, "grad_norm": 1.4847992658615112, "learning_rate": 7.211229582442898e-06, "loss": 2.845, "step": 230660 }, { "epoch": 0.0009472, "grad_norm": 0.8869988918304443, "learning_rate": 7.210456260970462e-06, "loss": 2.8061, "step": 230670 }, { "epoch": 0.0009728, "grad_norm": 0.8125653862953186, "learning_rate": 7.209682957587476e-06, "loss": 2.9321, "step": 230680 }, { "epoch": 0.0009984, "grad_norm": 0.7484506368637085, "learning_rate": 7.208909672298956e-06, "loss": 2.6302, "step": 230690 }, { "epoch": 0.001024, "grad_norm": 0.796440064907074, "learning_rate": 7.2081364051099155e-06, "loss": 2.7389, "step": 230700 }, { "epoch": 0.0010496, "grad_norm": 0.7588115930557251, "learning_rate": 7.2073631560253665e-06, "loss": 2.628, "step": 230710 }, { "epoch": 0.0010752, "grad_norm": 0.8778071403503418, "learning_rate": 7.206589925050324e-06, "loss": 2.6742, "step": 230720 }, { "epoch": 0.0011008, "grad_norm": 0.743353009223938, "learning_rate": 7.205816712189806e-06, "loss": 2.655, "step": 230730 }, { "epoch": 0.0011264, "grad_norm": 0.9568703174591064, "learning_rate": 7.205043517448826e-06, "loss": 2.9488, "step": 230740 }, { "epoch": 0.001152, "grad_norm": 0.8247199058532715, "learning_rate": 7.204270340832392e-06, "loss": 2.5696, "step": 230750 }, { "epoch": 0.0011776, "grad_norm": 0.9269512295722961, "learning_rate": 7.203497182345522e-06, "loss": 2.8103, "step": 230760 }, { "epoch": 0.0012032, "grad_norm": 0.8063720464706421, "learning_rate": 7.202724041993233e-06, "loss": 2.8291, "step": 230770 }, { "epoch": 0.0012288, "grad_norm": 0.789323091506958, "learning_rate": 7.201950919780533e-06, "loss": 2.7346, "step": 230780 }, { "epoch": 0.0012544, "grad_norm": 0.8448060750961304, "learning_rate": 7.201177815712437e-06, "loss": 2.9253, "step": 230790 }, { "epoch": 0.00128, "grad_norm": 0.7920942902565002, "learning_rate": 7.20040472979396e-06, "loss": 2.5156, "step": 230800 }, { "epoch": 0.0013056, "grad_norm": 0.8269293904304504, "learning_rate": 7.199631662030111e-06, "loss": 2.556, "step": 230810 }, { "epoch": 0.0013312, "grad_norm": 0.8257313370704651, "learning_rate": 7.198858612425908e-06, "loss": 2.838, "step": 230820 }, { "epoch": 0.0013568, "grad_norm": 0.8503432869911194, "learning_rate": 7.198085580986365e-06, "loss": 2.8171, "step": 230830 }, { "epoch": 0.0013824, "grad_norm": 0.886343777179718, "learning_rate": 7.197312567716491e-06, "loss": 2.6692, "step": 230840 }, { "epoch": 0.001408, "grad_norm": 0.784568727016449, "learning_rate": 7.196539572621299e-06, "loss": 3.1566, "step": 230850 }, { "epoch": 0.0014336, "grad_norm": 0.8039040565490723, "learning_rate": 7.195766595705806e-06, "loss": 2.6779, "step": 230860 }, { "epoch": 0.0014592, "grad_norm": 0.7660006284713745, "learning_rate": 7.1949936369750196e-06, "loss": 2.4674, "step": 230870 }, { "epoch": 0.0014848, "grad_norm": 0.7465877532958984, "learning_rate": 7.194220696433953e-06, "loss": 2.5908, "step": 230880 }, { "epoch": 0.0015104, "grad_norm": 0.9869976043701172, "learning_rate": 7.19344777408762e-06, "loss": 2.469, "step": 230890 }, { "epoch": 0.001536, "grad_norm": 0.893920361995697, "learning_rate": 7.192674869941034e-06, "loss": 2.397, "step": 230900 }, { "epoch": 2.56e-05, "grad_norm": 0.8645808100700378, "learning_rate": 7.191901983999203e-06, "loss": 1.9981, "step": 230910 }, { "epoch": 5.12e-05, "grad_norm": 0.9017747640609741, "learning_rate": 7.191129116267141e-06, "loss": 2.8782, "step": 230920 }, { "epoch": 7.68e-05, "grad_norm": 0.8058680891990662, "learning_rate": 7.190356266749865e-06, "loss": 2.5396, "step": 230930 }, { "epoch": 0.0001024, "grad_norm": 0.8447227478027344, "learning_rate": 7.1895834354523765e-06, "loss": 2.7643, "step": 230940 }, { "epoch": 0.000128, "grad_norm": 0.8871160745620728, "learning_rate": 7.188810622379694e-06, "loss": 3.077, "step": 230950 }, { "epoch": 0.0001536, "grad_norm": 0.8381379246711731, "learning_rate": 7.188037827536828e-06, "loss": 2.866, "step": 230960 }, { "epoch": 0.0001792, "grad_norm": 0.9692832827568054, "learning_rate": 7.187265050928791e-06, "loss": 2.802, "step": 230970 }, { "epoch": 0.0002048, "grad_norm": 0.8630526065826416, "learning_rate": 7.186492292560585e-06, "loss": 2.7724, "step": 230980 }, { "epoch": 0.0002304, "grad_norm": 0.8841404318809509, "learning_rate": 7.185719552437237e-06, "loss": 2.7889, "step": 230990 }, { "epoch": 0.000256, "grad_norm": 0.9597904086112976, "learning_rate": 7.184946830563746e-06, "loss": 2.7248, "step": 231000 }, { "epoch": 0.0002816, "grad_norm": 0.7598053216934204, "learning_rate": 7.184174126945128e-06, "loss": 2.6954, "step": 231010 }, { "epoch": 0.0003072, "grad_norm": 0.7441298365592957, "learning_rate": 7.183401441586392e-06, "loss": 2.6724, "step": 231020 }, { "epoch": 0.0003328, "grad_norm": 0.844735860824585, "learning_rate": 7.1826287744925505e-06, "loss": 2.7442, "step": 231030 }, { "epoch": 0.0003584, "grad_norm": 0.7837151288986206, "learning_rate": 7.181856125668611e-06, "loss": 2.7623, "step": 231040 }, { "epoch": 0.000384, "grad_norm": 0.7672107219696045, "learning_rate": 7.181083495119586e-06, "loss": 2.6465, "step": 231050 }, { "epoch": 0.0004096, "grad_norm": 0.7397451400756836, "learning_rate": 7.180310882850487e-06, "loss": 2.6534, "step": 231060 }, { "epoch": 0.0004352, "grad_norm": 0.834987223148346, "learning_rate": 7.179538288866321e-06, "loss": 3.009, "step": 231070 }, { "epoch": 0.0004608, "grad_norm": 0.839194655418396, "learning_rate": 7.178765713172099e-06, "loss": 2.6521, "step": 231080 }, { "epoch": 0.0004864, "grad_norm": 0.7857689261436462, "learning_rate": 7.177993155772832e-06, "loss": 2.64, "step": 231090 }, { "epoch": 0.000512, "grad_norm": 0.8248437643051147, "learning_rate": 7.17722061667353e-06, "loss": 2.5825, "step": 231100 }, { "epoch": 0.0005376, "grad_norm": 0.7552452087402344, "learning_rate": 7.1764480958792025e-06, "loss": 2.6911, "step": 231110 }, { "epoch": 0.0005632, "grad_norm": 0.7901179790496826, "learning_rate": 7.175675593394858e-06, "loss": 2.5957, "step": 231120 }, { "epoch": 0.0005888, "grad_norm": 0.8156298995018005, "learning_rate": 7.174903109225507e-06, "loss": 2.6496, "step": 231130 }, { "epoch": 0.0006144, "grad_norm": 0.9751638174057007, "learning_rate": 7.174130643376158e-06, "loss": 2.7253, "step": 231140 }, { "epoch": 0.00064, "grad_norm": 0.8965615630149841, "learning_rate": 7.1733581958518186e-06, "loss": 2.6831, "step": 231150 }, { "epoch": 0.0006656, "grad_norm": 0.8182575702667236, "learning_rate": 7.172585766657501e-06, "loss": 2.8129, "step": 231160 }, { "epoch": 0.0006912, "grad_norm": 0.8429032564163208, "learning_rate": 7.171813355798214e-06, "loss": 2.7888, "step": 231170 }, { "epoch": 0.0007168, "grad_norm": 0.8669342398643494, "learning_rate": 7.171040963278965e-06, "loss": 2.8059, "step": 231180 }, { "epoch": 0.0007424, "grad_norm": 0.798252284526825, "learning_rate": 7.170268589104767e-06, "loss": 2.6788, "step": 231190 }, { "epoch": 0.000768, "grad_norm": 0.8781235218048096, "learning_rate": 7.169496233280623e-06, "loss": 2.7397, "step": 231200 }, { "epoch": 0.0007936, "grad_norm": 0.806671142578125, "learning_rate": 7.168723895811543e-06, "loss": 2.4231, "step": 231210 }, { "epoch": 0.0008192, "grad_norm": 0.7356632947921753, "learning_rate": 7.167951576702536e-06, "loss": 2.6514, "step": 231220 }, { "epoch": 0.0008448, "grad_norm": 1.0675209760665894, "learning_rate": 7.1671792759586135e-06, "loss": 2.8103, "step": 231230 }, { "epoch": 0.0008704, "grad_norm": 0.8386954665184021, "learning_rate": 7.166406993584777e-06, "loss": 2.5024, "step": 231240 }, { "epoch": 0.000896, "grad_norm": 0.7574783563613892, "learning_rate": 7.165634729586038e-06, "loss": 2.4564, "step": 231250 }, { "epoch": 0.0009216, "grad_norm": 0.6657108068466187, "learning_rate": 7.1648624839674075e-06, "loss": 2.5556, "step": 231260 }, { "epoch": 0.0009472, "grad_norm": 0.8047541379928589, "learning_rate": 7.164090256733889e-06, "loss": 2.5541, "step": 231270 }, { "epoch": 0.0009728, "grad_norm": 0.8109795451164246, "learning_rate": 7.1633180478904896e-06, "loss": 2.5605, "step": 231280 }, { "epoch": 0.0009984, "grad_norm": 0.9509296417236328, "learning_rate": 7.16254585744222e-06, "loss": 2.674, "step": 231290 }, { "epoch": 0.001024, "grad_norm": 0.8318233489990234, "learning_rate": 7.161773685394088e-06, "loss": 2.788, "step": 231300 }, { "epoch": 0.0010496, "grad_norm": 0.8188078999519348, "learning_rate": 7.1610015317510974e-06, "loss": 2.7561, "step": 231310 }, { "epoch": 0.0010752, "grad_norm": 0.8513484597206116, "learning_rate": 7.160229396518257e-06, "loss": 2.6992, "step": 231320 }, { "epoch": 0.0011008, "grad_norm": 0.8445260524749756, "learning_rate": 7.159457279700572e-06, "loss": 2.6875, "step": 231330 }, { "epoch": 0.0011264, "grad_norm": 0.7935020327568054, "learning_rate": 7.158685181303054e-06, "loss": 2.991, "step": 231340 }, { "epoch": 0.001152, "grad_norm": 0.7857428789138794, "learning_rate": 7.15791310133071e-06, "loss": 2.6001, "step": 231350 }, { "epoch": 0.0011776, "grad_norm": 0.7448939085006714, "learning_rate": 7.157141039788542e-06, "loss": 2.7226, "step": 231360 }, { "epoch": 0.0012032, "grad_norm": 0.8317792415618896, "learning_rate": 7.156368996681559e-06, "loss": 3.0065, "step": 231370 }, { "epoch": 0.0012288, "grad_norm": 0.816429853439331, "learning_rate": 7.1555969720147665e-06, "loss": 2.7474, "step": 231380 }, { "epoch": 0.0012544, "grad_norm": 0.8255901336669922, "learning_rate": 7.154824965793175e-06, "loss": 2.8793, "step": 231390 }, { "epoch": 0.00128, "grad_norm": 1.0146287679672241, "learning_rate": 7.154052978021784e-06, "loss": 2.9093, "step": 231400 }, { "epoch": 0.0013056, "grad_norm": 0.824729859828949, "learning_rate": 7.153281008705603e-06, "loss": 2.6978, "step": 231410 }, { "epoch": 0.0013312, "grad_norm": 1.0043540000915527, "learning_rate": 7.152509057849639e-06, "loss": 2.4999, "step": 231420 }, { "epoch": 0.0013568, "grad_norm": 0.8434889316558838, "learning_rate": 7.1517371254589e-06, "loss": 2.8623, "step": 231430 }, { "epoch": 0.0013824, "grad_norm": 0.7986765503883362, "learning_rate": 7.150965211538385e-06, "loss": 2.4236, "step": 231440 }, { "epoch": 0.001408, "grad_norm": 0.7177174091339111, "learning_rate": 7.150193316093104e-06, "loss": 2.8684, "step": 231450 }, { "epoch": 0.0014336, "grad_norm": 0.832953155040741, "learning_rate": 7.149421439128064e-06, "loss": 2.5559, "step": 231460 }, { "epoch": 0.0014592, "grad_norm": 0.8144771456718445, "learning_rate": 7.148649580648266e-06, "loss": 2.566, "step": 231470 }, { "epoch": 0.0014848, "grad_norm": 1.0628728866577148, "learning_rate": 7.147877740658717e-06, "loss": 2.202, "step": 231480 }, { "epoch": 0.0015104, "grad_norm": 0.8714337348937988, "learning_rate": 7.147105919164424e-06, "loss": 2.6216, "step": 231490 }, { "epoch": 0.001536, "grad_norm": 0.7594228982925415, "learning_rate": 7.146334116170389e-06, "loss": 2.6961, "step": 231500 }, { "epoch": 2.56e-05, "grad_norm": 0.8284110426902771, "learning_rate": 7.1455623316816215e-06, "loss": 2.1344, "step": 231510 }, { "epoch": 5.12e-05, "grad_norm": 0.7984039783477783, "learning_rate": 7.144790565703124e-06, "loss": 2.8166, "step": 231520 }, { "epoch": 7.68e-05, "grad_norm": 0.7275964617729187, "learning_rate": 7.144018818239899e-06, "loss": 2.5242, "step": 231530 }, { "epoch": 0.0001024, "grad_norm": 0.7305235266685486, "learning_rate": 7.143247089296954e-06, "loss": 2.5556, "step": 231540 }, { "epoch": 0.000128, "grad_norm": 0.8378111720085144, "learning_rate": 7.142475378879295e-06, "loss": 2.7558, "step": 231550 }, { "epoch": 0.0001536, "grad_norm": 0.7604773044586182, "learning_rate": 7.1417036869919186e-06, "loss": 2.7116, "step": 231560 }, { "epoch": 0.0001792, "grad_norm": 0.9170324802398682, "learning_rate": 7.1409320136398365e-06, "loss": 2.5813, "step": 231570 }, { "epoch": 0.0002048, "grad_norm": 0.8492838144302368, "learning_rate": 7.14016035882805e-06, "loss": 2.7509, "step": 231580 }, { "epoch": 0.0002304, "grad_norm": 0.848038375377655, "learning_rate": 7.139388722561567e-06, "loss": 2.7028, "step": 231590 }, { "epoch": 0.000256, "grad_norm": 0.8567869067192078, "learning_rate": 7.138617104845384e-06, "loss": 2.722, "step": 231600 }, { "epoch": 0.0002816, "grad_norm": 0.8552805185317993, "learning_rate": 7.137845505684508e-06, "loss": 2.6758, "step": 231610 }, { "epoch": 0.0003072, "grad_norm": 0.851074755191803, "learning_rate": 7.137073925083945e-06, "loss": 2.5342, "step": 231620 }, { "epoch": 0.0003328, "grad_norm": 0.816504955291748, "learning_rate": 7.1363023630486984e-06, "loss": 2.9, "step": 231630 }, { "epoch": 0.0003584, "grad_norm": 0.8041521310806274, "learning_rate": 7.135530819583767e-06, "loss": 2.8861, "step": 231640 }, { "epoch": 0.000384, "grad_norm": 0.9597709774971008, "learning_rate": 7.134759294694158e-06, "loss": 2.6534, "step": 231650 }, { "epoch": 0.0004096, "grad_norm": 0.9026467800140381, "learning_rate": 7.133987788384875e-06, "loss": 2.774, "step": 231660 }, { "epoch": 0.0004352, "grad_norm": 0.9768804311752319, "learning_rate": 7.133216300660917e-06, "loss": 2.6355, "step": 231670 }, { "epoch": 0.0004608, "grad_norm": 0.7857440114021301, "learning_rate": 7.1324448315272875e-06, "loss": 2.646, "step": 231680 }, { "epoch": 0.0004864, "grad_norm": 0.9606079459190369, "learning_rate": 7.131673380988993e-06, "loss": 2.95, "step": 231690 }, { "epoch": 0.000512, "grad_norm": 0.7594825625419617, "learning_rate": 7.130901949051035e-06, "loss": 2.4701, "step": 231700 }, { "epoch": 0.0005376, "grad_norm": 0.820013701915741, "learning_rate": 7.130130535718416e-06, "loss": 2.6557, "step": 231710 }, { "epoch": 0.0005632, "grad_norm": 0.7681928873062134, "learning_rate": 7.129359140996139e-06, "loss": 2.527, "step": 231720 }, { "epoch": 0.0005888, "grad_norm": 0.840410053730011, "learning_rate": 7.128587764889203e-06, "loss": 2.4605, "step": 231730 }, { "epoch": 0.0006144, "grad_norm": 0.8515318036079407, "learning_rate": 7.127816407402612e-06, "loss": 2.8074, "step": 231740 }, { "epoch": 0.00064, "grad_norm": 1.013410210609436, "learning_rate": 7.12704506854137e-06, "loss": 2.9678, "step": 231750 }, { "epoch": 0.0006656, "grad_norm": 0.8566061854362488, "learning_rate": 7.126273748310476e-06, "loss": 2.8026, "step": 231760 }, { "epoch": 0.0006912, "grad_norm": 0.7686049342155457, "learning_rate": 7.1255024467149335e-06, "loss": 3.1544, "step": 231770 }, { "epoch": 0.0007168, "grad_norm": 0.8237780332565308, "learning_rate": 7.124731163759741e-06, "loss": 2.6662, "step": 231780 }, { "epoch": 0.0007424, "grad_norm": 0.7792670726776123, "learning_rate": 7.1239598994499085e-06, "loss": 2.5558, "step": 231790 }, { "epoch": 0.000768, "grad_norm": 0.8292564749717712, "learning_rate": 7.1231886537904275e-06, "loss": 2.5986, "step": 231800 }, { "epoch": 0.0007936, "grad_norm": 0.8209124803543091, "learning_rate": 7.122417426786303e-06, "loss": 2.7111, "step": 231810 }, { "epoch": 0.0008192, "grad_norm": 0.8032427430152893, "learning_rate": 7.1216462184425394e-06, "loss": 2.68, "step": 231820 }, { "epoch": 0.0008448, "grad_norm": 0.8121094107627869, "learning_rate": 7.120875028764133e-06, "loss": 2.6474, "step": 231830 }, { "epoch": 0.0008704, "grad_norm": 0.7518374919891357, "learning_rate": 7.120103857756088e-06, "loss": 2.687, "step": 231840 }, { "epoch": 0.000896, "grad_norm": 0.8404531478881836, "learning_rate": 7.1193327054234006e-06, "loss": 2.6325, "step": 231850 }, { "epoch": 0.0009216, "grad_norm": 0.8195745348930359, "learning_rate": 7.118561571771077e-06, "loss": 2.5488, "step": 231860 }, { "epoch": 0.0009472, "grad_norm": 0.7054311037063599, "learning_rate": 7.117790456804117e-06, "loss": 2.2952, "step": 231870 }, { "epoch": 0.0009728, "grad_norm": 0.845890462398529, "learning_rate": 7.117019360527522e-06, "loss": 2.6949, "step": 231880 }, { "epoch": 0.0009984, "grad_norm": 0.8153350353240967, "learning_rate": 7.116248282946288e-06, "loss": 2.9829, "step": 231890 }, { "epoch": 0.001024, "grad_norm": 0.9575459361076355, "learning_rate": 7.115477224065416e-06, "loss": 2.9849, "step": 231900 }, { "epoch": 0.0010496, "grad_norm": 0.8557880520820618, "learning_rate": 7.114706183889908e-06, "loss": 2.7349, "step": 231910 }, { "epoch": 0.0010752, "grad_norm": 0.9404239058494568, "learning_rate": 7.113935162424767e-06, "loss": 3.0028, "step": 231920 }, { "epoch": 0.0011008, "grad_norm": 0.7833629846572876, "learning_rate": 7.113164159674985e-06, "loss": 2.7592, "step": 231930 }, { "epoch": 0.0011264, "grad_norm": 0.7493608593940735, "learning_rate": 7.112393175645569e-06, "loss": 2.7264, "step": 231940 }, { "epoch": 0.001152, "grad_norm": 0.7759078145027161, "learning_rate": 7.111622210341517e-06, "loss": 2.9242, "step": 231950 }, { "epoch": 0.0011776, "grad_norm": 0.8002473711967468, "learning_rate": 7.110851263767824e-06, "loss": 2.9313, "step": 231960 }, { "epoch": 0.0012032, "grad_norm": 0.8035088777542114, "learning_rate": 7.110080335929494e-06, "loss": 2.8287, "step": 231970 }, { "epoch": 0.0012288, "grad_norm": 0.8459159731864929, "learning_rate": 7.109309426831525e-06, "loss": 2.7956, "step": 231980 }, { "epoch": 0.0012544, "grad_norm": 0.9150171875953674, "learning_rate": 7.108538536478917e-06, "loss": 2.547, "step": 231990 }, { "epoch": 0.00128, "grad_norm": 0.830092191696167, "learning_rate": 7.107767664876668e-06, "loss": 2.6371, "step": 232000 }, { "epoch": 0.0013056, "grad_norm": 0.721409797668457, "learning_rate": 7.106996812029776e-06, "loss": 2.5785, "step": 232010 }, { "epoch": 0.0013312, "grad_norm": 0.8739949464797974, "learning_rate": 7.106225977943238e-06, "loss": 2.511, "step": 232020 }, { "epoch": 0.0013568, "grad_norm": 0.9490314722061157, "learning_rate": 7.105455162622058e-06, "loss": 2.7033, "step": 232030 }, { "epoch": 0.0013824, "grad_norm": 0.8108749389648438, "learning_rate": 7.104684366071235e-06, "loss": 2.8488, "step": 232040 }, { "epoch": 0.001408, "grad_norm": 0.8059520125389099, "learning_rate": 7.103913588295761e-06, "loss": 2.6208, "step": 232050 }, { "epoch": 0.0014336, "grad_norm": 0.7784737944602966, "learning_rate": 7.1031428293006375e-06, "loss": 2.7928, "step": 232060 }, { "epoch": 0.0014592, "grad_norm": 0.8091699481010437, "learning_rate": 7.1023720890908644e-06, "loss": 2.6405, "step": 232070 }, { "epoch": 0.0014848, "grad_norm": 0.9327775239944458, "learning_rate": 7.10160136767144e-06, "loss": 2.4239, "step": 232080 }, { "epoch": 0.0015104, "grad_norm": 0.871576726436615, "learning_rate": 7.100830665047358e-06, "loss": 2.3919, "step": 232090 }, { "epoch": 0.001536, "grad_norm": 0.8916705250740051, "learning_rate": 7.100059981223617e-06, "loss": 2.4953, "step": 232100 }, { "epoch": 2.56e-05, "grad_norm": 0.7680124640464783, "learning_rate": 7.099289316205218e-06, "loss": 1.9586, "step": 232110 }, { "epoch": 5.12e-05, "grad_norm": 0.9224613308906555, "learning_rate": 7.098518669997158e-06, "loss": 2.846, "step": 232120 }, { "epoch": 7.68e-05, "grad_norm": 0.7393319010734558, "learning_rate": 7.097748042604433e-06, "loss": 2.7033, "step": 232130 }, { "epoch": 0.0001024, "grad_norm": 0.9510853886604309, "learning_rate": 7.09697743403204e-06, "loss": 2.7257, "step": 232140 }, { "epoch": 0.000128, "grad_norm": 0.851387083530426, "learning_rate": 7.096206844284979e-06, "loss": 2.7274, "step": 232150 }, { "epoch": 0.0001536, "grad_norm": 0.8903861045837402, "learning_rate": 7.095436273368241e-06, "loss": 2.9318, "step": 232160 }, { "epoch": 0.0001792, "grad_norm": 0.8609397411346436, "learning_rate": 7.094665721286829e-06, "loss": 2.7431, "step": 232170 }, { "epoch": 0.0002048, "grad_norm": 0.8148273825645447, "learning_rate": 7.093895188045736e-06, "loss": 2.9193, "step": 232180 }, { "epoch": 0.0002304, "grad_norm": 0.8051674962043762, "learning_rate": 7.093124673649964e-06, "loss": 2.6028, "step": 232190 }, { "epoch": 0.000256, "grad_norm": 0.7421225905418396, "learning_rate": 7.092354178104499e-06, "loss": 2.7785, "step": 232200 }, { "epoch": 0.0002816, "grad_norm": 0.7737879157066345, "learning_rate": 7.091583701414351e-06, "loss": 2.5679, "step": 232210 }, { "epoch": 0.0003072, "grad_norm": 0.769432544708252, "learning_rate": 7.090813243584507e-06, "loss": 2.6007, "step": 232220 }, { "epoch": 0.0003328, "grad_norm": 0.8067291975021362, "learning_rate": 7.090042804619965e-06, "loss": 2.5547, "step": 232230 }, { "epoch": 0.0003584, "grad_norm": 0.799823522567749, "learning_rate": 7.089272384525725e-06, "loss": 2.8409, "step": 232240 }, { "epoch": 0.000384, "grad_norm": 0.8225572109222412, "learning_rate": 7.088501983306779e-06, "loss": 2.6974, "step": 232250 }, { "epoch": 0.0004096, "grad_norm": 0.8104720115661621, "learning_rate": 7.087731600968122e-06, "loss": 2.5341, "step": 232260 }, { "epoch": 0.0004352, "grad_norm": 0.7834747433662415, "learning_rate": 7.086961237514754e-06, "loss": 2.7327, "step": 232270 }, { "epoch": 0.0004608, "grad_norm": 0.7889052033424377, "learning_rate": 7.086190892951667e-06, "loss": 2.4917, "step": 232280 }, { "epoch": 0.0004864, "grad_norm": 0.9767167568206787, "learning_rate": 7.085420567283858e-06, "loss": 2.645, "step": 232290 }, { "epoch": 0.000512, "grad_norm": 0.7745198607444763, "learning_rate": 7.084650260516321e-06, "loss": 2.5706, "step": 232300 }, { "epoch": 0.0005376, "grad_norm": 0.8691017627716064, "learning_rate": 7.083879972654053e-06, "loss": 2.5272, "step": 232310 }, { "epoch": 0.0005632, "grad_norm": 0.8709973096847534, "learning_rate": 7.0831097037020515e-06, "loss": 2.8869, "step": 232320 }, { "epoch": 0.0005888, "grad_norm": 0.8204741477966309, "learning_rate": 7.082339453665304e-06, "loss": 2.4723, "step": 232330 }, { "epoch": 0.0006144, "grad_norm": 0.9541375041007996, "learning_rate": 7.081569222548811e-06, "loss": 2.6743, "step": 232340 }, { "epoch": 0.00064, "grad_norm": 0.7964534163475037, "learning_rate": 7.080799010357568e-06, "loss": 2.7568, "step": 232350 }, { "epoch": 0.0006656, "grad_norm": 0.8975979685783386, "learning_rate": 7.080028817096565e-06, "loss": 2.9438, "step": 232360 }, { "epoch": 0.0006912, "grad_norm": 0.8648867607116699, "learning_rate": 7.079258642770796e-06, "loss": 3.0465, "step": 232370 }, { "epoch": 0.0007168, "grad_norm": 0.992206335067749, "learning_rate": 7.078488487385262e-06, "loss": 2.6538, "step": 232380 }, { "epoch": 0.0007424, "grad_norm": 0.7927573323249817, "learning_rate": 7.077718350944954e-06, "loss": 2.8466, "step": 232390 }, { "epoch": 0.000768, "grad_norm": 0.8503648042678833, "learning_rate": 7.076948233454866e-06, "loss": 2.862, "step": 232400 }, { "epoch": 0.0007936, "grad_norm": 0.788664698600769, "learning_rate": 7.076178134919993e-06, "loss": 2.8885, "step": 232410 }, { "epoch": 0.0008192, "grad_norm": 0.8112935423851013, "learning_rate": 7.075408055345325e-06, "loss": 2.8455, "step": 232420 }, { "epoch": 0.0008448, "grad_norm": 0.8091871738433838, "learning_rate": 7.07463799473586e-06, "loss": 2.6216, "step": 232430 }, { "epoch": 0.0008704, "grad_norm": 0.8463128805160522, "learning_rate": 7.073867953096592e-06, "loss": 2.6858, "step": 232440 }, { "epoch": 0.000896, "grad_norm": 0.8907524943351746, "learning_rate": 7.073097930432509e-06, "loss": 2.6703, "step": 232450 }, { "epoch": 0.0009216, "grad_norm": 0.842997133731842, "learning_rate": 7.07232792674861e-06, "loss": 2.3437, "step": 232460 }, { "epoch": 0.0009472, "grad_norm": 0.8348447680473328, "learning_rate": 7.071557942049885e-06, "loss": 2.837, "step": 232470 }, { "epoch": 0.0009728, "grad_norm": 0.8215102553367615, "learning_rate": 7.070787976341331e-06, "loss": 2.9893, "step": 232480 }, { "epoch": 0.0009984, "grad_norm": 0.7948623299598694, "learning_rate": 7.070018029627936e-06, "loss": 2.7212, "step": 232490 }, { "epoch": 0.001024, "grad_norm": 0.7172789573669434, "learning_rate": 7.069248101914696e-06, "loss": 2.3348, "step": 232500 }, { "epoch": 0.0010496, "grad_norm": 0.8991668224334717, "learning_rate": 7.068478193206604e-06, "loss": 2.8112, "step": 232510 }, { "epoch": 0.0010752, "grad_norm": 0.8462167978286743, "learning_rate": 7.067708303508654e-06, "loss": 2.7637, "step": 232520 }, { "epoch": 0.0011008, "grad_norm": 0.8383526802062988, "learning_rate": 7.066938432825831e-06, "loss": 2.6559, "step": 232530 }, { "epoch": 0.0011264, "grad_norm": 0.8427352905273438, "learning_rate": 7.066168581163136e-06, "loss": 2.7228, "step": 232540 }, { "epoch": 0.001152, "grad_norm": 0.9058263301849365, "learning_rate": 7.0653987485255546e-06, "loss": 2.7102, "step": 232550 }, { "epoch": 0.0011776, "grad_norm": 0.9300446510314941, "learning_rate": 7.064628934918085e-06, "loss": 2.9704, "step": 232560 }, { "epoch": 0.0012032, "grad_norm": 0.8376397490501404, "learning_rate": 7.063859140345719e-06, "loss": 2.7866, "step": 232570 }, { "epoch": 0.0012288, "grad_norm": 0.7874274253845215, "learning_rate": 7.063089364813442e-06, "loss": 2.9744, "step": 232580 }, { "epoch": 0.0012544, "grad_norm": 0.80373215675354, "learning_rate": 7.062319608326253e-06, "loss": 2.8995, "step": 232590 }, { "epoch": 0.00128, "grad_norm": 0.7828577160835266, "learning_rate": 7.061549870889138e-06, "loss": 2.5478, "step": 232600 }, { "epoch": 0.0013056, "grad_norm": 0.8819389343261719, "learning_rate": 7.060780152507094e-06, "loss": 2.8531, "step": 232610 }, { "epoch": 0.0013312, "grad_norm": 0.9305280447006226, "learning_rate": 7.060010453185109e-06, "loss": 2.8365, "step": 232620 }, { "epoch": 0.0013568, "grad_norm": 0.7845356464385986, "learning_rate": 7.059240772928173e-06, "loss": 2.5388, "step": 232630 }, { "epoch": 0.0013824, "grad_norm": 0.7917231917381287, "learning_rate": 7.058471111741282e-06, "loss": 2.7775, "step": 232640 }, { "epoch": 0.001408, "grad_norm": 0.8749911189079285, "learning_rate": 7.057701469629421e-06, "loss": 2.786, "step": 232650 }, { "epoch": 0.0014336, "grad_norm": 0.768580436706543, "learning_rate": 7.056931846597586e-06, "loss": 2.8061, "step": 232660 }, { "epoch": 0.0014592, "grad_norm": 0.8749316334724426, "learning_rate": 7.056162242650764e-06, "loss": 2.8155, "step": 232670 }, { "epoch": 0.0014848, "grad_norm": 0.7015243172645569, "learning_rate": 7.055392657793949e-06, "loss": 2.5498, "step": 232680 }, { "epoch": 0.0015104, "grad_norm": 0.8959083557128906, "learning_rate": 7.054623092032129e-06, "loss": 2.2578, "step": 232690 }, { "epoch": 0.001536, "grad_norm": 0.9036973118782043, "learning_rate": 7.053853545370296e-06, "loss": 2.0977, "step": 232700 }, { "epoch": 0.0015616, "grad_norm": 0.9030726552009583, "learning_rate": 7.053084017813438e-06, "loss": 2.8735, "step": 232710 }, { "epoch": 0.0015872, "grad_norm": 0.8436724543571472, "learning_rate": 7.052314509366547e-06, "loss": 2.7243, "step": 232720 }, { "epoch": 0.0016128, "grad_norm": 0.8302525877952576, "learning_rate": 7.051545020034614e-06, "loss": 2.8543, "step": 232730 }, { "epoch": 0.0016384, "grad_norm": 0.8106706738471985, "learning_rate": 7.05077554982263e-06, "loss": 2.5498, "step": 232740 }, { "epoch": 0.001664, "grad_norm": 0.8679559826850891, "learning_rate": 7.050006098735582e-06, "loss": 2.8264, "step": 232750 }, { "epoch": 0.0016896, "grad_norm": 0.8661714196205139, "learning_rate": 7.049236666778458e-06, "loss": 2.8333, "step": 232760 }, { "epoch": 0.0017152, "grad_norm": 0.8357275724411011, "learning_rate": 7.0484672539562545e-06, "loss": 2.7722, "step": 232770 }, { "epoch": 0.0017408, "grad_norm": 0.8139961957931519, "learning_rate": 7.047697860273953e-06, "loss": 2.5888, "step": 232780 }, { "epoch": 0.0017664, "grad_norm": 0.9464243650436401, "learning_rate": 7.046928485736546e-06, "loss": 2.7331, "step": 232790 }, { "epoch": 0.001792, "grad_norm": 0.7934463620185852, "learning_rate": 7.046159130349024e-06, "loss": 2.592, "step": 232800 }, { "epoch": 2.56e-05, "grad_norm": 0.8058181405067444, "learning_rate": 7.045389794116378e-06, "loss": 1.844, "step": 232810 }, { "epoch": 5.12e-05, "grad_norm": 0.8827943205833435, "learning_rate": 7.044620477043589e-06, "loss": 2.7578, "step": 232820 }, { "epoch": 7.68e-05, "grad_norm": 0.8156419396400452, "learning_rate": 7.043851179135652e-06, "loss": 2.7516, "step": 232830 }, { "epoch": 0.0001024, "grad_norm": 1.0404136180877686, "learning_rate": 7.043081900397557e-06, "loss": 2.5854, "step": 232840 }, { "epoch": 0.000128, "grad_norm": 0.9278523921966553, "learning_rate": 7.042312640834288e-06, "loss": 2.6839, "step": 232850 }, { "epoch": 0.0001536, "grad_norm": 0.7991564273834229, "learning_rate": 7.041543400450834e-06, "loss": 2.7286, "step": 232860 }, { "epoch": 0.0001792, "grad_norm": 0.6744595766067505, "learning_rate": 7.040774179252186e-06, "loss": 2.4939, "step": 232870 }, { "epoch": 0.0002048, "grad_norm": 0.8708304166793823, "learning_rate": 7.040004977243333e-06, "loss": 2.7613, "step": 232880 }, { "epoch": 0.0002304, "grad_norm": 0.8085336089134216, "learning_rate": 7.0392357944292555e-06, "loss": 2.5958, "step": 232890 }, { "epoch": 0.000256, "grad_norm": 0.8579778075218201, "learning_rate": 7.0384666308149515e-06, "loss": 2.8765, "step": 232900 }, { "epoch": 0.0002816, "grad_norm": 0.8689493536949158, "learning_rate": 7.0376974864054035e-06, "loss": 2.5726, "step": 232910 }, { "epoch": 0.0003072, "grad_norm": 0.807266891002655, "learning_rate": 7.0369283612056e-06, "loss": 2.6602, "step": 232920 }, { "epoch": 0.0003328, "grad_norm": 0.7977423071861267, "learning_rate": 7.036159255220528e-06, "loss": 2.5168, "step": 232930 }, { "epoch": 0.0003584, "grad_norm": 0.8861188888549805, "learning_rate": 7.035390168455178e-06, "loss": 2.818, "step": 232940 }, { "epoch": 0.000384, "grad_norm": 0.8273731470108032, "learning_rate": 7.034621100914532e-06, "loss": 2.5089, "step": 232950 }, { "epoch": 0.0004096, "grad_norm": 0.891812801361084, "learning_rate": 7.033852052603583e-06, "loss": 2.7832, "step": 232960 }, { "epoch": 0.0004352, "grad_norm": 1.0128980875015259, "learning_rate": 7.033083023527314e-06, "loss": 2.7131, "step": 232970 }, { "epoch": 0.0004608, "grad_norm": 0.8031839728355408, "learning_rate": 7.032314013690711e-06, "loss": 2.6565, "step": 232980 }, { "epoch": 0.0004864, "grad_norm": 0.9371721744537354, "learning_rate": 7.031545023098765e-06, "loss": 2.6684, "step": 232990 }, { "epoch": 0.000512, "grad_norm": 0.9212530851364136, "learning_rate": 7.030776051756458e-06, "loss": 2.7654, "step": 233000 }, { "epoch": 0.0005376, "grad_norm": 0.7252604961395264, "learning_rate": 7.030007099668783e-06, "loss": 2.3962, "step": 233010 }, { "epoch": 0.0005632, "grad_norm": 0.8828787803649902, "learning_rate": 7.029238166840722e-06, "loss": 2.7981, "step": 233020 }, { "epoch": 0.0005888, "grad_norm": 0.9385031461715698, "learning_rate": 7.028469253277259e-06, "loss": 2.344, "step": 233030 }, { "epoch": 0.0006144, "grad_norm": 0.8420653343200684, "learning_rate": 7.027700358983387e-06, "loss": 2.4848, "step": 233040 }, { "epoch": 0.00064, "grad_norm": 0.8736659288406372, "learning_rate": 7.026931483964086e-06, "loss": 2.6618, "step": 233050 }, { "epoch": 0.0006656, "grad_norm": 0.8574842810630798, "learning_rate": 7.026162628224344e-06, "loss": 2.8288, "step": 233060 }, { "epoch": 0.0006912, "grad_norm": 0.8291305899620056, "learning_rate": 7.025393791769144e-06, "loss": 2.849, "step": 233070 }, { "epoch": 0.0007168, "grad_norm": 0.8948954939842224, "learning_rate": 7.024624974603478e-06, "loss": 2.5463, "step": 233080 }, { "epoch": 0.0007424, "grad_norm": 0.7879743576049805, "learning_rate": 7.023856176732327e-06, "loss": 2.7278, "step": 233090 }, { "epoch": 0.000768, "grad_norm": 0.9176305532455444, "learning_rate": 7.02308739816068e-06, "loss": 2.7393, "step": 233100 }, { "epoch": 0.0007936, "grad_norm": 0.8868340849876404, "learning_rate": 7.0223186388935195e-06, "loss": 2.768, "step": 233110 }, { "epoch": 0.0008192, "grad_norm": 0.8512806296348572, "learning_rate": 7.021549898935831e-06, "loss": 2.765, "step": 233120 }, { "epoch": 0.0008448, "grad_norm": 0.8121416568756104, "learning_rate": 7.020781178292601e-06, "loss": 2.7025, "step": 233130 }, { "epoch": 0.0008704, "grad_norm": 0.829376757144928, "learning_rate": 7.020012476968811e-06, "loss": 2.7959, "step": 233140 }, { "epoch": 0.000896, "grad_norm": 0.9747993350028992, "learning_rate": 7.019243794969448e-06, "loss": 2.7341, "step": 233150 }, { "epoch": 0.0009216, "grad_norm": 0.8647096753120422, "learning_rate": 7.018475132299498e-06, "loss": 2.6169, "step": 233160 }, { "epoch": 0.0009472, "grad_norm": 0.8833145499229431, "learning_rate": 7.017706488963946e-06, "loss": 2.5665, "step": 233170 }, { "epoch": 0.0009728, "grad_norm": 0.7863320112228394, "learning_rate": 7.016937864967772e-06, "loss": 2.815, "step": 233180 }, { "epoch": 0.0009984, "grad_norm": 0.9571518301963806, "learning_rate": 7.016169260315963e-06, "loss": 2.8955, "step": 233190 }, { "epoch": 0.001024, "grad_norm": 0.9160623550415039, "learning_rate": 7.015400675013505e-06, "loss": 2.561, "step": 233200 }, { "epoch": 0.0010496, "grad_norm": 0.8877311944961548, "learning_rate": 7.014632109065382e-06, "loss": 2.6402, "step": 233210 }, { "epoch": 0.0010752, "grad_norm": 0.944274365901947, "learning_rate": 7.0138635624765726e-06, "loss": 2.6622, "step": 233220 }, { "epoch": 0.0011008, "grad_norm": 0.7996581792831421, "learning_rate": 7.0130950352520665e-06, "loss": 2.6953, "step": 233230 }, { "epoch": 0.0011264, "grad_norm": 1.2044380903244019, "learning_rate": 7.0123265273968425e-06, "loss": 2.6979, "step": 233240 }, { "epoch": 0.001152, "grad_norm": 0.7762665152549744, "learning_rate": 7.011558038915889e-06, "loss": 2.8222, "step": 233250 }, { "epoch": 0.0011776, "grad_norm": 0.8404411673545837, "learning_rate": 7.010789569814189e-06, "loss": 2.7899, "step": 233260 }, { "epoch": 0.0012032, "grad_norm": 0.7997880578041077, "learning_rate": 7.010021120096722e-06, "loss": 3.0022, "step": 233270 }, { "epoch": 0.0012288, "grad_norm": 0.7815568447113037, "learning_rate": 7.009252689768473e-06, "loss": 2.6732, "step": 233280 }, { "epoch": 0.0012544, "grad_norm": 0.8312505483627319, "learning_rate": 7.008484278834426e-06, "loss": 2.7652, "step": 233290 }, { "epoch": 0.00128, "grad_norm": 0.781539797782898, "learning_rate": 7.007715887299566e-06, "loss": 2.8228, "step": 233300 }, { "epoch": 0.0013056, "grad_norm": 0.8849239945411682, "learning_rate": 7.0069475151688716e-06, "loss": 2.7902, "step": 233310 }, { "epoch": 0.0013312, "grad_norm": 0.7847746014595032, "learning_rate": 7.006179162447326e-06, "loss": 2.7087, "step": 233320 }, { "epoch": 0.0013568, "grad_norm": 0.8821688890457153, "learning_rate": 7.0054108291399155e-06, "loss": 2.5036, "step": 233330 }, { "epoch": 0.0013824, "grad_norm": 0.9069758653640747, "learning_rate": 7.004642515251617e-06, "loss": 2.8829, "step": 233340 }, { "epoch": 0.001408, "grad_norm": 0.7759695649147034, "learning_rate": 7.003874220787416e-06, "loss": 2.711, "step": 233350 }, { "epoch": 0.0014336, "grad_norm": 0.8900223970413208, "learning_rate": 7.0031059457522955e-06, "loss": 2.5171, "step": 233360 }, { "epoch": 0.0014592, "grad_norm": 0.8206783533096313, "learning_rate": 7.002337690151238e-06, "loss": 2.8023, "step": 233370 }, { "epoch": 0.0014848, "grad_norm": 0.8193807005882263, "learning_rate": 7.001569453989221e-06, "loss": 2.3766, "step": 233380 }, { "epoch": 0.0015104, "grad_norm": 0.798739492893219, "learning_rate": 7.000801237271231e-06, "loss": 2.4929, "step": 233390 }, { "epoch": 0.001536, "grad_norm": 0.8888989090919495, "learning_rate": 7.000033040002245e-06, "loss": 2.2887, "step": 233400 }, { "epoch": 0.0015616, "grad_norm": 0.8651706576347351, "learning_rate": 6.999264862187252e-06, "loss": 2.6727, "step": 233410 }, { "epoch": 0.0015872, "grad_norm": 0.9714741110801697, "learning_rate": 6.998496703831221e-06, "loss": 2.8885, "step": 233420 }, { "epoch": 0.0016128, "grad_norm": 0.7932536602020264, "learning_rate": 6.997728564939149e-06, "loss": 3.0685, "step": 233430 }, { "epoch": 0.0016384, "grad_norm": 0.8375908136367798, "learning_rate": 6.996960445516007e-06, "loss": 2.8492, "step": 233440 }, { "epoch": 0.001664, "grad_norm": 0.8217241168022156, "learning_rate": 6.996192345566778e-06, "loss": 2.9334, "step": 233450 }, { "epoch": 0.0016896, "grad_norm": 0.8241358995437622, "learning_rate": 6.995424265096444e-06, "loss": 2.7831, "step": 233460 }, { "epoch": 0.0017152, "grad_norm": 0.8351585865020752, "learning_rate": 6.994656204109986e-06, "loss": 2.6547, "step": 233470 }, { "epoch": 0.0017408, "grad_norm": 0.8612504005432129, "learning_rate": 6.993888162612381e-06, "loss": 2.4402, "step": 233480 }, { "epoch": 0.0017664, "grad_norm": 0.9133037328720093, "learning_rate": 6.993120140608613e-06, "loss": 2.9341, "step": 233490 }, { "epoch": 0.001792, "grad_norm": 0.8529041409492493, "learning_rate": 6.992352138103665e-06, "loss": 2.9454, "step": 233500 }, { "epoch": 2.56e-05, "grad_norm": 0.8341432213783264, "learning_rate": 6.991584155102511e-06, "loss": 1.9959, "step": 233510 }, { "epoch": 5.12e-05, "grad_norm": 0.8454518914222717, "learning_rate": 6.990816191610135e-06, "loss": 2.7172, "step": 233520 }, { "epoch": 7.68e-05, "grad_norm": 0.7979486584663391, "learning_rate": 6.990048247631518e-06, "loss": 2.6919, "step": 233530 }, { "epoch": 0.0001024, "grad_norm": 0.841562807559967, "learning_rate": 6.989280323171637e-06, "loss": 2.646, "step": 233540 }, { "epoch": 0.000128, "grad_norm": 0.7835842967033386, "learning_rate": 6.988512418235471e-06, "loss": 2.539, "step": 233550 }, { "epoch": 0.0001536, "grad_norm": 0.8480894565582275, "learning_rate": 6.987744532828003e-06, "loss": 2.8327, "step": 233560 }, { "epoch": 0.0001792, "grad_norm": 0.7924591302871704, "learning_rate": 6.986976666954213e-06, "loss": 2.7164, "step": 233570 }, { "epoch": 0.0002048, "grad_norm": 1.397138237953186, "learning_rate": 6.9862088206190756e-06, "loss": 2.6264, "step": 233580 }, { "epoch": 0.0002304, "grad_norm": 0.8167298436164856, "learning_rate": 6.9854409938275715e-06, "loss": 2.6745, "step": 233590 }, { "epoch": 0.000256, "grad_norm": 0.7956943511962891, "learning_rate": 6.9846731865846824e-06, "loss": 2.7259, "step": 233600 }, { "epoch": 0.0002816, "grad_norm": 0.8959904313087463, "learning_rate": 6.983905398895387e-06, "loss": 2.6007, "step": 233610 }, { "epoch": 0.0003072, "grad_norm": 0.8611029982566833, "learning_rate": 6.9831376307646645e-06, "loss": 2.3816, "step": 233620 }, { "epoch": 0.0003328, "grad_norm": 0.8764241933822632, "learning_rate": 6.982369882197495e-06, "loss": 2.705, "step": 233630 }, { "epoch": 0.0003584, "grad_norm": 0.9094473123550415, "learning_rate": 6.981602153198852e-06, "loss": 2.6983, "step": 233640 }, { "epoch": 0.000384, "grad_norm": 0.8416604995727539, "learning_rate": 6.980834443773716e-06, "loss": 2.568, "step": 233650 }, { "epoch": 0.0004096, "grad_norm": 0.9003143906593323, "learning_rate": 6.980066753927069e-06, "loss": 2.6757, "step": 233660 }, { "epoch": 0.0004352, "grad_norm": 0.8604199886322021, "learning_rate": 6.9792990836638845e-06, "loss": 3.1491, "step": 233670 }, { "epoch": 0.0004608, "grad_norm": 0.7491870522499084, "learning_rate": 6.978531432989143e-06, "loss": 2.7213, "step": 233680 }, { "epoch": 0.0004864, "grad_norm": 0.8854911923408508, "learning_rate": 6.977763801907822e-06, "loss": 2.8148, "step": 233690 }, { "epoch": 0.000512, "grad_norm": 0.7562501430511475, "learning_rate": 6.976996190424901e-06, "loss": 2.6615, "step": 233700 }, { "epoch": 0.0005376, "grad_norm": 0.8293057680130005, "learning_rate": 6.9762285985453545e-06, "loss": 2.4671, "step": 233710 }, { "epoch": 0.0005632, "grad_norm": 0.9181168675422668, "learning_rate": 6.975461026274162e-06, "loss": 2.8014, "step": 233720 }, { "epoch": 0.0005888, "grad_norm": 0.9226995706558228, "learning_rate": 6.974693473616303e-06, "loss": 2.3157, "step": 233730 }, { "epoch": 0.0006144, "grad_norm": 0.9152562022209167, "learning_rate": 6.9739259405767514e-06, "loss": 2.6418, "step": 233740 }, { "epoch": 0.00064, "grad_norm": 0.8020469546318054, "learning_rate": 6.9731584271604844e-06, "loss": 2.6726, "step": 233750 }, { "epoch": 0.0006656, "grad_norm": 0.8576669692993164, "learning_rate": 6.972390933372479e-06, "loss": 2.5893, "step": 233760 }, { "epoch": 0.0006912, "grad_norm": 0.873123049736023, "learning_rate": 6.971623459217717e-06, "loss": 2.5863, "step": 233770 }, { "epoch": 0.0007168, "grad_norm": 0.8004098534584045, "learning_rate": 6.970856004701171e-06, "loss": 2.6227, "step": 233780 }, { "epoch": 0.0007424, "grad_norm": 0.8494545221328735, "learning_rate": 6.97008856982782e-06, "loss": 2.7913, "step": 233790 }, { "epoch": 0.000768, "grad_norm": 0.8921388387680054, "learning_rate": 6.969321154602639e-06, "loss": 2.6641, "step": 233800 }, { "epoch": 0.0007936, "grad_norm": 0.8747520446777344, "learning_rate": 6.968553759030604e-06, "loss": 2.6542, "step": 233810 }, { "epoch": 0.0008192, "grad_norm": 0.836607813835144, "learning_rate": 6.967786383116692e-06, "loss": 2.3051, "step": 233820 }, { "epoch": 0.0008448, "grad_norm": 0.788557767868042, "learning_rate": 6.9670190268658825e-06, "loss": 2.5148, "step": 233830 }, { "epoch": 0.0008704, "grad_norm": 0.8967320322990417, "learning_rate": 6.966251690283146e-06, "loss": 2.7155, "step": 233840 }, { "epoch": 0.000896, "grad_norm": 0.7812309265136719, "learning_rate": 6.96548437337346e-06, "loss": 2.5347, "step": 233850 }, { "epoch": 0.0009216, "grad_norm": 0.8290796875953674, "learning_rate": 6.9647170761418044e-06, "loss": 2.5717, "step": 233860 }, { "epoch": 0.0009472, "grad_norm": 0.8701794147491455, "learning_rate": 6.963949798593149e-06, "loss": 2.7032, "step": 233870 }, { "epoch": 0.0009728, "grad_norm": 0.8406051993370056, "learning_rate": 6.963182540732474e-06, "loss": 2.6682, "step": 233880 }, { "epoch": 0.0009984, "grad_norm": 0.8479194641113281, "learning_rate": 6.962415302564751e-06, "loss": 2.619, "step": 233890 }, { "epoch": 0.001024, "grad_norm": 0.865516185760498, "learning_rate": 6.961648084094961e-06, "loss": 2.6997, "step": 233900 }, { "epoch": 0.0010496, "grad_norm": 0.9221714735031128, "learning_rate": 6.9608808853280725e-06, "loss": 2.6554, "step": 233910 }, { "epoch": 0.0010752, "grad_norm": 0.9478402137756348, "learning_rate": 6.960113706269065e-06, "loss": 2.5587, "step": 233920 }, { "epoch": 0.0011008, "grad_norm": 0.7895127534866333, "learning_rate": 6.959346546922914e-06, "loss": 2.6844, "step": 233930 }, { "epoch": 0.0011264, "grad_norm": 0.8569363951683044, "learning_rate": 6.9585794072945855e-06, "loss": 2.8403, "step": 233940 }, { "epoch": 0.001152, "grad_norm": 0.7551025152206421, "learning_rate": 6.9578122873890675e-06, "loss": 2.6887, "step": 233950 }, { "epoch": 0.0011776, "grad_norm": 0.7719548344612122, "learning_rate": 6.957045187211327e-06, "loss": 2.5912, "step": 233960 }, { "epoch": 0.0012032, "grad_norm": 0.876585066318512, "learning_rate": 6.95627810676634e-06, "loss": 2.7342, "step": 233970 }, { "epoch": 0.0012288, "grad_norm": 0.8147241473197937, "learning_rate": 6.95551104605908e-06, "loss": 2.9567, "step": 233980 }, { "epoch": 0.0012544, "grad_norm": 0.8218995928764343, "learning_rate": 6.954744005094524e-06, "loss": 2.9097, "step": 233990 }, { "epoch": 0.00128, "grad_norm": 0.7688713669776917, "learning_rate": 6.953976983877642e-06, "loss": 2.5878, "step": 234000 }, { "epoch": 0.0013056, "grad_norm": 0.9697936773300171, "learning_rate": 6.9532099824134094e-06, "loss": 2.702, "step": 234010 }, { "epoch": 0.0013312, "grad_norm": 0.855536937713623, "learning_rate": 6.952443000706801e-06, "loss": 2.8466, "step": 234020 }, { "epoch": 0.0013568, "grad_norm": 0.9581269025802612, "learning_rate": 6.951676038762789e-06, "loss": 2.7224, "step": 234030 }, { "epoch": 0.0013824, "grad_norm": 0.8359749913215637, "learning_rate": 6.950909096586348e-06, "loss": 2.6384, "step": 234040 }, { "epoch": 0.001408, "grad_norm": 0.8952873945236206, "learning_rate": 6.9501421741824505e-06, "loss": 2.7306, "step": 234050 }, { "epoch": 0.0014336, "grad_norm": 0.8050683736801147, "learning_rate": 6.949375271556073e-06, "loss": 2.6787, "step": 234060 }, { "epoch": 0.0014592, "grad_norm": 0.8527023196220398, "learning_rate": 6.948608388712184e-06, "loss": 2.7397, "step": 234070 }, { "epoch": 0.0014848, "grad_norm": 0.907547116279602, "learning_rate": 6.947841525655759e-06, "loss": 2.6576, "step": 234080 }, { "epoch": 0.0015104, "grad_norm": 0.7792013883590698, "learning_rate": 6.94707468239177e-06, "loss": 2.5648, "step": 234090 }, { "epoch": 0.001536, "grad_norm": 1.018087387084961, "learning_rate": 6.946307858925192e-06, "loss": 1.8619, "step": 234100 }, { "epoch": 0.0015616, "grad_norm": 0.9605078101158142, "learning_rate": 6.9455410552609905e-06, "loss": 2.5663, "step": 234110 }, { "epoch": 0.0015872, "grad_norm": 0.8311392664909363, "learning_rate": 6.9447742714041486e-06, "loss": 2.491, "step": 234120 }, { "epoch": 0.0016128, "grad_norm": 0.8108864426612854, "learning_rate": 6.944007507359633e-06, "loss": 2.6907, "step": 234130 }, { "epoch": 0.0016384, "grad_norm": 0.8524082899093628, "learning_rate": 6.943240763132416e-06, "loss": 2.5392, "step": 234140 }, { "epoch": 0.001664, "grad_norm": 1.0033026933670044, "learning_rate": 6.942474038727473e-06, "loss": 2.7151, "step": 234150 }, { "epoch": 0.0016896, "grad_norm": 0.8094969987869263, "learning_rate": 6.9417073341497695e-06, "loss": 2.5712, "step": 234160 }, { "epoch": 0.0017152, "grad_norm": 1.0258482694625854, "learning_rate": 6.940940649404282e-06, "loss": 2.595, "step": 234170 }, { "epoch": 0.0017408, "grad_norm": 0.8270221948623657, "learning_rate": 6.9401739844959815e-06, "loss": 2.6417, "step": 234180 }, { "epoch": 0.0017664, "grad_norm": 0.8390604853630066, "learning_rate": 6.939407339429843e-06, "loss": 2.8008, "step": 234190 }, { "epoch": 0.001792, "grad_norm": 0.8432414531707764, "learning_rate": 6.938640714210829e-06, "loss": 2.8391, "step": 234200 }, { "epoch": 0.0018176, "grad_norm": 0.8715373277664185, "learning_rate": 6.937874108843918e-06, "loss": 2.6686, "step": 234210 }, { "epoch": 0.0018432, "grad_norm": 0.8238497376441956, "learning_rate": 6.937107523334084e-06, "loss": 2.8718, "step": 234220 }, { "epoch": 0.0018688, "grad_norm": 0.7356290221214294, "learning_rate": 6.936340957686288e-06, "loss": 2.4448, "step": 234230 }, { "epoch": 0.0018944, "grad_norm": 0.8490942716598511, "learning_rate": 6.935574411905508e-06, "loss": 3.0045, "step": 234240 }, { "epoch": 0.00192, "grad_norm": 0.8373507261276245, "learning_rate": 6.934807885996715e-06, "loss": 2.61, "step": 234250 }, { "epoch": 0.0019456, "grad_norm": 0.8259606957435608, "learning_rate": 6.9340413799648776e-06, "loss": 2.7756, "step": 234260 }, { "epoch": 0.0019712, "grad_norm": 0.749359667301178, "learning_rate": 6.933274893814967e-06, "loss": 2.5926, "step": 234270 }, { "epoch": 0.0019968, "grad_norm": 0.823721706867218, "learning_rate": 6.932508427551952e-06, "loss": 2.4596, "step": 234280 }, { "epoch": 0.0020224, "grad_norm": 0.8850231170654297, "learning_rate": 6.931741981180804e-06, "loss": 2.7447, "step": 234290 }, { "epoch": 0.002048, "grad_norm": 0.7715412974357605, "learning_rate": 6.930975554706496e-06, "loss": 2.5469, "step": 234300 }, { "epoch": 0.0020736, "grad_norm": 0.8849248886108398, "learning_rate": 6.930209148133995e-06, "loss": 2.6173, "step": 234310 }, { "epoch": 0.0020992, "grad_norm": 0.8364054560661316, "learning_rate": 6.929442761468275e-06, "loss": 2.4367, "step": 234320 }, { "epoch": 0.0021248, "grad_norm": 0.9712650775909424, "learning_rate": 6.9286763947142995e-06, "loss": 2.6254, "step": 234330 }, { "epoch": 0.0021504, "grad_norm": 0.7554362416267395, "learning_rate": 6.927910047877042e-06, "loss": 2.8453, "step": 234340 }, { "epoch": 0.002176, "grad_norm": 0.8881562948226929, "learning_rate": 6.927143720961473e-06, "loss": 2.8124, "step": 234350 }, { "epoch": 0.0022016, "grad_norm": 1.308800220489502, "learning_rate": 6.926377413972558e-06, "loss": 2.8667, "step": 234360 }, { "epoch": 0.0022272, "grad_norm": 0.7857517004013062, "learning_rate": 6.925611126915268e-06, "loss": 2.9322, "step": 234370 }, { "epoch": 0.0022528, "grad_norm": 0.7066246271133423, "learning_rate": 6.924844859794575e-06, "loss": 2.5652, "step": 234380 }, { "epoch": 0.0022784, "grad_norm": 0.8445597887039185, "learning_rate": 6.924078612615446e-06, "loss": 2.8254, "step": 234390 }, { "epoch": 0.002304, "grad_norm": 1.177329659461975, "learning_rate": 6.923312385382848e-06, "loss": 3.1557, "step": 234400 }, { "epoch": 0.0023296, "grad_norm": 0.7597833275794983, "learning_rate": 6.922546178101752e-06, "loss": 2.5057, "step": 234410 }, { "epoch": 0.0023552, "grad_norm": 1.1097933053970337, "learning_rate": 6.921779990777127e-06, "loss": 2.8204, "step": 234420 }, { "epoch": 0.0023808, "grad_norm": 0.8404260873794556, "learning_rate": 6.92101382341394e-06, "loss": 2.7879, "step": 234430 }, { "epoch": 0.0024064, "grad_norm": 0.7895416617393494, "learning_rate": 6.92024767601716e-06, "loss": 2.7665, "step": 234440 }, { "epoch": 0.002432, "grad_norm": 0.7436951398849487, "learning_rate": 6.919481548591756e-06, "loss": 2.6269, "step": 234450 }, { "epoch": 0.0024576, "grad_norm": 0.7987615466117859, "learning_rate": 6.918715441142692e-06, "loss": 2.8209, "step": 234460 }, { "epoch": 0.0024832, "grad_norm": 0.7251514792442322, "learning_rate": 6.91794935367494e-06, "loss": 2.9087, "step": 234470 }, { "epoch": 0.0025088, "grad_norm": 0.7856795787811279, "learning_rate": 6.917183286193472e-06, "loss": 2.7586, "step": 234480 }, { "epoch": 0.0025344, "grad_norm": 0.9336908459663391, "learning_rate": 6.916417238703248e-06, "loss": 2.9281, "step": 234490 }, { "epoch": 0.00256, "grad_norm": 0.796265184879303, "learning_rate": 6.915651211209239e-06, "loss": 2.8658, "step": 234500 }, { "epoch": 0.0025856, "grad_norm": 1.0580511093139648, "learning_rate": 6.914885203716412e-06, "loss": 2.7567, "step": 234510 }, { "epoch": 0.0026112, "grad_norm": 0.8015888929367065, "learning_rate": 6.914119216229736e-06, "loss": 2.4878, "step": 234520 }, { "epoch": 0.0026368, "grad_norm": 0.7870532870292664, "learning_rate": 6.913353248754176e-06, "loss": 2.8835, "step": 234530 }, { "epoch": 0.0026624, "grad_norm": 0.8536500930786133, "learning_rate": 6.9125873012947e-06, "loss": 2.972, "step": 234540 }, { "epoch": 0.002688, "grad_norm": 1.040372610092163, "learning_rate": 6.911821373856276e-06, "loss": 3.0336, "step": 234550 }, { "epoch": 0.0027136, "grad_norm": 0.8053653240203857, "learning_rate": 6.911055466443867e-06, "loss": 2.7702, "step": 234560 }, { "epoch": 0.0027392, "grad_norm": 0.9295529723167419, "learning_rate": 6.910289579062443e-06, "loss": 2.5155, "step": 234570 }, { "epoch": 0.0027648, "grad_norm": 0.8709783554077148, "learning_rate": 6.90952371171697e-06, "loss": 2.9795, "step": 234580 }, { "epoch": 0.0027904, "grad_norm": 1.0627143383026123, "learning_rate": 6.908757864412416e-06, "loss": 2.7561, "step": 234590 }, { "epoch": 0.002816, "grad_norm": 0.7365731000900269, "learning_rate": 6.907992037153744e-06, "loss": 2.9165, "step": 234600 }, { "epoch": 0.0028416, "grad_norm": 0.9039601683616638, "learning_rate": 6.907226229945922e-06, "loss": 2.9189, "step": 234610 }, { "epoch": 0.0028672, "grad_norm": 0.8080585598945618, "learning_rate": 6.906460442793918e-06, "loss": 2.9417, "step": 234620 }, { "epoch": 0.0028928, "grad_norm": 0.8338820338249207, "learning_rate": 6.90569467570269e-06, "loss": 2.9241, "step": 234630 }, { "epoch": 0.0029184, "grad_norm": 0.8397356867790222, "learning_rate": 6.904928928677214e-06, "loss": 2.9296, "step": 234640 }, { "epoch": 0.002944, "grad_norm": 0.8870421648025513, "learning_rate": 6.904163201722451e-06, "loss": 3.1255, "step": 234650 }, { "epoch": 0.0029696, "grad_norm": 0.8785773515701294, "learning_rate": 6.903397494843367e-06, "loss": 2.9103, "step": 234660 }, { "epoch": 0.0029952, "grad_norm": 0.765664279460907, "learning_rate": 6.902631808044926e-06, "loss": 2.7556, "step": 234670 }, { "epoch": 0.0030208, "grad_norm": 0.7689655423164368, "learning_rate": 6.9018661413320986e-06, "loss": 2.8268, "step": 234680 }, { "epoch": 0.0030464, "grad_norm": 0.7666209936141968, "learning_rate": 6.901100494709843e-06, "loss": 2.6847, "step": 234690 }, { "epoch": 0.003072, "grad_norm": 0.8165966868400574, "learning_rate": 6.9003348681831265e-06, "loss": 3.0974, "step": 234700 }, { "epoch": 0.0030976, "grad_norm": 0.7385879755020142, "learning_rate": 6.899569261756915e-06, "loss": 2.8368, "step": 234710 }, { "epoch": 0.0031232, "grad_norm": 0.8984041213989258, "learning_rate": 6.898803675436174e-06, "loss": 2.9547, "step": 234720 }, { "epoch": 0.0031488, "grad_norm": 0.7884451746940613, "learning_rate": 6.898038109225866e-06, "loss": 2.7957, "step": 234730 }, { "epoch": 0.0031744, "grad_norm": 0.820537805557251, "learning_rate": 6.897272563130957e-06, "loss": 2.9192, "step": 234740 }, { "epoch": 0.0032, "grad_norm": 0.7581765055656433, "learning_rate": 6.8965070371564105e-06, "loss": 2.7148, "step": 234750 }, { "epoch": 0.0032256, "grad_norm": 0.8090218305587769, "learning_rate": 6.895741531307192e-06, "loss": 2.9271, "step": 234760 }, { "epoch": 0.0032512, "grad_norm": 0.8689702153205872, "learning_rate": 6.894976045588262e-06, "loss": 2.9189, "step": 234770 }, { "epoch": 0.0032768, "grad_norm": 0.8977057933807373, "learning_rate": 6.894210580004587e-06, "loss": 2.8398, "step": 234780 }, { "epoch": 0.0033024, "grad_norm": 0.795421838760376, "learning_rate": 6.893445134561136e-06, "loss": 2.9427, "step": 234790 }, { "epoch": 0.003328, "grad_norm": 0.9182202219963074, "learning_rate": 6.892679709262864e-06, "loss": 2.7662, "step": 234800 }, { "epoch": 0.0033536, "grad_norm": 1.0035603046417236, "learning_rate": 6.891914304114735e-06, "loss": 3.0221, "step": 234810 }, { "epoch": 0.0033792, "grad_norm": 0.7706133127212524, "learning_rate": 6.8911489191217196e-06, "loss": 2.757, "step": 234820 }, { "epoch": 0.0034048, "grad_norm": 0.8022343516349792, "learning_rate": 6.890383554288777e-06, "loss": 2.7635, "step": 234830 }, { "epoch": 0.0034304, "grad_norm": 0.8418096899986267, "learning_rate": 6.889618209620872e-06, "loss": 2.6606, "step": 234840 }, { "epoch": 0.003456, "grad_norm": 0.7525208592414856, "learning_rate": 6.888852885122964e-06, "loss": 2.9802, "step": 234850 }, { "epoch": 0.0034816, "grad_norm": 0.8462871313095093, "learning_rate": 6.888087580800018e-06, "loss": 2.8104, "step": 234860 }, { "epoch": 0.0035072, "grad_norm": 0.7491713762283325, "learning_rate": 6.887322296656997e-06, "loss": 2.8908, "step": 234870 }, { "epoch": 0.0035328, "grad_norm": 0.7836775779724121, "learning_rate": 6.886557032698868e-06, "loss": 2.8318, "step": 234880 }, { "epoch": 0.0035584, "grad_norm": 0.7401611804962158, "learning_rate": 6.885791788930585e-06, "loss": 2.9079, "step": 234890 }, { "epoch": 0.003584, "grad_norm": 0.9223432540893555, "learning_rate": 6.885026565357115e-06, "loss": 2.841, "step": 234900 }, { "epoch": 0.0036096, "grad_norm": 0.7868599891662598, "learning_rate": 6.88426136198342e-06, "loss": 3.0475, "step": 234910 }, { "epoch": 0.0036352, "grad_norm": 0.8116159439086914, "learning_rate": 6.883496178814464e-06, "loss": 3.0169, "step": 234920 }, { "epoch": 0.0036608, "grad_norm": 0.7959586977958679, "learning_rate": 6.882731015855204e-06, "loss": 2.8063, "step": 234930 }, { "epoch": 0.0036864, "grad_norm": 0.846500813961029, "learning_rate": 6.881965873110606e-06, "loss": 2.9006, "step": 234940 }, { "epoch": 0.003712, "grad_norm": 0.818846583366394, "learning_rate": 6.8812007505856325e-06, "loss": 2.7291, "step": 234950 }, { "epoch": 0.0037376, "grad_norm": 0.8121203184127808, "learning_rate": 6.880435648285242e-06, "loss": 3.0611, "step": 234960 }, { "epoch": 0.0037632, "grad_norm": 0.7814087271690369, "learning_rate": 6.879670566214396e-06, "loss": 2.8742, "step": 234970 }, { "epoch": 0.0037888, "grad_norm": 0.798576831817627, "learning_rate": 6.878905504378055e-06, "loss": 2.9103, "step": 234980 }, { "epoch": 0.0038144, "grad_norm": 0.95786052942276, "learning_rate": 6.878140462781185e-06, "loss": 2.9395, "step": 234990 }, { "epoch": 0.00384, "grad_norm": 0.8836155533790588, "learning_rate": 6.877375441428744e-06, "loss": 2.9507, "step": 235000 }, { "epoch": 0.0038656, "grad_norm": 0.8677904009819031, "learning_rate": 6.876610440325696e-06, "loss": 2.7982, "step": 235010 }, { "epoch": 0.0038912, "grad_norm": 0.7507909536361694, "learning_rate": 6.8758454594769974e-06, "loss": 2.7258, "step": 235020 }, { "epoch": 0.0039168, "grad_norm": 0.795636773109436, "learning_rate": 6.87508049888761e-06, "loss": 2.9212, "step": 235030 }, { "epoch": 0.0039424, "grad_norm": 0.9375107288360596, "learning_rate": 6.874315558562496e-06, "loss": 2.7992, "step": 235040 }, { "epoch": 0.003968, "grad_norm": 0.8399636149406433, "learning_rate": 6.873550638506614e-06, "loss": 3.0123, "step": 235050 }, { "epoch": 0.0039936, "grad_norm": 0.7556323409080505, "learning_rate": 6.872785738724925e-06, "loss": 2.9287, "step": 235060 }, { "epoch": 0.0040192, "grad_norm": 0.7825124859809875, "learning_rate": 6.87202085922239e-06, "loss": 2.7683, "step": 235070 }, { "epoch": 0.0040448, "grad_norm": 0.7897646427154541, "learning_rate": 6.871256000003971e-06, "loss": 2.7664, "step": 235080 }, { "epoch": 0.0040704, "grad_norm": 1.241477131843567, "learning_rate": 6.870491161074622e-06, "loss": 3.0307, "step": 235090 }, { "epoch": 0.004096, "grad_norm": 1.0006550550460815, "learning_rate": 6.8697263424393065e-06, "loss": 2.9859, "step": 235100 }, { "epoch": 0.0041216, "grad_norm": 0.7462039589881897, "learning_rate": 6.868961544102986e-06, "loss": 2.8974, "step": 235110 }, { "epoch": 0.0041472, "grad_norm": 1.0342824459075928, "learning_rate": 6.868196766070616e-06, "loss": 3.002, "step": 235120 }, { "epoch": 0.0041728, "grad_norm": 0.795272946357727, "learning_rate": 6.867432008347157e-06, "loss": 2.7214, "step": 235130 }, { "epoch": 0.0041984, "grad_norm": 0.7479667663574219, "learning_rate": 6.866667270937568e-06, "loss": 2.7856, "step": 235140 }, { "epoch": 0.004224, "grad_norm": 0.7667438983917236, "learning_rate": 6.865902553846813e-06, "loss": 2.8604, "step": 235150 }, { "epoch": 0.0042496, "grad_norm": 0.9394447207450867, "learning_rate": 6.86513785707984e-06, "loss": 2.848, "step": 235160 }, { "epoch": 0.0042752, "grad_norm": 0.8444170355796814, "learning_rate": 6.86437318064162e-06, "loss": 3.0237, "step": 235170 }, { "epoch": 0.0043008, "grad_norm": 0.8079456686973572, "learning_rate": 6.863608524537105e-06, "loss": 2.8751, "step": 235180 }, { "epoch": 0.0043264, "grad_norm": 0.8943831920623779, "learning_rate": 6.862843888771255e-06, "loss": 2.7215, "step": 235190 }, { "epoch": 0.004352, "grad_norm": 1.2438057661056519, "learning_rate": 6.8620792733490305e-06, "loss": 3.1603, "step": 235200 }, { "epoch": 0.0043776, "grad_norm": 0.7638301253318787, "learning_rate": 6.861314678275388e-06, "loss": 2.7251, "step": 235210 }, { "epoch": 0.0044032, "grad_norm": 0.7477211952209473, "learning_rate": 6.860550103555284e-06, "loss": 2.8505, "step": 235220 }, { "epoch": 0.0044288, "grad_norm": 0.7670096755027771, "learning_rate": 6.859785549193679e-06, "loss": 2.9253, "step": 235230 }, { "epoch": 0.0044544, "grad_norm": 0.9042599201202393, "learning_rate": 6.859021015195533e-06, "loss": 2.8709, "step": 235240 }, { "epoch": 0.00448, "grad_norm": 0.7699570655822754, "learning_rate": 6.8582565015657965e-06, "loss": 2.7062, "step": 235250 }, { "epoch": 0.0045056, "grad_norm": 0.7982677817344666, "learning_rate": 6.857492008309432e-06, "loss": 2.8918, "step": 235260 }, { "epoch": 0.0045312, "grad_norm": 2.19453501701355, "learning_rate": 6.856727535431398e-06, "loss": 3.0306, "step": 235270 }, { "epoch": 0.0045568, "grad_norm": 0.8227958679199219, "learning_rate": 6.855963082936652e-06, "loss": 2.7878, "step": 235280 }, { "epoch": 0.0045824, "grad_norm": 2.2006285190582275, "learning_rate": 6.855198650830148e-06, "loss": 2.886, "step": 235290 }, { "epoch": 0.004608, "grad_norm": 0.8550253510475159, "learning_rate": 6.854434239116845e-06, "loss": 3.0687, "step": 235300 }, { "epoch": 0.0046336, "grad_norm": 0.7555240988731384, "learning_rate": 6.8536698478017e-06, "loss": 2.8637, "step": 235310 }, { "epoch": 0.0046592, "grad_norm": 0.9240103960037231, "learning_rate": 6.852905476889671e-06, "loss": 2.8251, "step": 235320 }, { "epoch": 0.0046848, "grad_norm": 0.8348448276519775, "learning_rate": 6.8521411263857095e-06, "loss": 3.1223, "step": 235330 }, { "epoch": 0.0047104, "grad_norm": 1.0025163888931274, "learning_rate": 6.85137679629478e-06, "loss": 3.0477, "step": 235340 }, { "epoch": 0.004736, "grad_norm": 0.8275322318077087, "learning_rate": 6.850612486621833e-06, "loss": 2.9841, "step": 235350 }, { "epoch": 0.0047616, "grad_norm": 0.9590789675712585, "learning_rate": 6.849848197371827e-06, "loss": 2.8408, "step": 235360 }, { "epoch": 0.0047872, "grad_norm": 0.9216554760932922, "learning_rate": 6.849083928549722e-06, "loss": 2.7107, "step": 235370 }, { "epoch": 0.0048128, "grad_norm": 0.7485522627830505, "learning_rate": 6.848319680160468e-06, "loss": 2.9249, "step": 235380 }, { "epoch": 0.0048384, "grad_norm": 0.8731443285942078, "learning_rate": 6.847555452209022e-06, "loss": 2.8038, "step": 235390 }, { "epoch": 0.004864, "grad_norm": 0.8239305019378662, "learning_rate": 6.8467912447003415e-06, "loss": 2.8407, "step": 235400 }, { "epoch": 0.0048896, "grad_norm": 0.7502994537353516, "learning_rate": 6.846027057639385e-06, "loss": 2.7079, "step": 235410 }, { "epoch": 0.0049152, "grad_norm": 0.8457599878311157, "learning_rate": 6.845262891031101e-06, "loss": 2.7792, "step": 235420 }, { "epoch": 0.0049408, "grad_norm": 0.7700570225715637, "learning_rate": 6.84449874488045e-06, "loss": 2.8515, "step": 235430 }, { "epoch": 0.0049664, "grad_norm": 0.7475781440734863, "learning_rate": 6.843734619192388e-06, "loss": 2.6361, "step": 235440 }, { "epoch": 0.004992, "grad_norm": 0.7307766079902649, "learning_rate": 6.842970513971866e-06, "loss": 2.7684, "step": 235450 }, { "epoch": 0.0050176, "grad_norm": 1.0085039138793945, "learning_rate": 6.842206429223841e-06, "loss": 2.972, "step": 235460 }, { "epoch": 0.0050432, "grad_norm": 0.7681459784507751, "learning_rate": 6.841442364953268e-06, "loss": 2.8859, "step": 235470 }, { "epoch": 0.0050688, "grad_norm": 1.4737964868545532, "learning_rate": 6.840678321165104e-06, "loss": 2.7794, "step": 235480 }, { "epoch": 0.0050944, "grad_norm": 0.7710636258125305, "learning_rate": 6.839914297864299e-06, "loss": 2.6596, "step": 235490 }, { "epoch": 0.00512, "grad_norm": 0.7751798033714294, "learning_rate": 6.839150295055808e-06, "loss": 2.8032, "step": 235500 }, { "epoch": 0.0051456, "grad_norm": 0.9748345017433167, "learning_rate": 6.838386312744589e-06, "loss": 2.9186, "step": 235510 }, { "epoch": 0.0051712, "grad_norm": 0.7218871116638184, "learning_rate": 6.837622350935596e-06, "loss": 2.7648, "step": 235520 }, { "epoch": 0.0051968, "grad_norm": 0.9802530407905579, "learning_rate": 6.836858409633782e-06, "loss": 2.8956, "step": 235530 }, { "epoch": 0.0052224, "grad_norm": 0.8095552921295166, "learning_rate": 6.8360944888441e-06, "loss": 2.8409, "step": 235540 }, { "epoch": 0.005248, "grad_norm": 1.003595232963562, "learning_rate": 6.8353305885715025e-06, "loss": 3.1378, "step": 235550 }, { "epoch": 0.0052736, "grad_norm": 3.3910117149353027, "learning_rate": 6.8345667088209466e-06, "loss": 2.9832, "step": 235560 }, { "epoch": 0.0052992, "grad_norm": 0.9597769975662231, "learning_rate": 6.833802849597386e-06, "loss": 2.8678, "step": 235570 }, { "epoch": 0.0053248, "grad_norm": 0.7737454175949097, "learning_rate": 6.833039010905771e-06, "loss": 2.7538, "step": 235580 }, { "epoch": 0.0053504, "grad_norm": 0.7734951972961426, "learning_rate": 6.832275192751054e-06, "loss": 2.9308, "step": 235590 }, { "epoch": 0.005376, "grad_norm": 0.8842957019805908, "learning_rate": 6.831511395138193e-06, "loss": 2.8836, "step": 235600 }, { "epoch": 0.0054016, "grad_norm": 0.814301609992981, "learning_rate": 6.83074761807214e-06, "loss": 2.9667, "step": 235610 }, { "epoch": 0.0054272, "grad_norm": 0.7960047125816345, "learning_rate": 6.829983861557843e-06, "loss": 2.7038, "step": 235620 }, { "epoch": 0.0054528, "grad_norm": 0.928648054599762, "learning_rate": 6.82922012560026e-06, "loss": 2.9667, "step": 235630 }, { "epoch": 0.0054784, "grad_norm": 0.7793529629707336, "learning_rate": 6.8284564102043425e-06, "loss": 2.9394, "step": 235640 }, { "epoch": 0.005504, "grad_norm": 0.7980348467826843, "learning_rate": 6.827692715375041e-06, "loss": 2.7084, "step": 235650 }, { "epoch": 0.0055296, "grad_norm": 0.7872781753540039, "learning_rate": 6.826929041117309e-06, "loss": 2.7543, "step": 235660 }, { "epoch": 0.0055552, "grad_norm": 0.8884873390197754, "learning_rate": 6.826165387436099e-06, "loss": 2.8585, "step": 235670 }, { "epoch": 0.0055808, "grad_norm": 0.78144770860672, "learning_rate": 6.82540175433636e-06, "loss": 2.6761, "step": 235680 }, { "epoch": 0.0056064, "grad_norm": 0.9659698605537415, "learning_rate": 6.82463814182305e-06, "loss": 2.8044, "step": 235690 }, { "epoch": 0.005632, "grad_norm": 0.8381772637367249, "learning_rate": 6.823874549901119e-06, "loss": 2.7536, "step": 235700 }, { "epoch": 0.0056576, "grad_norm": 0.7724138498306274, "learning_rate": 6.8231109785755154e-06, "loss": 3.0963, "step": 235710 }, { "epoch": 0.0056832, "grad_norm": 0.85874342918396, "learning_rate": 6.8223474278511945e-06, "loss": 2.9373, "step": 235720 }, { "epoch": 0.0057088, "grad_norm": 0.8162028193473816, "learning_rate": 6.8215838977331066e-06, "loss": 2.7074, "step": 235730 }, { "epoch": 0.0057344, "grad_norm": 0.7608163952827454, "learning_rate": 6.8208203882262e-06, "loss": 2.9331, "step": 235740 }, { "epoch": 0.00576, "grad_norm": 0.7282572984695435, "learning_rate": 6.82005689933543e-06, "loss": 2.6798, "step": 235750 }, { "epoch": 0.0057856, "grad_norm": 0.7926065325737, "learning_rate": 6.819293431065743e-06, "loss": 2.8421, "step": 235760 }, { "epoch": 0.0058112, "grad_norm": 0.8163644075393677, "learning_rate": 6.818529983422098e-06, "loss": 2.8052, "step": 235770 }, { "epoch": 0.0058368, "grad_norm": 0.7888874411582947, "learning_rate": 6.817766556409437e-06, "loss": 2.7028, "step": 235780 }, { "epoch": 0.0058624, "grad_norm": 0.7565310001373291, "learning_rate": 6.817003150032712e-06, "loss": 2.6544, "step": 235790 }, { "epoch": 0.005888, "grad_norm": 1.0218042135238647, "learning_rate": 6.816239764296879e-06, "loss": 2.857, "step": 235800 }, { "epoch": 0.0059136, "grad_norm": 0.7494388818740845, "learning_rate": 6.815476399206886e-06, "loss": 2.8957, "step": 235810 }, { "epoch": 0.0059392, "grad_norm": 0.7404144406318665, "learning_rate": 6.81471305476768e-06, "loss": 2.7113, "step": 235820 }, { "epoch": 0.0059648, "grad_norm": 0.7640352845191956, "learning_rate": 6.813949730984213e-06, "loss": 2.8764, "step": 235830 }, { "epoch": 0.0059904, "grad_norm": 0.8554592132568359, "learning_rate": 6.813186427861438e-06, "loss": 2.8794, "step": 235840 }, { "epoch": 0.006016, "grad_norm": 0.7751770615577698, "learning_rate": 6.8124231454042945e-06, "loss": 2.7914, "step": 235850 }, { "epoch": 0.0060416, "grad_norm": 0.7796262502670288, "learning_rate": 6.811659883617747e-06, "loss": 2.7083, "step": 235860 }, { "epoch": 0.0060672, "grad_norm": 0.9439787268638611, "learning_rate": 6.8108966425067355e-06, "loss": 2.6578, "step": 235870 }, { "epoch": 0.0060928, "grad_norm": 0.8082603812217712, "learning_rate": 6.810133422076212e-06, "loss": 2.9051, "step": 235880 }, { "epoch": 0.0061184, "grad_norm": 0.7804611325263977, "learning_rate": 6.809370222331126e-06, "loss": 2.617, "step": 235890 }, { "epoch": 0.006144, "grad_norm": 0.8228448629379272, "learning_rate": 6.808607043276427e-06, "loss": 2.7903, "step": 235900 }, { "epoch": 0.0061696, "grad_norm": 0.7918440699577332, "learning_rate": 6.807843884917061e-06, "loss": 2.735, "step": 235910 }, { "epoch": 0.0061952, "grad_norm": 0.7670903205871582, "learning_rate": 6.80708074725798e-06, "loss": 2.8652, "step": 235920 }, { "epoch": 0.0062208, "grad_norm": 0.8110995888710022, "learning_rate": 6.806317630304133e-06, "loss": 2.867, "step": 235930 }, { "epoch": 0.0062464, "grad_norm": 0.7129989862442017, "learning_rate": 6.805554534060466e-06, "loss": 2.843, "step": 235940 }, { "epoch": 0.006272, "grad_norm": 0.8175245523452759, "learning_rate": 6.80479145853193e-06, "loss": 2.7088, "step": 235950 }, { "epoch": 0.0062976, "grad_norm": 0.9129753112792969, "learning_rate": 6.8040284037234705e-06, "loss": 2.8975, "step": 235960 }, { "epoch": 0.0063232, "grad_norm": 0.7936801314353943, "learning_rate": 6.80326536964004e-06, "loss": 2.974, "step": 235970 }, { "epoch": 0.0063488, "grad_norm": 0.8271650075912476, "learning_rate": 6.802502356286581e-06, "loss": 2.8337, "step": 235980 }, { "epoch": 0.0063744, "grad_norm": 0.8537148237228394, "learning_rate": 6.801739363668045e-06, "loss": 2.7439, "step": 235990 }, { "epoch": 0.0064, "grad_norm": 0.9836148619651794, "learning_rate": 6.800976391789382e-06, "loss": 2.7834, "step": 236000 }, { "epoch": 0.0064256, "grad_norm": 0.8207709789276123, "learning_rate": 6.800213440655534e-06, "loss": 2.7343, "step": 236010 }, { "epoch": 0.0064512, "grad_norm": 0.7514338493347168, "learning_rate": 6.799450510271451e-06, "loss": 2.8496, "step": 236020 }, { "epoch": 0.0064768, "grad_norm": 1.0763107538223267, "learning_rate": 6.7986876006420795e-06, "loss": 2.7954, "step": 236030 }, { "epoch": 0.0065024, "grad_norm": 0.8899896144866943, "learning_rate": 6.7979247117723705e-06, "loss": 2.7309, "step": 236040 }, { "epoch": 0.006528, "grad_norm": 0.8019543290138245, "learning_rate": 6.7971618436672685e-06, "loss": 2.9154, "step": 236050 }, { "epoch": 0.0065536, "grad_norm": 0.7878087162971497, "learning_rate": 6.7963989963317235e-06, "loss": 2.711, "step": 236060 }, { "epoch": 0.0065792, "grad_norm": 0.7483411431312561, "learning_rate": 6.795636169770676e-06, "loss": 2.8784, "step": 236070 }, { "epoch": 0.0066048, "grad_norm": 0.8506212830543518, "learning_rate": 6.794873363989076e-06, "loss": 2.9543, "step": 236080 }, { "epoch": 0.0066304, "grad_norm": 0.7162821888923645, "learning_rate": 6.794110578991873e-06, "loss": 2.6509, "step": 236090 }, { "epoch": 0.006656, "grad_norm": 0.941387951374054, "learning_rate": 6.79334781478401e-06, "loss": 3.1592, "step": 236100 }, { "epoch": 0.0066816, "grad_norm": 0.9446196556091309, "learning_rate": 6.792585071370434e-06, "loss": 2.9388, "step": 236110 }, { "epoch": 0.0067072, "grad_norm": 0.7507122159004211, "learning_rate": 6.791822348756092e-06, "loss": 2.8662, "step": 236120 }, { "epoch": 0.0067328, "grad_norm": 0.8929161429405212, "learning_rate": 6.79105964694593e-06, "loss": 2.8855, "step": 236130 }, { "epoch": 0.0067584, "grad_norm": 0.748494565486908, "learning_rate": 6.7902969659448914e-06, "loss": 2.7306, "step": 236140 }, { "epoch": 0.006784, "grad_norm": 0.7358953952789307, "learning_rate": 6.7895343057579244e-06, "loss": 2.5676, "step": 236150 }, { "epoch": 0.0068096, "grad_norm": 0.8506143689155579, "learning_rate": 6.788771666389974e-06, "loss": 2.9947, "step": 236160 }, { "epoch": 0.0068352, "grad_norm": 0.7862493991851807, "learning_rate": 6.788009047845988e-06, "loss": 2.9075, "step": 236170 }, { "epoch": 0.0068608, "grad_norm": 1.1436760425567627, "learning_rate": 6.787246450130908e-06, "loss": 2.7346, "step": 236180 }, { "epoch": 0.0068864, "grad_norm": 0.8102016448974609, "learning_rate": 6.78648387324968e-06, "loss": 2.8864, "step": 236190 }, { "epoch": 0.006912, "grad_norm": 0.7731093764305115, "learning_rate": 6.785721317207248e-06, "loss": 2.7645, "step": 236200 }, { "epoch": 0.0069376, "grad_norm": 0.835671603679657, "learning_rate": 6.784958782008562e-06, "loss": 2.7046, "step": 236210 }, { "epoch": 0.0069632, "grad_norm": 0.7667710185050964, "learning_rate": 6.784196267658565e-06, "loss": 2.8807, "step": 236220 }, { "epoch": 0.0069888, "grad_norm": 1.104300856590271, "learning_rate": 6.7834337741622e-06, "loss": 3.2574, "step": 236230 }, { "epoch": 0.0070144, "grad_norm": 0.904937207698822, "learning_rate": 6.78267130152441e-06, "loss": 2.9545, "step": 236240 }, { "epoch": 0.00704, "grad_norm": 0.9134624004364014, "learning_rate": 6.781908849750143e-06, "loss": 2.8304, "step": 236250 }, { "epoch": 0.0070656, "grad_norm": 0.7057270407676697, "learning_rate": 6.781146418844344e-06, "loss": 2.799, "step": 236260 }, { "epoch": 0.0070912, "grad_norm": 0.7317230105400085, "learning_rate": 6.780384008811952e-06, "loss": 2.9202, "step": 236270 }, { "epoch": 0.0071168, "grad_norm": 1.0501106977462769, "learning_rate": 6.779621619657914e-06, "loss": 2.8054, "step": 236280 }, { "epoch": 0.0071424, "grad_norm": 0.7888480424880981, "learning_rate": 6.7788592513871755e-06, "loss": 2.8922, "step": 236290 }, { "epoch": 0.007168, "grad_norm": 0.7479458451271057, "learning_rate": 6.778096904004679e-06, "loss": 2.9312, "step": 236300 }, { "epoch": 0.0071936, "grad_norm": 0.8943301439285278, "learning_rate": 6.777334577515367e-06, "loss": 3.0145, "step": 236310 }, { "epoch": 0.0072192, "grad_norm": 0.7146400213241577, "learning_rate": 6.776572271924182e-06, "loss": 2.8799, "step": 236320 }, { "epoch": 0.0072448, "grad_norm": 0.7954562902450562, "learning_rate": 6.775809987236073e-06, "loss": 2.8765, "step": 236330 }, { "epoch": 0.0072704, "grad_norm": 0.936475396156311, "learning_rate": 6.775047723455976e-06, "loss": 2.8693, "step": 236340 }, { "epoch": 0.007296, "grad_norm": 0.7575796246528625, "learning_rate": 6.774285480588839e-06, "loss": 2.8938, "step": 236350 }, { "epoch": 0.0073216, "grad_norm": 0.9413476586341858, "learning_rate": 6.773523258639603e-06, "loss": 2.8457, "step": 236360 }, { "epoch": 0.0073472, "grad_norm": 0.8706610798835754, "learning_rate": 6.772761057613209e-06, "loss": 2.7815, "step": 236370 }, { "epoch": 0.0073728, "grad_norm": 0.7710934281349182, "learning_rate": 6.771998877514603e-06, "loss": 2.8325, "step": 236380 }, { "epoch": 0.0073984, "grad_norm": 0.7656084895133972, "learning_rate": 6.771236718348729e-06, "loss": 2.8936, "step": 236390 }, { "epoch": 0.007424, "grad_norm": 0.7196746468544006, "learning_rate": 6.770474580120524e-06, "loss": 2.8723, "step": 236400 }, { "epoch": 0.0074496, "grad_norm": 0.8875367641448975, "learning_rate": 6.769712462834934e-06, "loss": 2.6747, "step": 236410 }, { "epoch": 0.0074752, "grad_norm": 0.954362154006958, "learning_rate": 6.768950366496902e-06, "loss": 2.9641, "step": 236420 }, { "epoch": 0.0075008, "grad_norm": 1.2960307598114014, "learning_rate": 6.768188291111366e-06, "loss": 2.9147, "step": 236430 }, { "epoch": 0.0075264, "grad_norm": 0.787317156791687, "learning_rate": 6.76742623668327e-06, "loss": 2.8871, "step": 236440 }, { "epoch": 0.007552, "grad_norm": 0.7756245136260986, "learning_rate": 6.766664203217556e-06, "loss": 2.8862, "step": 236450 }, { "epoch": 0.0075776, "grad_norm": 0.8219406604766846, "learning_rate": 6.765902190719167e-06, "loss": 2.6724, "step": 236460 }, { "epoch": 0.0076032, "grad_norm": 0.7627509236335754, "learning_rate": 6.765140199193041e-06, "loss": 2.7507, "step": 236470 }, { "epoch": 0.0076288, "grad_norm": 0.7857176065444946, "learning_rate": 6.7643782286441205e-06, "loss": 2.9617, "step": 236480 }, { "epoch": 0.0076544, "grad_norm": 0.77397221326828, "learning_rate": 6.763616279077347e-06, "loss": 2.9511, "step": 236490 }, { "epoch": 0.00768, "grad_norm": 0.8121401071548462, "learning_rate": 6.762854350497665e-06, "loss": 2.7912, "step": 236500 }, { "epoch": 0.0077056, "grad_norm": 0.8017376065254211, "learning_rate": 6.76209244291001e-06, "loss": 2.9306, "step": 236510 }, { "epoch": 0.0077312, "grad_norm": 0.723939061164856, "learning_rate": 6.761330556319326e-06, "loss": 2.8692, "step": 236520 }, { "epoch": 0.0077568, "grad_norm": 1.0049083232879639, "learning_rate": 6.760568690730552e-06, "loss": 2.7192, "step": 236530 }, { "epoch": 0.0077824, "grad_norm": 0.8257355690002441, "learning_rate": 6.7598068461486286e-06, "loss": 2.8257, "step": 236540 }, { "epoch": 0.007808, "grad_norm": 0.8432038426399231, "learning_rate": 6.759045022578494e-06, "loss": 3.0196, "step": 236550 }, { "epoch": 0.0078336, "grad_norm": 1.040366291999817, "learning_rate": 6.758283220025093e-06, "loss": 2.9424, "step": 236560 }, { "epoch": 0.0078592, "grad_norm": 1.016858458518982, "learning_rate": 6.7575214384933655e-06, "loss": 2.8539, "step": 236570 }, { "epoch": 0.0078848, "grad_norm": 0.830888032913208, "learning_rate": 6.756759677988248e-06, "loss": 2.8732, "step": 236580 }, { "epoch": 0.0079104, "grad_norm": 0.9490156769752502, "learning_rate": 6.7559979385146845e-06, "loss": 2.9467, "step": 236590 }, { "epoch": 0.007936, "grad_norm": 0.8691385984420776, "learning_rate": 6.75523622007761e-06, "loss": 2.704, "step": 236600 }, { "epoch": 0.0079616, "grad_norm": 1.0094629526138306, "learning_rate": 6.754474522681967e-06, "loss": 2.918, "step": 236610 }, { "epoch": 0.0079872, "grad_norm": 0.972903311252594, "learning_rate": 6.753712846332696e-06, "loss": 2.8572, "step": 236620 }, { "epoch": 0.0080128, "grad_norm": 0.7242955565452576, "learning_rate": 6.752951191034733e-06, "loss": 2.776, "step": 236630 }, { "epoch": 0.0080384, "grad_norm": 0.8909826278686523, "learning_rate": 6.752189556793016e-06, "loss": 2.7867, "step": 236640 }, { "epoch": 0.008064, "grad_norm": 0.7992689609527588, "learning_rate": 6.75142794361249e-06, "loss": 2.764, "step": 236650 }, { "epoch": 0.0080896, "grad_norm": 0.8260809779167175, "learning_rate": 6.750666351498089e-06, "loss": 2.9698, "step": 236660 }, { "epoch": 0.0081152, "grad_norm": 0.7663139700889587, "learning_rate": 6.7499047804547546e-06, "loss": 2.7944, "step": 236670 }, { "epoch": 0.0081408, "grad_norm": 0.763913094997406, "learning_rate": 6.7491432304874215e-06, "loss": 2.9517, "step": 236680 }, { "epoch": 0.0081664, "grad_norm": 0.7808148264884949, "learning_rate": 6.748381701601031e-06, "loss": 2.965, "step": 236690 }, { "epoch": 0.008192, "grad_norm": 0.77100670337677, "learning_rate": 6.7476201938005235e-06, "loss": 2.8791, "step": 236700 }, { "epoch": 0.0082176, "grad_norm": 2.334894895553589, "learning_rate": 6.746858707090833e-06, "loss": 2.6632, "step": 236710 }, { "epoch": 0.0082432, "grad_norm": 0.8593084812164307, "learning_rate": 6.746097241476897e-06, "loss": 2.8299, "step": 236720 }, { "epoch": 0.0082688, "grad_norm": 0.8110367655754089, "learning_rate": 6.745335796963657e-06, "loss": 2.9669, "step": 236730 }, { "epoch": 0.0082944, "grad_norm": 0.7627100348472595, "learning_rate": 6.744574373556049e-06, "loss": 2.6489, "step": 236740 }, { "epoch": 0.00832, "grad_norm": 0.8588494062423706, "learning_rate": 6.743812971259014e-06, "loss": 2.9475, "step": 236750 }, { "epoch": 0.0083456, "grad_norm": 0.8395514488220215, "learning_rate": 6.743051590077483e-06, "loss": 2.9916, "step": 236760 }, { "epoch": 0.0083712, "grad_norm": 0.7209861874580383, "learning_rate": 6.742290230016399e-06, "loss": 2.8347, "step": 236770 }, { "epoch": 0.0083968, "grad_norm": 0.8710430264472961, "learning_rate": 6.741528891080696e-06, "loss": 2.8657, "step": 236780 }, { "epoch": 0.0084224, "grad_norm": 0.779486894607544, "learning_rate": 6.740767573275313e-06, "loss": 2.8263, "step": 236790 }, { "epoch": 0.008448, "grad_norm": 0.8118667602539062, "learning_rate": 6.740006276605186e-06, "loss": 2.9468, "step": 236800 }, { "epoch": 0.0084736, "grad_norm": 0.7947338819503784, "learning_rate": 6.739245001075251e-06, "loss": 2.7124, "step": 236810 }, { "epoch": 0.0084992, "grad_norm": 0.7312729954719543, "learning_rate": 6.738483746690447e-06, "loss": 2.9036, "step": 236820 }, { "epoch": 0.0085248, "grad_norm": 0.7738308310508728, "learning_rate": 6.737722513455709e-06, "loss": 3.0048, "step": 236830 }, { "epoch": 0.0085504, "grad_norm": 0.7943831086158752, "learning_rate": 6.736961301375972e-06, "loss": 3.0885, "step": 236840 }, { "epoch": 0.008576, "grad_norm": 0.9029412269592285, "learning_rate": 6.7362001104561736e-06, "loss": 2.9253, "step": 236850 }, { "epoch": 0.0086016, "grad_norm": 0.7447136044502258, "learning_rate": 6.735438940701252e-06, "loss": 2.7704, "step": 236860 }, { "epoch": 0.0086272, "grad_norm": 0.8857337236404419, "learning_rate": 6.73467779211614e-06, "loss": 2.8832, "step": 236870 }, { "epoch": 0.0086528, "grad_norm": 0.7130148410797119, "learning_rate": 6.733916664705775e-06, "loss": 2.9326, "step": 236880 }, { "epoch": 0.0086784, "grad_norm": 0.8687919974327087, "learning_rate": 6.733155558475091e-06, "loss": 2.8253, "step": 236890 }, { "epoch": 0.008704, "grad_norm": 0.8035460710525513, "learning_rate": 6.732394473429025e-06, "loss": 2.7931, "step": 236900 }, { "epoch": 0.0087296, "grad_norm": 0.8296636343002319, "learning_rate": 6.7316334095725126e-06, "loss": 2.7581, "step": 236910 }, { "epoch": 0.0087552, "grad_norm": 0.7617623209953308, "learning_rate": 6.730872366910491e-06, "loss": 2.8526, "step": 236920 }, { "epoch": 0.0087808, "grad_norm": 0.7511303424835205, "learning_rate": 6.730111345447892e-06, "loss": 2.9285, "step": 236930 }, { "epoch": 0.0088064, "grad_norm": 0.8448120355606079, "learning_rate": 6.729350345189652e-06, "loss": 2.9406, "step": 236940 }, { "epoch": 0.008832, "grad_norm": 0.7516356706619263, "learning_rate": 6.7285893661407074e-06, "loss": 2.9127, "step": 236950 }, { "epoch": 0.0088576, "grad_norm": 0.7740426063537598, "learning_rate": 6.72782840830599e-06, "loss": 2.7489, "step": 236960 }, { "epoch": 0.0088832, "grad_norm": 0.7956497073173523, "learning_rate": 6.727067471690434e-06, "loss": 2.66, "step": 236970 }, { "epoch": 0.0089088, "grad_norm": 0.7426387667655945, "learning_rate": 6.726306556298977e-06, "loss": 2.8787, "step": 236980 }, { "epoch": 0.0089344, "grad_norm": 0.7654321193695068, "learning_rate": 6.7255456621365545e-06, "loss": 2.8745, "step": 236990 }, { "epoch": 0.00896, "grad_norm": 0.8549219369888306, "learning_rate": 6.724784789208094e-06, "loss": 3.0036, "step": 237000 }, { "epoch": 0.0089856, "grad_norm": 1.138397216796875, "learning_rate": 6.724023937518536e-06, "loss": 2.5635, "step": 237010 }, { "epoch": 0.0090112, "grad_norm": 0.7801642417907715, "learning_rate": 6.723263107072812e-06, "loss": 2.9338, "step": 237020 }, { "epoch": 0.0090368, "grad_norm": 0.7825180292129517, "learning_rate": 6.7225022978758556e-06, "loss": 3.1342, "step": 237030 }, { "epoch": 0.0090624, "grad_norm": 0.745024561882019, "learning_rate": 6.7217415099326e-06, "loss": 2.7902, "step": 237040 }, { "epoch": 0.009088, "grad_norm": 0.9312148094177246, "learning_rate": 6.72098074324798e-06, "loss": 2.849, "step": 237050 }, { "epoch": 0.0091136, "grad_norm": 0.82979416847229, "learning_rate": 6.72021999782693e-06, "loss": 2.8249, "step": 237060 }, { "epoch": 0.0091392, "grad_norm": 0.7602632641792297, "learning_rate": 6.719459273674376e-06, "loss": 2.927, "step": 237070 }, { "epoch": 0.0091648, "grad_norm": 0.9446355104446411, "learning_rate": 6.7186985707952625e-06, "loss": 2.885, "step": 237080 }, { "epoch": 0.0091904, "grad_norm": 0.7401157021522522, "learning_rate": 6.7179378891945145e-06, "loss": 3.0076, "step": 237090 }, { "epoch": 0.009216, "grad_norm": 0.7381459474563599, "learning_rate": 6.717177228877068e-06, "loss": 2.902, "step": 237100 }, { "epoch": 0.0092416, "grad_norm": 0.7789002060890198, "learning_rate": 6.7164165898478526e-06, "loss": 2.7627, "step": 237110 }, { "epoch": 0.0092672, "grad_norm": 0.7923112511634827, "learning_rate": 6.7156559721118065e-06, "loss": 2.9335, "step": 237120 }, { "epoch": 0.0092928, "grad_norm": 0.8455477952957153, "learning_rate": 6.714895375673856e-06, "loss": 2.709, "step": 237130 }, { "epoch": 0.0093184, "grad_norm": 0.7847411036491394, "learning_rate": 6.714134800538936e-06, "loss": 2.9744, "step": 237140 }, { "epoch": 0.009344, "grad_norm": 0.8106456995010376, "learning_rate": 6.713374246711981e-06, "loss": 2.9183, "step": 237150 }, { "epoch": 0.0093696, "grad_norm": 1.939729928970337, "learning_rate": 6.712613714197919e-06, "loss": 3.313, "step": 237160 }, { "epoch": 0.0093952, "grad_norm": 0.7810055017471313, "learning_rate": 6.711853203001681e-06, "loss": 3.0177, "step": 237170 }, { "epoch": 0.0094208, "grad_norm": 0.7378543019294739, "learning_rate": 6.7110927131282045e-06, "loss": 2.92, "step": 237180 }, { "epoch": 0.0094464, "grad_norm": 0.7678629159927368, "learning_rate": 6.710332244582417e-06, "loss": 2.8836, "step": 237190 }, { "epoch": 0.009472, "grad_norm": 0.8424858450889587, "learning_rate": 6.70957179736925e-06, "loss": 2.679, "step": 237200 }, { "epoch": 0.0094976, "grad_norm": 0.7749304175376892, "learning_rate": 6.7088113714936354e-06, "loss": 2.9405, "step": 237210 }, { "epoch": 0.0095232, "grad_norm": 0.9682620167732239, "learning_rate": 6.7080509669605044e-06, "loss": 2.9181, "step": 237220 }, { "epoch": 0.0095488, "grad_norm": 0.794719398021698, "learning_rate": 6.707290583774788e-06, "loss": 3.1094, "step": 237230 }, { "epoch": 0.0095744, "grad_norm": 0.8116098046302795, "learning_rate": 6.706530221941414e-06, "loss": 2.7119, "step": 237240 }, { "epoch": 0.0096, "grad_norm": 0.7870550751686096, "learning_rate": 6.705769881465319e-06, "loss": 2.9524, "step": 237250 }, { "epoch": 0.0096256, "grad_norm": 0.8798168897628784, "learning_rate": 6.70500956235143e-06, "loss": 2.8683, "step": 237260 }, { "epoch": 0.0096512, "grad_norm": 0.7793520092964172, "learning_rate": 6.70424926460468e-06, "loss": 2.7798, "step": 237270 }, { "epoch": 0.0096768, "grad_norm": 0.9958526492118835, "learning_rate": 6.703488988229998e-06, "loss": 3.0184, "step": 237280 }, { "epoch": 0.0097024, "grad_norm": 0.7732696533203125, "learning_rate": 6.7027287332323125e-06, "loss": 2.9282, "step": 237290 }, { "epoch": 0.009728, "grad_norm": 0.8466508388519287, "learning_rate": 6.701968499616555e-06, "loss": 2.9768, "step": 237300 }, { "epoch": 0.0097536, "grad_norm": 0.7695175409317017, "learning_rate": 6.701208287387656e-06, "loss": 2.8453, "step": 237310 }, { "epoch": 0.0097792, "grad_norm": 0.8319741487503052, "learning_rate": 6.700448096550544e-06, "loss": 2.8116, "step": 237320 }, { "epoch": 0.0098048, "grad_norm": 0.8815202116966248, "learning_rate": 6.699687927110148e-06, "loss": 2.8194, "step": 237330 }, { "epoch": 0.0098304, "grad_norm": 0.8724138140678406, "learning_rate": 6.6989277790714e-06, "loss": 2.7871, "step": 237340 }, { "epoch": 0.009856, "grad_norm": 0.8185709714889526, "learning_rate": 6.698167652439229e-06, "loss": 2.7277, "step": 237350 }, { "epoch": 0.0098816, "grad_norm": 0.8061392903327942, "learning_rate": 6.697407547218562e-06, "loss": 2.8889, "step": 237360 }, { "epoch": 0.0099072, "grad_norm": 0.7754219770431519, "learning_rate": 6.696647463414327e-06, "loss": 2.7384, "step": 237370 }, { "epoch": 0.0099328, "grad_norm": 1.2267762422561646, "learning_rate": 6.695887401031458e-06, "loss": 2.8806, "step": 237380 }, { "epoch": 0.0099584, "grad_norm": 0.7305707931518555, "learning_rate": 6.695127360074883e-06, "loss": 2.9942, "step": 237390 }, { "epoch": 0.009984, "grad_norm": 0.7992395758628845, "learning_rate": 6.694367340549526e-06, "loss": 3.1945, "step": 237400 }, { "epoch": 0.0100096, "grad_norm": 0.9041965007781982, "learning_rate": 6.693607342460318e-06, "loss": 2.7606, "step": 237410 }, { "epoch": 0.0100352, "grad_norm": 0.7598696351051331, "learning_rate": 6.6928473658121855e-06, "loss": 2.7704, "step": 237420 }, { "epoch": 0.0100608, "grad_norm": 0.7118558287620544, "learning_rate": 6.692087410610062e-06, "loss": 2.9111, "step": 237430 }, { "epoch": 0.0100864, "grad_norm": 2.364391326904297, "learning_rate": 6.6913274768588735e-06, "loss": 2.9008, "step": 237440 }, { "epoch": 0.010112, "grad_norm": 0.6886769533157349, "learning_rate": 6.690567564563545e-06, "loss": 2.9869, "step": 237450 }, { "epoch": 0.0101376, "grad_norm": 0.9131158590316772, "learning_rate": 6.689807673729007e-06, "loss": 2.8466, "step": 237460 }, { "epoch": 0.0101632, "grad_norm": 0.9166131019592285, "learning_rate": 6.689047804360185e-06, "loss": 2.9704, "step": 237470 }, { "epoch": 0.0101888, "grad_norm": 0.756111204624176, "learning_rate": 6.6882879564620115e-06, "loss": 2.9452, "step": 237480 }, { "epoch": 0.0102144, "grad_norm": 0.8848692774772644, "learning_rate": 6.687528130039408e-06, "loss": 2.7989, "step": 237490 }, { "epoch": 0.01024, "grad_norm": 0.9819674491882324, "learning_rate": 6.686768325097304e-06, "loss": 2.6175, "step": 237500 }, { "epoch": 0.0102656, "grad_norm": 0.8531841039657593, "learning_rate": 6.686008541640629e-06, "loss": 2.7259, "step": 237510 }, { "epoch": 0.0102912, "grad_norm": 0.8018609285354614, "learning_rate": 6.685248779674307e-06, "loss": 2.9225, "step": 237520 }, { "epoch": 0.0103168, "grad_norm": 0.7877097725868225, "learning_rate": 6.684489039203263e-06, "loss": 2.7545, "step": 237530 }, { "epoch": 0.0103424, "grad_norm": 1.0056298971176147, "learning_rate": 6.68372932023243e-06, "loss": 2.8135, "step": 237540 }, { "epoch": 0.010368, "grad_norm": 1.0054404735565186, "learning_rate": 6.68296962276673e-06, "loss": 2.9952, "step": 237550 }, { "epoch": 0.0103936, "grad_norm": 0.748988926410675, "learning_rate": 6.68220994681109e-06, "loss": 2.7735, "step": 237560 }, { "epoch": 0.0104192, "grad_norm": 0.747715950012207, "learning_rate": 6.681450292370437e-06, "loss": 2.8465, "step": 237570 }, { "epoch": 0.0104448, "grad_norm": 0.8400607705116272, "learning_rate": 6.680690659449695e-06, "loss": 2.996, "step": 237580 }, { "epoch": 0.0104704, "grad_norm": 0.8396015167236328, "learning_rate": 6.679931048053792e-06, "loss": 3.1032, "step": 237590 }, { "epoch": 0.010496, "grad_norm": 0.9335473775863647, "learning_rate": 6.679171458187654e-06, "loss": 2.9657, "step": 237600 }, { "epoch": 0.0105216, "grad_norm": 0.9934448599815369, "learning_rate": 6.67841188985621e-06, "loss": 2.915, "step": 237610 }, { "epoch": 0.0105472, "grad_norm": 0.767137885093689, "learning_rate": 6.677652343064379e-06, "loss": 2.9099, "step": 237620 }, { "epoch": 0.0105728, "grad_norm": 0.8490960597991943, "learning_rate": 6.676892817817091e-06, "loss": 2.719, "step": 237630 }, { "epoch": 0.0105984, "grad_norm": 1.0439094305038452, "learning_rate": 6.6761333141192706e-06, "loss": 2.8984, "step": 237640 }, { "epoch": 0.010624, "grad_norm": 0.9713204503059387, "learning_rate": 6.67537383197584e-06, "loss": 2.9845, "step": 237650 }, { "epoch": 0.0106496, "grad_norm": 0.7344627380371094, "learning_rate": 6.6746143713917275e-06, "loss": 2.8664, "step": 237660 }, { "epoch": 0.0106752, "grad_norm": 0.7661029100418091, "learning_rate": 6.673854932371857e-06, "loss": 2.9067, "step": 237670 }, { "epoch": 0.0107008, "grad_norm": 0.9830145835876465, "learning_rate": 6.673095514921155e-06, "loss": 3.0668, "step": 237680 }, { "epoch": 0.0107264, "grad_norm": 0.7586302757263184, "learning_rate": 6.672336119044542e-06, "loss": 2.725, "step": 237690 }, { "epoch": 0.010752, "grad_norm": 0.9816341400146484, "learning_rate": 6.671576744746946e-06, "loss": 2.8361, "step": 237700 }, { "epoch": 0.0107776, "grad_norm": 0.9187764525413513, "learning_rate": 6.670817392033291e-06, "loss": 2.9179, "step": 237710 }, { "epoch": 0.0108032, "grad_norm": 0.8680880665779114, "learning_rate": 6.670058060908499e-06, "loss": 3.2284, "step": 237720 }, { "epoch": 0.0108288, "grad_norm": 1.127237319946289, "learning_rate": 6.6692987513774955e-06, "loss": 2.838, "step": 237730 }, { "epoch": 0.0108544, "grad_norm": 0.8271328806877136, "learning_rate": 6.668539463445205e-06, "loss": 2.8558, "step": 237740 }, { "epoch": 0.01088, "grad_norm": 0.8827256560325623, "learning_rate": 6.667780197116551e-06, "loss": 2.9013, "step": 237750 }, { "epoch": 0.0109056, "grad_norm": 0.9082823395729065, "learning_rate": 6.667020952396457e-06, "loss": 2.8697, "step": 237760 }, { "epoch": 0.0109312, "grad_norm": 3.904360294342041, "learning_rate": 6.666261729289842e-06, "loss": 2.8186, "step": 237770 }, { "epoch": 0.0109568, "grad_norm": 0.8606892228126526, "learning_rate": 6.665502527801637e-06, "loss": 2.9189, "step": 237780 }, { "epoch": 0.0109824, "grad_norm": 0.8554335832595825, "learning_rate": 6.664743347936761e-06, "loss": 3.0615, "step": 237790 }, { "epoch": 0.011008, "grad_norm": 0.7925039529800415, "learning_rate": 6.663984189700139e-06, "loss": 2.8378, "step": 237800 }, { "epoch": 0.0110336, "grad_norm": 0.8327353000640869, "learning_rate": 6.663225053096696e-06, "loss": 3.1542, "step": 237810 }, { "epoch": 0.0110592, "grad_norm": 0.8629691004753113, "learning_rate": 6.662465938131347e-06, "loss": 2.8668, "step": 237820 }, { "epoch": 0.0110848, "grad_norm": 0.757530927658081, "learning_rate": 6.6617068448090214e-06, "loss": 2.8667, "step": 237830 }, { "epoch": 0.0111104, "grad_norm": 0.7825107574462891, "learning_rate": 6.6609477731346406e-06, "loss": 2.7056, "step": 237840 }, { "epoch": 0.011136, "grad_norm": 0.8097912073135376, "learning_rate": 6.660188723113125e-06, "loss": 2.7385, "step": 237850 }, { "epoch": 0.0111616, "grad_norm": 0.8213067054748535, "learning_rate": 6.659429694749398e-06, "loss": 2.8489, "step": 237860 }, { "epoch": 0.0111872, "grad_norm": 0.7955726981163025, "learning_rate": 6.65867068804838e-06, "loss": 2.8484, "step": 237870 }, { "epoch": 0.0112128, "grad_norm": 0.994694709777832, "learning_rate": 6.657911703014998e-06, "loss": 2.8253, "step": 237880 }, { "epoch": 0.0112384, "grad_norm": 0.7971354722976685, "learning_rate": 6.657152739654167e-06, "loss": 2.7956, "step": 237890 }, { "epoch": 0.011264, "grad_norm": 0.7866383194923401, "learning_rate": 6.656393797970814e-06, "loss": 2.7934, "step": 237900 }, { "epoch": 0.0112896, "grad_norm": 0.7675338387489319, "learning_rate": 6.655634877969859e-06, "loss": 2.87, "step": 237910 }, { "epoch": 0.0113152, "grad_norm": 0.8410564661026001, "learning_rate": 6.654875979656222e-06, "loss": 2.8842, "step": 237920 }, { "epoch": 0.0113408, "grad_norm": 0.8666383624076843, "learning_rate": 6.654117103034825e-06, "loss": 3.0024, "step": 237930 }, { "epoch": 0.0113664, "grad_norm": 0.9737293720245361, "learning_rate": 6.653358248110586e-06, "loss": 2.6869, "step": 237940 }, { "epoch": 0.011392, "grad_norm": 0.8070398569107056, "learning_rate": 6.6525994148884345e-06, "loss": 2.8577, "step": 237950 }, { "epoch": 0.0114176, "grad_norm": 0.7566013336181641, "learning_rate": 6.651840603373283e-06, "loss": 2.7798, "step": 237960 }, { "epoch": 0.0114432, "grad_norm": 0.7995551228523254, "learning_rate": 6.65108181357006e-06, "loss": 2.6828, "step": 237970 }, { "epoch": 0.0114688, "grad_norm": 0.7420449256896973, "learning_rate": 6.650323045483678e-06, "loss": 2.8246, "step": 237980 }, { "epoch": 0.0114944, "grad_norm": 0.846599280834198, "learning_rate": 6.64956429911906e-06, "loss": 2.8274, "step": 237990 }, { "epoch": 0.01152, "grad_norm": 0.8384715914726257, "learning_rate": 6.648805574481127e-06, "loss": 2.8892, "step": 238000 }, { "epoch": 0.0115456, "grad_norm": 0.8839398622512817, "learning_rate": 6.6480468715748025e-06, "loss": 2.7071, "step": 238010 }, { "epoch": 0.0115712, "grad_norm": 0.9119644165039062, "learning_rate": 6.647288190405002e-06, "loss": 2.9442, "step": 238020 }, { "epoch": 0.0115968, "grad_norm": 0.8902599811553955, "learning_rate": 6.646529530976644e-06, "loss": 3.0576, "step": 238030 }, { "epoch": 0.0116224, "grad_norm": 0.9024169445037842, "learning_rate": 6.645770893294654e-06, "loss": 2.8343, "step": 238040 }, { "epoch": 0.011648, "grad_norm": 0.9078211188316345, "learning_rate": 6.645012277363947e-06, "loss": 2.6799, "step": 238050 }, { "epoch": 0.0116736, "grad_norm": 0.7518585324287415, "learning_rate": 6.644253683189443e-06, "loss": 2.9284, "step": 238060 }, { "epoch": 0.0116992, "grad_norm": 1.0980149507522583, "learning_rate": 6.643495110776061e-06, "loss": 2.8281, "step": 238070 }, { "epoch": 0.0117248, "grad_norm": 0.9008972644805908, "learning_rate": 6.642736560128725e-06, "loss": 2.9003, "step": 238080 }, { "epoch": 0.0117504, "grad_norm": 0.9869228005409241, "learning_rate": 6.641978031252348e-06, "loss": 3.0044, "step": 238090 }, { "epoch": 0.011776, "grad_norm": 0.7456342577934265, "learning_rate": 6.6412195241518494e-06, "loss": 2.719, "step": 238100 }, { "epoch": 0.0118016, "grad_norm": 0.7414336800575256, "learning_rate": 6.6404610388321486e-06, "loss": 2.8237, "step": 238110 }, { "epoch": 0.0118272, "grad_norm": 0.7478236556053162, "learning_rate": 6.639702575298168e-06, "loss": 2.7506, "step": 238120 }, { "epoch": 0.0118528, "grad_norm": 0.7712421417236328, "learning_rate": 6.638944133554824e-06, "loss": 2.9303, "step": 238130 }, { "epoch": 0.0118784, "grad_norm": 0.7990104556083679, "learning_rate": 6.6381857136070325e-06, "loss": 2.7618, "step": 238140 }, { "epoch": 0.011904, "grad_norm": 0.7860149145126343, "learning_rate": 6.637427315459712e-06, "loss": 2.8768, "step": 238150 }, { "epoch": 0.0119296, "grad_norm": 1.146856665611267, "learning_rate": 6.636668939117783e-06, "loss": 3.0742, "step": 238160 }, { "epoch": 0.0119552, "grad_norm": 0.7737164497375488, "learning_rate": 6.635910584586165e-06, "loss": 2.9814, "step": 238170 }, { "epoch": 0.0119808, "grad_norm": 0.8114365935325623, "learning_rate": 6.635152251869768e-06, "loss": 2.9408, "step": 238180 }, { "epoch": 0.0120064, "grad_norm": 0.8102784156799316, "learning_rate": 6.634393940973516e-06, "loss": 2.6286, "step": 238190 }, { "epoch": 0.012032, "grad_norm": 0.7599424123764038, "learning_rate": 6.633635651902327e-06, "loss": 2.6344, "step": 238200 }, { "epoch": 0.0120576, "grad_norm": 1.132742166519165, "learning_rate": 6.632877384661114e-06, "loss": 2.8825, "step": 238210 }, { "epoch": 0.0120832, "grad_norm": 0.7874583005905151, "learning_rate": 6.632119139254796e-06, "loss": 2.7037, "step": 238220 }, { "epoch": 0.0121088, "grad_norm": 0.758825957775116, "learning_rate": 6.631360915688291e-06, "loss": 2.8529, "step": 238230 }, { "epoch": 0.0121344, "grad_norm": 1.0942925214767456, "learning_rate": 6.630602713966516e-06, "loss": 2.9231, "step": 238240 }, { "epoch": 0.01216, "grad_norm": 0.7503662109375, "learning_rate": 6.629844534094385e-06, "loss": 2.8902, "step": 238250 }, { "epoch": 0.0121856, "grad_norm": 0.7775386571884155, "learning_rate": 6.629086376076818e-06, "loss": 2.8067, "step": 238260 }, { "epoch": 0.0122112, "grad_norm": 0.9318820238113403, "learning_rate": 6.6283282399187285e-06, "loss": 2.8737, "step": 238270 }, { "epoch": 0.0122368, "grad_norm": 0.8363392949104309, "learning_rate": 6.627570125625037e-06, "loss": 2.9612, "step": 238280 }, { "epoch": 0.0122624, "grad_norm": 1.2266349792480469, "learning_rate": 6.62681203320065e-06, "loss": 2.8743, "step": 238290 }, { "epoch": 0.012288, "grad_norm": 0.784140944480896, "learning_rate": 6.626053962650498e-06, "loss": 2.8692, "step": 238300 }, { "epoch": 0.0123136, "grad_norm": 0.7958219051361084, "learning_rate": 6.625295913979486e-06, "loss": 2.8523, "step": 238310 }, { "epoch": 0.0123392, "grad_norm": 0.8362147212028503, "learning_rate": 6.6245378871925346e-06, "loss": 2.8905, "step": 238320 }, { "epoch": 0.0123648, "grad_norm": 1.7096906900405884, "learning_rate": 6.623779882294559e-06, "loss": 2.9842, "step": 238330 }, { "epoch": 0.0123904, "grad_norm": 1.3579444885253906, "learning_rate": 6.623021899290471e-06, "loss": 2.8645, "step": 238340 }, { "epoch": 0.012416, "grad_norm": 0.7154728174209595, "learning_rate": 6.62226393818519e-06, "loss": 2.9994, "step": 238350 }, { "epoch": 0.0124416, "grad_norm": 0.838693380355835, "learning_rate": 6.621505998983628e-06, "loss": 2.9982, "step": 238360 }, { "epoch": 0.0124672, "grad_norm": 0.7978742718696594, "learning_rate": 6.620748081690705e-06, "loss": 2.8686, "step": 238370 }, { "epoch": 0.0124928, "grad_norm": 0.7654860615730286, "learning_rate": 6.619990186311329e-06, "loss": 2.8196, "step": 238380 }, { "epoch": 0.0125184, "grad_norm": 0.9871559143066406, "learning_rate": 6.61923231285042e-06, "loss": 3.0425, "step": 238390 }, { "epoch": 0.012544, "grad_norm": 1.0783642530441284, "learning_rate": 6.618474461312891e-06, "loss": 3.1121, "step": 238400 }, { "epoch": 0.0125696, "grad_norm": 0.798799991607666, "learning_rate": 6.617716631703657e-06, "loss": 2.9729, "step": 238410 }, { "epoch": 0.0125952, "grad_norm": 0.752898097038269, "learning_rate": 6.61695882402763e-06, "loss": 2.9717, "step": 238420 }, { "epoch": 0.0126208, "grad_norm": 0.8723381161689758, "learning_rate": 6.616201038289727e-06, "loss": 2.8752, "step": 238430 }, { "epoch": 0.0126464, "grad_norm": 0.7940933704376221, "learning_rate": 6.6154432744948615e-06, "loss": 2.8261, "step": 238440 }, { "epoch": 0.012672, "grad_norm": 0.8729709386825562, "learning_rate": 6.614685532647946e-06, "loss": 2.8269, "step": 238450 }, { "epoch": 0.0126976, "grad_norm": 0.7146526575088501, "learning_rate": 6.613927812753893e-06, "loss": 2.6447, "step": 238460 }, { "epoch": 0.0127232, "grad_norm": 0.8183557987213135, "learning_rate": 6.61317011481762e-06, "loss": 2.9755, "step": 238470 }, { "epoch": 0.0127488, "grad_norm": 0.8235167264938354, "learning_rate": 6.612412438844038e-06, "loss": 2.7577, "step": 238480 }, { "epoch": 0.0127744, "grad_norm": 0.7318621873855591, "learning_rate": 6.611654784838062e-06, "loss": 2.9505, "step": 238490 }, { "epoch": 0.0128, "grad_norm": 1.0687240362167358, "learning_rate": 6.610897152804606e-06, "loss": 2.8593, "step": 238500 }, { "epoch": 0.0128256, "grad_norm": 0.9441061615943909, "learning_rate": 6.610139542748579e-06, "loss": 2.9176, "step": 238510 }, { "epoch": 0.0128512, "grad_norm": 0.7636157274246216, "learning_rate": 6.609381954674896e-06, "loss": 2.883, "step": 238520 }, { "epoch": 0.0128768, "grad_norm": 0.8062474131584167, "learning_rate": 6.608624388588474e-06, "loss": 2.9304, "step": 238530 }, { "epoch": 0.0129024, "grad_norm": 0.8289771676063538, "learning_rate": 6.607866844494217e-06, "loss": 2.9643, "step": 238540 }, { "epoch": 0.012928, "grad_norm": 0.9705694317817688, "learning_rate": 6.607109322397043e-06, "loss": 2.746, "step": 238550 }, { "epoch": 0.0129536, "grad_norm": 1.3501046895980835, "learning_rate": 6.606351822301863e-06, "loss": 2.8245, "step": 238560 }, { "epoch": 0.0129792, "grad_norm": 1.0210988521575928, "learning_rate": 6.605594344213592e-06, "loss": 2.9479, "step": 238570 }, { "epoch": 0.0130048, "grad_norm": 0.8481458425521851, "learning_rate": 6.604836888137136e-06, "loss": 2.9966, "step": 238580 }, { "epoch": 0.0130304, "grad_norm": 0.7993905544281006, "learning_rate": 6.604079454077412e-06, "loss": 2.7651, "step": 238590 }, { "epoch": 0.013056, "grad_norm": 1.1718552112579346, "learning_rate": 6.603322042039333e-06, "loss": 2.9053, "step": 238600 }, { "epoch": 0.0130816, "grad_norm": 0.7836899161338806, "learning_rate": 6.602564652027803e-06, "loss": 3.2056, "step": 238610 }, { "epoch": 0.0131072, "grad_norm": 1.064947247505188, "learning_rate": 6.601807284047741e-06, "loss": 3.0394, "step": 238620 }, { "epoch": 0.0131328, "grad_norm": 0.8212024569511414, "learning_rate": 6.6010499381040534e-06, "loss": 2.8576, "step": 238630 }, { "epoch": 0.0131584, "grad_norm": 1.0949745178222656, "learning_rate": 6.600292614201653e-06, "loss": 3.1545, "step": 238640 }, { "epoch": 0.013184, "grad_norm": 0.8497124910354614, "learning_rate": 6.599535312345453e-06, "loss": 2.7928, "step": 238650 }, { "epoch": 0.0132096, "grad_norm": 1.3404638767242432, "learning_rate": 6.598778032540366e-06, "loss": 2.9683, "step": 238660 }, { "epoch": 0.0132352, "grad_norm": 0.7785758972167969, "learning_rate": 6.598020774791296e-06, "loss": 3.0646, "step": 238670 }, { "epoch": 0.0132608, "grad_norm": 0.7149701118469238, "learning_rate": 6.597263539103158e-06, "loss": 2.7364, "step": 238680 }, { "epoch": 0.0132864, "grad_norm": 0.7821182012557983, "learning_rate": 6.596506325480862e-06, "loss": 2.6681, "step": 238690 }, { "epoch": 0.013312, "grad_norm": 0.9053922891616821, "learning_rate": 6.595749133929319e-06, "loss": 2.9541, "step": 238700 }, { "epoch": 0.0133376, "grad_norm": 0.7891550660133362, "learning_rate": 6.594991964453437e-06, "loss": 2.8722, "step": 238710 }, { "epoch": 0.0133632, "grad_norm": 0.8323475122451782, "learning_rate": 6.594234817058125e-06, "loss": 2.8963, "step": 238720 }, { "epoch": 0.0133888, "grad_norm": 1.0708028078079224, "learning_rate": 6.5934776917483e-06, "loss": 2.9023, "step": 238730 }, { "epoch": 0.0134144, "grad_norm": 0.9015355706214905, "learning_rate": 6.592720588528863e-06, "loss": 2.8766, "step": 238740 }, { "epoch": 0.01344, "grad_norm": 0.8273393511772156, "learning_rate": 6.591963507404728e-06, "loss": 2.8538, "step": 238750 }, { "epoch": 0.0134656, "grad_norm": 0.7876372933387756, "learning_rate": 6.591206448380803e-06, "loss": 2.6739, "step": 238760 }, { "epoch": 0.0134912, "grad_norm": 0.7297579050064087, "learning_rate": 6.590449411462001e-06, "loss": 2.9535, "step": 238770 }, { "epoch": 0.0135168, "grad_norm": 1.3273985385894775, "learning_rate": 6.5896923966532264e-06, "loss": 2.9852, "step": 238780 }, { "epoch": 0.0135424, "grad_norm": 0.9105395674705505, "learning_rate": 6.588935403959391e-06, "loss": 2.8902, "step": 238790 }, { "epoch": 0.013568, "grad_norm": 0.7975258827209473, "learning_rate": 6.588178433385404e-06, "loss": 3.0054, "step": 238800 }, { "epoch": 0.0135936, "grad_norm": 0.8538989424705505, "learning_rate": 6.587421484936167e-06, "loss": 2.7985, "step": 238810 }, { "epoch": 0.0136192, "grad_norm": 1.116540789604187, "learning_rate": 6.586664558616601e-06, "loss": 3.0253, "step": 238820 }, { "epoch": 0.0136448, "grad_norm": 1.0356072187423706, "learning_rate": 6.585907654431607e-06, "loss": 2.7935, "step": 238830 }, { "epoch": 0.0136704, "grad_norm": 0.9359905123710632, "learning_rate": 6.585150772386093e-06, "loss": 2.8863, "step": 238840 }, { "epoch": 0.013696, "grad_norm": 0.7984997630119324, "learning_rate": 6.5843939124849695e-06, "loss": 2.7459, "step": 238850 }, { "epoch": 0.0137216, "grad_norm": 0.7860941290855408, "learning_rate": 6.5836370747331455e-06, "loss": 2.9196, "step": 238860 }, { "epoch": 0.0137472, "grad_norm": 0.8543375134468079, "learning_rate": 6.582880259135526e-06, "loss": 2.9029, "step": 238870 }, { "epoch": 0.0137728, "grad_norm": 0.8922789692878723, "learning_rate": 6.582123465697019e-06, "loss": 2.9891, "step": 238880 }, { "epoch": 0.0137984, "grad_norm": 0.7908408045768738, "learning_rate": 6.581366694422533e-06, "loss": 3.0082, "step": 238890 }, { "epoch": 0.013824, "grad_norm": 0.7171127796173096, "learning_rate": 6.580609945316978e-06, "loss": 2.844, "step": 238900 }, { "epoch": 0.0138496, "grad_norm": 1.115809679031372, "learning_rate": 6.579853218385255e-06, "loss": 2.8448, "step": 238910 }, { "epoch": 0.0138752, "grad_norm": 1.2257750034332275, "learning_rate": 6.579096513632277e-06, "loss": 2.895, "step": 238920 }, { "epoch": 0.0139008, "grad_norm": 0.8062114715576172, "learning_rate": 6.57833983106295e-06, "loss": 2.9924, "step": 238930 }, { "epoch": 0.0139264, "grad_norm": 0.8141754865646362, "learning_rate": 6.577583170682178e-06, "loss": 2.9215, "step": 238940 }, { "epoch": 0.013952, "grad_norm": 0.8236470818519592, "learning_rate": 6.576826532494868e-06, "loss": 2.6055, "step": 238950 }, { "epoch": 0.0139776, "grad_norm": 0.7887414693832397, "learning_rate": 6.57606991650593e-06, "loss": 2.7754, "step": 238960 }, { "epoch": 0.0140032, "grad_norm": 0.8433213233947754, "learning_rate": 6.575313322720271e-06, "loss": 2.8763, "step": 238970 }, { "epoch": 0.0140288, "grad_norm": 0.9017269611358643, "learning_rate": 6.574556751142789e-06, "loss": 2.9113, "step": 238980 }, { "epoch": 0.0140544, "grad_norm": 0.8836965560913086, "learning_rate": 6.573800201778399e-06, "loss": 2.8288, "step": 238990 }, { "epoch": 0.01408, "grad_norm": 0.7527217864990234, "learning_rate": 6.573043674632006e-06, "loss": 3.0004, "step": 239000 }, { "epoch": 0.0141056, "grad_norm": 0.7560350298881531, "learning_rate": 6.572287169708512e-06, "loss": 2.7943, "step": 239010 }, { "epoch": 0.0141312, "grad_norm": 0.8684869408607483, "learning_rate": 6.571530687012828e-06, "loss": 2.9831, "step": 239020 }, { "epoch": 0.0141568, "grad_norm": 0.912182092666626, "learning_rate": 6.570774226549853e-06, "loss": 2.5981, "step": 239030 }, { "epoch": 0.0141824, "grad_norm": 0.9476915597915649, "learning_rate": 6.570017788324496e-06, "loss": 2.6768, "step": 239040 }, { "epoch": 0.014208, "grad_norm": 0.8235728740692139, "learning_rate": 6.569261372341662e-06, "loss": 2.9374, "step": 239050 }, { "epoch": 0.0142336, "grad_norm": 0.9021913409233093, "learning_rate": 6.568504978606258e-06, "loss": 3.0324, "step": 239060 }, { "epoch": 0.0142592, "grad_norm": 1.0898205041885376, "learning_rate": 6.567748607123187e-06, "loss": 2.8698, "step": 239070 }, { "epoch": 0.0142848, "grad_norm": 0.8052015900611877, "learning_rate": 6.566992257897352e-06, "loss": 3.0871, "step": 239080 }, { "epoch": 0.0143104, "grad_norm": 1.3827526569366455, "learning_rate": 6.566235930933661e-06, "loss": 3.5588, "step": 239090 }, { "epoch": 0.014336, "grad_norm": 0.9012904763221741, "learning_rate": 6.565479626237019e-06, "loss": 2.9328, "step": 239100 }, { "epoch": 0.0143616, "grad_norm": 0.7293872833251953, "learning_rate": 6.564723343812327e-06, "loss": 2.7744, "step": 239110 }, { "epoch": 0.0143872, "grad_norm": 0.771384596824646, "learning_rate": 6.563967083664491e-06, "loss": 3.1706, "step": 239120 }, { "epoch": 0.0144128, "grad_norm": 0.8496600389480591, "learning_rate": 6.563210845798418e-06, "loss": 3.0105, "step": 239130 }, { "epoch": 0.0144384, "grad_norm": 0.8300013542175293, "learning_rate": 6.5624546302190065e-06, "loss": 2.8478, "step": 239140 }, { "epoch": 0.014464, "grad_norm": 0.8912471532821655, "learning_rate": 6.561698436931163e-06, "loss": 2.8082, "step": 239150 }, { "epoch": 0.0144896, "grad_norm": 0.7085486054420471, "learning_rate": 6.5609422659397895e-06, "loss": 2.8875, "step": 239160 }, { "epoch": 0.0145152, "grad_norm": 0.8469105362892151, "learning_rate": 6.560186117249794e-06, "loss": 2.9444, "step": 239170 }, { "epoch": 0.0145408, "grad_norm": 0.6827659010887146, "learning_rate": 6.559429990866075e-06, "loss": 2.7561, "step": 239180 }, { "epoch": 0.0145664, "grad_norm": 0.950782299041748, "learning_rate": 6.558673886793543e-06, "loss": 2.8267, "step": 239190 }, { "epoch": 0.014592, "grad_norm": 0.7218373417854309, "learning_rate": 6.557917805037093e-06, "loss": 2.7825, "step": 239200 }, { "epoch": 0.0146176, "grad_norm": 0.8336629271507263, "learning_rate": 6.5571617456016315e-06, "loss": 2.9131, "step": 239210 }, { "epoch": 0.0146432, "grad_norm": 0.8115389347076416, "learning_rate": 6.556405708492063e-06, "loss": 2.9315, "step": 239220 }, { "epoch": 0.0146688, "grad_norm": 0.8803179264068604, "learning_rate": 6.5556496937132866e-06, "loss": 2.8759, "step": 239230 }, { "epoch": 0.0146944, "grad_norm": 1.0243041515350342, "learning_rate": 6.5548937012702065e-06, "loss": 3.0296, "step": 239240 }, { "epoch": 0.01472, "grad_norm": 0.8117198348045349, "learning_rate": 6.554137731167725e-06, "loss": 2.8658, "step": 239250 }, { "epoch": 0.0147456, "grad_norm": 0.8133671879768372, "learning_rate": 6.553381783410746e-06, "loss": 2.8234, "step": 239260 }, { "epoch": 0.0147712, "grad_norm": 0.9023573994636536, "learning_rate": 6.5526258580041676e-06, "loss": 2.7612, "step": 239270 }, { "epoch": 0.0147968, "grad_norm": 0.8519150018692017, "learning_rate": 6.551869954952896e-06, "loss": 2.8762, "step": 239280 }, { "epoch": 0.0148224, "grad_norm": 0.872333824634552, "learning_rate": 6.551114074261833e-06, "loss": 2.9635, "step": 239290 }, { "epoch": 0.014848, "grad_norm": 0.7212115526199341, "learning_rate": 6.5503582159358746e-06, "loss": 2.8714, "step": 239300 }, { "epoch": 0.0148736, "grad_norm": 1.0638856887817383, "learning_rate": 6.549602379979929e-06, "loss": 2.9184, "step": 239310 }, { "epoch": 0.0148992, "grad_norm": 0.765315055847168, "learning_rate": 6.5488465663988945e-06, "loss": 2.9229, "step": 239320 }, { "epoch": 0.0149248, "grad_norm": 0.6929014921188354, "learning_rate": 6.548090775197669e-06, "loss": 2.906, "step": 239330 }, { "epoch": 0.0149504, "grad_norm": 0.7568346261978149, "learning_rate": 6.5473350063811616e-06, "loss": 2.7164, "step": 239340 }, { "epoch": 0.014976, "grad_norm": 0.7849207520484924, "learning_rate": 6.5465792599542686e-06, "loss": 2.889, "step": 239350 }, { "epoch": 0.0150016, "grad_norm": 0.7543129324913025, "learning_rate": 6.545823535921892e-06, "loss": 2.7033, "step": 239360 }, { "epoch": 0.0150272, "grad_norm": 0.928558886051178, "learning_rate": 6.54506783428893e-06, "loss": 2.8111, "step": 239370 }, { "epoch": 0.0150528, "grad_norm": 0.8506290912628174, "learning_rate": 6.544312155060286e-06, "loss": 2.9644, "step": 239380 }, { "epoch": 0.0150784, "grad_norm": 0.8582215905189514, "learning_rate": 6.54355649824086e-06, "loss": 2.7768, "step": 239390 }, { "epoch": 0.015104, "grad_norm": 0.8235539793968201, "learning_rate": 6.54280086383555e-06, "loss": 2.8262, "step": 239400 }, { "epoch": 0.0151296, "grad_norm": 0.7712265253067017, "learning_rate": 6.542045251849258e-06, "loss": 2.8024, "step": 239410 }, { "epoch": 0.0151552, "grad_norm": 1.2149289846420288, "learning_rate": 6.5412896622868845e-06, "loss": 2.7598, "step": 239420 }, { "epoch": 0.0151808, "grad_norm": 0.8343316316604614, "learning_rate": 6.540534095153328e-06, "loss": 2.5861, "step": 239430 }, { "epoch": 0.0152064, "grad_norm": 0.7991174459457397, "learning_rate": 6.539778550453489e-06, "loss": 2.8865, "step": 239440 }, { "epoch": 0.015232, "grad_norm": 0.8699343800544739, "learning_rate": 6.539023028192264e-06, "loss": 2.847, "step": 239450 }, { "epoch": 0.0152576, "grad_norm": 0.8260142207145691, "learning_rate": 6.538267528374558e-06, "loss": 2.8766, "step": 239460 }, { "epoch": 0.0152832, "grad_norm": 0.892512857913971, "learning_rate": 6.537512051005265e-06, "loss": 2.9168, "step": 239470 }, { "epoch": 0.0153088, "grad_norm": 0.860701322555542, "learning_rate": 6.536756596089286e-06, "loss": 2.8589, "step": 239480 }, { "epoch": 0.0153344, "grad_norm": 0.7990941405296326, "learning_rate": 6.536001163631522e-06, "loss": 3.3112, "step": 239490 }, { "epoch": 0.01536, "grad_norm": 0.9486439824104309, "learning_rate": 6.535245753636867e-06, "loss": 2.927, "step": 239500 }, { "epoch": 0.0153856, "grad_norm": 0.7342494130134583, "learning_rate": 6.5344903661102226e-06, "loss": 2.5926, "step": 239510 }, { "epoch": 0.0154112, "grad_norm": 0.7905156016349792, "learning_rate": 6.533735001056487e-06, "loss": 2.7531, "step": 239520 }, { "epoch": 0.0154368, "grad_norm": 0.8497311472892761, "learning_rate": 6.53297965848056e-06, "loss": 2.8639, "step": 239530 }, { "epoch": 0.0154624, "grad_norm": 0.8121259808540344, "learning_rate": 6.532224338387339e-06, "loss": 2.7725, "step": 239540 }, { "epoch": 0.015488, "grad_norm": 0.7711806297302246, "learning_rate": 6.531469040781723e-06, "loss": 2.7757, "step": 239550 }, { "epoch": 0.0155136, "grad_norm": 0.8013837933540344, "learning_rate": 6.530713765668606e-06, "loss": 2.9292, "step": 239560 }, { "epoch": 0.0155392, "grad_norm": 1.125182032585144, "learning_rate": 6.529958513052888e-06, "loss": 2.8215, "step": 239570 }, { "epoch": 0.0155648, "grad_norm": 0.818798840045929, "learning_rate": 6.529203282939467e-06, "loss": 2.8779, "step": 239580 }, { "epoch": 0.0155904, "grad_norm": 0.7911282181739807, "learning_rate": 6.528448075333242e-06, "loss": 2.4939, "step": 239590 }, { "epoch": 0.015616, "grad_norm": 0.9051757454872131, "learning_rate": 6.527692890239108e-06, "loss": 2.8032, "step": 239600 }, { "epoch": 0.0156416, "grad_norm": 0.7458904981613159, "learning_rate": 6.526937727661962e-06, "loss": 2.8166, "step": 239610 }, { "epoch": 0.0156672, "grad_norm": 0.8111796379089355, "learning_rate": 6.526182587606704e-06, "loss": 2.8866, "step": 239620 }, { "epoch": 0.0156928, "grad_norm": 0.7861413359642029, "learning_rate": 6.525427470078226e-06, "loss": 2.6599, "step": 239630 }, { "epoch": 0.0157184, "grad_norm": 0.8713036775588989, "learning_rate": 6.524672375081428e-06, "loss": 2.881, "step": 239640 }, { "epoch": 0.015744, "grad_norm": 0.8283236622810364, "learning_rate": 6.523917302621205e-06, "loss": 2.8278, "step": 239650 }, { "epoch": 0.0157696, "grad_norm": 0.8806777000427246, "learning_rate": 6.5231622527024575e-06, "loss": 2.9597, "step": 239660 }, { "epoch": 0.0157952, "grad_norm": 0.7895262837409973, "learning_rate": 6.522407225330076e-06, "loss": 2.8414, "step": 239670 }, { "epoch": 0.0158208, "grad_norm": 1.4896717071533203, "learning_rate": 6.521652220508958e-06, "loss": 2.9159, "step": 239680 }, { "epoch": 0.0158464, "grad_norm": 0.8189912438392639, "learning_rate": 6.520897238244004e-06, "loss": 2.7497, "step": 239690 }, { "epoch": 0.015872, "grad_norm": 0.778428852558136, "learning_rate": 6.5201422785401045e-06, "loss": 3.0354, "step": 239700 }, { "epoch": 0.0158976, "grad_norm": 0.9060018062591553, "learning_rate": 6.5193873414021614e-06, "loss": 2.8474, "step": 239710 }, { "epoch": 0.0159232, "grad_norm": 0.7221596837043762, "learning_rate": 6.518632426835063e-06, "loss": 2.64, "step": 239720 }, { "epoch": 0.0159488, "grad_norm": 0.8663894534111023, "learning_rate": 6.517877534843708e-06, "loss": 2.9836, "step": 239730 }, { "epoch": 0.0159744, "grad_norm": 0.8759427070617676, "learning_rate": 6.517122665432993e-06, "loss": 2.8836, "step": 239740 }, { "epoch": 0.016, "grad_norm": 0.8109322786331177, "learning_rate": 6.516367818607814e-06, "loss": 3.0327, "step": 239750 }, { "epoch": 0.0160256, "grad_norm": 1.2126502990722656, "learning_rate": 6.5156129943730615e-06, "loss": 2.9456, "step": 239760 }, { "epoch": 0.0160512, "grad_norm": 0.6984370946884155, "learning_rate": 6.514858192733632e-06, "loss": 2.8647, "step": 239770 }, { "epoch": 0.0160768, "grad_norm": 0.7250746488571167, "learning_rate": 6.514103413694422e-06, "loss": 2.7585, "step": 239780 }, { "epoch": 0.0161024, "grad_norm": 1.5454822778701782, "learning_rate": 6.513348657260325e-06, "loss": 2.6891, "step": 239790 }, { "epoch": 0.016128, "grad_norm": 0.9182811975479126, "learning_rate": 6.512593923436235e-06, "loss": 2.7774, "step": 239800 }, { "epoch": 0.0161536, "grad_norm": 0.7562435269355774, "learning_rate": 6.511839212227046e-06, "loss": 2.9001, "step": 239810 }, { "epoch": 0.0161792, "grad_norm": 0.8226099610328674, "learning_rate": 6.511084523637654e-06, "loss": 2.9159, "step": 239820 }, { "epoch": 0.0162048, "grad_norm": 0.7707860469818115, "learning_rate": 6.51032985767295e-06, "loss": 2.9346, "step": 239830 }, { "epoch": 0.0162304, "grad_norm": 0.7846006751060486, "learning_rate": 6.509575214337831e-06, "loss": 2.9415, "step": 239840 }, { "epoch": 0.016256, "grad_norm": 0.7463361620903015, "learning_rate": 6.5088205936371855e-06, "loss": 2.835, "step": 239850 }, { "epoch": 0.0162816, "grad_norm": 0.8179194927215576, "learning_rate": 6.508065995575913e-06, "loss": 2.9122, "step": 239860 }, { "epoch": 0.0163072, "grad_norm": 0.8205379843711853, "learning_rate": 6.507311420158904e-06, "loss": 2.8434, "step": 239870 }, { "epoch": 0.0163328, "grad_norm": 0.8183853030204773, "learning_rate": 6.506556867391054e-06, "loss": 2.7166, "step": 239880 }, { "epoch": 0.0163584, "grad_norm": 0.8132006525993347, "learning_rate": 6.505802337277254e-06, "loss": 2.8115, "step": 239890 }, { "epoch": 0.016384, "grad_norm": 0.8816971182823181, "learning_rate": 6.5050478298223955e-06, "loss": 3.1258, "step": 239900 }, { "epoch": 0.0164096, "grad_norm": 0.8151007890701294, "learning_rate": 6.504293345031376e-06, "loss": 2.8651, "step": 239910 }, { "epoch": 0.0164352, "grad_norm": 0.7745792269706726, "learning_rate": 6.503538882909081e-06, "loss": 2.802, "step": 239920 }, { "epoch": 0.0164608, "grad_norm": 0.7458685636520386, "learning_rate": 6.50278444346041e-06, "loss": 2.5757, "step": 239930 }, { "epoch": 0.0164864, "grad_norm": 1.0928518772125244, "learning_rate": 6.502030026690251e-06, "loss": 2.7221, "step": 239940 }, { "epoch": 0.016512, "grad_norm": 0.8881527185440063, "learning_rate": 6.501275632603501e-06, "loss": 2.736, "step": 239950 }, { "epoch": 0.0165376, "grad_norm": 0.7991508841514587, "learning_rate": 6.500521261205044e-06, "loss": 3.161, "step": 239960 }, { "epoch": 0.0165632, "grad_norm": 0.7550405859947205, "learning_rate": 6.4997669124997785e-06, "loss": 2.9818, "step": 239970 }, { "epoch": 0.0165888, "grad_norm": 0.8741347789764404, "learning_rate": 6.499012586492594e-06, "loss": 2.8153, "step": 239980 }, { "epoch": 0.0166144, "grad_norm": 0.8149539232254028, "learning_rate": 6.498258283188385e-06, "loss": 2.9275, "step": 239990 }, { "epoch": 0.01664, "grad_norm": 0.774946391582489, "learning_rate": 6.497504002592037e-06, "loss": 2.8655, "step": 240000 }, { "epoch": 0.0166656, "grad_norm": 0.829675018787384, "learning_rate": 6.496749744708446e-06, "loss": 2.9903, "step": 240010 }, { "epoch": 0.0166912, "grad_norm": 1.0719505548477173, "learning_rate": 6.495995509542503e-06, "loss": 2.851, "step": 240020 }, { "epoch": 0.0167168, "grad_norm": 0.8796150088310242, "learning_rate": 6.495241297099092e-06, "loss": 2.7703, "step": 240030 }, { "epoch": 0.0167424, "grad_norm": 0.7915138006210327, "learning_rate": 6.4944871073831166e-06, "loss": 3.1074, "step": 240040 }, { "epoch": 0.016768, "grad_norm": 0.851318359375, "learning_rate": 6.493732940399457e-06, "loss": 2.9605, "step": 240050 }, { "epoch": 0.0167936, "grad_norm": 0.7592301368713379, "learning_rate": 6.492978796153009e-06, "loss": 2.9284, "step": 240060 }, { "epoch": 0.0168192, "grad_norm": 0.8308855295181274, "learning_rate": 6.492224674648659e-06, "loss": 3.0052, "step": 240070 }, { "epoch": 0.0168448, "grad_norm": 0.7814799547195435, "learning_rate": 6.491470575891304e-06, "loss": 2.8548, "step": 240080 }, { "epoch": 0.0168704, "grad_norm": 1.2615489959716797, "learning_rate": 6.490716499885827e-06, "loss": 3.008, "step": 240090 }, { "epoch": 0.016896, "grad_norm": 0.8204050660133362, "learning_rate": 6.489962446637121e-06, "loss": 2.6876, "step": 240100 }, { "epoch": 0.0169216, "grad_norm": 0.8392716646194458, "learning_rate": 6.489208416150079e-06, "loss": 2.8412, "step": 240110 }, { "epoch": 0.0169472, "grad_norm": 0.7886818647384644, "learning_rate": 6.488454408429584e-06, "loss": 2.8842, "step": 240120 }, { "epoch": 0.0169728, "grad_norm": 0.8005940914154053, "learning_rate": 6.4877004234805295e-06, "loss": 3.3287, "step": 240130 }, { "epoch": 0.0169984, "grad_norm": 1.295143723487854, "learning_rate": 6.486946461307804e-06, "loss": 2.9021, "step": 240140 }, { "epoch": 0.017024, "grad_norm": 0.694399356842041, "learning_rate": 6.486192521916299e-06, "loss": 2.8497, "step": 240150 }, { "epoch": 0.0170496, "grad_norm": 0.8037976026535034, "learning_rate": 6.4854386053109e-06, "loss": 2.8426, "step": 240160 }, { "epoch": 0.0170752, "grad_norm": 0.9794929027557373, "learning_rate": 6.484684711496496e-06, "loss": 2.8512, "step": 240170 }, { "epoch": 0.0171008, "grad_norm": 0.8648514747619629, "learning_rate": 6.483930840477981e-06, "loss": 2.7838, "step": 240180 }, { "epoch": 0.0171264, "grad_norm": 1.1013333797454834, "learning_rate": 6.483176992260239e-06, "loss": 2.746, "step": 240190 }, { "epoch": 0.017152, "grad_norm": 0.8532304763793945, "learning_rate": 6.482423166848153e-06, "loss": 2.738, "step": 240200 }, { "epoch": 0.0171776, "grad_norm": 0.7539172172546387, "learning_rate": 6.481669364246626e-06, "loss": 2.7978, "step": 240210 }, { "epoch": 0.0172032, "grad_norm": 0.7990268468856812, "learning_rate": 6.480915584460534e-06, "loss": 2.8132, "step": 240220 }, { "epoch": 0.0172288, "grad_norm": 1.9830130338668823, "learning_rate": 6.4801618274947705e-06, "loss": 3.2289, "step": 240230 }, { "epoch": 0.0172544, "grad_norm": 0.7862645983695984, "learning_rate": 6.479408093354225e-06, "loss": 2.8712, "step": 240240 }, { "epoch": 0.01728, "grad_norm": 0.9179694056510925, "learning_rate": 6.478654382043778e-06, "loss": 2.8632, "step": 240250 }, { "epoch": 0.0173056, "grad_norm": 0.8140431046485901, "learning_rate": 6.477900693568323e-06, "loss": 2.9622, "step": 240260 }, { "epoch": 0.0173312, "grad_norm": 0.9132986664772034, "learning_rate": 6.4771470279327466e-06, "loss": 2.7721, "step": 240270 }, { "epoch": 0.0173568, "grad_norm": 0.8120219707489014, "learning_rate": 6.476393385141936e-06, "loss": 2.9171, "step": 240280 }, { "epoch": 0.0173824, "grad_norm": 0.9697892665863037, "learning_rate": 6.475639765200776e-06, "loss": 2.8776, "step": 240290 }, { "epoch": 0.017408, "grad_norm": 0.7991726398468018, "learning_rate": 6.474886168114156e-06, "loss": 2.969, "step": 240300 }, { "epoch": 0.0174336, "grad_norm": 0.7904735207557678, "learning_rate": 6.474132593886965e-06, "loss": 3.0433, "step": 240310 }, { "epoch": 0.0174592, "grad_norm": 0.7754824161529541, "learning_rate": 6.4733790425240836e-06, "loss": 2.7921, "step": 240320 }, { "epoch": 0.0174848, "grad_norm": 0.8508201837539673, "learning_rate": 6.472625514030403e-06, "loss": 2.9247, "step": 240330 }, { "epoch": 0.0175104, "grad_norm": 0.8009639382362366, "learning_rate": 6.4718720084108075e-06, "loss": 2.8068, "step": 240340 }, { "epoch": 0.017536, "grad_norm": 0.85893315076828, "learning_rate": 6.471118525670188e-06, "loss": 2.9384, "step": 240350 }, { "epoch": 0.0175616, "grad_norm": 0.807310938835144, "learning_rate": 6.470365065813423e-06, "loss": 2.8895, "step": 240360 }, { "epoch": 0.0175872, "grad_norm": 0.8752646446228027, "learning_rate": 6.469611628845403e-06, "loss": 2.77, "step": 240370 }, { "epoch": 0.0176128, "grad_norm": 0.8907200694084167, "learning_rate": 6.468858214771012e-06, "loss": 2.8817, "step": 240380 }, { "epoch": 0.0176384, "grad_norm": 0.7784531712532043, "learning_rate": 6.468104823595138e-06, "loss": 2.879, "step": 240390 }, { "epoch": 0.017664, "grad_norm": 0.7231698632240295, "learning_rate": 6.4673514553226695e-06, "loss": 2.8751, "step": 240400 }, { "epoch": 0.0176896, "grad_norm": 0.8278190493583679, "learning_rate": 6.466598109958484e-06, "loss": 2.967, "step": 240410 }, { "epoch": 0.0177152, "grad_norm": 0.8169938921928406, "learning_rate": 6.465844787507472e-06, "loss": 2.8937, "step": 240420 }, { "epoch": 0.0177408, "grad_norm": 0.7437689304351807, "learning_rate": 6.465091487974516e-06, "loss": 2.8879, "step": 240430 }, { "epoch": 0.0177664, "grad_norm": 0.7824997901916504, "learning_rate": 6.4643382113645044e-06, "loss": 2.8898, "step": 240440 }, { "epoch": 0.017792, "grad_norm": 0.8710627555847168, "learning_rate": 6.463584957682319e-06, "loss": 2.8866, "step": 240450 }, { "epoch": 0.0178176, "grad_norm": 0.9564561247825623, "learning_rate": 6.4628317269328445e-06, "loss": 2.7306, "step": 240460 }, { "epoch": 0.0178432, "grad_norm": 0.7763599157333374, "learning_rate": 6.4620785191209665e-06, "loss": 2.9768, "step": 240470 }, { "epoch": 0.0178688, "grad_norm": 0.7522062063217163, "learning_rate": 6.461325334251571e-06, "loss": 2.9693, "step": 240480 }, { "epoch": 0.0178944, "grad_norm": 0.8745169043540955, "learning_rate": 6.460572172329538e-06, "loss": 2.8079, "step": 240490 }, { "epoch": 0.01792, "grad_norm": 0.7873097062110901, "learning_rate": 6.459819033359753e-06, "loss": 2.8806, "step": 240500 }, { "epoch": 0.0179456, "grad_norm": 2.0893301963806152, "learning_rate": 6.459065917347105e-06, "loss": 2.9628, "step": 240510 }, { "epoch": 0.0179712, "grad_norm": 0.9191226363182068, "learning_rate": 6.458312824296469e-06, "loss": 2.8673, "step": 240520 }, { "epoch": 0.0179968, "grad_norm": 0.7704469561576843, "learning_rate": 6.457559754212734e-06, "loss": 2.8733, "step": 240530 }, { "epoch": 0.0180224, "grad_norm": 0.7991830110549927, "learning_rate": 6.456806707100783e-06, "loss": 2.8371, "step": 240540 }, { "epoch": 0.018048, "grad_norm": 0.8156581521034241, "learning_rate": 6.456053682965497e-06, "loss": 2.8997, "step": 240550 }, { "epoch": 0.0180736, "grad_norm": 0.6922664642333984, "learning_rate": 6.455300681811763e-06, "loss": 2.8806, "step": 240560 }, { "epoch": 0.0180992, "grad_norm": 0.8228298425674438, "learning_rate": 6.454547703644462e-06, "loss": 2.9361, "step": 240570 }, { "epoch": 0.0181248, "grad_norm": 0.82185959815979, "learning_rate": 6.453794748468476e-06, "loss": 3.0325, "step": 240580 }, { "epoch": 0.0181504, "grad_norm": 0.8330591917037964, "learning_rate": 6.45304181628869e-06, "loss": 2.7838, "step": 240590 }, { "epoch": 0.018176, "grad_norm": 0.7588150501251221, "learning_rate": 6.452288907109986e-06, "loss": 3.005, "step": 240600 }, { "epoch": 0.0182016, "grad_norm": 0.767791748046875, "learning_rate": 6.4515360209372434e-06, "loss": 2.8426, "step": 240610 }, { "epoch": 0.0182272, "grad_norm": 0.8445079326629639, "learning_rate": 6.450783157775347e-06, "loss": 3.1902, "step": 240620 }, { "epoch": 0.0182528, "grad_norm": 0.8586742877960205, "learning_rate": 6.450030317629179e-06, "loss": 2.7687, "step": 240630 }, { "epoch": 0.0182784, "grad_norm": 0.8369491100311279, "learning_rate": 6.449277500503622e-06, "loss": 2.8622, "step": 240640 }, { "epoch": 0.018304, "grad_norm": 0.7875193953514099, "learning_rate": 6.448524706403555e-06, "loss": 2.8371, "step": 240650 }, { "epoch": 0.0183296, "grad_norm": 0.7707622051239014, "learning_rate": 6.447771935333862e-06, "loss": 2.8184, "step": 240660 }, { "epoch": 0.0183552, "grad_norm": 0.7624616026878357, "learning_rate": 6.4470191872994224e-06, "loss": 3.0417, "step": 240670 }, { "epoch": 0.0183808, "grad_norm": 0.8803941607475281, "learning_rate": 6.446266462305122e-06, "loss": 2.7715, "step": 240680 }, { "epoch": 0.0184064, "grad_norm": 0.8305469751358032, "learning_rate": 6.445513760355837e-06, "loss": 2.8939, "step": 240690 }, { "epoch": 0.018432, "grad_norm": 0.823560357093811, "learning_rate": 6.444761081456452e-06, "loss": 2.9401, "step": 240700 }, { "epoch": 0.0184576, "grad_norm": 0.8055459260940552, "learning_rate": 6.4440084256118475e-06, "loss": 2.9175, "step": 240710 }, { "epoch": 0.0184832, "grad_norm": 0.7919787764549255, "learning_rate": 6.443255792826897e-06, "loss": 2.7207, "step": 240720 }, { "epoch": 0.0185088, "grad_norm": 0.7958343029022217, "learning_rate": 6.442503183106493e-06, "loss": 2.8207, "step": 240730 }, { "epoch": 0.0185344, "grad_norm": 0.8034170866012573, "learning_rate": 6.4417505964555096e-06, "loss": 2.7612, "step": 240740 }, { "epoch": 0.01856, "grad_norm": 0.7557187676429749, "learning_rate": 6.440998032878827e-06, "loss": 2.9276, "step": 240750 }, { "epoch": 0.0185856, "grad_norm": 0.9123915433883667, "learning_rate": 6.440245492381327e-06, "loss": 2.7367, "step": 240760 }, { "epoch": 0.0186112, "grad_norm": 0.7917447090148926, "learning_rate": 6.43949297496789e-06, "loss": 3.0651, "step": 240770 }, { "epoch": 0.0186368, "grad_norm": 0.7917206287384033, "learning_rate": 6.438740480643393e-06, "loss": 2.8104, "step": 240780 }, { "epoch": 0.0186624, "grad_norm": 0.8681412935256958, "learning_rate": 6.437988009412718e-06, "loss": 2.8416, "step": 240790 }, { "epoch": 0.018688, "grad_norm": 0.7985097765922546, "learning_rate": 6.437235561280745e-06, "loss": 2.8399, "step": 240800 }, { "epoch": 0.0187136, "grad_norm": 0.8284551501274109, "learning_rate": 6.436483136252351e-06, "loss": 2.9253, "step": 240810 }, { "epoch": 0.0187392, "grad_norm": 0.8097102046012878, "learning_rate": 6.43573073433242e-06, "loss": 3.0852, "step": 240820 }, { "epoch": 0.0187648, "grad_norm": 0.7341601848602295, "learning_rate": 6.434978355525824e-06, "loss": 2.9047, "step": 240830 }, { "epoch": 0.0187904, "grad_norm": 0.7484716773033142, "learning_rate": 6.434225999837449e-06, "loss": 2.8476, "step": 240840 }, { "epoch": 0.018816, "grad_norm": 0.7962043285369873, "learning_rate": 6.4334736672721676e-06, "loss": 2.9435, "step": 240850 }, { "epoch": 0.0188416, "grad_norm": 0.8176029324531555, "learning_rate": 6.432721357834863e-06, "loss": 2.826, "step": 240860 }, { "epoch": 0.0188672, "grad_norm": 0.9178715348243713, "learning_rate": 6.431969071530413e-06, "loss": 2.83, "step": 240870 }, { "epoch": 0.0188928, "grad_norm": 0.9333153367042542, "learning_rate": 6.431216808363696e-06, "loss": 2.8938, "step": 240880 }, { "epoch": 0.0189184, "grad_norm": 0.8192073106765747, "learning_rate": 6.430464568339587e-06, "loss": 2.8697, "step": 240890 }, { "epoch": 0.018944, "grad_norm": 0.8334628939628601, "learning_rate": 6.429712351462967e-06, "loss": 2.8851, "step": 240900 }, { "epoch": 0.0189696, "grad_norm": 0.865519106388092, "learning_rate": 6.428960157738713e-06, "loss": 2.9183, "step": 240910 }, { "epoch": 0.0189952, "grad_norm": 0.7986482977867126, "learning_rate": 6.428207987171704e-06, "loss": 2.9433, "step": 240920 }, { "epoch": 0.0190208, "grad_norm": 0.7806711196899414, "learning_rate": 6.427455839766819e-06, "loss": 2.9684, "step": 240930 }, { "epoch": 0.0190464, "grad_norm": 1.2806154489517212, "learning_rate": 6.426703715528932e-06, "loss": 2.9376, "step": 240940 }, { "epoch": 0.019072, "grad_norm": 0.7485227584838867, "learning_rate": 6.425951614462922e-06, "loss": 2.8214, "step": 240950 }, { "epoch": 0.0190976, "grad_norm": 0.9631085395812988, "learning_rate": 6.425199536573665e-06, "loss": 2.9531, "step": 240960 }, { "epoch": 0.0191232, "grad_norm": 0.7232502698898315, "learning_rate": 6.424447481866041e-06, "loss": 3.0282, "step": 240970 }, { "epoch": 0.0191488, "grad_norm": 0.7912684082984924, "learning_rate": 6.423695450344923e-06, "loss": 2.8794, "step": 240980 }, { "epoch": 0.0191744, "grad_norm": 0.821621835231781, "learning_rate": 6.4229434420151905e-06, "loss": 2.9245, "step": 240990 }, { "epoch": 0.0192, "grad_norm": 0.7735702991485596, "learning_rate": 6.42219145688172e-06, "loss": 2.8541, "step": 241000 }, { "epoch": 0.0192256, "grad_norm": 0.8135802149772644, "learning_rate": 6.421439494949385e-06, "loss": 2.9766, "step": 241010 }, { "epoch": 0.0192512, "grad_norm": 0.8199110627174377, "learning_rate": 6.420687556223064e-06, "loss": 3.1809, "step": 241020 }, { "epoch": 0.0192768, "grad_norm": 0.8805087804794312, "learning_rate": 6.419935640707633e-06, "loss": 2.8522, "step": 241030 }, { "epoch": 0.0193024, "grad_norm": 0.7757567763328552, "learning_rate": 6.41918374840797e-06, "loss": 2.997, "step": 241040 }, { "epoch": 0.019328, "grad_norm": 0.7810851335525513, "learning_rate": 6.4184318793289465e-06, "loss": 2.9102, "step": 241050 }, { "epoch": 0.0193536, "grad_norm": 0.7679665088653564, "learning_rate": 6.417680033475441e-06, "loss": 3.093, "step": 241060 }, { "epoch": 0.0193792, "grad_norm": 0.7624624967575073, "learning_rate": 6.416928210852326e-06, "loss": 2.6511, "step": 241070 }, { "epoch": 0.0194048, "grad_norm": 0.8402001261711121, "learning_rate": 6.416176411464481e-06, "loss": 2.6731, "step": 241080 }, { "epoch": 0.0194304, "grad_norm": 0.6778231263160706, "learning_rate": 6.4154246353167794e-06, "loss": 2.4953, "step": 241090 }, { "epoch": 0.019456, "grad_norm": 0.7993025779724121, "learning_rate": 6.4146728824140994e-06, "loss": 2.5832, "step": 241100 }, { "epoch": 0.0194816, "grad_norm": 0.8010671138763428, "learning_rate": 6.413921152761309e-06, "loss": 2.4752, "step": 241110 }, { "epoch": 0.0195072, "grad_norm": 0.7773032784461975, "learning_rate": 6.413169446363288e-06, "loss": 2.5201, "step": 241120 }, { "epoch": 0.0195328, "grad_norm": 0.8500436544418335, "learning_rate": 6.412417763224912e-06, "loss": 2.9244, "step": 241130 }, { "epoch": 0.0195584, "grad_norm": 0.8233545422554016, "learning_rate": 6.411666103351052e-06, "loss": 2.8061, "step": 241140 }, { "epoch": 0.019584, "grad_norm": 0.8312751054763794, "learning_rate": 6.410914466746583e-06, "loss": 2.727, "step": 241150 }, { "epoch": 0.0196096, "grad_norm": 0.7632245421409607, "learning_rate": 6.410162853416379e-06, "loss": 2.5786, "step": 241160 }, { "epoch": 0.0196352, "grad_norm": 0.7827788591384888, "learning_rate": 6.4094112633653176e-06, "loss": 2.5883, "step": 241170 }, { "epoch": 0.0196608, "grad_norm": 0.9610710740089417, "learning_rate": 6.408659696598268e-06, "loss": 2.8311, "step": 241180 }, { "epoch": 0.0196864, "grad_norm": 0.8584707379341125, "learning_rate": 6.407908153120106e-06, "loss": 2.7619, "step": 241190 }, { "epoch": 0.019712, "grad_norm": 0.8021632432937622, "learning_rate": 6.407156632935707e-06, "loss": 2.5685, "step": 241200 }, { "epoch": 0.0197376, "grad_norm": 0.9782227277755737, "learning_rate": 6.40640513604994e-06, "loss": 2.5028, "step": 241210 }, { "epoch": 0.0197632, "grad_norm": 0.8552857041358948, "learning_rate": 6.40565366246768e-06, "loss": 2.8264, "step": 241220 }, { "epoch": 0.0197888, "grad_norm": 0.8877294659614563, "learning_rate": 6.404902212193802e-06, "loss": 2.7521, "step": 241230 }, { "epoch": 0.0198144, "grad_norm": 0.8061519265174866, "learning_rate": 6.404150785233179e-06, "loss": 2.6768, "step": 241240 }, { "epoch": 0.01984, "grad_norm": 1.0305864810943604, "learning_rate": 6.403399381590676e-06, "loss": 2.6553, "step": 241250 }, { "epoch": 0.0198656, "grad_norm": 0.7434669137001038, "learning_rate": 6.402648001271179e-06, "loss": 2.6101, "step": 241260 }, { "epoch": 0.0198912, "grad_norm": 0.7616145014762878, "learning_rate": 6.40189664427955e-06, "loss": 2.715, "step": 241270 }, { "epoch": 0.0199168, "grad_norm": 0.8155465722084045, "learning_rate": 6.401145310620667e-06, "loss": 2.7072, "step": 241280 }, { "epoch": 0.0199424, "grad_norm": 0.7565096020698547, "learning_rate": 6.400394000299399e-06, "loss": 2.6478, "step": 241290 }, { "epoch": 0.019968, "grad_norm": 0.8733323812484741, "learning_rate": 6.399642713320623e-06, "loss": 2.8065, "step": 241300 }, { "epoch": 0.0199936, "grad_norm": 0.8874194622039795, "learning_rate": 6.398891449689204e-06, "loss": 3.0515, "step": 241310 }, { "epoch": 0.0200192, "grad_norm": 0.7824500203132629, "learning_rate": 6.398140209410017e-06, "loss": 2.8739, "step": 241320 }, { "epoch": 0.0200448, "grad_norm": 0.7443954348564148, "learning_rate": 6.397388992487934e-06, "loss": 2.7246, "step": 241330 }, { "epoch": 0.0200704, "grad_norm": 0.763928234577179, "learning_rate": 6.396637798927826e-06, "loss": 2.6293, "step": 241340 }, { "epoch": 0.020096, "grad_norm": 0.7521589398384094, "learning_rate": 6.395886628734563e-06, "loss": 2.4365, "step": 241350 }, { "epoch": 0.0201216, "grad_norm": 0.7891272306442261, "learning_rate": 6.395135481913019e-06, "loss": 2.6753, "step": 241360 }, { "epoch": 0.0201472, "grad_norm": 0.8024390935897827, "learning_rate": 6.394384358468064e-06, "loss": 2.4261, "step": 241370 }, { "epoch": 0.0201728, "grad_norm": 0.8263373374938965, "learning_rate": 6.393633258404566e-06, "loss": 2.5703, "step": 241380 }, { "epoch": 0.0201984, "grad_norm": 0.7916993498802185, "learning_rate": 6.392882181727398e-06, "loss": 2.7218, "step": 241390 }, { "epoch": 0.020224, "grad_norm": 1.024030089378357, "learning_rate": 6.3921311284414325e-06, "loss": 2.6574, "step": 241400 }, { "epoch": 0.0202496, "grad_norm": 0.8969850540161133, "learning_rate": 6.391380098551536e-06, "loss": 2.4344, "step": 241410 }, { "epoch": 0.0202752, "grad_norm": 0.6537690162658691, "learning_rate": 6.390629092062578e-06, "loss": 2.5783, "step": 241420 }, { "epoch": 0.0203008, "grad_norm": 0.7728793621063232, "learning_rate": 6.389878108979434e-06, "loss": 2.5281, "step": 241430 }, { "epoch": 0.0203264, "grad_norm": 0.9146013855934143, "learning_rate": 6.38912714930697e-06, "loss": 2.7197, "step": 241440 }, { "epoch": 0.020352, "grad_norm": 0.9826264977455139, "learning_rate": 6.388376213050057e-06, "loss": 2.643, "step": 241450 }, { "epoch": 0.0203776, "grad_norm": 1.1704944372177124, "learning_rate": 6.387625300213567e-06, "loss": 2.6991, "step": 241460 }, { "epoch": 0.0204032, "grad_norm": 0.8039276003837585, "learning_rate": 6.386874410802364e-06, "loss": 2.7127, "step": 241470 }, { "epoch": 0.0204288, "grad_norm": 0.7833680510520935, "learning_rate": 6.386123544821321e-06, "loss": 2.6495, "step": 241480 }, { "epoch": 0.0204544, "grad_norm": 0.8524758815765381, "learning_rate": 6.385372702275308e-06, "loss": 2.7495, "step": 241490 }, { "epoch": 0.02048, "grad_norm": 0.9572325348854065, "learning_rate": 6.384621883169191e-06, "loss": 2.7829, "step": 241500 }, { "epoch": 0.0205056, "grad_norm": 0.7342213988304138, "learning_rate": 6.383871087507839e-06, "loss": 2.5782, "step": 241510 }, { "epoch": 0.0205312, "grad_norm": 0.8701971769332886, "learning_rate": 6.383120315296123e-06, "loss": 2.9682, "step": 241520 }, { "epoch": 0.0205568, "grad_norm": 0.7582829594612122, "learning_rate": 6.382369566538912e-06, "loss": 2.5884, "step": 241530 }, { "epoch": 0.0205824, "grad_norm": 0.8391811847686768, "learning_rate": 6.381618841241071e-06, "loss": 3.1312, "step": 241540 }, { "epoch": 0.020608, "grad_norm": 0.720500648021698, "learning_rate": 6.38086813940747e-06, "loss": 2.5703, "step": 241550 }, { "epoch": 0.0206336, "grad_norm": 0.8910625576972961, "learning_rate": 6.380117461042978e-06, "loss": 2.6238, "step": 241560 }, { "epoch": 0.0206592, "grad_norm": 0.8965273499488831, "learning_rate": 6.379366806152463e-06, "loss": 2.7004, "step": 241570 }, { "epoch": 0.0206848, "grad_norm": 0.862365186214447, "learning_rate": 6.37861617474079e-06, "loss": 2.6887, "step": 241580 }, { "epoch": 0.0207104, "grad_norm": 0.8264251947402954, "learning_rate": 6.377865566812828e-06, "loss": 2.7534, "step": 241590 }, { "epoch": 0.020736, "grad_norm": 0.7840147614479065, "learning_rate": 6.3771149823734465e-06, "loss": 2.6343, "step": 241600 }, { "epoch": 0.0207616, "grad_norm": 0.8185315132141113, "learning_rate": 6.376364421427512e-06, "loss": 2.8041, "step": 241610 }, { "epoch": 0.0207872, "grad_norm": 0.810184895992279, "learning_rate": 6.375613883979892e-06, "loss": 2.724, "step": 241620 }, { "epoch": 0.0208128, "grad_norm": 0.7416930198669434, "learning_rate": 6.374863370035451e-06, "loss": 2.4844, "step": 241630 }, { "epoch": 0.0208384, "grad_norm": 0.8292123675346375, "learning_rate": 6.374112879599059e-06, "loss": 2.6525, "step": 241640 }, { "epoch": 0.020864, "grad_norm": 0.850098729133606, "learning_rate": 6.37336241267558e-06, "loss": 2.832, "step": 241650 }, { "epoch": 0.0208896, "grad_norm": 0.8846346139907837, "learning_rate": 6.372611969269884e-06, "loss": 2.1639, "step": 241660 }, { "epoch": 0.0209152, "grad_norm": 0.9358989000320435, "learning_rate": 6.371861549386834e-06, "loss": 2.8406, "step": 241670 }, { "epoch": 0.0209408, "grad_norm": 0.9228748679161072, "learning_rate": 6.371111153031298e-06, "loss": 2.582, "step": 241680 }, { "epoch": 0.0209664, "grad_norm": 0.8643438220024109, "learning_rate": 6.370360780208143e-06, "loss": 2.7554, "step": 241690 }, { "epoch": 0.020992, "grad_norm": 0.8450843095779419, "learning_rate": 6.369610430922231e-06, "loss": 2.5373, "step": 241700 }, { "epoch": 0.0210176, "grad_norm": 0.8430025577545166, "learning_rate": 6.368860105178431e-06, "loss": 2.6449, "step": 241710 }, { "epoch": 0.0210432, "grad_norm": 0.8081871271133423, "learning_rate": 6.368109802981608e-06, "loss": 2.7145, "step": 241720 }, { "epoch": 0.0210688, "grad_norm": 0.8510885834693909, "learning_rate": 6.367359524336629e-06, "loss": 2.6111, "step": 241730 }, { "epoch": 0.0210944, "grad_norm": 0.9137452840805054, "learning_rate": 6.366609269248358e-06, "loss": 2.7756, "step": 241740 }, { "epoch": 0.02112, "grad_norm": 1.1049563884735107, "learning_rate": 6.365859037721659e-06, "loss": 3.0026, "step": 241750 }, { "epoch": 0.0211456, "grad_norm": 0.8796895146369934, "learning_rate": 6.3651088297614e-06, "loss": 2.9415, "step": 241760 }, { "epoch": 0.0211712, "grad_norm": 0.8511321544647217, "learning_rate": 6.364358645372439e-06, "loss": 2.5661, "step": 241770 }, { "epoch": 0.0211968, "grad_norm": 0.8037406206130981, "learning_rate": 6.36360848455965e-06, "loss": 2.6089, "step": 241780 }, { "epoch": 0.0212224, "grad_norm": 1.2179722785949707, "learning_rate": 6.362858347327896e-06, "loss": 1.9722, "step": 241790 }, { "epoch": 0.021248, "grad_norm": 0.8450062274932861, "learning_rate": 6.362108233682036e-06, "loss": 2.8077, "step": 241800 }, { "epoch": 0.0212736, "grad_norm": 0.9512060284614563, "learning_rate": 6.361358143626937e-06, "loss": 2.927, "step": 241810 }, { "epoch": 0.0212992, "grad_norm": 0.7444392442703247, "learning_rate": 6.360608077167466e-06, "loss": 2.6557, "step": 241820 }, { "epoch": 0.0213248, "grad_norm": 0.8946385383605957, "learning_rate": 6.3598580343084825e-06, "loss": 3.0516, "step": 241830 }, { "epoch": 0.0213504, "grad_norm": 1.2173491716384888, "learning_rate": 6.359108015054851e-06, "loss": 2.5762, "step": 241840 }, { "epoch": 0.021376, "grad_norm": 0.869301974773407, "learning_rate": 6.358358019411438e-06, "loss": 2.839, "step": 241850 }, { "epoch": 0.0214016, "grad_norm": 0.8513619899749756, "learning_rate": 6.357608047383107e-06, "loss": 2.4343, "step": 241860 }, { "epoch": 0.0214272, "grad_norm": 0.790263831615448, "learning_rate": 6.3568580989747176e-06, "loss": 2.5465, "step": 241870 }, { "epoch": 0.0214528, "grad_norm": 0.7888185977935791, "learning_rate": 6.356108174191136e-06, "loss": 2.6278, "step": 241880 }, { "epoch": 0.0214784, "grad_norm": 0.7682559490203857, "learning_rate": 6.3553582730372245e-06, "loss": 2.9377, "step": 241890 }, { "epoch": 0.021504, "grad_norm": 0.8168535232543945, "learning_rate": 6.354608395517845e-06, "loss": 2.9134, "step": 241900 }, { "epoch": 0.0215296, "grad_norm": 0.8104252815246582, "learning_rate": 6.353858541637862e-06, "loss": 2.9866, "step": 241910 }, { "epoch": 0.0215552, "grad_norm": 0.8542898893356323, "learning_rate": 6.353108711402136e-06, "loss": 2.962, "step": 241920 }, { "epoch": 0.0215808, "grad_norm": 0.7723912000656128, "learning_rate": 6.352358904815535e-06, "loss": 2.6504, "step": 241930 }, { "epoch": 0.0216064, "grad_norm": 2.7133615016937256, "learning_rate": 6.351609121882909e-06, "loss": 3.0227, "step": 241940 }, { "epoch": 0.021632, "grad_norm": 0.729575514793396, "learning_rate": 6.350859362609134e-06, "loss": 2.9637, "step": 241950 }, { "epoch": 0.0216576, "grad_norm": 0.8313182592391968, "learning_rate": 6.350109626999066e-06, "loss": 2.7288, "step": 241960 }, { "epoch": 0.0216832, "grad_norm": 0.7999184131622314, "learning_rate": 6.3493599150575645e-06, "loss": 2.8759, "step": 241970 }, { "epoch": 0.0217088, "grad_norm": 0.8434196710586548, "learning_rate": 6.348610226789496e-06, "loss": 2.3029, "step": 241980 }, { "epoch": 0.0217344, "grad_norm": 0.7598303556442261, "learning_rate": 6.347860562199722e-06, "loss": 2.4732, "step": 241990 }, { "epoch": 0.02176, "grad_norm": 0.9423151612281799, "learning_rate": 6.347110921293098e-06, "loss": 2.6947, "step": 242000 }, { "epoch": 0.0217856, "grad_norm": 0.9463802576065063, "learning_rate": 6.346361304074489e-06, "loss": 2.935, "step": 242010 }, { "epoch": 0.0218112, "grad_norm": 0.7606985569000244, "learning_rate": 6.345611710548758e-06, "loss": 2.832, "step": 242020 }, { "epoch": 0.0218368, "grad_norm": 0.8395470976829529, "learning_rate": 6.344862140720762e-06, "loss": 2.7088, "step": 242030 }, { "epoch": 0.0218624, "grad_norm": 0.858974277973175, "learning_rate": 6.344112594595364e-06, "loss": 2.8459, "step": 242040 }, { "epoch": 0.021888, "grad_norm": 0.77288818359375, "learning_rate": 6.343363072177425e-06, "loss": 2.8068, "step": 242050 }, { "epoch": 0.0219136, "grad_norm": 0.9305662512779236, "learning_rate": 6.342613573471805e-06, "loss": 2.6996, "step": 242060 }, { "epoch": 0.0219392, "grad_norm": 0.81998610496521, "learning_rate": 6.341864098483364e-06, "loss": 2.6469, "step": 242070 }, { "epoch": 0.0219648, "grad_norm": 0.8520130515098572, "learning_rate": 6.3411146472169595e-06, "loss": 2.8402, "step": 242080 }, { "epoch": 0.0219904, "grad_norm": 0.881865382194519, "learning_rate": 6.340365219677458e-06, "loss": 2.9927, "step": 242090 }, { "epoch": 0.022016, "grad_norm": 0.7894597053527832, "learning_rate": 6.339615815869712e-06, "loss": 2.9199, "step": 242100 }, { "epoch": 0.0220416, "grad_norm": 0.7534925937652588, "learning_rate": 6.338866435798585e-06, "loss": 2.9414, "step": 242110 }, { "epoch": 0.0220672, "grad_norm": 0.8318463563919067, "learning_rate": 6.338117079468935e-06, "loss": 2.8328, "step": 242120 }, { "epoch": 0.0220928, "grad_norm": 0.8570569157600403, "learning_rate": 6.337367746885623e-06, "loss": 2.5692, "step": 242130 }, { "epoch": 0.0221184, "grad_norm": 0.7742176055908203, "learning_rate": 6.33661843805351e-06, "loss": 2.6079, "step": 242140 }, { "epoch": 0.022144, "grad_norm": 1.0132349729537964, "learning_rate": 6.335869152977454e-06, "loss": 2.7403, "step": 242150 }, { "epoch": 0.0221696, "grad_norm": 0.8284965753555298, "learning_rate": 6.3351198916623094e-06, "loss": 2.9144, "step": 242160 }, { "epoch": 0.0221952, "grad_norm": 0.8233247399330139, "learning_rate": 6.33437065411294e-06, "loss": 2.8308, "step": 242170 }, { "epoch": 0.0222208, "grad_norm": 0.8854260444641113, "learning_rate": 6.333621440334201e-06, "loss": 2.7528, "step": 242180 }, { "epoch": 0.0222464, "grad_norm": 0.9083479046821594, "learning_rate": 6.332872250330957e-06, "loss": 2.9788, "step": 242190 }, { "epoch": 0.022272, "grad_norm": 0.9018794298171997, "learning_rate": 6.332123084108057e-06, "loss": 2.8396, "step": 242200 }, { "epoch": 0.0222976, "grad_norm": 0.8409528732299805, "learning_rate": 6.331373941670367e-06, "loss": 2.8269, "step": 242210 }, { "epoch": 0.0223232, "grad_norm": 0.7286116480827332, "learning_rate": 6.330624823022742e-06, "loss": 2.9392, "step": 242220 }, { "epoch": 0.0223488, "grad_norm": 0.7625986337661743, "learning_rate": 6.329875728170037e-06, "loss": 2.8777, "step": 242230 }, { "epoch": 0.0223744, "grad_norm": 0.901439368724823, "learning_rate": 6.329126657117115e-06, "loss": 2.8648, "step": 242240 }, { "epoch": 0.0224, "grad_norm": 0.7857005000114441, "learning_rate": 6.328377609868829e-06, "loss": 2.8478, "step": 242250 }, { "epoch": 0.0224256, "grad_norm": 0.8177239298820496, "learning_rate": 6.3276285864300415e-06, "loss": 2.8338, "step": 242260 }, { "epoch": 0.0224512, "grad_norm": 0.849847137928009, "learning_rate": 6.3268795868056035e-06, "loss": 2.9841, "step": 242270 }, { "epoch": 0.0224768, "grad_norm": 0.8300850987434387, "learning_rate": 6.3261306110003766e-06, "loss": 2.9833, "step": 242280 }, { "epoch": 0.0225024, "grad_norm": 1.0519084930419922, "learning_rate": 6.325381659019213e-06, "loss": 2.8649, "step": 242290 }, { "epoch": 0.022528, "grad_norm": 0.7995509505271912, "learning_rate": 6.324632730866975e-06, "loss": 2.7234, "step": 242300 }, { "epoch": 0.0225536, "grad_norm": 0.9051342010498047, "learning_rate": 6.32388382654852e-06, "loss": 2.7843, "step": 242310 }, { "epoch": 0.0225792, "grad_norm": 0.9705117344856262, "learning_rate": 6.323134946068699e-06, "loss": 2.7721, "step": 242320 }, { "epoch": 0.0226048, "grad_norm": 0.7988047003746033, "learning_rate": 6.322386089432369e-06, "loss": 3.0283, "step": 242330 }, { "epoch": 0.0226304, "grad_norm": 0.7559856176376343, "learning_rate": 6.321637256644391e-06, "loss": 2.7677, "step": 242340 }, { "epoch": 0.022656, "grad_norm": 0.759038507938385, "learning_rate": 6.320888447709616e-06, "loss": 2.9404, "step": 242350 }, { "epoch": 0.0226816, "grad_norm": 0.9021894335746765, "learning_rate": 6.320139662632903e-06, "loss": 2.9274, "step": 242360 }, { "epoch": 0.0227072, "grad_norm": 0.8798137903213501, "learning_rate": 6.319390901419104e-06, "loss": 2.9869, "step": 242370 }, { "epoch": 0.0227328, "grad_norm": 0.7782102823257446, "learning_rate": 6.318642164073081e-06, "loss": 2.6342, "step": 242380 }, { "epoch": 0.0227584, "grad_norm": 0.9033834934234619, "learning_rate": 6.317893450599681e-06, "loss": 2.8428, "step": 242390 }, { "epoch": 0.022784, "grad_norm": 0.8260931372642517, "learning_rate": 6.317144761003764e-06, "loss": 2.8573, "step": 242400 }, { "epoch": 0.0228096, "grad_norm": 0.764988899230957, "learning_rate": 6.316396095290185e-06, "loss": 2.9675, "step": 242410 }, { "epoch": 0.0228352, "grad_norm": 0.7442395687103271, "learning_rate": 6.3156474534638e-06, "loss": 2.7612, "step": 242420 }, { "epoch": 0.0228608, "grad_norm": 0.8424559235572815, "learning_rate": 6.31489883552946e-06, "loss": 2.7566, "step": 242430 }, { "epoch": 0.0228864, "grad_norm": 0.7711079716682434, "learning_rate": 6.314150241492022e-06, "loss": 2.6976, "step": 242440 }, { "epoch": 0.022912, "grad_norm": 0.8627309203147888, "learning_rate": 6.313401671356341e-06, "loss": 3.1558, "step": 242450 }, { "epoch": 0.0229376, "grad_norm": 0.7990326881408691, "learning_rate": 6.312653125127266e-06, "loss": 3.0371, "step": 242460 }, { "epoch": 0.0229632, "grad_norm": 0.9296841621398926, "learning_rate": 6.311904602809659e-06, "loss": 2.913, "step": 242470 }, { "epoch": 0.0229888, "grad_norm": 0.8271118402481079, "learning_rate": 6.311156104408372e-06, "loss": 3.043, "step": 242480 }, { "epoch": 0.0230144, "grad_norm": 0.7641383409500122, "learning_rate": 6.310407629928256e-06, "loss": 2.941, "step": 242490 }, { "epoch": 0.02304, "grad_norm": 0.8272345662117004, "learning_rate": 6.309659179374166e-06, "loss": 3.0368, "step": 242500 }, { "epoch": 0.0230656, "grad_norm": 0.8915419578552246, "learning_rate": 6.3089107527509564e-06, "loss": 2.7708, "step": 242510 }, { "epoch": 0.0230912, "grad_norm": 0.7544513940811157, "learning_rate": 6.308162350063478e-06, "loss": 2.8588, "step": 242520 }, { "epoch": 0.0231168, "grad_norm": 0.7428746819496155, "learning_rate": 6.307413971316585e-06, "loss": 2.7752, "step": 242530 }, { "epoch": 0.0231424, "grad_norm": 0.8165407776832581, "learning_rate": 6.306665616515133e-06, "loss": 3.0163, "step": 242540 }, { "epoch": 0.023168, "grad_norm": 0.7642818689346313, "learning_rate": 6.305917285663974e-06, "loss": 2.8774, "step": 242550 }, { "epoch": 0.0231936, "grad_norm": 0.8155698776245117, "learning_rate": 6.3051689787679575e-06, "loss": 3.0172, "step": 242560 }, { "epoch": 0.0232192, "grad_norm": 0.7146037817001343, "learning_rate": 6.304420695831937e-06, "loss": 2.8177, "step": 242570 }, { "epoch": 0.0232448, "grad_norm": 0.7790893912315369, "learning_rate": 6.303672436860769e-06, "loss": 2.9263, "step": 242580 }, { "epoch": 0.0232704, "grad_norm": 0.773955762386322, "learning_rate": 6.302924201859302e-06, "loss": 2.8901, "step": 242590 }, { "epoch": 0.023296, "grad_norm": 0.8366961479187012, "learning_rate": 6.302175990832387e-06, "loss": 2.9416, "step": 242600 }, { "epoch": 0.0233216, "grad_norm": 0.8889588713645935, "learning_rate": 6.301427803784879e-06, "loss": 2.7666, "step": 242610 }, { "epoch": 0.0233472, "grad_norm": 0.8962512016296387, "learning_rate": 6.300679640721631e-06, "loss": 2.8612, "step": 242620 }, { "epoch": 0.0233728, "grad_norm": 0.7357283234596252, "learning_rate": 6.299931501647489e-06, "loss": 2.8943, "step": 242630 }, { "epoch": 0.0233984, "grad_norm": 0.7998353838920593, "learning_rate": 6.2991833865673065e-06, "loss": 2.9108, "step": 242640 }, { "epoch": 0.023424, "grad_norm": 0.8492012619972229, "learning_rate": 6.29843529548594e-06, "loss": 2.791, "step": 242650 }, { "epoch": 0.0234496, "grad_norm": 0.8024083971977234, "learning_rate": 6.297687228408235e-06, "loss": 2.7744, "step": 242660 }, { "epoch": 0.0234752, "grad_norm": 0.9631216526031494, "learning_rate": 6.2969391853390435e-06, "loss": 3.0524, "step": 242670 }, { "epoch": 0.0235008, "grad_norm": 1.270646572113037, "learning_rate": 6.296191166283221e-06, "loss": 2.9125, "step": 242680 }, { "epoch": 0.0235264, "grad_norm": 0.8426502346992493, "learning_rate": 6.295443171245612e-06, "loss": 2.9278, "step": 242690 }, { "epoch": 0.023552, "grad_norm": 0.853417158126831, "learning_rate": 6.29469520023107e-06, "loss": 2.9837, "step": 242700 }, { "epoch": 0.0235776, "grad_norm": 0.8003159165382385, "learning_rate": 6.2939472532444454e-06, "loss": 2.8908, "step": 242710 }, { "epoch": 0.0236032, "grad_norm": 0.8726856112480164, "learning_rate": 6.2931993302905866e-06, "loss": 3.0329, "step": 242720 }, { "epoch": 0.0236288, "grad_norm": 0.7263809442520142, "learning_rate": 6.292451431374347e-06, "loss": 2.7168, "step": 242730 }, { "epoch": 0.0236544, "grad_norm": 0.8176568150520325, "learning_rate": 6.291703556500573e-06, "loss": 3.1335, "step": 242740 }, { "epoch": 0.02368, "grad_norm": 0.8923645615577698, "learning_rate": 6.290955705674116e-06, "loss": 2.8252, "step": 242750 }, { "epoch": 0.0237056, "grad_norm": 0.7726889252662659, "learning_rate": 6.2902078788998255e-06, "loss": 2.8411, "step": 242760 }, { "epoch": 0.0237312, "grad_norm": 0.8239255547523499, "learning_rate": 6.289460076182551e-06, "loss": 2.8494, "step": 242770 }, { "epoch": 0.0237568, "grad_norm": 0.7373718619346619, "learning_rate": 6.288712297527143e-06, "loss": 2.681, "step": 242780 }, { "epoch": 0.0237824, "grad_norm": 0.7486975193023682, "learning_rate": 6.287964542938447e-06, "loss": 2.7514, "step": 242790 }, { "epoch": 0.023808, "grad_norm": 0.8790756464004517, "learning_rate": 6.287216812421315e-06, "loss": 2.817, "step": 242800 }, { "epoch": 0.0238336, "grad_norm": 0.7366830110549927, "learning_rate": 6.286469105980595e-06, "loss": 2.9387, "step": 242810 }, { "epoch": 0.0238592, "grad_norm": 0.7557868957519531, "learning_rate": 6.285721423621136e-06, "loss": 2.7843, "step": 242820 }, { "epoch": 0.0238848, "grad_norm": 0.7792617082595825, "learning_rate": 6.284973765347786e-06, "loss": 2.9037, "step": 242830 }, { "epoch": 0.0239104, "grad_norm": 0.8416740298271179, "learning_rate": 6.284226131165397e-06, "loss": 2.8128, "step": 242840 }, { "epoch": 0.023936, "grad_norm": 1.4450913667678833, "learning_rate": 6.283478521078812e-06, "loss": 2.8201, "step": 242850 }, { "epoch": 0.0239616, "grad_norm": 0.8798723220825195, "learning_rate": 6.28273093509288e-06, "loss": 2.8008, "step": 242860 }, { "epoch": 0.0239872, "grad_norm": 0.7657660841941833, "learning_rate": 6.28198337321245e-06, "loss": 2.9026, "step": 242870 }, { "epoch": 0.0240128, "grad_norm": 0.9176483154296875, "learning_rate": 6.281235835442372e-06, "loss": 3.0116, "step": 242880 }, { "epoch": 0.0240384, "grad_norm": 0.8225199580192566, "learning_rate": 6.2804883217874905e-06, "loss": 2.7842, "step": 242890 }, { "epoch": 0.024064, "grad_norm": 0.7576464414596558, "learning_rate": 6.279740832252652e-06, "loss": 2.9898, "step": 242900 }, { "epoch": 0.0240896, "grad_norm": 1.0117143392562866, "learning_rate": 6.278993366842708e-06, "loss": 2.7238, "step": 242910 }, { "epoch": 0.0241152, "grad_norm": 0.9309149384498596, "learning_rate": 6.278245925562502e-06, "loss": 3.0139, "step": 242920 }, { "epoch": 0.0241408, "grad_norm": 0.7570151686668396, "learning_rate": 6.2774985084168805e-06, "loss": 2.9034, "step": 242930 }, { "epoch": 0.0241664, "grad_norm": 1.0054396390914917, "learning_rate": 6.276751115410693e-06, "loss": 2.8112, "step": 242940 }, { "epoch": 0.024192, "grad_norm": 0.7132278084754944, "learning_rate": 6.276003746548786e-06, "loss": 2.8907, "step": 242950 }, { "epoch": 0.0242176, "grad_norm": 0.8303020596504211, "learning_rate": 6.275256401836004e-06, "loss": 2.9521, "step": 242960 }, { "epoch": 0.0242432, "grad_norm": 0.8097726702690125, "learning_rate": 6.274509081277193e-06, "loss": 2.9371, "step": 242970 }, { "epoch": 0.0242688, "grad_norm": 0.7303343415260315, "learning_rate": 6.2737617848772045e-06, "loss": 2.6924, "step": 242980 }, { "epoch": 0.0242944, "grad_norm": 0.8435531258583069, "learning_rate": 6.273014512640874e-06, "loss": 2.8425, "step": 242990 }, { "epoch": 0.02432, "grad_norm": 0.7958796620368958, "learning_rate": 6.272267264573058e-06, "loss": 2.755, "step": 243000 }, { "epoch": 0.0243456, "grad_norm": 0.7736408710479736, "learning_rate": 6.271520040678598e-06, "loss": 2.8259, "step": 243010 }, { "epoch": 0.0243712, "grad_norm": 1.027687907218933, "learning_rate": 6.2707728409623394e-06, "loss": 2.8028, "step": 243020 }, { "epoch": 0.0243968, "grad_norm": 0.8220931887626648, "learning_rate": 6.270025665429127e-06, "loss": 2.6984, "step": 243030 }, { "epoch": 0.0244224, "grad_norm": 0.8256177306175232, "learning_rate": 6.26927851408381e-06, "loss": 2.9386, "step": 243040 }, { "epoch": 0.024448, "grad_norm": 0.868943452835083, "learning_rate": 6.268531386931227e-06, "loss": 2.7607, "step": 243050 }, { "epoch": 0.0244736, "grad_norm": 0.7839624881744385, "learning_rate": 6.267784283976227e-06, "loss": 2.9256, "step": 243060 }, { "epoch": 0.0244992, "grad_norm": 0.9395226240158081, "learning_rate": 6.267037205223654e-06, "loss": 2.7171, "step": 243070 }, { "epoch": 0.0245248, "grad_norm": 0.7696196436882019, "learning_rate": 6.2662901506783545e-06, "loss": 2.6961, "step": 243080 }, { "epoch": 0.0245504, "grad_norm": 1.012357234954834, "learning_rate": 6.265543120345169e-06, "loss": 2.7249, "step": 243090 }, { "epoch": 0.024576, "grad_norm": 0.8223379850387573, "learning_rate": 6.2647961142289435e-06, "loss": 2.8442, "step": 243100 }, { "epoch": 0.0246016, "grad_norm": 0.7494214773178101, "learning_rate": 6.264049132334525e-06, "loss": 2.9104, "step": 243110 }, { "epoch": 0.0246272, "grad_norm": 0.8259183168411255, "learning_rate": 6.263302174666753e-06, "loss": 2.7718, "step": 243120 }, { "epoch": 0.0246528, "grad_norm": 1.0116506814956665, "learning_rate": 6.262555241230473e-06, "loss": 2.8541, "step": 243130 }, { "epoch": 0.0246784, "grad_norm": 0.7834542393684387, "learning_rate": 6.26180833203053e-06, "loss": 3.0203, "step": 243140 }, { "epoch": 0.024704, "grad_norm": 0.7912849187850952, "learning_rate": 6.261061447071768e-06, "loss": 2.9895, "step": 243150 }, { "epoch": 0.0247296, "grad_norm": 0.7623744606971741, "learning_rate": 6.260314586359022e-06, "loss": 2.6965, "step": 243160 }, { "epoch": 0.0247552, "grad_norm": 0.8351172208786011, "learning_rate": 6.2595677498971485e-06, "loss": 2.6959, "step": 243170 }, { "epoch": 0.0247808, "grad_norm": 0.806041955947876, "learning_rate": 6.258820937690982e-06, "loss": 2.6674, "step": 243180 }, { "epoch": 0.0248064, "grad_norm": 0.7890927791595459, "learning_rate": 6.258074149745368e-06, "loss": 2.8364, "step": 243190 }, { "epoch": 0.024832, "grad_norm": 0.7730260491371155, "learning_rate": 6.25732738606515e-06, "loss": 2.6664, "step": 243200 }, { "epoch": 0.0248576, "grad_norm": 0.7698234915733337, "learning_rate": 6.256580646655167e-06, "loss": 2.7489, "step": 243210 }, { "epoch": 0.0248832, "grad_norm": 0.8417500257492065, "learning_rate": 6.2558339315202636e-06, "loss": 2.7899, "step": 243220 }, { "epoch": 0.0249088, "grad_norm": 0.8411496877670288, "learning_rate": 6.255087240665283e-06, "loss": 2.8443, "step": 243230 }, { "epoch": 0.0249344, "grad_norm": 0.7857900857925415, "learning_rate": 6.254340574095068e-06, "loss": 2.9651, "step": 243240 }, { "epoch": 0.02496, "grad_norm": 0.881924569606781, "learning_rate": 6.253593931814457e-06, "loss": 3.0161, "step": 243250 }, { "epoch": 0.0249856, "grad_norm": 0.7956781983375549, "learning_rate": 6.252847313828292e-06, "loss": 2.6434, "step": 243260 }, { "epoch": 0.0250112, "grad_norm": 0.7339916229248047, "learning_rate": 6.252100720141419e-06, "loss": 2.6605, "step": 243270 }, { "epoch": 0.0250368, "grad_norm": 0.8521314859390259, "learning_rate": 6.251354150758677e-06, "loss": 2.6451, "step": 243280 }, { "epoch": 0.0250624, "grad_norm": 0.774533748626709, "learning_rate": 6.250607605684905e-06, "loss": 2.8727, "step": 243290 }, { "epoch": 0.025088, "grad_norm": 0.7056618332862854, "learning_rate": 6.249861084924946e-06, "loss": 2.8318, "step": 243300 }, { "epoch": 0.0251136, "grad_norm": 0.9176211357116699, "learning_rate": 6.249114588483645e-06, "loss": 2.7787, "step": 243310 }, { "epoch": 0.0251392, "grad_norm": 0.8349877595901489, "learning_rate": 6.248368116365836e-06, "loss": 2.8202, "step": 243320 }, { "epoch": 0.0251648, "grad_norm": 0.7803732752799988, "learning_rate": 6.2476216685763606e-06, "loss": 2.9056, "step": 243330 }, { "epoch": 0.0251904, "grad_norm": 1.614595890045166, "learning_rate": 6.246875245120063e-06, "loss": 3.0622, "step": 243340 }, { "epoch": 0.025216, "grad_norm": 0.7445797324180603, "learning_rate": 6.246128846001782e-06, "loss": 2.7992, "step": 243350 }, { "epoch": 0.0252416, "grad_norm": 0.7517947554588318, "learning_rate": 6.2453824712263585e-06, "loss": 2.7291, "step": 243360 }, { "epoch": 0.0252672, "grad_norm": 0.7816565632820129, "learning_rate": 6.244636120798635e-06, "loss": 2.8079, "step": 243370 }, { "epoch": 0.0252928, "grad_norm": 0.9137887358665466, "learning_rate": 6.243889794723444e-06, "loss": 2.9619, "step": 243380 }, { "epoch": 0.0253184, "grad_norm": 0.8757987022399902, "learning_rate": 6.24314349300563e-06, "loss": 2.7477, "step": 243390 }, { "epoch": 0.025344, "grad_norm": 0.8023154735565186, "learning_rate": 6.242397215650035e-06, "loss": 3.0794, "step": 243400 }, { "epoch": 0.0253696, "grad_norm": 0.821875810623169, "learning_rate": 6.241650962661492e-06, "loss": 2.869, "step": 243410 }, { "epoch": 0.0253952, "grad_norm": 0.8736046552658081, "learning_rate": 6.240904734044844e-06, "loss": 2.8805, "step": 243420 }, { "epoch": 0.0254208, "grad_norm": 0.8493333458900452, "learning_rate": 6.240158529804929e-06, "loss": 2.8289, "step": 243430 }, { "epoch": 0.0254464, "grad_norm": 0.7328749299049377, "learning_rate": 6.239412349946591e-06, "loss": 2.8036, "step": 243440 }, { "epoch": 0.025472, "grad_norm": 0.723412275314331, "learning_rate": 6.23866619447466e-06, "loss": 2.7035, "step": 243450 }, { "epoch": 0.0254976, "grad_norm": 0.7340456247329712, "learning_rate": 6.237920063393981e-06, "loss": 2.7197, "step": 243460 }, { "epoch": 0.0255232, "grad_norm": 1.0646202564239502, "learning_rate": 6.237173956709393e-06, "loss": 2.5266, "step": 243470 }, { "epoch": 0.0255488, "grad_norm": 0.7661302089691162, "learning_rate": 6.2364278744257275e-06, "loss": 2.7801, "step": 243480 }, { "epoch": 0.0255744, "grad_norm": 0.7961762547492981, "learning_rate": 6.235681816547828e-06, "loss": 3.0027, "step": 243490 }, { "epoch": 0.0256, "grad_norm": 0.9123858213424683, "learning_rate": 6.234935783080531e-06, "loss": 2.7839, "step": 243500 }, { "epoch": 0.0256256, "grad_norm": 0.7918736338615417, "learning_rate": 6.234189774028674e-06, "loss": 2.849, "step": 243510 }, { "epoch": 0.0256512, "grad_norm": 0.7422581911087036, "learning_rate": 6.2334437893970965e-06, "loss": 2.8846, "step": 243520 }, { "epoch": 0.0256768, "grad_norm": 0.7865049839019775, "learning_rate": 6.232697829190637e-06, "loss": 2.9491, "step": 243530 }, { "epoch": 0.0257024, "grad_norm": 1.177509069442749, "learning_rate": 6.23195189341413e-06, "loss": 2.7931, "step": 243540 }, { "epoch": 0.025728, "grad_norm": 0.9647371172904968, "learning_rate": 6.2312059820724125e-06, "loss": 2.621, "step": 243550 }, { "epoch": 0.0257536, "grad_norm": 0.79997318983078, "learning_rate": 6.230460095170323e-06, "loss": 2.8967, "step": 243560 }, { "epoch": 0.0257792, "grad_norm": 0.7548099160194397, "learning_rate": 6.2297142327127e-06, "loss": 2.7007, "step": 243570 }, { "epoch": 0.0258048, "grad_norm": 0.7772356271743774, "learning_rate": 6.228968394704375e-06, "loss": 2.7858, "step": 243580 }, { "epoch": 0.0258304, "grad_norm": 0.764771580696106, "learning_rate": 6.2282225811501895e-06, "loss": 2.9262, "step": 243590 }, { "epoch": 0.025856, "grad_norm": 0.9595831036567688, "learning_rate": 6.22747679205498e-06, "loss": 2.9553, "step": 243600 }, { "epoch": 0.0258816, "grad_norm": 1.0719542503356934, "learning_rate": 6.226731027423578e-06, "loss": 2.8982, "step": 243610 }, { "epoch": 0.0259072, "grad_norm": 0.7487960457801819, "learning_rate": 6.2259852872608225e-06, "loss": 2.7924, "step": 243620 }, { "epoch": 0.0259328, "grad_norm": 0.750139594078064, "learning_rate": 6.22523957157155e-06, "loss": 2.6781, "step": 243630 }, { "epoch": 0.0259584, "grad_norm": 0.8099033236503601, "learning_rate": 6.224493880360597e-06, "loss": 2.893, "step": 243640 }, { "epoch": 0.025984, "grad_norm": 0.8260031342506409, "learning_rate": 6.223748213632797e-06, "loss": 2.7878, "step": 243650 }, { "epoch": 0.0260096, "grad_norm": 0.8361855745315552, "learning_rate": 6.223002571392984e-06, "loss": 2.8506, "step": 243660 }, { "epoch": 0.0260352, "grad_norm": 1.106593132019043, "learning_rate": 6.2222569536459996e-06, "loss": 2.7721, "step": 243670 }, { "epoch": 0.0260608, "grad_norm": 0.7825963497161865, "learning_rate": 6.221511360396668e-06, "loss": 2.802, "step": 243680 }, { "epoch": 0.0260864, "grad_norm": 0.8381059169769287, "learning_rate": 6.220765791649837e-06, "loss": 2.7925, "step": 243690 }, { "epoch": 0.026112, "grad_norm": 1.1106250286102295, "learning_rate": 6.220020247410333e-06, "loss": 2.8014, "step": 243700 }, { "epoch": 0.0261376, "grad_norm": 0.7626208066940308, "learning_rate": 6.219274727682993e-06, "loss": 2.6884, "step": 243710 }, { "epoch": 0.0261632, "grad_norm": 0.774734377861023, "learning_rate": 6.218529232472653e-06, "loss": 2.819, "step": 243720 }, { "epoch": 0.0261888, "grad_norm": 0.7513295412063599, "learning_rate": 6.217783761784147e-06, "loss": 2.8269, "step": 243730 }, { "epoch": 0.0262144, "grad_norm": 0.7507427930831909, "learning_rate": 6.217038315622307e-06, "loss": 2.8753, "step": 243740 }, { "epoch": 0.02624, "grad_norm": 0.8144906163215637, "learning_rate": 6.216292893991967e-06, "loss": 2.7255, "step": 243750 }, { "epoch": 0.0262656, "grad_norm": 0.7530681490898132, "learning_rate": 6.215547496897962e-06, "loss": 2.7242, "step": 243760 }, { "epoch": 0.0262912, "grad_norm": 0.8805318474769592, "learning_rate": 6.214802124345129e-06, "loss": 3.0307, "step": 243770 }, { "epoch": 0.0263168, "grad_norm": 0.8206403255462646, "learning_rate": 6.214056776338295e-06, "loss": 2.6937, "step": 243780 }, { "epoch": 0.0263424, "grad_norm": 0.7549155354499817, "learning_rate": 6.2133114528822965e-06, "loss": 2.7259, "step": 243790 }, { "epoch": 0.026368, "grad_norm": 0.8217437863349915, "learning_rate": 6.2125661539819695e-06, "loss": 2.9421, "step": 243800 }, { "epoch": 0.0263936, "grad_norm": 0.7536881566047668, "learning_rate": 6.211820879642143e-06, "loss": 2.8524, "step": 243810 }, { "epoch": 0.0264192, "grad_norm": 0.7992450594902039, "learning_rate": 6.21107562986765e-06, "loss": 2.8421, "step": 243820 }, { "epoch": 0.0264448, "grad_norm": 0.8556357622146606, "learning_rate": 6.210330404663325e-06, "loss": 2.6624, "step": 243830 }, { "epoch": 0.0264704, "grad_norm": 0.7863959074020386, "learning_rate": 6.209585204034003e-06, "loss": 2.6948, "step": 243840 }, { "epoch": 0.026496, "grad_norm": 0.747818648815155, "learning_rate": 6.208840027984511e-06, "loss": 2.8515, "step": 243850 }, { "epoch": 0.0265216, "grad_norm": 0.9050994515419006, "learning_rate": 6.208094876519681e-06, "loss": 2.9467, "step": 243860 }, { "epoch": 0.0265472, "grad_norm": 0.7907360196113586, "learning_rate": 6.207349749644351e-06, "loss": 2.8985, "step": 243870 }, { "epoch": 0.0265728, "grad_norm": 0.9136433601379395, "learning_rate": 6.206604647363349e-06, "loss": 3.0991, "step": 243880 }, { "epoch": 0.0265984, "grad_norm": 0.8457773923873901, "learning_rate": 6.20585956968151e-06, "loss": 2.929, "step": 243890 }, { "epoch": 0.026624, "grad_norm": 0.8181098699569702, "learning_rate": 6.2051145166036616e-06, "loss": 3.0483, "step": 243900 }, { "epoch": 0.0266496, "grad_norm": 0.8569298386573792, "learning_rate": 6.204369488134637e-06, "loss": 2.8848, "step": 243910 }, { "epoch": 0.0266752, "grad_norm": 0.7454953193664551, "learning_rate": 6.203624484279268e-06, "loss": 2.8697, "step": 243920 }, { "epoch": 0.0267008, "grad_norm": 0.9335304498672485, "learning_rate": 6.202879505042387e-06, "loss": 2.8753, "step": 243930 }, { "epoch": 0.0267264, "grad_norm": 0.7655903697013855, "learning_rate": 6.20213455042882e-06, "loss": 3.056, "step": 243940 }, { "epoch": 0.026752, "grad_norm": 0.7333716750144958, "learning_rate": 6.201389620443404e-06, "loss": 2.5906, "step": 243950 }, { "epoch": 0.0267776, "grad_norm": 0.91160649061203, "learning_rate": 6.200644715090965e-06, "loss": 2.9397, "step": 243960 }, { "epoch": 0.0268032, "grad_norm": 0.936332106590271, "learning_rate": 6.1998998343763365e-06, "loss": 3.0182, "step": 243970 }, { "epoch": 0.0268288, "grad_norm": 1.1925079822540283, "learning_rate": 6.1991549783043455e-06, "loss": 2.9419, "step": 243980 }, { "epoch": 0.0268544, "grad_norm": 0.8325220942497253, "learning_rate": 6.198410146879825e-06, "loss": 2.8308, "step": 243990 }, { "epoch": 0.02688, "grad_norm": 0.7177299857139587, "learning_rate": 6.197665340107606e-06, "loss": 2.8882, "step": 244000 }, { "epoch": 0.0269056, "grad_norm": 0.8861700296401978, "learning_rate": 6.196920557992517e-06, "loss": 2.7763, "step": 244010 }, { "epoch": 0.0269312, "grad_norm": 0.7905041575431824, "learning_rate": 6.196175800539385e-06, "loss": 2.8319, "step": 244020 }, { "epoch": 0.0269568, "grad_norm": 0.8092795610427856, "learning_rate": 6.1954310677530396e-06, "loss": 2.9212, "step": 244030 }, { "epoch": 0.0269824, "grad_norm": 0.8067270517349243, "learning_rate": 6.194686359638316e-06, "loss": 2.7979, "step": 244040 }, { "epoch": 0.027008, "grad_norm": 1.442391037940979, "learning_rate": 6.19394167620004e-06, "loss": 2.9848, "step": 244050 }, { "epoch": 0.0270336, "grad_norm": 0.7870419025421143, "learning_rate": 6.193197017443042e-06, "loss": 2.7639, "step": 244060 }, { "epoch": 0.0270592, "grad_norm": 0.8086248636245728, "learning_rate": 6.1924523833721475e-06, "loss": 2.9116, "step": 244070 }, { "epoch": 0.0270848, "grad_norm": 0.9754043221473694, "learning_rate": 6.191707773992189e-06, "loss": 2.882, "step": 244080 }, { "epoch": 0.0271104, "grad_norm": 0.7808353304862976, "learning_rate": 6.190963189307994e-06, "loss": 2.8885, "step": 244090 }, { "epoch": 0.027136, "grad_norm": 0.8002735376358032, "learning_rate": 6.190218629324388e-06, "loss": 2.8906, "step": 244100 }, { "epoch": 0.0271616, "grad_norm": 0.9207584857940674, "learning_rate": 6.189474094046203e-06, "loss": 2.8848, "step": 244110 }, { "epoch": 0.0271872, "grad_norm": 0.7768929600715637, "learning_rate": 6.1887295834782644e-06, "loss": 3.0119, "step": 244120 }, { "epoch": 0.0272128, "grad_norm": 0.789505660533905, "learning_rate": 6.187985097625404e-06, "loss": 2.7225, "step": 244130 }, { "epoch": 0.0272384, "grad_norm": 0.7890512943267822, "learning_rate": 6.187240636492447e-06, "loss": 2.8389, "step": 244140 }, { "epoch": 0.027264, "grad_norm": 0.781434178352356, "learning_rate": 6.1864962000842185e-06, "loss": 2.812, "step": 244150 }, { "epoch": 0.0272896, "grad_norm": 0.9452897310256958, "learning_rate": 6.185751788405549e-06, "loss": 2.9394, "step": 244160 }, { "epoch": 0.0273152, "grad_norm": 0.8263645172119141, "learning_rate": 6.18500740146127e-06, "loss": 2.7144, "step": 244170 }, { "epoch": 0.0273408, "grad_norm": 1.0049676895141602, "learning_rate": 6.1842630392562e-06, "loss": 2.8505, "step": 244180 }, { "epoch": 0.0273664, "grad_norm": 0.7166780233383179, "learning_rate": 6.18351870179517e-06, "loss": 2.869, "step": 244190 }, { "epoch": 0.027392, "grad_norm": 0.8402950167655945, "learning_rate": 6.182774389083005e-06, "loss": 2.8422, "step": 244200 }, { "epoch": 0.0274176, "grad_norm": 0.844175398349762, "learning_rate": 6.182030101124535e-06, "loss": 3.028, "step": 244210 }, { "epoch": 0.0274432, "grad_norm": 0.739915132522583, "learning_rate": 6.181285837924589e-06, "loss": 2.7162, "step": 244220 }, { "epoch": 0.0274688, "grad_norm": 0.7667557597160339, "learning_rate": 6.180541599487986e-06, "loss": 2.9372, "step": 244230 }, { "epoch": 0.0274944, "grad_norm": 0.7696612477302551, "learning_rate": 6.179797385819556e-06, "loss": 3.0552, "step": 244240 }, { "epoch": 0.02752, "grad_norm": 0.8749727010726929, "learning_rate": 6.1790531969241255e-06, "loss": 2.9728, "step": 244250 }, { "epoch": 0.0275456, "grad_norm": 0.8505927920341492, "learning_rate": 6.17830903280652e-06, "loss": 2.8923, "step": 244260 }, { "epoch": 0.0275712, "grad_norm": 0.8345351815223694, "learning_rate": 6.177564893471564e-06, "loss": 2.8835, "step": 244270 }, { "epoch": 0.0275968, "grad_norm": 0.7374905347824097, "learning_rate": 6.176820778924082e-06, "loss": 2.7674, "step": 244280 }, { "epoch": 0.0276224, "grad_norm": 0.8110021948814392, "learning_rate": 6.176076689168905e-06, "loss": 2.8716, "step": 244290 }, { "epoch": 0.027648, "grad_norm": 0.9081444144248962, "learning_rate": 6.1753326242108525e-06, "loss": 2.9215, "step": 244300 }, { "epoch": 0.0276736, "grad_norm": 1.0405656099319458, "learning_rate": 6.174588584054751e-06, "loss": 2.9367, "step": 244310 }, { "epoch": 0.0276992, "grad_norm": 0.8128254413604736, "learning_rate": 6.173844568705425e-06, "loss": 2.8436, "step": 244320 }, { "epoch": 0.0277248, "grad_norm": 0.9007399082183838, "learning_rate": 6.173100578167704e-06, "loss": 2.8139, "step": 244330 }, { "epoch": 0.0277504, "grad_norm": 0.8597893714904785, "learning_rate": 6.172356612446404e-06, "loss": 2.9073, "step": 244340 }, { "epoch": 0.027776, "grad_norm": 0.7371692657470703, "learning_rate": 6.171612671546355e-06, "loss": 2.7194, "step": 244350 }, { "epoch": 0.0278016, "grad_norm": 0.7197123765945435, "learning_rate": 6.170868755472383e-06, "loss": 2.797, "step": 244360 }, { "epoch": 0.0278272, "grad_norm": 0.7791103720664978, "learning_rate": 6.170124864229308e-06, "loss": 3.0257, "step": 244370 }, { "epoch": 0.0278528, "grad_norm": 0.7671465873718262, "learning_rate": 6.16938099782195e-06, "loss": 2.8393, "step": 244380 }, { "epoch": 0.0278784, "grad_norm": 0.7770779132843018, "learning_rate": 6.168637156255145e-06, "loss": 2.7048, "step": 244390 }, { "epoch": 0.027904, "grad_norm": 0.7867962121963501, "learning_rate": 6.167893339533708e-06, "loss": 2.9028, "step": 244400 }, { "epoch": 0.0279296, "grad_norm": 0.762269139289856, "learning_rate": 6.167149547662463e-06, "loss": 2.9077, "step": 244410 }, { "epoch": 0.0279552, "grad_norm": 0.7286902666091919, "learning_rate": 6.166405780646239e-06, "loss": 2.824, "step": 244420 }, { "epoch": 0.0279808, "grad_norm": 0.7540156841278076, "learning_rate": 6.16566203848985e-06, "loss": 2.8371, "step": 244430 }, { "epoch": 0.0280064, "grad_norm": 0.7744352221488953, "learning_rate": 6.164918321198126e-06, "loss": 2.7657, "step": 244440 }, { "epoch": 0.028032, "grad_norm": 0.7992241382598877, "learning_rate": 6.164174628775886e-06, "loss": 2.982, "step": 244450 }, { "epoch": 0.0280576, "grad_norm": 0.7183424830436707, "learning_rate": 6.163430961227956e-06, "loss": 2.8217, "step": 244460 }, { "epoch": 0.0280832, "grad_norm": 1.0058250427246094, "learning_rate": 6.1626873185591555e-06, "loss": 2.9904, "step": 244470 }, { "epoch": 0.0281088, "grad_norm": 0.7708268761634827, "learning_rate": 6.161943700774307e-06, "loss": 2.8557, "step": 244480 }, { "epoch": 0.0281344, "grad_norm": 1.2722675800323486, "learning_rate": 6.161200107878236e-06, "loss": 2.9442, "step": 244490 }, { "epoch": 0.02816, "grad_norm": 0.7601764798164368, "learning_rate": 6.160456539875759e-06, "loss": 2.8045, "step": 244500 }, { "epoch": 0.0281856, "grad_norm": 0.8149182200431824, "learning_rate": 6.159712996771703e-06, "loss": 2.7858, "step": 244510 }, { "epoch": 0.0282112, "grad_norm": 0.8009532690048218, "learning_rate": 6.158969478570885e-06, "loss": 2.7748, "step": 244520 }, { "epoch": 0.0282368, "grad_norm": 0.8631523251533508, "learning_rate": 6.158225985278132e-06, "loss": 2.8033, "step": 244530 }, { "epoch": 0.0282624, "grad_norm": 0.8060715198516846, "learning_rate": 6.157482516898262e-06, "loss": 2.9345, "step": 244540 }, { "epoch": 0.028288, "grad_norm": 0.8006880283355713, "learning_rate": 6.156739073436093e-06, "loss": 2.9358, "step": 244550 }, { "epoch": 0.0283136, "grad_norm": 0.8561672568321228, "learning_rate": 6.155995654896453e-06, "loss": 2.6371, "step": 244560 }, { "epoch": 0.0283392, "grad_norm": 0.8405375480651855, "learning_rate": 6.155252261284159e-06, "loss": 2.8423, "step": 244570 }, { "epoch": 0.0283648, "grad_norm": 0.7352131605148315, "learning_rate": 6.154508892604034e-06, "loss": 2.983, "step": 244580 }, { "epoch": 0.0283904, "grad_norm": 0.8110713362693787, "learning_rate": 6.153765548860895e-06, "loss": 2.8006, "step": 244590 }, { "epoch": 0.028416, "grad_norm": 0.8148738741874695, "learning_rate": 6.153022230059564e-06, "loss": 3.2383, "step": 244600 }, { "epoch": 0.0284416, "grad_norm": 0.8107163310050964, "learning_rate": 6.152278936204862e-06, "loss": 2.99, "step": 244610 }, { "epoch": 0.0284672, "grad_norm": 0.8235703110694885, "learning_rate": 6.15153566730161e-06, "loss": 2.8497, "step": 244620 }, { "epoch": 0.0284928, "grad_norm": 0.8518666625022888, "learning_rate": 6.1507924233546245e-06, "loss": 2.8947, "step": 244630 }, { "epoch": 0.0285184, "grad_norm": 0.7302799224853516, "learning_rate": 6.150049204368728e-06, "loss": 2.7832, "step": 244640 }, { "epoch": 0.028544, "grad_norm": 0.8067908883094788, "learning_rate": 6.149306010348738e-06, "loss": 3.0749, "step": 244650 }, { "epoch": 0.0285696, "grad_norm": 0.7916924953460693, "learning_rate": 6.148562841299477e-06, "loss": 2.7644, "step": 244660 }, { "epoch": 0.0285952, "grad_norm": 0.7390806674957275, "learning_rate": 6.147819697225761e-06, "loss": 3.0264, "step": 244670 }, { "epoch": 0.0286208, "grad_norm": 0.7847756147384644, "learning_rate": 6.14707657813241e-06, "loss": 2.608, "step": 244680 }, { "epoch": 0.0286464, "grad_norm": 0.7199379801750183, "learning_rate": 6.146333484024247e-06, "loss": 2.8643, "step": 244690 }, { "epoch": 0.028672, "grad_norm": 0.7340168356895447, "learning_rate": 6.145590414906084e-06, "loss": 2.931, "step": 244700 }, { "epoch": 0.0286976, "grad_norm": 0.7089138627052307, "learning_rate": 6.144847370782743e-06, "loss": 2.8202, "step": 244710 }, { "epoch": 0.0287232, "grad_norm": 0.7657575011253357, "learning_rate": 6.144104351659041e-06, "loss": 2.9156, "step": 244720 }, { "epoch": 0.0287488, "grad_norm": 0.8172904253005981, "learning_rate": 6.143361357539797e-06, "loss": 2.9785, "step": 244730 }, { "epoch": 0.0287744, "grad_norm": 0.7705270648002625, "learning_rate": 6.142618388429831e-06, "loss": 2.893, "step": 244740 }, { "epoch": 0.0288, "grad_norm": 0.9600790143013, "learning_rate": 6.141875444333961e-06, "loss": 2.8167, "step": 244750 }, { "epoch": 0.0288256, "grad_norm": 0.9089979529380798, "learning_rate": 6.141132525257003e-06, "loss": 2.8892, "step": 244760 }, { "epoch": 0.0288512, "grad_norm": 0.7625777125358582, "learning_rate": 6.140389631203774e-06, "loss": 3.1763, "step": 244770 }, { "epoch": 0.0288768, "grad_norm": 0.9816999435424805, "learning_rate": 6.139646762179095e-06, "loss": 2.8834, "step": 244780 }, { "epoch": 0.0289024, "grad_norm": 0.7946075201034546, "learning_rate": 6.138903918187778e-06, "loss": 2.8687, "step": 244790 }, { "epoch": 0.028928, "grad_norm": 0.8519230484962463, "learning_rate": 6.138161099234643e-06, "loss": 2.8899, "step": 244800 }, { "epoch": 0.0289536, "grad_norm": 0.8824313282966614, "learning_rate": 6.137418305324507e-06, "loss": 2.9539, "step": 244810 }, { "epoch": 0.0289792, "grad_norm": 0.8247466087341309, "learning_rate": 6.136675536462189e-06, "loss": 2.8418, "step": 244820 }, { "epoch": 0.0290048, "grad_norm": 1.0325531959533691, "learning_rate": 6.135932792652501e-06, "loss": 2.9878, "step": 244830 }, { "epoch": 0.0290304, "grad_norm": 0.8082541227340698, "learning_rate": 6.135190073900262e-06, "loss": 2.939, "step": 244840 }, { "epoch": 0.029056, "grad_norm": 0.8542763590812683, "learning_rate": 6.134447380210289e-06, "loss": 2.7177, "step": 244850 }, { "epoch": 0.0290816, "grad_norm": 0.9528343081474304, "learning_rate": 6.133704711587398e-06, "loss": 2.8106, "step": 244860 }, { "epoch": 0.0291072, "grad_norm": 0.7948898673057556, "learning_rate": 6.132962068036403e-06, "loss": 2.9304, "step": 244870 }, { "epoch": 0.0291328, "grad_norm": 0.9414973258972168, "learning_rate": 6.1322194495621225e-06, "loss": 2.8144, "step": 244880 }, { "epoch": 0.0291584, "grad_norm": 0.8328084945678711, "learning_rate": 6.131476856169371e-06, "loss": 2.8207, "step": 244890 }, { "epoch": 0.029184, "grad_norm": 0.8913581371307373, "learning_rate": 6.1307342878629605e-06, "loss": 2.9801, "step": 244900 }, { "epoch": 0.0292096, "grad_norm": 0.8255535960197449, "learning_rate": 6.1299917446477145e-06, "loss": 2.8624, "step": 244910 }, { "epoch": 0.0292352, "grad_norm": 0.8226228952407837, "learning_rate": 6.1292492265284395e-06, "loss": 3.1515, "step": 244920 }, { "epoch": 0.0292608, "grad_norm": 0.9642396569252014, "learning_rate": 6.128506733509958e-06, "loss": 2.8621, "step": 244930 }, { "epoch": 0.0292864, "grad_norm": 0.8157593011856079, "learning_rate": 6.127764265597078e-06, "loss": 2.9636, "step": 244940 }, { "epoch": 0.029312, "grad_norm": 0.8647530674934387, "learning_rate": 6.1270218227946225e-06, "loss": 2.8081, "step": 244950 }, { "epoch": 0.0293376, "grad_norm": 1.0494823455810547, "learning_rate": 6.1262794051073985e-06, "loss": 2.7388, "step": 244960 }, { "epoch": 0.0293632, "grad_norm": 0.7928919196128845, "learning_rate": 6.125537012540222e-06, "loss": 2.8646, "step": 244970 }, { "epoch": 0.0293888, "grad_norm": 0.8697530031204224, "learning_rate": 6.124794645097911e-06, "loss": 2.9933, "step": 244980 }, { "epoch": 0.0294144, "grad_norm": 0.9309591054916382, "learning_rate": 6.124052302785274e-06, "loss": 2.7883, "step": 244990 }, { "epoch": 0.02944, "grad_norm": 0.8079898953437805, "learning_rate": 6.123309985607129e-06, "loss": 2.9922, "step": 245000 }, { "epoch": 0.0294656, "grad_norm": 0.7769358158111572, "learning_rate": 6.122567693568286e-06, "loss": 2.8421, "step": 245010 }, { "epoch": 0.0294912, "grad_norm": 0.8104208707809448, "learning_rate": 6.121825426673567e-06, "loss": 3.0259, "step": 245020 }, { "epoch": 0.0295168, "grad_norm": 0.7528789639472961, "learning_rate": 6.121083184927774e-06, "loss": 2.9731, "step": 245030 }, { "epoch": 0.0295424, "grad_norm": 0.7330652475357056, "learning_rate": 6.1203409683357275e-06, "loss": 2.8273, "step": 245040 }, { "epoch": 0.029568, "grad_norm": 0.756044328212738, "learning_rate": 6.119598776902238e-06, "loss": 2.7654, "step": 245050 }, { "epoch": 0.0295936, "grad_norm": 0.768613338470459, "learning_rate": 6.11885661063212e-06, "loss": 2.8021, "step": 245060 }, { "epoch": 0.0296192, "grad_norm": 0.7885865569114685, "learning_rate": 6.1181144695301806e-06, "loss": 2.9517, "step": 245070 }, { "epoch": 0.0296448, "grad_norm": 0.9035263061523438, "learning_rate": 6.117372353601244e-06, "loss": 2.9621, "step": 245080 }, { "epoch": 0.0296704, "grad_norm": 0.8001052737236023, "learning_rate": 6.116630262850113e-06, "loss": 3.0711, "step": 245090 }, { "epoch": 0.029696, "grad_norm": 0.7917382121086121, "learning_rate": 6.115888197281603e-06, "loss": 2.9073, "step": 245100 }, { "epoch": 0.0297216, "grad_norm": 1.1369812488555908, "learning_rate": 6.115146156900527e-06, "loss": 2.9267, "step": 245110 }, { "epoch": 0.0297472, "grad_norm": 0.8762112259864807, "learning_rate": 6.114404141711694e-06, "loss": 2.8353, "step": 245120 }, { "epoch": 0.0297728, "grad_norm": 0.8820763826370239, "learning_rate": 6.113662151719918e-06, "loss": 2.7724, "step": 245130 }, { "epoch": 0.0297984, "grad_norm": 0.9536830186843872, "learning_rate": 6.112920186930009e-06, "loss": 2.8349, "step": 245140 }, { "epoch": 0.029824, "grad_norm": 0.8690072298049927, "learning_rate": 6.112178247346783e-06, "loss": 2.933, "step": 245150 }, { "epoch": 0.0298496, "grad_norm": 0.8023616671562195, "learning_rate": 6.111436332975045e-06, "loss": 2.8288, "step": 245160 }, { "epoch": 0.0298752, "grad_norm": 1.7597851753234863, "learning_rate": 6.1106944438196095e-06, "loss": 3.3256, "step": 245170 }, { "epoch": 0.0299008, "grad_norm": 0.8900044560432434, "learning_rate": 6.1099525798852875e-06, "loss": 2.8744, "step": 245180 }, { "epoch": 0.0299264, "grad_norm": 0.8618457913398743, "learning_rate": 6.109210741176887e-06, "loss": 2.7803, "step": 245190 }, { "epoch": 0.029952, "grad_norm": 0.7416192293167114, "learning_rate": 6.108468927699223e-06, "loss": 2.9317, "step": 245200 }, { "epoch": 0.0299776, "grad_norm": 0.8977958559989929, "learning_rate": 6.1077271394571025e-06, "loss": 2.8856, "step": 245210 }, { "epoch": 0.0300032, "grad_norm": 0.8021206855773926, "learning_rate": 6.106985376455339e-06, "loss": 2.8589, "step": 245220 }, { "epoch": 0.0300288, "grad_norm": 0.7544930577278137, "learning_rate": 6.106243638698739e-06, "loss": 2.8131, "step": 245230 }, { "epoch": 0.0300544, "grad_norm": 0.8387850522994995, "learning_rate": 6.1055019261921145e-06, "loss": 2.8613, "step": 245240 }, { "epoch": 0.03008, "grad_norm": 0.8107517957687378, "learning_rate": 6.104760238940271e-06, "loss": 2.7052, "step": 245250 }, { "epoch": 0.0301056, "grad_norm": 0.7433024644851685, "learning_rate": 6.104018576948026e-06, "loss": 3.0361, "step": 245260 }, { "epoch": 0.0301312, "grad_norm": 0.7178364396095276, "learning_rate": 6.103276940220184e-06, "loss": 2.8854, "step": 245270 }, { "epoch": 0.0301568, "grad_norm": 0.7599639892578125, "learning_rate": 6.102535328761559e-06, "loss": 2.8689, "step": 245280 }, { "epoch": 0.0301824, "grad_norm": 0.7608810663223267, "learning_rate": 6.101793742576953e-06, "loss": 2.8958, "step": 245290 }, { "epoch": 0.030208, "grad_norm": 0.76060551404953, "learning_rate": 6.1010521816711796e-06, "loss": 2.9562, "step": 245300 }, { "epoch": 0.0302336, "grad_norm": 0.8110736012458801, "learning_rate": 6.100310646049048e-06, "loss": 2.6692, "step": 245310 }, { "epoch": 0.0302592, "grad_norm": 1.1296870708465576, "learning_rate": 6.099569135715364e-06, "loss": 2.8709, "step": 245320 }, { "epoch": 0.0302848, "grad_norm": 2.5378551483154297, "learning_rate": 6.098827650674936e-06, "loss": 2.8683, "step": 245330 }, { "epoch": 0.0303104, "grad_norm": 0.8119137287139893, "learning_rate": 6.098086190932576e-06, "loss": 3.0068, "step": 245340 }, { "epoch": 0.030336, "grad_norm": 0.7964549660682678, "learning_rate": 6.097344756493091e-06, "loss": 2.8563, "step": 245350 }, { "epoch": 0.0303616, "grad_norm": 0.8481720685958862, "learning_rate": 6.096603347361287e-06, "loss": 2.9164, "step": 245360 }, { "epoch": 0.0303872, "grad_norm": 0.9519334435462952, "learning_rate": 6.095861963541971e-06, "loss": 2.8459, "step": 245370 }, { "epoch": 0.0304128, "grad_norm": 0.7874445915222168, "learning_rate": 6.095120605039957e-06, "loss": 2.7261, "step": 245380 }, { "epoch": 0.0304384, "grad_norm": 0.7819799780845642, "learning_rate": 6.0943792718600446e-06, "loss": 2.8941, "step": 245390 }, { "epoch": 0.030464, "grad_norm": 0.7395832538604736, "learning_rate": 6.093637964007044e-06, "loss": 2.7739, "step": 245400 }, { "epoch": 0.0304896, "grad_norm": 0.7928323745727539, "learning_rate": 6.092896681485765e-06, "loss": 2.7513, "step": 245410 }, { "epoch": 0.0305152, "grad_norm": 0.9296362996101379, "learning_rate": 6.092155424301008e-06, "loss": 2.8667, "step": 245420 }, { "epoch": 0.0305408, "grad_norm": 1.133980393409729, "learning_rate": 6.091414192457588e-06, "loss": 2.8174, "step": 245430 }, { "epoch": 0.0305664, "grad_norm": 0.8534303903579712, "learning_rate": 6.090672985960311e-06, "loss": 2.7547, "step": 245440 }, { "epoch": 0.030592, "grad_norm": 0.7471953630447388, "learning_rate": 6.089931804813978e-06, "loss": 2.7166, "step": 245450 }, { "epoch": 0.0306176, "grad_norm": 2.21710467338562, "learning_rate": 6.089190649023398e-06, "loss": 3.0096, "step": 245460 }, { "epoch": 0.0306432, "grad_norm": 0.8312378525733948, "learning_rate": 6.088449518593376e-06, "loss": 2.8906, "step": 245470 }, { "epoch": 0.0306688, "grad_norm": 0.770573079586029, "learning_rate": 6.087708413528723e-06, "loss": 3.0116, "step": 245480 }, { "epoch": 0.0306944, "grad_norm": 0.7255908846855164, "learning_rate": 6.086967333834238e-06, "loss": 2.891, "step": 245490 }, { "epoch": 0.03072, "grad_norm": 0.7978062033653259, "learning_rate": 6.0862262795147305e-06, "loss": 2.9157, "step": 245500 }, { "epoch": 0.0307456, "grad_norm": 0.7897313237190247, "learning_rate": 6.0854852505750075e-06, "loss": 2.872, "step": 245510 }, { "epoch": 0.0307712, "grad_norm": 0.7691380381584167, "learning_rate": 6.08474424701987e-06, "loss": 2.8432, "step": 245520 }, { "epoch": 0.0307968, "grad_norm": 0.76466965675354, "learning_rate": 6.084003268854126e-06, "loss": 2.7982, "step": 245530 }, { "epoch": 0.0308224, "grad_norm": 0.8897693753242493, "learning_rate": 6.08326231608258e-06, "loss": 2.8612, "step": 245540 }, { "epoch": 0.030848, "grad_norm": 0.804569661617279, "learning_rate": 6.082521388710039e-06, "loss": 2.7756, "step": 245550 }, { "epoch": 0.0308736, "grad_norm": 0.8625764846801758, "learning_rate": 6.0817804867413034e-06, "loss": 2.9189, "step": 245560 }, { "epoch": 0.0308992, "grad_norm": 0.919986367225647, "learning_rate": 6.08103961018118e-06, "loss": 2.9094, "step": 245570 }, { "epoch": 0.0309248, "grad_norm": 0.7651991248130798, "learning_rate": 6.080298759034475e-06, "loss": 2.9368, "step": 245580 }, { "epoch": 0.0309504, "grad_norm": 0.7460322380065918, "learning_rate": 6.079557933305988e-06, "loss": 3.1489, "step": 245590 }, { "epoch": 0.030976, "grad_norm": 0.9205493927001953, "learning_rate": 6.0788171330005255e-06, "loss": 2.9771, "step": 245600 }, { "epoch": 0.0310016, "grad_norm": 0.7854907512664795, "learning_rate": 6.078076358122893e-06, "loss": 2.9096, "step": 245610 }, { "epoch": 0.0310272, "grad_norm": 0.7724418640136719, "learning_rate": 6.077335608677893e-06, "loss": 2.6918, "step": 245620 }, { "epoch": 0.0310528, "grad_norm": 0.8937361836433411, "learning_rate": 6.076594884670329e-06, "loss": 2.7401, "step": 245630 }, { "epoch": 0.0310784, "grad_norm": 0.8071098923683167, "learning_rate": 6.075854186105008e-06, "loss": 2.8751, "step": 245640 }, { "epoch": 0.031104, "grad_norm": 0.7564698457717896, "learning_rate": 6.075113512986725e-06, "loss": 2.7493, "step": 245650 }, { "epoch": 0.0311296, "grad_norm": 0.7900705337524414, "learning_rate": 6.07437286532029e-06, "loss": 2.8848, "step": 245660 }, { "epoch": 0.0311552, "grad_norm": 0.8937413692474365, "learning_rate": 6.073632243110505e-06, "loss": 2.7746, "step": 245670 }, { "epoch": 0.0311808, "grad_norm": 0.7785166501998901, "learning_rate": 6.072891646362169e-06, "loss": 2.6835, "step": 245680 }, { "epoch": 0.0312064, "grad_norm": 0.8363127708435059, "learning_rate": 6.072151075080086e-06, "loss": 2.9721, "step": 245690 }, { "epoch": 0.031232, "grad_norm": 0.8104649186134338, "learning_rate": 6.071410529269062e-06, "loss": 2.9379, "step": 245700 }, { "epoch": 0.0312576, "grad_norm": 0.8688958287239075, "learning_rate": 6.070670008933897e-06, "loss": 2.9077, "step": 245710 }, { "epoch": 0.0312832, "grad_norm": 0.7496514916419983, "learning_rate": 6.06992951407939e-06, "loss": 2.9769, "step": 245720 }, { "epoch": 0.0313088, "grad_norm": 0.8157836198806763, "learning_rate": 6.069189044710346e-06, "loss": 3.0154, "step": 245730 }, { "epoch": 0.0313344, "grad_norm": 0.8062602281570435, "learning_rate": 6.0684486008315664e-06, "loss": 2.7057, "step": 245740 }, { "epoch": 0.03136, "grad_norm": 0.9342703223228455, "learning_rate": 6.067708182447856e-06, "loss": 3.1447, "step": 245750 }, { "epoch": 0.0313856, "grad_norm": 0.8227588534355164, "learning_rate": 6.06696778956401e-06, "loss": 2.7654, "step": 245760 }, { "epoch": 0.0314112, "grad_norm": 0.9088490605354309, "learning_rate": 6.066227422184831e-06, "loss": 2.9141, "step": 245770 }, { "epoch": 0.0314368, "grad_norm": 0.8285780549049377, "learning_rate": 6.065487080315124e-06, "loss": 2.761, "step": 245780 }, { "epoch": 0.0314624, "grad_norm": 0.7135452628135681, "learning_rate": 6.064746763959686e-06, "loss": 2.8036, "step": 245790 }, { "epoch": 0.031488, "grad_norm": 0.7637873291969299, "learning_rate": 6.064006473123323e-06, "loss": 2.8909, "step": 245800 }, { "epoch": 0.0315136, "grad_norm": 0.8168124556541443, "learning_rate": 6.06326620781083e-06, "loss": 2.9167, "step": 245810 }, { "epoch": 0.0315392, "grad_norm": 2.6520016193389893, "learning_rate": 6.062525968027008e-06, "loss": 2.8006, "step": 245820 }, { "epoch": 0.0315648, "grad_norm": 0.7725346088409424, "learning_rate": 6.0617857537766585e-06, "loss": 2.9306, "step": 245830 }, { "epoch": 0.0315904, "grad_norm": 1.5555920600891113, "learning_rate": 6.061045565064585e-06, "loss": 2.8958, "step": 245840 }, { "epoch": 0.031616, "grad_norm": 0.8363433480262756, "learning_rate": 6.060305401895582e-06, "loss": 2.9062, "step": 245850 }, { "epoch": 0.0316416, "grad_norm": 0.7412242293357849, "learning_rate": 6.059565264274452e-06, "loss": 2.8147, "step": 245860 }, { "epoch": 0.0316672, "grad_norm": 0.7771860957145691, "learning_rate": 6.058825152205995e-06, "loss": 2.9851, "step": 245870 }, { "epoch": 0.0316928, "grad_norm": 1.0832408666610718, "learning_rate": 6.0580850656950075e-06, "loss": 3.0349, "step": 245880 }, { "epoch": 0.0317184, "grad_norm": 0.9291951656341553, "learning_rate": 6.057345004746291e-06, "loss": 3.1661, "step": 245890 }, { "epoch": 0.031744, "grad_norm": 0.8917133808135986, "learning_rate": 6.056604969364644e-06, "loss": 3.1208, "step": 245900 }, { "epoch": 0.0317696, "grad_norm": 0.8343630433082581, "learning_rate": 6.055864959554868e-06, "loss": 2.8649, "step": 245910 }, { "epoch": 0.0317952, "grad_norm": 0.8445956110954285, "learning_rate": 6.055124975321758e-06, "loss": 2.9187, "step": 245920 }, { "epoch": 0.0318208, "grad_norm": 0.9011286497116089, "learning_rate": 6.054385016670112e-06, "loss": 3.3218, "step": 245930 }, { "epoch": 0.0318464, "grad_norm": 0.9045751690864563, "learning_rate": 6.053645083604731e-06, "loss": 2.723, "step": 245940 }, { "epoch": 0.031872, "grad_norm": 0.6955339908599854, "learning_rate": 6.052905176130414e-06, "loss": 2.8651, "step": 245950 }, { "epoch": 0.0318976, "grad_norm": 0.9008058309555054, "learning_rate": 6.0521652942519575e-06, "loss": 2.8076, "step": 245960 }, { "epoch": 0.0319232, "grad_norm": 0.7808674573898315, "learning_rate": 6.0514254379741625e-06, "loss": 2.6632, "step": 245970 }, { "epoch": 0.0319488, "grad_norm": 0.7861518859863281, "learning_rate": 6.050685607301822e-06, "loss": 2.8466, "step": 245980 }, { "epoch": 0.0319744, "grad_norm": 0.7612864375114441, "learning_rate": 6.049945802239735e-06, "loss": 3.0668, "step": 245990 }, { "epoch": 0.032, "grad_norm": 0.9385107159614563, "learning_rate": 6.049206022792702e-06, "loss": 2.7673, "step": 246000 }, { "epoch": 0.0320256, "grad_norm": 2.3374388217926025, "learning_rate": 6.048466268965516e-06, "loss": 2.7371, "step": 246010 }, { "epoch": 0.0320512, "grad_norm": 0.9252533316612244, "learning_rate": 6.047726540762977e-06, "loss": 3.0945, "step": 246020 }, { "epoch": 0.0320768, "grad_norm": 0.7949145436286926, "learning_rate": 6.046986838189881e-06, "loss": 2.8635, "step": 246030 }, { "epoch": 0.0321024, "grad_norm": 1.2697454690933228, "learning_rate": 6.046247161251025e-06, "loss": 2.996, "step": 246040 }, { "epoch": 0.032128, "grad_norm": 1.1836051940917969, "learning_rate": 6.045507509951205e-06, "loss": 3.1328, "step": 246050 }, { "epoch": 0.0321536, "grad_norm": 0.7501325607299805, "learning_rate": 6.044767884295217e-06, "loss": 2.9064, "step": 246060 }, { "epoch": 0.0321792, "grad_norm": 0.7231862545013428, "learning_rate": 6.04402828428786e-06, "loss": 2.836, "step": 246070 }, { "epoch": 0.0322048, "grad_norm": 0.7812439799308777, "learning_rate": 6.0432887099339265e-06, "loss": 2.9309, "step": 246080 }, { "epoch": 0.0322304, "grad_norm": 0.8124454617500305, "learning_rate": 6.042549161238214e-06, "loss": 3.0771, "step": 246090 }, { "epoch": 0.032256, "grad_norm": 1.0763076543807983, "learning_rate": 6.041809638205519e-06, "loss": 2.9562, "step": 246100 }, { "epoch": 0.0322816, "grad_norm": 0.8909285664558411, "learning_rate": 6.041070140840638e-06, "loss": 2.7889, "step": 246110 }, { "epoch": 0.0323072, "grad_norm": 0.7715656161308289, "learning_rate": 6.04033066914836e-06, "loss": 2.8537, "step": 246120 }, { "epoch": 0.0323328, "grad_norm": 0.8061790466308594, "learning_rate": 6.03959122313349e-06, "loss": 2.9793, "step": 246130 }, { "epoch": 0.0323584, "grad_norm": 0.7822531461715698, "learning_rate": 6.038851802800816e-06, "loss": 2.981, "step": 246140 }, { "epoch": 0.032384, "grad_norm": 0.7643556594848633, "learning_rate": 6.038112408155137e-06, "loss": 2.8012, "step": 246150 }, { "epoch": 0.0324096, "grad_norm": 0.79597008228302, "learning_rate": 6.037373039201244e-06, "loss": 2.7534, "step": 246160 }, { "epoch": 0.0324352, "grad_norm": 1.036983609199524, "learning_rate": 6.036633695943937e-06, "loss": 2.9119, "step": 246170 }, { "epoch": 0.0324608, "grad_norm": 0.789993941783905, "learning_rate": 6.035894378388005e-06, "loss": 2.7756, "step": 246180 }, { "epoch": 0.0324864, "grad_norm": 0.7456997036933899, "learning_rate": 6.035155086538245e-06, "loss": 2.927, "step": 246190 }, { "epoch": 0.032512, "grad_norm": 0.7832419872283936, "learning_rate": 6.034415820399451e-06, "loss": 2.7351, "step": 246200 }, { "epoch": 0.0325376, "grad_norm": 0.8412899971008301, "learning_rate": 6.033676579976415e-06, "loss": 2.8488, "step": 246210 }, { "epoch": 0.0325632, "grad_norm": 0.7416368126869202, "learning_rate": 6.032937365273934e-06, "loss": 2.8595, "step": 246220 }, { "epoch": 0.0325888, "grad_norm": 0.6930353045463562, "learning_rate": 6.0321981762967976e-06, "loss": 2.8757, "step": 246230 }, { "epoch": 0.0326144, "grad_norm": 0.8234835267066956, "learning_rate": 6.031459013049806e-06, "loss": 2.9222, "step": 246240 }, { "epoch": 0.03264, "grad_norm": 0.8423734903335571, "learning_rate": 6.030719875537743e-06, "loss": 3.1217, "step": 246250 }, { "epoch": 0.0326656, "grad_norm": 0.9330283999443054, "learning_rate": 6.029980763765408e-06, "loss": 2.8501, "step": 246260 }, { "epoch": 0.0326912, "grad_norm": 0.8034653663635254, "learning_rate": 6.029241677737596e-06, "loss": 2.7388, "step": 246270 }, { "epoch": 0.0327168, "grad_norm": 0.7233569622039795, "learning_rate": 6.028502617459093e-06, "loss": 2.8225, "step": 246280 }, { "epoch": 0.0327424, "grad_norm": 0.874870240688324, "learning_rate": 6.027763582934693e-06, "loss": 2.8464, "step": 246290 }, { "epoch": 0.032768, "grad_norm": 0.9181732535362244, "learning_rate": 6.027024574169193e-06, "loss": 2.8118, "step": 246300 }, { "epoch": 0.0327936, "grad_norm": 0.8439907431602478, "learning_rate": 6.026285591167382e-06, "loss": 2.7626, "step": 246310 }, { "epoch": 0.0328192, "grad_norm": 0.7577452659606934, "learning_rate": 6.025546633934053e-06, "loss": 2.7683, "step": 246320 }, { "epoch": 0.0328448, "grad_norm": 0.8467839956283569, "learning_rate": 6.024807702474e-06, "loss": 2.8561, "step": 246330 }, { "epoch": 0.0328704, "grad_norm": 1.9068273305892944, "learning_rate": 6.024068796792011e-06, "loss": 3.1841, "step": 246340 }, { "epoch": 0.032896, "grad_norm": 0.7853800654411316, "learning_rate": 6.0233299168928795e-06, "loss": 2.9402, "step": 246350 }, { "epoch": 0.0329216, "grad_norm": 0.7806881070137024, "learning_rate": 6.022591062781395e-06, "loss": 2.776, "step": 246360 }, { "epoch": 0.0329472, "grad_norm": 1.0421983003616333, "learning_rate": 6.021852234462353e-06, "loss": 3.0555, "step": 246370 }, { "epoch": 0.0329728, "grad_norm": 0.8857645392417908, "learning_rate": 6.02111343194054e-06, "loss": 2.7868, "step": 246380 }, { "epoch": 0.0329984, "grad_norm": 3.0608527660369873, "learning_rate": 6.0203746552207496e-06, "loss": 2.9747, "step": 246390 }, { "epoch": 0.033024, "grad_norm": 0.840213418006897, "learning_rate": 6.019635904307774e-06, "loss": 2.6173, "step": 246400 }, { "epoch": 0.0330496, "grad_norm": 0.8766409158706665, "learning_rate": 6.0188971792064e-06, "loss": 2.7971, "step": 246410 }, { "epoch": 0.0330752, "grad_norm": 0.7102655172348022, "learning_rate": 6.018158479921419e-06, "loss": 2.7589, "step": 246420 }, { "epoch": 0.0331008, "grad_norm": 0.7880553603172302, "learning_rate": 6.017419806457621e-06, "loss": 3.0051, "step": 246430 }, { "epoch": 0.0331264, "grad_norm": 0.8702762722969055, "learning_rate": 6.016681158819801e-06, "loss": 2.9396, "step": 246440 }, { "epoch": 0.033152, "grad_norm": 0.7803434729576111, "learning_rate": 6.015942537012742e-06, "loss": 2.9002, "step": 246450 }, { "epoch": 0.0331776, "grad_norm": 0.8030877113342285, "learning_rate": 6.015203941041236e-06, "loss": 2.8552, "step": 246460 }, { "epoch": 0.0332032, "grad_norm": 0.9481542706489563, "learning_rate": 6.0144653709100735e-06, "loss": 3.0013, "step": 246470 }, { "epoch": 0.0332288, "grad_norm": 0.7835856676101685, "learning_rate": 6.013726826624044e-06, "loss": 2.9563, "step": 246480 }, { "epoch": 0.0332544, "grad_norm": 0.7853192687034607, "learning_rate": 6.012988308187939e-06, "loss": 2.7112, "step": 246490 }, { "epoch": 0.03328, "grad_norm": 0.9111436605453491, "learning_rate": 6.012249815606545e-06, "loss": 2.7582, "step": 246500 }, { "epoch": 0.0333056, "grad_norm": 0.8502109050750732, "learning_rate": 6.011511348884649e-06, "loss": 2.9463, "step": 246510 }, { "epoch": 0.0333312, "grad_norm": 1.0670058727264404, "learning_rate": 6.010772908027043e-06, "loss": 2.6315, "step": 246520 }, { "epoch": 0.0333568, "grad_norm": 0.8260675668716431, "learning_rate": 6.010034493038516e-06, "loss": 2.7553, "step": 246530 }, { "epoch": 0.0333824, "grad_norm": 0.969069242477417, "learning_rate": 6.009296103923851e-06, "loss": 2.9613, "step": 246540 }, { "epoch": 0.033408, "grad_norm": 1.2403103113174438, "learning_rate": 6.008557740687844e-06, "loss": 2.7283, "step": 246550 }, { "epoch": 0.0334336, "grad_norm": 0.9412757754325867, "learning_rate": 6.00781940333528e-06, "loss": 2.874, "step": 246560 }, { "epoch": 0.0334592, "grad_norm": 0.8473026156425476, "learning_rate": 6.007081091870944e-06, "loss": 3.1361, "step": 246570 }, { "epoch": 0.0334848, "grad_norm": 1.1876633167266846, "learning_rate": 6.006342806299624e-06, "loss": 2.7203, "step": 246580 }, { "epoch": 0.0335104, "grad_norm": 0.7716090083122253, "learning_rate": 6.005604546626111e-06, "loss": 2.8084, "step": 246590 }, { "epoch": 0.033536, "grad_norm": 0.8366588950157166, "learning_rate": 6.004866312855193e-06, "loss": 2.8146, "step": 246600 }, { "epoch": 0.0335616, "grad_norm": 0.7199811935424805, "learning_rate": 6.004128104991653e-06, "loss": 2.681, "step": 246610 }, { "epoch": 0.0335872, "grad_norm": 0.8413639664649963, "learning_rate": 6.003389923040281e-06, "loss": 3.0308, "step": 246620 }, { "epoch": 0.0336128, "grad_norm": 0.964239239692688, "learning_rate": 6.0026517670058625e-06, "loss": 2.9113, "step": 246630 }, { "epoch": 0.0336384, "grad_norm": 0.7928479313850403, "learning_rate": 6.001913636893182e-06, "loss": 2.7929, "step": 246640 }, { "epoch": 0.033664, "grad_norm": 0.9333795309066772, "learning_rate": 6.001175532707033e-06, "loss": 3.0334, "step": 246650 }, { "epoch": 0.0336896, "grad_norm": 0.851218044757843, "learning_rate": 6.000437454452198e-06, "loss": 2.9328, "step": 246660 }, { "epoch": 0.0337152, "grad_norm": 0.7932664155960083, "learning_rate": 5.999699402133463e-06, "loss": 2.9471, "step": 246670 }, { "epoch": 0.0337408, "grad_norm": 0.807580828666687, "learning_rate": 5.998961375755612e-06, "loss": 2.9952, "step": 246680 }, { "epoch": 0.0337664, "grad_norm": 0.8300000429153442, "learning_rate": 5.998223375323437e-06, "loss": 2.854, "step": 246690 }, { "epoch": 0.033792, "grad_norm": 0.7746009230613708, "learning_rate": 5.997485400841717e-06, "loss": 2.9701, "step": 246700 }, { "epoch": 0.0338176, "grad_norm": 0.7725432515144348, "learning_rate": 5.996747452315241e-06, "loss": 3.0154, "step": 246710 }, { "epoch": 0.0338432, "grad_norm": 0.906450092792511, "learning_rate": 5.996009529748794e-06, "loss": 3.1467, "step": 246720 }, { "epoch": 0.0338688, "grad_norm": 0.8846757411956787, "learning_rate": 5.995271633147162e-06, "loss": 2.9201, "step": 246730 }, { "epoch": 0.0338944, "grad_norm": 1.0040481090545654, "learning_rate": 5.9945337625151276e-06, "loss": 2.9621, "step": 246740 }, { "epoch": 0.03392, "grad_norm": 0.8431209921836853, "learning_rate": 5.993795917857478e-06, "loss": 2.6484, "step": 246750 }, { "epoch": 0.0339456, "grad_norm": 1.1502971649169922, "learning_rate": 5.993058099178999e-06, "loss": 3.018, "step": 246760 }, { "epoch": 0.0339712, "grad_norm": 0.8835542798042297, "learning_rate": 5.992320306484469e-06, "loss": 2.9286, "step": 246770 }, { "epoch": 0.0339968, "grad_norm": 0.9262946248054504, "learning_rate": 5.99158253977868e-06, "loss": 2.8672, "step": 246780 }, { "epoch": 0.0340224, "grad_norm": 0.7509827613830566, "learning_rate": 5.990844799066411e-06, "loss": 2.9927, "step": 246790 }, { "epoch": 0.034048, "grad_norm": 0.7487938404083252, "learning_rate": 5.99010708435245e-06, "loss": 2.7964, "step": 246800 }, { "epoch": 0.0340736, "grad_norm": 0.7628417611122131, "learning_rate": 5.989369395641573e-06, "loss": 2.7149, "step": 246810 }, { "epoch": 0.0340992, "grad_norm": 0.95144122838974, "learning_rate": 5.988631732938577e-06, "loss": 2.8851, "step": 246820 }, { "epoch": 0.0341248, "grad_norm": 1.2254323959350586, "learning_rate": 5.9878940962482345e-06, "loss": 3.11, "step": 246830 }, { "epoch": 0.0341504, "grad_norm": 0.7892371416091919, "learning_rate": 5.987156485575333e-06, "loss": 2.9726, "step": 246840 }, { "epoch": 0.034176, "grad_norm": 0.7215209603309631, "learning_rate": 5.986418900924655e-06, "loss": 2.9552, "step": 246850 }, { "epoch": 0.0342016, "grad_norm": 0.6972098350524902, "learning_rate": 5.985681342300986e-06, "loss": 2.5954, "step": 246860 }, { "epoch": 0.0342272, "grad_norm": 0.9423560500144958, "learning_rate": 5.984943809709105e-06, "loss": 2.6058, "step": 246870 }, { "epoch": 0.0342528, "grad_norm": 0.9035907983779907, "learning_rate": 5.984206303153795e-06, "loss": 2.8664, "step": 246880 }, { "epoch": 0.0342784, "grad_norm": 0.9086513519287109, "learning_rate": 5.983468822639843e-06, "loss": 3.1827, "step": 246890 }, { "epoch": 0.034304, "grad_norm": 0.7788592576980591, "learning_rate": 5.982731368172025e-06, "loss": 2.8352, "step": 246900 }, { "epoch": 0.0343296, "grad_norm": 0.7921188473701477, "learning_rate": 5.981993939755129e-06, "loss": 2.7957, "step": 246910 }, { "epoch": 0.0343552, "grad_norm": 0.8703538179397583, "learning_rate": 5.981256537393931e-06, "loss": 2.8081, "step": 246920 }, { "epoch": 0.0343808, "grad_norm": 0.8346367478370667, "learning_rate": 5.980519161093221e-06, "loss": 2.7945, "step": 246930 }, { "epoch": 0.0344064, "grad_norm": 0.8272809386253357, "learning_rate": 5.979781810857773e-06, "loss": 2.8376, "step": 246940 }, { "epoch": 0.034432, "grad_norm": 0.9259575009346008, "learning_rate": 5.979044486692371e-06, "loss": 2.9416, "step": 246950 }, { "epoch": 0.0344576, "grad_norm": 2.0293092727661133, "learning_rate": 5.978307188601799e-06, "loss": 3.2717, "step": 246960 }, { "epoch": 0.0344832, "grad_norm": 1.0279827117919922, "learning_rate": 5.977569916590835e-06, "loss": 2.9013, "step": 246970 }, { "epoch": 0.0345088, "grad_norm": 0.8528785109519958, "learning_rate": 5.976832670664261e-06, "loss": 2.7005, "step": 246980 }, { "epoch": 0.0345344, "grad_norm": 0.8198003172874451, "learning_rate": 5.976095450826855e-06, "loss": 2.9131, "step": 246990 }, { "epoch": 0.03456, "grad_norm": 0.7660831809043884, "learning_rate": 5.975358257083403e-06, "loss": 2.8208, "step": 247000 }, { "epoch": 0.0345856, "grad_norm": 0.7767842411994934, "learning_rate": 5.9746210894386835e-06, "loss": 2.7721, "step": 247010 }, { "epoch": 0.0346112, "grad_norm": 0.6980008482933044, "learning_rate": 5.973883947897479e-06, "loss": 2.8451, "step": 247020 }, { "epoch": 0.0346368, "grad_norm": 0.8424957990646362, "learning_rate": 5.973146832464562e-06, "loss": 3.0819, "step": 247030 }, { "epoch": 0.0346624, "grad_norm": 0.8185490369796753, "learning_rate": 5.972409743144721e-06, "loss": 2.7848, "step": 247040 }, { "epoch": 0.034688, "grad_norm": 0.761040985584259, "learning_rate": 5.97167267994273e-06, "loss": 2.7593, "step": 247050 }, { "epoch": 0.0347136, "grad_norm": 0.9273964166641235, "learning_rate": 5.970935642863375e-06, "loss": 2.6745, "step": 247060 }, { "epoch": 0.0347392, "grad_norm": 1.0204498767852783, "learning_rate": 5.970198631911428e-06, "loss": 3.0275, "step": 247070 }, { "epoch": 0.0347648, "grad_norm": 0.8782755732536316, "learning_rate": 5.969461647091673e-06, "loss": 3.031, "step": 247080 }, { "epoch": 0.0347904, "grad_norm": 0.9004955291748047, "learning_rate": 5.96872468840889e-06, "loss": 3.0667, "step": 247090 }, { "epoch": 0.034816, "grad_norm": 0.898695707321167, "learning_rate": 5.967987755867853e-06, "loss": 2.9627, "step": 247100 }, { "epoch": 0.0348416, "grad_norm": 0.8092656135559082, "learning_rate": 5.967250849473345e-06, "loss": 2.8632, "step": 247110 }, { "epoch": 0.0348672, "grad_norm": 0.8496132493019104, "learning_rate": 5.966513969230144e-06, "loss": 2.725, "step": 247120 }, { "epoch": 0.0348928, "grad_norm": 0.7766447067260742, "learning_rate": 5.965777115143028e-06, "loss": 2.9479, "step": 247130 }, { "epoch": 0.0349184, "grad_norm": 0.716416597366333, "learning_rate": 5.965040287216774e-06, "loss": 2.9536, "step": 247140 }, { "epoch": 0.034944, "grad_norm": 0.7781287431716919, "learning_rate": 5.964303485456161e-06, "loss": 2.8927, "step": 247150 }, { "epoch": 0.0349696, "grad_norm": 0.7343887090682983, "learning_rate": 5.963566709865967e-06, "loss": 2.7385, "step": 247160 }, { "epoch": 0.0349952, "grad_norm": 0.8615061044692993, "learning_rate": 5.96282996045097e-06, "loss": 2.8078, "step": 247170 }, { "epoch": 0.0350208, "grad_norm": 0.7548163533210754, "learning_rate": 5.962093237215951e-06, "loss": 2.8705, "step": 247180 }, { "epoch": 0.0350464, "grad_norm": 0.7595424652099609, "learning_rate": 5.961356540165681e-06, "loss": 2.9169, "step": 247190 }, { "epoch": 0.035072, "grad_norm": 0.7409695386886597, "learning_rate": 5.960619869304942e-06, "loss": 3.0576, "step": 247200 }, { "epoch": 0.0350976, "grad_norm": 0.7649259567260742, "learning_rate": 5.95988322463851e-06, "loss": 2.668, "step": 247210 }, { "epoch": 0.0351232, "grad_norm": 0.853929340839386, "learning_rate": 5.959146606171163e-06, "loss": 2.8584, "step": 247220 }, { "epoch": 0.0351488, "grad_norm": 0.7807649970054626, "learning_rate": 5.958410013907671e-06, "loss": 2.9555, "step": 247230 }, { "epoch": 0.0351744, "grad_norm": 0.9689576029777527, "learning_rate": 5.957673447852821e-06, "loss": 2.8136, "step": 247240 }, { "epoch": 0.0352, "grad_norm": 0.7900054454803467, "learning_rate": 5.956936908011382e-06, "loss": 2.9578, "step": 247250 }, { "epoch": 0.0352256, "grad_norm": 0.7952377200126648, "learning_rate": 5.956200394388134e-06, "loss": 2.7045, "step": 247260 }, { "epoch": 0.0352512, "grad_norm": 0.8354195952415466, "learning_rate": 5.955463906987851e-06, "loss": 2.7269, "step": 247270 }, { "epoch": 0.0352768, "grad_norm": 1.2499696016311646, "learning_rate": 5.954727445815309e-06, "loss": 2.9701, "step": 247280 }, { "epoch": 0.0353024, "grad_norm": 0.9525722861289978, "learning_rate": 5.953991010875286e-06, "loss": 2.9733, "step": 247290 }, { "epoch": 0.035328, "grad_norm": 0.8004167675971985, "learning_rate": 5.9532546021725555e-06, "loss": 2.9454, "step": 247300 }, { "epoch": 0.0353536, "grad_norm": 0.7653794288635254, "learning_rate": 5.9525182197118925e-06, "loss": 2.8996, "step": 247310 }, { "epoch": 0.0353792, "grad_norm": 1.4333072900772095, "learning_rate": 5.951781863498073e-06, "loss": 2.8524, "step": 247320 }, { "epoch": 0.0354048, "grad_norm": 0.7696841955184937, "learning_rate": 5.951045533535874e-06, "loss": 2.8063, "step": 247330 }, { "epoch": 0.0354304, "grad_norm": 0.7916185259819031, "learning_rate": 5.950309229830063e-06, "loss": 2.7148, "step": 247340 }, { "epoch": 0.035456, "grad_norm": 0.9148409366607666, "learning_rate": 5.949572952385427e-06, "loss": 2.8094, "step": 247350 }, { "epoch": 0.0354816, "grad_norm": 0.8641766309738159, "learning_rate": 5.948836701206731e-06, "loss": 3.0126, "step": 247360 }, { "epoch": 0.0355072, "grad_norm": 0.7758622169494629, "learning_rate": 5.948100476298753e-06, "loss": 3.129, "step": 247370 }, { "epoch": 0.0355328, "grad_norm": 0.767149031162262, "learning_rate": 5.9473642776662676e-06, "loss": 2.9395, "step": 247380 }, { "epoch": 0.0355584, "grad_norm": 0.8407775163650513, "learning_rate": 5.946628105314047e-06, "loss": 2.7541, "step": 247390 }, { "epoch": 0.035584, "grad_norm": 0.9411744475364685, "learning_rate": 5.9458919592468655e-06, "loss": 2.7762, "step": 247400 }, { "epoch": 0.0356096, "grad_norm": 0.8227999806404114, "learning_rate": 5.945155839469497e-06, "loss": 2.7607, "step": 247410 }, { "epoch": 0.0356352, "grad_norm": 0.9987416863441467, "learning_rate": 5.944419745986716e-06, "loss": 2.7973, "step": 247420 }, { "epoch": 0.0356608, "grad_norm": 0.795612096786499, "learning_rate": 5.9436836788032945e-06, "loss": 2.716, "step": 247430 }, { "epoch": 0.0356864, "grad_norm": 1.0542213916778564, "learning_rate": 5.942947637924006e-06, "loss": 2.8548, "step": 247440 }, { "epoch": 0.035712, "grad_norm": 0.7365071773529053, "learning_rate": 5.942211623353624e-06, "loss": 2.7719, "step": 247450 }, { "epoch": 0.0357376, "grad_norm": 0.9248928427696228, "learning_rate": 5.941475635096924e-06, "loss": 2.9422, "step": 247460 }, { "epoch": 0.0357632, "grad_norm": 0.7660371661186218, "learning_rate": 5.940739673158673e-06, "loss": 2.8398, "step": 247470 }, { "epoch": 0.0357888, "grad_norm": 0.8462291359901428, "learning_rate": 5.940003737543646e-06, "loss": 2.8471, "step": 247480 }, { "epoch": 0.0358144, "grad_norm": 1.7705622911453247, "learning_rate": 5.939267828256619e-06, "loss": 2.8631, "step": 247490 }, { "epoch": 0.03584, "grad_norm": 0.7752551436424255, "learning_rate": 5.938531945302357e-06, "loss": 2.8584, "step": 247500 }, { "epoch": 0.0358656, "grad_norm": 0.7953210473060608, "learning_rate": 5.9377960886856346e-06, "loss": 2.7993, "step": 247510 }, { "epoch": 0.0358912, "grad_norm": 0.7399654984474182, "learning_rate": 5.937060258411228e-06, "loss": 2.9965, "step": 247520 }, { "epoch": 0.0359168, "grad_norm": 0.9060981273651123, "learning_rate": 5.936324454483905e-06, "loss": 2.9888, "step": 247530 }, { "epoch": 0.0359424, "grad_norm": 0.8932938575744629, "learning_rate": 5.935588676908438e-06, "loss": 2.8641, "step": 247540 }, { "epoch": 0.035968, "grad_norm": 0.8853943943977356, "learning_rate": 5.9348529256896e-06, "loss": 2.987, "step": 247550 }, { "epoch": 0.0359936, "grad_norm": 0.8386957049369812, "learning_rate": 5.934117200832159e-06, "loss": 2.8111, "step": 247560 }, { "epoch": 0.0360192, "grad_norm": 0.8256089687347412, "learning_rate": 5.9333815023408865e-06, "loss": 2.9505, "step": 247570 }, { "epoch": 0.0360448, "grad_norm": 0.7072153687477112, "learning_rate": 5.932645830220558e-06, "loss": 2.9545, "step": 247580 }, { "epoch": 0.0360704, "grad_norm": 0.756129264831543, "learning_rate": 5.931910184475937e-06, "loss": 2.9609, "step": 247590 }, { "epoch": 0.036096, "grad_norm": 0.8684787154197693, "learning_rate": 5.9311745651117965e-06, "loss": 2.8972, "step": 247600 }, { "epoch": 0.0361216, "grad_norm": 0.8051882982254028, "learning_rate": 5.930438972132909e-06, "loss": 2.7647, "step": 247610 }, { "epoch": 0.0361472, "grad_norm": 0.8188988566398621, "learning_rate": 5.929703405544045e-06, "loss": 3.1113, "step": 247620 }, { "epoch": 0.0361728, "grad_norm": 0.8099727034568787, "learning_rate": 5.92896786534997e-06, "loss": 2.8435, "step": 247630 }, { "epoch": 0.0361984, "grad_norm": 0.8014773726463318, "learning_rate": 5.928232351555457e-06, "loss": 2.9058, "step": 247640 }, { "epoch": 0.036224, "grad_norm": 0.7489780783653259, "learning_rate": 5.9274968641652765e-06, "loss": 2.8774, "step": 247650 }, { "epoch": 0.0362496, "grad_norm": 0.8060518503189087, "learning_rate": 5.926761403184195e-06, "loss": 3.3598, "step": 247660 }, { "epoch": 0.0362752, "grad_norm": 0.741709291934967, "learning_rate": 5.926025968616983e-06, "loss": 3.0603, "step": 247670 }, { "epoch": 0.0363008, "grad_norm": 0.7959344387054443, "learning_rate": 5.925290560468409e-06, "loss": 2.809, "step": 247680 }, { "epoch": 0.0363264, "grad_norm": 1.0178065299987793, "learning_rate": 5.924555178743243e-06, "loss": 2.9141, "step": 247690 }, { "epoch": 0.036352, "grad_norm": 0.8404181599617004, "learning_rate": 5.923819823446254e-06, "loss": 2.8206, "step": 247700 }, { "epoch": 0.0363776, "grad_norm": 1.0268067121505737, "learning_rate": 5.923084494582212e-06, "loss": 2.8379, "step": 247710 }, { "epoch": 0.0364032, "grad_norm": 0.8324753046035767, "learning_rate": 5.922349192155883e-06, "loss": 2.8598, "step": 247720 }, { "epoch": 0.0364288, "grad_norm": 0.8106198906898499, "learning_rate": 5.921613916172034e-06, "loss": 2.8445, "step": 247730 }, { "epoch": 0.0364544, "grad_norm": 0.7925618290901184, "learning_rate": 5.9208786666354355e-06, "loss": 2.7131, "step": 247740 }, { "epoch": 0.03648, "grad_norm": 0.7988266944885254, "learning_rate": 5.9201434435508565e-06, "loss": 2.6968, "step": 247750 }, { "epoch": 0.0365056, "grad_norm": 0.840715229511261, "learning_rate": 5.919408246923062e-06, "loss": 2.8297, "step": 247760 }, { "epoch": 0.0365312, "grad_norm": 0.770697295665741, "learning_rate": 5.9186730767568205e-06, "loss": 2.9661, "step": 247770 }, { "epoch": 0.0365568, "grad_norm": 1.0405207872390747, "learning_rate": 5.917937933056899e-06, "loss": 2.8852, "step": 247780 }, { "epoch": 0.0365824, "grad_norm": 1.3175740242004395, "learning_rate": 5.917202815828065e-06, "loss": 3.1034, "step": 247790 }, { "epoch": 0.036608, "grad_norm": 0.9191152453422546, "learning_rate": 5.916467725075086e-06, "loss": 3.1934, "step": 247800 }, { "epoch": 0.0366336, "grad_norm": 0.8708665370941162, "learning_rate": 5.915732660802729e-06, "loss": 3.1423, "step": 247810 }, { "epoch": 0.0366592, "grad_norm": 0.7202258706092834, "learning_rate": 5.91499762301576e-06, "loss": 2.9127, "step": 247820 }, { "epoch": 0.0366848, "grad_norm": 0.8025045394897461, "learning_rate": 5.914262611718946e-06, "loss": 2.9323, "step": 247830 }, { "epoch": 0.0367104, "grad_norm": 0.7926526069641113, "learning_rate": 5.9135276269170515e-06, "loss": 2.9929, "step": 247840 }, { "epoch": 0.036736, "grad_norm": 0.8692866563796997, "learning_rate": 5.9127926686148464e-06, "loss": 2.8967, "step": 247850 }, { "epoch": 0.0367616, "grad_norm": 0.7896663546562195, "learning_rate": 5.912057736817089e-06, "loss": 2.7882, "step": 247860 }, { "epoch": 0.0367872, "grad_norm": 0.8045694828033447, "learning_rate": 5.911322831528558e-06, "loss": 2.7337, "step": 247870 }, { "epoch": 0.0368128, "grad_norm": 0.7652718424797058, "learning_rate": 5.910587952754008e-06, "loss": 2.8704, "step": 247880 }, { "epoch": 0.0368384, "grad_norm": 0.7732508778572083, "learning_rate": 5.909853100498208e-06, "loss": 2.9782, "step": 247890 }, { "epoch": 0.036864, "grad_norm": 0.8998091220855713, "learning_rate": 5.909118274765926e-06, "loss": 3.0704, "step": 247900 }, { "epoch": 0.0368896, "grad_norm": 0.96812903881073, "learning_rate": 5.908383475561925e-06, "loss": 2.9314, "step": 247910 }, { "epoch": 0.0369152, "grad_norm": 0.8475438952445984, "learning_rate": 5.9076487028909675e-06, "loss": 2.9176, "step": 247920 }, { "epoch": 0.0369408, "grad_norm": 0.8272233009338379, "learning_rate": 5.906913956757821e-06, "loss": 3.0262, "step": 247930 }, { "epoch": 0.0369664, "grad_norm": 1.0258581638336182, "learning_rate": 5.90617923716725e-06, "loss": 2.9486, "step": 247940 }, { "epoch": 0.036992, "grad_norm": 0.8305932283401489, "learning_rate": 5.90544454412402e-06, "loss": 3.0438, "step": 247950 }, { "epoch": 0.0370176, "grad_norm": 0.7462217211723328, "learning_rate": 5.9047098776328926e-06, "loss": 2.9615, "step": 247960 }, { "epoch": 0.0370432, "grad_norm": 0.8497454524040222, "learning_rate": 5.903975237698634e-06, "loss": 2.9048, "step": 247970 }, { "epoch": 0.0370688, "grad_norm": 0.7313283681869507, "learning_rate": 5.903240624326007e-06, "loss": 2.9716, "step": 247980 }, { "epoch": 0.0370944, "grad_norm": 0.8498904705047607, "learning_rate": 5.9025060375197754e-06, "loss": 2.6201, "step": 247990 }, { "epoch": 0.03712, "grad_norm": 0.8245441913604736, "learning_rate": 5.901771477284704e-06, "loss": 3.1067, "step": 248000 }, { "epoch": 0.0371456, "grad_norm": 0.7546050548553467, "learning_rate": 5.901036943625553e-06, "loss": 2.8939, "step": 248010 }, { "epoch": 0.0371712, "grad_norm": 0.7640741467475891, "learning_rate": 5.900302436547092e-06, "loss": 2.7107, "step": 248020 }, { "epoch": 0.0371968, "grad_norm": 0.7787065505981445, "learning_rate": 5.899567956054073e-06, "loss": 2.8991, "step": 248030 }, { "epoch": 0.0372224, "grad_norm": 0.8471642732620239, "learning_rate": 5.898833502151273e-06, "loss": 2.9802, "step": 248040 }, { "epoch": 0.037248, "grad_norm": 0.8706129789352417, "learning_rate": 5.898099074843445e-06, "loss": 2.9004, "step": 248050 }, { "epoch": 0.0372736, "grad_norm": 0.9124633073806763, "learning_rate": 5.897364674135355e-06, "loss": 2.9053, "step": 248060 }, { "epoch": 0.0372992, "grad_norm": 0.7229591608047485, "learning_rate": 5.896630300031766e-06, "loss": 2.848, "step": 248070 }, { "epoch": 0.0373248, "grad_norm": 0.7933399677276611, "learning_rate": 5.895895952537437e-06, "loss": 2.8387, "step": 248080 }, { "epoch": 0.0373504, "grad_norm": 0.7910750508308411, "learning_rate": 5.8951616316571326e-06, "loss": 2.8814, "step": 248090 }, { "epoch": 0.037376, "grad_norm": 0.8586826920509338, "learning_rate": 5.894427337395614e-06, "loss": 2.8792, "step": 248100 }, { "epoch": 0.0374016, "grad_norm": 0.7448665499687195, "learning_rate": 5.893693069757644e-06, "loss": 2.834, "step": 248110 }, { "epoch": 0.0374272, "grad_norm": 0.8265507221221924, "learning_rate": 5.892958828747983e-06, "loss": 2.7721, "step": 248120 }, { "epoch": 0.0374528, "grad_norm": 0.9134870767593384, "learning_rate": 5.8922246143713916e-06, "loss": 2.623, "step": 248130 }, { "epoch": 0.0374784, "grad_norm": 0.982964038848877, "learning_rate": 5.891490426632631e-06, "loss": 3.0372, "step": 248140 }, { "epoch": 0.037504, "grad_norm": 0.9557311534881592, "learning_rate": 5.890756265536466e-06, "loss": 2.9442, "step": 248150 }, { "epoch": 0.0375296, "grad_norm": 0.7281880974769592, "learning_rate": 5.890022131087651e-06, "loss": 2.8933, "step": 248160 }, { "epoch": 0.0375552, "grad_norm": 0.9290975332260132, "learning_rate": 5.889288023290952e-06, "loss": 3.0133, "step": 248170 }, { "epoch": 0.0375808, "grad_norm": 0.9024880528450012, "learning_rate": 5.888553942151129e-06, "loss": 2.7798, "step": 248180 }, { "epoch": 0.0376064, "grad_norm": 0.8578115701675415, "learning_rate": 5.8878198876729385e-06, "loss": 2.9567, "step": 248190 }, { "epoch": 0.037632, "grad_norm": 0.8139041066169739, "learning_rate": 5.887085859861145e-06, "loss": 2.895, "step": 248200 }, { "epoch": 0.0376576, "grad_norm": 0.7439284324645996, "learning_rate": 5.886351858720503e-06, "loss": 3.0668, "step": 248210 }, { "epoch": 0.0376832, "grad_norm": 0.8388254642486572, "learning_rate": 5.885617884255779e-06, "loss": 3.1832, "step": 248220 }, { "epoch": 0.0377088, "grad_norm": 0.7724881768226624, "learning_rate": 5.884883936471728e-06, "loss": 2.7856, "step": 248230 }, { "epoch": 0.0377344, "grad_norm": 1.654720664024353, "learning_rate": 5.884150015373114e-06, "loss": 2.8232, "step": 248240 }, { "epoch": 0.03776, "grad_norm": 0.8899338841438293, "learning_rate": 5.883416120964691e-06, "loss": 2.8165, "step": 248250 }, { "epoch": 0.0377856, "grad_norm": 0.7919434309005737, "learning_rate": 5.882682253251221e-06, "loss": 2.8391, "step": 248260 }, { "epoch": 0.0378112, "grad_norm": 0.791506826877594, "learning_rate": 5.881948412237463e-06, "loss": 2.9406, "step": 248270 }, { "epoch": 0.0378368, "grad_norm": 0.8983154892921448, "learning_rate": 5.8812145979281745e-06, "loss": 2.9627, "step": 248280 }, { "epoch": 0.0378624, "grad_norm": 0.7709835767745972, "learning_rate": 5.880480810328113e-06, "loss": 2.8353, "step": 248290 }, { "epoch": 0.037888, "grad_norm": 0.8221229314804077, "learning_rate": 5.879747049442041e-06, "loss": 2.9016, "step": 248300 }, { "epoch": 0.0379136, "grad_norm": 0.814567506313324, "learning_rate": 5.879013315274715e-06, "loss": 2.8284, "step": 248310 }, { "epoch": 0.0379392, "grad_norm": 0.7591407895088196, "learning_rate": 5.8782796078308906e-06, "loss": 2.9296, "step": 248320 }, { "epoch": 0.0379648, "grad_norm": 0.7821623682975769, "learning_rate": 5.877545927115326e-06, "loss": 2.96, "step": 248330 }, { "epoch": 0.0379904, "grad_norm": 1.031684398651123, "learning_rate": 5.876812273132783e-06, "loss": 2.8283, "step": 248340 }, { "epoch": 0.038016, "grad_norm": 0.7943613529205322, "learning_rate": 5.876078645888017e-06, "loss": 2.8471, "step": 248350 }, { "epoch": 0.0380416, "grad_norm": 0.8494152426719666, "learning_rate": 5.875345045385783e-06, "loss": 2.9367, "step": 248360 }, { "epoch": 0.0380672, "grad_norm": 0.8857992887496948, "learning_rate": 5.874611471630841e-06, "loss": 2.983, "step": 248370 }, { "epoch": 0.0380928, "grad_norm": 0.7906984686851501, "learning_rate": 5.873877924627945e-06, "loss": 2.9636, "step": 248380 }, { "epoch": 0.0381184, "grad_norm": 0.8816412091255188, "learning_rate": 5.873144404381856e-06, "loss": 2.8348, "step": 248390 }, { "epoch": 0.038144, "grad_norm": 0.895348846912384, "learning_rate": 5.8724109108973305e-06, "loss": 2.8857, "step": 248400 }, { "epoch": 0.0381696, "grad_norm": 1.0311665534973145, "learning_rate": 5.871677444179123e-06, "loss": 3.0938, "step": 248410 }, { "epoch": 0.0381952, "grad_norm": 0.8295584321022034, "learning_rate": 5.870944004231986e-06, "loss": 2.8525, "step": 248420 }, { "epoch": 0.0382208, "grad_norm": 0.9557947516441345, "learning_rate": 5.870210591060683e-06, "loss": 3.1126, "step": 248430 }, { "epoch": 0.0382464, "grad_norm": 0.8012545108795166, "learning_rate": 5.8694772046699685e-06, "loss": 2.839, "step": 248440 }, { "epoch": 0.038272, "grad_norm": 0.8316527605056763, "learning_rate": 5.868743845064594e-06, "loss": 2.8449, "step": 248450 }, { "epoch": 0.0382976, "grad_norm": 0.9711632132530212, "learning_rate": 5.868010512249318e-06, "loss": 2.8155, "step": 248460 }, { "epoch": 0.0383232, "grad_norm": 0.8598636388778687, "learning_rate": 5.867277206228897e-06, "loss": 2.8562, "step": 248470 }, { "epoch": 0.0383488, "grad_norm": 0.8525871634483337, "learning_rate": 5.8665439270080835e-06, "loss": 3.0074, "step": 248480 }, { "epoch": 0.0383744, "grad_norm": 0.7828537225723267, "learning_rate": 5.865810674591634e-06, "loss": 2.9085, "step": 248490 }, { "epoch": 0.0384, "grad_norm": 0.7628176212310791, "learning_rate": 5.865077448984304e-06, "loss": 2.8874, "step": 248500 }, { "epoch": 0.0384256, "grad_norm": 0.7519339323043823, "learning_rate": 5.86434425019085e-06, "loss": 2.9144, "step": 248510 }, { "epoch": 0.0384512, "grad_norm": 0.7759193778038025, "learning_rate": 5.86361107821602e-06, "loss": 2.9034, "step": 248520 }, { "epoch": 0.0384768, "grad_norm": 0.7584382891654968, "learning_rate": 5.862877933064576e-06, "loss": 2.9126, "step": 248530 }, { "epoch": 0.0385024, "grad_norm": 0.7325407862663269, "learning_rate": 5.862144814741269e-06, "loss": 2.6977, "step": 248540 }, { "epoch": 0.038528, "grad_norm": 0.8484264612197876, "learning_rate": 5.8614117232508486e-06, "loss": 2.8687, "step": 248550 }, { "epoch": 0.0385536, "grad_norm": 1.3888843059539795, "learning_rate": 5.8606786585980755e-06, "loss": 2.9544, "step": 248560 }, { "epoch": 0.0385792, "grad_norm": 0.8154997825622559, "learning_rate": 5.8599456207877045e-06, "loss": 2.8325, "step": 248570 }, { "epoch": 0.0386048, "grad_norm": 0.7846381068229675, "learning_rate": 5.859212609824484e-06, "loss": 3.1117, "step": 248580 }, { "epoch": 0.0386304, "grad_norm": 0.8717289566993713, "learning_rate": 5.8584796257131696e-06, "loss": 2.8198, "step": 248590 }, { "epoch": 0.038656, "grad_norm": 1.7766191959381104, "learning_rate": 5.857746668458515e-06, "loss": 2.1963, "step": 248600 }, { "epoch": 0.0386816, "grad_norm": 0.7868359088897705, "learning_rate": 5.85701373806527e-06, "loss": 2.6337, "step": 248610 }, { "epoch": 0.0387072, "grad_norm": 0.751844048500061, "learning_rate": 5.85628083453819e-06, "loss": 2.731, "step": 248620 }, { "epoch": 0.0387328, "grad_norm": 2.24831485748291, "learning_rate": 5.8555479578820284e-06, "loss": 2.8066, "step": 248630 }, { "epoch": 0.0387584, "grad_norm": 0.77434903383255, "learning_rate": 5.854815108101539e-06, "loss": 2.6855, "step": 248640 }, { "epoch": 0.038784, "grad_norm": 0.7514477372169495, "learning_rate": 5.85408228520147e-06, "loss": 2.546, "step": 248650 }, { "epoch": 0.0388096, "grad_norm": 0.8826143741607666, "learning_rate": 5.853349489186575e-06, "loss": 2.7305, "step": 248660 }, { "epoch": 0.0388352, "grad_norm": 0.9061204791069031, "learning_rate": 5.852616720061608e-06, "loss": 2.7281, "step": 248670 }, { "epoch": 0.0388608, "grad_norm": 1.2551947832107544, "learning_rate": 5.851883977831318e-06, "loss": 2.6255, "step": 248680 }, { "epoch": 0.0388864, "grad_norm": 0.864134669303894, "learning_rate": 5.851151262500459e-06, "loss": 2.6682, "step": 248690 }, { "epoch": 0.038912, "grad_norm": 0.9218107461929321, "learning_rate": 5.85041857407378e-06, "loss": 2.5734, "step": 248700 }, { "epoch": 0.0389376, "grad_norm": 0.8438498377799988, "learning_rate": 5.8496859125560354e-06, "loss": 2.4927, "step": 248710 }, { "epoch": 0.0389632, "grad_norm": 1.6143759489059448, "learning_rate": 5.848953277951973e-06, "loss": 2.689, "step": 248720 }, { "epoch": 0.0389888, "grad_norm": 0.8249414563179016, "learning_rate": 5.848220670266344e-06, "loss": 2.7734, "step": 248730 }, { "epoch": 0.0390144, "grad_norm": 0.8129360675811768, "learning_rate": 5.847488089503902e-06, "loss": 2.9476, "step": 248740 }, { "epoch": 0.03904, "grad_norm": 0.8103604912757874, "learning_rate": 5.846755535669396e-06, "loss": 2.4514, "step": 248750 }, { "epoch": 0.0390656, "grad_norm": 0.7589678168296814, "learning_rate": 5.846023008767579e-06, "loss": 2.7916, "step": 248760 }, { "epoch": 0.0390912, "grad_norm": 0.7535825371742249, "learning_rate": 5.845290508803196e-06, "loss": 3.2158, "step": 248770 }, { "epoch": 0.0391168, "grad_norm": 0.944585382938385, "learning_rate": 5.844558035781002e-06, "loss": 2.8538, "step": 248780 }, { "epoch": 0.0391424, "grad_norm": 0.8053461909294128, "learning_rate": 5.843825589705743e-06, "loss": 2.7867, "step": 248790 }, { "epoch": 0.039168, "grad_norm": 0.8379564881324768, "learning_rate": 5.843093170582172e-06, "loss": 2.5237, "step": 248800 }, { "epoch": 0.0391936, "grad_norm": 0.7577686905860901, "learning_rate": 5.842360778415036e-06, "loss": 2.5085, "step": 248810 }, { "epoch": 0.0392192, "grad_norm": 0.8565651178359985, "learning_rate": 5.841628413209085e-06, "loss": 2.3616, "step": 248820 }, { "epoch": 0.0392448, "grad_norm": 0.8635993003845215, "learning_rate": 5.8408960749690694e-06, "loss": 2.6311, "step": 248830 }, { "epoch": 0.0392704, "grad_norm": 0.7614710330963135, "learning_rate": 5.840163763699738e-06, "loss": 2.649, "step": 248840 }, { "epoch": 0.039296, "grad_norm": 0.7904678583145142, "learning_rate": 5.839431479405838e-06, "loss": 2.7261, "step": 248850 }, { "epoch": 0.0393216, "grad_norm": 0.8607455492019653, "learning_rate": 5.838699222092119e-06, "loss": 2.7568, "step": 248860 }, { "epoch": 0.0393472, "grad_norm": 0.8163661956787109, "learning_rate": 5.837966991763328e-06, "loss": 2.6514, "step": 248870 }, { "epoch": 0.0393728, "grad_norm": 0.9254552125930786, "learning_rate": 5.837234788424217e-06, "loss": 2.8698, "step": 248880 }, { "epoch": 0.0393984, "grad_norm": 0.7239981889724731, "learning_rate": 5.836502612079535e-06, "loss": 2.6688, "step": 248890 }, { "epoch": 0.039424, "grad_norm": 0.883226752281189, "learning_rate": 5.835770462734021e-06, "loss": 2.4783, "step": 248900 }, { "epoch": 0.0394496, "grad_norm": 1.5952231884002686, "learning_rate": 5.835038340392432e-06, "loss": 2.2389, "step": 248910 }, { "epoch": 0.0394752, "grad_norm": 0.8501927256584167, "learning_rate": 5.8343062450595116e-06, "loss": 2.573, "step": 248920 }, { "epoch": 0.0395008, "grad_norm": 1.1028451919555664, "learning_rate": 5.833574176740007e-06, "loss": 2.789, "step": 248930 }, { "epoch": 0.0395264, "grad_norm": 0.7180627584457397, "learning_rate": 5.832842135438668e-06, "loss": 2.4339, "step": 248940 }, { "epoch": 0.039552, "grad_norm": 0.746472179889679, "learning_rate": 5.832110121160244e-06, "loss": 2.6351, "step": 248950 }, { "epoch": 0.0395776, "grad_norm": 0.8045094609260559, "learning_rate": 5.8313781339094745e-06, "loss": 2.7656, "step": 248960 }, { "epoch": 0.0396032, "grad_norm": 0.7604789733886719, "learning_rate": 5.8306461736911105e-06, "loss": 2.6126, "step": 248970 }, { "epoch": 0.0396288, "grad_norm": 0.7302208542823792, "learning_rate": 5.829914240509896e-06, "loss": 2.522, "step": 248980 }, { "epoch": 0.0396544, "grad_norm": 0.7493511438369751, "learning_rate": 5.82918233437058e-06, "loss": 2.677, "step": 248990 }, { "epoch": 0.03968, "grad_norm": 0.8108726143836975, "learning_rate": 5.8284504552779095e-06, "loss": 2.7382, "step": 249000 }, { "epoch": 0.0397056, "grad_norm": 0.913537859916687, "learning_rate": 5.827718603236627e-06, "loss": 2.5981, "step": 249010 }, { "epoch": 0.0397312, "grad_norm": 0.7747339606285095, "learning_rate": 5.826986778251487e-06, "loss": 2.609, "step": 249020 }, { "epoch": 0.0397568, "grad_norm": 0.8037036657333374, "learning_rate": 5.826254980327224e-06, "loss": 2.8067, "step": 249030 }, { "epoch": 0.0397824, "grad_norm": 0.8059662580490112, "learning_rate": 5.8255232094685864e-06, "loss": 2.4758, "step": 249040 }, { "epoch": 0.039808, "grad_norm": 0.8546116948127747, "learning_rate": 5.824791465680323e-06, "loss": 3.027, "step": 249050 }, { "epoch": 0.0398336, "grad_norm": 0.9825249314308167, "learning_rate": 5.824059748967177e-06, "loss": 2.7872, "step": 249060 }, { "epoch": 0.0398592, "grad_norm": 0.7942522168159485, "learning_rate": 5.823328059333893e-06, "loss": 2.7924, "step": 249070 }, { "epoch": 0.0398848, "grad_norm": 0.992071270942688, "learning_rate": 5.822596396785217e-06, "loss": 2.7055, "step": 249080 }, { "epoch": 0.0399104, "grad_norm": 0.7350842952728271, "learning_rate": 5.821864761325894e-06, "loss": 2.5762, "step": 249090 }, { "epoch": 0.039936, "grad_norm": 1.4079875946044922, "learning_rate": 5.821133152960667e-06, "loss": 2.9537, "step": 249100 }, { "epoch": 2.56e-05, "grad_norm": 0.797319233417511, "learning_rate": 5.820401571694281e-06, "loss": 2.0972, "step": 249110 }, { "epoch": 5.12e-05, "grad_norm": 0.8166006207466125, "learning_rate": 5.819670017531479e-06, "loss": 2.6924, "step": 249120 }, { "epoch": 7.68e-05, "grad_norm": 1.0441590547561646, "learning_rate": 5.8189384904770066e-06, "loss": 2.7936, "step": 249130 }, { "epoch": 0.0001024, "grad_norm": 0.7603439688682556, "learning_rate": 5.818206990535607e-06, "loss": 2.5419, "step": 249140 }, { "epoch": 0.000128, "grad_norm": 0.7672567367553711, "learning_rate": 5.8174755177120276e-06, "loss": 2.5244, "step": 249150 }, { "epoch": 0.0001536, "grad_norm": 0.7711187601089478, "learning_rate": 5.816744072011005e-06, "loss": 2.6741, "step": 249160 }, { "epoch": 0.0001792, "grad_norm": 0.7867550849914551, "learning_rate": 5.816012653437283e-06, "loss": 2.546, "step": 249170 }, { "epoch": 0.0002048, "grad_norm": 0.7878385186195374, "learning_rate": 5.815281261995609e-06, "loss": 2.6374, "step": 249180 }, { "epoch": 0.0002304, "grad_norm": 0.7535017728805542, "learning_rate": 5.814549897690722e-06, "loss": 2.6036, "step": 249190 }, { "epoch": 0.000256, "grad_norm": 0.7731413841247559, "learning_rate": 5.8138185605273675e-06, "loss": 2.8, "step": 249200 }, { "epoch": 0.0002816, "grad_norm": 0.7992144823074341, "learning_rate": 5.813087250510288e-06, "loss": 2.5782, "step": 249210 }, { "epoch": 0.0003072, "grad_norm": 0.7456404566764832, "learning_rate": 5.8123559676442275e-06, "loss": 2.4891, "step": 249220 }, { "epoch": 0.0003328, "grad_norm": 0.8002533912658691, "learning_rate": 5.811624711933923e-06, "loss": 2.6826, "step": 249230 }, { "epoch": 0.0003584, "grad_norm": 0.986612856388092, "learning_rate": 5.810893483384118e-06, "loss": 2.7682, "step": 249240 }, { "epoch": 0.000384, "grad_norm": 0.8581728339195251, "learning_rate": 5.810162281999552e-06, "loss": 2.6925, "step": 249250 }, { "epoch": 0.0004096, "grad_norm": 0.7684454321861267, "learning_rate": 5.809431107784973e-06, "loss": 2.5646, "step": 249260 }, { "epoch": 0.0004352, "grad_norm": 0.8953800201416016, "learning_rate": 5.808699960745125e-06, "loss": 2.852, "step": 249270 }, { "epoch": 0.0004608, "grad_norm": 0.8416187763214111, "learning_rate": 5.807968840884739e-06, "loss": 2.877, "step": 249280 }, { "epoch": 0.0004864, "grad_norm": 1.0038516521453857, "learning_rate": 5.807237748208561e-06, "loss": 2.7991, "step": 249290 }, { "epoch": 0.000512, "grad_norm": 0.7969757318496704, "learning_rate": 5.806506682721332e-06, "loss": 2.5773, "step": 249300 }, { "epoch": 0.0005376, "grad_norm": 0.9502347707748413, "learning_rate": 5.805775644427794e-06, "loss": 2.4533, "step": 249310 }, { "epoch": 0.0005632, "grad_norm": 0.8796488046646118, "learning_rate": 5.805044633332684e-06, "loss": 2.739, "step": 249320 }, { "epoch": 0.0005888, "grad_norm": 0.970037043094635, "learning_rate": 5.804313649440746e-06, "loss": 2.3729, "step": 249330 }, { "epoch": 0.0006144, "grad_norm": 0.8899552822113037, "learning_rate": 5.803582692756718e-06, "loss": 2.6376, "step": 249340 }, { "epoch": 0.00064, "grad_norm": 1.039807677268982, "learning_rate": 5.802851763285346e-06, "loss": 2.8799, "step": 249350 }, { "epoch": 0.0006656, "grad_norm": 0.8300620913505554, "learning_rate": 5.802120861031361e-06, "loss": 2.8411, "step": 249360 }, { "epoch": 0.0006912, "grad_norm": 0.8274714350700378, "learning_rate": 5.801389985999506e-06, "loss": 3.0445, "step": 249370 }, { "epoch": 0.0007168, "grad_norm": 0.8634915351867676, "learning_rate": 5.80065913819452e-06, "loss": 2.7632, "step": 249380 }, { "epoch": 0.0007424, "grad_norm": 0.9084657430648804, "learning_rate": 5.799928317621145e-06, "loss": 2.7424, "step": 249390 }, { "epoch": 0.000768, "grad_norm": 0.697940468788147, "learning_rate": 5.799197524284118e-06, "loss": 2.6167, "step": 249400 }, { "epoch": 0.0007936, "grad_norm": 0.8086650371551514, "learning_rate": 5.798466758188179e-06, "loss": 2.7516, "step": 249410 }, { "epoch": 0.0008192, "grad_norm": 1.296993374824524, "learning_rate": 5.797736019338066e-06, "loss": 2.1151, "step": 249420 }, { "epoch": 0.0008448, "grad_norm": 0.803306519985199, "learning_rate": 5.797005307738518e-06, "loss": 2.6917, "step": 249430 }, { "epoch": 0.0008704, "grad_norm": 0.9706259965896606, "learning_rate": 5.796274623394273e-06, "loss": 2.568, "step": 249440 }, { "epoch": 0.000896, "grad_norm": 0.8298355937004089, "learning_rate": 5.795543966310071e-06, "loss": 2.6423, "step": 249450 }, { "epoch": 0.0009216, "grad_norm": 0.9649028778076172, "learning_rate": 5.79481333649065e-06, "loss": 2.6104, "step": 249460 }, { "epoch": 0.0009472, "grad_norm": 1.0273646116256714, "learning_rate": 5.794082733940749e-06, "loss": 2.6217, "step": 249470 }, { "epoch": 0.0009728, "grad_norm": 0.7422587275505066, "learning_rate": 5.7933521586651e-06, "loss": 2.5434, "step": 249480 }, { "epoch": 0.0009984, "grad_norm": 0.9303723573684692, "learning_rate": 5.792621610668445e-06, "loss": 2.6341, "step": 249490 }, { "epoch": 0.001024, "grad_norm": 0.81159508228302, "learning_rate": 5.791891089955519e-06, "loss": 2.5447, "step": 249500 }, { "epoch": 0.0010496, "grad_norm": 0.921545147895813, "learning_rate": 5.791160596531062e-06, "loss": 2.5838, "step": 249510 }, { "epoch": 0.0010752, "grad_norm": 0.886893630027771, "learning_rate": 5.790430130399811e-06, "loss": 2.8069, "step": 249520 }, { "epoch": 0.0011008, "grad_norm": 0.7306959629058838, "learning_rate": 5.7896996915665e-06, "loss": 2.579, "step": 249530 }, { "epoch": 0.0011264, "grad_norm": 0.8859989643096924, "learning_rate": 5.788969280035869e-06, "loss": 2.6661, "step": 249540 }, { "epoch": 0.001152, "grad_norm": 0.757183849811554, "learning_rate": 5.788238895812657e-06, "loss": 2.7594, "step": 249550 }, { "epoch": 0.0011776, "grad_norm": 0.9500589370727539, "learning_rate": 5.787508538901591e-06, "loss": 2.5508, "step": 249560 }, { "epoch": 0.0012032, "grad_norm": 0.9118767976760864, "learning_rate": 5.786778209307413e-06, "loss": 2.9302, "step": 249570 }, { "epoch": 0.0012288, "grad_norm": 0.7440682053565979, "learning_rate": 5.78604790703486e-06, "loss": 2.9027, "step": 249580 }, { "epoch": 0.0012544, "grad_norm": 1.0861408710479736, "learning_rate": 5.785317632088665e-06, "loss": 2.791, "step": 249590 }, { "epoch": 0.00128, "grad_norm": 0.736195981502533, "learning_rate": 5.7845873844735656e-06, "loss": 2.5912, "step": 249600 }, { "epoch": 0.0013056, "grad_norm": 0.8088458776473999, "learning_rate": 5.783857164194296e-06, "loss": 2.526, "step": 249610 }, { "epoch": 0.0013312, "grad_norm": 0.9073466658592224, "learning_rate": 5.7831269712555925e-06, "loss": 2.9749, "step": 249620 }, { "epoch": 0.0013568, "grad_norm": 1.2213823795318604, "learning_rate": 5.782396805662189e-06, "loss": 2.9549, "step": 249630 }, { "epoch": 0.0013824, "grad_norm": 0.9455797076225281, "learning_rate": 5.7816666674188234e-06, "loss": 2.3597, "step": 249640 }, { "epoch": 0.001408, "grad_norm": 0.8919442892074585, "learning_rate": 5.780936556530227e-06, "loss": 2.6549, "step": 249650 }, { "epoch": 0.0014336, "grad_norm": 0.793249249458313, "learning_rate": 5.780206473001135e-06, "loss": 2.5359, "step": 249660 }, { "epoch": 0.0014592, "grad_norm": 0.8596956133842468, "learning_rate": 5.7794764168362895e-06, "loss": 2.619, "step": 249670 }, { "epoch": 0.0014848, "grad_norm": 0.8033567667007446, "learning_rate": 5.778746388040411e-06, "loss": 2.5162, "step": 249680 }, { "epoch": 0.0015104, "grad_norm": 0.8124760985374451, "learning_rate": 5.778016386618241e-06, "loss": 2.9413, "step": 249690 }, { "epoch": 0.001536, "grad_norm": 0.8625249266624451, "learning_rate": 5.777286412574513e-06, "loss": 2.5613, "step": 249700 }, { "epoch": 0.0015616, "grad_norm": 0.791652262210846, "learning_rate": 5.77655646591396e-06, "loss": 2.3558, "step": 249710 }, { "epoch": 0.0015872, "grad_norm": 0.8087891936302185, "learning_rate": 5.775826546641317e-06, "loss": 2.6852, "step": 249720 }, { "epoch": 0.0016128, "grad_norm": 0.8067068457603455, "learning_rate": 5.775096654761314e-06, "loss": 2.7459, "step": 249730 }, { "epoch": 0.0016384, "grad_norm": 0.998184084892273, "learning_rate": 5.774366790278692e-06, "loss": 2.6886, "step": 249740 }, { "epoch": 0.001664, "grad_norm": 0.7974511981010437, "learning_rate": 5.773636953198174e-06, "loss": 2.7805, "step": 249750 }, { "epoch": 0.0016896, "grad_norm": 0.874559760093689, "learning_rate": 5.772907143524497e-06, "loss": 2.7572, "step": 249760 }, { "epoch": 0.0017152, "grad_norm": 0.7690868973731995, "learning_rate": 5.7721773612623895e-06, "loss": 2.667, "step": 249770 }, { "epoch": 0.0017408, "grad_norm": 0.9316548705101013, "learning_rate": 5.7714476064165924e-06, "loss": 2.6102, "step": 249780 }, { "epoch": 0.0017664, "grad_norm": 0.8567599058151245, "learning_rate": 5.7707178789918336e-06, "loss": 2.523, "step": 249790 }, { "epoch": 0.001792, "grad_norm": 0.7681041359901428, "learning_rate": 5.769988178992849e-06, "loss": 2.8674, "step": 249800 }, { "epoch": 2.56e-05, "grad_norm": 0.8006198406219482, "learning_rate": 5.769258506424364e-06, "loss": 1.8977, "step": 249810 }, { "epoch": 5.12e-05, "grad_norm": 0.9628517627716064, "learning_rate": 5.768528861291113e-06, "loss": 2.7207, "step": 249820 }, { "epoch": 7.68e-05, "grad_norm": 0.9298821091651917, "learning_rate": 5.767799243597827e-06, "loss": 3.0202, "step": 249830 }, { "epoch": 0.0001024, "grad_norm": 0.7986998558044434, "learning_rate": 5.767069653349239e-06, "loss": 2.6537, "step": 249840 }, { "epoch": 0.000128, "grad_norm": 0.8582561016082764, "learning_rate": 5.766340090550077e-06, "loss": 2.9754, "step": 249850 }, { "epoch": 0.0001536, "grad_norm": 0.7499070763587952, "learning_rate": 5.765610555205077e-06, "loss": 2.7558, "step": 249860 }, { "epoch": 0.0001792, "grad_norm": 0.711100697517395, "learning_rate": 5.764881047318971e-06, "loss": 2.5196, "step": 249870 }, { "epoch": 0.0002048, "grad_norm": 0.894171416759491, "learning_rate": 5.764151566896481e-06, "loss": 2.8085, "step": 249880 }, { "epoch": 0.0002304, "grad_norm": 0.807715654373169, "learning_rate": 5.763422113942343e-06, "loss": 2.667, "step": 249890 }, { "epoch": 0.000256, "grad_norm": 0.9488003253936768, "learning_rate": 5.762692688461285e-06, "loss": 2.8111, "step": 249900 }, { "epoch": 0.0002816, "grad_norm": 0.824400007724762, "learning_rate": 5.76196329045804e-06, "loss": 2.6936, "step": 249910 }, { "epoch": 0.0003072, "grad_norm": 0.7917752265930176, "learning_rate": 5.761233919937337e-06, "loss": 2.4992, "step": 249920 }, { "epoch": 0.0003328, "grad_norm": 0.8485068082809448, "learning_rate": 5.760504576903905e-06, "loss": 2.4299, "step": 249930 }, { "epoch": 0.0003584, "grad_norm": 0.8849816918373108, "learning_rate": 5.759775261362479e-06, "loss": 2.813, "step": 249940 }, { "epoch": 0.000384, "grad_norm": 0.9016855955123901, "learning_rate": 5.759045973317774e-06, "loss": 2.7902, "step": 249950 }, { "epoch": 0.0004096, "grad_norm": 0.772830605506897, "learning_rate": 5.758316712774535e-06, "loss": 2.7314, "step": 249960 }, { "epoch": 0.0004352, "grad_norm": 0.7686393857002258, "learning_rate": 5.757587479737482e-06, "loss": 2.6235, "step": 249970 }, { "epoch": 0.0004608, "grad_norm": 0.8097503185272217, "learning_rate": 5.756858274211347e-06, "loss": 2.4163, "step": 249980 }, { "epoch": 0.0004864, "grad_norm": 0.81865394115448, "learning_rate": 5.75612909620086e-06, "loss": 2.7534, "step": 249990 }, { "epoch": 0.000512, "grad_norm": 0.8812549114227295, "learning_rate": 5.755399945710751e-06, "loss": 2.6724, "step": 250000 }, { "epoch": 0.0005376, "grad_norm": 0.7475103735923767, "learning_rate": 5.754670822745743e-06, "loss": 2.589, "step": 250010 }, { "epoch": 0.0005632, "grad_norm": 0.854483962059021, "learning_rate": 5.753941727310563e-06, "loss": 2.1468, "step": 250020 }, { "epoch": 0.0005888, "grad_norm": 0.8447445034980774, "learning_rate": 5.753212659409946e-06, "loss": 2.728, "step": 250030 }, { "epoch": 0.0006144, "grad_norm": 0.7941365242004395, "learning_rate": 5.752483619048614e-06, "loss": 2.4727, "step": 250040 }, { "epoch": 0.00064, "grad_norm": 0.8735479116439819, "learning_rate": 5.751754606231298e-06, "loss": 2.8392, "step": 250050 }, { "epoch": 0.0006656, "grad_norm": 0.8168923854827881, "learning_rate": 5.751025620962724e-06, "loss": 2.9221, "step": 250060 }, { "epoch": 0.0006912, "grad_norm": 0.7913432121276855, "learning_rate": 5.750296663247623e-06, "loss": 2.7173, "step": 250070 }, { "epoch": 0.0007168, "grad_norm": 0.8760449290275574, "learning_rate": 5.749567733090715e-06, "loss": 2.6973, "step": 250080 }, { "epoch": 0.0007424, "grad_norm": 0.8282016515731812, "learning_rate": 5.748838830496731e-06, "loss": 2.773, "step": 250090 }, { "epoch": 0.000768, "grad_norm": 0.7630759477615356, "learning_rate": 5.748109955470398e-06, "loss": 2.5733, "step": 250100 }, { "epoch": 0.0007936, "grad_norm": 0.7991570830345154, "learning_rate": 5.747381108016441e-06, "loss": 2.7948, "step": 250110 }, { "epoch": 0.0008192, "grad_norm": 0.8489035964012146, "learning_rate": 5.746652288139583e-06, "loss": 2.8186, "step": 250120 }, { "epoch": 0.0008448, "grad_norm": 0.7955869436264038, "learning_rate": 5.745923495844563e-06, "loss": 2.7016, "step": 250130 }, { "epoch": 0.0008704, "grad_norm": 0.801302433013916, "learning_rate": 5.745194731136095e-06, "loss": 2.6381, "step": 250140 }, { "epoch": 0.000896, "grad_norm": 0.9346930384635925, "learning_rate": 5.744465994018907e-06, "loss": 2.8356, "step": 250150 }, { "epoch": 0.0009216, "grad_norm": 1.3120274543762207, "learning_rate": 5.743737284497728e-06, "loss": 3.1266, "step": 250160 }, { "epoch": 0.0009472, "grad_norm": 0.6897042393684387, "learning_rate": 5.7430086025772805e-06, "loss": 2.5451, "step": 250170 }, { "epoch": 0.0009728, "grad_norm": 0.7767043709754944, "learning_rate": 5.74227994826229e-06, "loss": 2.7679, "step": 250180 }, { "epoch": 0.0009984, "grad_norm": 0.9722742438316345, "learning_rate": 5.741551321557483e-06, "loss": 2.6638, "step": 250190 }, { "epoch": 0.001024, "grad_norm": 0.9528529047966003, "learning_rate": 5.740822722467589e-06, "loss": 2.4206, "step": 250200 }, { "epoch": 0.0010496, "grad_norm": 0.8870466351509094, "learning_rate": 5.740094150997322e-06, "loss": 2.6935, "step": 250210 }, { "epoch": 0.0010752, "grad_norm": 0.8201764225959778, "learning_rate": 5.739365607151412e-06, "loss": 2.7212, "step": 250220 }, { "epoch": 0.0011008, "grad_norm": 0.9022127985954285, "learning_rate": 5.738637090934587e-06, "loss": 2.8351, "step": 250230 }, { "epoch": 0.0011264, "grad_norm": 0.8922038674354553, "learning_rate": 5.7379086023515655e-06, "loss": 2.7798, "step": 250240 }, { "epoch": 0.001152, "grad_norm": 0.8280287384986877, "learning_rate": 5.737180141407074e-06, "loss": 2.8946, "step": 250250 }, { "epoch": 0.0011776, "grad_norm": 0.7808349132537842, "learning_rate": 5.7364517081058365e-06, "loss": 2.9396, "step": 250260 }, { "epoch": 0.0012032, "grad_norm": 0.7640113830566406, "learning_rate": 5.735723302452581e-06, "loss": 2.7906, "step": 250270 }, { "epoch": 0.0012288, "grad_norm": 0.8188612461090088, "learning_rate": 5.734994924452022e-06, "loss": 2.9165, "step": 250280 }, { "epoch": 0.0012544, "grad_norm": 0.7910318374633789, "learning_rate": 5.734266574108883e-06, "loss": 2.6328, "step": 250290 }, { "epoch": 0.00128, "grad_norm": 0.8492656946182251, "learning_rate": 5.733538251427897e-06, "loss": 2.9757, "step": 250300 }, { "epoch": 0.0013056, "grad_norm": 0.8221877217292786, "learning_rate": 5.73280995641378e-06, "loss": 2.9183, "step": 250310 }, { "epoch": 0.0013312, "grad_norm": 0.9464103579521179, "learning_rate": 5.732081689071257e-06, "loss": 2.4906, "step": 250320 }, { "epoch": 0.0013568, "grad_norm": 0.841974139213562, "learning_rate": 5.731353449405055e-06, "loss": 2.5847, "step": 250330 }, { "epoch": 0.0013824, "grad_norm": 0.7790971398353577, "learning_rate": 5.730625237419886e-06, "loss": 2.8382, "step": 250340 }, { "epoch": 0.001408, "grad_norm": 0.7881439328193665, "learning_rate": 5.729897053120478e-06, "loss": 2.7705, "step": 250350 }, { "epoch": 0.0014336, "grad_norm": 0.7745099663734436, "learning_rate": 5.729168896511552e-06, "loss": 2.7895, "step": 250360 }, { "epoch": 0.0014592, "grad_norm": 0.854263961315155, "learning_rate": 5.728440767597833e-06, "loss": 2.6426, "step": 250370 }, { "epoch": 0.0014848, "grad_norm": 0.8113903403282166, "learning_rate": 5.727712666384037e-06, "loss": 2.6506, "step": 250380 }, { "epoch": 0.0015104, "grad_norm": 0.9782209396362305, "learning_rate": 5.726984592874892e-06, "loss": 2.2448, "step": 250390 }, { "epoch": 0.001536, "grad_norm": 0.9131900072097778, "learning_rate": 5.72625654707512e-06, "loss": 2.1067, "step": 250400 }, { "epoch": 0.0015616, "grad_norm": 0.8975584506988525, "learning_rate": 5.7255285289894325e-06, "loss": 2.6041, "step": 250410 }, { "epoch": 0.0015872, "grad_norm": 0.9348223805427551, "learning_rate": 5.724800538622558e-06, "loss": 2.8955, "step": 250420 }, { "epoch": 0.0016128, "grad_norm": 0.8707572221755981, "learning_rate": 5.724072575979216e-06, "loss": 2.8706, "step": 250430 }, { "epoch": 0.0016384, "grad_norm": 0.798159122467041, "learning_rate": 5.723344641064126e-06, "loss": 2.4637, "step": 250440 }, { "epoch": 0.001664, "grad_norm": 0.8366182446479797, "learning_rate": 5.722616733882009e-06, "loss": 2.6519, "step": 250450 }, { "epoch": 0.0016896, "grad_norm": 0.9116507172584534, "learning_rate": 5.721888854437587e-06, "loss": 2.7672, "step": 250460 }, { "epoch": 0.0017152, "grad_norm": 0.7933275699615479, "learning_rate": 5.721161002735578e-06, "loss": 2.6744, "step": 250470 }, { "epoch": 0.0017408, "grad_norm": 0.8458976745605469, "learning_rate": 5.720433178780704e-06, "loss": 3.0557, "step": 250480 }, { "epoch": 0.0017664, "grad_norm": 0.8224613666534424, "learning_rate": 5.7197053825776814e-06, "loss": 2.5845, "step": 250490 }, { "epoch": 0.001792, "grad_norm": 0.7472386956214905, "learning_rate": 5.7189776141312335e-06, "loss": 2.7321, "step": 250500 }, { "epoch": 2.56e-05, "grad_norm": 0.7720825672149658, "learning_rate": 5.718249873446077e-06, "loss": 1.7668, "step": 250510 }, { "epoch": 5.12e-05, "grad_norm": 0.8707046508789062, "learning_rate": 5.717522160526931e-06, "loss": 2.5914, "step": 250520 }, { "epoch": 7.68e-05, "grad_norm": 0.9744386076927185, "learning_rate": 5.716794475378522e-06, "loss": 2.8173, "step": 250530 }, { "epoch": 0.0001024, "grad_norm": 0.9141551852226257, "learning_rate": 5.716066818005558e-06, "loss": 2.7474, "step": 250540 }, { "epoch": 0.000128, "grad_norm": 0.807899534702301, "learning_rate": 5.715339188412762e-06, "loss": 2.6847, "step": 250550 }, { "epoch": 0.0001536, "grad_norm": 0.9074775576591492, "learning_rate": 5.7146115866048524e-06, "loss": 2.7628, "step": 250560 }, { "epoch": 0.0001792, "grad_norm": 0.7893065810203552, "learning_rate": 5.713884012586548e-06, "loss": 2.8015, "step": 250570 }, { "epoch": 0.0002048, "grad_norm": 0.7584040760993958, "learning_rate": 5.713156466362566e-06, "loss": 2.7478, "step": 250580 }, { "epoch": 0.0002304, "grad_norm": 0.8483837246894836, "learning_rate": 5.712428947937625e-06, "loss": 2.6785, "step": 250590 }, { "epoch": 0.000256, "grad_norm": 0.9784862399101257, "learning_rate": 5.711701457316448e-06, "loss": 2.6454, "step": 250600 }, { "epoch": 0.0002816, "grad_norm": 0.8224061727523804, "learning_rate": 5.710973994503741e-06, "loss": 2.6669, "step": 250610 }, { "epoch": 0.0003072, "grad_norm": 0.8483554720878601, "learning_rate": 5.71024655950423e-06, "loss": 2.4129, "step": 250620 }, { "epoch": 0.0003328, "grad_norm": 0.823150634765625, "learning_rate": 5.709519152322629e-06, "loss": 2.6298, "step": 250630 }, { "epoch": 0.0003584, "grad_norm": 0.942540168762207, "learning_rate": 5.708791772963653e-06, "loss": 2.9151, "step": 250640 }, { "epoch": 0.000384, "grad_norm": 0.7567484974861145, "learning_rate": 5.708064421432025e-06, "loss": 2.6225, "step": 250650 }, { "epoch": 0.0004096, "grad_norm": 0.7752779126167297, "learning_rate": 5.707337097732463e-06, "loss": 2.4691, "step": 250660 }, { "epoch": 0.0004352, "grad_norm": 0.811873733997345, "learning_rate": 5.706609801869673e-06, "loss": 2.5803, "step": 250670 }, { "epoch": 0.0004608, "grad_norm": 0.832188606262207, "learning_rate": 5.7058825338483805e-06, "loss": 2.6777, "step": 250680 }, { "epoch": 0.0004864, "grad_norm": 0.8346327543258667, "learning_rate": 5.705155293673297e-06, "loss": 2.7831, "step": 250690 }, { "epoch": 0.000512, "grad_norm": 0.7794786691665649, "learning_rate": 5.704428081349141e-06, "loss": 2.6759, "step": 250700 }, { "epoch": 0.0005376, "grad_norm": 0.966620147228241, "learning_rate": 5.703700896880626e-06, "loss": 2.5785, "step": 250710 }, { "epoch": 0.0005632, "grad_norm": 0.8074309825897217, "learning_rate": 5.70297374027247e-06, "loss": 2.6586, "step": 250720 }, { "epoch": 0.0005888, "grad_norm": 0.90766441822052, "learning_rate": 5.70224661152939e-06, "loss": 2.6412, "step": 250730 }, { "epoch": 0.0006144, "grad_norm": 0.7673394083976746, "learning_rate": 5.701519510656095e-06, "loss": 2.7575, "step": 250740 }, { "epoch": 0.00064, "grad_norm": 0.8672343492507935, "learning_rate": 5.700792437657303e-06, "loss": 2.7036, "step": 250750 }, { "epoch": 0.0006656, "grad_norm": 0.7971817255020142, "learning_rate": 5.70006539253773e-06, "loss": 3.0713, "step": 250760 }, { "epoch": 0.0006912, "grad_norm": 0.8166465759277344, "learning_rate": 5.6993383753020915e-06, "loss": 2.6035, "step": 250770 }, { "epoch": 0.0007168, "grad_norm": 0.835363507270813, "learning_rate": 5.698611385955099e-06, "loss": 2.6609, "step": 250780 }, { "epoch": 0.0007424, "grad_norm": 0.902379035949707, "learning_rate": 5.69788442450147e-06, "loss": 2.7699, "step": 250790 }, { "epoch": 0.000768, "grad_norm": 0.8655834794044495, "learning_rate": 5.697157490945919e-06, "loss": 2.7646, "step": 250800 }, { "epoch": 0.0007936, "grad_norm": 0.8975052833557129, "learning_rate": 5.696430585293155e-06, "loss": 2.5443, "step": 250810 }, { "epoch": 0.0008192, "grad_norm": 0.7165122628211975, "learning_rate": 5.695703707547892e-06, "loss": 2.6807, "step": 250820 }, { "epoch": 0.0008448, "grad_norm": 0.8583021759986877, "learning_rate": 5.694976857714849e-06, "loss": 2.5432, "step": 250830 }, { "epoch": 0.0008704, "grad_norm": 0.8437660932540894, "learning_rate": 5.6942500357987385e-06, "loss": 2.5463, "step": 250840 }, { "epoch": 0.000896, "grad_norm": 0.8377701044082642, "learning_rate": 5.693523241804275e-06, "loss": 2.5186, "step": 250850 }, { "epoch": 0.0009216, "grad_norm": 0.7776591777801514, "learning_rate": 5.692796475736164e-06, "loss": 2.7017, "step": 250860 }, { "epoch": 0.0009472, "grad_norm": 0.9261879920959473, "learning_rate": 5.692069737599124e-06, "loss": 2.6376, "step": 250870 }, { "epoch": 0.0009728, "grad_norm": 0.8005128502845764, "learning_rate": 5.691343027397868e-06, "loss": 2.776, "step": 250880 }, { "epoch": 0.0009984, "grad_norm": 0.8440898656845093, "learning_rate": 5.6906163451371055e-06, "loss": 2.7259, "step": 250890 }, { "epoch": 0.001024, "grad_norm": 0.8147785067558289, "learning_rate": 5.689889690821552e-06, "loss": 2.7009, "step": 250900 }, { "epoch": 0.0010496, "grad_norm": 0.8129860758781433, "learning_rate": 5.689163064455918e-06, "loss": 2.7639, "step": 250910 }, { "epoch": 0.0010752, "grad_norm": 0.9012199640274048, "learning_rate": 5.688436466044915e-06, "loss": 2.7318, "step": 250920 }, { "epoch": 0.0011008, "grad_norm": 0.8255957365036011, "learning_rate": 5.687709895593261e-06, "loss": 2.7955, "step": 250930 }, { "epoch": 0.0011264, "grad_norm": 0.958888828754425, "learning_rate": 5.686983353105657e-06, "loss": 2.6553, "step": 250940 }, { "epoch": 0.001152, "grad_norm": 0.8673452734947205, "learning_rate": 5.6862568385868215e-06, "loss": 2.9064, "step": 250950 }, { "epoch": 0.0011776, "grad_norm": 0.9116180539131165, "learning_rate": 5.6855303520414614e-06, "loss": 2.7711, "step": 250960 }, { "epoch": 0.0012032, "grad_norm": 0.7889728546142578, "learning_rate": 5.684803893474293e-06, "loss": 2.7697, "step": 250970 }, { "epoch": 0.0012288, "grad_norm": 0.8885542154312134, "learning_rate": 5.684077462890023e-06, "loss": 2.8406, "step": 250980 }, { "epoch": 0.0012544, "grad_norm": 0.9181637763977051, "learning_rate": 5.6833510602933635e-06, "loss": 2.8131, "step": 250990 }, { "epoch": 0.00128, "grad_norm": 0.761827826499939, "learning_rate": 5.682624685689026e-06, "loss": 2.852, "step": 251000 }, { "epoch": 0.0013056, "grad_norm": 0.7953298687934875, "learning_rate": 5.68189833908172e-06, "loss": 2.8334, "step": 251010 }, { "epoch": 0.0013312, "grad_norm": 0.8254378437995911, "learning_rate": 5.681172020476154e-06, "loss": 2.6171, "step": 251020 }, { "epoch": 0.0013568, "grad_norm": 0.8299893140792847, "learning_rate": 5.680445729877041e-06, "loss": 2.7034, "step": 251030 }, { "epoch": 0.0013824, "grad_norm": 0.799452543258667, "learning_rate": 5.6797194672890885e-06, "loss": 2.774, "step": 251040 }, { "epoch": 0.001408, "grad_norm": 0.7561438083648682, "learning_rate": 5.678993232717013e-06, "loss": 2.7418, "step": 251050 }, { "epoch": 0.0014336, "grad_norm": 0.889094889163971, "learning_rate": 5.678267026165513e-06, "loss": 2.8124, "step": 251060 }, { "epoch": 0.0014592, "grad_norm": 0.7488831877708435, "learning_rate": 5.677540847639302e-06, "loss": 2.6694, "step": 251070 }, { "epoch": 0.0014848, "grad_norm": 0.8740824460983276, "learning_rate": 5.676814697143091e-06, "loss": 2.7928, "step": 251080 }, { "epoch": 0.0015104, "grad_norm": 0.8289759159088135, "learning_rate": 5.676088574681586e-06, "loss": 2.3166, "step": 251090 }, { "epoch": 0.001536, "grad_norm": 0.8872091770172119, "learning_rate": 5.675362480259499e-06, "loss": 2.6779, "step": 251100 }, { "epoch": 2.56e-05, "grad_norm": 0.8197603225708008, "learning_rate": 5.674636413881536e-06, "loss": 1.9092, "step": 251110 }, { "epoch": 5.12e-05, "grad_norm": 0.9169403910636902, "learning_rate": 5.673910375552407e-06, "loss": 2.833, "step": 251120 }, { "epoch": 7.68e-05, "grad_norm": 1.6720291376113892, "learning_rate": 5.673184365276824e-06, "loss": 2.8008, "step": 251130 }, { "epoch": 0.0001024, "grad_norm": 0.8171964883804321, "learning_rate": 5.672458383059485e-06, "loss": 2.4935, "step": 251140 }, { "epoch": 0.000128, "grad_norm": 0.7411216497421265, "learning_rate": 5.6717324289051036e-06, "loss": 2.5375, "step": 251150 }, { "epoch": 0.0001536, "grad_norm": 0.7843350172042847, "learning_rate": 5.671006502818383e-06, "loss": 2.6879, "step": 251160 }, { "epoch": 0.0001792, "grad_norm": 0.8704582452774048, "learning_rate": 5.67028060480404e-06, "loss": 2.4488, "step": 251170 }, { "epoch": 0.0002048, "grad_norm": 1.170534610748291, "learning_rate": 5.66955473486678e-06, "loss": 2.6174, "step": 251180 }, { "epoch": 0.0002304, "grad_norm": 0.8150536417961121, "learning_rate": 5.668828893011302e-06, "loss": 2.5149, "step": 251190 }, { "epoch": 0.000256, "grad_norm": 0.8317787051200867, "learning_rate": 5.668103079242319e-06, "loss": 2.5096, "step": 251200 }, { "epoch": 0.0002816, "grad_norm": 0.7808004021644592, "learning_rate": 5.6673772935645354e-06, "loss": 2.5722, "step": 251210 }, { "epoch": 0.0003072, "grad_norm": 0.7762937545776367, "learning_rate": 5.666651535982658e-06, "loss": 2.4728, "step": 251220 }, { "epoch": 0.0003328, "grad_norm": 1.0508809089660645, "learning_rate": 5.665925806501394e-06, "loss": 2.6568, "step": 251230 }, { "epoch": 0.0003584, "grad_norm": 1.072281837463379, "learning_rate": 5.665200105125449e-06, "loss": 2.6349, "step": 251240 }, { "epoch": 0.000384, "grad_norm": 0.9085014462471008, "learning_rate": 5.664474431859535e-06, "loss": 2.5609, "step": 251250 }, { "epoch": 0.0004096, "grad_norm": 0.9405331015586853, "learning_rate": 5.663748786708345e-06, "loss": 2.672, "step": 251260 }, { "epoch": 0.0004352, "grad_norm": 0.9198781847953796, "learning_rate": 5.663023169676593e-06, "loss": 2.7115, "step": 251270 }, { "epoch": 0.0004608, "grad_norm": 0.7940968871116638, "learning_rate": 5.662297580768985e-06, "loss": 2.4698, "step": 251280 }, { "epoch": 0.0004864, "grad_norm": 0.9283983707427979, "learning_rate": 5.6615720199902226e-06, "loss": 2.6202, "step": 251290 }, { "epoch": 0.000512, "grad_norm": 0.9377641677856445, "learning_rate": 5.660846487345013e-06, "loss": 2.6626, "step": 251300 }, { "epoch": 0.0005376, "grad_norm": 0.9325255751609802, "learning_rate": 5.66012098283806e-06, "loss": 2.5544, "step": 251310 }, { "epoch": 0.0005632, "grad_norm": 0.8519333600997925, "learning_rate": 5.65939550647407e-06, "loss": 2.6408, "step": 251320 }, { "epoch": 0.0005888, "grad_norm": 0.9802019596099854, "learning_rate": 5.658670058257751e-06, "loss": 2.8103, "step": 251330 }, { "epoch": 0.0006144, "grad_norm": 0.809746265411377, "learning_rate": 5.657944638193793e-06, "loss": 2.6243, "step": 251340 }, { "epoch": 0.00064, "grad_norm": 0.9932870864868164, "learning_rate": 5.657219246286916e-06, "loss": 2.7187, "step": 251350 }, { "epoch": 0.0006656, "grad_norm": 0.8515182733535767, "learning_rate": 5.656493882541817e-06, "loss": 2.8731, "step": 251360 }, { "epoch": 0.0006912, "grad_norm": 0.9736989140510559, "learning_rate": 5.6557685469632e-06, "loss": 2.6927, "step": 251370 }, { "epoch": 0.0007168, "grad_norm": 0.8543142676353455, "learning_rate": 5.6550432395557755e-06, "loss": 2.7811, "step": 251380 }, { "epoch": 0.0007424, "grad_norm": 0.9764845371246338, "learning_rate": 5.654317960324236e-06, "loss": 2.7755, "step": 251390 }, { "epoch": 0.000768, "grad_norm": 0.7656722664833069, "learning_rate": 5.653592709273289e-06, "loss": 2.5209, "step": 251400 }, { "epoch": 0.0007936, "grad_norm": 0.8767892122268677, "learning_rate": 5.6528674864076385e-06, "loss": 2.597, "step": 251410 }, { "epoch": 0.0008192, "grad_norm": 0.8009806871414185, "learning_rate": 5.652142291731987e-06, "loss": 2.6691, "step": 251420 }, { "epoch": 0.0008448, "grad_norm": 0.8074222207069397, "learning_rate": 5.651417125251038e-06, "loss": 2.6715, "step": 251430 }, { "epoch": 0.0008704, "grad_norm": 0.7761775851249695, "learning_rate": 5.650691986969492e-06, "loss": 2.601, "step": 251440 }, { "epoch": 0.000896, "grad_norm": 0.8278769254684448, "learning_rate": 5.649966876892057e-06, "loss": 2.576, "step": 251450 }, { "epoch": 0.0009216, "grad_norm": 0.8870946168899536, "learning_rate": 5.649241795023426e-06, "loss": 2.4155, "step": 251460 }, { "epoch": 0.0009472, "grad_norm": 0.8427074551582336, "learning_rate": 5.648516741368306e-06, "loss": 2.6756, "step": 251470 }, { "epoch": 0.0009728, "grad_norm": 0.7415788769721985, "learning_rate": 5.647791715931399e-06, "loss": 2.2975, "step": 251480 }, { "epoch": 0.0009984, "grad_norm": 1.0861393213272095, "learning_rate": 5.647066718717407e-06, "loss": 2.4965, "step": 251490 }, { "epoch": 0.001024, "grad_norm": 1.0020185708999634, "learning_rate": 5.646341749731028e-06, "loss": 2.6867, "step": 251500 }, { "epoch": 0.0010496, "grad_norm": 1.074581265449524, "learning_rate": 5.645616808976968e-06, "loss": 2.8248, "step": 251510 }, { "epoch": 0.0010752, "grad_norm": 0.8766847252845764, "learning_rate": 5.644891896459924e-06, "loss": 2.7291, "step": 251520 }, { "epoch": 0.0011008, "grad_norm": 0.8619017601013184, "learning_rate": 5.644167012184599e-06, "loss": 2.887, "step": 251530 }, { "epoch": 0.0011264, "grad_norm": 0.8278294801712036, "learning_rate": 5.643442156155694e-06, "loss": 2.7498, "step": 251540 }, { "epoch": 0.001152, "grad_norm": 0.7566702961921692, "learning_rate": 5.642717328377909e-06, "loss": 2.6061, "step": 251550 }, { "epoch": 0.0011776, "grad_norm": 0.7451735734939575, "learning_rate": 5.641992528855943e-06, "loss": 2.5342, "step": 251560 }, { "epoch": 0.0012032, "grad_norm": 0.8897053599357605, "learning_rate": 5.641267757594497e-06, "loss": 3.0501, "step": 251570 }, { "epoch": 0.0012288, "grad_norm": 0.8126269578933716, "learning_rate": 5.640543014598276e-06, "loss": 2.7611, "step": 251580 }, { "epoch": 0.0012544, "grad_norm": 1.066567063331604, "learning_rate": 5.639818299871971e-06, "loss": 2.6971, "step": 251590 }, { "epoch": 0.00128, "grad_norm": 0.810336709022522, "learning_rate": 5.639093613420285e-06, "loss": 2.6451, "step": 251600 }, { "epoch": 0.0013056, "grad_norm": 0.9207715392112732, "learning_rate": 5.638368955247919e-06, "loss": 2.807, "step": 251610 }, { "epoch": 0.0013312, "grad_norm": 1.2538968324661255, "learning_rate": 5.63764432535957e-06, "loss": 2.6987, "step": 251620 }, { "epoch": 0.0013568, "grad_norm": 0.8196830153465271, "learning_rate": 5.636919723759939e-06, "loss": 2.6752, "step": 251630 }, { "epoch": 0.0013824, "grad_norm": 0.946433424949646, "learning_rate": 5.636195150453724e-06, "loss": 2.645, "step": 251640 }, { "epoch": 0.001408, "grad_norm": 0.747045636177063, "learning_rate": 5.635470605445628e-06, "loss": 2.8689, "step": 251650 }, { "epoch": 0.0014336, "grad_norm": 0.9053503274917603, "learning_rate": 5.63474608874034e-06, "loss": 2.6392, "step": 251660 }, { "epoch": 0.0014592, "grad_norm": 0.7486228942871094, "learning_rate": 5.634021600342563e-06, "loss": 2.6105, "step": 251670 }, { "epoch": 0.0014848, "grad_norm": 0.871229887008667, "learning_rate": 5.633297140256997e-06, "loss": 2.6478, "step": 251680 }, { "epoch": 0.0015104, "grad_norm": 0.7908428311347961, "learning_rate": 5.632572708488334e-06, "loss": 2.377, "step": 251690 }, { "epoch": 0.001536, "grad_norm": 1.1203826665878296, "learning_rate": 5.631848305041281e-06, "loss": 2.1382, "step": 251700 }, { "epoch": 0.0015616, "grad_norm": 0.8869194984436035, "learning_rate": 5.631123929920534e-06, "loss": 2.6579, "step": 251710 }, { "epoch": 0.0015872, "grad_norm": 0.832618236541748, "learning_rate": 5.630399583130784e-06, "loss": 2.7048, "step": 251720 }, { "epoch": 0.0016128, "grad_norm": 0.8985399007797241, "learning_rate": 5.629675264676731e-06, "loss": 3.0304, "step": 251730 }, { "epoch": 0.0016384, "grad_norm": 0.7864298820495605, "learning_rate": 5.628950974563071e-06, "loss": 2.8138, "step": 251740 }, { "epoch": 0.001664, "grad_norm": 0.9117552042007446, "learning_rate": 5.6282267127945034e-06, "loss": 2.9982, "step": 251750 }, { "epoch": 0.0016896, "grad_norm": 0.8178107738494873, "learning_rate": 5.627502479375724e-06, "loss": 2.5595, "step": 251760 }, { "epoch": 0.0017152, "grad_norm": 0.8937843441963196, "learning_rate": 5.626778274311428e-06, "loss": 2.6523, "step": 251770 }, { "epoch": 0.0017408, "grad_norm": 0.8570241928100586, "learning_rate": 5.626054097606317e-06, "loss": 2.6345, "step": 251780 }, { "epoch": 0.0017664, "grad_norm": 0.7392259836196899, "learning_rate": 5.625329949265079e-06, "loss": 2.7148, "step": 251790 }, { "epoch": 0.001792, "grad_norm": 0.8599974513053894, "learning_rate": 5.624605829292415e-06, "loss": 2.8006, "step": 251800 }, { "epoch": 0.0018176, "grad_norm": 0.9000558257102966, "learning_rate": 5.6238817376930176e-06, "loss": 2.4317, "step": 251810 }, { "epoch": 0.0018432, "grad_norm": 0.7710070013999939, "learning_rate": 5.623157674471584e-06, "loss": 2.863, "step": 251820 }, { "epoch": 0.0018688, "grad_norm": 0.7482039928436279, "learning_rate": 5.62243363963281e-06, "loss": 2.4538, "step": 251830 }, { "epoch": 0.0018944, "grad_norm": 1.108917236328125, "learning_rate": 5.621709633181392e-06, "loss": 2.7861, "step": 251840 }, { "epoch": 0.00192, "grad_norm": 1.3947474956512451, "learning_rate": 5.620985655122025e-06, "loss": 2.1033, "step": 251850 }, { "epoch": 0.0019456, "grad_norm": 0.8057843446731567, "learning_rate": 5.620261705459396e-06, "loss": 2.8624, "step": 251860 }, { "epoch": 0.0019712, "grad_norm": 0.8732367157936096, "learning_rate": 5.619537784198208e-06, "loss": 2.8096, "step": 251870 }, { "epoch": 0.0019968, "grad_norm": 0.8967739939689636, "learning_rate": 5.618813891343156e-06, "loss": 2.5142, "step": 251880 }, { "epoch": 0.0020224, "grad_norm": 0.7951758503913879, "learning_rate": 5.618090026898928e-06, "loss": 2.6099, "step": 251890 }, { "epoch": 0.002048, "grad_norm": 0.7338095307350159, "learning_rate": 5.617366190870224e-06, "loss": 2.8007, "step": 251900 }, { "epoch": 0.0020736, "grad_norm": 0.7970879673957825, "learning_rate": 5.616642383261739e-06, "loss": 2.4707, "step": 251910 }, { "epoch": 0.0020992, "grad_norm": 0.8537598848342896, "learning_rate": 5.615918604078159e-06, "loss": 2.6617, "step": 251920 }, { "epoch": 0.0021248, "grad_norm": 0.808557391166687, "learning_rate": 5.6151948533241814e-06, "loss": 2.785, "step": 251930 }, { "epoch": 0.0021504, "grad_norm": 0.8052600622177124, "learning_rate": 5.614471131004502e-06, "loss": 2.9926, "step": 251940 }, { "epoch": 0.002176, "grad_norm": 0.8121561408042908, "learning_rate": 5.6137474371238084e-06, "loss": 2.7284, "step": 251950 }, { "epoch": 0.0022016, "grad_norm": 0.7954313158988953, "learning_rate": 5.6130237716868e-06, "loss": 3.015, "step": 251960 }, { "epoch": 0.0022272, "grad_norm": 0.8266835808753967, "learning_rate": 5.6123001346981655e-06, "loss": 2.902, "step": 251970 }, { "epoch": 0.0022528, "grad_norm": 0.7276848554611206, "learning_rate": 5.611576526162601e-06, "loss": 2.5998, "step": 251980 }, { "epoch": 0.0022784, "grad_norm": 0.8845354914665222, "learning_rate": 5.610852946084794e-06, "loss": 3.1115, "step": 251990 }, { "epoch": 0.002304, "grad_norm": 0.7916066646575928, "learning_rate": 5.610129394469439e-06, "loss": 2.4854, "step": 252000 }, { "epoch": 0.0023296, "grad_norm": 0.8181713223457336, "learning_rate": 5.609405871321228e-06, "loss": 2.31, "step": 252010 }, { "epoch": 0.0023552, "grad_norm": 1.0123165845870972, "learning_rate": 5.608682376644853e-06, "loss": 2.7809, "step": 252020 }, { "epoch": 0.0023808, "grad_norm": 0.8031031489372253, "learning_rate": 5.607958910445002e-06, "loss": 2.3141, "step": 252030 }, { "epoch": 0.0024064, "grad_norm": 0.7982204556465149, "learning_rate": 5.6072354727263776e-06, "loss": 2.755, "step": 252040 }, { "epoch": 0.002432, "grad_norm": 0.9588553309440613, "learning_rate": 5.60651206349366e-06, "loss": 2.7227, "step": 252050 }, { "epoch": 0.0024576, "grad_norm": 0.7507890462875366, "learning_rate": 5.605788682751543e-06, "loss": 2.5858, "step": 252060 }, { "epoch": 0.0024832, "grad_norm": 0.7927099466323853, "learning_rate": 5.60506533050472e-06, "loss": 2.5954, "step": 252070 }, { "epoch": 0.0025088, "grad_norm": 0.8024464249610901, "learning_rate": 5.60434200675788e-06, "loss": 2.7619, "step": 252080 }, { "epoch": 0.0025344, "grad_norm": 0.8096972107887268, "learning_rate": 5.603618711515713e-06, "loss": 2.8209, "step": 252090 }, { "epoch": 0.00256, "grad_norm": 0.8291794061660767, "learning_rate": 5.602895444782912e-06, "loss": 2.7121, "step": 252100 }, { "epoch": 0.0025856, "grad_norm": 0.8772698640823364, "learning_rate": 5.6021722065641685e-06, "loss": 2.7256, "step": 252110 }, { "epoch": 0.0026112, "grad_norm": 0.9261296391487122, "learning_rate": 5.601448996864165e-06, "loss": 2.7145, "step": 252120 }, { "epoch": 0.0026368, "grad_norm": 0.7888281941413879, "learning_rate": 5.600725815687596e-06, "loss": 2.6172, "step": 252130 }, { "epoch": 0.0026624, "grad_norm": 0.745463490486145, "learning_rate": 5.600002663039151e-06, "loss": 2.7399, "step": 252140 }, { "epoch": 0.002688, "grad_norm": 0.823722779750824, "learning_rate": 5.599279538923519e-06, "loss": 2.769, "step": 252150 }, { "epoch": 0.0027136, "grad_norm": 0.9725760817527771, "learning_rate": 5.598556443345391e-06, "loss": 2.8675, "step": 252160 }, { "epoch": 0.0027392, "grad_norm": 0.9554885029792786, "learning_rate": 5.597833376309454e-06, "loss": 3.077, "step": 252170 }, { "epoch": 0.0027648, "grad_norm": 0.8172775506973267, "learning_rate": 5.5971103378204e-06, "loss": 2.9454, "step": 252180 }, { "epoch": 0.0027904, "grad_norm": 1.216816782951355, "learning_rate": 5.596387327882913e-06, "loss": 2.7847, "step": 252190 }, { "epoch": 0.002816, "grad_norm": 1.0757381916046143, "learning_rate": 5.595664346501684e-06, "loss": 2.9973, "step": 252200 }, { "epoch": 0.0028416, "grad_norm": 0.7728971838951111, "learning_rate": 5.594941393681397e-06, "loss": 2.7576, "step": 252210 }, { "epoch": 0.0028672, "grad_norm": 0.8290134072303772, "learning_rate": 5.594218469426747e-06, "loss": 2.8395, "step": 252220 }, { "epoch": 0.0028928, "grad_norm": 0.8468372225761414, "learning_rate": 5.59349557374242e-06, "loss": 2.8457, "step": 252230 }, { "epoch": 0.0029184, "grad_norm": 0.8827172517776489, "learning_rate": 5.592772706633107e-06, "loss": 2.8435, "step": 252240 }, { "epoch": 0.002944, "grad_norm": 0.8209259510040283, "learning_rate": 5.592049868103489e-06, "loss": 3.0159, "step": 252250 }, { "epoch": 0.0029696, "grad_norm": 0.8882826566696167, "learning_rate": 5.591327058158254e-06, "loss": 2.7951, "step": 252260 }, { "epoch": 0.0029952, "grad_norm": 0.8513098359107971, "learning_rate": 5.590604276802093e-06, "loss": 2.8511, "step": 252270 }, { "epoch": 0.0030208, "grad_norm": 0.8210028409957886, "learning_rate": 5.589881524039691e-06, "loss": 2.8553, "step": 252280 }, { "epoch": 0.0030464, "grad_norm": 0.9225528836250305, "learning_rate": 5.5891587998757354e-06, "loss": 2.7773, "step": 252290 }, { "epoch": 0.003072, "grad_norm": 0.8499186038970947, "learning_rate": 5.588436104314912e-06, "loss": 2.8989, "step": 252300 }, { "epoch": 0.0030976, "grad_norm": 1.086411476135254, "learning_rate": 5.587713437361912e-06, "loss": 2.8948, "step": 252310 }, { "epoch": 0.0031232, "grad_norm": 0.8191314935684204, "learning_rate": 5.586990799021415e-06, "loss": 2.8179, "step": 252320 }, { "epoch": 0.0031488, "grad_norm": 0.7813053727149963, "learning_rate": 5.586268189298109e-06, "loss": 2.952, "step": 252330 }, { "epoch": 0.0031744, "grad_norm": 0.7826826572418213, "learning_rate": 5.585545608196681e-06, "loss": 3.1824, "step": 252340 }, { "epoch": 0.0032, "grad_norm": 0.8516574501991272, "learning_rate": 5.584823055721815e-06, "loss": 2.8965, "step": 252350 }, { "epoch": 0.0032256, "grad_norm": 1.0766922235488892, "learning_rate": 5.5841005318782e-06, "loss": 2.967, "step": 252360 }, { "epoch": 0.0032512, "grad_norm": 0.7793176174163818, "learning_rate": 5.583378036670519e-06, "loss": 2.9136, "step": 252370 }, { "epoch": 0.0032768, "grad_norm": 0.8300490975379944, "learning_rate": 5.582655570103457e-06, "loss": 3.0922, "step": 252380 }, { "epoch": 0.0033024, "grad_norm": 0.8516215085983276, "learning_rate": 5.581933132181701e-06, "loss": 2.7392, "step": 252390 }, { "epoch": 0.003328, "grad_norm": 1.1526188850402832, "learning_rate": 5.581210722909933e-06, "loss": 2.7904, "step": 252400 }, { "epoch": 0.0033536, "grad_norm": 0.8510170578956604, "learning_rate": 5.58048834229284e-06, "loss": 3.0983, "step": 252410 }, { "epoch": 0.0033792, "grad_norm": 0.7951958179473877, "learning_rate": 5.579765990335104e-06, "loss": 2.8375, "step": 252420 }, { "epoch": 0.0034048, "grad_norm": 0.8797087073326111, "learning_rate": 5.579043667041413e-06, "loss": 3.0007, "step": 252430 }, { "epoch": 0.0034304, "grad_norm": 0.8502205014228821, "learning_rate": 5.578321372416451e-06, "loss": 3.0784, "step": 252440 }, { "epoch": 0.003456, "grad_norm": 0.7689149379730225, "learning_rate": 5.577599106464897e-06, "loss": 2.5504, "step": 252450 }, { "epoch": 0.0034816, "grad_norm": 0.7628046870231628, "learning_rate": 5.576876869191437e-06, "loss": 2.8751, "step": 252460 }, { "epoch": 0.0035072, "grad_norm": 0.790644109249115, "learning_rate": 5.576154660600754e-06, "loss": 2.8134, "step": 252470 }, { "epoch": 0.0035328, "grad_norm": 0.8595258593559265, "learning_rate": 5.575432480697534e-06, "loss": 2.6523, "step": 252480 }, { "epoch": 0.0035584, "grad_norm": 0.8792805671691895, "learning_rate": 5.574710329486458e-06, "loss": 2.7168, "step": 252490 }, { "epoch": 0.003584, "grad_norm": 1.0391374826431274, "learning_rate": 5.573988206972209e-06, "loss": 3.0067, "step": 252500 }, { "epoch": 0.0036096, "grad_norm": 0.8758174777030945, "learning_rate": 5.573266113159474e-06, "loss": 2.7691, "step": 252510 }, { "epoch": 0.0036352, "grad_norm": 0.8538258671760559, "learning_rate": 5.5725440480529295e-06, "loss": 2.7534, "step": 252520 }, { "epoch": 0.0036608, "grad_norm": 0.8828665614128113, "learning_rate": 5.571822011657258e-06, "loss": 2.8059, "step": 252530 }, { "epoch": 0.0036864, "grad_norm": 0.778547465801239, "learning_rate": 5.571100003977146e-06, "loss": 2.8577, "step": 252540 }, { "epoch": 0.003712, "grad_norm": 0.7882676720619202, "learning_rate": 5.570378025017271e-06, "loss": 2.7582, "step": 252550 }, { "epoch": 0.0037376, "grad_norm": 0.8383966684341431, "learning_rate": 5.5696560747823195e-06, "loss": 2.9247, "step": 252560 }, { "epoch": 0.0037632, "grad_norm": 0.8791062831878662, "learning_rate": 5.568934153276968e-06, "loss": 2.6027, "step": 252570 }, { "epoch": 0.0037888, "grad_norm": 0.817534327507019, "learning_rate": 5.568212260505903e-06, "loss": 2.8461, "step": 252580 }, { "epoch": 0.0038144, "grad_norm": 0.7768037915229797, "learning_rate": 5.567490396473804e-06, "loss": 3.0511, "step": 252590 }, { "epoch": 0.00384, "grad_norm": 0.8504695296287537, "learning_rate": 5.566768561185351e-06, "loss": 2.9655, "step": 252600 }, { "epoch": 0.0038656, "grad_norm": 0.8237985372543335, "learning_rate": 5.566046754645225e-06, "loss": 2.7374, "step": 252610 }, { "epoch": 0.0038912, "grad_norm": 0.8778043389320374, "learning_rate": 5.565324976858109e-06, "loss": 3.0526, "step": 252620 }, { "epoch": 0.0039168, "grad_norm": 0.9217174649238586, "learning_rate": 5.56460322782868e-06, "loss": 2.6849, "step": 252630 }, { "epoch": 0.0039424, "grad_norm": 0.763072669506073, "learning_rate": 5.563881507561626e-06, "loss": 2.932, "step": 252640 }, { "epoch": 0.003968, "grad_norm": 0.7670396566390991, "learning_rate": 5.563159816061616e-06, "loss": 2.9266, "step": 252650 }, { "epoch": 0.0039936, "grad_norm": 1.3014153242111206, "learning_rate": 5.562438153333337e-06, "loss": 2.6279, "step": 252660 }, { "epoch": 0.0040192, "grad_norm": 0.8985641002655029, "learning_rate": 5.561716519381466e-06, "loss": 2.914, "step": 252670 }, { "epoch": 0.0040448, "grad_norm": 0.8045148253440857, "learning_rate": 5.560994914210685e-06, "loss": 2.8583, "step": 252680 }, { "epoch": 0.0040704, "grad_norm": 0.7857016324996948, "learning_rate": 5.560273337825671e-06, "loss": 2.9927, "step": 252690 }, { "epoch": 0.004096, "grad_norm": 0.874780535697937, "learning_rate": 5.559551790231106e-06, "loss": 2.6623, "step": 252700 }, { "epoch": 0.0041216, "grad_norm": 1.1400573253631592, "learning_rate": 5.55883027143167e-06, "loss": 2.9742, "step": 252710 }, { "epoch": 0.0041472, "grad_norm": 0.9472581148147583, "learning_rate": 5.558108781432036e-06, "loss": 2.8151, "step": 252720 }, { "epoch": 0.0041728, "grad_norm": 0.9494690895080566, "learning_rate": 5.557387320236881e-06, "loss": 2.8288, "step": 252730 }, { "epoch": 0.0041984, "grad_norm": 0.8012667298316956, "learning_rate": 5.556665887850894e-06, "loss": 2.8356, "step": 252740 }, { "epoch": 0.004224, "grad_norm": 5.138787269592285, "learning_rate": 5.555944484278747e-06, "loss": 2.9192, "step": 252750 }, { "epoch": 0.0042496, "grad_norm": 0.7528912425041199, "learning_rate": 5.555223109525123e-06, "loss": 2.8585, "step": 252760 }, { "epoch": 0.0042752, "grad_norm": 0.8609076738357544, "learning_rate": 5.554501763594692e-06, "loss": 2.8455, "step": 252770 }, { "epoch": 0.0043008, "grad_norm": 0.974615752696991, "learning_rate": 5.553780446492136e-06, "loss": 2.8199, "step": 252780 }, { "epoch": 0.0043264, "grad_norm": 1.24714195728302, "learning_rate": 5.5530591582221315e-06, "loss": 2.9014, "step": 252790 }, { "epoch": 0.004352, "grad_norm": 0.9018168449401855, "learning_rate": 5.552337898789357e-06, "loss": 2.7652, "step": 252800 }, { "epoch": 0.0043776, "grad_norm": 0.7629708051681519, "learning_rate": 5.551616668198488e-06, "loss": 3.0547, "step": 252810 }, { "epoch": 0.0044032, "grad_norm": 0.8360867500305176, "learning_rate": 5.550895466454204e-06, "loss": 2.8338, "step": 252820 }, { "epoch": 0.0044288, "grad_norm": 0.830823540687561, "learning_rate": 5.5501742935611835e-06, "loss": 3.0273, "step": 252830 }, { "epoch": 0.0044544, "grad_norm": 1.361641526222229, "learning_rate": 5.5494531495240976e-06, "loss": 2.5971, "step": 252840 }, { "epoch": 0.00448, "grad_norm": 0.8569912314414978, "learning_rate": 5.548732034347624e-06, "loss": 2.9793, "step": 252850 }, { "epoch": 0.0045056, "grad_norm": 1.0205622911453247, "learning_rate": 5.54801094803644e-06, "loss": 2.9599, "step": 252860 }, { "epoch": 0.0045312, "grad_norm": 0.9617211818695068, "learning_rate": 5.5472898905952225e-06, "loss": 2.6414, "step": 252870 }, { "epoch": 0.0045568, "grad_norm": 0.8978268504142761, "learning_rate": 5.546568862028646e-06, "loss": 2.7418, "step": 252880 }, { "epoch": 0.0045824, "grad_norm": 0.8080927729606628, "learning_rate": 5.545847862341387e-06, "loss": 2.7285, "step": 252890 }, { "epoch": 0.004608, "grad_norm": 0.8656396269798279, "learning_rate": 5.545126891538122e-06, "loss": 2.6538, "step": 252900 }, { "epoch": 0.0046336, "grad_norm": 0.822719395160675, "learning_rate": 5.544405949623524e-06, "loss": 2.9848, "step": 252910 }, { "epoch": 0.0046592, "grad_norm": 1.6629468202590942, "learning_rate": 5.543685036602269e-06, "loss": 2.9857, "step": 252920 }, { "epoch": 0.0046848, "grad_norm": 0.8307455778121948, "learning_rate": 5.542964152479033e-06, "loss": 2.7038, "step": 252930 }, { "epoch": 0.0047104, "grad_norm": 0.9712585210800171, "learning_rate": 5.54224329725849e-06, "loss": 3.0291, "step": 252940 }, { "epoch": 0.004736, "grad_norm": 1.063805341720581, "learning_rate": 5.541522470945313e-06, "loss": 2.9323, "step": 252950 }, { "epoch": 0.0047616, "grad_norm": 0.7797554731369019, "learning_rate": 5.540801673544183e-06, "loss": 2.8321, "step": 252960 }, { "epoch": 0.0047872, "grad_norm": 0.8829580545425415, "learning_rate": 5.540080905059766e-06, "loss": 2.864, "step": 252970 }, { "epoch": 0.0048128, "grad_norm": 0.7779219150543213, "learning_rate": 5.5393601654967375e-06, "loss": 2.6729, "step": 252980 }, { "epoch": 0.0048384, "grad_norm": 0.8184079527854919, "learning_rate": 5.538639454859773e-06, "loss": 2.9167, "step": 252990 }, { "epoch": 0.004864, "grad_norm": 0.9360586404800415, "learning_rate": 5.537918773153547e-06, "loss": 2.7659, "step": 253000 }, { "epoch": 0.0048896, "grad_norm": 0.7798564434051514, "learning_rate": 5.537198120382731e-06, "loss": 2.6832, "step": 253010 }, { "epoch": 0.0049152, "grad_norm": 0.8280392289161682, "learning_rate": 5.5364774965519995e-06, "loss": 2.72, "step": 253020 }, { "epoch": 0.0049408, "grad_norm": 0.7333475351333618, "learning_rate": 5.535756901666029e-06, "loss": 2.6965, "step": 253030 }, { "epoch": 0.0049664, "grad_norm": 0.8765440583229065, "learning_rate": 5.535036335729485e-06, "loss": 3.1285, "step": 253040 }, { "epoch": 0.004992, "grad_norm": 0.7551947832107544, "learning_rate": 5.534315798747042e-06, "loss": 2.8031, "step": 253050 }, { "epoch": 0.0050176, "grad_norm": 0.8152309060096741, "learning_rate": 5.533595290723375e-06, "loss": 2.8048, "step": 253060 }, { "epoch": 0.0050432, "grad_norm": 0.7546278238296509, "learning_rate": 5.532874811663156e-06, "loss": 2.8649, "step": 253070 }, { "epoch": 0.0050688, "grad_norm": 0.8296380043029785, "learning_rate": 5.532154361571054e-06, "loss": 2.9176, "step": 253080 }, { "epoch": 0.0050944, "grad_norm": 0.7554542422294617, "learning_rate": 5.531433940451749e-06, "loss": 2.745, "step": 253090 }, { "epoch": 0.00512, "grad_norm": 0.8869355916976929, "learning_rate": 5.530713548309905e-06, "loss": 2.8256, "step": 253100 }, { "epoch": 0.0051456, "grad_norm": 0.7666786313056946, "learning_rate": 5.529993185150196e-06, "loss": 2.8601, "step": 253110 }, { "epoch": 0.0051712, "grad_norm": 0.7713285684585571, "learning_rate": 5.529272850977292e-06, "loss": 2.6989, "step": 253120 }, { "epoch": 0.0051968, "grad_norm": 0.8237841725349426, "learning_rate": 5.5285525457958645e-06, "loss": 2.9749, "step": 253130 }, { "epoch": 0.0052224, "grad_norm": 0.9573932886123657, "learning_rate": 5.527832269610587e-06, "loss": 2.8091, "step": 253140 }, { "epoch": 0.005248, "grad_norm": 0.7770906686782837, "learning_rate": 5.5271120224261285e-06, "loss": 2.6792, "step": 253150 }, { "epoch": 0.0052736, "grad_norm": 1.074803352355957, "learning_rate": 5.526391804247164e-06, "loss": 2.6113, "step": 253160 }, { "epoch": 0.0052992, "grad_norm": 1.104288101196289, "learning_rate": 5.525671615078355e-06, "loss": 2.8206, "step": 253170 }, { "epoch": 0.0053248, "grad_norm": 0.9730214476585388, "learning_rate": 5.524951454924376e-06, "loss": 2.5986, "step": 253180 }, { "epoch": 0.0053504, "grad_norm": 0.7861621975898743, "learning_rate": 5.5242313237898995e-06, "loss": 2.9159, "step": 253190 }, { "epoch": 0.005376, "grad_norm": 0.7121822237968445, "learning_rate": 5.523511221679592e-06, "loss": 2.8671, "step": 253200 }, { "epoch": 0.0054016, "grad_norm": 1.0176060199737549, "learning_rate": 5.522791148598124e-06, "loss": 2.8388, "step": 253210 }, { "epoch": 0.0054272, "grad_norm": 0.760061502456665, "learning_rate": 5.522071104550166e-06, "loss": 2.8961, "step": 253220 }, { "epoch": 0.0054528, "grad_norm": 0.8095670342445374, "learning_rate": 5.521351089540392e-06, "loss": 2.8681, "step": 253230 }, { "epoch": 0.0054784, "grad_norm": 0.9253013134002686, "learning_rate": 5.520631103573461e-06, "loss": 2.8969, "step": 253240 }, { "epoch": 0.005504, "grad_norm": 0.8017123937606812, "learning_rate": 5.519911146654042e-06, "loss": 2.6758, "step": 253250 }, { "epoch": 0.0055296, "grad_norm": 0.7806591987609863, "learning_rate": 5.519191218786814e-06, "loss": 2.821, "step": 253260 }, { "epoch": 0.0055552, "grad_norm": 0.8069443106651306, "learning_rate": 5.51847131997644e-06, "loss": 2.7577, "step": 253270 }, { "epoch": 0.0055808, "grad_norm": 0.8705791234970093, "learning_rate": 5.5177514502275865e-06, "loss": 2.7986, "step": 253280 }, { "epoch": 0.0056064, "grad_norm": 0.7925358414649963, "learning_rate": 5.51703160954493e-06, "loss": 2.8833, "step": 253290 }, { "epoch": 0.005632, "grad_norm": 0.8236842751502991, "learning_rate": 5.516311797933126e-06, "loss": 2.857, "step": 253300 }, { "epoch": 0.0056576, "grad_norm": 0.7986200451850891, "learning_rate": 5.515592015396849e-06, "loss": 3.0409, "step": 253310 }, { "epoch": 0.0056832, "grad_norm": 0.9360299110412598, "learning_rate": 5.514872261940767e-06, "loss": 2.7829, "step": 253320 }, { "epoch": 0.0057088, "grad_norm": 0.8600350618362427, "learning_rate": 5.5141525375695455e-06, "loss": 2.8549, "step": 253330 }, { "epoch": 0.0057344, "grad_norm": 0.8461736440658569, "learning_rate": 5.513432842287852e-06, "loss": 2.8296, "step": 253340 }, { "epoch": 0.00576, "grad_norm": 0.8061182498931885, "learning_rate": 5.512713176100355e-06, "loss": 2.5882, "step": 253350 }, { "epoch": 0.0057856, "grad_norm": 0.7790619134902954, "learning_rate": 5.511993539011724e-06, "loss": 2.6623, "step": 253360 }, { "epoch": 0.0058112, "grad_norm": 0.8316906094551086, "learning_rate": 5.511273931026619e-06, "loss": 2.7156, "step": 253370 }, { "epoch": 0.0058368, "grad_norm": 0.7768601775169373, "learning_rate": 5.510554352149709e-06, "loss": 2.8258, "step": 253380 }, { "epoch": 0.0058624, "grad_norm": 0.7457306981086731, "learning_rate": 5.509834802385661e-06, "loss": 2.7382, "step": 253390 }, { "epoch": 0.005888, "grad_norm": 0.7780801653862, "learning_rate": 5.5091152817391415e-06, "loss": 2.876, "step": 253400 }, { "epoch": 0.0059136, "grad_norm": 0.7811010479927063, "learning_rate": 5.508395790214816e-06, "loss": 2.8524, "step": 253410 }, { "epoch": 0.0059392, "grad_norm": 0.877833366394043, "learning_rate": 5.507676327817349e-06, "loss": 2.8246, "step": 253420 }, { "epoch": 0.0059648, "grad_norm": 0.8709690570831299, "learning_rate": 5.506956894551407e-06, "loss": 2.7195, "step": 253430 }, { "epoch": 0.0059904, "grad_norm": 0.8119745850563049, "learning_rate": 5.506237490421656e-06, "loss": 3.0221, "step": 253440 }, { "epoch": 0.006016, "grad_norm": 0.7887313365936279, "learning_rate": 5.50551811543276e-06, "loss": 2.7704, "step": 253450 }, { "epoch": 0.0060416, "grad_norm": 0.9061846137046814, "learning_rate": 5.504798769589385e-06, "loss": 2.6564, "step": 253460 }, { "epoch": 0.0060672, "grad_norm": 0.8577406406402588, "learning_rate": 5.504079452896196e-06, "loss": 2.6424, "step": 253470 }, { "epoch": 0.0060928, "grad_norm": 0.9177100658416748, "learning_rate": 5.503360165357856e-06, "loss": 2.8165, "step": 253480 }, { "epoch": 0.0061184, "grad_norm": 0.8471518754959106, "learning_rate": 5.502640906979034e-06, "loss": 2.9656, "step": 253490 }, { "epoch": 0.006144, "grad_norm": 0.8078407645225525, "learning_rate": 5.501921677764387e-06, "loss": 2.6901, "step": 253500 }, { "epoch": 0.0061696, "grad_norm": 0.8548108339309692, "learning_rate": 5.5012024777185825e-06, "loss": 2.6893, "step": 253510 }, { "epoch": 0.0061952, "grad_norm": 1.7464463710784912, "learning_rate": 5.5004833068462845e-06, "loss": 2.7437, "step": 253520 }, { "epoch": 0.0062208, "grad_norm": 0.9118075370788574, "learning_rate": 5.499764165152156e-06, "loss": 2.97, "step": 253530 }, { "epoch": 0.0062464, "grad_norm": 0.7467394471168518, "learning_rate": 5.4990450526408615e-06, "loss": 3.1446, "step": 253540 }, { "epoch": 0.006272, "grad_norm": 0.7494953870773315, "learning_rate": 5.498325969317062e-06, "loss": 2.7282, "step": 253550 }, { "epoch": 0.0062976, "grad_norm": 2.615227699279785, "learning_rate": 5.497606915185427e-06, "loss": 3.3845, "step": 253560 }, { "epoch": 0.0063232, "grad_norm": 0.7797377705574036, "learning_rate": 5.496887890250611e-06, "loss": 2.9351, "step": 253570 }, { "epoch": 0.0063488, "grad_norm": 0.801129162311554, "learning_rate": 5.4961688945172805e-06, "loss": 2.6643, "step": 253580 }, { "epoch": 0.0063744, "grad_norm": 0.9196310043334961, "learning_rate": 5.495449927990097e-06, "loss": 2.9185, "step": 253590 }, { "epoch": 0.0064, "grad_norm": 0.8358805775642395, "learning_rate": 5.49473099067372e-06, "loss": 2.9671, "step": 253600 }, { "epoch": 0.0064256, "grad_norm": 0.7659972906112671, "learning_rate": 5.494012082572819e-06, "loss": 2.8994, "step": 253610 }, { "epoch": 0.0064512, "grad_norm": 1.370888352394104, "learning_rate": 5.493293203692056e-06, "loss": 3.1487, "step": 253620 }, { "epoch": 0.0064768, "grad_norm": 0.7994032502174377, "learning_rate": 5.492574354036086e-06, "loss": 2.7719, "step": 253630 }, { "epoch": 0.0065024, "grad_norm": 0.7677698135375977, "learning_rate": 5.491855533609572e-06, "loss": 2.7687, "step": 253640 }, { "epoch": 0.006528, "grad_norm": 0.851245105266571, "learning_rate": 5.491136742417177e-06, "loss": 2.6831, "step": 253650 }, { "epoch": 0.0065536, "grad_norm": 0.8253768086433411, "learning_rate": 5.490417980463563e-06, "loss": 2.941, "step": 253660 }, { "epoch": 0.0065792, "grad_norm": 1.6068521738052368, "learning_rate": 5.489699247753388e-06, "loss": 2.8279, "step": 253670 }, { "epoch": 0.0066048, "grad_norm": 0.9181848764419556, "learning_rate": 5.488980544291316e-06, "loss": 2.5454, "step": 253680 }, { "epoch": 0.0066304, "grad_norm": 0.8136492371559143, "learning_rate": 5.48826187008201e-06, "loss": 2.8253, "step": 253690 }, { "epoch": 0.006656, "grad_norm": 0.8973974585533142, "learning_rate": 5.487543225130122e-06, "loss": 3.0138, "step": 253700 }, { "epoch": 0.0066816, "grad_norm": 0.7880903482437134, "learning_rate": 5.486824609440318e-06, "loss": 2.7828, "step": 253710 }, { "epoch": 0.0067072, "grad_norm": 0.9381313920021057, "learning_rate": 5.486106023017257e-06, "loss": 2.8951, "step": 253720 }, { "epoch": 0.0067328, "grad_norm": 0.8262411952018738, "learning_rate": 5.485387465865597e-06, "loss": 2.9631, "step": 253730 }, { "epoch": 0.0067584, "grad_norm": 0.8432076573371887, "learning_rate": 5.484668937990001e-06, "loss": 2.8965, "step": 253740 }, { "epoch": 0.006784, "grad_norm": 0.7519770860671997, "learning_rate": 5.483950439395127e-06, "loss": 2.6669, "step": 253750 }, { "epoch": 0.0068096, "grad_norm": 0.8268483281135559, "learning_rate": 5.483231970085638e-06, "loss": 2.8176, "step": 253760 }, { "epoch": 0.0068352, "grad_norm": 0.7631959915161133, "learning_rate": 5.4825135300661806e-06, "loss": 2.6869, "step": 253770 }, { "epoch": 0.0068608, "grad_norm": 0.7870575189590454, "learning_rate": 5.481795119341427e-06, "loss": 2.8461, "step": 253780 }, { "epoch": 0.0068864, "grad_norm": 0.8133823275566101, "learning_rate": 5.481076737916031e-06, "loss": 2.6336, "step": 253790 }, { "epoch": 0.006912, "grad_norm": 0.813419759273529, "learning_rate": 5.480358385794651e-06, "loss": 2.7266, "step": 253800 }, { "epoch": 0.0069376, "grad_norm": 0.853204607963562, "learning_rate": 5.479640062981946e-06, "loss": 2.5103, "step": 253810 }, { "epoch": 0.0069632, "grad_norm": 0.8005762100219727, "learning_rate": 5.478921769482578e-06, "loss": 2.92, "step": 253820 }, { "epoch": 0.0069888, "grad_norm": 0.7646763920783997, "learning_rate": 5.478203505301196e-06, "loss": 2.85, "step": 253830 }, { "epoch": 0.0070144, "grad_norm": 0.8635514974594116, "learning_rate": 5.477485270442464e-06, "loss": 3.0049, "step": 253840 }, { "epoch": 0.00704, "grad_norm": 0.9726852178573608, "learning_rate": 5.476767064911037e-06, "loss": 2.875, "step": 253850 }, { "epoch": 0.0070656, "grad_norm": 0.9263727068901062, "learning_rate": 5.476048888711574e-06, "loss": 3.0128, "step": 253860 }, { "epoch": 0.0070912, "grad_norm": 0.7424221038818359, "learning_rate": 5.4753307418487324e-06, "loss": 2.6318, "step": 253870 }, { "epoch": 0.0071168, "grad_norm": 0.8096689581871033, "learning_rate": 5.474612624327167e-06, "loss": 2.9835, "step": 253880 }, { "epoch": 0.0071424, "grad_norm": 0.7935922145843506, "learning_rate": 5.473894536151541e-06, "loss": 3.0296, "step": 253890 }, { "epoch": 0.007168, "grad_norm": 0.9139288663864136, "learning_rate": 5.4731764773265e-06, "loss": 2.8774, "step": 253900 }, { "epoch": 0.0071936, "grad_norm": 0.790340781211853, "learning_rate": 5.472458447856708e-06, "loss": 2.7552, "step": 253910 }, { "epoch": 0.0072192, "grad_norm": 0.9034151434898376, "learning_rate": 5.471740447746821e-06, "loss": 2.8418, "step": 253920 }, { "epoch": 0.0072448, "grad_norm": 0.8950718641281128, "learning_rate": 5.471022477001491e-06, "loss": 2.8874, "step": 253930 }, { "epoch": 0.0072704, "grad_norm": 0.8220866918563843, "learning_rate": 5.470304535625378e-06, "loss": 3.0291, "step": 253940 }, { "epoch": 0.007296, "grad_norm": 0.8572131395339966, "learning_rate": 5.469586623623136e-06, "loss": 2.7304, "step": 253950 }, { "epoch": 0.0073216, "grad_norm": 0.7961885929107666, "learning_rate": 5.468868740999421e-06, "loss": 2.8274, "step": 253960 }, { "epoch": 0.0073472, "grad_norm": 0.8823227286338806, "learning_rate": 5.468150887758889e-06, "loss": 2.9557, "step": 253970 }, { "epoch": 0.0073728, "grad_norm": 0.7390563488006592, "learning_rate": 5.467433063906192e-06, "loss": 2.6369, "step": 253980 }, { "epoch": 0.0073984, "grad_norm": 0.7438360452651978, "learning_rate": 5.466715269445988e-06, "loss": 2.9924, "step": 253990 }, { "epoch": 0.007424, "grad_norm": 0.7742618918418884, "learning_rate": 5.46599750438293e-06, "loss": 2.565, "step": 254000 }, { "epoch": 0.0074496, "grad_norm": 0.8933141231536865, "learning_rate": 5.465279768721674e-06, "loss": 2.9567, "step": 254010 }, { "epoch": 0.0074752, "grad_norm": 0.905235230922699, "learning_rate": 5.464562062466879e-06, "loss": 2.9232, "step": 254020 }, { "epoch": 0.0075008, "grad_norm": 0.8677110075950623, "learning_rate": 5.463844385623187e-06, "loss": 2.9053, "step": 254030 }, { "epoch": 0.0075264, "grad_norm": 0.7924414277076721, "learning_rate": 5.463126738195259e-06, "loss": 3.0152, "step": 254040 }, { "epoch": 0.007552, "grad_norm": 0.8186025023460388, "learning_rate": 5.462409120187748e-06, "loss": 2.6968, "step": 254050 }, { "epoch": 0.0075776, "grad_norm": 0.7986893057823181, "learning_rate": 5.46169153160531e-06, "loss": 2.7675, "step": 254060 }, { "epoch": 0.0076032, "grad_norm": 0.7633858919143677, "learning_rate": 5.460973972452594e-06, "loss": 2.9507, "step": 254070 }, { "epoch": 0.0076288, "grad_norm": 0.7382264137268066, "learning_rate": 5.4602564427342575e-06, "loss": 2.6248, "step": 254080 }, { "epoch": 0.0076544, "grad_norm": 0.8125272393226624, "learning_rate": 5.459538942454954e-06, "loss": 2.937, "step": 254090 }, { "epoch": 0.00768, "grad_norm": 0.800579309463501, "learning_rate": 5.458821471619331e-06, "loss": 2.839, "step": 254100 }, { "epoch": 0.0077056, "grad_norm": 0.7209054231643677, "learning_rate": 5.458104030232043e-06, "loss": 2.991, "step": 254110 }, { "epoch": 0.0077312, "grad_norm": 0.7248134016990662, "learning_rate": 5.45738661829774e-06, "loss": 2.8969, "step": 254120 }, { "epoch": 0.0077568, "grad_norm": 0.7627120614051819, "learning_rate": 5.456669235821079e-06, "loss": 2.9374, "step": 254130 }, { "epoch": 0.0077824, "grad_norm": 0.8042708039283752, "learning_rate": 5.455951882806717e-06, "loss": 2.8553, "step": 254140 }, { "epoch": 0.007808, "grad_norm": 0.8859302997589111, "learning_rate": 5.455234559259295e-06, "loss": 2.9309, "step": 254150 }, { "epoch": 0.0078336, "grad_norm": 0.7900108695030212, "learning_rate": 5.4545172651834685e-06, "loss": 2.964, "step": 254160 }, { "epoch": 0.0078592, "grad_norm": 0.858005940914154, "learning_rate": 5.4538000005838905e-06, "loss": 2.9497, "step": 254170 }, { "epoch": 0.0078848, "grad_norm": 0.8716608881950378, "learning_rate": 5.4530827654652095e-06, "loss": 2.8587, "step": 254180 }, { "epoch": 0.0079104, "grad_norm": 1.008164644241333, "learning_rate": 5.45236555983208e-06, "loss": 2.7243, "step": 254190 }, { "epoch": 0.007936, "grad_norm": 0.8623702526092529, "learning_rate": 5.45164838368915e-06, "loss": 3.1265, "step": 254200 }, { "epoch": 0.0079616, "grad_norm": 0.7833138704299927, "learning_rate": 5.450931237041073e-06, "loss": 2.8865, "step": 254210 }, { "epoch": 0.0079872, "grad_norm": 0.8919860124588013, "learning_rate": 5.450214119892501e-06, "loss": 2.9378, "step": 254220 }, { "epoch": 0.0080128, "grad_norm": 0.7696031928062439, "learning_rate": 5.449497032248077e-06, "loss": 2.7395, "step": 254230 }, { "epoch": 0.0080384, "grad_norm": 0.7660887241363525, "learning_rate": 5.4487799741124544e-06, "loss": 2.7369, "step": 254240 }, { "epoch": 0.008064, "grad_norm": 0.8301268219947815, "learning_rate": 5.4480629454902865e-06, "loss": 2.9634, "step": 254250 }, { "epoch": 0.0080896, "grad_norm": 0.9134442210197449, "learning_rate": 5.44734594638622e-06, "loss": 2.8402, "step": 254260 }, { "epoch": 0.0081152, "grad_norm": 1.383583664894104, "learning_rate": 5.446628976804904e-06, "loss": 2.8104, "step": 254270 }, { "epoch": 0.0081408, "grad_norm": 0.8216174840927124, "learning_rate": 5.445912036750989e-06, "loss": 2.7318, "step": 254280 }, { "epoch": 0.0081664, "grad_norm": 1.3581105470657349, "learning_rate": 5.445195126229128e-06, "loss": 2.7655, "step": 254290 }, { "epoch": 0.008192, "grad_norm": 0.7536939382553101, "learning_rate": 5.444478245243959e-06, "loss": 3.0601, "step": 254300 }, { "epoch": 0.0082176, "grad_norm": 0.8753871917724609, "learning_rate": 5.443761393800142e-06, "loss": 2.9142, "step": 254310 }, { "epoch": 0.0082432, "grad_norm": 0.7656180262565613, "learning_rate": 5.443044571902321e-06, "loss": 2.9858, "step": 254320 }, { "epoch": 0.0082688, "grad_norm": 0.824992835521698, "learning_rate": 5.442327779555144e-06, "loss": 2.8406, "step": 254330 }, { "epoch": 0.0082944, "grad_norm": 0.753383457660675, "learning_rate": 5.4416110167632655e-06, "loss": 2.9395, "step": 254340 }, { "epoch": 0.00832, "grad_norm": 0.8279721140861511, "learning_rate": 5.440894283531324e-06, "loss": 2.9989, "step": 254350 }, { "epoch": 0.0083456, "grad_norm": 0.7542093992233276, "learning_rate": 5.44017757986397e-06, "loss": 2.907, "step": 254360 }, { "epoch": 0.0083712, "grad_norm": 1.0026785135269165, "learning_rate": 5.439460905765854e-06, "loss": 2.8231, "step": 254370 }, { "epoch": 0.0083968, "grad_norm": 1.74116849899292, "learning_rate": 5.438744261241622e-06, "loss": 2.9977, "step": 254380 }, { "epoch": 0.0084224, "grad_norm": 0.8059913516044617, "learning_rate": 5.43802764629592e-06, "loss": 2.9254, "step": 254390 }, { "epoch": 0.008448, "grad_norm": 0.757011353969574, "learning_rate": 5.437311060933397e-06, "loss": 2.7477, "step": 254400 }, { "epoch": 0.0084736, "grad_norm": 0.8045397400856018, "learning_rate": 5.436594505158699e-06, "loss": 2.7269, "step": 254410 }, { "epoch": 0.0084992, "grad_norm": 0.7133564352989197, "learning_rate": 5.435877978976476e-06, "loss": 2.6885, "step": 254420 }, { "epoch": 0.0085248, "grad_norm": 0.7764645218849182, "learning_rate": 5.435161482391369e-06, "loss": 3.0003, "step": 254430 }, { "epoch": 0.0085504, "grad_norm": 0.7598362565040588, "learning_rate": 5.4344450154080254e-06, "loss": 2.8811, "step": 254440 }, { "epoch": 0.008576, "grad_norm": 0.97175532579422, "learning_rate": 5.433728578031093e-06, "loss": 2.796, "step": 254450 }, { "epoch": 0.0086016, "grad_norm": 0.7921191453933716, "learning_rate": 5.433012170265217e-06, "loss": 2.9095, "step": 254460 }, { "epoch": 0.0086272, "grad_norm": 0.8782895803451538, "learning_rate": 5.432295792115043e-06, "loss": 2.8963, "step": 254470 }, { "epoch": 0.0086528, "grad_norm": 1.000443935394287, "learning_rate": 5.4315794435852175e-06, "loss": 3.0246, "step": 254480 }, { "epoch": 0.0086784, "grad_norm": 0.8376747965812683, "learning_rate": 5.430863124680386e-06, "loss": 2.897, "step": 254490 }, { "epoch": 0.008704, "grad_norm": 0.7752490043640137, "learning_rate": 5.430146835405191e-06, "loss": 2.7426, "step": 254500 }, { "epoch": 0.0087296, "grad_norm": 0.7682108283042908, "learning_rate": 5.429430575764281e-06, "loss": 2.754, "step": 254510 }, { "epoch": 0.0087552, "grad_norm": 0.91030353307724, "learning_rate": 5.428714345762297e-06, "loss": 2.8186, "step": 254520 }, { "epoch": 0.0087808, "grad_norm": 0.8506553173065186, "learning_rate": 5.427998145403886e-06, "loss": 2.8384, "step": 254530 }, { "epoch": 0.0088064, "grad_norm": 0.8278641104698181, "learning_rate": 5.427281974693697e-06, "loss": 2.8788, "step": 254540 }, { "epoch": 0.008832, "grad_norm": 0.8660897612571716, "learning_rate": 5.426565833636365e-06, "loss": 2.896, "step": 254550 }, { "epoch": 0.0088576, "grad_norm": 0.7660538554191589, "learning_rate": 5.425849722236537e-06, "loss": 2.8054, "step": 254560 }, { "epoch": 0.0088832, "grad_norm": 0.7329092621803284, "learning_rate": 5.425133640498858e-06, "loss": 2.6673, "step": 254570 }, { "epoch": 0.0089088, "grad_norm": 0.8160979151725769, "learning_rate": 5.424417588427973e-06, "loss": 2.637, "step": 254580 }, { "epoch": 0.0089344, "grad_norm": 0.8082692623138428, "learning_rate": 5.423701566028523e-06, "loss": 2.8798, "step": 254590 }, { "epoch": 0.00896, "grad_norm": 0.7979830503463745, "learning_rate": 5.422985573305152e-06, "loss": 3.0374, "step": 254600 }, { "epoch": 0.0089856, "grad_norm": 0.8514770269393921, "learning_rate": 5.4222696102625025e-06, "loss": 2.5059, "step": 254610 }, { "epoch": 0.0090112, "grad_norm": 1.3085639476776123, "learning_rate": 5.421553676905222e-06, "loss": 3.192, "step": 254620 }, { "epoch": 0.0090368, "grad_norm": 0.8345706462860107, "learning_rate": 5.420837773237947e-06, "loss": 2.8353, "step": 254630 }, { "epoch": 0.0090624, "grad_norm": 0.8344944715499878, "learning_rate": 5.420121899265316e-06, "loss": 2.7732, "step": 254640 }, { "epoch": 0.009088, "grad_norm": 0.8960566520690918, "learning_rate": 5.419406054991983e-06, "loss": 3.0125, "step": 254650 }, { "epoch": 0.0091136, "grad_norm": 0.7398132085800171, "learning_rate": 5.4186902404225835e-06, "loss": 2.9122, "step": 254660 }, { "epoch": 0.0091392, "grad_norm": 0.7664998769760132, "learning_rate": 5.417974455561764e-06, "loss": 2.9636, "step": 254670 }, { "epoch": 0.0091648, "grad_norm": 0.8071070909500122, "learning_rate": 5.417258700414159e-06, "loss": 2.9389, "step": 254680 }, { "epoch": 0.0091904, "grad_norm": 0.8457467555999756, "learning_rate": 5.416542974984412e-06, "loss": 2.9637, "step": 254690 }, { "epoch": 0.009216, "grad_norm": 0.7598708868026733, "learning_rate": 5.415827279277167e-06, "loss": 2.8682, "step": 254700 }, { "epoch": 0.0092416, "grad_norm": 0.8877251744270325, "learning_rate": 5.4151116132970635e-06, "loss": 2.7697, "step": 254710 }, { "epoch": 0.0092672, "grad_norm": 0.8721135854721069, "learning_rate": 5.414395977048743e-06, "loss": 2.8662, "step": 254720 }, { "epoch": 0.0092928, "grad_norm": 0.7946755290031433, "learning_rate": 5.4136803705368445e-06, "loss": 2.802, "step": 254730 }, { "epoch": 0.0093184, "grad_norm": 0.7494450211524963, "learning_rate": 5.412964793766015e-06, "loss": 2.7453, "step": 254740 }, { "epoch": 0.009344, "grad_norm": 0.8088334798812866, "learning_rate": 5.412249246740886e-06, "loss": 2.8474, "step": 254750 }, { "epoch": 0.0093696, "grad_norm": 0.7858192920684814, "learning_rate": 5.411533729466101e-06, "loss": 2.8561, "step": 254760 }, { "epoch": 0.0093952, "grad_norm": 0.731217086315155, "learning_rate": 5.4108182419463005e-06, "loss": 2.8135, "step": 254770 }, { "epoch": 0.0094208, "grad_norm": 0.7820490002632141, "learning_rate": 5.410102784186123e-06, "loss": 2.8978, "step": 254780 }, { "epoch": 0.0094464, "grad_norm": 0.787456750869751, "learning_rate": 5.40938735619021e-06, "loss": 2.7764, "step": 254790 }, { "epoch": 0.009472, "grad_norm": 0.8675233125686646, "learning_rate": 5.4086719579632e-06, "loss": 2.8761, "step": 254800 }, { "epoch": 0.0094976, "grad_norm": 0.8460472822189331, "learning_rate": 5.407956589509736e-06, "loss": 2.8343, "step": 254810 }, { "epoch": 0.0095232, "grad_norm": 0.8182120323181152, "learning_rate": 5.407241250834445e-06, "loss": 3.1616, "step": 254820 }, { "epoch": 0.0095488, "grad_norm": 0.9156132340431213, "learning_rate": 5.406525941941978e-06, "loss": 2.9172, "step": 254830 }, { "epoch": 0.0095744, "grad_norm": 0.815350353717804, "learning_rate": 5.40581066283697e-06, "loss": 2.8716, "step": 254840 }, { "epoch": 0.0096, "grad_norm": 0.7802117466926575, "learning_rate": 5.405095413524057e-06, "loss": 2.9206, "step": 254850 }, { "epoch": 0.0096256, "grad_norm": 0.8679704666137695, "learning_rate": 5.404380194007881e-06, "loss": 2.9165, "step": 254860 }, { "epoch": 0.0096512, "grad_norm": 2.132781744003296, "learning_rate": 5.403665004293082e-06, "loss": 2.9866, "step": 254870 }, { "epoch": 0.0096768, "grad_norm": 0.8082001805305481, "learning_rate": 5.402949844384289e-06, "loss": 2.9603, "step": 254880 }, { "epoch": 0.0097024, "grad_norm": 1.20448637008667, "learning_rate": 5.402234714286144e-06, "loss": 2.778, "step": 254890 }, { "epoch": 0.009728, "grad_norm": 0.9650071859359741, "learning_rate": 5.401519614003286e-06, "loss": 2.7049, "step": 254900 }, { "epoch": 0.0097536, "grad_norm": 0.9270169138908386, "learning_rate": 5.400804543540351e-06, "loss": 2.7776, "step": 254910 }, { "epoch": 0.0097792, "grad_norm": 0.7633162140846252, "learning_rate": 5.400089502901976e-06, "loss": 2.6178, "step": 254920 }, { "epoch": 0.0098048, "grad_norm": 0.8710430860519409, "learning_rate": 5.399374492092798e-06, "loss": 2.9368, "step": 254930 }, { "epoch": 0.0098304, "grad_norm": 0.8750624060630798, "learning_rate": 5.398659511117459e-06, "loss": 2.9541, "step": 254940 }, { "epoch": 0.009856, "grad_norm": 0.7344175577163696, "learning_rate": 5.397944559980584e-06, "loss": 2.8837, "step": 254950 }, { "epoch": 0.0098816, "grad_norm": 1.0061460733413696, "learning_rate": 5.3972296386868166e-06, "loss": 2.7904, "step": 254960 }, { "epoch": 0.0099072, "grad_norm": 0.8056373596191406, "learning_rate": 5.396514747240792e-06, "loss": 2.8473, "step": 254970 }, { "epoch": 0.0099328, "grad_norm": 0.7414243817329407, "learning_rate": 5.395799885647146e-06, "loss": 2.9967, "step": 254980 }, { "epoch": 0.0099584, "grad_norm": 1.0194029808044434, "learning_rate": 5.39508505391051e-06, "loss": 3.2203, "step": 254990 }, { "epoch": 0.009984, "grad_norm": 0.755955159664154, "learning_rate": 5.39437025203553e-06, "loss": 2.8092, "step": 255000 }, { "epoch": 0.0100096, "grad_norm": 0.7744314074516296, "learning_rate": 5.3936554800268315e-06, "loss": 2.9464, "step": 255010 }, { "epoch": 0.0100352, "grad_norm": 0.7898524403572083, "learning_rate": 5.392940737889054e-06, "loss": 2.9575, "step": 255020 }, { "epoch": 0.0100608, "grad_norm": 0.72966468334198, "learning_rate": 5.392226025626831e-06, "loss": 2.8895, "step": 255030 }, { "epoch": 0.0100864, "grad_norm": 0.8906839489936829, "learning_rate": 5.391511343244798e-06, "loss": 2.7368, "step": 255040 }, { "epoch": 0.010112, "grad_norm": 0.8892248868942261, "learning_rate": 5.3907966907475885e-06, "loss": 3.0297, "step": 255050 }, { "epoch": 0.0101376, "grad_norm": 0.7482385039329529, "learning_rate": 5.390082068139837e-06, "loss": 2.7256, "step": 255060 }, { "epoch": 0.0101632, "grad_norm": 0.8121020197868347, "learning_rate": 5.389367475426184e-06, "loss": 2.9842, "step": 255070 }, { "epoch": 0.0101888, "grad_norm": 0.7710288166999817, "learning_rate": 5.388652912611252e-06, "loss": 2.9854, "step": 255080 }, { "epoch": 0.0102144, "grad_norm": 0.7454388737678528, "learning_rate": 5.38793837969968e-06, "loss": 2.7777, "step": 255090 }, { "epoch": 0.01024, "grad_norm": 0.933393657207489, "learning_rate": 5.387223876696102e-06, "loss": 2.9568, "step": 255100 }, { "epoch": 0.0102656, "grad_norm": 0.7984071373939514, "learning_rate": 5.386509403605152e-06, "loss": 2.8964, "step": 255110 }, { "epoch": 0.0102912, "grad_norm": 0.8791545033454895, "learning_rate": 5.385794960431463e-06, "loss": 2.95, "step": 255120 }, { "epoch": 0.0103168, "grad_norm": 0.8025632500648499, "learning_rate": 5.385080547179666e-06, "loss": 3.0655, "step": 255130 }, { "epoch": 0.0103424, "grad_norm": 0.8614419102668762, "learning_rate": 5.384366163854399e-06, "loss": 3.0918, "step": 255140 }, { "epoch": 0.010368, "grad_norm": 0.8091189861297607, "learning_rate": 5.383651810460289e-06, "loss": 2.9659, "step": 255150 }, { "epoch": 0.0103936, "grad_norm": 1.8385896682739258, "learning_rate": 5.3829374870019675e-06, "loss": 3.218, "step": 255160 }, { "epoch": 0.0104192, "grad_norm": 0.8350193500518799, "learning_rate": 5.382223193484066e-06, "loss": 3.2455, "step": 255170 }, { "epoch": 0.0104448, "grad_norm": 0.7762564420700073, "learning_rate": 5.381508929911226e-06, "loss": 2.8609, "step": 255180 }, { "epoch": 0.0104704, "grad_norm": 0.9379992485046387, "learning_rate": 5.38079469628807e-06, "loss": 2.8404, "step": 255190 }, { "epoch": 0.010496, "grad_norm": 0.7654432058334351, "learning_rate": 5.380080492619238e-06, "loss": 2.9257, "step": 255200 }, { "epoch": 0.0105216, "grad_norm": 0.8181043267250061, "learning_rate": 5.379366318909352e-06, "loss": 2.8673, "step": 255210 }, { "epoch": 0.0105472, "grad_norm": 0.7695027589797974, "learning_rate": 5.3786521751630484e-06, "loss": 2.7263, "step": 255220 }, { "epoch": 0.0105728, "grad_norm": 0.9951304793357849, "learning_rate": 5.377938061384957e-06, "loss": 2.8074, "step": 255230 }, { "epoch": 0.0105984, "grad_norm": 0.7965419888496399, "learning_rate": 5.3772239775797085e-06, "loss": 2.7785, "step": 255240 }, { "epoch": 0.010624, "grad_norm": 0.816730797290802, "learning_rate": 5.376509923751934e-06, "loss": 2.806, "step": 255250 }, { "epoch": 0.0106496, "grad_norm": 0.784660279750824, "learning_rate": 5.375795899906263e-06, "loss": 2.8297, "step": 255260 }, { "epoch": 0.0106752, "grad_norm": 0.7697584629058838, "learning_rate": 5.375081906047331e-06, "loss": 2.9306, "step": 255270 }, { "epoch": 0.0107008, "grad_norm": 0.7619170546531677, "learning_rate": 5.37436794217976e-06, "loss": 2.8847, "step": 255280 }, { "epoch": 0.0107264, "grad_norm": 0.852239727973938, "learning_rate": 5.373654008308184e-06, "loss": 2.622, "step": 255290 }, { "epoch": 0.010752, "grad_norm": 0.8021697998046875, "learning_rate": 5.372940104437232e-06, "loss": 3.2518, "step": 255300 }, { "epoch": 0.0107776, "grad_norm": 0.738312304019928, "learning_rate": 5.3722262305715334e-06, "loss": 2.9922, "step": 255310 }, { "epoch": 0.0108032, "grad_norm": 0.7796955704689026, "learning_rate": 5.371512386715718e-06, "loss": 2.8431, "step": 255320 }, { "epoch": 0.0108288, "grad_norm": 0.7666494846343994, "learning_rate": 5.370798572874414e-06, "loss": 2.631, "step": 255330 }, { "epoch": 0.0108544, "grad_norm": 0.8818960189819336, "learning_rate": 5.370084789052252e-06, "loss": 3.0039, "step": 255340 }, { "epoch": 0.01088, "grad_norm": 0.7585851550102234, "learning_rate": 5.369371035253859e-06, "loss": 2.7942, "step": 255350 }, { "epoch": 0.0109056, "grad_norm": 0.7015970945358276, "learning_rate": 5.368657311483865e-06, "loss": 2.8919, "step": 255360 }, { "epoch": 0.0109312, "grad_norm": 0.804915189743042, "learning_rate": 5.367943617746896e-06, "loss": 2.8636, "step": 255370 }, { "epoch": 0.0109568, "grad_norm": 0.7672659158706665, "learning_rate": 5.367229954047581e-06, "loss": 3.0056, "step": 255380 }, { "epoch": 0.0109824, "grad_norm": 0.7566434741020203, "learning_rate": 5.366516320390551e-06, "loss": 2.8509, "step": 255390 }, { "epoch": 0.011008, "grad_norm": 0.85648113489151, "learning_rate": 5.365802716780434e-06, "loss": 4.4332, "step": 255400 }, { "epoch": 0.0110336, "grad_norm": 0.7045837640762329, "learning_rate": 5.365089143221851e-06, "loss": 2.8625, "step": 255410 }, { "epoch": 0.0110592, "grad_norm": 0.8531771302223206, "learning_rate": 5.364375599719433e-06, "loss": 2.8525, "step": 255420 }, { "epoch": 0.0110848, "grad_norm": 0.7946866750717163, "learning_rate": 5.363662086277808e-06, "loss": 2.7881, "step": 255430 }, { "epoch": 0.0111104, "grad_norm": 0.8448289632797241, "learning_rate": 5.362948602901601e-06, "loss": 2.812, "step": 255440 }, { "epoch": 0.011136, "grad_norm": 0.811938464641571, "learning_rate": 5.362235149595439e-06, "loss": 2.8646, "step": 255450 }, { "epoch": 0.0111616, "grad_norm": 0.8469358682632446, "learning_rate": 5.361521726363952e-06, "loss": 2.7883, "step": 255460 }, { "epoch": 0.0111872, "grad_norm": 0.8217935562133789, "learning_rate": 5.360808333211767e-06, "loss": 3.1084, "step": 255470 }, { "epoch": 0.0112128, "grad_norm": 0.9088783860206604, "learning_rate": 5.360094970143501e-06, "loss": 2.8886, "step": 255480 }, { "epoch": 0.0112384, "grad_norm": 0.9801842570304871, "learning_rate": 5.3593816371637875e-06, "loss": 2.8613, "step": 255490 }, { "epoch": 0.011264, "grad_norm": 0.8392219543457031, "learning_rate": 5.358668334277251e-06, "loss": 2.7907, "step": 255500 }, { "epoch": 0.0112896, "grad_norm": 0.7978240847587585, "learning_rate": 5.357955061488513e-06, "loss": 2.8188, "step": 255510 }, { "epoch": 0.0113152, "grad_norm": 0.8889535069465637, "learning_rate": 5.357241818802207e-06, "loss": 2.9787, "step": 255520 }, { "epoch": 0.0113408, "grad_norm": 0.8925661444664001, "learning_rate": 5.356528606222957e-06, "loss": 2.7562, "step": 255530 }, { "epoch": 0.0113664, "grad_norm": 1.3534351587295532, "learning_rate": 5.355815423755382e-06, "loss": 2.8417, "step": 255540 }, { "epoch": 0.011392, "grad_norm": 0.8428630828857422, "learning_rate": 5.355102271404109e-06, "loss": 2.7094, "step": 255550 }, { "epoch": 0.0114176, "grad_norm": 0.8754501342773438, "learning_rate": 5.354389149173762e-06, "loss": 2.9951, "step": 255560 }, { "epoch": 0.0114432, "grad_norm": 0.9693076610565186, "learning_rate": 5.353676057068968e-06, "loss": 2.9654, "step": 255570 }, { "epoch": 0.0114688, "grad_norm": 0.970628023147583, "learning_rate": 5.352962995094351e-06, "loss": 2.6881, "step": 255580 }, { "epoch": 0.0114944, "grad_norm": 0.8380241990089417, "learning_rate": 5.352249963254532e-06, "loss": 2.8279, "step": 255590 }, { "epoch": 0.01152, "grad_norm": 0.8125719428062439, "learning_rate": 5.3515369615541426e-06, "loss": 2.8936, "step": 255600 }, { "epoch": 0.0115456, "grad_norm": 1.0274641513824463, "learning_rate": 5.350823989997794e-06, "loss": 2.7783, "step": 255610 }, { "epoch": 0.0115712, "grad_norm": 0.8053932189941406, "learning_rate": 5.350111048590119e-06, "loss": 2.8854, "step": 255620 }, { "epoch": 0.0115968, "grad_norm": 0.8779410719871521, "learning_rate": 5.349398137335736e-06, "loss": 2.9061, "step": 255630 }, { "epoch": 0.0116224, "grad_norm": 0.7651969790458679, "learning_rate": 5.348685256239271e-06, "loss": 2.7297, "step": 255640 }, { "epoch": 0.011648, "grad_norm": 0.7484729886054993, "learning_rate": 5.347972405305346e-06, "loss": 2.8776, "step": 255650 }, { "epoch": 0.0116736, "grad_norm": 0.8609609007835388, "learning_rate": 5.347259584538583e-06, "loss": 2.7772, "step": 255660 }, { "epoch": 0.0116992, "grad_norm": 0.9566562175750732, "learning_rate": 5.34654679394361e-06, "loss": 3.0125, "step": 255670 }, { "epoch": 0.0117248, "grad_norm": 0.8733117580413818, "learning_rate": 5.345834033525039e-06, "loss": 2.9817, "step": 255680 }, { "epoch": 0.0117504, "grad_norm": 0.7368727922439575, "learning_rate": 5.345121303287495e-06, "loss": 2.9764, "step": 255690 }, { "epoch": 0.011776, "grad_norm": 0.8888649940490723, "learning_rate": 5.344408603235606e-06, "loss": 2.8231, "step": 255700 }, { "epoch": 0.0118016, "grad_norm": 1.0205609798431396, "learning_rate": 5.343695933373989e-06, "loss": 2.9607, "step": 255710 }, { "epoch": 0.0118272, "grad_norm": 0.7825377583503723, "learning_rate": 5.342983293707266e-06, "loss": 2.9565, "step": 255720 }, { "epoch": 0.0118528, "grad_norm": 0.8638370633125305, "learning_rate": 5.342270684240064e-06, "loss": 2.8446, "step": 255730 }, { "epoch": 0.0118784, "grad_norm": 0.9265411496162415, "learning_rate": 5.3415581049769935e-06, "loss": 3.0666, "step": 255740 }, { "epoch": 0.011904, "grad_norm": 0.8765382766723633, "learning_rate": 5.340845555922682e-06, "loss": 2.8838, "step": 255750 }, { "epoch": 0.0119296, "grad_norm": 0.7616429924964905, "learning_rate": 5.340133037081747e-06, "loss": 2.5856, "step": 255760 }, { "epoch": 0.0119552, "grad_norm": 0.7712927460670471, "learning_rate": 5.339420548458813e-06, "loss": 2.8848, "step": 255770 }, { "epoch": 0.0119808, "grad_norm": 0.7542492747306824, "learning_rate": 5.338708090058497e-06, "loss": 2.8805, "step": 255780 }, { "epoch": 0.0120064, "grad_norm": 0.7512916326522827, "learning_rate": 5.33799566188542e-06, "loss": 2.6785, "step": 255790 }, { "epoch": 0.012032, "grad_norm": 0.7421396970748901, "learning_rate": 5.337283263944206e-06, "loss": 2.8579, "step": 255800 }, { "epoch": 0.0120576, "grad_norm": 0.8656833171844482, "learning_rate": 5.3365708962394665e-06, "loss": 2.8731, "step": 255810 }, { "epoch": 0.0120832, "grad_norm": 0.7905917167663574, "learning_rate": 5.335858558775826e-06, "loss": 2.8363, "step": 255820 }, { "epoch": 0.0121088, "grad_norm": 0.7760031819343567, "learning_rate": 5.3351462515579035e-06, "loss": 2.773, "step": 255830 }, { "epoch": 0.0121344, "grad_norm": 0.9707251191139221, "learning_rate": 5.3344339745903175e-06, "loss": 2.8864, "step": 255840 }, { "epoch": 0.01216, "grad_norm": 0.7917926907539368, "learning_rate": 5.333721727877688e-06, "loss": 2.9526, "step": 255850 }, { "epoch": 0.0121856, "grad_norm": 3.891606092453003, "learning_rate": 5.333009511424632e-06, "loss": 2.5856, "step": 255860 }, { "epoch": 0.0122112, "grad_norm": 1.0081418752670288, "learning_rate": 5.332297325235768e-06, "loss": 3.0068, "step": 255870 }, { "epoch": 0.0122368, "grad_norm": 0.9342308044433594, "learning_rate": 5.331585169315716e-06, "loss": 2.8761, "step": 255880 }, { "epoch": 0.0122624, "grad_norm": 0.7823627591133118, "learning_rate": 5.330873043669094e-06, "loss": 2.9953, "step": 255890 }, { "epoch": 0.012288, "grad_norm": 0.7375440001487732, "learning_rate": 5.33016094830052e-06, "loss": 2.8975, "step": 255900 }, { "epoch": 0.0123136, "grad_norm": 0.7417622804641724, "learning_rate": 5.32944888321461e-06, "loss": 2.9638, "step": 255910 }, { "epoch": 0.0123392, "grad_norm": 1.9589147567749023, "learning_rate": 5.328736848415983e-06, "loss": 3.0731, "step": 255920 }, { "epoch": 0.0123648, "grad_norm": 0.7772890329360962, "learning_rate": 5.328024843909261e-06, "loss": 2.9471, "step": 255930 }, { "epoch": 0.0123904, "grad_norm": 1.0172293186187744, "learning_rate": 5.327312869699051e-06, "loss": 2.8669, "step": 255940 }, { "epoch": 0.012416, "grad_norm": 0.7415482401847839, "learning_rate": 5.326600925789975e-06, "loss": 2.7495, "step": 255950 }, { "epoch": 0.0124416, "grad_norm": 0.8113489747047424, "learning_rate": 5.325889012186651e-06, "loss": 2.7732, "step": 255960 }, { "epoch": 0.0124672, "grad_norm": 1.01914381980896, "learning_rate": 5.325177128893694e-06, "loss": 3.0545, "step": 255970 }, { "epoch": 0.0124928, "grad_norm": 0.8331207633018494, "learning_rate": 5.32446527591572e-06, "loss": 2.7245, "step": 255980 }, { "epoch": 0.0125184, "grad_norm": 0.8041173815727234, "learning_rate": 5.3237534532573474e-06, "loss": 2.8682, "step": 255990 }, { "epoch": 0.012544, "grad_norm": 1.199450135231018, "learning_rate": 5.323041660923195e-06, "loss": 2.7763, "step": 256000 }, { "epoch": 0.0125696, "grad_norm": 0.923504650592804, "learning_rate": 5.322329898917869e-06, "loss": 2.7482, "step": 256010 }, { "epoch": 0.0125952, "grad_norm": 0.8038642406463623, "learning_rate": 5.321618167245992e-06, "loss": 2.9471, "step": 256020 }, { "epoch": 0.0126208, "grad_norm": 0.9549691081047058, "learning_rate": 5.320906465912177e-06, "loss": 3.0536, "step": 256030 }, { "epoch": 0.0126464, "grad_norm": 0.8004876971244812, "learning_rate": 5.320194794921036e-06, "loss": 2.7884, "step": 256040 }, { "epoch": 0.012672, "grad_norm": 1.2010762691497803, "learning_rate": 5.319483154277196e-06, "loss": 3.1037, "step": 256050 }, { "epoch": 0.0126976, "grad_norm": 1.2885372638702393, "learning_rate": 5.318771543985261e-06, "loss": 2.8361, "step": 256060 }, { "epoch": 0.0127232, "grad_norm": 0.8874810338020325, "learning_rate": 5.318059964049846e-06, "loss": 2.9247, "step": 256070 }, { "epoch": 0.0127488, "grad_norm": 0.7637718915939331, "learning_rate": 5.3173484144755704e-06, "loss": 2.8771, "step": 256080 }, { "epoch": 0.0127744, "grad_norm": 0.8712933659553528, "learning_rate": 5.316636895267045e-06, "loss": 2.9599, "step": 256090 }, { "epoch": 0.0128, "grad_norm": 0.9455146193504333, "learning_rate": 5.3159254064288835e-06, "loss": 2.9911, "step": 256100 }, { "epoch": 0.0128256, "grad_norm": 0.811829686164856, "learning_rate": 5.3152139479657025e-06, "loss": 2.9276, "step": 256110 }, { "epoch": 0.0128512, "grad_norm": 1.0042107105255127, "learning_rate": 5.314502519882116e-06, "loss": 2.9677, "step": 256120 }, { "epoch": 0.0128768, "grad_norm": 0.8723130226135254, "learning_rate": 5.313791122182734e-06, "loss": 2.9488, "step": 256130 }, { "epoch": 0.0129024, "grad_norm": 0.7861865162849426, "learning_rate": 5.313079754872169e-06, "loss": 2.984, "step": 256140 }, { "epoch": 0.012928, "grad_norm": 0.7117831707000732, "learning_rate": 5.3123684179550375e-06, "loss": 2.8536, "step": 256150 }, { "epoch": 0.0129536, "grad_norm": 0.764431893825531, "learning_rate": 5.31165711143595e-06, "loss": 2.8786, "step": 256160 }, { "epoch": 0.0129792, "grad_norm": 0.8518897294998169, "learning_rate": 5.310945835319521e-06, "loss": 2.8959, "step": 256170 }, { "epoch": 0.0130048, "grad_norm": 1.0214279890060425, "learning_rate": 5.310234589610361e-06, "loss": 3.0663, "step": 256180 }, { "epoch": 0.0130304, "grad_norm": 0.8268762230873108, "learning_rate": 5.309523374313084e-06, "loss": 2.8052, "step": 256190 }, { "epoch": 0.013056, "grad_norm": 0.7730439305305481, "learning_rate": 5.308812189432305e-06, "loss": 2.7645, "step": 256200 }, { "epoch": 0.0130816, "grad_norm": 0.798141360282898, "learning_rate": 5.3081010349726236e-06, "loss": 2.7085, "step": 256210 }, { "epoch": 0.0131072, "grad_norm": 0.8246241807937622, "learning_rate": 5.307389910938665e-06, "loss": 2.7577, "step": 256220 }, { "epoch": 0.0131328, "grad_norm": 0.855529248714447, "learning_rate": 5.3066788173350355e-06, "loss": 2.8619, "step": 256230 }, { "epoch": 0.0131584, "grad_norm": 0.8208242654800415, "learning_rate": 5.305967754166348e-06, "loss": 2.7737, "step": 256240 }, { "epoch": 0.013184, "grad_norm": 0.781068742275238, "learning_rate": 5.305256721437213e-06, "loss": 2.7217, "step": 256250 }, { "epoch": 0.0132096, "grad_norm": 0.8151941299438477, "learning_rate": 5.304545719152239e-06, "loss": 2.8363, "step": 256260 }, { "epoch": 0.0132352, "grad_norm": 0.8517744541168213, "learning_rate": 5.303834747316036e-06, "loss": 2.8296, "step": 256270 }, { "epoch": 0.0132608, "grad_norm": 0.8399896025657654, "learning_rate": 5.303123805933219e-06, "loss": 3.0226, "step": 256280 }, { "epoch": 0.0132864, "grad_norm": 0.7706466317176819, "learning_rate": 5.302412895008394e-06, "loss": 2.8688, "step": 256290 }, { "epoch": 0.013312, "grad_norm": 0.78352952003479, "learning_rate": 5.301702014546173e-06, "loss": 3.655, "step": 256300 }, { "epoch": 0.0133376, "grad_norm": 0.809078574180603, "learning_rate": 5.300991164551167e-06, "loss": 2.8378, "step": 256310 }, { "epoch": 0.0133632, "grad_norm": 0.8450784683227539, "learning_rate": 5.3002803450279886e-06, "loss": 2.8719, "step": 256320 }, { "epoch": 0.0133888, "grad_norm": 0.8853555917739868, "learning_rate": 5.299569555981237e-06, "loss": 2.965, "step": 256330 }, { "epoch": 0.0134144, "grad_norm": 0.7356374859809875, "learning_rate": 5.2988587974155295e-06, "loss": 2.8361, "step": 256340 }, { "epoch": 0.01344, "grad_norm": 0.8518565893173218, "learning_rate": 5.2981480693354716e-06, "loss": 2.7982, "step": 256350 }, { "epoch": 0.0134656, "grad_norm": 0.7327706813812256, "learning_rate": 5.297437371745675e-06, "loss": 2.949, "step": 256360 }, { "epoch": 0.0134912, "grad_norm": 0.8708344101905823, "learning_rate": 5.296726704650747e-06, "loss": 2.641, "step": 256370 }, { "epoch": 0.0135168, "grad_norm": 0.7623043656349182, "learning_rate": 5.296016068055296e-06, "loss": 2.95, "step": 256380 }, { "epoch": 0.0135424, "grad_norm": 0.7640691995620728, "learning_rate": 5.295305461963932e-06, "loss": 2.9077, "step": 256390 }, { "epoch": 0.013568, "grad_norm": 2.291898250579834, "learning_rate": 5.294594886381261e-06, "loss": 3.026, "step": 256400 }, { "epoch": 0.0135936, "grad_norm": 0.8480849862098694, "learning_rate": 5.293884341311891e-06, "loss": 2.7348, "step": 256410 }, { "epoch": 0.0136192, "grad_norm": 0.8009179830551147, "learning_rate": 5.2931738267604316e-06, "loss": 3.0137, "step": 256420 }, { "epoch": 0.0136448, "grad_norm": 0.7713523507118225, "learning_rate": 5.2924633427314874e-06, "loss": 2.7467, "step": 256430 }, { "epoch": 0.0136704, "grad_norm": 0.773215651512146, "learning_rate": 5.291752889229669e-06, "loss": 2.9549, "step": 256440 }, { "epoch": 0.013696, "grad_norm": 0.8500131368637085, "learning_rate": 5.291042466259584e-06, "loss": 2.827, "step": 256450 }, { "epoch": 0.0137216, "grad_norm": 0.7934130430221558, "learning_rate": 5.290332073825835e-06, "loss": 2.846, "step": 256460 }, { "epoch": 0.0137472, "grad_norm": 0.8525474071502686, "learning_rate": 5.289621711933029e-06, "loss": 3.1816, "step": 256470 }, { "epoch": 0.0137728, "grad_norm": 0.8313926458358765, "learning_rate": 5.288911380585775e-06, "loss": 3.0745, "step": 256480 }, { "epoch": 0.0137984, "grad_norm": 0.8329505920410156, "learning_rate": 5.28820107978868e-06, "loss": 2.9193, "step": 256490 }, { "epoch": 0.013824, "grad_norm": 0.9308491349220276, "learning_rate": 5.287490809546346e-06, "loss": 2.951, "step": 256500 }, { "epoch": 0.0138496, "grad_norm": 0.900080680847168, "learning_rate": 5.286780569863383e-06, "loss": 2.7779, "step": 256510 }, { "epoch": 0.0138752, "grad_norm": 0.8525245785713196, "learning_rate": 5.2860703607444e-06, "loss": 2.9022, "step": 256520 }, { "epoch": 0.0139008, "grad_norm": 0.7132829427719116, "learning_rate": 5.2853601821939934e-06, "loss": 2.5843, "step": 256530 }, { "epoch": 0.0139264, "grad_norm": 0.7044272422790527, "learning_rate": 5.284650034216774e-06, "loss": 2.8146, "step": 256540 }, { "epoch": 0.013952, "grad_norm": 0.940919041633606, "learning_rate": 5.283939916817344e-06, "loss": 3.0212, "step": 256550 }, { "epoch": 0.0139776, "grad_norm": 0.8333229422569275, "learning_rate": 5.283229830000307e-06, "loss": 2.8472, "step": 256560 }, { "epoch": 0.0140032, "grad_norm": 0.8036003708839417, "learning_rate": 5.282519773770276e-06, "loss": 2.8097, "step": 256570 }, { "epoch": 0.0140288, "grad_norm": 0.9874554872512817, "learning_rate": 5.281809748131854e-06, "loss": 2.7862, "step": 256580 }, { "epoch": 0.0140544, "grad_norm": 0.8280710577964783, "learning_rate": 5.281099753089638e-06, "loss": 2.9951, "step": 256590 }, { "epoch": 0.01408, "grad_norm": 0.8152764439582825, "learning_rate": 5.280389788648234e-06, "loss": 2.9149, "step": 256600 }, { "epoch": 0.0141056, "grad_norm": 0.788305401802063, "learning_rate": 5.27967985481225e-06, "loss": 2.9301, "step": 256610 }, { "epoch": 0.0141312, "grad_norm": 0.9205355048179626, "learning_rate": 5.278969951586288e-06, "loss": 2.8026, "step": 256620 }, { "epoch": 0.0141568, "grad_norm": 0.7281591892242432, "learning_rate": 5.27826007897495e-06, "loss": 3.035, "step": 256630 }, { "epoch": 0.0141824, "grad_norm": 0.8485949039459229, "learning_rate": 5.27755023698284e-06, "loss": 3.0422, "step": 256640 }, { "epoch": 0.014208, "grad_norm": 0.9169301390647888, "learning_rate": 5.276840425614567e-06, "loss": 2.8052, "step": 256650 }, { "epoch": 0.0142336, "grad_norm": 0.840732991695404, "learning_rate": 5.276130644874724e-06, "loss": 2.6851, "step": 256660 }, { "epoch": 0.0142592, "grad_norm": 0.7586621642112732, "learning_rate": 5.275420894767918e-06, "loss": 2.7734, "step": 256670 }, { "epoch": 0.0142848, "grad_norm": 5.159088134765625, "learning_rate": 5.274711175298752e-06, "loss": 2.9827, "step": 256680 }, { "epoch": 0.0143104, "grad_norm": 1.0134878158569336, "learning_rate": 5.274001486471829e-06, "loss": 2.8125, "step": 256690 }, { "epoch": 0.014336, "grad_norm": 0.9385380148887634, "learning_rate": 5.273291828291749e-06, "loss": 3.0046, "step": 256700 }, { "epoch": 0.0143616, "grad_norm": 1.199568748474121, "learning_rate": 5.2725822007631165e-06, "loss": 2.921, "step": 256710 }, { "epoch": 0.0143872, "grad_norm": 1.7568105459213257, "learning_rate": 5.271872603890536e-06, "loss": 2.8939, "step": 256720 }, { "epoch": 0.0144128, "grad_norm": 0.8090635538101196, "learning_rate": 5.271163037678596e-06, "loss": 2.806, "step": 256730 }, { "epoch": 0.0144384, "grad_norm": 0.7688023447990417, "learning_rate": 5.270453502131912e-06, "loss": 2.8847, "step": 256740 }, { "epoch": 0.014464, "grad_norm": 0.8677545785903931, "learning_rate": 5.269743997255079e-06, "loss": 2.8511, "step": 256750 }, { "epoch": 0.0144896, "grad_norm": 0.876140296459198, "learning_rate": 5.2690345230527005e-06, "loss": 2.8226, "step": 256760 }, { "epoch": 0.0145152, "grad_norm": 0.794144332408905, "learning_rate": 5.268325079529375e-06, "loss": 2.9549, "step": 256770 }, { "epoch": 0.0145408, "grad_norm": 0.9785053133964539, "learning_rate": 5.267615666689707e-06, "loss": 2.8678, "step": 256780 }, { "epoch": 0.0145664, "grad_norm": 0.7833499908447266, "learning_rate": 5.266906284538291e-06, "loss": 2.8048, "step": 256790 }, { "epoch": 0.014592, "grad_norm": 1.0150158405303955, "learning_rate": 5.266196933079728e-06, "loss": 3.0022, "step": 256800 }, { "epoch": 0.0146176, "grad_norm": 0.8226448893547058, "learning_rate": 5.265487612318622e-06, "loss": 2.7998, "step": 256810 }, { "epoch": 0.0146432, "grad_norm": 0.7122786641120911, "learning_rate": 5.26477832225957e-06, "loss": 2.7923, "step": 256820 }, { "epoch": 0.0146688, "grad_norm": 0.8186473250389099, "learning_rate": 5.264069062907171e-06, "loss": 2.9119, "step": 256830 }, { "epoch": 0.0146944, "grad_norm": 0.8083987236022949, "learning_rate": 5.263359834266026e-06, "loss": 2.8822, "step": 256840 }, { "epoch": 0.01472, "grad_norm": 0.9180100560188293, "learning_rate": 5.262650636340738e-06, "loss": 2.6496, "step": 256850 }, { "epoch": 0.0147456, "grad_norm": 0.9318389892578125, "learning_rate": 5.261941469135898e-06, "loss": 2.9747, "step": 256860 }, { "epoch": 0.0147712, "grad_norm": 0.739833414554596, "learning_rate": 5.261232332656108e-06, "loss": 2.9725, "step": 256870 }, { "epoch": 0.0147968, "grad_norm": 0.8242554068565369, "learning_rate": 5.260523226905966e-06, "loss": 2.9257, "step": 256880 }, { "epoch": 0.0148224, "grad_norm": 1.028207540512085, "learning_rate": 5.259814151890073e-06, "loss": 2.7783, "step": 256890 }, { "epoch": 0.014848, "grad_norm": 0.8380058407783508, "learning_rate": 5.2591051076130225e-06, "loss": 2.9157, "step": 256900 }, { "epoch": 0.0148736, "grad_norm": 0.7869480848312378, "learning_rate": 5.258396094079419e-06, "loss": 2.8903, "step": 256910 }, { "epoch": 0.0148992, "grad_norm": 1.0561352968215942, "learning_rate": 5.257687111293854e-06, "loss": 2.8773, "step": 256920 }, { "epoch": 0.0149248, "grad_norm": 0.9462451338768005, "learning_rate": 5.256978159260928e-06, "loss": 3.0934, "step": 256930 }, { "epoch": 0.0149504, "grad_norm": 0.9341544508934021, "learning_rate": 5.256269237985238e-06, "loss": 2.7191, "step": 256940 }, { "epoch": 0.014976, "grad_norm": 0.7813514471054077, "learning_rate": 5.2555603474713835e-06, "loss": 2.8231, "step": 256950 }, { "epoch": 0.0150016, "grad_norm": 0.8499619960784912, "learning_rate": 5.2548514877239575e-06, "loss": 2.7198, "step": 256960 }, { "epoch": 0.0150272, "grad_norm": 0.7796741127967834, "learning_rate": 5.254142658747559e-06, "loss": 2.7356, "step": 256970 }, { "epoch": 0.0150528, "grad_norm": 0.8627537488937378, "learning_rate": 5.2534338605467885e-06, "loss": 2.9204, "step": 256980 }, { "epoch": 0.0150784, "grad_norm": 0.9876757264137268, "learning_rate": 5.252725093126232e-06, "loss": 2.7695, "step": 256990 }, { "epoch": 0.015104, "grad_norm": 0.8539504408836365, "learning_rate": 5.2520163564904925e-06, "loss": 2.7546, "step": 257000 }, { "epoch": 0.0151296, "grad_norm": 0.7706423997879028, "learning_rate": 5.2513076506441665e-06, "loss": 2.556, "step": 257010 }, { "epoch": 0.0151552, "grad_norm": 0.8955048322677612, "learning_rate": 5.250598975591848e-06, "loss": 2.8716, "step": 257020 }, { "epoch": 0.0151808, "grad_norm": 0.8880419135093689, "learning_rate": 5.24989033133813e-06, "loss": 3.0294, "step": 257030 }, { "epoch": 0.0152064, "grad_norm": 0.9216420650482178, "learning_rate": 5.249181717887614e-06, "loss": 2.7889, "step": 257040 }, { "epoch": 0.015232, "grad_norm": 0.8195436596870422, "learning_rate": 5.248473135244894e-06, "loss": 2.9427, "step": 257050 }, { "epoch": 0.0152576, "grad_norm": 0.8092547655105591, "learning_rate": 5.247764583414561e-06, "loss": 3.1337, "step": 257060 }, { "epoch": 0.0152832, "grad_norm": 0.9728691577911377, "learning_rate": 5.24705606240121e-06, "loss": 2.9607, "step": 257070 }, { "epoch": 0.0153088, "grad_norm": 0.8044329881668091, "learning_rate": 5.246347572209433e-06, "loss": 3.0636, "step": 257080 }, { "epoch": 0.0153344, "grad_norm": 1.1226990222930908, "learning_rate": 5.245639112843834e-06, "loss": 2.8035, "step": 257090 }, { "epoch": 0.01536, "grad_norm": 0.8400368094444275, "learning_rate": 5.2449306843090005e-06, "loss": 2.919, "step": 257100 }, { "epoch": 0.0153856, "grad_norm": 0.7823745012283325, "learning_rate": 5.244222286609533e-06, "loss": 2.9683, "step": 257110 }, { "epoch": 0.0154112, "grad_norm": 1.0224529504776, "learning_rate": 5.243513919750015e-06, "loss": 2.8895, "step": 257120 }, { "epoch": 0.0154368, "grad_norm": 0.7177509069442749, "learning_rate": 5.2428055837350455e-06, "loss": 2.9058, "step": 257130 }, { "epoch": 0.0154624, "grad_norm": 0.8792719841003418, "learning_rate": 5.242097278569218e-06, "loss": 2.7267, "step": 257140 }, { "epoch": 0.015488, "grad_norm": 0.7320239543914795, "learning_rate": 5.241389004257124e-06, "loss": 2.8348, "step": 257150 }, { "epoch": 0.0155136, "grad_norm": 0.917519748210907, "learning_rate": 5.240680760803358e-06, "loss": 2.9255, "step": 257160 }, { "epoch": 0.0155392, "grad_norm": 0.8774864673614502, "learning_rate": 5.239972548212513e-06, "loss": 2.9229, "step": 257170 }, { "epoch": 0.0155648, "grad_norm": 1.0451384782791138, "learning_rate": 5.239264366489185e-06, "loss": 2.976, "step": 257180 }, { "epoch": 0.0155904, "grad_norm": 0.7188464999198914, "learning_rate": 5.2385562156379575e-06, "loss": 2.6556, "step": 257190 }, { "epoch": 0.015616, "grad_norm": 0.9457322955131531, "learning_rate": 5.237848095663428e-06, "loss": 2.9385, "step": 257200 }, { "epoch": 0.0156416, "grad_norm": 0.8443065285682678, "learning_rate": 5.237140006570187e-06, "loss": 2.8401, "step": 257210 }, { "epoch": 0.0156672, "grad_norm": 0.7219653725624084, "learning_rate": 5.236431948362829e-06, "loss": 2.7457, "step": 257220 }, { "epoch": 0.0156928, "grad_norm": 0.8318017721176147, "learning_rate": 5.2357239210459435e-06, "loss": 2.9429, "step": 257230 }, { "epoch": 0.0157184, "grad_norm": 0.8213587403297424, "learning_rate": 5.2350159246241206e-06, "loss": 2.8524, "step": 257240 }, { "epoch": 0.015744, "grad_norm": 0.8495520353317261, "learning_rate": 5.234307959101954e-06, "loss": 3.2178, "step": 257250 }, { "epoch": 0.0157696, "grad_norm": 0.7284195423126221, "learning_rate": 5.233600024484035e-06, "loss": 2.8029, "step": 257260 }, { "epoch": 0.0157952, "grad_norm": 0.8011943697929382, "learning_rate": 5.232892120774951e-06, "loss": 2.8001, "step": 257270 }, { "epoch": 0.0158208, "grad_norm": 0.8358827829360962, "learning_rate": 5.232184247979296e-06, "loss": 2.9837, "step": 257280 }, { "epoch": 0.0158464, "grad_norm": 0.7337135672569275, "learning_rate": 5.231476406101658e-06, "loss": 2.8578, "step": 257290 }, { "epoch": 0.015872, "grad_norm": 0.789686381816864, "learning_rate": 5.2307685951466295e-06, "loss": 2.9316, "step": 257300 }, { "epoch": 0.0158976, "grad_norm": 0.7550737261772156, "learning_rate": 5.230060815118802e-06, "loss": 2.798, "step": 257310 }, { "epoch": 0.0159232, "grad_norm": 0.8372502326965332, "learning_rate": 5.229353066022759e-06, "loss": 2.7451, "step": 257320 }, { "epoch": 0.0159488, "grad_norm": 0.825238049030304, "learning_rate": 5.228645347863092e-06, "loss": 2.8327, "step": 257330 }, { "epoch": 0.0159744, "grad_norm": 0.7414020895957947, "learning_rate": 5.227937660644394e-06, "loss": 2.8792, "step": 257340 }, { "epoch": 0.016, "grad_norm": 0.8321099281311035, "learning_rate": 5.227230004371251e-06, "loss": 3.1579, "step": 257350 }, { "epoch": 0.0160256, "grad_norm": 0.7628365755081177, "learning_rate": 5.226522379048254e-06, "loss": 2.8116, "step": 257360 }, { "epoch": 0.0160512, "grad_norm": 0.7351198792457581, "learning_rate": 5.22581478467999e-06, "loss": 2.7323, "step": 257370 }, { "epoch": 0.0160768, "grad_norm": 0.8311527371406555, "learning_rate": 5.2251072212710505e-06, "loss": 3.0732, "step": 257380 }, { "epoch": 0.0161024, "grad_norm": 0.762843132019043, "learning_rate": 5.224399688826021e-06, "loss": 3.0587, "step": 257390 }, { "epoch": 0.016128, "grad_norm": 0.8130677342414856, "learning_rate": 5.223692187349487e-06, "loss": 2.8221, "step": 257400 }, { "epoch": 0.0161536, "grad_norm": 0.7279256582260132, "learning_rate": 5.222984716846043e-06, "loss": 2.7104, "step": 257410 }, { "epoch": 0.0161792, "grad_norm": 0.7546041011810303, "learning_rate": 5.222277277320271e-06, "loss": 2.7953, "step": 257420 }, { "epoch": 0.0162048, "grad_norm": 1.0166934728622437, "learning_rate": 5.221569868776763e-06, "loss": 2.8586, "step": 257430 }, { "epoch": 0.0162304, "grad_norm": 0.9766255021095276, "learning_rate": 5.220862491220102e-06, "loss": 2.7532, "step": 257440 }, { "epoch": 0.016256, "grad_norm": 0.8946823477745056, "learning_rate": 5.22015514465488e-06, "loss": 2.915, "step": 257450 }, { "epoch": 0.0162816, "grad_norm": 0.8309634327888489, "learning_rate": 5.21944782908568e-06, "loss": 2.7801, "step": 257460 }, { "epoch": 0.0163072, "grad_norm": 0.878088116645813, "learning_rate": 5.218740544517091e-06, "loss": 2.7901, "step": 257470 }, { "epoch": 0.0163328, "grad_norm": 0.8362464904785156, "learning_rate": 5.218033290953699e-06, "loss": 2.7913, "step": 257480 }, { "epoch": 0.0163584, "grad_norm": 0.7447080016136169, "learning_rate": 5.21732606840009e-06, "loss": 3.0317, "step": 257490 }, { "epoch": 0.016384, "grad_norm": 1.6483877897262573, "learning_rate": 5.21661887686085e-06, "loss": 2.8919, "step": 257500 }, { "epoch": 0.0164096, "grad_norm": 0.7625596523284912, "learning_rate": 5.21591171634057e-06, "loss": 2.8387, "step": 257510 }, { "epoch": 0.0164352, "grad_norm": 2.7539970874786377, "learning_rate": 5.215204586843827e-06, "loss": 2.4454, "step": 257520 }, { "epoch": 0.0164608, "grad_norm": 0.8084027767181396, "learning_rate": 5.214497488375209e-06, "loss": 2.7044, "step": 257530 }, { "epoch": 0.0164864, "grad_norm": 0.8960153460502625, "learning_rate": 5.2137904209393065e-06, "loss": 2.9911, "step": 257540 }, { "epoch": 0.016512, "grad_norm": 1.5994361639022827, "learning_rate": 5.213083384540698e-06, "loss": 2.8525, "step": 257550 }, { "epoch": 0.0165376, "grad_norm": 0.7781383395195007, "learning_rate": 5.212376379183972e-06, "loss": 2.748, "step": 257560 }, { "epoch": 0.0165632, "grad_norm": 0.9599398970603943, "learning_rate": 5.211669404873714e-06, "loss": 2.9837, "step": 257570 }, { "epoch": 0.0165888, "grad_norm": 1.021482229232788, "learning_rate": 5.21096246161451e-06, "loss": 3.0741, "step": 257580 }, { "epoch": 0.0166144, "grad_norm": 0.9305902123451233, "learning_rate": 5.210255549410938e-06, "loss": 3.099, "step": 257590 }, { "epoch": 0.01664, "grad_norm": 0.7244032025337219, "learning_rate": 5.209548668267583e-06, "loss": 2.8935, "step": 257600 }, { "epoch": 0.0166656, "grad_norm": 0.8842983245849609, "learning_rate": 5.208841818189035e-06, "loss": 2.8458, "step": 257610 }, { "epoch": 0.0166912, "grad_norm": 0.8233888745307922, "learning_rate": 5.208134999179873e-06, "loss": 3.025, "step": 257620 }, { "epoch": 0.0167168, "grad_norm": 1.144218921661377, "learning_rate": 5.207428211244687e-06, "loss": 2.8412, "step": 257630 }, { "epoch": 0.0167424, "grad_norm": 0.7813051342964172, "learning_rate": 5.206721454388053e-06, "loss": 3.1081, "step": 257640 }, { "epoch": 0.016768, "grad_norm": 0.832568347454071, "learning_rate": 5.2060147286145544e-06, "loss": 2.7649, "step": 257650 }, { "epoch": 0.0167936, "grad_norm": 0.9554557204246521, "learning_rate": 5.2053080339287765e-06, "loss": 3.3792, "step": 257660 }, { "epoch": 0.0168192, "grad_norm": 0.8220101594924927, "learning_rate": 5.204601370335303e-06, "loss": 2.9416, "step": 257670 }, { "epoch": 0.0168448, "grad_norm": 0.7653006315231323, "learning_rate": 5.203894737838713e-06, "loss": 2.7533, "step": 257680 }, { "epoch": 0.0168704, "grad_norm": 0.7995777726173401, "learning_rate": 5.2031881364435934e-06, "loss": 2.899, "step": 257690 }, { "epoch": 0.016896, "grad_norm": 0.9250782132148743, "learning_rate": 5.2024815661545225e-06, "loss": 2.7147, "step": 257700 }, { "epoch": 0.0169216, "grad_norm": 0.7572040557861328, "learning_rate": 5.201775026976087e-06, "loss": 2.9769, "step": 257710 }, { "epoch": 0.0169472, "grad_norm": 0.8253322243690491, "learning_rate": 5.201068518912861e-06, "loss": 2.9262, "step": 257720 }, { "epoch": 0.0169728, "grad_norm": 0.9166663289070129, "learning_rate": 5.200362041969433e-06, "loss": 2.8757, "step": 257730 }, { "epoch": 0.0169984, "grad_norm": 0.795560896396637, "learning_rate": 5.1996555961503795e-06, "loss": 2.8155, "step": 257740 }, { "epoch": 0.017024, "grad_norm": 0.8976984620094299, "learning_rate": 5.198949181460284e-06, "loss": 3.115, "step": 257750 }, { "epoch": 0.0170496, "grad_norm": 0.8047394752502441, "learning_rate": 5.198242797903727e-06, "loss": 2.9125, "step": 257760 }, { "epoch": 0.0170752, "grad_norm": 0.9738863706588745, "learning_rate": 5.1975364454852895e-06, "loss": 3.0199, "step": 257770 }, { "epoch": 0.0171008, "grad_norm": 0.814636766910553, "learning_rate": 5.196830124209551e-06, "loss": 2.9546, "step": 257780 }, { "epoch": 0.0171264, "grad_norm": 1.0360357761383057, "learning_rate": 5.196123834081094e-06, "loss": 2.8735, "step": 257790 }, { "epoch": 0.017152, "grad_norm": 0.9126702547073364, "learning_rate": 5.195417575104497e-06, "loss": 2.8046, "step": 257800 }, { "epoch": 0.0171776, "grad_norm": 0.92905193567276, "learning_rate": 5.194711347284339e-06, "loss": 2.9344, "step": 257810 }, { "epoch": 0.0172032, "grad_norm": 0.9456483125686646, "learning_rate": 5.194005150625201e-06, "loss": 2.9405, "step": 257820 }, { "epoch": 0.0172288, "grad_norm": 0.8861285448074341, "learning_rate": 5.193298985131667e-06, "loss": 2.8208, "step": 257830 }, { "epoch": 0.0172544, "grad_norm": 0.7891942262649536, "learning_rate": 5.192592850808308e-06, "loss": 2.8227, "step": 257840 }, { "epoch": 0.01728, "grad_norm": 0.7958211302757263, "learning_rate": 5.191886747659707e-06, "loss": 3.0065, "step": 257850 }, { "epoch": 0.0173056, "grad_norm": 0.9224005937576294, "learning_rate": 5.191180675690442e-06, "loss": 2.9563, "step": 257860 }, { "epoch": 0.0173312, "grad_norm": 0.8585735559463501, "learning_rate": 5.190474634905092e-06, "loss": 2.8256, "step": 257870 }, { "epoch": 0.0173568, "grad_norm": 0.8184717297554016, "learning_rate": 5.189768625308234e-06, "loss": 2.7981, "step": 257880 }, { "epoch": 0.0173824, "grad_norm": 0.9056628942489624, "learning_rate": 5.18906264690445e-06, "loss": 2.8179, "step": 257890 }, { "epoch": 0.017408, "grad_norm": 0.8336194157600403, "learning_rate": 5.188356699698316e-06, "loss": 3.2207, "step": 257900 }, { "epoch": 0.0174336, "grad_norm": 0.815994918346405, "learning_rate": 5.187650783694415e-06, "loss": 2.7551, "step": 257910 }, { "epoch": 0.0174592, "grad_norm": 0.8044771552085876, "learning_rate": 5.186944898897315e-06, "loss": 2.7932, "step": 257920 }, { "epoch": 0.0174848, "grad_norm": 0.8041329383850098, "learning_rate": 5.186239045311596e-06, "loss": 2.8651, "step": 257930 }, { "epoch": 0.0175104, "grad_norm": 0.78834068775177, "learning_rate": 5.185533222941841e-06, "loss": 2.8694, "step": 257940 }, { "epoch": 0.017536, "grad_norm": 0.7946574091911316, "learning_rate": 5.184827431792617e-06, "loss": 3.0538, "step": 257950 }, { "epoch": 0.0175616, "grad_norm": 0.8502477407455444, "learning_rate": 5.184121671868516e-06, "loss": 2.9529, "step": 257960 }, { "epoch": 0.0175872, "grad_norm": 0.835500180721283, "learning_rate": 5.183415943174103e-06, "loss": 2.8842, "step": 257970 }, { "epoch": 0.0176128, "grad_norm": 1.0084668397903442, "learning_rate": 5.182710245713957e-06, "loss": 2.9374, "step": 257980 }, { "epoch": 0.0176384, "grad_norm": 0.8925259113311768, "learning_rate": 5.182004579492654e-06, "loss": 2.8659, "step": 257990 }, { "epoch": 0.017664, "grad_norm": 0.7747271060943604, "learning_rate": 5.181298944514772e-06, "loss": 2.9367, "step": 258000 }, { "epoch": 0.0176896, "grad_norm": 0.8675567507743835, "learning_rate": 5.180593340784885e-06, "loss": 2.7723, "step": 258010 }, { "epoch": 0.0177152, "grad_norm": 0.8456153273582458, "learning_rate": 5.17988776830757e-06, "loss": 2.8316, "step": 258020 }, { "epoch": 0.0177408, "grad_norm": 0.8399059176445007, "learning_rate": 5.179182227087405e-06, "loss": 2.8319, "step": 258030 }, { "epoch": 0.0177664, "grad_norm": 0.7477583289146423, "learning_rate": 5.178476717128957e-06, "loss": 2.8394, "step": 258040 }, { "epoch": 0.017792, "grad_norm": 1.1374295949935913, "learning_rate": 5.177771238436809e-06, "loss": 2.8047, "step": 258050 }, { "epoch": 0.0178176, "grad_norm": 0.7968860864639282, "learning_rate": 5.1770657910155296e-06, "loss": 2.9784, "step": 258060 }, { "epoch": 0.0178432, "grad_norm": 0.9001362323760986, "learning_rate": 5.176360374869699e-06, "loss": 3.0151, "step": 258070 }, { "epoch": 0.0178688, "grad_norm": 0.7993898391723633, "learning_rate": 5.1756549900038886e-06, "loss": 2.9598, "step": 258080 }, { "epoch": 0.0178944, "grad_norm": 0.8145325779914856, "learning_rate": 5.174949636422674e-06, "loss": 2.8181, "step": 258090 }, { "epoch": 0.01792, "grad_norm": 1.3389334678649902, "learning_rate": 5.174244314130632e-06, "loss": 2.8472, "step": 258100 }, { "epoch": 0.0179456, "grad_norm": 0.9192495346069336, "learning_rate": 5.17353902313233e-06, "loss": 3.0711, "step": 258110 }, { "epoch": 0.0179712, "grad_norm": 0.8639244437217712, "learning_rate": 5.17283376343234e-06, "loss": 2.6987, "step": 258120 }, { "epoch": 0.0179968, "grad_norm": 0.8058681488037109, "learning_rate": 5.172128535035243e-06, "loss": 2.937, "step": 258130 }, { "epoch": 0.0180224, "grad_norm": 0.7509017586708069, "learning_rate": 5.171423337945611e-06, "loss": 2.992, "step": 258140 }, { "epoch": 0.018048, "grad_norm": 0.8301562666893005, "learning_rate": 5.1707181721680155e-06, "loss": 2.8684, "step": 258150 }, { "epoch": 0.0180736, "grad_norm": 0.8381595015525818, "learning_rate": 5.170013037707033e-06, "loss": 2.8404, "step": 258160 }, { "epoch": 0.0180992, "grad_norm": 0.8502675294876099, "learning_rate": 5.169307934567227e-06, "loss": 2.9087, "step": 258170 }, { "epoch": 0.0181248, "grad_norm": 0.8727567791938782, "learning_rate": 5.168602862753177e-06, "loss": 2.826, "step": 258180 }, { "epoch": 0.0181504, "grad_norm": 0.7412412762641907, "learning_rate": 5.167897822269452e-06, "loss": 2.8034, "step": 258190 }, { "epoch": 0.018176, "grad_norm": 0.9018987417221069, "learning_rate": 5.167192813120627e-06, "loss": 3.0507, "step": 258200 }, { "epoch": 0.0182016, "grad_norm": 0.8297238349914551, "learning_rate": 5.1664878353112715e-06, "loss": 2.8185, "step": 258210 }, { "epoch": 0.0182272, "grad_norm": 0.7993646860122681, "learning_rate": 5.165782888845956e-06, "loss": 2.897, "step": 258220 }, { "epoch": 0.0182528, "grad_norm": 0.7892695069313049, "learning_rate": 5.16507797372926e-06, "loss": 2.9669, "step": 258230 }, { "epoch": 0.0182784, "grad_norm": 0.8452648520469666, "learning_rate": 5.164373089965744e-06, "loss": 2.7923, "step": 258240 }, { "epoch": 0.018304, "grad_norm": 0.7525089383125305, "learning_rate": 5.163668237559982e-06, "loss": 2.9017, "step": 258250 }, { "epoch": 0.0183296, "grad_norm": 1.0162194967269897, "learning_rate": 5.162963416516547e-06, "loss": 2.9605, "step": 258260 }, { "epoch": 0.0183552, "grad_norm": 0.7725000381469727, "learning_rate": 5.162258626840009e-06, "loss": 2.9643, "step": 258270 }, { "epoch": 0.0183808, "grad_norm": 0.7704557776451111, "learning_rate": 5.161553868534938e-06, "loss": 2.9567, "step": 258280 }, { "epoch": 0.0184064, "grad_norm": 0.9182236790657043, "learning_rate": 5.160849141605904e-06, "loss": 2.9481, "step": 258290 }, { "epoch": 0.018432, "grad_norm": 0.7806254625320435, "learning_rate": 5.1601444460574776e-06, "loss": 2.8934, "step": 258300 }, { "epoch": 0.0184576, "grad_norm": 0.9387162923812866, "learning_rate": 5.159439781894227e-06, "loss": 2.8739, "step": 258310 }, { "epoch": 0.0184832, "grad_norm": 0.8626856803894043, "learning_rate": 5.158735149120724e-06, "loss": 2.7813, "step": 258320 }, { "epoch": 0.0185088, "grad_norm": 0.7265539765357971, "learning_rate": 5.158030547741537e-06, "loss": 2.8922, "step": 258330 }, { "epoch": 0.0185344, "grad_norm": 0.778622031211853, "learning_rate": 5.157325977761235e-06, "loss": 3.0259, "step": 258340 }, { "epoch": 0.01856, "grad_norm": 0.7597065567970276, "learning_rate": 5.156621439184386e-06, "loss": 3.0418, "step": 258350 }, { "epoch": 0.0185856, "grad_norm": 0.9854649305343628, "learning_rate": 5.155916932015565e-06, "loss": 2.9097, "step": 258360 }, { "epoch": 0.0186112, "grad_norm": 0.7182836532592773, "learning_rate": 5.155212456259331e-06, "loss": 2.5232, "step": 258370 }, { "epoch": 0.0186368, "grad_norm": 0.8256639838218689, "learning_rate": 5.154508011920256e-06, "loss": 3.0593, "step": 258380 }, { "epoch": 0.0186624, "grad_norm": 0.7994951009750366, "learning_rate": 5.15380359900291e-06, "loss": 2.8809, "step": 258390 }, { "epoch": 0.018688, "grad_norm": 0.7374458909034729, "learning_rate": 5.1530992175118585e-06, "loss": 3.1068, "step": 258400 }, { "epoch": 0.0187136, "grad_norm": 0.7642658352851868, "learning_rate": 5.152394867451671e-06, "loss": 2.7546, "step": 258410 }, { "epoch": 0.0187392, "grad_norm": 0.8053738474845886, "learning_rate": 5.151690548826915e-06, "loss": 3.0038, "step": 258420 }, { "epoch": 0.0187648, "grad_norm": 0.8731200098991394, "learning_rate": 5.150986261642161e-06, "loss": 2.8748, "step": 258430 }, { "epoch": 0.0187904, "grad_norm": 0.9363320469856262, "learning_rate": 5.150282005901968e-06, "loss": 2.6988, "step": 258440 }, { "epoch": 0.018816, "grad_norm": 1.0404528379440308, "learning_rate": 5.149577781610906e-06, "loss": 2.9878, "step": 258450 }, { "epoch": 0.0188416, "grad_norm": 0.7995098233222961, "learning_rate": 5.148873588773544e-06, "loss": 2.8171, "step": 258460 }, { "epoch": 0.0188672, "grad_norm": 0.7819361090660095, "learning_rate": 5.148169427394445e-06, "loss": 2.9146, "step": 258470 }, { "epoch": 0.0188928, "grad_norm": 0.9598925709724426, "learning_rate": 5.147465297478181e-06, "loss": 2.7498, "step": 258480 }, { "epoch": 0.0189184, "grad_norm": 0.7668570876121521, "learning_rate": 5.1467611990293175e-06, "loss": 2.9067, "step": 258490 }, { "epoch": 0.018944, "grad_norm": 0.9010541439056396, "learning_rate": 5.146057132052414e-06, "loss": 2.7867, "step": 258500 }, { "epoch": 0.0189696, "grad_norm": 0.7938972115516663, "learning_rate": 5.145353096552042e-06, "loss": 2.9563, "step": 258510 }, { "epoch": 0.0189952, "grad_norm": 0.786846935749054, "learning_rate": 5.144649092532763e-06, "loss": 2.9092, "step": 258520 }, { "epoch": 0.0190208, "grad_norm": 0.7921646237373352, "learning_rate": 5.1439451199991455e-06, "loss": 2.7604, "step": 258530 }, { "epoch": 0.0190464, "grad_norm": 0.8759034872055054, "learning_rate": 5.143241178955752e-06, "loss": 2.8349, "step": 258540 }, { "epoch": 0.019072, "grad_norm": 0.7512286305427551, "learning_rate": 5.14253726940715e-06, "loss": 2.7875, "step": 258550 }, { "epoch": 0.0190976, "grad_norm": 0.8004568815231323, "learning_rate": 5.1418333913579055e-06, "loss": 2.964, "step": 258560 }, { "epoch": 0.0191232, "grad_norm": 0.8366153240203857, "learning_rate": 5.1411295448125776e-06, "loss": 2.9881, "step": 258570 }, { "epoch": 0.0191488, "grad_norm": 0.8205683827400208, "learning_rate": 5.1404257297757335e-06, "loss": 2.8799, "step": 258580 }, { "epoch": 0.0191744, "grad_norm": 0.9330136179924011, "learning_rate": 5.1397219462519354e-06, "loss": 2.9279, "step": 258590 }, { "epoch": 0.0192, "grad_norm": 0.7133328914642334, "learning_rate": 5.13901819424575e-06, "loss": 2.715, "step": 258600 }, { "epoch": 0.0192256, "grad_norm": 0.9063575863838196, "learning_rate": 5.138314473761738e-06, "loss": 2.701, "step": 258610 }, { "epoch": 0.0192512, "grad_norm": 0.7400994300842285, "learning_rate": 5.137610784804467e-06, "loss": 2.8453, "step": 258620 }, { "epoch": 0.0192768, "grad_norm": 0.8683338165283203, "learning_rate": 5.136907127378501e-06, "loss": 2.9285, "step": 258630 }, { "epoch": 0.0193024, "grad_norm": 0.8914021849632263, "learning_rate": 5.136203501488395e-06, "loss": 3.005, "step": 258640 }, { "epoch": 0.019328, "grad_norm": 0.9361183643341064, "learning_rate": 5.135499907138713e-06, "loss": 1.9762, "step": 258650 }, { "epoch": 0.0193536, "grad_norm": 0.8225476741790771, "learning_rate": 5.1347963443340256e-06, "loss": 2.8015, "step": 258660 }, { "epoch": 0.0193792, "grad_norm": 1.0103076696395874, "learning_rate": 5.13409281307889e-06, "loss": 2.7021, "step": 258670 }, { "epoch": 0.0194048, "grad_norm": 0.7633051872253418, "learning_rate": 5.13338931337787e-06, "loss": 2.4951, "step": 258680 }, { "epoch": 0.0194304, "grad_norm": 0.7954878807067871, "learning_rate": 5.132685845235532e-06, "loss": 2.8192, "step": 258690 }, { "epoch": 0.019456, "grad_norm": 0.8134709000587463, "learning_rate": 5.1319824086564275e-06, "loss": 2.5986, "step": 258700 }, { "epoch": 0.0194816, "grad_norm": 0.8998852968215942, "learning_rate": 5.131279003645123e-06, "loss": 2.6272, "step": 258710 }, { "epoch": 0.0195072, "grad_norm": 0.7421534657478333, "learning_rate": 5.13057563020618e-06, "loss": 2.7094, "step": 258720 }, { "epoch": 0.0195328, "grad_norm": 0.900518000125885, "learning_rate": 5.129872288344162e-06, "loss": 2.541, "step": 258730 }, { "epoch": 0.0195584, "grad_norm": 0.8954731225967407, "learning_rate": 5.129168978063626e-06, "loss": 2.7032, "step": 258740 }, { "epoch": 0.019584, "grad_norm": 0.798043966293335, "learning_rate": 5.128465699369135e-06, "loss": 2.6508, "step": 258750 }, { "epoch": 0.0196096, "grad_norm": 0.7608800530433655, "learning_rate": 5.1277624522652535e-06, "loss": 2.3651, "step": 258760 }, { "epoch": 0.0196352, "grad_norm": 0.8823991417884827, "learning_rate": 5.127059236756533e-06, "loss": 2.9305, "step": 258770 }, { "epoch": 0.0196608, "grad_norm": 0.8133881688117981, "learning_rate": 5.126356052847539e-06, "loss": 2.6244, "step": 258780 }, { "epoch": 0.0196864, "grad_norm": 0.80521559715271, "learning_rate": 5.12565290054283e-06, "loss": 2.6245, "step": 258790 }, { "epoch": 0.019712, "grad_norm": 0.8229536414146423, "learning_rate": 5.124949779846969e-06, "loss": 2.7089, "step": 258800 }, { "epoch": 0.0197376, "grad_norm": 0.7866284251213074, "learning_rate": 5.12424669076451e-06, "loss": 2.534, "step": 258810 }, { "epoch": 0.0197632, "grad_norm": 0.7701188921928406, "learning_rate": 5.123543633300017e-06, "loss": 2.514, "step": 258820 }, { "epoch": 0.0197888, "grad_norm": 0.8590112328529358, "learning_rate": 5.1228406074580465e-06, "loss": 2.7972, "step": 258830 }, { "epoch": 0.0198144, "grad_norm": 0.7878186702728271, "learning_rate": 5.122137613243159e-06, "loss": 2.6754, "step": 258840 }, { "epoch": 0.01984, "grad_norm": 0.75508052110672, "learning_rate": 5.121434650659912e-06, "loss": 2.6795, "step": 258850 }, { "epoch": 0.0198656, "grad_norm": 0.8161941170692444, "learning_rate": 5.120731719712866e-06, "loss": 2.6148, "step": 258860 }, { "epoch": 0.0198912, "grad_norm": 0.8273772597312927, "learning_rate": 5.1200288204065776e-06, "loss": 2.5712, "step": 258870 }, { "epoch": 0.0199168, "grad_norm": 0.8072066903114319, "learning_rate": 5.119325952745604e-06, "loss": 2.4343, "step": 258880 }, { "epoch": 0.0199424, "grad_norm": 0.9042631983757019, "learning_rate": 5.1186231167345105e-06, "loss": 2.6919, "step": 258890 }, { "epoch": 0.019968, "grad_norm": 0.8596765995025635, "learning_rate": 5.117920312377844e-06, "loss": 2.6737, "step": 258900 }, { "epoch": 0.0199936, "grad_norm": 0.7897853851318359, "learning_rate": 5.117217539680166e-06, "loss": 2.6385, "step": 258910 }, { "epoch": 0.0200192, "grad_norm": 0.8592568635940552, "learning_rate": 5.116514798646035e-06, "loss": 2.9402, "step": 258920 }, { "epoch": 0.0200448, "grad_norm": 1.0876494646072388, "learning_rate": 5.115812089280007e-06, "loss": 2.681, "step": 258930 }, { "epoch": 0.0200704, "grad_norm": 0.7572733759880066, "learning_rate": 5.115109411586641e-06, "loss": 2.6897, "step": 258940 }, { "epoch": 0.020096, "grad_norm": 0.8043922781944275, "learning_rate": 5.114406765570492e-06, "loss": 2.6521, "step": 258950 }, { "epoch": 0.0201216, "grad_norm": 0.7672666311264038, "learning_rate": 5.113704151236121e-06, "loss": 2.7641, "step": 258960 }, { "epoch": 0.0201472, "grad_norm": 0.827333390712738, "learning_rate": 5.113001568588075e-06, "loss": 2.5989, "step": 258970 }, { "epoch": 0.0201728, "grad_norm": 0.9144535660743713, "learning_rate": 5.112299017630915e-06, "loss": 2.8231, "step": 258980 }, { "epoch": 0.0201984, "grad_norm": 0.8970019221305847, "learning_rate": 5.111596498369194e-06, "loss": 2.7282, "step": 258990 }, { "epoch": 0.020224, "grad_norm": 1.3344950675964355, "learning_rate": 5.110894010807474e-06, "loss": 3.0314, "step": 259000 }, { "epoch": 0.0202496, "grad_norm": 0.6637527346611023, "learning_rate": 5.110191554950306e-06, "loss": 2.475, "step": 259010 }, { "epoch": 0.0202752, "grad_norm": 0.7844653725624084, "learning_rate": 5.109489130802251e-06, "loss": 2.6357, "step": 259020 }, { "epoch": 0.0203008, "grad_norm": 0.8095828890800476, "learning_rate": 5.108786738367855e-06, "loss": 2.6084, "step": 259030 }, { "epoch": 0.0203264, "grad_norm": 0.8733289241790771, "learning_rate": 5.108084377651677e-06, "loss": 2.8837, "step": 259040 }, { "epoch": 0.020352, "grad_norm": 0.8417125344276428, "learning_rate": 5.107382048658271e-06, "loss": 2.7898, "step": 259050 }, { "epoch": 0.0203776, "grad_norm": 0.7802391648292542, "learning_rate": 5.106679751392194e-06, "loss": 2.6303, "step": 259060 }, { "epoch": 0.0204032, "grad_norm": 0.7852710485458374, "learning_rate": 5.105977485857998e-06, "loss": 2.4974, "step": 259070 }, { "epoch": 0.0204288, "grad_norm": 0.7440178990364075, "learning_rate": 5.1052752520602356e-06, "loss": 2.4398, "step": 259080 }, { "epoch": 0.0204544, "grad_norm": 0.9914191365242004, "learning_rate": 5.104573050003467e-06, "loss": 2.7433, "step": 259090 }, { "epoch": 0.02048, "grad_norm": 0.7969359755516052, "learning_rate": 5.103870879692239e-06, "loss": 3.0071, "step": 259100 }, { "epoch": 0.0205056, "grad_norm": 1.063472032546997, "learning_rate": 5.103168741131106e-06, "loss": 2.7042, "step": 259110 }, { "epoch": 0.0205312, "grad_norm": 1.0620026588439941, "learning_rate": 5.102466634324621e-06, "loss": 2.9476, "step": 259120 }, { "epoch": 0.0205568, "grad_norm": 0.8670602440834045, "learning_rate": 5.1017645592773405e-06, "loss": 2.7771, "step": 259130 }, { "epoch": 0.0205824, "grad_norm": 0.773442268371582, "learning_rate": 5.101062515993813e-06, "loss": 2.9154, "step": 259140 }, { "epoch": 0.020608, "grad_norm": 0.8348127603530884, "learning_rate": 5.100360504478594e-06, "loss": 2.4988, "step": 259150 }, { "epoch": 0.0206336, "grad_norm": 0.9212081432342529, "learning_rate": 5.099658524736239e-06, "loss": 2.7546, "step": 259160 }, { "epoch": 0.0206592, "grad_norm": 0.852956235408783, "learning_rate": 5.098956576771288e-06, "loss": 2.584, "step": 259170 }, { "epoch": 0.0206848, "grad_norm": 0.9440213441848755, "learning_rate": 5.0982546605883044e-06, "loss": 2.7497, "step": 259180 }, { "epoch": 0.0207104, "grad_norm": 0.9608725309371948, "learning_rate": 5.097552776191837e-06, "loss": 2.8242, "step": 259190 }, { "epoch": 0.020736, "grad_norm": 0.8516179919242859, "learning_rate": 5.0968509235864376e-06, "loss": 2.7054, "step": 259200 }, { "epoch": 0.0207616, "grad_norm": 0.9078637957572937, "learning_rate": 5.09614910277666e-06, "loss": 2.7595, "step": 259210 }, { "epoch": 0.0207872, "grad_norm": 0.7468504905700684, "learning_rate": 5.095447313767048e-06, "loss": 2.3967, "step": 259220 }, { "epoch": 0.0208128, "grad_norm": 0.7944355607032776, "learning_rate": 5.094745556562157e-06, "loss": 2.4265, "step": 259230 }, { "epoch": 0.0208384, "grad_norm": 0.8207970261573792, "learning_rate": 5.094043831166538e-06, "loss": 2.2639, "step": 259240 }, { "epoch": 0.020864, "grad_norm": 0.9722916483879089, "learning_rate": 5.093342137584741e-06, "loss": 2.747, "step": 259250 }, { "epoch": 0.0208896, "grad_norm": 0.8381938338279724, "learning_rate": 5.092640475821316e-06, "loss": 2.671, "step": 259260 }, { "epoch": 0.0209152, "grad_norm": 0.7953352332115173, "learning_rate": 5.091938845880813e-06, "loss": 2.4938, "step": 259270 }, { "epoch": 0.0209408, "grad_norm": 0.8332078456878662, "learning_rate": 5.091237247767782e-06, "loss": 2.5759, "step": 259280 }, { "epoch": 0.0209664, "grad_norm": 0.8432919383049011, "learning_rate": 5.090535681486777e-06, "loss": 2.6893, "step": 259290 }, { "epoch": 0.020992, "grad_norm": 0.8388484120368958, "learning_rate": 5.089834147042338e-06, "loss": 2.8147, "step": 259300 }, { "epoch": 2.56e-05, "grad_norm": 0.81269371509552, "learning_rate": 5.089132644439022e-06, "loss": 2.0371, "step": 259310 }, { "epoch": 5.12e-05, "grad_norm": 0.8241093158721924, "learning_rate": 5.088431173681375e-06, "loss": 2.7111, "step": 259320 }, { "epoch": 7.68e-05, "grad_norm": 0.7681488394737244, "learning_rate": 5.087729734773946e-06, "loss": 2.5629, "step": 259330 }, { "epoch": 0.0001024, "grad_norm": 0.8588348627090454, "learning_rate": 5.087028327721284e-06, "loss": 2.7711, "step": 259340 }, { "epoch": 0.000128, "grad_norm": 0.8190664649009705, "learning_rate": 5.086326952527938e-06, "loss": 2.8151, "step": 259350 }, { "epoch": 0.0001536, "grad_norm": 0.8243175148963928, "learning_rate": 5.0856256091984565e-06, "loss": 2.753, "step": 259360 }, { "epoch": 0.0001792, "grad_norm": 0.8426821827888489, "learning_rate": 5.084924297737386e-06, "loss": 2.6224, "step": 259370 }, { "epoch": 0.0002048, "grad_norm": 0.8241744041442871, "learning_rate": 5.084223018149275e-06, "loss": 2.7411, "step": 259380 }, { "epoch": 0.0002304, "grad_norm": 0.7995104789733887, "learning_rate": 5.08352177043867e-06, "loss": 2.526, "step": 259390 }, { "epoch": 0.000256, "grad_norm": 0.9151366949081421, "learning_rate": 5.082820554610122e-06, "loss": 2.571, "step": 259400 }, { "epoch": 0.0002816, "grad_norm": 0.9095807671546936, "learning_rate": 5.082119370668179e-06, "loss": 2.4156, "step": 259410 }, { "epoch": 0.0003072, "grad_norm": 0.8040733337402344, "learning_rate": 5.081418218617381e-06, "loss": 2.3186, "step": 259420 }, { "epoch": 0.0003328, "grad_norm": 0.8887538909912109, "learning_rate": 5.080717098462279e-06, "loss": 2.3483, "step": 259430 }, { "epoch": 0.0003584, "grad_norm": 0.8842424750328064, "learning_rate": 5.080016010207418e-06, "loss": 2.7499, "step": 259440 }, { "epoch": 0.000384, "grad_norm": 0.8392354249954224, "learning_rate": 5.079314953857346e-06, "loss": 2.7696, "step": 259450 }, { "epoch": 0.0004096, "grad_norm": 0.8794389963150024, "learning_rate": 5.078613929416609e-06, "loss": 2.4298, "step": 259460 }, { "epoch": 0.0004352, "grad_norm": 0.7681273221969604, "learning_rate": 5.077912936889754e-06, "loss": 2.7625, "step": 259470 }, { "epoch": 0.0004608, "grad_norm": 0.7874820232391357, "learning_rate": 5.077211976281325e-06, "loss": 2.5495, "step": 259480 }, { "epoch": 0.0004864, "grad_norm": 0.8623689413070679, "learning_rate": 5.07651104759587e-06, "loss": 2.6506, "step": 259490 }, { "epoch": 0.000512, "grad_norm": 0.7541446089744568, "learning_rate": 5.07581015083793e-06, "loss": 2.5207, "step": 259500 }, { "epoch": 0.0005376, "grad_norm": 0.8013522624969482, "learning_rate": 5.075109286012052e-06, "loss": 2.7531, "step": 259510 }, { "epoch": 0.0005632, "grad_norm": 0.812432050704956, "learning_rate": 5.0744084531227776e-06, "loss": 2.6709, "step": 259520 }, { "epoch": 0.0005888, "grad_norm": 0.7619944214820862, "learning_rate": 5.0737076521746595e-06, "loss": 2.5301, "step": 259530 }, { "epoch": 0.0006144, "grad_norm": 0.8930966854095459, "learning_rate": 5.0730068831722425e-06, "loss": 2.6832, "step": 259540 }, { "epoch": 0.00064, "grad_norm": 0.8536672592163086, "learning_rate": 5.072306146120061e-06, "loss": 2.5955, "step": 259550 }, { "epoch": 0.0006656, "grad_norm": 0.8451758623123169, "learning_rate": 5.0716054410226645e-06, "loss": 2.8772, "step": 259560 }, { "epoch": 0.0006912, "grad_norm": 0.8651977777481079, "learning_rate": 5.070904767884599e-06, "loss": 2.7341, "step": 259570 }, { "epoch": 0.0007168, "grad_norm": 0.7772446274757385, "learning_rate": 5.070204126710404e-06, "loss": 2.7014, "step": 259580 }, { "epoch": 0.0007424, "grad_norm": 0.8538591861724854, "learning_rate": 5.069503517504626e-06, "loss": 2.7978, "step": 259590 }, { "epoch": 0.000768, "grad_norm": 0.9057173132896423, "learning_rate": 5.068802940271807e-06, "loss": 2.8769, "step": 259600 }, { "epoch": 0.0007936, "grad_norm": 0.8393978476524353, "learning_rate": 5.068102395016494e-06, "loss": 2.6132, "step": 259610 }, { "epoch": 0.0008192, "grad_norm": 0.8070728778839111, "learning_rate": 5.0674018817432235e-06, "loss": 2.6826, "step": 259620 }, { "epoch": 0.0008448, "grad_norm": 0.9203263521194458, "learning_rate": 5.06670140045654e-06, "loss": 2.5971, "step": 259630 }, { "epoch": 0.0008704, "grad_norm": 0.8388210535049438, "learning_rate": 5.066000951160986e-06, "loss": 2.706, "step": 259640 }, { "epoch": 0.000896, "grad_norm": 0.841425895690918, "learning_rate": 5.065300533861106e-06, "loss": 2.6842, "step": 259650 }, { "epoch": 0.0009216, "grad_norm": 0.8550062775611877, "learning_rate": 5.06460014856144e-06, "loss": 2.6375, "step": 259660 }, { "epoch": 0.0009472, "grad_norm": 0.7769721746444702, "learning_rate": 5.06389979526653e-06, "loss": 2.7034, "step": 259670 }, { "epoch": 0.0009728, "grad_norm": 0.7753776907920837, "learning_rate": 5.0631994739809185e-06, "loss": 2.7998, "step": 259680 }, { "epoch": 0.0009984, "grad_norm": 0.8500498533248901, "learning_rate": 5.062499184709147e-06, "loss": 2.7732, "step": 259690 }, { "epoch": 0.001024, "grad_norm": 0.8660815358161926, "learning_rate": 5.061798927455754e-06, "loss": 2.7036, "step": 259700 }, { "epoch": 0.0010496, "grad_norm": 0.8799214363098145, "learning_rate": 5.061098702225285e-06, "loss": 2.5422, "step": 259710 }, { "epoch": 0.0010752, "grad_norm": 0.7763156294822693, "learning_rate": 5.060398509022276e-06, "loss": 2.6142, "step": 259720 }, { "epoch": 0.0011008, "grad_norm": 0.8579939007759094, "learning_rate": 5.0596983478512716e-06, "loss": 2.7243, "step": 259730 }, { "epoch": 0.0011264, "grad_norm": 0.7575529217720032, "learning_rate": 5.058998218716813e-06, "loss": 2.7399, "step": 259740 }, { "epoch": 0.001152, "grad_norm": 1.110063910484314, "learning_rate": 5.058298121623435e-06, "loss": 2.5653, "step": 259750 }, { "epoch": 0.0011776, "grad_norm": 0.799325704574585, "learning_rate": 5.05759805657568e-06, "loss": 2.9334, "step": 259760 }, { "epoch": 0.0012032, "grad_norm": 0.780531108379364, "learning_rate": 5.056898023578087e-06, "loss": 2.7923, "step": 259770 }, { "epoch": 0.0012288, "grad_norm": 0.9068752527236938, "learning_rate": 5.056198022635199e-06, "loss": 2.9552, "step": 259780 }, { "epoch": 0.0012544, "grad_norm": 0.8391881585121155, "learning_rate": 5.055498053751551e-06, "loss": 2.7646, "step": 259790 }, { "epoch": 0.00128, "grad_norm": 0.9247125387191772, "learning_rate": 5.054798116931685e-06, "loss": 2.7248, "step": 259800 }, { "epoch": 0.0013056, "grad_norm": 0.7942794561386108, "learning_rate": 5.054098212180142e-06, "loss": 2.7618, "step": 259810 }, { "epoch": 0.0013312, "grad_norm": 0.9175695180892944, "learning_rate": 5.053398339501454e-06, "loss": 2.8421, "step": 259820 }, { "epoch": 0.0013568, "grad_norm": 0.8517060279846191, "learning_rate": 5.052698498900164e-06, "loss": 2.8283, "step": 259830 }, { "epoch": 0.0013824, "grad_norm": 0.9246159195899963, "learning_rate": 5.0519986903808084e-06, "loss": 2.8783, "step": 259840 }, { "epoch": 0.001408, "grad_norm": 0.7538278698921204, "learning_rate": 5.0512989139479264e-06, "loss": 2.687, "step": 259850 }, { "epoch": 0.0014336, "grad_norm": 0.9075462222099304, "learning_rate": 5.050599169606052e-06, "loss": 2.6895, "step": 259860 }, { "epoch": 0.0014592, "grad_norm": 0.8203365206718445, "learning_rate": 5.049899457359734e-06, "loss": 2.6275, "step": 259870 }, { "epoch": 0.0014848, "grad_norm": 0.8687867522239685, "learning_rate": 5.049199777213498e-06, "loss": 2.4352, "step": 259880 }, { "epoch": 0.0015104, "grad_norm": 0.9826704263687134, "learning_rate": 5.0485001291718875e-06, "loss": 2.2563, "step": 259890 }, { "epoch": 0.001536, "grad_norm": 0.8015427589416504, "learning_rate": 5.047800513239436e-06, "loss": 2.4831, "step": 259900 }, { "epoch": 2.56e-05, "grad_norm": 0.8050435781478882, "learning_rate": 5.047100929420683e-06, "loss": 1.8082, "step": 259910 }, { "epoch": 5.12e-05, "grad_norm": 0.8690884113311768, "learning_rate": 5.046401377720165e-06, "loss": 2.7225, "step": 259920 }, { "epoch": 7.68e-05, "grad_norm": 0.7963715195655823, "learning_rate": 5.0457018581424155e-06, "loss": 2.5947, "step": 259930 }, { "epoch": 0.0001024, "grad_norm": 0.8283849954605103, "learning_rate": 5.045002370691978e-06, "loss": 2.8121, "step": 259940 }, { "epoch": 0.000128, "grad_norm": 0.8085004091262817, "learning_rate": 5.044302915373378e-06, "loss": 2.7232, "step": 259950 }, { "epoch": 0.0001536, "grad_norm": 0.8275526165962219, "learning_rate": 5.043603492191158e-06, "loss": 2.8372, "step": 259960 }, { "epoch": 0.0001792, "grad_norm": 0.8607044219970703, "learning_rate": 5.042904101149853e-06, "loss": 2.6447, "step": 259970 }, { "epoch": 0.0002048, "grad_norm": 0.77863609790802, "learning_rate": 5.042204742253996e-06, "loss": 2.5642, "step": 259980 }, { "epoch": 0.0002304, "grad_norm": 0.8170357346534729, "learning_rate": 5.041505415508123e-06, "loss": 2.725, "step": 259990 }, { "epoch": 0.000256, "grad_norm": 0.8687834739685059, "learning_rate": 5.04080612091677e-06, "loss": 2.7746, "step": 260000 }, { "epoch": 0.0002816, "grad_norm": 0.9999313950538635, "learning_rate": 5.040106858484474e-06, "loss": 2.5907, "step": 260010 }, { "epoch": 0.0003072, "grad_norm": 0.8468592762947083, "learning_rate": 5.039407628215767e-06, "loss": 2.4809, "step": 260020 }, { "epoch": 0.0003328, "grad_norm": 0.8080777525901794, "learning_rate": 5.03870843011518e-06, "loss": 2.9822, "step": 260030 }, { "epoch": 0.0003584, "grad_norm": 0.849205493927002, "learning_rate": 5.038009264187247e-06, "loss": 2.6881, "step": 260040 }, { "epoch": 0.000384, "grad_norm": 0.8256101012229919, "learning_rate": 5.0373101304365076e-06, "loss": 2.7378, "step": 260050 }, { "epoch": 0.0004096, "grad_norm": 0.7459841966629028, "learning_rate": 5.0366110288674955e-06, "loss": 2.4742, "step": 260060 }, { "epoch": 0.0004352, "grad_norm": 0.8371037840843201, "learning_rate": 5.035911959484745e-06, "loss": 2.7742, "step": 260070 }, { "epoch": 0.0004608, "grad_norm": 0.7059394717216492, "learning_rate": 5.035212922292782e-06, "loss": 2.5897, "step": 260080 }, { "epoch": 0.0004864, "grad_norm": 0.9000685811042786, "learning_rate": 5.034513917296143e-06, "loss": 2.7209, "step": 260090 }, { "epoch": 0.000512, "grad_norm": 0.8991557955741882, "learning_rate": 5.033814944499362e-06, "loss": 2.5837, "step": 260100 }, { "epoch": 0.0005376, "grad_norm": 0.8423190712928772, "learning_rate": 5.033116003906972e-06, "loss": 2.807, "step": 260110 }, { "epoch": 0.0005632, "grad_norm": 0.8163170218467712, "learning_rate": 5.032417095523504e-06, "loss": 2.5258, "step": 260120 }, { "epoch": 0.0005888, "grad_norm": 0.9823347330093384, "learning_rate": 5.031718219353492e-06, "loss": 2.3934, "step": 260130 }, { "epoch": 0.0006144, "grad_norm": 0.8220852613449097, "learning_rate": 5.031019375401469e-06, "loss": 2.6091, "step": 260140 }, { "epoch": 0.00064, "grad_norm": 0.9081718921661377, "learning_rate": 5.030320563671962e-06, "loss": 2.9239, "step": 260150 }, { "epoch": 0.0006656, "grad_norm": 0.8783406019210815, "learning_rate": 5.029621784169505e-06, "loss": 3.0839, "step": 260160 }, { "epoch": 0.0006912, "grad_norm": 0.8170782923698425, "learning_rate": 5.0289230368986295e-06, "loss": 2.788, "step": 260170 }, { "epoch": 0.0007168, "grad_norm": 0.8315532803535461, "learning_rate": 5.028224321863869e-06, "loss": 2.6334, "step": 260180 }, { "epoch": 0.0007424, "grad_norm": 0.896939754486084, "learning_rate": 5.02752563906975e-06, "loss": 2.6265, "step": 260190 }, { "epoch": 0.000768, "grad_norm": 0.744107723236084, "learning_rate": 5.026826988520806e-06, "loss": 2.5677, "step": 260200 }, { "epoch": 0.0007936, "grad_norm": 0.8576726317405701, "learning_rate": 5.026128370221568e-06, "loss": 2.759, "step": 260210 }, { "epoch": 0.0008192, "grad_norm": 0.8227545022964478, "learning_rate": 5.0254297841765655e-06, "loss": 2.6818, "step": 260220 }, { "epoch": 0.0008448, "grad_norm": 0.7947182059288025, "learning_rate": 5.024731230390328e-06, "loss": 2.6253, "step": 260230 }, { "epoch": 0.0008704, "grad_norm": 0.7973987460136414, "learning_rate": 5.024032708867388e-06, "loss": 2.5888, "step": 260240 }, { "epoch": 0.000896, "grad_norm": 0.8235663175582886, "learning_rate": 5.023334219612271e-06, "loss": 2.3597, "step": 260250 }, { "epoch": 0.0009216, "grad_norm": 0.8684105277061462, "learning_rate": 5.0226357626295106e-06, "loss": 2.7204, "step": 260260 }, { "epoch": 0.0009472, "grad_norm": 0.8137446045875549, "learning_rate": 5.021937337923638e-06, "loss": 2.5675, "step": 260270 }, { "epoch": 0.0009728, "grad_norm": 0.7793560028076172, "learning_rate": 5.021238945499174e-06, "loss": 2.7194, "step": 260280 }, { "epoch": 0.0009984, "grad_norm": 0.90975421667099, "learning_rate": 5.020540585360654e-06, "loss": 2.5144, "step": 260290 }, { "epoch": 0.001024, "grad_norm": 0.8060032725334167, "learning_rate": 5.019842257512604e-06, "loss": 2.5253, "step": 260300 }, { "epoch": 0.0010496, "grad_norm": 0.8098061680793762, "learning_rate": 5.019143961959554e-06, "loss": 2.6635, "step": 260310 }, { "epoch": 0.0010752, "grad_norm": 0.9830915331840515, "learning_rate": 5.01844569870603e-06, "loss": 2.6632, "step": 260320 }, { "epoch": 0.0011008, "grad_norm": 0.8086047172546387, "learning_rate": 5.017747467756563e-06, "loss": 2.8552, "step": 260330 }, { "epoch": 0.0011264, "grad_norm": 0.807221531867981, "learning_rate": 5.017049269115684e-06, "loss": 2.5089, "step": 260340 }, { "epoch": 0.001152, "grad_norm": 0.8360836505889893, "learning_rate": 5.016351102787911e-06, "loss": 2.5982, "step": 260350 }, { "epoch": 0.0011776, "grad_norm": 0.8289716243743896, "learning_rate": 5.015652968777777e-06, "loss": 2.776, "step": 260360 }, { "epoch": 0.0012032, "grad_norm": 0.8422760963439941, "learning_rate": 5.0149548670898085e-06, "loss": 2.7691, "step": 260370 }, { "epoch": 0.0012288, "grad_norm": 0.7996336221694946, "learning_rate": 5.014256797728534e-06, "loss": 2.5486, "step": 260380 }, { "epoch": 0.0012544, "grad_norm": 0.8490316867828369, "learning_rate": 5.013558760698474e-06, "loss": 3.0522, "step": 260390 }, { "epoch": 0.00128, "grad_norm": 0.7762674689292908, "learning_rate": 5.012860756004168e-06, "loss": 2.4497, "step": 260400 }, { "epoch": 0.0013056, "grad_norm": 0.8380627036094666, "learning_rate": 5.012162783650131e-06, "loss": 2.8614, "step": 260410 }, { "epoch": 0.0013312, "grad_norm": 0.9249323010444641, "learning_rate": 5.011464843640893e-06, "loss": 2.901, "step": 260420 }, { "epoch": 0.0013568, "grad_norm": 0.8668707609176636, "learning_rate": 5.01076693598098e-06, "loss": 2.9626, "step": 260430 }, { "epoch": 0.0013824, "grad_norm": 0.774468719959259, "learning_rate": 5.0100690606749155e-06, "loss": 2.8113, "step": 260440 }, { "epoch": 0.001408, "grad_norm": 0.8183870911598206, "learning_rate": 5.009371217727229e-06, "loss": 2.4767, "step": 260450 }, { "epoch": 0.0014336, "grad_norm": 0.8485498428344727, "learning_rate": 5.008673407142444e-06, "loss": 2.8681, "step": 260460 }, { "epoch": 0.0014592, "grad_norm": 0.84125155210495, "learning_rate": 5.007975628925088e-06, "loss": 2.4918, "step": 260470 }, { "epoch": 0.0014848, "grad_norm": 0.8117474317550659, "learning_rate": 5.007277883079679e-06, "loss": 2.4448, "step": 260480 }, { "epoch": 0.0015104, "grad_norm": 0.9241840839385986, "learning_rate": 5.006580169610749e-06, "loss": 2.2035, "step": 260490 }, { "epoch": 0.001536, "grad_norm": 1.1006230115890503, "learning_rate": 5.0058824885228165e-06, "loss": 2.2286, "step": 260500 }, { "epoch": 0.0015616, "grad_norm": 0.8899348378181458, "learning_rate": 5.0051848398204096e-06, "loss": 2.8472, "step": 260510 }, { "epoch": 0.0015872, "grad_norm": 0.892876386642456, "learning_rate": 5.004487223508052e-06, "loss": 2.5395, "step": 260520 }, { "epoch": 0.0016128, "grad_norm": 0.850951075553894, "learning_rate": 5.003789639590266e-06, "loss": 2.4337, "step": 260530 }, { "epoch": 0.0016384, "grad_norm": 0.852016031742096, "learning_rate": 5.003092088071582e-06, "loss": 2.4392, "step": 260540 }, { "epoch": 0.001664, "grad_norm": 0.8203999996185303, "learning_rate": 5.002394568956512e-06, "loss": 2.9065, "step": 260550 }, { "epoch": 0.0016896, "grad_norm": 0.809718668460846, "learning_rate": 5.0016970822495835e-06, "loss": 2.6768, "step": 260560 }, { "epoch": 0.0017152, "grad_norm": 0.8142222762107849, "learning_rate": 5.000999627955323e-06, "loss": 2.7546, "step": 260570 }, { "epoch": 0.0017408, "grad_norm": 1.1106078624725342, "learning_rate": 5.000302206078251e-06, "loss": 2.7953, "step": 260580 }, { "epoch": 0.0017664, "grad_norm": 0.8362298607826233, "learning_rate": 4.999604816622893e-06, "loss": 2.8095, "step": 260590 }, { "epoch": 0.001792, "grad_norm": 0.9311718344688416, "learning_rate": 4.998907459593771e-06, "loss": 2.6164, "step": 260600 }, { "epoch": 0.0018176, "grad_norm": 0.8087295889854431, "learning_rate": 4.9982101349954005e-06, "loss": 2.5506, "step": 260610 }, { "epoch": 0.0018432, "grad_norm": 0.7783082127571106, "learning_rate": 4.997512842832308e-06, "loss": 2.5346, "step": 260620 }, { "epoch": 0.0018688, "grad_norm": 0.7116954922676086, "learning_rate": 4.996815583109016e-06, "loss": 2.5006, "step": 260630 }, { "epoch": 0.0018944, "grad_norm": 0.8055993914604187, "learning_rate": 4.996118355830045e-06, "loss": 2.7312, "step": 260640 }, { "epoch": 0.00192, "grad_norm": 0.8653737902641296, "learning_rate": 4.995421160999917e-06, "loss": 2.7542, "step": 260650 }, { "epoch": 0.0019456, "grad_norm": 0.7899976968765259, "learning_rate": 4.994723998623152e-06, "loss": 2.678, "step": 260660 }, { "epoch": 0.0019712, "grad_norm": 0.8395774364471436, "learning_rate": 4.994026868704276e-06, "loss": 2.8384, "step": 260670 }, { "epoch": 0.0019968, "grad_norm": 1.2239080667495728, "learning_rate": 4.993329771247801e-06, "loss": 2.2531, "step": 260680 }, { "epoch": 0.0020224, "grad_norm": 0.7879367470741272, "learning_rate": 4.992632706258253e-06, "loss": 2.5999, "step": 260690 }, { "epoch": 0.002048, "grad_norm": 0.8197283744812012, "learning_rate": 4.991935673740151e-06, "loss": 2.705, "step": 260700 }, { "epoch": 0.0020736, "grad_norm": 0.7801401615142822, "learning_rate": 4.991238673698013e-06, "loss": 2.5159, "step": 260710 }, { "epoch": 0.0020992, "grad_norm": 0.825166642665863, "learning_rate": 4.990541706136362e-06, "loss": 2.7067, "step": 260720 }, { "epoch": 0.0021248, "grad_norm": 0.7452365159988403, "learning_rate": 4.9898447710597175e-06, "loss": 2.7335, "step": 260730 }, { "epoch": 0.0021504, "grad_norm": 0.7921704053878784, "learning_rate": 4.989147868472598e-06, "loss": 2.7479, "step": 260740 }, { "epoch": 0.002176, "grad_norm": 0.7829078435897827, "learning_rate": 4.9884509983795225e-06, "loss": 2.9128, "step": 260750 }, { "epoch": 0.0022016, "grad_norm": 1.0015628337860107, "learning_rate": 4.987754160785011e-06, "loss": 2.8585, "step": 260760 }, { "epoch": 0.0022272, "grad_norm": 0.950157880783081, "learning_rate": 4.987057355693581e-06, "loss": 2.9713, "step": 260770 }, { "epoch": 0.0022528, "grad_norm": 0.8035308718681335, "learning_rate": 4.986360583109752e-06, "loss": 2.8951, "step": 260780 }, { "epoch": 0.0022784, "grad_norm": 0.9242753386497498, "learning_rate": 4.9856638430380426e-06, "loss": 2.6366, "step": 260790 }, { "epoch": 0.002304, "grad_norm": 0.799828827381134, "learning_rate": 4.9849671354829736e-06, "loss": 2.8808, "step": 260800 }, { "epoch": 0.0023296, "grad_norm": 0.9402490854263306, "learning_rate": 4.984270460449055e-06, "loss": 2.9249, "step": 260810 }, { "epoch": 0.0023552, "grad_norm": 0.9429343342781067, "learning_rate": 4.983573817940811e-06, "loss": 2.5984, "step": 260820 }, { "epoch": 0.0023808, "grad_norm": 0.8896597027778625, "learning_rate": 4.982877207962757e-06, "loss": 2.6403, "step": 260830 }, { "epoch": 0.0024064, "grad_norm": 0.8912819623947144, "learning_rate": 4.982180630519412e-06, "loss": 2.6972, "step": 260840 }, { "epoch": 0.002432, "grad_norm": 0.7922418117523193, "learning_rate": 4.981484085615291e-06, "loss": 2.6081, "step": 260850 }, { "epoch": 0.0024576, "grad_norm": 0.7753576636314392, "learning_rate": 4.980787573254912e-06, "loss": 2.6593, "step": 260860 }, { "epoch": 0.0024832, "grad_norm": 0.777284562587738, "learning_rate": 4.9800910934427955e-06, "loss": 2.806, "step": 260870 }, { "epoch": 0.0025088, "grad_norm": 0.8560603857040405, "learning_rate": 4.979394646183451e-06, "loss": 2.851, "step": 260880 }, { "epoch": 0.0025344, "grad_norm": 0.7522865533828735, "learning_rate": 4.978698231481397e-06, "loss": 2.7661, "step": 260890 }, { "epoch": 0.00256, "grad_norm": 2.004275321960449, "learning_rate": 4.978001849341151e-06, "loss": 2.8757, "step": 260900 }, { "epoch": 0.0025856, "grad_norm": 0.8959113955497742, "learning_rate": 4.977305499767224e-06, "loss": 2.4774, "step": 260910 }, { "epoch": 0.0026112, "grad_norm": 0.7438504695892334, "learning_rate": 4.976609182764145e-06, "loss": 2.7463, "step": 260920 }, { "epoch": 0.0026368, "grad_norm": 0.837760329246521, "learning_rate": 4.9759128983364145e-06, "loss": 2.9616, "step": 260930 }, { "epoch": 0.0026624, "grad_norm": 0.7860109806060791, "learning_rate": 4.975216646488556e-06, "loss": 2.7625, "step": 260940 }, { "epoch": 0.002688, "grad_norm": 0.831943154335022, "learning_rate": 4.9745204272250804e-06, "loss": 2.7512, "step": 260950 }, { "epoch": 0.0027136, "grad_norm": 0.9055236577987671, "learning_rate": 4.973824240550504e-06, "loss": 2.8865, "step": 260960 }, { "epoch": 0.0027392, "grad_norm": 0.9152436256408691, "learning_rate": 4.973128086469342e-06, "loss": 3.0577, "step": 260970 }, { "epoch": 0.0027648, "grad_norm": 0.7488582730293274, "learning_rate": 4.9724319649861085e-06, "loss": 2.5761, "step": 260980 }, { "epoch": 0.0027904, "grad_norm": 0.7745081782341003, "learning_rate": 4.971735876105318e-06, "loss": 2.8226, "step": 260990 }, { "epoch": 0.002816, "grad_norm": 0.9366905689239502, "learning_rate": 4.9710398198314866e-06, "loss": 2.9211, "step": 261000 }, { "epoch": 0.0028416, "grad_norm": 0.8118077516555786, "learning_rate": 4.9703437961691225e-06, "loss": 2.7256, "step": 261010 }, { "epoch": 0.0028672, "grad_norm": 0.8754675984382629, "learning_rate": 4.969647805122742e-06, "loss": 2.799, "step": 261020 }, { "epoch": 0.0028928, "grad_norm": 0.8385509252548218, "learning_rate": 4.968951846696859e-06, "loss": 2.973, "step": 261030 }, { "epoch": 0.0029184, "grad_norm": 1.1757005453109741, "learning_rate": 4.968255920895986e-06, "loss": 2.8597, "step": 261040 }, { "epoch": 0.002944, "grad_norm": 0.7387265563011169, "learning_rate": 4.967560027724636e-06, "loss": 2.7406, "step": 261050 }, { "epoch": 0.0029696, "grad_norm": 0.7391137480735779, "learning_rate": 4.966864167187322e-06, "loss": 2.895, "step": 261060 }, { "epoch": 0.0029952, "grad_norm": 1.1452727317810059, "learning_rate": 4.966168339288559e-06, "loss": 2.8571, "step": 261070 }, { "epoch": 0.0030208, "grad_norm": 0.7677956223487854, "learning_rate": 4.96547254403285e-06, "loss": 2.7408, "step": 261080 }, { "epoch": 0.0030464, "grad_norm": 0.8760393857955933, "learning_rate": 4.964776781424716e-06, "loss": 2.8309, "step": 261090 }, { "epoch": 0.003072, "grad_norm": 0.8058046698570251, "learning_rate": 4.964081051468668e-06, "loss": 3.0339, "step": 261100 }, { "epoch": 0.0030976, "grad_norm": 0.825387716293335, "learning_rate": 4.963385354169216e-06, "loss": 2.777, "step": 261110 }, { "epoch": 0.0031232, "grad_norm": 0.7788547277450562, "learning_rate": 4.962689689530874e-06, "loss": 2.8186, "step": 261120 }, { "epoch": 0.0031488, "grad_norm": 0.8290858268737793, "learning_rate": 4.961994057558147e-06, "loss": 2.7978, "step": 261130 }, { "epoch": 0.0031744, "grad_norm": 0.8507273197174072, "learning_rate": 4.96129845825555e-06, "loss": 2.7087, "step": 261140 }, { "epoch": 0.0032, "grad_norm": 0.8393129110336304, "learning_rate": 4.960602891627593e-06, "loss": 2.7041, "step": 261150 }, { "epoch": 0.0032256, "grad_norm": 0.8293592929840088, "learning_rate": 4.959907357678787e-06, "loss": 2.9588, "step": 261160 }, { "epoch": 0.0032512, "grad_norm": 0.8380524516105652, "learning_rate": 4.959211856413642e-06, "loss": 2.926, "step": 261170 }, { "epoch": 0.0032768, "grad_norm": 0.8059473037719727, "learning_rate": 4.958516387836669e-06, "loss": 2.8694, "step": 261180 }, { "epoch": 0.0033024, "grad_norm": 0.7596539258956909, "learning_rate": 4.9578209519523815e-06, "loss": 2.9702, "step": 261190 }, { "epoch": 0.003328, "grad_norm": 0.795074999332428, "learning_rate": 4.957125548765281e-06, "loss": 2.6781, "step": 261200 }, { "epoch": 0.0033536, "grad_norm": 0.7468567490577698, "learning_rate": 4.9564301782798805e-06, "loss": 2.6542, "step": 261210 }, { "epoch": 0.0033792, "grad_norm": 0.7737159132957458, "learning_rate": 4.955734840500691e-06, "loss": 2.8221, "step": 261220 }, { "epoch": 0.0034048, "grad_norm": 1.0043694972991943, "learning_rate": 4.955039535432219e-06, "loss": 2.8534, "step": 261230 }, { "epoch": 0.0034304, "grad_norm": 0.7261490225791931, "learning_rate": 4.954344263078976e-06, "loss": 2.8481, "step": 261240 }, { "epoch": 0.003456, "grad_norm": 0.9850482940673828, "learning_rate": 4.953649023445469e-06, "loss": 2.8417, "step": 261250 }, { "epoch": 0.0034816, "grad_norm": 0.7876828908920288, "learning_rate": 4.952953816536207e-06, "loss": 2.9448, "step": 261260 }, { "epoch": 0.0035072, "grad_norm": 0.8237312436103821, "learning_rate": 4.9522586423556985e-06, "loss": 2.6862, "step": 261270 }, { "epoch": 0.0035328, "grad_norm": 0.8372952342033386, "learning_rate": 4.951563500908451e-06, "loss": 2.662, "step": 261280 }, { "epoch": 0.0035584, "grad_norm": 0.9371205568313599, "learning_rate": 4.950868392198972e-06, "loss": 2.7737, "step": 261290 }, { "epoch": 0.003584, "grad_norm": 1.0653510093688965, "learning_rate": 4.95017331623177e-06, "loss": 2.7828, "step": 261300 }, { "epoch": 0.0036096, "grad_norm": 0.8245903253555298, "learning_rate": 4.949478273011351e-06, "loss": 3.2258, "step": 261310 }, { "epoch": 0.0036352, "grad_norm": 0.961317241191864, "learning_rate": 4.9487832625422286e-06, "loss": 2.4832, "step": 261320 }, { "epoch": 0.0036608, "grad_norm": 0.7827318906784058, "learning_rate": 4.948088284828899e-06, "loss": 2.8462, "step": 261330 }, { "epoch": 0.0036864, "grad_norm": 0.7594543099403381, "learning_rate": 4.947393339875875e-06, "loss": 2.773, "step": 261340 }, { "epoch": 0.003712, "grad_norm": 0.8803157210350037, "learning_rate": 4.946698427687662e-06, "loss": 2.9825, "step": 261350 }, { "epoch": 0.0037376, "grad_norm": 0.8888145685195923, "learning_rate": 4.946003548268767e-06, "loss": 2.7435, "step": 261360 }, { "epoch": 0.0037632, "grad_norm": 0.8982081413269043, "learning_rate": 4.945308701623696e-06, "loss": 2.5915, "step": 261370 }, { "epoch": 0.0037888, "grad_norm": 0.8041176795959473, "learning_rate": 4.944613887756954e-06, "loss": 2.8528, "step": 261380 }, { "epoch": 0.0038144, "grad_norm": 0.8037723302841187, "learning_rate": 4.943919106673052e-06, "loss": 2.9189, "step": 261390 }, { "epoch": 0.00384, "grad_norm": 0.7975918650627136, "learning_rate": 4.9432243583764875e-06, "loss": 2.8698, "step": 261400 }, { "epoch": 0.0038656, "grad_norm": 0.7701942920684814, "learning_rate": 4.9425296428717674e-06, "loss": 2.7488, "step": 261410 }, { "epoch": 0.0038912, "grad_norm": 0.8021337985992432, "learning_rate": 4.9418349601634e-06, "loss": 2.9656, "step": 261420 }, { "epoch": 0.0039168, "grad_norm": 0.8300039172172546, "learning_rate": 4.941140310255883e-06, "loss": 2.7941, "step": 261430 }, { "epoch": 0.0039424, "grad_norm": 0.9637582898139954, "learning_rate": 4.940445693153731e-06, "loss": 2.7753, "step": 261440 }, { "epoch": 0.003968, "grad_norm": 0.7503750920295715, "learning_rate": 4.939751108861449e-06, "loss": 2.9882, "step": 261450 }, { "epoch": 0.0039936, "grad_norm": 0.819658100605011, "learning_rate": 4.939056557383531e-06, "loss": 2.8863, "step": 261460 }, { "epoch": 0.0040192, "grad_norm": 1.7542659044265747, "learning_rate": 4.938362038724488e-06, "loss": 2.8026, "step": 261470 }, { "epoch": 0.0040448, "grad_norm": 0.8634411692619324, "learning_rate": 4.937667552888821e-06, "loss": 2.919, "step": 261480 }, { "epoch": 0.0040704, "grad_norm": 0.763616681098938, "learning_rate": 4.936973099881034e-06, "loss": 2.9303, "step": 261490 }, { "epoch": 0.004096, "grad_norm": 0.7815954089164734, "learning_rate": 4.936278679705632e-06, "loss": 2.9572, "step": 261500 }, { "epoch": 0.0041216, "grad_norm": 0.8508173823356628, "learning_rate": 4.935584292367116e-06, "loss": 2.7631, "step": 261510 }, { "epoch": 0.0041472, "grad_norm": 0.8535917401313782, "learning_rate": 4.934889937869994e-06, "loss": 2.9438, "step": 261520 }, { "epoch": 0.0041728, "grad_norm": 0.8732794523239136, "learning_rate": 4.93419561621876e-06, "loss": 2.9006, "step": 261530 }, { "epoch": 0.0041984, "grad_norm": 0.8559924960136414, "learning_rate": 4.933501327417922e-06, "loss": 2.5679, "step": 261540 }, { "epoch": 0.004224, "grad_norm": 0.8153512477874756, "learning_rate": 4.9328070714719815e-06, "loss": 2.7109, "step": 261550 }, { "epoch": 0.0042496, "grad_norm": 0.8880622386932373, "learning_rate": 4.9321128483854396e-06, "loss": 3.1645, "step": 261560 }, { "epoch": 0.0042752, "grad_norm": 0.7784890532493591, "learning_rate": 4.9314186581628e-06, "loss": 3.0052, "step": 261570 }, { "epoch": 0.0043008, "grad_norm": 0.821674644947052, "learning_rate": 4.930724500808563e-06, "loss": 2.8538, "step": 261580 }, { "epoch": 0.0043264, "grad_norm": 0.9099636077880859, "learning_rate": 4.930030376327234e-06, "loss": 2.8778, "step": 261590 }, { "epoch": 0.004352, "grad_norm": 0.8895342946052551, "learning_rate": 4.929336284723303e-06, "loss": 2.8562, "step": 261600 }, { "epoch": 0.0043776, "grad_norm": 0.8471095561981201, "learning_rate": 4.928642226001283e-06, "loss": 2.9009, "step": 261610 }, { "epoch": 0.0044032, "grad_norm": 0.8194845914840698, "learning_rate": 4.927948200165669e-06, "loss": 2.8139, "step": 261620 }, { "epoch": 0.0044288, "grad_norm": 0.8051743507385254, "learning_rate": 4.927254207220964e-06, "loss": 2.6657, "step": 261630 }, { "epoch": 0.0044544, "grad_norm": 0.8992008566856384, "learning_rate": 4.926560247171666e-06, "loss": 2.9976, "step": 261640 }, { "epoch": 0.00448, "grad_norm": 0.805984616279602, "learning_rate": 4.9258663200222816e-06, "loss": 2.9523, "step": 261650 }, { "epoch": 0.0045056, "grad_norm": 0.8182432055473328, "learning_rate": 4.925172425777301e-06, "loss": 2.9568, "step": 261660 }, { "epoch": 0.0045312, "grad_norm": 0.7870587110519409, "learning_rate": 4.924478564441229e-06, "loss": 3.0328, "step": 261670 }, { "epoch": 0.0045568, "grad_norm": 0.8108694553375244, "learning_rate": 4.923784736018565e-06, "loss": 2.5256, "step": 261680 }, { "epoch": 0.0045824, "grad_norm": 0.8814663887023926, "learning_rate": 4.923090940513807e-06, "loss": 2.5524, "step": 261690 }, { "epoch": 0.004608, "grad_norm": 1.6747004985809326, "learning_rate": 4.922397177931455e-06, "loss": 3.0552, "step": 261700 }, { "epoch": 0.0046336, "grad_norm": 0.8250153064727783, "learning_rate": 4.921703448276009e-06, "loss": 2.9652, "step": 261710 }, { "epoch": 0.0046592, "grad_norm": 0.8705180287361145, "learning_rate": 4.921009751551969e-06, "loss": 2.8988, "step": 261720 }, { "epoch": 0.0046848, "grad_norm": 0.7889481782913208, "learning_rate": 4.920316087763828e-06, "loss": 2.8778, "step": 261730 }, { "epoch": 0.0047104, "grad_norm": 0.7880713939666748, "learning_rate": 4.919622456916087e-06, "loss": 2.8895, "step": 261740 }, { "epoch": 0.004736, "grad_norm": 0.8356634378433228, "learning_rate": 4.918928859013244e-06, "loss": 2.7837, "step": 261750 }, { "epoch": 0.0047616, "grad_norm": 0.7325760126113892, "learning_rate": 4.918235294059796e-06, "loss": 2.7638, "step": 261760 }, { "epoch": 0.0047872, "grad_norm": 1.0279005765914917, "learning_rate": 4.917541762060243e-06, "loss": 2.6232, "step": 261770 }, { "epoch": 0.0048128, "grad_norm": 0.7464640140533447, "learning_rate": 4.91684826301908e-06, "loss": 2.9337, "step": 261780 }, { "epoch": 0.0048384, "grad_norm": 0.7719635963439941, "learning_rate": 4.916154796940805e-06, "loss": 2.8409, "step": 261790 }, { "epoch": 0.004864, "grad_norm": 0.822034478187561, "learning_rate": 4.915461363829916e-06, "loss": 2.8801, "step": 261800 }, { "epoch": 0.0048896, "grad_norm": 0.8495726585388184, "learning_rate": 4.914767963690908e-06, "loss": 3.1063, "step": 261810 }, { "epoch": 0.0049152, "grad_norm": 0.8343591690063477, "learning_rate": 4.914074596528279e-06, "loss": 2.8322, "step": 261820 }, { "epoch": 0.0049408, "grad_norm": 0.808997631072998, "learning_rate": 4.913381262346524e-06, "loss": 2.6887, "step": 261830 }, { "epoch": 0.0049664, "grad_norm": 0.765123188495636, "learning_rate": 4.9126879611501385e-06, "loss": 2.8551, "step": 261840 }, { "epoch": 0.004992, "grad_norm": 0.8203936815261841, "learning_rate": 4.911994692943625e-06, "loss": 2.6214, "step": 261850 }, { "epoch": 0.0050176, "grad_norm": 0.7603069543838501, "learning_rate": 4.911301457731468e-06, "loss": 2.7752, "step": 261860 }, { "epoch": 0.0050432, "grad_norm": 0.7835657000541687, "learning_rate": 4.910608255518171e-06, "loss": 2.9227, "step": 261870 }, { "epoch": 0.0050688, "grad_norm": 0.9744693636894226, "learning_rate": 4.909915086308226e-06, "loss": 2.7062, "step": 261880 }, { "epoch": 0.0050944, "grad_norm": 0.7894196510314941, "learning_rate": 4.909221950106129e-06, "loss": 2.8325, "step": 261890 }, { "epoch": 0.00512, "grad_norm": 0.8934434652328491, "learning_rate": 4.908528846916374e-06, "loss": 2.6633, "step": 261900 }, { "epoch": 0.0051456, "grad_norm": 0.8265752196311951, "learning_rate": 4.907835776743457e-06, "loss": 2.8702, "step": 261910 }, { "epoch": 0.0051712, "grad_norm": 0.801323413848877, "learning_rate": 4.9071427395918746e-06, "loss": 2.9375, "step": 261920 }, { "epoch": 0.0051968, "grad_norm": 0.9569956064224243, "learning_rate": 4.906449735466115e-06, "loss": 2.85, "step": 261930 }, { "epoch": 0.0052224, "grad_norm": 0.7932955622673035, "learning_rate": 4.905756764370675e-06, "loss": 2.9453, "step": 261940 }, { "epoch": 0.005248, "grad_norm": 1.0791332721710205, "learning_rate": 4.905063826310045e-06, "loss": 2.6706, "step": 261950 }, { "epoch": 0.0052736, "grad_norm": 0.8378022909164429, "learning_rate": 4.904370921288725e-06, "loss": 2.6461, "step": 261960 }, { "epoch": 0.0052992, "grad_norm": 0.7595360279083252, "learning_rate": 4.9036780493112055e-06, "loss": 2.7982, "step": 261970 }, { "epoch": 0.0053248, "grad_norm": 0.9316574931144714, "learning_rate": 4.9029852103819846e-06, "loss": 2.9641, "step": 261980 }, { "epoch": 0.0053504, "grad_norm": 0.7701011300086975, "learning_rate": 4.9022924045055444e-06, "loss": 2.7626, "step": 261990 }, { "epoch": 0.005376, "grad_norm": 0.7356091737747192, "learning_rate": 4.901599631686384e-06, "loss": 2.87, "step": 262000 }, { "epoch": 0.0054016, "grad_norm": 0.689937174320221, "learning_rate": 4.900906891928995e-06, "loss": 2.8336, "step": 262010 }, { "epoch": 0.0054272, "grad_norm": 0.7793341875076294, "learning_rate": 4.900214185237869e-06, "loss": 2.7696, "step": 262020 }, { "epoch": 0.0054528, "grad_norm": 0.7978051900863647, "learning_rate": 4.899521511617499e-06, "loss": 2.8165, "step": 262030 }, { "epoch": 0.0054784, "grad_norm": 0.8403940796852112, "learning_rate": 4.898828871072376e-06, "loss": 2.7842, "step": 262040 }, { "epoch": 0.005504, "grad_norm": 0.8552291989326477, "learning_rate": 4.898136263606997e-06, "loss": 2.7204, "step": 262050 }, { "epoch": 0.0055296, "grad_norm": 0.9306896328926086, "learning_rate": 4.897443689225844e-06, "loss": 2.7038, "step": 262060 }, { "epoch": 0.0055552, "grad_norm": 0.7766335010528564, "learning_rate": 4.896751147933413e-06, "loss": 2.679, "step": 262070 }, { "epoch": 0.0055808, "grad_norm": 0.7767395377159119, "learning_rate": 4.896058639734194e-06, "loss": 2.7771, "step": 262080 }, { "epoch": 0.0056064, "grad_norm": 1.0040217638015747, "learning_rate": 4.895366164632678e-06, "loss": 2.821, "step": 262090 }, { "epoch": 0.005632, "grad_norm": 0.8248155117034912, "learning_rate": 4.894673722633356e-06, "loss": 2.9006, "step": 262100 }, { "epoch": 0.0056576, "grad_norm": 0.8006454110145569, "learning_rate": 4.893981313740719e-06, "loss": 2.8669, "step": 262110 }, { "epoch": 0.0056832, "grad_norm": 0.8179599046707153, "learning_rate": 4.893288937959258e-06, "loss": 2.806, "step": 262120 }, { "epoch": 0.0057088, "grad_norm": 0.8944426774978638, "learning_rate": 4.892596595293455e-06, "loss": 2.6609, "step": 262130 }, { "epoch": 0.0057344, "grad_norm": 0.8007418513298035, "learning_rate": 4.89190428574781e-06, "loss": 2.8873, "step": 262140 }, { "epoch": 0.00576, "grad_norm": 0.8110442757606506, "learning_rate": 4.891212009326807e-06, "loss": 2.7352, "step": 262150 }, { "epoch": 0.0057856, "grad_norm": 1.0254716873168945, "learning_rate": 4.890519766034936e-06, "loss": 3.0021, "step": 262160 }, { "epoch": 0.0058112, "grad_norm": 0.7878443598747253, "learning_rate": 4.889827555876687e-06, "loss": 2.6717, "step": 262170 }, { "epoch": 0.0058368, "grad_norm": 0.7670473456382751, "learning_rate": 4.8891353788565525e-06, "loss": 2.6983, "step": 262180 }, { "epoch": 0.0058624, "grad_norm": 0.7748332619667053, "learning_rate": 4.888443234979012e-06, "loss": 2.7879, "step": 262190 }, { "epoch": 0.005888, "grad_norm": 0.7956174612045288, "learning_rate": 4.887751124248559e-06, "loss": 2.7451, "step": 262200 }, { "epoch": 0.0059136, "grad_norm": 0.803021252155304, "learning_rate": 4.887059046669681e-06, "loss": 2.7049, "step": 262210 }, { "epoch": 0.0059392, "grad_norm": 0.9821768403053284, "learning_rate": 4.886367002246867e-06, "loss": 2.8154, "step": 262220 }, { "epoch": 0.0059648, "grad_norm": 1.376573085784912, "learning_rate": 4.885674990984604e-06, "loss": 2.896, "step": 262230 }, { "epoch": 0.0059904, "grad_norm": 0.7475369572639465, "learning_rate": 4.884983012887378e-06, "loss": 2.9481, "step": 262240 }, { "epoch": 0.006016, "grad_norm": 0.7841987013816833, "learning_rate": 4.884291067959683e-06, "loss": 2.8506, "step": 262250 }, { "epoch": 0.0060416, "grad_norm": 0.7714530229568481, "learning_rate": 4.883599156205996e-06, "loss": 2.9911, "step": 262260 }, { "epoch": 0.0060672, "grad_norm": 0.8030843734741211, "learning_rate": 4.8829072776308095e-06, "loss": 2.8315, "step": 262270 }, { "epoch": 0.0060928, "grad_norm": 0.7610888481140137, "learning_rate": 4.88221543223861e-06, "loss": 2.3379, "step": 262280 }, { "epoch": 0.0061184, "grad_norm": 0.8354301452636719, "learning_rate": 4.881523620033881e-06, "loss": 2.8112, "step": 262290 }, { "epoch": 0.006144, "grad_norm": 0.7694317698478699, "learning_rate": 4.880831841021113e-06, "loss": 2.7028, "step": 262300 }, { "epoch": 0.0061696, "grad_norm": 0.7224324345588684, "learning_rate": 4.880140095204789e-06, "loss": 2.6203, "step": 262310 }, { "epoch": 0.0061952, "grad_norm": 0.8317619562149048, "learning_rate": 4.879448382589397e-06, "loss": 2.8685, "step": 262320 }, { "epoch": 0.0062208, "grad_norm": 0.8833781480789185, "learning_rate": 4.87875670317942e-06, "loss": 2.688, "step": 262330 }, { "epoch": 0.0062464, "grad_norm": 0.7521949410438538, "learning_rate": 4.878065056979346e-06, "loss": 2.6769, "step": 262340 }, { "epoch": 0.006272, "grad_norm": 0.8191972970962524, "learning_rate": 4.877373443993657e-06, "loss": 2.7625, "step": 262350 }, { "epoch": 0.0062976, "grad_norm": 0.8068232536315918, "learning_rate": 4.876681864226841e-06, "loss": 2.7117, "step": 262360 }, { "epoch": 0.0063232, "grad_norm": 1.280279517173767, "learning_rate": 4.875990317683381e-06, "loss": 2.7034, "step": 262370 }, { "epoch": 0.0063488, "grad_norm": 0.8134846687316895, "learning_rate": 4.8752988043677665e-06, "loss": 2.6672, "step": 262380 }, { "epoch": 0.0063744, "grad_norm": 0.7918553948402405, "learning_rate": 4.874607324284474e-06, "loss": 3.1698, "step": 262390 }, { "epoch": 0.0064, "grad_norm": 1.0232590436935425, "learning_rate": 4.873915877437989e-06, "loss": 2.6368, "step": 262400 }, { "epoch": 0.0064256, "grad_norm": 0.7948594689369202, "learning_rate": 4.873224463832798e-06, "loss": 2.8142, "step": 262410 }, { "epoch": 0.0064512, "grad_norm": 0.8810107111930847, "learning_rate": 4.872533083473384e-06, "loss": 3.0279, "step": 262420 }, { "epoch": 0.0064768, "grad_norm": 0.83307284116745, "learning_rate": 4.87184173636423e-06, "loss": 2.8956, "step": 262430 }, { "epoch": 0.0065024, "grad_norm": 0.7871333956718445, "learning_rate": 4.871150422509821e-06, "loss": 2.9047, "step": 262440 }, { "epoch": 0.006528, "grad_norm": 0.8127316236495972, "learning_rate": 4.87045914191464e-06, "loss": 2.7247, "step": 262450 }, { "epoch": 0.0065536, "grad_norm": 0.9480254054069519, "learning_rate": 4.869767894583165e-06, "loss": 2.7163, "step": 262460 }, { "epoch": 0.0065792, "grad_norm": 0.7995740175247192, "learning_rate": 4.869076680519879e-06, "loss": 2.8151, "step": 262470 }, { "epoch": 0.0066048, "grad_norm": 0.7518104910850525, "learning_rate": 4.86838549972927e-06, "loss": 2.7759, "step": 262480 }, { "epoch": 0.0066304, "grad_norm": 0.7435190677642822, "learning_rate": 4.867694352215817e-06, "loss": 2.9923, "step": 262490 }, { "epoch": 0.006656, "grad_norm": 0.8045051693916321, "learning_rate": 4.867003237984006e-06, "loss": 2.7725, "step": 262500 }, { "epoch": 0.0066816, "grad_norm": 1.099801778793335, "learning_rate": 4.866312157038312e-06, "loss": 2.6904, "step": 262510 }, { "epoch": 0.0067072, "grad_norm": 0.8723674416542053, "learning_rate": 4.865621109383219e-06, "loss": 2.901, "step": 262520 }, { "epoch": 0.0067328, "grad_norm": 0.7671525478363037, "learning_rate": 4.864930095023207e-06, "loss": 2.9045, "step": 262530 }, { "epoch": 0.0067584, "grad_norm": 0.7947373986244202, "learning_rate": 4.8642391139627595e-06, "loss": 2.7786, "step": 262540 }, { "epoch": 0.006784, "grad_norm": 0.838346004486084, "learning_rate": 4.863548166206356e-06, "loss": 2.7565, "step": 262550 }, { "epoch": 0.0068096, "grad_norm": 0.7606129050254822, "learning_rate": 4.8628572517584796e-06, "loss": 2.7927, "step": 262560 }, { "epoch": 0.0068352, "grad_norm": 0.8776338696479797, "learning_rate": 4.862166370623606e-06, "loss": 2.8269, "step": 262570 }, { "epoch": 0.0068608, "grad_norm": 1.7679989337921143, "learning_rate": 4.861475522806224e-06, "loss": 2.8973, "step": 262580 }, { "epoch": 0.0068864, "grad_norm": 0.8109639286994934, "learning_rate": 4.860784708310803e-06, "loss": 2.7245, "step": 262590 }, { "epoch": 0.006912, "grad_norm": 0.7951434850692749, "learning_rate": 4.860093927141827e-06, "loss": 2.8341, "step": 262600 }, { "epoch": 0.0069376, "grad_norm": 0.8487917184829712, "learning_rate": 4.859403179303776e-06, "loss": 2.7249, "step": 262610 }, { "epoch": 0.0069632, "grad_norm": 0.7538968920707703, "learning_rate": 4.858712464801129e-06, "loss": 2.9568, "step": 262620 }, { "epoch": 0.0069888, "grad_norm": 1.2034446001052856, "learning_rate": 4.8580217836383645e-06, "loss": 2.7118, "step": 262630 }, { "epoch": 0.0070144, "grad_norm": 0.7848944067955017, "learning_rate": 4.8573311358199635e-06, "loss": 2.707, "step": 262640 }, { "epoch": 0.00704, "grad_norm": 0.7832714319229126, "learning_rate": 4.856640521350403e-06, "loss": 2.7753, "step": 262650 }, { "epoch": 0.0070656, "grad_norm": 1.173567533493042, "learning_rate": 4.855949940234161e-06, "loss": 2.9627, "step": 262660 }, { "epoch": 0.0070912, "grad_norm": 1.4904505014419556, "learning_rate": 4.855259392475718e-06, "loss": 2.9566, "step": 262670 }, { "epoch": 0.0071168, "grad_norm": 0.9582147598266602, "learning_rate": 4.85456887807955e-06, "loss": 2.7882, "step": 262680 }, { "epoch": 0.0071424, "grad_norm": 0.8110358715057373, "learning_rate": 4.853878397050135e-06, "loss": 2.7684, "step": 262690 }, { "epoch": 0.007168, "grad_norm": 0.8102409243583679, "learning_rate": 4.853187949391954e-06, "loss": 2.6754, "step": 262700 }, { "epoch": 0.0071936, "grad_norm": 0.8125007748603821, "learning_rate": 4.85249753510948e-06, "loss": 2.8529, "step": 262710 }, { "epoch": 0.0072192, "grad_norm": 1.1732090711593628, "learning_rate": 4.851807154207189e-06, "loss": 3.0081, "step": 262720 }, { "epoch": 0.0072448, "grad_norm": 0.7947466373443604, "learning_rate": 4.851116806689561e-06, "loss": 2.7442, "step": 262730 }, { "epoch": 0.0072704, "grad_norm": 0.8825553059577942, "learning_rate": 4.850426492561072e-06, "loss": 2.8675, "step": 262740 }, { "epoch": 0.007296, "grad_norm": 0.8123030662536621, "learning_rate": 4.849736211826198e-06, "loss": 2.7175, "step": 262750 }, { "epoch": 0.0073216, "grad_norm": 0.7891181111335754, "learning_rate": 4.849045964489417e-06, "loss": 3.0001, "step": 262760 }, { "epoch": 0.0073472, "grad_norm": 0.9022513628005981, "learning_rate": 4.848355750555204e-06, "loss": 2.8656, "step": 262770 }, { "epoch": 0.0073728, "grad_norm": 0.7887265682220459, "learning_rate": 4.847665570028038e-06, "loss": 3.0409, "step": 262780 }, { "epoch": 0.0073984, "grad_norm": 0.9183505177497864, "learning_rate": 4.846975422912388e-06, "loss": 2.7875, "step": 262790 }, { "epoch": 0.007424, "grad_norm": 0.8144945502281189, "learning_rate": 4.846285309212733e-06, "loss": 2.8419, "step": 262800 }, { "epoch": 0.0074496, "grad_norm": 0.9519970417022705, "learning_rate": 4.845595228933547e-06, "loss": 2.686, "step": 262810 }, { "epoch": 0.0074752, "grad_norm": 0.7447562217712402, "learning_rate": 4.844905182079304e-06, "loss": 2.7346, "step": 262820 }, { "epoch": 0.0075008, "grad_norm": 0.8145484924316406, "learning_rate": 4.844215168654488e-06, "loss": 2.9603, "step": 262830 }, { "epoch": 0.0075264, "grad_norm": 0.9024884104728699, "learning_rate": 4.843525188663561e-06, "loss": 2.9402, "step": 262840 }, { "epoch": 0.007552, "grad_norm": 0.803023099899292, "learning_rate": 4.842835242111005e-06, "loss": 2.8265, "step": 262850 }, { "epoch": 0.0075776, "grad_norm": 0.7574452757835388, "learning_rate": 4.84214532900129e-06, "loss": 2.8661, "step": 262860 }, { "epoch": 0.0076032, "grad_norm": 0.8418099880218506, "learning_rate": 4.8414554493388934e-06, "loss": 2.8492, "step": 262870 }, { "epoch": 0.0076288, "grad_norm": 0.909307062625885, "learning_rate": 4.840765603128286e-06, "loss": 2.7998, "step": 262880 }, { "epoch": 0.0076544, "grad_norm": 0.817790687084198, "learning_rate": 4.840075790373942e-06, "loss": 2.9651, "step": 262890 }, { "epoch": 0.00768, "grad_norm": 0.7624891996383667, "learning_rate": 4.83938601108034e-06, "loss": 2.8105, "step": 262900 }, { "epoch": 0.0077056, "grad_norm": 0.7895187139511108, "learning_rate": 4.838696265251944e-06, "loss": 2.8174, "step": 262910 }, { "epoch": 0.0077312, "grad_norm": 0.7584384083747864, "learning_rate": 4.838006552893232e-06, "loss": 3.0426, "step": 262920 }, { "epoch": 0.0077568, "grad_norm": 0.8021336197853088, "learning_rate": 4.837316874008674e-06, "loss": 2.8627, "step": 262930 }, { "epoch": 0.0077824, "grad_norm": 0.7201144099235535, "learning_rate": 4.836627228602745e-06, "loss": 2.9739, "step": 262940 }, { "epoch": 0.007808, "grad_norm": 0.7586404085159302, "learning_rate": 4.835937616679916e-06, "loss": 2.9061, "step": 262950 }, { "epoch": 0.0078336, "grad_norm": 0.8348206281661987, "learning_rate": 4.8352480382446586e-06, "loss": 2.9017, "step": 262960 }, { "epoch": 0.0078592, "grad_norm": 0.8048037886619568, "learning_rate": 4.834558493301446e-06, "loss": 3.0041, "step": 262970 }, { "epoch": 0.0078848, "grad_norm": 0.838154137134552, "learning_rate": 4.833868981854752e-06, "loss": 2.9352, "step": 262980 }, { "epoch": 0.0079104, "grad_norm": 0.7915177345275879, "learning_rate": 4.833179503909041e-06, "loss": 2.8986, "step": 262990 }, { "epoch": 0.007936, "grad_norm": 0.7825537919998169, "learning_rate": 4.832490059468784e-06, "loss": 2.8698, "step": 263000 }, { "epoch": 0.0079616, "grad_norm": 0.8439725041389465, "learning_rate": 4.8318006485384596e-06, "loss": 2.8129, "step": 263010 }, { "epoch": 0.0079872, "grad_norm": 0.8022933006286621, "learning_rate": 4.831111271122534e-06, "loss": 2.8673, "step": 263020 }, { "epoch": 0.0080128, "grad_norm": 0.8182438015937805, "learning_rate": 4.830421927225482e-06, "loss": 2.9348, "step": 263030 }, { "epoch": 0.0080384, "grad_norm": 0.779262363910675, "learning_rate": 4.829732616851766e-06, "loss": 2.7835, "step": 263040 }, { "epoch": 0.008064, "grad_norm": 1.019164800643921, "learning_rate": 4.829043340005859e-06, "loss": 2.8244, "step": 263050 }, { "epoch": 0.0080896, "grad_norm": 0.9187606573104858, "learning_rate": 4.828354096692232e-06, "loss": 2.8453, "step": 263060 }, { "epoch": 0.0081152, "grad_norm": 0.8434553146362305, "learning_rate": 4.827664886915357e-06, "loss": 2.9195, "step": 263070 }, { "epoch": 0.0081408, "grad_norm": 1.1077003479003906, "learning_rate": 4.826975710679698e-06, "loss": 2.8596, "step": 263080 }, { "epoch": 0.0081664, "grad_norm": 0.7401200532913208, "learning_rate": 4.826286567989727e-06, "loss": 2.9641, "step": 263090 }, { "epoch": 0.008192, "grad_norm": 0.7731555104255676, "learning_rate": 4.825597458849915e-06, "loss": 2.8007, "step": 263100 }, { "epoch": 0.0082176, "grad_norm": 0.8265126347541809, "learning_rate": 4.824908383264727e-06, "loss": 2.773, "step": 263110 }, { "epoch": 0.0082432, "grad_norm": 1.0231882333755493, "learning_rate": 4.824219341238632e-06, "loss": 2.8555, "step": 263120 }, { "epoch": 0.0082688, "grad_norm": 0.8092761039733887, "learning_rate": 4.823530332776098e-06, "loss": 2.9344, "step": 263130 }, { "epoch": 0.0082944, "grad_norm": 0.8488980531692505, "learning_rate": 4.822841357881596e-06, "loss": 2.6957, "step": 263140 }, { "epoch": 0.00832, "grad_norm": 0.8799371719360352, "learning_rate": 4.82215241655959e-06, "loss": 2.8053, "step": 263150 }, { "epoch": 0.0083456, "grad_norm": 0.9175581336021423, "learning_rate": 4.82146350881455e-06, "loss": 3.0501, "step": 263160 }, { "epoch": 0.0083712, "grad_norm": 0.8796755075454712, "learning_rate": 4.820774634650943e-06, "loss": 2.9346, "step": 263170 }, { "epoch": 0.0083968, "grad_norm": 0.8095020055770874, "learning_rate": 4.820085794073236e-06, "loss": 2.7879, "step": 263180 }, { "epoch": 0.0084224, "grad_norm": 0.7754899859428406, "learning_rate": 4.819396987085896e-06, "loss": 2.9469, "step": 263190 }, { "epoch": 0.008448, "grad_norm": 0.7357217073440552, "learning_rate": 4.818708213693391e-06, "loss": 3.1411, "step": 263200 }, { "epoch": 0.0084736, "grad_norm": 0.8225201368331909, "learning_rate": 4.818019473900184e-06, "loss": 2.6804, "step": 263210 }, { "epoch": 0.0084992, "grad_norm": 0.981291651725769, "learning_rate": 4.817330767710744e-06, "loss": 2.8969, "step": 263220 }, { "epoch": 0.0085248, "grad_norm": 0.8042603731155396, "learning_rate": 4.8166420951295415e-06, "loss": 2.7612, "step": 263230 }, { "epoch": 0.0085504, "grad_norm": 0.8285202383995056, "learning_rate": 4.815953456161032e-06, "loss": 2.8538, "step": 263240 }, { "epoch": 0.008576, "grad_norm": 0.8007301688194275, "learning_rate": 4.815264850809688e-06, "loss": 2.9211, "step": 263250 }, { "epoch": 0.0086016, "grad_norm": 0.7773743271827698, "learning_rate": 4.814576279079973e-06, "loss": 2.8502, "step": 263260 }, { "epoch": 0.0086272, "grad_norm": 0.9005755186080933, "learning_rate": 4.813887740976353e-06, "loss": 3.0142, "step": 263270 }, { "epoch": 0.0086528, "grad_norm": 0.7847978472709656, "learning_rate": 4.813199236503292e-06, "loss": 2.8349, "step": 263280 }, { "epoch": 0.0086784, "grad_norm": 0.8983844518661499, "learning_rate": 4.8125107656652555e-06, "loss": 2.9625, "step": 263290 }, { "epoch": 0.008704, "grad_norm": 0.9452351331710815, "learning_rate": 4.811822328466713e-06, "loss": 2.946, "step": 263300 }, { "epoch": 0.0087296, "grad_norm": 0.9030247926712036, "learning_rate": 4.811133924912119e-06, "loss": 2.7607, "step": 263310 }, { "epoch": 0.0087552, "grad_norm": 0.7977635264396667, "learning_rate": 4.810445555005943e-06, "loss": 2.7133, "step": 263320 }, { "epoch": 0.0087808, "grad_norm": 0.8297029137611389, "learning_rate": 4.809757218752649e-06, "loss": 2.7836, "step": 263330 }, { "epoch": 0.0088064, "grad_norm": 0.7198085188865662, "learning_rate": 4.809068916156695e-06, "loss": 2.9678, "step": 263340 }, { "epoch": 0.008832, "grad_norm": 0.7967691421508789, "learning_rate": 4.808380647222554e-06, "loss": 2.7261, "step": 263350 }, { "epoch": 0.0088576, "grad_norm": 0.742732048034668, "learning_rate": 4.807692411954688e-06, "loss": 2.877, "step": 263360 }, { "epoch": 0.0088832, "grad_norm": 0.8208882808685303, "learning_rate": 4.807004210357552e-06, "loss": 2.8948, "step": 263370 }, { "epoch": 0.0089088, "grad_norm": 0.8408832550048828, "learning_rate": 4.806316042435615e-06, "loss": 2.6393, "step": 263380 }, { "epoch": 0.0089344, "grad_norm": 0.7671428322792053, "learning_rate": 4.805627908193338e-06, "loss": 2.5711, "step": 263390 }, { "epoch": 0.00896, "grad_norm": 0.8237773776054382, "learning_rate": 4.804939807635183e-06, "loss": 2.8757, "step": 263400 }, { "epoch": 0.0089856, "grad_norm": 0.7921761870384216, "learning_rate": 4.804251740765613e-06, "loss": 2.8476, "step": 263410 }, { "epoch": 0.0090112, "grad_norm": 0.8430052399635315, "learning_rate": 4.803563707589089e-06, "loss": 2.5712, "step": 263420 }, { "epoch": 0.0090368, "grad_norm": 0.9585695266723633, "learning_rate": 4.8028757081100785e-06, "loss": 2.7538, "step": 263430 }, { "epoch": 0.0090624, "grad_norm": 0.8229129910469055, "learning_rate": 4.802187742333033e-06, "loss": 2.8789, "step": 263440 }, { "epoch": 0.009088, "grad_norm": 0.907913327217102, "learning_rate": 4.801499810262419e-06, "loss": 2.7787, "step": 263450 }, { "epoch": 0.0091136, "grad_norm": 0.9306819438934326, "learning_rate": 4.800811911902698e-06, "loss": 2.9748, "step": 263460 }, { "epoch": 0.0091392, "grad_norm": 0.888481616973877, "learning_rate": 4.800124047258328e-06, "loss": 2.8995, "step": 263470 }, { "epoch": 0.0091648, "grad_norm": 0.8041516542434692, "learning_rate": 4.799436216333773e-06, "loss": 2.977, "step": 263480 }, { "epoch": 0.0091904, "grad_norm": 0.7373694181442261, "learning_rate": 4.798748419133491e-06, "loss": 2.769, "step": 263490 }, { "epoch": 0.009216, "grad_norm": 0.7912337183952332, "learning_rate": 4.798060655661947e-06, "loss": 2.9299, "step": 263500 }, { "epoch": 0.0092416, "grad_norm": 0.93208247423172, "learning_rate": 4.7973729259235945e-06, "loss": 2.9124, "step": 263510 }, { "epoch": 0.0092672, "grad_norm": 0.7509072422981262, "learning_rate": 4.7966852299228905e-06, "loss": 3.0241, "step": 263520 }, { "epoch": 0.0092928, "grad_norm": 1.0343241691589355, "learning_rate": 4.7959975676643056e-06, "loss": 2.6729, "step": 263530 }, { "epoch": 0.0093184, "grad_norm": 0.8393169641494751, "learning_rate": 4.795309939152292e-06, "loss": 2.8969, "step": 263540 }, { "epoch": 0.009344, "grad_norm": 0.8377876281738281, "learning_rate": 4.794622344391311e-06, "loss": 2.9017, "step": 263550 }, { "epoch": 0.0093696, "grad_norm": 1.00262451171875, "learning_rate": 4.793934783385824e-06, "loss": 2.8206, "step": 263560 }, { "epoch": 0.0093952, "grad_norm": 0.7948088645935059, "learning_rate": 4.793247256140282e-06, "loss": 2.7926, "step": 263570 }, { "epoch": 0.0094208, "grad_norm": 0.7658308148384094, "learning_rate": 4.792559762659149e-06, "loss": 2.9957, "step": 263580 }, { "epoch": 0.0094464, "grad_norm": 0.8893796801567078, "learning_rate": 4.791872302946881e-06, "loss": 2.8921, "step": 263590 }, { "epoch": 0.009472, "grad_norm": 0.7842429876327515, "learning_rate": 4.791184877007938e-06, "loss": 3.0157, "step": 263600 }, { "epoch": 0.0094976, "grad_norm": 0.7455933690071106, "learning_rate": 4.790497484846775e-06, "loss": 2.9559, "step": 263610 }, { "epoch": 0.0095232, "grad_norm": 0.7749631404876709, "learning_rate": 4.7898101264678535e-06, "loss": 2.7726, "step": 263620 }, { "epoch": 0.0095488, "grad_norm": 0.7838962078094482, "learning_rate": 4.789122801875631e-06, "loss": 2.8348, "step": 263630 }, { "epoch": 0.0095744, "grad_norm": 0.8178287744522095, "learning_rate": 4.788435511074559e-06, "loss": 2.9863, "step": 263640 }, { "epoch": 0.0096, "grad_norm": 0.8260414004325867, "learning_rate": 4.7877482540690965e-06, "loss": 2.8704, "step": 263650 }, { "epoch": 0.0096256, "grad_norm": 0.7624117732048035, "learning_rate": 4.787061030863702e-06, "loss": 2.932, "step": 263660 }, { "epoch": 0.0096512, "grad_norm": 0.8155379295349121, "learning_rate": 4.786373841462831e-06, "loss": 2.9268, "step": 263670 }, { "epoch": 0.0096768, "grad_norm": 1.0099948644638062, "learning_rate": 4.785686685870941e-06, "loss": 2.9664, "step": 263680 }, { "epoch": 0.0097024, "grad_norm": 0.8859341740608215, "learning_rate": 4.784999564092486e-06, "loss": 2.8074, "step": 263690 }, { "epoch": 0.009728, "grad_norm": 0.943435549736023, "learning_rate": 4.784312476131924e-06, "loss": 2.7423, "step": 263700 }, { "epoch": 0.0097536, "grad_norm": 0.9518463015556335, "learning_rate": 4.783625421993709e-06, "loss": 2.9332, "step": 263710 }, { "epoch": 0.0097792, "grad_norm": 0.773953378200531, "learning_rate": 4.782938401682296e-06, "loss": 2.8139, "step": 263720 }, { "epoch": 0.0098048, "grad_norm": 0.7907480597496033, "learning_rate": 4.7822514152021414e-06, "loss": 2.6135, "step": 263730 }, { "epoch": 0.0098304, "grad_norm": 0.7624431848526001, "learning_rate": 4.781564462557699e-06, "loss": 2.7634, "step": 263740 }, { "epoch": 0.009856, "grad_norm": 0.7409116625785828, "learning_rate": 4.780877543753426e-06, "loss": 2.8878, "step": 263750 }, { "epoch": 0.0098816, "grad_norm": 0.966031014919281, "learning_rate": 4.780190658793778e-06, "loss": 2.7967, "step": 263760 }, { "epoch": 0.0099072, "grad_norm": 0.8116340637207031, "learning_rate": 4.779503807683202e-06, "loss": 2.739, "step": 263770 }, { "epoch": 0.0099328, "grad_norm": 0.9606654047966003, "learning_rate": 4.778816990426157e-06, "loss": 2.904, "step": 263780 }, { "epoch": 0.0099584, "grad_norm": 0.7789889574050903, "learning_rate": 4.778130207027095e-06, "loss": 2.7836, "step": 263790 }, { "epoch": 0.009984, "grad_norm": 0.9218695163726807, "learning_rate": 4.7774434574904725e-06, "loss": 2.8164, "step": 263800 }, { "epoch": 0.0100096, "grad_norm": 0.8818469643592834, "learning_rate": 4.7767567418207405e-06, "loss": 2.9012, "step": 263810 }, { "epoch": 0.0100352, "grad_norm": 0.8378807902336121, "learning_rate": 4.776070060022352e-06, "loss": 2.9531, "step": 263820 }, { "epoch": 0.0100608, "grad_norm": 1.0306261777877808, "learning_rate": 4.775383412099766e-06, "loss": 2.9005, "step": 263830 }, { "epoch": 0.0100864, "grad_norm": 0.8331193923950195, "learning_rate": 4.774696798057425e-06, "loss": 2.8255, "step": 263840 }, { "epoch": 0.010112, "grad_norm": 0.8714050054550171, "learning_rate": 4.7740102178997875e-06, "loss": 2.7045, "step": 263850 }, { "epoch": 0.0101376, "grad_norm": 0.7470344305038452, "learning_rate": 4.773323671631304e-06, "loss": 2.8955, "step": 263860 }, { "epoch": 0.0101632, "grad_norm": 0.8501931428909302, "learning_rate": 4.7726371592564245e-06, "loss": 2.7687, "step": 263870 }, { "epoch": 0.0101888, "grad_norm": 0.8303055763244629, "learning_rate": 4.771950680779607e-06, "loss": 2.943, "step": 263880 }, { "epoch": 0.0102144, "grad_norm": 0.8316789865493774, "learning_rate": 4.7712642362053045e-06, "loss": 2.8846, "step": 263890 }, { "epoch": 0.01024, "grad_norm": 0.9230756759643555, "learning_rate": 4.770577825537958e-06, "loss": 2.9753, "step": 263900 }, { "epoch": 0.0102656, "grad_norm": 0.9123333096504211, "learning_rate": 4.769891448782026e-06, "loss": 2.608, "step": 263910 }, { "epoch": 0.0102912, "grad_norm": 0.8731915354728699, "learning_rate": 4.769205105941958e-06, "loss": 2.7523, "step": 263920 }, { "epoch": 0.0103168, "grad_norm": 0.844977855682373, "learning_rate": 4.768518797022205e-06, "loss": 2.9469, "step": 263930 }, { "epoch": 0.0103424, "grad_norm": 0.7884578108787537, "learning_rate": 4.7678325220272155e-06, "loss": 2.9106, "step": 263940 }, { "epoch": 0.010368, "grad_norm": 0.8198543787002563, "learning_rate": 4.767146280961443e-06, "loss": 2.9548, "step": 263950 }, { "epoch": 0.0103936, "grad_norm": 0.8692910075187683, "learning_rate": 4.76646007382934e-06, "loss": 2.7865, "step": 263960 }, { "epoch": 0.0104192, "grad_norm": 0.8466352820396423, "learning_rate": 4.765773900635348e-06, "loss": 2.9594, "step": 263970 }, { "epoch": 0.0104448, "grad_norm": 0.8289999961853027, "learning_rate": 4.765087761383922e-06, "loss": 2.9441, "step": 263980 }, { "epoch": 0.0104704, "grad_norm": 1.3083840608596802, "learning_rate": 4.764401656079511e-06, "loss": 2.9364, "step": 263990 }, { "epoch": 0.010496, "grad_norm": 0.769531786441803, "learning_rate": 4.763715584726562e-06, "loss": 2.907, "step": 264000 }, { "epoch": 0.0105216, "grad_norm": 0.7174090147018433, "learning_rate": 4.763029547329526e-06, "loss": 2.8115, "step": 264010 }, { "epoch": 0.0105472, "grad_norm": 0.8071773648262024, "learning_rate": 4.762343543892853e-06, "loss": 2.8498, "step": 264020 }, { "epoch": 0.0105728, "grad_norm": 0.7723903059959412, "learning_rate": 4.761657574420993e-06, "loss": 2.8729, "step": 264030 }, { "epoch": 0.0105984, "grad_norm": 0.8656411170959473, "learning_rate": 4.760971638918385e-06, "loss": 2.9745, "step": 264040 }, { "epoch": 0.010624, "grad_norm": 0.8411847352981567, "learning_rate": 4.760285737389486e-06, "loss": 2.8544, "step": 264050 }, { "epoch": 0.0106496, "grad_norm": 0.8179494142532349, "learning_rate": 4.759599869838742e-06, "loss": 2.8873, "step": 264060 }, { "epoch": 0.0106752, "grad_norm": 0.8320645093917847, "learning_rate": 4.7589140362706e-06, "loss": 2.8213, "step": 264070 }, { "epoch": 0.0107008, "grad_norm": 0.954226553440094, "learning_rate": 4.758228236689507e-06, "loss": 2.872, "step": 264080 }, { "epoch": 0.0107264, "grad_norm": 0.8036489486694336, "learning_rate": 4.757542471099915e-06, "loss": 2.857, "step": 264090 }, { "epoch": 0.010752, "grad_norm": 0.8249059915542603, "learning_rate": 4.756856739506264e-06, "loss": 2.8806, "step": 264100 }, { "epoch": 0.0107776, "grad_norm": 0.8509751558303833, "learning_rate": 4.756171041913002e-06, "loss": 2.8936, "step": 264110 }, { "epoch": 0.0108032, "grad_norm": 0.9219454526901245, "learning_rate": 4.755485378324578e-06, "loss": 2.7885, "step": 264120 }, { "epoch": 0.0108288, "grad_norm": 0.770317554473877, "learning_rate": 4.7547997487454366e-06, "loss": 2.9285, "step": 264130 }, { "epoch": 0.0108544, "grad_norm": 0.7776609659194946, "learning_rate": 4.754114153180026e-06, "loss": 2.6374, "step": 264140 }, { "epoch": 0.01088, "grad_norm": 0.8939977884292603, "learning_rate": 4.7534285916327905e-06, "loss": 2.7374, "step": 264150 }, { "epoch": 0.0109056, "grad_norm": 0.7828552722930908, "learning_rate": 4.75274306410818e-06, "loss": 2.7823, "step": 264160 }, { "epoch": 0.0109312, "grad_norm": 0.8868138790130615, "learning_rate": 4.7520575706106335e-06, "loss": 2.8074, "step": 264170 }, { "epoch": 0.0109568, "grad_norm": 0.7293012738227844, "learning_rate": 4.751372111144596e-06, "loss": 2.7675, "step": 264180 }, { "epoch": 0.0109824, "grad_norm": 0.7764026522636414, "learning_rate": 4.750686685714518e-06, "loss": 2.9275, "step": 264190 }, { "epoch": 0.011008, "grad_norm": 0.877758264541626, "learning_rate": 4.750001294324841e-06, "loss": 2.9128, "step": 264200 }, { "epoch": 0.0110336, "grad_norm": 0.7813438177108765, "learning_rate": 4.74931593698001e-06, "loss": 2.9805, "step": 264210 }, { "epoch": 0.0110592, "grad_norm": 0.7749430537223816, "learning_rate": 4.748630613684471e-06, "loss": 2.8228, "step": 264220 }, { "epoch": 0.0110848, "grad_norm": 0.9128813743591309, "learning_rate": 4.747945324442666e-06, "loss": 3.0768, "step": 264230 }, { "epoch": 0.0111104, "grad_norm": 0.8405775427818298, "learning_rate": 4.747260069259039e-06, "loss": 2.917, "step": 264240 }, { "epoch": 0.011136, "grad_norm": 0.8484730124473572, "learning_rate": 4.746574848138035e-06, "loss": 2.7404, "step": 264250 }, { "epoch": 0.0111616, "grad_norm": 0.8249212503433228, "learning_rate": 4.745889661084096e-06, "loss": 2.6723, "step": 264260 }, { "epoch": 0.0111872, "grad_norm": 0.8281247019767761, "learning_rate": 4.745204508101668e-06, "loss": 2.7478, "step": 264270 }, { "epoch": 0.0112128, "grad_norm": 0.8533861041069031, "learning_rate": 4.7445193891951945e-06, "loss": 2.8204, "step": 264280 }, { "epoch": 0.0112384, "grad_norm": 0.7646121382713318, "learning_rate": 4.743834304369113e-06, "loss": 2.774, "step": 264290 }, { "epoch": 0.011264, "grad_norm": 1.493703007698059, "learning_rate": 4.743149253627869e-06, "loss": 2.9341, "step": 264300 }, { "epoch": 0.0112896, "grad_norm": 0.9403733015060425, "learning_rate": 4.742464236975906e-06, "loss": 2.8899, "step": 264310 }, { "epoch": 0.0113152, "grad_norm": 0.9228437542915344, "learning_rate": 4.741779254417663e-06, "loss": 3.0584, "step": 264320 }, { "epoch": 0.0113408, "grad_norm": 0.8753408193588257, "learning_rate": 4.741094305957585e-06, "loss": 2.8848, "step": 264330 }, { "epoch": 0.0113664, "grad_norm": 0.754368245601654, "learning_rate": 4.740409391600114e-06, "loss": 2.9224, "step": 264340 }, { "epoch": 0.011392, "grad_norm": 0.9017702341079712, "learning_rate": 4.7397245113496884e-06, "loss": 2.9903, "step": 264350 }, { "epoch": 0.0114176, "grad_norm": 0.7875017523765564, "learning_rate": 4.739039665210756e-06, "loss": 2.8706, "step": 264360 }, { "epoch": 0.0114432, "grad_norm": 0.9407398104667664, "learning_rate": 4.738354853187749e-06, "loss": 2.9325, "step": 264370 }, { "epoch": 0.0114688, "grad_norm": 0.7535969614982605, "learning_rate": 4.737670075285113e-06, "loss": 2.7315, "step": 264380 }, { "epoch": 0.0114944, "grad_norm": 0.7857102751731873, "learning_rate": 4.736985331507284e-06, "loss": 2.762, "step": 264390 }, { "epoch": 0.01152, "grad_norm": 0.7759318947792053, "learning_rate": 4.7363006218587105e-06, "loss": 2.8401, "step": 264400 }, { "epoch": 0.0115456, "grad_norm": 0.7748493552207947, "learning_rate": 4.735615946343831e-06, "loss": 2.9735, "step": 264410 }, { "epoch": 0.0115712, "grad_norm": 0.8496517539024353, "learning_rate": 4.734931304967081e-06, "loss": 2.82, "step": 264420 }, { "epoch": 0.0115968, "grad_norm": 1.5857113599777222, "learning_rate": 4.7342466977329e-06, "loss": 2.9942, "step": 264430 }, { "epoch": 0.0116224, "grad_norm": 0.7796471118927002, "learning_rate": 4.733562124645732e-06, "loss": 2.7469, "step": 264440 }, { "epoch": 0.011648, "grad_norm": 0.8036671876907349, "learning_rate": 4.732877585710013e-06, "loss": 2.8785, "step": 264450 }, { "epoch": 0.0116736, "grad_norm": 0.7457982897758484, "learning_rate": 4.732193080930183e-06, "loss": 2.8406, "step": 264460 }, { "epoch": 0.0116992, "grad_norm": 0.8150416016578674, "learning_rate": 4.73150861031068e-06, "loss": 2.806, "step": 264470 }, { "epoch": 0.0117248, "grad_norm": 0.7840436697006226, "learning_rate": 4.730824173855949e-06, "loss": 2.7268, "step": 264480 }, { "epoch": 0.0117504, "grad_norm": 0.8711058497428894, "learning_rate": 4.730139771570418e-06, "loss": 2.8819, "step": 264490 }, { "epoch": 0.011776, "grad_norm": 0.8854064345359802, "learning_rate": 4.72945540345853e-06, "loss": 2.9194, "step": 264500 }, { "epoch": 0.0118016, "grad_norm": 0.8068715333938599, "learning_rate": 4.728771069524722e-06, "loss": 3.0272, "step": 264510 }, { "epoch": 0.0118272, "grad_norm": 0.7758327722549438, "learning_rate": 4.7280867697734345e-06, "loss": 2.8788, "step": 264520 }, { "epoch": 0.0118528, "grad_norm": 0.8565075397491455, "learning_rate": 4.727402504209103e-06, "loss": 2.924, "step": 264530 }, { "epoch": 0.0118784, "grad_norm": 0.8736703991889954, "learning_rate": 4.726718272836164e-06, "loss": 2.9706, "step": 264540 }, { "epoch": 0.011904, "grad_norm": 0.7613547444343567, "learning_rate": 4.726034075659056e-06, "loss": 2.9028, "step": 264550 }, { "epoch": 0.0119296, "grad_norm": 0.7784790992736816, "learning_rate": 4.725349912682216e-06, "loss": 2.681, "step": 264560 }, { "epoch": 0.0119552, "grad_norm": 0.8801794648170471, "learning_rate": 4.724665783910079e-06, "loss": 2.8324, "step": 264570 }, { "epoch": 0.0119808, "grad_norm": 0.7610428333282471, "learning_rate": 4.723981689347083e-06, "loss": 2.9269, "step": 264580 }, { "epoch": 0.0120064, "grad_norm": 0.7354708909988403, "learning_rate": 4.723297628997663e-06, "loss": 2.8553, "step": 264590 }, { "epoch": 0.012032, "grad_norm": 0.7995479106903076, "learning_rate": 4.722613602866256e-06, "loss": 2.9458, "step": 264600 }, { "epoch": 0.0120576, "grad_norm": 0.7494317889213562, "learning_rate": 4.721929610957301e-06, "loss": 2.8966, "step": 264610 }, { "epoch": 0.0120832, "grad_norm": 0.8856359124183655, "learning_rate": 4.721245653275226e-06, "loss": 2.8321, "step": 264620 }, { "epoch": 0.0121088, "grad_norm": 0.8800643682479858, "learning_rate": 4.7205617298244686e-06, "loss": 2.7332, "step": 264630 }, { "epoch": 0.0121344, "grad_norm": 0.9492300152778625, "learning_rate": 4.719877840609467e-06, "loss": 2.9748, "step": 264640 }, { "epoch": 0.01216, "grad_norm": 0.9070111513137817, "learning_rate": 4.719193985634654e-06, "loss": 2.8108, "step": 264650 }, { "epoch": 0.0121856, "grad_norm": 0.8055896162986755, "learning_rate": 4.718510164904464e-06, "loss": 2.8348, "step": 264660 }, { "epoch": 0.0122112, "grad_norm": 0.7415534853935242, "learning_rate": 4.717826378423332e-06, "loss": 3.0699, "step": 264670 }, { "epoch": 0.0122368, "grad_norm": 0.7985686659812927, "learning_rate": 4.717142626195696e-06, "loss": 2.7988, "step": 264680 }, { "epoch": 0.0122624, "grad_norm": 0.9500386118888855, "learning_rate": 4.7164589082259826e-06, "loss": 2.7995, "step": 264690 }, { "epoch": 0.012288, "grad_norm": 0.9508262872695923, "learning_rate": 4.715775224518629e-06, "loss": 2.835, "step": 264700 }, { "epoch": 0.0123136, "grad_norm": 0.8177362084388733, "learning_rate": 4.715091575078068e-06, "loss": 2.7815, "step": 264710 }, { "epoch": 0.0123392, "grad_norm": 0.8343334794044495, "learning_rate": 4.714407959908734e-06, "loss": 2.8695, "step": 264720 }, { "epoch": 0.0123648, "grad_norm": 1.4211981296539307, "learning_rate": 4.71372437901506e-06, "loss": 2.9809, "step": 264730 }, { "epoch": 0.0123904, "grad_norm": 0.7449261546134949, "learning_rate": 4.713040832401478e-06, "loss": 2.8891, "step": 264740 }, { "epoch": 0.012416, "grad_norm": 1.0805929899215698, "learning_rate": 4.7123573200724215e-06, "loss": 2.9627, "step": 264750 }, { "epoch": 0.0124416, "grad_norm": 0.8272106051445007, "learning_rate": 4.711673842032322e-06, "loss": 2.8254, "step": 264760 }, { "epoch": 0.0124672, "grad_norm": 0.8021241426467896, "learning_rate": 4.710990398285612e-06, "loss": 2.9701, "step": 264770 }, { "epoch": 0.0124928, "grad_norm": 0.7440956234931946, "learning_rate": 4.7103069888367245e-06, "loss": 2.6473, "step": 264780 }, { "epoch": 0.0125184, "grad_norm": 0.9677755236625671, "learning_rate": 4.709623613690089e-06, "loss": 2.7681, "step": 264790 }, { "epoch": 0.012544, "grad_norm": 0.829105019569397, "learning_rate": 4.70894027285014e-06, "loss": 2.7937, "step": 264800 }, { "epoch": 0.0125696, "grad_norm": 0.764153242111206, "learning_rate": 4.70825696632131e-06, "loss": 3.0557, "step": 264810 }, { "epoch": 0.0125952, "grad_norm": 1.025241494178772, "learning_rate": 4.707573694108024e-06, "loss": 2.7829, "step": 264820 }, { "epoch": 0.0126208, "grad_norm": 1.043730616569519, "learning_rate": 4.706890456214715e-06, "loss": 2.7109, "step": 264830 }, { "epoch": 0.0126464, "grad_norm": 0.7357205748558044, "learning_rate": 4.706207252645816e-06, "loss": 2.7478, "step": 264840 }, { "epoch": 0.012672, "grad_norm": 1.5000630617141724, "learning_rate": 4.7055240834057545e-06, "loss": 2.9426, "step": 264850 }, { "epoch": 0.0126976, "grad_norm": 0.8375683426856995, "learning_rate": 4.7048409484989635e-06, "loss": 2.644, "step": 264860 }, { "epoch": 0.0127232, "grad_norm": 0.8433275818824768, "learning_rate": 4.704157847929871e-06, "loss": 3.0185, "step": 264870 }, { "epoch": 0.0127488, "grad_norm": 0.8201912045478821, "learning_rate": 4.703474781702911e-06, "loss": 2.9114, "step": 264880 }, { "epoch": 0.0127744, "grad_norm": 0.882042646408081, "learning_rate": 4.702791749822506e-06, "loss": 2.9588, "step": 264890 }, { "epoch": 0.0128, "grad_norm": 0.7429812550544739, "learning_rate": 4.7021087522930885e-06, "loss": 2.8807, "step": 264900 }, { "epoch": 0.0128256, "grad_norm": 0.7678806185722351, "learning_rate": 4.701425789119084e-06, "loss": 2.7716, "step": 264910 }, { "epoch": 0.0128512, "grad_norm": 0.7883074283599854, "learning_rate": 4.700742860304929e-06, "loss": 3.0356, "step": 264920 }, { "epoch": 0.0128768, "grad_norm": 0.9589511752128601, "learning_rate": 4.700059965855048e-06, "loss": 2.8245, "step": 264930 }, { "epoch": 0.0129024, "grad_norm": 0.8029040098190308, "learning_rate": 4.6993771057738736e-06, "loss": 2.8878, "step": 264940 }, { "epoch": 0.012928, "grad_norm": 0.8120481967926025, "learning_rate": 4.698694280065826e-06, "loss": 2.8276, "step": 264950 }, { "epoch": 0.0129536, "grad_norm": 0.8650121092796326, "learning_rate": 4.698011488735337e-06, "loss": 2.7993, "step": 264960 }, { "epoch": 0.0129792, "grad_norm": 0.7698165774345398, "learning_rate": 4.697328731786834e-06, "loss": 2.7477, "step": 264970 }, { "epoch": 0.0130048, "grad_norm": 0.7290380001068115, "learning_rate": 4.696646009224745e-06, "loss": 2.9052, "step": 264980 }, { "epoch": 0.0130304, "grad_norm": 1.046200156211853, "learning_rate": 4.6959633210534984e-06, "loss": 2.8998, "step": 264990 }, { "epoch": 0.013056, "grad_norm": 0.8195106983184814, "learning_rate": 4.6952806672775185e-06, "loss": 2.7974, "step": 265000 }, { "epoch": 0.0130816, "grad_norm": 0.9920405745506287, "learning_rate": 4.694598047901239e-06, "loss": 2.8328, "step": 265010 }, { "epoch": 0.0131072, "grad_norm": 0.9234384298324585, "learning_rate": 4.693915462929076e-06, "loss": 2.934, "step": 265020 }, { "epoch": 0.0131328, "grad_norm": 0.9715402722358704, "learning_rate": 4.693232912365462e-06, "loss": 2.8971, "step": 265030 }, { "epoch": 0.0131584, "grad_norm": 1.3113412857055664, "learning_rate": 4.692550396214821e-06, "loss": 3.19, "step": 265040 }, { "epoch": 0.013184, "grad_norm": 0.8839790225028992, "learning_rate": 4.691867914481581e-06, "loss": 2.6359, "step": 265050 }, { "epoch": 0.0132096, "grad_norm": 0.9073086380958557, "learning_rate": 4.691185467170166e-06, "loss": 2.9165, "step": 265060 }, { "epoch": 0.0132352, "grad_norm": 0.9221028089523315, "learning_rate": 4.690503054285004e-06, "loss": 2.774, "step": 265070 }, { "epoch": 0.0132608, "grad_norm": 1.0832931995391846, "learning_rate": 4.689820675830518e-06, "loss": 3.0916, "step": 265080 }, { "epoch": 0.0132864, "grad_norm": 0.7553728818893433, "learning_rate": 4.689138331811133e-06, "loss": 2.8545, "step": 265090 }, { "epoch": 0.013312, "grad_norm": 1.266324520111084, "learning_rate": 4.688456022231274e-06, "loss": 2.7702, "step": 265100 }, { "epoch": 0.0133376, "grad_norm": 0.8455448150634766, "learning_rate": 4.687773747095367e-06, "loss": 2.8125, "step": 265110 }, { "epoch": 0.0133632, "grad_norm": 0.7517982125282288, "learning_rate": 4.687091506407835e-06, "loss": 3.0076, "step": 265120 }, { "epoch": 0.0133888, "grad_norm": 0.9299426078796387, "learning_rate": 4.686409300173102e-06, "loss": 2.8259, "step": 265130 }, { "epoch": 0.0134144, "grad_norm": 0.8921471238136292, "learning_rate": 4.6857271283955974e-06, "loss": 2.7345, "step": 265140 }, { "epoch": 0.01344, "grad_norm": 0.8269410133361816, "learning_rate": 4.685044991079736e-06, "loss": 2.8295, "step": 265150 }, { "epoch": 0.0134656, "grad_norm": 0.8469983339309692, "learning_rate": 4.684362888229945e-06, "loss": 2.8572, "step": 265160 }, { "epoch": 0.0134912, "grad_norm": 0.7300184369087219, "learning_rate": 4.683680819850648e-06, "loss": 2.8021, "step": 265170 }, { "epoch": 0.0135168, "grad_norm": 0.7353792786598206, "learning_rate": 4.682998785946268e-06, "loss": 2.6905, "step": 265180 }, { "epoch": 0.0135424, "grad_norm": 0.8772777915000916, "learning_rate": 4.682316786521228e-06, "loss": 2.7893, "step": 265190 }, { "epoch": 0.013568, "grad_norm": 1.0812160968780518, "learning_rate": 4.68163482157995e-06, "loss": 2.8213, "step": 265200 }, { "epoch": 0.0135936, "grad_norm": 0.7424115538597107, "learning_rate": 4.680952891126861e-06, "loss": 2.8323, "step": 265210 }, { "epoch": 0.0136192, "grad_norm": 0.8105082511901855, "learning_rate": 4.680270995166375e-06, "loss": 2.8399, "step": 265220 }, { "epoch": 0.0136448, "grad_norm": 1.3820703029632568, "learning_rate": 4.679589133702917e-06, "loss": 2.8094, "step": 265230 }, { "epoch": 0.0136704, "grad_norm": 0.8555790781974792, "learning_rate": 4.678907306740911e-06, "loss": 2.8021, "step": 265240 }, { "epoch": 0.013696, "grad_norm": 0.7301345467567444, "learning_rate": 4.678225514284775e-06, "loss": 2.7689, "step": 265250 }, { "epoch": 0.0137216, "grad_norm": 0.7305338382720947, "learning_rate": 4.677543756338929e-06, "loss": 2.6601, "step": 265260 }, { "epoch": 0.0137472, "grad_norm": 0.9455430507659912, "learning_rate": 4.676862032907806e-06, "loss": 2.866, "step": 265270 }, { "epoch": 0.0137728, "grad_norm": 2.01468563079834, "learning_rate": 4.676180343995813e-06, "loss": 2.8223, "step": 265280 }, { "epoch": 0.0137984, "grad_norm": 0.9033623933792114, "learning_rate": 4.675498689607376e-06, "loss": 2.7445, "step": 265290 }, { "epoch": 0.013824, "grad_norm": 4.058364391326904, "learning_rate": 4.674817069746915e-06, "loss": 2.7915, "step": 265300 }, { "epoch": 0.0138496, "grad_norm": 0.8655925393104553, "learning_rate": 4.67413548441885e-06, "loss": 2.7847, "step": 265310 }, { "epoch": 0.0138752, "grad_norm": 0.9044963121414185, "learning_rate": 4.6734539336276005e-06, "loss": 2.8763, "step": 265320 }, { "epoch": 0.0139008, "grad_norm": 0.8027561902999878, "learning_rate": 4.672772417377587e-06, "loss": 2.9222, "step": 265330 }, { "epoch": 0.0139264, "grad_norm": 0.8230448961257935, "learning_rate": 4.672090935673232e-06, "loss": 2.8825, "step": 265340 }, { "epoch": 0.013952, "grad_norm": 0.7866798639297485, "learning_rate": 4.671409488518948e-06, "loss": 2.9141, "step": 265350 }, { "epoch": 0.0139776, "grad_norm": 0.8691102266311646, "learning_rate": 4.670728075919157e-06, "loss": 2.791, "step": 265360 }, { "epoch": 0.0140032, "grad_norm": 0.7967666983604431, "learning_rate": 4.670046697878279e-06, "loss": 3.0315, "step": 265370 }, { "epoch": 0.0140288, "grad_norm": 0.7921371459960938, "learning_rate": 4.66936535440073e-06, "loss": 2.7784, "step": 265380 }, { "epoch": 0.0140544, "grad_norm": 0.7811248898506165, "learning_rate": 4.668684045490931e-06, "loss": 2.9461, "step": 265390 }, { "epoch": 0.01408, "grad_norm": 4.6299729347229, "learning_rate": 4.668002771153299e-06, "loss": 2.833, "step": 265400 }, { "epoch": 0.0141056, "grad_norm": 0.8878771662712097, "learning_rate": 4.667321531392255e-06, "loss": 3.1073, "step": 265410 }, { "epoch": 0.0141312, "grad_norm": 0.8808308243751526, "learning_rate": 4.666640326212211e-06, "loss": 2.9061, "step": 265420 }, { "epoch": 0.0141568, "grad_norm": 0.7326937913894653, "learning_rate": 4.665959155617582e-06, "loss": 2.9043, "step": 265430 }, { "epoch": 0.0141824, "grad_norm": 1.0711336135864258, "learning_rate": 4.665278019612795e-06, "loss": 2.7483, "step": 265440 }, { "epoch": 0.014208, "grad_norm": 0.9134808778762817, "learning_rate": 4.664596918202262e-06, "loss": 2.983, "step": 265450 }, { "epoch": 0.0142336, "grad_norm": 0.817338228225708, "learning_rate": 4.6639158513904e-06, "loss": 2.7951, "step": 265460 }, { "epoch": 0.0142592, "grad_norm": 0.813245952129364, "learning_rate": 4.663234819181629e-06, "loss": 2.8998, "step": 265470 }, { "epoch": 0.0142848, "grad_norm": 0.828128457069397, "learning_rate": 4.662553821580359e-06, "loss": 3.1155, "step": 265480 }, { "epoch": 0.0143104, "grad_norm": 0.8498018980026245, "learning_rate": 4.661872858591008e-06, "loss": 2.9702, "step": 265490 }, { "epoch": 0.014336, "grad_norm": 0.8737629055976868, "learning_rate": 4.661191930217993e-06, "loss": 2.6146, "step": 265500 }, { "epoch": 0.0143616, "grad_norm": 0.8233938813209534, "learning_rate": 4.6605110364657295e-06, "loss": 2.9042, "step": 265510 }, { "epoch": 0.0143872, "grad_norm": 0.9192173480987549, "learning_rate": 4.659830177338633e-06, "loss": 2.7793, "step": 265520 }, { "epoch": 0.0144128, "grad_norm": 0.8874800205230713, "learning_rate": 4.659149352841118e-06, "loss": 3.4456, "step": 265530 }, { "epoch": 0.0144384, "grad_norm": 0.9654070734977722, "learning_rate": 4.658468562977605e-06, "loss": 2.8667, "step": 265540 }, { "epoch": 0.014464, "grad_norm": 1.1481071710586548, "learning_rate": 4.657787807752498e-06, "loss": 2.8574, "step": 265550 }, { "epoch": 0.0144896, "grad_norm": 0.7932468056678772, "learning_rate": 4.6571070871702196e-06, "loss": 3.0424, "step": 265560 }, { "epoch": 0.0145152, "grad_norm": 1.14216947555542, "learning_rate": 4.65642640123518e-06, "loss": 2.8294, "step": 265570 }, { "epoch": 0.0145408, "grad_norm": 0.8088092803955078, "learning_rate": 4.6557457499517965e-06, "loss": 2.8522, "step": 265580 }, { "epoch": 0.0145664, "grad_norm": 0.842769980430603, "learning_rate": 4.65506513332448e-06, "loss": 2.7699, "step": 265590 }, { "epoch": 0.014592, "grad_norm": 0.7237634658813477, "learning_rate": 4.654384551357647e-06, "loss": 2.9681, "step": 265600 }, { "epoch": 0.0146176, "grad_norm": 0.8703255653381348, "learning_rate": 4.653704004055708e-06, "loss": 2.7384, "step": 265610 }, { "epoch": 0.0146432, "grad_norm": 0.8195402026176453, "learning_rate": 4.6530234914230785e-06, "loss": 2.9532, "step": 265620 }, { "epoch": 0.0146688, "grad_norm": 0.7799012660980225, "learning_rate": 4.652343013464171e-06, "loss": 2.8534, "step": 265630 }, { "epoch": 0.0146944, "grad_norm": 0.8365017175674438, "learning_rate": 4.651662570183397e-06, "loss": 2.8206, "step": 265640 }, { "epoch": 0.01472, "grad_norm": 0.9321938157081604, "learning_rate": 4.65098216158517e-06, "loss": 2.9883, "step": 265650 }, { "epoch": 0.0147456, "grad_norm": 0.8274303674697876, "learning_rate": 4.650301787673903e-06, "loss": 2.8472, "step": 265660 }, { "epoch": 0.0147712, "grad_norm": 1.1058568954467773, "learning_rate": 4.649621448454011e-06, "loss": 2.8801, "step": 265670 }, { "epoch": 0.0147968, "grad_norm": 0.7670047283172607, "learning_rate": 4.648941143929896e-06, "loss": 2.7421, "step": 265680 }, { "epoch": 0.0148224, "grad_norm": 0.8227085471153259, "learning_rate": 4.6482608741059776e-06, "loss": 2.797, "step": 265690 }, { "epoch": 0.014848, "grad_norm": 0.6854421496391296, "learning_rate": 4.647580638986664e-06, "loss": 2.5555, "step": 265700 }, { "epoch": 0.0148736, "grad_norm": 0.8803887367248535, "learning_rate": 4.646900438576368e-06, "loss": 2.7925, "step": 265710 }, { "epoch": 0.0148992, "grad_norm": 1.2668819427490234, "learning_rate": 4.6462202728795e-06, "loss": 2.8835, "step": 265720 }, { "epoch": 0.0149248, "grad_norm": 0.8788548707962036, "learning_rate": 4.645540141900469e-06, "loss": 2.9582, "step": 265730 }, { "epoch": 0.0149504, "grad_norm": 0.7417997717857361, "learning_rate": 4.644860045643693e-06, "loss": 2.792, "step": 265740 }, { "epoch": 0.014976, "grad_norm": 0.8348346948623657, "learning_rate": 4.6441799841135725e-06, "loss": 2.9655, "step": 265750 }, { "epoch": 0.0150016, "grad_norm": 0.8196346163749695, "learning_rate": 4.643499957314521e-06, "loss": 2.9758, "step": 265760 }, { "epoch": 0.0150272, "grad_norm": 1.048292875289917, "learning_rate": 4.642819965250949e-06, "loss": 2.872, "step": 265770 }, { "epoch": 0.0150528, "grad_norm": 0.9855927228927612, "learning_rate": 4.642140007927261e-06, "loss": 2.9014, "step": 265780 }, { "epoch": 0.0150784, "grad_norm": 0.8946656584739685, "learning_rate": 4.64146008534788e-06, "loss": 2.7447, "step": 265790 }, { "epoch": 0.015104, "grad_norm": 0.8576677441596985, "learning_rate": 4.640780197517201e-06, "loss": 2.6084, "step": 265800 }, { "epoch": 0.0151296, "grad_norm": 0.7971984148025513, "learning_rate": 4.640100344439638e-06, "loss": 2.8336, "step": 265810 }, { "epoch": 0.0151552, "grad_norm": 0.7744091153144836, "learning_rate": 4.639420526119601e-06, "loss": 2.9141, "step": 265820 }, { "epoch": 0.0151808, "grad_norm": 0.8529818058013916, "learning_rate": 4.638740742561497e-06, "loss": 2.8065, "step": 265830 }, { "epoch": 0.0152064, "grad_norm": 0.9541472792625427, "learning_rate": 4.638060993769734e-06, "loss": 2.6569, "step": 265840 }, { "epoch": 0.015232, "grad_norm": 0.8767997026443481, "learning_rate": 4.637381279748721e-06, "loss": 2.7413, "step": 265850 }, { "epoch": 0.0152576, "grad_norm": 0.8827390074729919, "learning_rate": 4.6367016005028656e-06, "loss": 2.9927, "step": 265860 }, { "epoch": 0.0152832, "grad_norm": 0.7351444959640503, "learning_rate": 4.636021956036577e-06, "loss": 2.6132, "step": 265870 }, { "epoch": 0.0153088, "grad_norm": 0.786146342754364, "learning_rate": 4.6353423463542575e-06, "loss": 2.9523, "step": 265880 }, { "epoch": 0.0153344, "grad_norm": 0.8089346289634705, "learning_rate": 4.634662771460318e-06, "loss": 2.7978, "step": 265890 }, { "epoch": 0.01536, "grad_norm": 0.7683755159378052, "learning_rate": 4.633983231359163e-06, "loss": 2.7909, "step": 265900 }, { "epoch": 0.0153856, "grad_norm": 0.791817307472229, "learning_rate": 4.6333037260552015e-06, "loss": 2.7884, "step": 265910 }, { "epoch": 0.0154112, "grad_norm": 0.8347926139831543, "learning_rate": 4.632624255552839e-06, "loss": 2.9991, "step": 265920 }, { "epoch": 0.0154368, "grad_norm": 0.9051225781440735, "learning_rate": 4.631944819856481e-06, "loss": 2.9746, "step": 265930 }, { "epoch": 0.0154624, "grad_norm": 0.7965380549430847, "learning_rate": 4.6312654189705386e-06, "loss": 2.8527, "step": 265940 }, { "epoch": 0.015488, "grad_norm": 0.7874837517738342, "learning_rate": 4.630586052899406e-06, "loss": 2.92, "step": 265950 }, { "epoch": 0.0155136, "grad_norm": 0.8599544167518616, "learning_rate": 4.629906721647499e-06, "loss": 2.7088, "step": 265960 }, { "epoch": 0.0155392, "grad_norm": 0.8184391260147095, "learning_rate": 4.629227425219219e-06, "loss": 2.9634, "step": 265970 }, { "epoch": 0.0155648, "grad_norm": 0.8803422451019287, "learning_rate": 4.628548163618974e-06, "loss": 2.9972, "step": 265980 }, { "epoch": 0.0155904, "grad_norm": 0.951302707195282, "learning_rate": 4.627868936851168e-06, "loss": 2.8977, "step": 265990 }, { "epoch": 0.015616, "grad_norm": 0.8931394815444946, "learning_rate": 4.627189744920202e-06, "loss": 3.0034, "step": 266000 }, { "epoch": 0.0156416, "grad_norm": 0.7986830472946167, "learning_rate": 4.626510587830482e-06, "loss": 2.6937, "step": 266010 }, { "epoch": 0.0156672, "grad_norm": 0.8990470767021179, "learning_rate": 4.625831465586412e-06, "loss": 2.9963, "step": 266020 }, { "epoch": 0.0156928, "grad_norm": 0.9019079208374023, "learning_rate": 4.625152378192398e-06, "loss": 2.788, "step": 266030 }, { "epoch": 0.0157184, "grad_norm": 0.8377339839935303, "learning_rate": 4.624473325652841e-06, "loss": 2.7992, "step": 266040 }, { "epoch": 0.015744, "grad_norm": 0.9031291604042053, "learning_rate": 4.623794307972147e-06, "loss": 2.9732, "step": 266050 }, { "epoch": 0.0157696, "grad_norm": 1.0250781774520874, "learning_rate": 4.623115325154718e-06, "loss": 2.7342, "step": 266060 }, { "epoch": 0.0157952, "grad_norm": 0.7607418894767761, "learning_rate": 4.622436377204961e-06, "loss": 2.794, "step": 266070 }, { "epoch": 0.0158208, "grad_norm": 0.7782995700836182, "learning_rate": 4.62175746412727e-06, "loss": 2.8753, "step": 266080 }, { "epoch": 0.0158464, "grad_norm": 0.9122952818870544, "learning_rate": 4.621078585926053e-06, "loss": 2.7541, "step": 266090 }, { "epoch": 0.015872, "grad_norm": 0.9243373870849609, "learning_rate": 4.620399742605712e-06, "loss": 2.8424, "step": 266100 }, { "epoch": 0.0158976, "grad_norm": 2.4462456703186035, "learning_rate": 4.619720934170649e-06, "loss": 2.7128, "step": 266110 }, { "epoch": 0.0159232, "grad_norm": 0.924089252948761, "learning_rate": 4.619042160625266e-06, "loss": 2.9416, "step": 266120 }, { "epoch": 0.0159488, "grad_norm": 0.7574425935745239, "learning_rate": 4.618363421973964e-06, "loss": 2.9818, "step": 266130 }, { "epoch": 0.0159744, "grad_norm": 0.7717793583869934, "learning_rate": 4.617684718221146e-06, "loss": 2.9904, "step": 266140 }, { "epoch": 0.016, "grad_norm": 0.908672571182251, "learning_rate": 4.617006049371211e-06, "loss": 2.8372, "step": 266150 }, { "epoch": 0.0160256, "grad_norm": 0.924997866153717, "learning_rate": 4.616327415428561e-06, "loss": 2.8521, "step": 266160 }, { "epoch": 0.0160512, "grad_norm": 0.8918262124061584, "learning_rate": 4.615648816397598e-06, "loss": 2.8717, "step": 266170 }, { "epoch": 0.0160768, "grad_norm": 0.9625135660171509, "learning_rate": 4.614970252282721e-06, "loss": 3.0227, "step": 266180 }, { "epoch": 0.0161024, "grad_norm": 1.7418568134307861, "learning_rate": 4.614291723088334e-06, "loss": 2.8143, "step": 266190 }, { "epoch": 0.016128, "grad_norm": 1.253462553024292, "learning_rate": 4.6136132288188304e-06, "loss": 2.9363, "step": 266200 }, { "epoch": 0.0161536, "grad_norm": 0.7812591195106506, "learning_rate": 4.612934769478613e-06, "loss": 2.8958, "step": 266210 }, { "epoch": 0.0161792, "grad_norm": 0.8144238591194153, "learning_rate": 4.612256345072083e-06, "loss": 2.9994, "step": 266220 }, { "epoch": 0.0162048, "grad_norm": 0.7498958706855774, "learning_rate": 4.611577955603638e-06, "loss": 2.906, "step": 266230 }, { "epoch": 0.0162304, "grad_norm": 0.776374340057373, "learning_rate": 4.610899601077679e-06, "loss": 2.7944, "step": 266240 }, { "epoch": 0.016256, "grad_norm": 0.7660407423973083, "learning_rate": 4.610221281498602e-06, "loss": 2.81, "step": 266250 }, { "epoch": 0.0162816, "grad_norm": 0.7884504795074463, "learning_rate": 4.609542996870812e-06, "loss": 2.7627, "step": 266260 }, { "epoch": 0.0163072, "grad_norm": 0.7448206543922424, "learning_rate": 4.608864747198699e-06, "loss": 2.9029, "step": 266270 }, { "epoch": 0.0163328, "grad_norm": 0.8504055738449097, "learning_rate": 4.608186532486666e-06, "loss": 2.9277, "step": 266280 }, { "epoch": 0.0163584, "grad_norm": 0.8956514596939087, "learning_rate": 4.607508352739111e-06, "loss": 3.1506, "step": 266290 }, { "epoch": 0.016384, "grad_norm": 0.772939145565033, "learning_rate": 4.606830207960426e-06, "loss": 2.7281, "step": 266300 }, { "epoch": 0.0164096, "grad_norm": 0.80271977186203, "learning_rate": 4.606152098155018e-06, "loss": 2.8573, "step": 266310 }, { "epoch": 0.0164352, "grad_norm": 0.8400452733039856, "learning_rate": 4.605474023327285e-06, "loss": 2.8896, "step": 266320 }, { "epoch": 0.0164608, "grad_norm": 0.87086421251297, "learning_rate": 4.6047959834816135e-06, "loss": 2.9962, "step": 266330 }, { "epoch": 0.0164864, "grad_norm": 1.0052733421325684, "learning_rate": 4.604117978622408e-06, "loss": 2.8375, "step": 266340 }, { "epoch": 0.016512, "grad_norm": 1.2403626441955566, "learning_rate": 4.603440008754062e-06, "loss": 2.7362, "step": 266350 }, { "epoch": 0.0165376, "grad_norm": 0.9183814525604248, "learning_rate": 4.602762073880973e-06, "loss": 2.9628, "step": 266360 }, { "epoch": 0.0165632, "grad_norm": 0.8183991312980652, "learning_rate": 4.6020841740075385e-06, "loss": 2.9628, "step": 266370 }, { "epoch": 0.0165888, "grad_norm": 0.7935662269592285, "learning_rate": 4.601406309138153e-06, "loss": 2.9292, "step": 266380 }, { "epoch": 0.0166144, "grad_norm": 0.8857422471046448, "learning_rate": 4.600728479277218e-06, "loss": 2.8996, "step": 266390 }, { "epoch": 0.01664, "grad_norm": 0.9221318364143372, "learning_rate": 4.600050684429119e-06, "loss": 2.7202, "step": 266400 }, { "epoch": 0.0166656, "grad_norm": 0.8713972568511963, "learning_rate": 4.599372924598257e-06, "loss": 2.8693, "step": 266410 }, { "epoch": 0.0166912, "grad_norm": 0.9320632219314575, "learning_rate": 4.598695199789025e-06, "loss": 3.1199, "step": 266420 }, { "epoch": 0.0167168, "grad_norm": 0.779383659362793, "learning_rate": 4.598017510005819e-06, "loss": 3.0051, "step": 266430 }, { "epoch": 0.0167424, "grad_norm": 0.7688450217247009, "learning_rate": 4.5973398552530344e-06, "loss": 2.8641, "step": 266440 }, { "epoch": 0.016768, "grad_norm": 0.9777953624725342, "learning_rate": 4.5966622355350655e-06, "loss": 2.9236, "step": 266450 }, { "epoch": 0.0167936, "grad_norm": 0.7853045463562012, "learning_rate": 4.5959846508563085e-06, "loss": 2.8571, "step": 266460 }, { "epoch": 0.0168192, "grad_norm": 0.7956197261810303, "learning_rate": 4.595307101221153e-06, "loss": 3.0386, "step": 266470 }, { "epoch": 0.0168448, "grad_norm": 0.8403976559638977, "learning_rate": 4.5946295866339885e-06, "loss": 2.8839, "step": 266480 }, { "epoch": 0.0168704, "grad_norm": 0.7553597092628479, "learning_rate": 4.59395210709922e-06, "loss": 2.6431, "step": 266490 }, { "epoch": 0.016896, "grad_norm": 1.137031078338623, "learning_rate": 4.593274662621235e-06, "loss": 2.9223, "step": 266500 }, { "epoch": 0.0169216, "grad_norm": 0.8175393342971802, "learning_rate": 4.592597253204426e-06, "loss": 2.8419, "step": 266510 }, { "epoch": 0.0169472, "grad_norm": 0.7507602572441101, "learning_rate": 4.591919878853191e-06, "loss": 2.854, "step": 266520 }, { "epoch": 0.0169728, "grad_norm": 0.8017517924308777, "learning_rate": 4.591242539571915e-06, "loss": 2.6663, "step": 266530 }, { "epoch": 0.0169984, "grad_norm": 0.7707898020744324, "learning_rate": 4.590565235364993e-06, "loss": 2.8761, "step": 266540 }, { "epoch": 0.017024, "grad_norm": 0.8819889426231384, "learning_rate": 4.589887966236818e-06, "loss": 2.8102, "step": 266550 }, { "epoch": 0.0170496, "grad_norm": 0.8619113564491272, "learning_rate": 4.589210732191782e-06, "loss": 3.2376, "step": 266560 }, { "epoch": 0.0170752, "grad_norm": 0.7412703037261963, "learning_rate": 4.588533533234276e-06, "loss": 2.9057, "step": 266570 }, { "epoch": 0.0171008, "grad_norm": 0.8566879034042358, "learning_rate": 4.587856369368692e-06, "loss": 2.9093, "step": 266580 }, { "epoch": 0.0171264, "grad_norm": 0.8374587297439575, "learning_rate": 4.587179240599426e-06, "loss": 3.1334, "step": 266590 }, { "epoch": 0.017152, "grad_norm": 1.4439353942871094, "learning_rate": 4.586502146930858e-06, "loss": 2.3806, "step": 266600 }, { "epoch": 0.0171776, "grad_norm": 0.7420405149459839, "learning_rate": 4.585825088367386e-06, "loss": 2.7227, "step": 266610 }, { "epoch": 0.0172032, "grad_norm": 0.9244765639305115, "learning_rate": 4.585148064913399e-06, "loss": 2.8021, "step": 266620 }, { "epoch": 0.0172288, "grad_norm": 0.8104252219200134, "learning_rate": 4.5844710765732884e-06, "loss": 3.1537, "step": 266630 }, { "epoch": 0.0172544, "grad_norm": 0.7465360760688782, "learning_rate": 4.5837941233514425e-06, "loss": 2.8496, "step": 266640 }, { "epoch": 0.01728, "grad_norm": 0.8452523350715637, "learning_rate": 4.583117205252253e-06, "loss": 2.9327, "step": 266650 }, { "epoch": 0.0173056, "grad_norm": 1.3451391458511353, "learning_rate": 4.5824403222801095e-06, "loss": 3.135, "step": 266660 }, { "epoch": 0.0173312, "grad_norm": 1.2415364980697632, "learning_rate": 4.5817634744394e-06, "loss": 3.1103, "step": 266670 }, { "epoch": 0.0173568, "grad_norm": 0.8888317942619324, "learning_rate": 4.581086661734514e-06, "loss": 2.7358, "step": 266680 }, { "epoch": 0.0173824, "grad_norm": 0.7556589841842651, "learning_rate": 4.580409884169842e-06, "loss": 2.9926, "step": 266690 }, { "epoch": 0.017408, "grad_norm": 1.703312873840332, "learning_rate": 4.57973314174977e-06, "loss": 2.7259, "step": 266700 }, { "epoch": 0.0174336, "grad_norm": 1.3559585809707642, "learning_rate": 4.579056434478689e-06, "loss": 2.949, "step": 266710 }, { "epoch": 0.0174592, "grad_norm": 0.7645557522773743, "learning_rate": 4.5783797623609895e-06, "loss": 2.7477, "step": 266720 }, { "epoch": 0.0174848, "grad_norm": 0.7928648591041565, "learning_rate": 4.577703125401053e-06, "loss": 2.8422, "step": 266730 }, { "epoch": 0.0175104, "grad_norm": 0.7095249891281128, "learning_rate": 4.5770265236032706e-06, "loss": 2.7553, "step": 266740 }, { "epoch": 0.017536, "grad_norm": 0.8594553470611572, "learning_rate": 4.5763499569720305e-06, "loss": 2.7931, "step": 266750 }, { "epoch": 0.0175616, "grad_norm": 0.8659684658050537, "learning_rate": 4.575673425511719e-06, "loss": 2.9936, "step": 266760 }, { "epoch": 0.0175872, "grad_norm": 0.8223424553871155, "learning_rate": 4.574996929226724e-06, "loss": 2.7909, "step": 266770 }, { "epoch": 0.0176128, "grad_norm": 0.7418496012687683, "learning_rate": 4.574320468121431e-06, "loss": 3.0291, "step": 266780 }, { "epoch": 0.0176384, "grad_norm": 0.7770134806632996, "learning_rate": 4.573644042200233e-06, "loss": 2.8924, "step": 266790 }, { "epoch": 0.017664, "grad_norm": 0.8033969402313232, "learning_rate": 4.572967651467506e-06, "loss": 2.8932, "step": 266800 }, { "epoch": 0.0176896, "grad_norm": 0.8038050532341003, "learning_rate": 4.572291295927642e-06, "loss": 2.9207, "step": 266810 }, { "epoch": 0.0177152, "grad_norm": 0.8502261638641357, "learning_rate": 4.571614975585024e-06, "loss": 2.6711, "step": 266820 }, { "epoch": 0.0177408, "grad_norm": 0.8568108677864075, "learning_rate": 4.570938690444042e-06, "loss": 3.0584, "step": 266830 }, { "epoch": 0.0177664, "grad_norm": 0.7266066074371338, "learning_rate": 4.5702624405090814e-06, "loss": 2.9293, "step": 266840 }, { "epoch": 0.017792, "grad_norm": 0.805321455001831, "learning_rate": 4.5695862257845284e-06, "loss": 2.8218, "step": 266850 }, { "epoch": 0.0178176, "grad_norm": 0.792479932308197, "learning_rate": 4.568910046274761e-06, "loss": 2.8664, "step": 266860 }, { "epoch": 0.0178432, "grad_norm": 0.7497124075889587, "learning_rate": 4.568233901984171e-06, "loss": 2.8018, "step": 266870 }, { "epoch": 0.0178688, "grad_norm": 0.7512218356132507, "learning_rate": 4.567557792917139e-06, "loss": 2.6777, "step": 266880 }, { "epoch": 0.0178944, "grad_norm": 0.7870727181434631, "learning_rate": 4.566881719078051e-06, "loss": 2.8938, "step": 266890 }, { "epoch": 0.01792, "grad_norm": 1.2760446071624756, "learning_rate": 4.566205680471291e-06, "loss": 2.9776, "step": 266900 }, { "epoch": 0.0179456, "grad_norm": 0.7553344964981079, "learning_rate": 4.565529677101244e-06, "loss": 2.8988, "step": 266910 }, { "epoch": 0.0179712, "grad_norm": 0.8132312297821045, "learning_rate": 4.564853708972296e-06, "loss": 2.7505, "step": 266920 }, { "epoch": 0.0179968, "grad_norm": 0.7993513345718384, "learning_rate": 4.564177776088822e-06, "loss": 2.9966, "step": 266930 }, { "epoch": 0.0180224, "grad_norm": 1.001939296722412, "learning_rate": 4.563501878455212e-06, "loss": 2.8834, "step": 266940 }, { "epoch": 0.018048, "grad_norm": 0.7702880501747131, "learning_rate": 4.562826016075848e-06, "loss": 2.8247, "step": 266950 }, { "epoch": 0.0180736, "grad_norm": 0.7759977579116821, "learning_rate": 4.56215018895511e-06, "loss": 3.028, "step": 266960 }, { "epoch": 0.0180992, "grad_norm": 0.7776974439620972, "learning_rate": 4.5614743970973856e-06, "loss": 2.947, "step": 266970 }, { "epoch": 0.0181248, "grad_norm": 0.8182370662689209, "learning_rate": 4.560798640507052e-06, "loss": 2.8694, "step": 266980 }, { "epoch": 0.0181504, "grad_norm": 0.7851715683937073, "learning_rate": 4.560122919188499e-06, "loss": 2.862, "step": 266990 }, { "epoch": 0.018176, "grad_norm": 0.7471705675125122, "learning_rate": 4.559447233146095e-06, "loss": 2.8637, "step": 267000 }, { "epoch": 0.0182016, "grad_norm": 0.7378638386726379, "learning_rate": 4.558771582384233e-06, "loss": 2.7785, "step": 267010 }, { "epoch": 0.0182272, "grad_norm": 0.853305459022522, "learning_rate": 4.558095966907292e-06, "loss": 2.9488, "step": 267020 }, { "epoch": 0.0182528, "grad_norm": 0.7508935928344727, "learning_rate": 4.557420386719652e-06, "loss": 2.8793, "step": 267030 }, { "epoch": 0.0182784, "grad_norm": 0.7526723146438599, "learning_rate": 4.556744841825694e-06, "loss": 2.8092, "step": 267040 }, { "epoch": 0.018304, "grad_norm": 0.7938639521598816, "learning_rate": 4.556069332229803e-06, "loss": 2.8432, "step": 267050 }, { "epoch": 0.0183296, "grad_norm": 0.8894165754318237, "learning_rate": 4.555393857936352e-06, "loss": 2.9917, "step": 267060 }, { "epoch": 0.0183552, "grad_norm": 0.8404815793037415, "learning_rate": 4.554718418949725e-06, "loss": 2.8702, "step": 267070 }, { "epoch": 0.0183808, "grad_norm": 0.8477603793144226, "learning_rate": 4.554043015274302e-06, "loss": 2.7993, "step": 267080 }, { "epoch": 0.0184064, "grad_norm": 0.9994305372238159, "learning_rate": 4.553367646914462e-06, "loss": 3.1139, "step": 267090 }, { "epoch": 0.018432, "grad_norm": 0.7620731592178345, "learning_rate": 4.552692313874586e-06, "loss": 2.9403, "step": 267100 }, { "epoch": 0.0184576, "grad_norm": 0.7955332398414612, "learning_rate": 4.552017016159051e-06, "loss": 2.9085, "step": 267110 }, { "epoch": 0.0184832, "grad_norm": 0.8177194595336914, "learning_rate": 4.5513417537722434e-06, "loss": 2.9114, "step": 267120 }, { "epoch": 0.0185088, "grad_norm": 0.8018754124641418, "learning_rate": 4.550666526718532e-06, "loss": 2.6581, "step": 267130 }, { "epoch": 0.0185344, "grad_norm": 0.865988552570343, "learning_rate": 4.549991335002301e-06, "loss": 2.9895, "step": 267140 }, { "epoch": 0.01856, "grad_norm": 0.8879271745681763, "learning_rate": 4.549316178627926e-06, "loss": 2.9304, "step": 267150 }, { "epoch": 0.0185856, "grad_norm": 0.8013210296630859, "learning_rate": 4.5486410575997875e-06, "loss": 2.6704, "step": 267160 }, { "epoch": 0.0186112, "grad_norm": 0.842516303062439, "learning_rate": 4.54796597192226e-06, "loss": 2.9258, "step": 267170 }, { "epoch": 0.0186368, "grad_norm": 1.0011051893234253, "learning_rate": 4.547290921599731e-06, "loss": 2.7926, "step": 267180 }, { "epoch": 0.0186624, "grad_norm": 0.8003929257392883, "learning_rate": 4.546615906636568e-06, "loss": 2.9861, "step": 267190 }, { "epoch": 0.018688, "grad_norm": 0.8808223605155945, "learning_rate": 4.54594092703715e-06, "loss": 2.9129, "step": 267200 }, { "epoch": 0.0187136, "grad_norm": 0.8281540274620056, "learning_rate": 4.545265982805858e-06, "loss": 2.9689, "step": 267210 }, { "epoch": 0.0187392, "grad_norm": 0.7285675406455994, "learning_rate": 4.544591073947065e-06, "loss": 2.766, "step": 267220 }, { "epoch": 0.0187648, "grad_norm": 0.8528509140014648, "learning_rate": 4.543916200465148e-06, "loss": 2.8351, "step": 267230 }, { "epoch": 0.0187904, "grad_norm": 0.8435199856758118, "learning_rate": 4.543241362364486e-06, "loss": 2.7304, "step": 267240 }, { "epoch": 0.018816, "grad_norm": 1.119560718536377, "learning_rate": 4.5425665596494564e-06, "loss": 2.9404, "step": 267250 }, { "epoch": 0.0188416, "grad_norm": 0.8149279952049255, "learning_rate": 4.541891792324428e-06, "loss": 2.8413, "step": 267260 }, { "epoch": 0.0188672, "grad_norm": 0.798337996006012, "learning_rate": 4.5412170603937825e-06, "loss": 3.0131, "step": 267270 }, { "epoch": 0.0188928, "grad_norm": 0.7879531383514404, "learning_rate": 4.540542363861893e-06, "loss": 2.8838, "step": 267280 }, { "epoch": 0.0189184, "grad_norm": 0.7476959824562073, "learning_rate": 4.539867702733135e-06, "loss": 2.9909, "step": 267290 }, { "epoch": 0.018944, "grad_norm": 0.90083247423172, "learning_rate": 4.539193077011883e-06, "loss": 2.8565, "step": 267300 }, { "epoch": 0.0189696, "grad_norm": 0.7976469993591309, "learning_rate": 4.538518486702513e-06, "loss": 2.9955, "step": 267310 }, { "epoch": 0.0189952, "grad_norm": 0.7818500995635986, "learning_rate": 4.537843931809404e-06, "loss": 2.9455, "step": 267320 }, { "epoch": 0.0190208, "grad_norm": 0.8090285062789917, "learning_rate": 4.537169412336922e-06, "loss": 2.8313, "step": 267330 }, { "epoch": 0.0190464, "grad_norm": 0.7497215867042542, "learning_rate": 4.536494928289442e-06, "loss": 2.786, "step": 267340 }, { "epoch": 0.019072, "grad_norm": 0.7895327210426331, "learning_rate": 4.535820479671339e-06, "loss": 2.9027, "step": 267350 }, { "epoch": 0.0190976, "grad_norm": 1.112269639968872, "learning_rate": 4.535146066486991e-06, "loss": 2.8337, "step": 267360 }, { "epoch": 0.0191232, "grad_norm": 0.8855725526809692, "learning_rate": 4.534471688740768e-06, "loss": 2.8611, "step": 267370 }, { "epoch": 0.0191488, "grad_norm": 1.163731336593628, "learning_rate": 4.533797346437047e-06, "loss": 2.8468, "step": 267380 }, { "epoch": 0.0191744, "grad_norm": 0.7965706586837769, "learning_rate": 4.533123039580194e-06, "loss": 2.688, "step": 267390 }, { "epoch": 0.0192, "grad_norm": 0.9172622561454773, "learning_rate": 4.532448768174585e-06, "loss": 2.9584, "step": 267400 }, { "epoch": 0.0192256, "grad_norm": 0.9696881175041199, "learning_rate": 4.531774532224593e-06, "loss": 2.7945, "step": 267410 }, { "epoch": 0.0192512, "grad_norm": 1.1078698635101318, "learning_rate": 4.53110033173459e-06, "loss": 2.9626, "step": 267420 }, { "epoch": 0.0192768, "grad_norm": 0.7832490801811218, "learning_rate": 4.530426166708948e-06, "loss": 2.5517, "step": 267430 }, { "epoch": 0.0193024, "grad_norm": 0.8094643354415894, "learning_rate": 4.529752037152039e-06, "loss": 2.8074, "step": 267440 }, { "epoch": 0.019328, "grad_norm": 0.9378412365913391, "learning_rate": 4.5290779430682375e-06, "loss": 2.8682, "step": 267450 }, { "epoch": 0.0193536, "grad_norm": 0.8004348278045654, "learning_rate": 4.528403884461907e-06, "loss": 2.683, "step": 267460 }, { "epoch": 0.0193792, "grad_norm": 0.8339777588844299, "learning_rate": 4.5277298613374246e-06, "loss": 2.767, "step": 267470 }, { "epoch": 0.0194048, "grad_norm": 1.0508840084075928, "learning_rate": 4.5270558736991585e-06, "loss": 2.9377, "step": 267480 }, { "epoch": 0.0194304, "grad_norm": 0.8401446342468262, "learning_rate": 4.526381921551481e-06, "loss": 3.183, "step": 267490 }, { "epoch": 0.019456, "grad_norm": 0.7816187739372253, "learning_rate": 4.525708004898761e-06, "loss": 2.7417, "step": 267500 }, { "epoch": 0.0194816, "grad_norm": 0.8331683874130249, "learning_rate": 4.525034123745371e-06, "loss": 2.5288, "step": 267510 }, { "epoch": 0.0195072, "grad_norm": 0.7619444131851196, "learning_rate": 4.5243602780956795e-06, "loss": 2.6794, "step": 267520 }, { "epoch": 0.0195328, "grad_norm": 0.8547788858413696, "learning_rate": 4.523686467954056e-06, "loss": 2.7, "step": 267530 }, { "epoch": 0.0195584, "grad_norm": 0.8453103303909302, "learning_rate": 4.523012693324871e-06, "loss": 2.4319, "step": 267540 }, { "epoch": 0.019584, "grad_norm": 0.7474826574325562, "learning_rate": 4.522338954212492e-06, "loss": 2.7707, "step": 267550 }, { "epoch": 0.0196096, "grad_norm": 0.742668092250824, "learning_rate": 4.5216652506212896e-06, "loss": 2.7478, "step": 267560 }, { "epoch": 0.0196352, "grad_norm": 0.7584457993507385, "learning_rate": 4.520991582555636e-06, "loss": 2.5147, "step": 267570 }, { "epoch": 0.0196608, "grad_norm": 1.103275179862976, "learning_rate": 4.520317950019892e-06, "loss": 2.712, "step": 267580 }, { "epoch": 0.0196864, "grad_norm": 0.8320592641830444, "learning_rate": 4.51964435301843e-06, "loss": 2.4107, "step": 267590 }, { "epoch": 0.019712, "grad_norm": 0.8141682147979736, "learning_rate": 4.518970791555618e-06, "loss": 2.729, "step": 267600 }, { "epoch": 0.0197376, "grad_norm": 0.795664370059967, "learning_rate": 4.518297265635823e-06, "loss": 2.791, "step": 267610 }, { "epoch": 0.0197632, "grad_norm": 0.8942670226097107, "learning_rate": 4.517623775263413e-06, "loss": 2.7343, "step": 267620 }, { "epoch": 0.0197888, "grad_norm": 0.795211911201477, "learning_rate": 4.516950320442756e-06, "loss": 2.5632, "step": 267630 }, { "epoch": 0.0198144, "grad_norm": 0.84963059425354, "learning_rate": 4.51627690117822e-06, "loss": 2.7378, "step": 267640 }, { "epoch": 0.01984, "grad_norm": 0.8415331244468689, "learning_rate": 4.5156035174741745e-06, "loss": 2.8399, "step": 267650 }, { "epoch": 0.0198656, "grad_norm": 0.7513353228569031, "learning_rate": 4.514930169334979e-06, "loss": 2.5307, "step": 267660 }, { "epoch": 0.0198912, "grad_norm": 0.9077762365341187, "learning_rate": 4.514256856765002e-06, "loss": 2.6382, "step": 267670 }, { "epoch": 0.0199168, "grad_norm": 0.8627795577049255, "learning_rate": 4.513583579768613e-06, "loss": 2.7115, "step": 267680 }, { "epoch": 0.0199424, "grad_norm": 0.8041325211524963, "learning_rate": 4.5129103383501715e-06, "loss": 2.5956, "step": 267690 }, { "epoch": 0.019968, "grad_norm": 0.8313233256340027, "learning_rate": 4.512237132514057e-06, "loss": 2.3816, "step": 267700 }, { "epoch": 0.0199936, "grad_norm": 0.8399199843406677, "learning_rate": 4.511563962264621e-06, "loss": 2.5723, "step": 267710 }, { "epoch": 0.0200192, "grad_norm": 0.9226841926574707, "learning_rate": 4.510890827606234e-06, "loss": 2.6405, "step": 267720 }, { "epoch": 0.0200448, "grad_norm": 0.989078164100647, "learning_rate": 4.510217728543263e-06, "loss": 2.6264, "step": 267730 }, { "epoch": 0.0200704, "grad_norm": 0.847663164138794, "learning_rate": 4.50954466508007e-06, "loss": 2.8671, "step": 267740 }, { "epoch": 0.020096, "grad_norm": 0.8295560479164124, "learning_rate": 4.508871637221021e-06, "loss": 2.5982, "step": 267750 }, { "epoch": 0.0201216, "grad_norm": 0.7960400581359863, "learning_rate": 4.508198644970479e-06, "loss": 2.6712, "step": 267760 }, { "epoch": 0.0201472, "grad_norm": 0.7790676951408386, "learning_rate": 4.507525688332813e-06, "loss": 2.5341, "step": 267770 }, { "epoch": 0.0201728, "grad_norm": 0.8063475489616394, "learning_rate": 4.506852767312381e-06, "loss": 2.5968, "step": 267780 }, { "epoch": 0.0201984, "grad_norm": 0.7732113599777222, "learning_rate": 4.506179881913547e-06, "loss": 2.7013, "step": 267790 }, { "epoch": 0.020224, "grad_norm": 0.8446836471557617, "learning_rate": 4.505507032140678e-06, "loss": 2.5963, "step": 267800 }, { "epoch": 0.0202496, "grad_norm": 0.8557088375091553, "learning_rate": 4.504834217998134e-06, "loss": 2.6046, "step": 267810 }, { "epoch": 0.0202752, "grad_norm": 0.7778148055076599, "learning_rate": 4.504161439490281e-06, "loss": 2.7132, "step": 267820 }, { "epoch": 0.0203008, "grad_norm": 0.843986988067627, "learning_rate": 4.50348869662148e-06, "loss": 2.4805, "step": 267830 }, { "epoch": 0.0203264, "grad_norm": 0.7122617959976196, "learning_rate": 4.502815989396094e-06, "loss": 2.4976, "step": 267840 }, { "epoch": 0.020352, "grad_norm": 0.8346683382987976, "learning_rate": 4.50214331781849e-06, "loss": 2.7518, "step": 267850 }, { "epoch": 0.0203776, "grad_norm": 0.8712834119796753, "learning_rate": 4.50147068189302e-06, "loss": 2.5048, "step": 267860 }, { "epoch": 0.0204032, "grad_norm": 0.7347269058227539, "learning_rate": 4.5007980816240484e-06, "loss": 2.5543, "step": 267870 }, { "epoch": 0.0204288, "grad_norm": 0.7924768924713135, "learning_rate": 4.500125517015943e-06, "loss": 2.6167, "step": 267880 }, { "epoch": 0.0204544, "grad_norm": 0.8261469602584839, "learning_rate": 4.499452988073063e-06, "loss": 2.6918, "step": 267890 }, { "epoch": 0.02048, "grad_norm": 0.7220826745033264, "learning_rate": 4.498780494799772e-06, "loss": 2.7124, "step": 267900 }, { "epoch": 0.0205056, "grad_norm": 0.8194187879562378, "learning_rate": 4.498108037200423e-06, "loss": 2.6873, "step": 267910 }, { "epoch": 0.0205312, "grad_norm": 0.880897581577301, "learning_rate": 4.497435615279381e-06, "loss": 2.6676, "step": 267920 }, { "epoch": 0.0205568, "grad_norm": 0.854411244392395, "learning_rate": 4.496763229041007e-06, "loss": 2.7643, "step": 267930 }, { "epoch": 0.0205824, "grad_norm": 0.8490898609161377, "learning_rate": 4.496090878489661e-06, "loss": 2.8794, "step": 267940 }, { "epoch": 0.020608, "grad_norm": 0.8362969756126404, "learning_rate": 4.495418563629703e-06, "loss": 2.8731, "step": 267950 }, { "epoch": 0.0206336, "grad_norm": 0.8484517931938171, "learning_rate": 4.494746284465492e-06, "loss": 2.6762, "step": 267960 }, { "epoch": 0.0206592, "grad_norm": 0.7643757462501526, "learning_rate": 4.494074041001393e-06, "loss": 2.8164, "step": 267970 }, { "epoch": 0.0206848, "grad_norm": 0.8550894856452942, "learning_rate": 4.493401833241756e-06, "loss": 2.6113, "step": 267980 }, { "epoch": 0.0207104, "grad_norm": 0.8487599492073059, "learning_rate": 4.492729661190945e-06, "loss": 2.4708, "step": 267990 }, { "epoch": 0.020736, "grad_norm": 0.8436899185180664, "learning_rate": 4.492057524853318e-06, "loss": 2.4106, "step": 268000 }, { "epoch": 0.0207616, "grad_norm": 0.7878562808036804, "learning_rate": 4.491385424233235e-06, "loss": 2.5596, "step": 268010 }, { "epoch": 0.0207872, "grad_norm": 0.8289282321929932, "learning_rate": 4.490713359335053e-06, "loss": 2.7474, "step": 268020 }, { "epoch": 0.0208128, "grad_norm": 0.7901196479797363, "learning_rate": 4.490041330163131e-06, "loss": 2.7697, "step": 268030 }, { "epoch": 0.0208384, "grad_norm": 0.959800124168396, "learning_rate": 4.489369336721826e-06, "loss": 2.6671, "step": 268040 }, { "epoch": 0.020864, "grad_norm": 0.828263521194458, "learning_rate": 4.488697379015496e-06, "loss": 2.7559, "step": 268050 }, { "epoch": 0.0208896, "grad_norm": 0.9662021994590759, "learning_rate": 4.488025457048501e-06, "loss": 2.168, "step": 268060 }, { "epoch": 0.0209152, "grad_norm": 0.7835180759429932, "learning_rate": 4.487353570825193e-06, "loss": 2.4904, "step": 268070 }, { "epoch": 0.0209408, "grad_norm": 0.984660267829895, "learning_rate": 4.486681720349933e-06, "loss": 2.6552, "step": 268080 }, { "epoch": 0.0209664, "grad_norm": 0.9397792816162109, "learning_rate": 4.486009905627078e-06, "loss": 2.581, "step": 268090 }, { "epoch": 0.020992, "grad_norm": 0.8848846554756165, "learning_rate": 4.485338126660986e-06, "loss": 2.6189, "step": 268100 }, { "epoch": 0.0210176, "grad_norm": 0.8045523166656494, "learning_rate": 4.484666383456008e-06, "loss": 2.7434, "step": 268110 }, { "epoch": 0.0210432, "grad_norm": 0.7978660464286804, "learning_rate": 4.4839946760165006e-06, "loss": 2.7477, "step": 268120 }, { "epoch": 0.0210688, "grad_norm": 0.7979171276092529, "learning_rate": 4.4833230043468235e-06, "loss": 2.7552, "step": 268130 }, { "epoch": 0.0210944, "grad_norm": 0.8329349756240845, "learning_rate": 4.4826513684513306e-06, "loss": 2.7467, "step": 268140 }, { "epoch": 0.02112, "grad_norm": 0.878964900970459, "learning_rate": 4.4819797683343755e-06, "loss": 2.5699, "step": 268150 }, { "epoch": 0.0211456, "grad_norm": 0.9371177554130554, "learning_rate": 4.481308204000317e-06, "loss": 2.8186, "step": 268160 }, { "epoch": 0.0211712, "grad_norm": 0.9922018647193909, "learning_rate": 4.480636675453512e-06, "loss": 2.6083, "step": 268170 }, { "epoch": 0.0211968, "grad_norm": 0.8732150197029114, "learning_rate": 4.479965182698306e-06, "loss": 2.6461, "step": 268180 }, { "epoch": 0.0212224, "grad_norm": 0.7515074014663696, "learning_rate": 4.479293725739059e-06, "loss": 2.5657, "step": 268190 }, { "epoch": 0.021248, "grad_norm": 0.9564550518989563, "learning_rate": 4.4786223045801255e-06, "loss": 2.688, "step": 268200 }, { "epoch": 0.0212736, "grad_norm": 0.8299770355224609, "learning_rate": 4.4779509192258585e-06, "loss": 2.5938, "step": 268210 }, { "epoch": 0.0212992, "grad_norm": 0.7904135584831238, "learning_rate": 4.477279569680609e-06, "loss": 3.0333, "step": 268220 }, { "epoch": 0.0213248, "grad_norm": 0.8481311798095703, "learning_rate": 4.476608255948741e-06, "loss": 2.7747, "step": 268230 }, { "epoch": 0.0213504, "grad_norm": 0.7938883900642395, "learning_rate": 4.475936978034596e-06, "loss": 2.7219, "step": 268240 }, { "epoch": 0.021376, "grad_norm": 0.7864943742752075, "learning_rate": 4.4752657359425334e-06, "loss": 2.7274, "step": 268250 }, { "epoch": 0.0214016, "grad_norm": 0.9702358245849609, "learning_rate": 4.474594529676902e-06, "loss": 2.7788, "step": 268260 }, { "epoch": 0.0214272, "grad_norm": 1.007047176361084, "learning_rate": 4.4739233592420585e-06, "loss": 2.4554, "step": 268270 }, { "epoch": 0.0214528, "grad_norm": 0.8175709843635559, "learning_rate": 4.473252224642353e-06, "loss": 2.6212, "step": 268280 }, { "epoch": 0.0214784, "grad_norm": 0.9361078143119812, "learning_rate": 4.472581125882138e-06, "loss": 2.5772, "step": 268290 }, { "epoch": 0.021504, "grad_norm": 0.8168360590934753, "learning_rate": 4.471910062965767e-06, "loss": 2.6928, "step": 268300 }, { "epoch": 0.0215296, "grad_norm": 0.9645993113517761, "learning_rate": 4.471239035897589e-06, "loss": 2.5686, "step": 268310 }, { "epoch": 0.0215552, "grad_norm": 1.033023715019226, "learning_rate": 4.470568044681955e-06, "loss": 2.837, "step": 268320 }, { "epoch": 0.0215808, "grad_norm": 0.9287503361701965, "learning_rate": 4.4698970893232186e-06, "loss": 2.9468, "step": 268330 }, { "epoch": 0.0216064, "grad_norm": 0.8720455169677734, "learning_rate": 4.469226169825729e-06, "loss": 2.7068, "step": 268340 }, { "epoch": 0.021632, "grad_norm": 0.8120130896568298, "learning_rate": 4.468555286193839e-06, "loss": 2.8109, "step": 268350 }, { "epoch": 0.0216576, "grad_norm": 0.821504533290863, "learning_rate": 4.467884438431897e-06, "loss": 2.3844, "step": 268360 }, { "epoch": 0.0216832, "grad_norm": 1.0020824670791626, "learning_rate": 4.467213626544258e-06, "loss": 2.0531, "step": 268370 }, { "epoch": 0.0217088, "grad_norm": 0.9041292071342468, "learning_rate": 4.4665428505352636e-06, "loss": 2.8856, "step": 268380 }, { "epoch": 0.0217344, "grad_norm": 0.8152177333831787, "learning_rate": 4.4658721104092666e-06, "loss": 2.5594, "step": 268390 }, { "epoch": 0.02176, "grad_norm": 0.8305807709693909, "learning_rate": 4.46520140617062e-06, "loss": 2.3943, "step": 268400 }, { "epoch": 0.0217856, "grad_norm": 1.2707830667495728, "learning_rate": 4.464530737823671e-06, "loss": 2.8396, "step": 268410 }, { "epoch": 0.0218112, "grad_norm": 0.9302741289138794, "learning_rate": 4.46386010537277e-06, "loss": 2.856, "step": 268420 }, { "epoch": 0.0218368, "grad_norm": 0.7886301279067993, "learning_rate": 4.463189508822269e-06, "loss": 2.6812, "step": 268430 }, { "epoch": 0.0218624, "grad_norm": 0.7705427408218384, "learning_rate": 4.462518948176509e-06, "loss": 2.9645, "step": 268440 }, { "epoch": 0.021888, "grad_norm": 0.8702605366706848, "learning_rate": 4.461848423439841e-06, "loss": 2.7334, "step": 268450 }, { "epoch": 0.0219136, "grad_norm": 0.7881425023078918, "learning_rate": 4.461177934616614e-06, "loss": 2.8526, "step": 268460 }, { "epoch": 0.0219392, "grad_norm": 0.9242303371429443, "learning_rate": 4.460507481711177e-06, "loss": 2.7704, "step": 268470 }, { "epoch": 0.0219648, "grad_norm": 0.779961884021759, "learning_rate": 4.4598370647278756e-06, "loss": 2.5387, "step": 268480 }, { "epoch": 0.0219904, "grad_norm": 0.8520575761795044, "learning_rate": 4.459166683671059e-06, "loss": 2.6267, "step": 268490 }, { "epoch": 0.022016, "grad_norm": 0.8171792030334473, "learning_rate": 4.458496338545078e-06, "loss": 2.9423, "step": 268500 }, { "epoch": 0.0220416, "grad_norm": 0.9108584523200989, "learning_rate": 4.457826029354272e-06, "loss": 2.7586, "step": 268510 }, { "epoch": 0.0220672, "grad_norm": 0.8509511351585388, "learning_rate": 4.45715575610299e-06, "loss": 2.8237, "step": 268520 }, { "epoch": 0.0220928, "grad_norm": 0.8094779849052429, "learning_rate": 4.4564855187955806e-06, "loss": 2.8187, "step": 268530 }, { "epoch": 0.0221184, "grad_norm": 0.8249849677085876, "learning_rate": 4.455815317436389e-06, "loss": 2.8834, "step": 268540 }, { "epoch": 0.022144, "grad_norm": 0.9372179508209229, "learning_rate": 4.455145152029762e-06, "loss": 2.8891, "step": 268550 }, { "epoch": 0.0221696, "grad_norm": 0.7927029132843018, "learning_rate": 4.4544750225800445e-06, "loss": 2.832, "step": 268560 }, { "epoch": 0.0221952, "grad_norm": 0.8760881423950195, "learning_rate": 4.453804929091583e-06, "loss": 2.7978, "step": 268570 }, { "epoch": 0.0222208, "grad_norm": 0.8132011890411377, "learning_rate": 4.453134871568723e-06, "loss": 2.8886, "step": 268580 }, { "epoch": 0.0222464, "grad_norm": 0.8613794445991516, "learning_rate": 4.4524648500158085e-06, "loss": 2.7659, "step": 268590 }, { "epoch": 0.022272, "grad_norm": 0.9206904172897339, "learning_rate": 4.451794864437185e-06, "loss": 2.8192, "step": 268600 }, { "epoch": 0.0222976, "grad_norm": 0.8169381022453308, "learning_rate": 4.451124914837198e-06, "loss": 2.8194, "step": 268610 }, { "epoch": 0.0223232, "grad_norm": 0.7272748351097107, "learning_rate": 4.450455001220191e-06, "loss": 2.6984, "step": 268620 }, { "epoch": 0.0223488, "grad_norm": 0.9989567399024963, "learning_rate": 4.449785123590512e-06, "loss": 3.0157, "step": 268630 }, { "epoch": 0.0223744, "grad_norm": 0.9027867913246155, "learning_rate": 4.449115281952497e-06, "loss": 2.6063, "step": 268640 }, { "epoch": 0.0224, "grad_norm": 0.9273561835289001, "learning_rate": 4.448445476310496e-06, "loss": 2.9522, "step": 268650 }, { "epoch": 0.0224256, "grad_norm": 0.8534864783287048, "learning_rate": 4.447775706668848e-06, "loss": 2.9811, "step": 268660 }, { "epoch": 0.0224512, "grad_norm": 1.1156657934188843, "learning_rate": 4.447105973031901e-06, "loss": 3.0725, "step": 268670 }, { "epoch": 0.0224768, "grad_norm": 0.7959924340248108, "learning_rate": 4.446436275403996e-06, "loss": 2.8968, "step": 268680 }, { "epoch": 0.0225024, "grad_norm": 0.7122648358345032, "learning_rate": 4.445766613789475e-06, "loss": 2.8419, "step": 268690 }, { "epoch": 0.022528, "grad_norm": 0.8201043009757996, "learning_rate": 4.445096988192685e-06, "loss": 2.6963, "step": 268700 }, { "epoch": 0.0225536, "grad_norm": 0.7862223982810974, "learning_rate": 4.444427398617961e-06, "loss": 2.9089, "step": 268710 }, { "epoch": 0.0225792, "grad_norm": 0.8217008113861084, "learning_rate": 4.44375784506965e-06, "loss": 2.8504, "step": 268720 }, { "epoch": 0.0226048, "grad_norm": 0.8083928823471069, "learning_rate": 4.4430883275520905e-06, "loss": 2.8861, "step": 268730 }, { "epoch": 0.0226304, "grad_norm": 0.7869763374328613, "learning_rate": 4.4424188460696236e-06, "loss": 2.7609, "step": 268740 }, { "epoch": 0.022656, "grad_norm": 0.7469998002052307, "learning_rate": 4.441749400626598e-06, "loss": 2.6799, "step": 268750 }, { "epoch": 0.0226816, "grad_norm": 0.8357947468757629, "learning_rate": 4.441079991227354e-06, "loss": 2.9661, "step": 268760 }, { "epoch": 0.0227072, "grad_norm": 0.7826205492019653, "learning_rate": 4.440410617876224e-06, "loss": 2.9896, "step": 268770 }, { "epoch": 0.0227328, "grad_norm": 0.8385726809501648, "learning_rate": 4.439741280577554e-06, "loss": 2.849, "step": 268780 }, { "epoch": 0.0227584, "grad_norm": 0.8100746273994446, "learning_rate": 4.439071979335685e-06, "loss": 2.7642, "step": 268790 }, { "epoch": 0.022784, "grad_norm": 0.8583581447601318, "learning_rate": 4.438402714154955e-06, "loss": 2.7284, "step": 268800 }, { "epoch": 0.0228096, "grad_norm": 0.8320906162261963, "learning_rate": 4.437733485039707e-06, "loss": 2.8368, "step": 268810 }, { "epoch": 0.0228352, "grad_norm": 0.9534993171691895, "learning_rate": 4.437064291994277e-06, "loss": 2.8281, "step": 268820 }, { "epoch": 0.0228608, "grad_norm": 0.8002384305000305, "learning_rate": 4.436395135023012e-06, "loss": 2.6986, "step": 268830 }, { "epoch": 0.0228864, "grad_norm": 0.7758695483207703, "learning_rate": 4.435726014130242e-06, "loss": 2.8858, "step": 268840 }, { "epoch": 0.022912, "grad_norm": 0.9796104431152344, "learning_rate": 4.435056929320308e-06, "loss": 2.5785, "step": 268850 }, { "epoch": 0.0229376, "grad_norm": 0.8386833667755127, "learning_rate": 4.4343878805975524e-06, "loss": 3.0521, "step": 268860 }, { "epoch": 0.0229632, "grad_norm": 0.7958843111991882, "learning_rate": 4.4337188679663125e-06, "loss": 2.9659, "step": 268870 }, { "epoch": 0.0229888, "grad_norm": 0.7590956687927246, "learning_rate": 4.433049891430926e-06, "loss": 3.0237, "step": 268880 }, { "epoch": 0.0230144, "grad_norm": 0.8228144645690918, "learning_rate": 4.432380950995732e-06, "loss": 2.9487, "step": 268890 }, { "epoch": 0.02304, "grad_norm": 0.777251660823822, "learning_rate": 4.431712046665071e-06, "loss": 2.8517, "step": 268900 }, { "epoch": 0.0230656, "grad_norm": 0.9071947336196899, "learning_rate": 4.4310431784432694e-06, "loss": 3.0861, "step": 268910 }, { "epoch": 0.0230912, "grad_norm": 0.7447348833084106, "learning_rate": 4.430374346334678e-06, "loss": 2.619, "step": 268920 }, { "epoch": 0.0231168, "grad_norm": 0.8438158631324768, "learning_rate": 4.429705550343628e-06, "loss": 2.6713, "step": 268930 }, { "epoch": 0.0231424, "grad_norm": 0.8633637428283691, "learning_rate": 4.429036790474457e-06, "loss": 2.8611, "step": 268940 }, { "epoch": 0.023168, "grad_norm": 0.8136720657348633, "learning_rate": 4.428368066731502e-06, "loss": 2.8833, "step": 268950 }, { "epoch": 0.0231936, "grad_norm": 0.8234465718269348, "learning_rate": 4.4276993791191025e-06, "loss": 2.7384, "step": 268960 }, { "epoch": 0.0232192, "grad_norm": 0.7834864854812622, "learning_rate": 4.427030727641588e-06, "loss": 3.0865, "step": 268970 }, { "epoch": 0.0232448, "grad_norm": 0.9079717993736267, "learning_rate": 4.426362112303298e-06, "loss": 2.9601, "step": 268980 }, { "epoch": 0.0232704, "grad_norm": 0.767033040523529, "learning_rate": 4.425693533108569e-06, "loss": 2.7864, "step": 268990 }, { "epoch": 0.023296, "grad_norm": 0.7837134003639221, "learning_rate": 4.425024990061736e-06, "loss": 2.7256, "step": 269000 }, { "epoch": 0.0233216, "grad_norm": 1.2802119255065918, "learning_rate": 4.424356483167133e-06, "loss": 3.052, "step": 269010 }, { "epoch": 0.0233472, "grad_norm": 0.8016387224197388, "learning_rate": 4.423688012429098e-06, "loss": 2.7896, "step": 269020 }, { "epoch": 0.0233728, "grad_norm": 0.732499897480011, "learning_rate": 4.423019577851968e-06, "loss": 2.9419, "step": 269030 }, { "epoch": 0.0233984, "grad_norm": 1.4071537256240845, "learning_rate": 4.4223511794400694e-06, "loss": 2.9233, "step": 269040 }, { "epoch": 0.023424, "grad_norm": 0.9139420390129089, "learning_rate": 4.421682817197741e-06, "loss": 3.0811, "step": 269050 }, { "epoch": 0.0234496, "grad_norm": 0.8551510572433472, "learning_rate": 4.421014491129317e-06, "loss": 2.7111, "step": 269060 }, { "epoch": 0.0234752, "grad_norm": 0.7895559072494507, "learning_rate": 4.420346201239132e-06, "loss": 2.9677, "step": 269070 }, { "epoch": 0.0235008, "grad_norm": 0.864928662776947, "learning_rate": 4.419677947531517e-06, "loss": 2.726, "step": 269080 }, { "epoch": 0.0235264, "grad_norm": 0.7592706084251404, "learning_rate": 4.4190097300108106e-06, "loss": 2.7946, "step": 269090 }, { "epoch": 0.023552, "grad_norm": 0.952975869178772, "learning_rate": 4.418341548681341e-06, "loss": 3.0258, "step": 269100 }, { "epoch": 0.0235776, "grad_norm": 0.8447610139846802, "learning_rate": 4.417673403547442e-06, "loss": 2.8868, "step": 269110 }, { "epoch": 0.0236032, "grad_norm": 0.8133629560470581, "learning_rate": 4.417005294613449e-06, "loss": 2.806, "step": 269120 }, { "epoch": 0.0236288, "grad_norm": 0.9582303762435913, "learning_rate": 4.416337221883692e-06, "loss": 2.7516, "step": 269130 }, { "epoch": 0.0236544, "grad_norm": 0.7380181550979614, "learning_rate": 4.415669185362506e-06, "loss": 2.8257, "step": 269140 }, { "epoch": 0.02368, "grad_norm": 0.7975431680679321, "learning_rate": 4.415001185054218e-06, "loss": 3.1545, "step": 269150 }, { "epoch": 0.0237056, "grad_norm": 0.8092128038406372, "learning_rate": 4.41433322096317e-06, "loss": 2.7575, "step": 269160 }, { "epoch": 0.0237312, "grad_norm": 0.819102942943573, "learning_rate": 4.413665293093681e-06, "loss": 3.0249, "step": 269170 }, { "epoch": 0.0237568, "grad_norm": 1.2480086088180542, "learning_rate": 4.412997401450087e-06, "loss": 2.7884, "step": 269180 }, { "epoch": 0.0237824, "grad_norm": 0.7591166496276855, "learning_rate": 4.412329546036722e-06, "loss": 2.8009, "step": 269190 }, { "epoch": 0.023808, "grad_norm": 0.866544246673584, "learning_rate": 4.411661726857915e-06, "loss": 2.8096, "step": 269200 }, { "epoch": 0.0238336, "grad_norm": 0.8058000206947327, "learning_rate": 4.410993943917995e-06, "loss": 2.8958, "step": 269210 }, { "epoch": 0.0238592, "grad_norm": 0.7912917137145996, "learning_rate": 4.410326197221294e-06, "loss": 2.7291, "step": 269220 }, { "epoch": 0.0238848, "grad_norm": 0.7754306197166443, "learning_rate": 4.409658486772146e-06, "loss": 3.0183, "step": 269230 }, { "epoch": 0.0239104, "grad_norm": 0.7872558832168579, "learning_rate": 4.408990812574873e-06, "loss": 2.9443, "step": 269240 }, { "epoch": 0.023936, "grad_norm": 0.8545482754707336, "learning_rate": 4.40832317463381e-06, "loss": 2.8206, "step": 269250 }, { "epoch": 0.0239616, "grad_norm": 0.8398406505584717, "learning_rate": 4.407655572953281e-06, "loss": 2.8939, "step": 269260 }, { "epoch": 0.0239872, "grad_norm": 0.8149398565292358, "learning_rate": 4.4069880075376225e-06, "loss": 2.8249, "step": 269270 }, { "epoch": 0.0240128, "grad_norm": 0.9416255950927734, "learning_rate": 4.406320478391164e-06, "loss": 3.0279, "step": 269280 }, { "epoch": 0.0240384, "grad_norm": 0.9379673600196838, "learning_rate": 4.4056529855182275e-06, "loss": 2.9269, "step": 269290 }, { "epoch": 0.024064, "grad_norm": 0.8166030049324036, "learning_rate": 4.404985528923143e-06, "loss": 2.7539, "step": 269300 }, { "epoch": 0.0240896, "grad_norm": 1.0032737255096436, "learning_rate": 4.404318108610243e-06, "loss": 2.9984, "step": 269310 }, { "epoch": 0.0241152, "grad_norm": 0.8312200307846069, "learning_rate": 4.403650724583851e-06, "loss": 2.9983, "step": 269320 }, { "epoch": 0.0241408, "grad_norm": 0.7939977049827576, "learning_rate": 4.402983376848297e-06, "loss": 2.7269, "step": 269330 }, { "epoch": 0.0241664, "grad_norm": 0.8410510420799255, "learning_rate": 4.4023160654079085e-06, "loss": 2.9355, "step": 269340 }, { "epoch": 0.024192, "grad_norm": 0.8857991099357605, "learning_rate": 4.401648790267012e-06, "loss": 3.0995, "step": 269350 }, { "epoch": 0.0242176, "grad_norm": 0.7812699675559998, "learning_rate": 4.40098155142994e-06, "loss": 2.6882, "step": 269360 }, { "epoch": 0.0242432, "grad_norm": 0.7698540687561035, "learning_rate": 4.40031434890101e-06, "loss": 2.7826, "step": 269370 }, { "epoch": 0.0242688, "grad_norm": 1.3554787635803223, "learning_rate": 4.399647182684555e-06, "loss": 2.663, "step": 269380 }, { "epoch": 0.0242944, "grad_norm": 0.8583908081054688, "learning_rate": 4.398980052784897e-06, "loss": 3.0059, "step": 269390 }, { "epoch": 0.02432, "grad_norm": 0.9629957675933838, "learning_rate": 4.398312959206367e-06, "loss": 2.7932, "step": 269400 }, { "epoch": 0.0243456, "grad_norm": 0.7533034682273865, "learning_rate": 4.397645901953287e-06, "loss": 2.8373, "step": 269410 }, { "epoch": 0.0243712, "grad_norm": 6.174240589141846, "learning_rate": 4.396978881029985e-06, "loss": 2.7257, "step": 269420 }, { "epoch": 0.0243968, "grad_norm": 0.773915708065033, "learning_rate": 4.3963118964407855e-06, "loss": 2.7973, "step": 269430 }, { "epoch": 0.0244224, "grad_norm": 0.7636030316352844, "learning_rate": 4.395644948190016e-06, "loss": 2.634, "step": 269440 }, { "epoch": 0.024448, "grad_norm": 0.8317500352859497, "learning_rate": 4.394978036281997e-06, "loss": 2.9785, "step": 269450 }, { "epoch": 0.0244736, "grad_norm": 0.8956093788146973, "learning_rate": 4.394311160721057e-06, "loss": 2.8718, "step": 269460 }, { "epoch": 0.0244992, "grad_norm": 0.7460319399833679, "learning_rate": 4.39364432151152e-06, "loss": 2.8542, "step": 269470 }, { "epoch": 0.0245248, "grad_norm": 0.7966717481613159, "learning_rate": 4.392977518657712e-06, "loss": 2.9325, "step": 269480 }, { "epoch": 0.0245504, "grad_norm": 0.8498506546020508, "learning_rate": 4.392310752163952e-06, "loss": 2.7512, "step": 269490 }, { "epoch": 0.024576, "grad_norm": 0.8370991945266724, "learning_rate": 4.391644022034566e-06, "loss": 2.7305, "step": 269500 }, { "epoch": 0.0246016, "grad_norm": 0.7391327619552612, "learning_rate": 4.390977328273878e-06, "loss": 2.7542, "step": 269510 }, { "epoch": 0.0246272, "grad_norm": 0.9032117128372192, "learning_rate": 4.390310670886211e-06, "loss": 2.9918, "step": 269520 }, { "epoch": 0.0246528, "grad_norm": 0.7620027661323547, "learning_rate": 4.3896440498758874e-06, "loss": 2.8724, "step": 269530 }, { "epoch": 0.0246784, "grad_norm": 0.8215527534484863, "learning_rate": 4.3889774652472325e-06, "loss": 2.8275, "step": 269540 }, { "epoch": 0.024704, "grad_norm": 0.7504416704177856, "learning_rate": 4.388310917004571e-06, "loss": 2.7497, "step": 269550 }, { "epoch": 0.0247296, "grad_norm": 0.789312481880188, "learning_rate": 4.387644405152218e-06, "loss": 2.9711, "step": 269560 }, { "epoch": 0.0247552, "grad_norm": 1.0122836828231812, "learning_rate": 4.3869779296945005e-06, "loss": 2.7649, "step": 269570 }, { "epoch": 0.0247808, "grad_norm": 0.8535758256912231, "learning_rate": 4.386311490635737e-06, "loss": 2.7807, "step": 269580 }, { "epoch": 0.0248064, "grad_norm": 0.8127313852310181, "learning_rate": 4.385645087980254e-06, "loss": 2.8524, "step": 269590 }, { "epoch": 0.024832, "grad_norm": 0.7523118257522583, "learning_rate": 4.384978721732369e-06, "loss": 2.7389, "step": 269600 }, { "epoch": 0.0248576, "grad_norm": 1.805811882019043, "learning_rate": 4.384312391896406e-06, "loss": 2.7867, "step": 269610 }, { "epoch": 0.0248832, "grad_norm": 1.3365206718444824, "learning_rate": 4.383646098476683e-06, "loss": 2.7448, "step": 269620 }, { "epoch": 0.0249088, "grad_norm": 0.9138392806053162, "learning_rate": 4.3829798414775235e-06, "loss": 2.844, "step": 269630 }, { "epoch": 0.0249344, "grad_norm": 0.8269604444503784, "learning_rate": 4.3823136209032465e-06, "loss": 2.9109, "step": 269640 }, { "epoch": 0.02496, "grad_norm": 0.8759862184524536, "learning_rate": 4.381647436758173e-06, "loss": 2.8732, "step": 269650 }, { "epoch": 0.0249856, "grad_norm": 0.737548828125, "learning_rate": 4.380981289046622e-06, "loss": 2.7324, "step": 269660 }, { "epoch": 0.0250112, "grad_norm": 0.8488948941230774, "learning_rate": 4.380315177772915e-06, "loss": 2.6934, "step": 269670 }, { "epoch": 0.0250368, "grad_norm": 0.7703497409820557, "learning_rate": 4.379649102941374e-06, "loss": 2.8679, "step": 269680 }, { "epoch": 0.0250624, "grad_norm": 0.8445053696632385, "learning_rate": 4.37898306455631e-06, "loss": 2.8425, "step": 269690 }, { "epoch": 0.025088, "grad_norm": 0.7313134670257568, "learning_rate": 4.378317062622049e-06, "loss": 2.6887, "step": 269700 }, { "epoch": 0.0251136, "grad_norm": 0.7479186654090881, "learning_rate": 4.377651097142905e-06, "loss": 2.7778, "step": 269710 }, { "epoch": 0.0251392, "grad_norm": 1.0121068954467773, "learning_rate": 4.376985168123201e-06, "loss": 2.9437, "step": 269720 }, { "epoch": 0.0251648, "grad_norm": 0.7399611473083496, "learning_rate": 4.376319275567254e-06, "loss": 2.7647, "step": 269730 }, { "epoch": 0.0251904, "grad_norm": 0.8761504888534546, "learning_rate": 4.375653419479382e-06, "loss": 2.963, "step": 269740 }, { "epoch": 0.025216, "grad_norm": 0.7836953401565552, "learning_rate": 4.374987599863906e-06, "loss": 2.65, "step": 269750 }, { "epoch": 0.0252416, "grad_norm": 0.9403292536735535, "learning_rate": 4.3743218167251366e-06, "loss": 2.6755, "step": 269760 }, { "epoch": 0.0252672, "grad_norm": 0.7401154637336731, "learning_rate": 4.373656070067396e-06, "loss": 2.7607, "step": 269770 }, { "epoch": 0.0252928, "grad_norm": 0.7777808904647827, "learning_rate": 4.372990359894996e-06, "loss": 2.9819, "step": 269780 }, { "epoch": 0.0253184, "grad_norm": 1.161738395690918, "learning_rate": 4.372324686212262e-06, "loss": 2.6438, "step": 269790 }, { "epoch": 0.025344, "grad_norm": 0.8003179430961609, "learning_rate": 4.371659049023507e-06, "loss": 3.0725, "step": 269800 }, { "epoch": 0.0253696, "grad_norm": 0.7463340163230896, "learning_rate": 4.370993448333049e-06, "loss": 2.6708, "step": 269810 }, { "epoch": 0.0253952, "grad_norm": 0.8296038508415222, "learning_rate": 4.370327884145201e-06, "loss": 2.8039, "step": 269820 }, { "epoch": 0.0254208, "grad_norm": 0.8299642205238342, "learning_rate": 4.36966235646428e-06, "loss": 2.668, "step": 269830 }, { "epoch": 0.0254464, "grad_norm": 0.8624724745750427, "learning_rate": 4.368996865294602e-06, "loss": 2.6693, "step": 269840 }, { "epoch": 0.025472, "grad_norm": 0.7737617492675781, "learning_rate": 4.368331410640483e-06, "loss": 2.7016, "step": 269850 }, { "epoch": 0.0254976, "grad_norm": 0.8729997873306274, "learning_rate": 4.367665992506238e-06, "loss": 2.7517, "step": 269860 }, { "epoch": 0.0255232, "grad_norm": 0.916817307472229, "learning_rate": 4.367000610896183e-06, "loss": 2.843, "step": 269870 }, { "epoch": 0.0255488, "grad_norm": 0.8410384654998779, "learning_rate": 4.366335265814634e-06, "loss": 2.5833, "step": 269880 }, { "epoch": 0.0255744, "grad_norm": 0.7662664651870728, "learning_rate": 4.365669957265901e-06, "loss": 2.8654, "step": 269890 }, { "epoch": 0.0256, "grad_norm": 0.7996360063552856, "learning_rate": 4.3650046852543e-06, "loss": 2.805, "step": 269900 }, { "epoch": 0.0256256, "grad_norm": 0.7748528718948364, "learning_rate": 4.364339449784148e-06, "loss": 2.9141, "step": 269910 }, { "epoch": 0.0256512, "grad_norm": 0.9871317744255066, "learning_rate": 4.363674250859754e-06, "loss": 2.9366, "step": 269920 }, { "epoch": 0.0256768, "grad_norm": 0.7589327096939087, "learning_rate": 4.363009088485436e-06, "loss": 2.8101, "step": 269930 }, { "epoch": 0.0257024, "grad_norm": 0.7627909183502197, "learning_rate": 4.362343962665506e-06, "loss": 2.7203, "step": 269940 }, { "epoch": 0.025728, "grad_norm": 0.8818424344062805, "learning_rate": 4.361678873404281e-06, "loss": 2.6846, "step": 269950 }, { "epoch": 0.0257536, "grad_norm": 0.8286229968070984, "learning_rate": 4.361013820706063e-06, "loss": 2.6832, "step": 269960 }, { "epoch": 0.0257792, "grad_norm": 1.6215159893035889, "learning_rate": 4.360348804575174e-06, "loss": 3.0052, "step": 269970 }, { "epoch": 0.0258048, "grad_norm": 0.8179784417152405, "learning_rate": 4.359683825015924e-06, "loss": 2.6978, "step": 269980 }, { "epoch": 0.0258304, "grad_norm": 0.8013446927070618, "learning_rate": 4.3590188820326275e-06, "loss": 2.8346, "step": 269990 }, { "epoch": 0.025856, "grad_norm": 0.8665812015533447, "learning_rate": 4.358353975629592e-06, "loss": 2.8562, "step": 270000 }, { "epoch": 0.0258816, "grad_norm": 0.8174430727958679, "learning_rate": 4.357689105811137e-06, "loss": 2.7295, "step": 270010 }, { "epoch": 0.0259072, "grad_norm": 0.7972216010093689, "learning_rate": 4.357024272581565e-06, "loss": 2.6627, "step": 270020 }, { "epoch": 0.0259328, "grad_norm": 0.79078209400177, "learning_rate": 4.35635947594519e-06, "loss": 2.5453, "step": 270030 }, { "epoch": 0.0259584, "grad_norm": 0.7664099335670471, "learning_rate": 4.355694715906324e-06, "loss": 2.8449, "step": 270040 }, { "epoch": 0.025984, "grad_norm": 0.9498503804206848, "learning_rate": 4.355029992469278e-06, "loss": 2.6613, "step": 270050 }, { "epoch": 0.0260096, "grad_norm": 0.8297502398490906, "learning_rate": 4.354365305638361e-06, "loss": 2.7033, "step": 270060 }, { "epoch": 0.0260352, "grad_norm": 0.8759931325912476, "learning_rate": 4.353700655417886e-06, "loss": 2.87, "step": 270070 }, { "epoch": 0.0260608, "grad_norm": 0.807823121547699, "learning_rate": 4.353036041812165e-06, "loss": 2.7001, "step": 270080 }, { "epoch": 0.0260864, "grad_norm": 0.7926666736602783, "learning_rate": 4.352371464825501e-06, "loss": 2.8054, "step": 270090 }, { "epoch": 0.026112, "grad_norm": 0.7615487575531006, "learning_rate": 4.351706924462207e-06, "loss": 2.7898, "step": 270100 }, { "epoch": 0.0261376, "grad_norm": 0.8569207191467285, "learning_rate": 4.351042420726592e-06, "loss": 2.8538, "step": 270110 }, { "epoch": 0.0261632, "grad_norm": 1.100258231163025, "learning_rate": 4.350377953622966e-06, "loss": 2.9824, "step": 270120 }, { "epoch": 0.0261888, "grad_norm": 0.7703425884246826, "learning_rate": 4.349713523155633e-06, "loss": 2.8236, "step": 270130 }, { "epoch": 0.0262144, "grad_norm": 0.7725086808204651, "learning_rate": 4.349049129328915e-06, "loss": 2.7975, "step": 270140 }, { "epoch": 0.02624, "grad_norm": 1.061413288116455, "learning_rate": 4.348384772147106e-06, "loss": 2.6976, "step": 270150 }, { "epoch": 0.0262656, "grad_norm": 0.7585852742195129, "learning_rate": 4.347720451614521e-06, "loss": 2.7754, "step": 270160 }, { "epoch": 0.0262912, "grad_norm": 0.7814412713050842, "learning_rate": 4.347056167735467e-06, "loss": 2.6015, "step": 270170 }, { "epoch": 0.0263168, "grad_norm": 0.8985809683799744, "learning_rate": 4.3463919205142506e-06, "loss": 2.8102, "step": 270180 }, { "epoch": 0.0263424, "grad_norm": 0.8327707052230835, "learning_rate": 4.345727709955181e-06, "loss": 2.8239, "step": 270190 }, { "epoch": 0.026368, "grad_norm": 0.7890555262565613, "learning_rate": 4.345063536062563e-06, "loss": 2.8364, "step": 270200 }, { "epoch": 0.0263936, "grad_norm": 0.914241373538971, "learning_rate": 4.34439939884071e-06, "loss": 3.01, "step": 270210 }, { "epoch": 0.0264192, "grad_norm": 0.7891442775726318, "learning_rate": 4.34373529829392e-06, "loss": 2.629, "step": 270220 }, { "epoch": 0.0264448, "grad_norm": 0.821802020072937, "learning_rate": 4.343071234426503e-06, "loss": 2.9726, "step": 270230 }, { "epoch": 0.0264704, "grad_norm": 0.8519718647003174, "learning_rate": 4.3424072072427644e-06, "loss": 2.9291, "step": 270240 }, { "epoch": 0.026496, "grad_norm": 0.8402509689331055, "learning_rate": 4.341743216747013e-06, "loss": 2.8171, "step": 270250 }, { "epoch": 0.0265216, "grad_norm": 0.8448869585990906, "learning_rate": 4.3410792629435525e-06, "loss": 2.8532, "step": 270260 }, { "epoch": 0.0265472, "grad_norm": 0.8147057294845581, "learning_rate": 4.340415345836689e-06, "loss": 3.0549, "step": 270270 }, { "epoch": 0.0265728, "grad_norm": 1.3013859987258911, "learning_rate": 4.3397514654307315e-06, "loss": 2.7473, "step": 270280 }, { "epoch": 0.0265984, "grad_norm": 0.9221881031990051, "learning_rate": 4.3390876217299785e-06, "loss": 2.7624, "step": 270290 }, { "epoch": 0.026624, "grad_norm": 0.7902083992958069, "learning_rate": 4.338423814738734e-06, "loss": 2.8538, "step": 270300 }, { "epoch": 0.0266496, "grad_norm": 0.7460629940032959, "learning_rate": 4.337760044461309e-06, "loss": 2.8621, "step": 270310 }, { "epoch": 0.0266752, "grad_norm": 0.7748505473136902, "learning_rate": 4.337096310902006e-06, "loss": 2.7943, "step": 270320 }, { "epoch": 0.0267008, "grad_norm": 0.7999946475028992, "learning_rate": 4.336432614065129e-06, "loss": 2.9002, "step": 270330 }, { "epoch": 0.0267264, "grad_norm": 0.9339538812637329, "learning_rate": 4.335768953954984e-06, "loss": 2.7545, "step": 270340 }, { "epoch": 0.026752, "grad_norm": 0.7785249352455139, "learning_rate": 4.335105330575869e-06, "loss": 2.6467, "step": 270350 }, { "epoch": 0.0267776, "grad_norm": 0.8076238036155701, "learning_rate": 4.334441743932089e-06, "loss": 2.7556, "step": 270360 }, { "epoch": 0.0268032, "grad_norm": 0.9084799885749817, "learning_rate": 4.33377819402795e-06, "loss": 3.0451, "step": 270370 }, { "epoch": 0.0268288, "grad_norm": 0.7944127321243286, "learning_rate": 4.333114680867753e-06, "loss": 2.9449, "step": 270380 }, { "epoch": 0.0268544, "grad_norm": 0.9001702666282654, "learning_rate": 4.332451204455801e-06, "loss": 2.7972, "step": 270390 }, { "epoch": 0.02688, "grad_norm": 0.9285470247268677, "learning_rate": 4.331787764796397e-06, "loss": 2.8792, "step": 270400 }, { "epoch": 0.0269056, "grad_norm": 0.7406872510910034, "learning_rate": 4.331124361893847e-06, "loss": 2.8678, "step": 270410 }, { "epoch": 0.0269312, "grad_norm": 0.7512311339378357, "learning_rate": 4.330460995752444e-06, "loss": 2.7555, "step": 270420 }, { "epoch": 0.0269568, "grad_norm": 0.8402616381645203, "learning_rate": 4.329797666376496e-06, "loss": 2.896, "step": 270430 }, { "epoch": 0.0269824, "grad_norm": 0.8738567233085632, "learning_rate": 4.329134373770303e-06, "loss": 2.8654, "step": 270440 }, { "epoch": 0.027008, "grad_norm": 0.9401406645774841, "learning_rate": 4.3284711179381655e-06, "loss": 2.7746, "step": 270450 }, { "epoch": 0.0270336, "grad_norm": 0.7886061668395996, "learning_rate": 4.327807898884386e-06, "loss": 2.9398, "step": 270460 }, { "epoch": 0.0270592, "grad_norm": 0.7690539956092834, "learning_rate": 4.327144716613265e-06, "loss": 2.9039, "step": 270470 }, { "epoch": 0.0270848, "grad_norm": 0.9164430499076843, "learning_rate": 4.326481571129103e-06, "loss": 2.8115, "step": 270480 }, { "epoch": 0.0271104, "grad_norm": 0.8797364234924316, "learning_rate": 4.3258184624361995e-06, "loss": 2.8235, "step": 270490 }, { "epoch": 0.027136, "grad_norm": 0.8104826211929321, "learning_rate": 4.325155390538855e-06, "loss": 2.7954, "step": 270500 }, { "epoch": 0.0271616, "grad_norm": 0.7584193348884583, "learning_rate": 4.324492355441371e-06, "loss": 2.7956, "step": 270510 }, { "epoch": 0.0271872, "grad_norm": 0.8000727295875549, "learning_rate": 4.323829357148045e-06, "loss": 2.7674, "step": 270520 }, { "epoch": 0.0272128, "grad_norm": 0.7839937210083008, "learning_rate": 4.3231663956631765e-06, "loss": 2.8972, "step": 270530 }, { "epoch": 0.0272384, "grad_norm": 0.7305145859718323, "learning_rate": 4.32250347099107e-06, "loss": 2.8097, "step": 270540 }, { "epoch": 0.027264, "grad_norm": 0.9035284519195557, "learning_rate": 4.321840583136016e-06, "loss": 2.7964, "step": 270550 }, { "epoch": 0.0272896, "grad_norm": 0.7983044385910034, "learning_rate": 4.321177732102315e-06, "loss": 2.8184, "step": 270560 }, { "epoch": 0.0273152, "grad_norm": 0.8264080286026001, "learning_rate": 4.320514917894269e-06, "loss": 2.8068, "step": 270570 }, { "epoch": 0.0273408, "grad_norm": 0.8612797260284424, "learning_rate": 4.319852140516173e-06, "loss": 2.7222, "step": 270580 }, { "epoch": 0.0273664, "grad_norm": 0.8551177978515625, "learning_rate": 4.319189399972326e-06, "loss": 3.0889, "step": 270590 }, { "epoch": 0.027392, "grad_norm": 0.9341190457344055, "learning_rate": 4.318526696267027e-06, "loss": 2.9815, "step": 270600 }, { "epoch": 0.0274176, "grad_norm": 0.974995493888855, "learning_rate": 4.317864029404576e-06, "loss": 2.893, "step": 270610 }, { "epoch": 0.0274432, "grad_norm": 0.809030294418335, "learning_rate": 4.317201399389263e-06, "loss": 2.9426, "step": 270620 }, { "epoch": 0.0274688, "grad_norm": 0.8996726870536804, "learning_rate": 4.316538806225389e-06, "loss": 2.9223, "step": 270630 }, { "epoch": 0.0274944, "grad_norm": 0.8715442419052124, "learning_rate": 4.315876249917249e-06, "loss": 2.7342, "step": 270640 }, { "epoch": 0.02752, "grad_norm": 0.8754614591598511, "learning_rate": 4.315213730469138e-06, "loss": 2.8335, "step": 270650 }, { "epoch": 0.0275456, "grad_norm": 0.9038063287734985, "learning_rate": 4.314551247885362e-06, "loss": 2.9472, "step": 270660 }, { "epoch": 0.0275712, "grad_norm": 0.7497251629829407, "learning_rate": 4.313888802170207e-06, "loss": 2.8992, "step": 270670 }, { "epoch": 0.0275968, "grad_norm": 0.8032228946685791, "learning_rate": 4.313226393327973e-06, "loss": 2.7965, "step": 270680 }, { "epoch": 0.0276224, "grad_norm": 0.8456484079360962, "learning_rate": 4.312564021362953e-06, "loss": 2.8814, "step": 270690 }, { "epoch": 0.027648, "grad_norm": 0.8247681260108948, "learning_rate": 4.3119016862794435e-06, "loss": 2.7367, "step": 270700 }, { "epoch": 0.0276736, "grad_norm": 0.8150022029876709, "learning_rate": 4.311239388081741e-06, "loss": 2.8696, "step": 270710 }, { "epoch": 0.0276992, "grad_norm": 0.7416117191314697, "learning_rate": 4.310577126774138e-06, "loss": 2.9798, "step": 270720 }, { "epoch": 0.0277248, "grad_norm": 0.9508228302001953, "learning_rate": 4.309914902360931e-06, "loss": 2.7802, "step": 270730 }, { "epoch": 0.0277504, "grad_norm": 0.782746434211731, "learning_rate": 4.309252714846417e-06, "loss": 2.9075, "step": 270740 }, { "epoch": 0.027776, "grad_norm": 0.7909644842147827, "learning_rate": 4.308590564234884e-06, "loss": 2.6054, "step": 270750 }, { "epoch": 0.0278016, "grad_norm": 0.852971076965332, "learning_rate": 4.3079284505306275e-06, "loss": 2.7564, "step": 270760 }, { "epoch": 0.0278272, "grad_norm": 0.7775435447692871, "learning_rate": 4.307266373737943e-06, "loss": 2.9482, "step": 270770 }, { "epoch": 0.0278528, "grad_norm": 0.8741878867149353, "learning_rate": 4.3066043338611216e-06, "loss": 2.6587, "step": 270780 }, { "epoch": 0.0278784, "grad_norm": 0.7225689888000488, "learning_rate": 4.30594233090446e-06, "loss": 2.9146, "step": 270790 }, { "epoch": 0.027904, "grad_norm": 0.8475602865219116, "learning_rate": 4.305280364872247e-06, "loss": 2.8497, "step": 270800 }, { "epoch": 0.0279296, "grad_norm": 0.7256355285644531, "learning_rate": 4.304618435768783e-06, "loss": 3.0127, "step": 270810 }, { "epoch": 0.0279552, "grad_norm": 0.919914960861206, "learning_rate": 4.30395654359835e-06, "loss": 2.7831, "step": 270820 }, { "epoch": 0.0279808, "grad_norm": 0.7793561220169067, "learning_rate": 4.303294688365242e-06, "loss": 2.8679, "step": 270830 }, { "epoch": 0.0280064, "grad_norm": 0.827921450138092, "learning_rate": 4.302632870073757e-06, "loss": 2.5975, "step": 270840 }, { "epoch": 0.028032, "grad_norm": 0.8455668687820435, "learning_rate": 4.301971088728184e-06, "loss": 2.7529, "step": 270850 }, { "epoch": 0.0280576, "grad_norm": 1.0050221681594849, "learning_rate": 4.301309344332818e-06, "loss": 2.8023, "step": 270860 }, { "epoch": 0.0280832, "grad_norm": 0.813075840473175, "learning_rate": 4.300647636891942e-06, "loss": 2.7545, "step": 270870 }, { "epoch": 0.0281088, "grad_norm": 1.09579336643219, "learning_rate": 4.299985966409852e-06, "loss": 2.7771, "step": 270880 }, { "epoch": 0.0281344, "grad_norm": 1.1157841682434082, "learning_rate": 4.2993243328908375e-06, "loss": 2.7592, "step": 270890 }, { "epoch": 0.02816, "grad_norm": 0.8423797488212585, "learning_rate": 4.29866273633919e-06, "loss": 2.9901, "step": 270900 }, { "epoch": 0.0281856, "grad_norm": 0.8618876338005066, "learning_rate": 4.2980011767592e-06, "loss": 2.849, "step": 270910 }, { "epoch": 0.0282112, "grad_norm": 0.8301748633384705, "learning_rate": 4.297339654155156e-06, "loss": 2.6884, "step": 270920 }, { "epoch": 0.0282368, "grad_norm": 0.8623591661453247, "learning_rate": 4.29667816853135e-06, "loss": 2.805, "step": 270930 }, { "epoch": 0.0282624, "grad_norm": 0.9067925214767456, "learning_rate": 4.2960167198920725e-06, "loss": 2.8915, "step": 270940 }, { "epoch": 0.028288, "grad_norm": 0.8809365630149841, "learning_rate": 4.295355308241607e-06, "loss": 2.9544, "step": 270950 }, { "epoch": 0.0283136, "grad_norm": 0.9540452361106873, "learning_rate": 4.294693933584247e-06, "loss": 2.7976, "step": 270960 }, { "epoch": 0.0283392, "grad_norm": 0.8956319689750671, "learning_rate": 4.29403259592428e-06, "loss": 2.9022, "step": 270970 }, { "epoch": 0.0283648, "grad_norm": 0.8397217988967896, "learning_rate": 4.2933712952659955e-06, "loss": 2.9386, "step": 270980 }, { "epoch": 0.0283904, "grad_norm": 0.7764031887054443, "learning_rate": 4.292710031613681e-06, "loss": 3.0415, "step": 270990 }, { "epoch": 0.028416, "grad_norm": 0.7786691784858704, "learning_rate": 4.2920488049716266e-06, "loss": 3.0007, "step": 271000 }, { "epoch": 0.0284416, "grad_norm": 0.8210547566413879, "learning_rate": 4.291387615344118e-06, "loss": 2.916, "step": 271010 }, { "epoch": 0.0284672, "grad_norm": 0.8035166263580322, "learning_rate": 4.290726462735443e-06, "loss": 2.7579, "step": 271020 }, { "epoch": 0.0284928, "grad_norm": 1.01613450050354, "learning_rate": 4.29006534714989e-06, "loss": 2.6997, "step": 271030 }, { "epoch": 0.0285184, "grad_norm": 0.8268240690231323, "learning_rate": 4.2894042685917455e-06, "loss": 2.9437, "step": 271040 }, { "epoch": 0.028544, "grad_norm": 0.7802167534828186, "learning_rate": 4.288743227065297e-06, "loss": 2.9235, "step": 271050 }, { "epoch": 0.0285696, "grad_norm": 0.7752158641815186, "learning_rate": 4.2880822225748344e-06, "loss": 2.8973, "step": 271060 }, { "epoch": 0.0285952, "grad_norm": 0.7999876141548157, "learning_rate": 4.287421255124636e-06, "loss": 2.8778, "step": 271070 }, { "epoch": 0.0286208, "grad_norm": 0.8277575373649597, "learning_rate": 4.286760324718994e-06, "loss": 2.8538, "step": 271080 }, { "epoch": 0.0286464, "grad_norm": 0.782153308391571, "learning_rate": 4.286099431362192e-06, "loss": 2.8948, "step": 271090 }, { "epoch": 0.028672, "grad_norm": 0.8827675580978394, "learning_rate": 4.285438575058516e-06, "loss": 2.9239, "step": 271100 }, { "epoch": 0.0286976, "grad_norm": 0.799103856086731, "learning_rate": 4.284777755812253e-06, "loss": 2.7954, "step": 271110 }, { "epoch": 0.0287232, "grad_norm": 0.8332042098045349, "learning_rate": 4.284116973627687e-06, "loss": 2.7203, "step": 271120 }, { "epoch": 0.0287488, "grad_norm": 0.7998461127281189, "learning_rate": 4.283456228509103e-06, "loss": 2.7957, "step": 271130 }, { "epoch": 0.0287744, "grad_norm": 0.8925078511238098, "learning_rate": 4.282795520460791e-06, "loss": 2.9617, "step": 271140 }, { "epoch": 0.0288, "grad_norm": 0.8308320045471191, "learning_rate": 4.282134849487026e-06, "loss": 2.935, "step": 271150 }, { "epoch": 0.0288256, "grad_norm": 0.8418428897857666, "learning_rate": 4.281474215592096e-06, "loss": 2.9988, "step": 271160 }, { "epoch": 0.0288512, "grad_norm": 0.8228631615638733, "learning_rate": 4.280813618780284e-06, "loss": 2.9433, "step": 271170 }, { "epoch": 0.0288768, "grad_norm": 0.7953410744667053, "learning_rate": 4.280153059055878e-06, "loss": 2.8826, "step": 271180 }, { "epoch": 0.0289024, "grad_norm": 0.8219360113143921, "learning_rate": 4.279492536423164e-06, "loss": 2.8145, "step": 271190 }, { "epoch": 0.028928, "grad_norm": 0.8977381587028503, "learning_rate": 4.278832050886417e-06, "loss": 2.7651, "step": 271200 }, { "epoch": 0.0289536, "grad_norm": 0.83360755443573, "learning_rate": 4.278171602449923e-06, "loss": 2.8502, "step": 271210 }, { "epoch": 0.0289792, "grad_norm": 0.9716648459434509, "learning_rate": 4.277511191117966e-06, "loss": 2.8416, "step": 271220 }, { "epoch": 0.0290048, "grad_norm": 0.8101586699485779, "learning_rate": 4.276850816894827e-06, "loss": 2.7616, "step": 271230 }, { "epoch": 0.0290304, "grad_norm": 0.749574601650238, "learning_rate": 4.27619047978479e-06, "loss": 2.8329, "step": 271240 }, { "epoch": 0.029056, "grad_norm": 0.742027997970581, "learning_rate": 4.275530179792137e-06, "loss": 2.7907, "step": 271250 }, { "epoch": 0.0290816, "grad_norm": 0.7993668913841248, "learning_rate": 4.274869916921153e-06, "loss": 2.9596, "step": 271260 }, { "epoch": 0.0291072, "grad_norm": 0.8278310894966125, "learning_rate": 4.274209691176111e-06, "loss": 2.8168, "step": 271270 }, { "epoch": 0.0291328, "grad_norm": 0.7758366465568542, "learning_rate": 4.273549502561298e-06, "loss": 2.7088, "step": 271280 }, { "epoch": 0.0291584, "grad_norm": 0.8611191511154175, "learning_rate": 4.272889351080995e-06, "loss": 2.9208, "step": 271290 }, { "epoch": 0.029184, "grad_norm": 0.8317705392837524, "learning_rate": 4.272229236739484e-06, "loss": 3.0378, "step": 271300 }, { "epoch": 0.0292096, "grad_norm": 0.8127644062042236, "learning_rate": 4.271569159541041e-06, "loss": 2.7615, "step": 271310 }, { "epoch": 0.0292352, "grad_norm": 0.761311948299408, "learning_rate": 4.2709091194899524e-06, "loss": 2.9371, "step": 271320 }, { "epoch": 0.0292608, "grad_norm": 0.7920821309089661, "learning_rate": 4.270249116590494e-06, "loss": 2.9882, "step": 271330 }, { "epoch": 0.0292864, "grad_norm": 0.8289006948471069, "learning_rate": 4.269589150846951e-06, "loss": 2.8299, "step": 271340 }, { "epoch": 0.029312, "grad_norm": 1.4791781902313232, "learning_rate": 4.268929222263593e-06, "loss": 3.0142, "step": 271350 }, { "epoch": 0.0293376, "grad_norm": 0.7987555265426636, "learning_rate": 4.2682693308447085e-06, "loss": 2.8964, "step": 271360 }, { "epoch": 0.0293632, "grad_norm": 0.7575956583023071, "learning_rate": 4.2676094765945744e-06, "loss": 3.0619, "step": 271370 }, { "epoch": 0.0293888, "grad_norm": 0.8112315535545349, "learning_rate": 4.26694965951747e-06, "loss": 2.9847, "step": 271380 }, { "epoch": 0.0294144, "grad_norm": 1.634372353553772, "learning_rate": 4.2662898796176754e-06, "loss": 2.5708, "step": 271390 }, { "epoch": 0.02944, "grad_norm": 0.8065472841262817, "learning_rate": 4.265630136899465e-06, "loss": 3.0106, "step": 271400 }, { "epoch": 0.0294656, "grad_norm": 0.8073643445968628, "learning_rate": 4.264970431367118e-06, "loss": 2.7757, "step": 271410 }, { "epoch": 0.0294912, "grad_norm": 0.7629348635673523, "learning_rate": 4.264310763024915e-06, "loss": 2.6327, "step": 271420 }, { "epoch": 0.0295168, "grad_norm": 0.9033949971199036, "learning_rate": 4.263651131877131e-06, "loss": 2.9062, "step": 271430 }, { "epoch": 0.0295424, "grad_norm": 0.7633361220359802, "learning_rate": 4.262991537928046e-06, "loss": 2.8533, "step": 271440 }, { "epoch": 0.029568, "grad_norm": 0.9194744825363159, "learning_rate": 4.262331981181936e-06, "loss": 2.7897, "step": 271450 }, { "epoch": 0.0295936, "grad_norm": 0.8041872382164001, "learning_rate": 4.26167246164308e-06, "loss": 2.865, "step": 271460 }, { "epoch": 0.0296192, "grad_norm": 0.9206161499023438, "learning_rate": 4.261012979315752e-06, "loss": 2.8852, "step": 271470 }, { "epoch": 0.0296448, "grad_norm": 0.8673995733261108, "learning_rate": 4.260353534204228e-06, "loss": 2.9695, "step": 271480 }, { "epoch": 0.0296704, "grad_norm": 0.8091479539871216, "learning_rate": 4.2596941263127865e-06, "loss": 2.8857, "step": 271490 }, { "epoch": 0.029696, "grad_norm": 0.7911152839660645, "learning_rate": 4.259034755645703e-06, "loss": 2.9406, "step": 271500 }, { "epoch": 0.0297216, "grad_norm": 0.7841716408729553, "learning_rate": 4.258375422207253e-06, "loss": 2.927, "step": 271510 }, { "epoch": 0.0297472, "grad_norm": 0.7963041067123413, "learning_rate": 4.257716126001713e-06, "loss": 2.8337, "step": 271520 }, { "epoch": 0.0297728, "grad_norm": 0.7256750464439392, "learning_rate": 4.257056867033357e-06, "loss": 2.8559, "step": 271530 }, { "epoch": 0.0297984, "grad_norm": 0.982826828956604, "learning_rate": 4.256397645306462e-06, "loss": 3.0238, "step": 271540 }, { "epoch": 0.029824, "grad_norm": 0.9795306921005249, "learning_rate": 4.2557384608253006e-06, "loss": 2.8087, "step": 271550 }, { "epoch": 0.0298496, "grad_norm": 0.7786610722541809, "learning_rate": 4.2550793135941495e-06, "loss": 2.8112, "step": 271560 }, { "epoch": 0.0298752, "grad_norm": 0.8582621812820435, "learning_rate": 4.254420203617282e-06, "loss": 2.785, "step": 271570 }, { "epoch": 0.0299008, "grad_norm": 1.229454517364502, "learning_rate": 4.253761130898972e-06, "loss": 2.8793, "step": 271580 }, { "epoch": 0.0299264, "grad_norm": 0.8375540375709534, "learning_rate": 4.253102095443498e-06, "loss": 2.7373, "step": 271590 }, { "epoch": 0.029952, "grad_norm": 0.7664128541946411, "learning_rate": 4.2524430972551256e-06, "loss": 2.9569, "step": 271600 }, { "epoch": 0.0299776, "grad_norm": 0.7717020511627197, "learning_rate": 4.251784136338133e-06, "loss": 2.551, "step": 271610 }, { "epoch": 0.0300032, "grad_norm": 0.8212359547615051, "learning_rate": 4.2511252126967916e-06, "loss": 2.8561, "step": 271620 }, { "epoch": 0.0300288, "grad_norm": 0.7943127751350403, "learning_rate": 4.250466326335376e-06, "loss": 2.7866, "step": 271630 }, { "epoch": 0.0300544, "grad_norm": 0.7518441081047058, "learning_rate": 4.249807477258157e-06, "loss": 2.7773, "step": 271640 }, { "epoch": 0.03008, "grad_norm": 0.820397675037384, "learning_rate": 4.24914866546941e-06, "loss": 2.9393, "step": 271650 }, { "epoch": 0.0301056, "grad_norm": 0.9786913990974426, "learning_rate": 4.248489890973408e-06, "loss": 2.5181, "step": 271660 }, { "epoch": 0.0301312, "grad_norm": 0.8301033973693848, "learning_rate": 4.247831153774418e-06, "loss": 2.8378, "step": 271670 }, { "epoch": 0.0301568, "grad_norm": 0.8190876841545105, "learning_rate": 4.2471724538767135e-06, "loss": 2.6896, "step": 271680 }, { "epoch": 0.0301824, "grad_norm": 0.8043221831321716, "learning_rate": 4.246513791284566e-06, "loss": 2.8889, "step": 271690 }, { "epoch": 0.030208, "grad_norm": 0.7742021083831787, "learning_rate": 4.245855166002245e-06, "loss": 2.9742, "step": 271700 }, { "epoch": 0.0302336, "grad_norm": 0.7459447383880615, "learning_rate": 4.2451965780340275e-06, "loss": 2.8999, "step": 271710 }, { "epoch": 0.0302592, "grad_norm": 0.7748852968215942, "learning_rate": 4.244538027384184e-06, "loss": 2.7545, "step": 271720 }, { "epoch": 0.0302848, "grad_norm": 0.7910745739936829, "learning_rate": 4.243879514056979e-06, "loss": 3.0053, "step": 271730 }, { "epoch": 0.0303104, "grad_norm": 0.7353671193122864, "learning_rate": 4.243221038056684e-06, "loss": 2.86, "step": 271740 }, { "epoch": 0.030336, "grad_norm": 0.8704978823661804, "learning_rate": 4.242562599387572e-06, "loss": 2.9, "step": 271750 }, { "epoch": 0.0303616, "grad_norm": 0.8170037269592285, "learning_rate": 4.241904198053911e-06, "loss": 2.6796, "step": 271760 }, { "epoch": 0.0303872, "grad_norm": 0.8592857122421265, "learning_rate": 4.24124583405997e-06, "loss": 2.7661, "step": 271770 }, { "epoch": 0.0304128, "grad_norm": 1.1274062395095825, "learning_rate": 4.24058750741002e-06, "loss": 3.0646, "step": 271780 }, { "epoch": 0.0304384, "grad_norm": 0.8416872024536133, "learning_rate": 4.239929218108333e-06, "loss": 2.8652, "step": 271790 }, { "epoch": 0.030464, "grad_norm": 0.7756351232528687, "learning_rate": 4.239270966159171e-06, "loss": 2.9283, "step": 271800 }, { "epoch": 0.0304896, "grad_norm": 0.9005497694015503, "learning_rate": 4.238612751566805e-06, "loss": 2.9513, "step": 271810 }, { "epoch": 0.0305152, "grad_norm": 0.7999497056007385, "learning_rate": 4.237954574335504e-06, "loss": 2.9051, "step": 271820 }, { "epoch": 0.0305408, "grad_norm": 0.8452165722846985, "learning_rate": 4.237296434469538e-06, "loss": 2.8932, "step": 271830 }, { "epoch": 0.0305664, "grad_norm": 0.7924706339836121, "learning_rate": 4.236638331973171e-06, "loss": 3.0146, "step": 271840 }, { "epoch": 0.030592, "grad_norm": 0.9337350726127625, "learning_rate": 4.2359802668506735e-06, "loss": 2.855, "step": 271850 }, { "epoch": 0.0306176, "grad_norm": 0.8125534653663635, "learning_rate": 4.235322239106315e-06, "loss": 2.7616, "step": 271860 }, { "epoch": 0.0306432, "grad_norm": 0.9007594585418701, "learning_rate": 4.234664248744353e-06, "loss": 2.8743, "step": 271870 }, { "epoch": 0.0306688, "grad_norm": 0.8617162108421326, "learning_rate": 4.234006295769065e-06, "loss": 2.8298, "step": 271880 }, { "epoch": 0.0306944, "grad_norm": 0.9202273488044739, "learning_rate": 4.2333483801847145e-06, "loss": 2.8773, "step": 271890 }, { "epoch": 0.03072, "grad_norm": 0.9431347250938416, "learning_rate": 4.2326905019955656e-06, "loss": 2.929, "step": 271900 }, { "epoch": 0.0307456, "grad_norm": 1.0035855770111084, "learning_rate": 4.232032661205887e-06, "loss": 2.6246, "step": 271910 }, { "epoch": 0.0307712, "grad_norm": 1.3874355554580688, "learning_rate": 4.231374857819946e-06, "loss": 3.0153, "step": 271920 }, { "epoch": 0.0307968, "grad_norm": 0.8081599473953247, "learning_rate": 4.2307170918420035e-06, "loss": 2.9716, "step": 271930 }, { "epoch": 0.0308224, "grad_norm": 0.826727569103241, "learning_rate": 4.230059363276328e-06, "loss": 2.9463, "step": 271940 }, { "epoch": 0.030848, "grad_norm": 0.8133410215377808, "learning_rate": 4.229401672127183e-06, "loss": 2.8547, "step": 271950 }, { "epoch": 0.0308736, "grad_norm": 0.8295311331748962, "learning_rate": 4.228744018398835e-06, "loss": 2.8361, "step": 271960 }, { "epoch": 0.0308992, "grad_norm": 1.5169835090637207, "learning_rate": 4.228086402095547e-06, "loss": 2.7402, "step": 271970 }, { "epoch": 0.0309248, "grad_norm": 0.8203999996185303, "learning_rate": 4.227428823221585e-06, "loss": 2.9028, "step": 271980 }, { "epoch": 0.0309504, "grad_norm": 1.0646517276763916, "learning_rate": 4.226771281781218e-06, "loss": 2.8315, "step": 271990 }, { "epoch": 0.030976, "grad_norm": 0.7744703888893127, "learning_rate": 4.226113777778701e-06, "loss": 2.9318, "step": 272000 }, { "epoch": 0.0310016, "grad_norm": 0.8467718362808228, "learning_rate": 4.2254563112183e-06, "loss": 2.8128, "step": 272010 }, { "epoch": 0.0310272, "grad_norm": 0.7704817652702332, "learning_rate": 4.224798882104282e-06, "loss": 2.7797, "step": 272020 }, { "epoch": 0.0310528, "grad_norm": 0.8287115097045898, "learning_rate": 4.224141490440907e-06, "loss": 2.727, "step": 272030 }, { "epoch": 0.0310784, "grad_norm": 0.7660039067268372, "learning_rate": 4.223484136232436e-06, "loss": 2.9772, "step": 272040 }, { "epoch": 0.031104, "grad_norm": 0.7711772918701172, "learning_rate": 4.222826819483144e-06, "loss": 2.8538, "step": 272050 }, { "epoch": 0.0311296, "grad_norm": 0.8197442889213562, "learning_rate": 4.22216954019728e-06, "loss": 2.8747, "step": 272060 }, { "epoch": 0.0311552, "grad_norm": 0.8059545159339905, "learning_rate": 4.2215122983791104e-06, "loss": 2.8542, "step": 272070 }, { "epoch": 0.0311808, "grad_norm": 0.944795548915863, "learning_rate": 4.2208550940328975e-06, "loss": 2.8214, "step": 272080 }, { "epoch": 0.0312064, "grad_norm": 0.8849993348121643, "learning_rate": 4.220197927162904e-06, "loss": 2.8335, "step": 272090 }, { "epoch": 0.031232, "grad_norm": 1.0671169757843018, "learning_rate": 4.2195407977733916e-06, "loss": 2.9653, "step": 272100 }, { "epoch": 0.0312576, "grad_norm": 0.8114474415779114, "learning_rate": 4.21888370586862e-06, "loss": 2.7548, "step": 272110 }, { "epoch": 0.0312832, "grad_norm": 0.902470588684082, "learning_rate": 4.218226651452855e-06, "loss": 2.7981, "step": 272120 }, { "epoch": 0.0313088, "grad_norm": 0.7843896746635437, "learning_rate": 4.21756963453035e-06, "loss": 2.5825, "step": 272130 }, { "epoch": 0.0313344, "grad_norm": 0.8621875047683716, "learning_rate": 4.2169126551053684e-06, "loss": 2.9778, "step": 272140 }, { "epoch": 0.03136, "grad_norm": 0.8126943707466125, "learning_rate": 4.216255713182173e-06, "loss": 2.7896, "step": 272150 }, { "epoch": 0.0313856, "grad_norm": 4.115856170654297, "learning_rate": 4.2155988087650205e-06, "loss": 2.8254, "step": 272160 }, { "epoch": 0.0314112, "grad_norm": 0.7077510356903076, "learning_rate": 4.214941941858173e-06, "loss": 2.7165, "step": 272170 }, { "epoch": 0.0314368, "grad_norm": 0.8367580771446228, "learning_rate": 4.214285112465889e-06, "loss": 3.0943, "step": 272180 }, { "epoch": 0.0314624, "grad_norm": 1.6423968076705933, "learning_rate": 4.2136283205924335e-06, "loss": 2.8249, "step": 272190 }, { "epoch": 0.031488, "grad_norm": 0.8708357214927673, "learning_rate": 4.2129715662420555e-06, "loss": 2.6749, "step": 272200 }, { "epoch": 0.0315136, "grad_norm": 0.8340100049972534, "learning_rate": 4.212314849419019e-06, "loss": 2.7448, "step": 272210 }, { "epoch": 0.0315392, "grad_norm": 0.8867541551589966, "learning_rate": 4.21165817012758e-06, "loss": 2.9632, "step": 272220 }, { "epoch": 0.0315648, "grad_norm": 0.8729619383811951, "learning_rate": 4.211001528372002e-06, "loss": 2.7185, "step": 272230 }, { "epoch": 0.0315904, "grad_norm": 0.7657760977745056, "learning_rate": 4.210344924156539e-06, "loss": 2.9426, "step": 272240 }, { "epoch": 0.031616, "grad_norm": 0.898537814617157, "learning_rate": 4.209688357485457e-06, "loss": 2.7822, "step": 272250 }, { "epoch": 0.0316416, "grad_norm": 0.8825775980949402, "learning_rate": 4.2090318283630024e-06, "loss": 2.6439, "step": 272260 }, { "epoch": 0.0316672, "grad_norm": 0.8456906080245972, "learning_rate": 4.2083753367934375e-06, "loss": 2.7981, "step": 272270 }, { "epoch": 0.0316928, "grad_norm": 1.4719706773757935, "learning_rate": 4.207718882781019e-06, "loss": 2.9938, "step": 272280 }, { "epoch": 0.0317184, "grad_norm": 0.7482456564903259, "learning_rate": 4.207062466330004e-06, "loss": 2.8628, "step": 272290 }, { "epoch": 0.031744, "grad_norm": 0.8053088784217834, "learning_rate": 4.20640608744465e-06, "loss": 2.7229, "step": 272300 }, { "epoch": 0.0317696, "grad_norm": 0.7236805558204651, "learning_rate": 4.2057497461292115e-06, "loss": 2.9368, "step": 272310 }, { "epoch": 0.0317952, "grad_norm": 0.7835902571678162, "learning_rate": 4.205093442387951e-06, "loss": 2.815, "step": 272320 }, { "epoch": 0.0318208, "grad_norm": 0.887507975101471, "learning_rate": 4.204437176225114e-06, "loss": 2.9068, "step": 272330 }, { "epoch": 0.0318464, "grad_norm": 0.7896473407745361, "learning_rate": 4.203780947644964e-06, "loss": 2.7578, "step": 272340 }, { "epoch": 0.031872, "grad_norm": 0.7101094126701355, "learning_rate": 4.203124756651753e-06, "loss": 3.0113, "step": 272350 }, { "epoch": 0.0318976, "grad_norm": 0.908632218837738, "learning_rate": 4.202468603249737e-06, "loss": 2.7472, "step": 272360 }, { "epoch": 0.0319232, "grad_norm": 0.8051949739456177, "learning_rate": 4.201812487443172e-06, "loss": 2.682, "step": 272370 }, { "epoch": 0.0319488, "grad_norm": 0.7521450519561768, "learning_rate": 4.2011564092363115e-06, "loss": 2.8067, "step": 272380 }, { "epoch": 0.0319744, "grad_norm": 1.0171749591827393, "learning_rate": 4.2005003686334116e-06, "loss": 2.8838, "step": 272390 }, { "epoch": 0.032, "grad_norm": 0.8683339953422546, "learning_rate": 4.199844365638724e-06, "loss": 2.7147, "step": 272400 }, { "epoch": 0.0320256, "grad_norm": 0.8181086182594299, "learning_rate": 4.199188400256504e-06, "loss": 2.9334, "step": 272410 }, { "epoch": 0.0320512, "grad_norm": 0.9543250799179077, "learning_rate": 4.198532472491007e-06, "loss": 2.8095, "step": 272420 }, { "epoch": 0.0320768, "grad_norm": 0.8655940294265747, "learning_rate": 4.197876582346484e-06, "loss": 3.0104, "step": 272430 }, { "epoch": 0.0321024, "grad_norm": 0.7668570876121521, "learning_rate": 4.197220729827189e-06, "loss": 2.9997, "step": 272440 }, { "epoch": 0.032128, "grad_norm": 0.7842043042182922, "learning_rate": 4.196564914937381e-06, "loss": 2.8078, "step": 272450 }, { "epoch": 0.0321536, "grad_norm": 0.8684035539627075, "learning_rate": 4.195909137681302e-06, "loss": 2.7025, "step": 272460 }, { "epoch": 0.0321792, "grad_norm": 0.8757510185241699, "learning_rate": 4.195253398063211e-06, "loss": 2.8229, "step": 272470 }, { "epoch": 0.0322048, "grad_norm": 1.2417670488357544, "learning_rate": 4.194597696087357e-06, "loss": 2.9779, "step": 272480 }, { "epoch": 0.0322304, "grad_norm": 0.7995795011520386, "learning_rate": 4.193942031757995e-06, "loss": 2.807, "step": 272490 }, { "epoch": 0.032256, "grad_norm": 0.8354308605194092, "learning_rate": 4.193286405079375e-06, "loss": 2.8369, "step": 272500 }, { "epoch": 0.0322816, "grad_norm": 0.7916421294212341, "learning_rate": 4.192630816055752e-06, "loss": 2.7774, "step": 272510 }, { "epoch": 0.0323072, "grad_norm": 0.7958284020423889, "learning_rate": 4.191975264691376e-06, "loss": 2.8092, "step": 272520 }, { "epoch": 0.0323328, "grad_norm": 0.8013945817947388, "learning_rate": 4.191319750990493e-06, "loss": 2.8384, "step": 272530 }, { "epoch": 0.0323584, "grad_norm": 0.8749656677246094, "learning_rate": 4.190664274957359e-06, "loss": 3.1628, "step": 272540 }, { "epoch": 0.032384, "grad_norm": 0.7139562964439392, "learning_rate": 4.190008836596223e-06, "loss": 2.8133, "step": 272550 }, { "epoch": 0.0324096, "grad_norm": 0.9090393781661987, "learning_rate": 4.189353435911335e-06, "loss": 2.822, "step": 272560 }, { "epoch": 0.0324352, "grad_norm": 0.6957855224609375, "learning_rate": 4.188698072906946e-06, "loss": 2.7635, "step": 272570 }, { "epoch": 0.0324608, "grad_norm": 0.7850416302680969, "learning_rate": 4.188042747587305e-06, "loss": 2.7624, "step": 272580 }, { "epoch": 0.0324864, "grad_norm": 0.7792761325836182, "learning_rate": 4.187387459956663e-06, "loss": 2.8862, "step": 272590 }, { "epoch": 0.032512, "grad_norm": 0.8143225312232971, "learning_rate": 4.186732210019269e-06, "loss": 2.5565, "step": 272600 }, { "epoch": 0.0325376, "grad_norm": 0.8368592262268066, "learning_rate": 4.18607699777937e-06, "loss": 2.852, "step": 272610 }, { "epoch": 0.0325632, "grad_norm": 0.8245437741279602, "learning_rate": 4.185421823241218e-06, "loss": 2.8193, "step": 272620 }, { "epoch": 0.0325888, "grad_norm": 0.8373677730560303, "learning_rate": 4.18476668640906e-06, "loss": 2.8426, "step": 272630 }, { "epoch": 0.0326144, "grad_norm": 0.7633054852485657, "learning_rate": 4.184111587287148e-06, "loss": 2.8847, "step": 272640 }, { "epoch": 0.03264, "grad_norm": 0.7712593078613281, "learning_rate": 4.183456525879722e-06, "loss": 2.6736, "step": 272650 }, { "epoch": 0.0326656, "grad_norm": 0.8229111433029175, "learning_rate": 4.182801502191035e-06, "loss": 2.9011, "step": 272660 }, { "epoch": 0.0326912, "grad_norm": 0.7772979140281677, "learning_rate": 4.182146516225335e-06, "loss": 2.7357, "step": 272670 }, { "epoch": 0.0327168, "grad_norm": 0.8890111446380615, "learning_rate": 4.181491567986868e-06, "loss": 2.8814, "step": 272680 }, { "epoch": 0.0327424, "grad_norm": 0.9155759215354919, "learning_rate": 4.180836657479882e-06, "loss": 2.888, "step": 272690 }, { "epoch": 0.032768, "grad_norm": 1.2343803644180298, "learning_rate": 4.180181784708624e-06, "loss": 3.0253, "step": 272700 }, { "epoch": 0.0327936, "grad_norm": 0.9143255949020386, "learning_rate": 4.179526949677339e-06, "loss": 2.6399, "step": 272710 }, { "epoch": 0.0328192, "grad_norm": 0.7790368795394897, "learning_rate": 4.178872152390281e-06, "loss": 2.9138, "step": 272720 }, { "epoch": 0.0328448, "grad_norm": 0.8387163877487183, "learning_rate": 4.1782173928516845e-06, "loss": 2.8529, "step": 272730 }, { "epoch": 0.0328704, "grad_norm": 0.8461087346076965, "learning_rate": 4.177562671065798e-06, "loss": 2.8078, "step": 272740 }, { "epoch": 0.032896, "grad_norm": 0.9535697102546692, "learning_rate": 4.176907987036873e-06, "loss": 2.8842, "step": 272750 }, { "epoch": 0.0329216, "grad_norm": 0.9855450987815857, "learning_rate": 4.176253340769153e-06, "loss": 2.8407, "step": 272760 }, { "epoch": 0.0329472, "grad_norm": 0.7239554524421692, "learning_rate": 4.175598732266885e-06, "loss": 2.8179, "step": 272770 }, { "epoch": 0.0329728, "grad_norm": 0.8101738691329956, "learning_rate": 4.174944161534309e-06, "loss": 3.0009, "step": 272780 }, { "epoch": 0.0329984, "grad_norm": 0.7277031540870667, "learning_rate": 4.174289628575672e-06, "loss": 2.7595, "step": 272790 }, { "epoch": 0.033024, "grad_norm": 0.7783127427101135, "learning_rate": 4.1736351333952175e-06, "loss": 2.9188, "step": 272800 }, { "epoch": 0.0330496, "grad_norm": 0.9076414108276367, "learning_rate": 4.17298067599719e-06, "loss": 2.8077, "step": 272810 }, { "epoch": 0.0330752, "grad_norm": 0.776233971118927, "learning_rate": 4.172326256385836e-06, "loss": 2.7973, "step": 272820 }, { "epoch": 0.0331008, "grad_norm": 1.045093059539795, "learning_rate": 4.171671874565396e-06, "loss": 2.7841, "step": 272830 }, { "epoch": 0.0331264, "grad_norm": 0.808302104473114, "learning_rate": 4.171017530540119e-06, "loss": 2.8394, "step": 272840 }, { "epoch": 0.033152, "grad_norm": 0.8196648359298706, "learning_rate": 4.170363224314241e-06, "loss": 2.8861, "step": 272850 }, { "epoch": 0.0331776, "grad_norm": 0.7939946055412292, "learning_rate": 4.1697089558920075e-06, "loss": 2.739, "step": 272860 }, { "epoch": 0.0332032, "grad_norm": 0.7442807555198669, "learning_rate": 4.169054725277663e-06, "loss": 2.7712, "step": 272870 }, { "epoch": 0.0332288, "grad_norm": 0.8464484214782715, "learning_rate": 4.168400532475448e-06, "loss": 2.8547, "step": 272880 }, { "epoch": 0.0332544, "grad_norm": 0.8028318881988525, "learning_rate": 4.167746377489605e-06, "loss": 2.9528, "step": 272890 }, { "epoch": 0.03328, "grad_norm": 0.796690046787262, "learning_rate": 4.167092260324378e-06, "loss": 2.9242, "step": 272900 }, { "epoch": 0.0333056, "grad_norm": 0.7330526113510132, "learning_rate": 4.166438180984007e-06, "loss": 2.6529, "step": 272910 }, { "epoch": 0.0333312, "grad_norm": 0.9792318344116211, "learning_rate": 4.1657841394727335e-06, "loss": 2.9604, "step": 272920 }, { "epoch": 0.0333568, "grad_norm": 0.8929740786552429, "learning_rate": 4.1651301357948005e-06, "loss": 3.1019, "step": 272930 }, { "epoch": 0.0333824, "grad_norm": 0.7663324475288391, "learning_rate": 4.164476169954448e-06, "loss": 2.8209, "step": 272940 }, { "epoch": 0.033408, "grad_norm": 0.7747218012809753, "learning_rate": 4.163822241955915e-06, "loss": 2.7318, "step": 272950 }, { "epoch": 0.0334336, "grad_norm": 0.9067963361740112, "learning_rate": 4.163168351803445e-06, "loss": 3.0085, "step": 272960 }, { "epoch": 0.0334592, "grad_norm": 0.7452452182769775, "learning_rate": 4.16251449950128e-06, "loss": 2.8079, "step": 272970 }, { "epoch": 0.0334848, "grad_norm": 0.9283996820449829, "learning_rate": 4.161860685053654e-06, "loss": 2.751, "step": 272980 }, { "epoch": 0.0335104, "grad_norm": 0.8196980953216553, "learning_rate": 4.161206908464809e-06, "loss": 2.8386, "step": 272990 }, { "epoch": 0.033536, "grad_norm": 0.8988714814186096, "learning_rate": 4.160553169738986e-06, "loss": 3.3485, "step": 273000 }, { "epoch": 0.0335616, "grad_norm": 0.9639438390731812, "learning_rate": 4.159899468880424e-06, "loss": 2.9851, "step": 273010 }, { "epoch": 0.0335872, "grad_norm": 0.7494848966598511, "learning_rate": 4.159245805893361e-06, "loss": 2.911, "step": 273020 }, { "epoch": 0.0336128, "grad_norm": 0.7240285277366638, "learning_rate": 4.158592180782037e-06, "loss": 3.0291, "step": 273030 }, { "epoch": 0.0336384, "grad_norm": 0.8184041380882263, "learning_rate": 4.157938593550694e-06, "loss": 2.8463, "step": 273040 }, { "epoch": 0.033664, "grad_norm": 0.911130428314209, "learning_rate": 4.157285044203563e-06, "loss": 2.8323, "step": 273050 }, { "epoch": 0.0336896, "grad_norm": 0.8039289712905884, "learning_rate": 4.156631532744886e-06, "loss": 2.7965, "step": 273060 }, { "epoch": 0.0337152, "grad_norm": 0.8059295415878296, "learning_rate": 4.1559780591789e-06, "loss": 2.8316, "step": 273070 }, { "epoch": 0.0337408, "grad_norm": 0.7764996290206909, "learning_rate": 4.1553246235098434e-06, "loss": 2.7767, "step": 273080 }, { "epoch": 0.0337664, "grad_norm": 0.9041740298271179, "learning_rate": 4.1546712257419495e-06, "loss": 2.9509, "step": 273090 }, { "epoch": 0.033792, "grad_norm": 0.8073201775550842, "learning_rate": 4.154017865879467e-06, "loss": 2.9082, "step": 273100 }, { "epoch": 0.0338176, "grad_norm": 0.8282181024551392, "learning_rate": 4.15336454392662e-06, "loss": 2.7333, "step": 273110 }, { "epoch": 0.0338432, "grad_norm": 0.8414103388786316, "learning_rate": 4.152711259887652e-06, "loss": 2.8299, "step": 273120 }, { "epoch": 0.0338688, "grad_norm": 0.797768235206604, "learning_rate": 4.1520580137667966e-06, "loss": 2.9282, "step": 273130 }, { "epoch": 0.0338944, "grad_norm": 0.925683856010437, "learning_rate": 4.15140480556829e-06, "loss": 2.9206, "step": 273140 }, { "epoch": 0.03392, "grad_norm": 0.8716447353363037, "learning_rate": 4.150751635296369e-06, "loss": 2.9354, "step": 273150 }, { "epoch": 0.0339456, "grad_norm": 0.7665970325469971, "learning_rate": 4.1500985029552695e-06, "loss": 2.873, "step": 273160 }, { "epoch": 0.0339712, "grad_norm": 0.8653565645217896, "learning_rate": 4.1494454085492306e-06, "loss": 2.9077, "step": 273170 }, { "epoch": 0.0339968, "grad_norm": 0.8534305691719055, "learning_rate": 4.148792352082479e-06, "loss": 2.7418, "step": 273180 }, { "epoch": 0.0340224, "grad_norm": 0.8048848509788513, "learning_rate": 4.148139333559255e-06, "loss": 2.5614, "step": 273190 }, { "epoch": 0.034048, "grad_norm": 0.8856821060180664, "learning_rate": 4.147486352983791e-06, "loss": 2.8794, "step": 273200 }, { "epoch": 0.0340736, "grad_norm": 0.9068118929862976, "learning_rate": 4.1468334103603224e-06, "loss": 2.9014, "step": 273210 }, { "epoch": 0.0340992, "grad_norm": 0.7404183149337769, "learning_rate": 4.146180505693085e-06, "loss": 2.7035, "step": 273220 }, { "epoch": 0.0341248, "grad_norm": 0.7614738941192627, "learning_rate": 4.14552763898631e-06, "loss": 2.9075, "step": 273230 }, { "epoch": 0.0341504, "grad_norm": 0.8701255917549133, "learning_rate": 4.144874810244236e-06, "loss": 2.9484, "step": 273240 }, { "epoch": 0.034176, "grad_norm": 0.8079026937484741, "learning_rate": 4.1442220194710905e-06, "loss": 2.7791, "step": 273250 }, { "epoch": 0.0342016, "grad_norm": 0.77217698097229, "learning_rate": 4.143569266671104e-06, "loss": 2.7908, "step": 273260 }, { "epoch": 0.0342272, "grad_norm": 1.5023541450500488, "learning_rate": 4.142916551848518e-06, "loss": 2.8642, "step": 273270 }, { "epoch": 0.0342528, "grad_norm": 0.8957668542861938, "learning_rate": 4.142263875007562e-06, "loss": 3.2845, "step": 273280 }, { "epoch": 0.0342784, "grad_norm": 0.7738516926765442, "learning_rate": 4.141611236152468e-06, "loss": 2.9054, "step": 273290 }, { "epoch": 0.034304, "grad_norm": 0.7761545777320862, "learning_rate": 4.140958635287471e-06, "loss": 2.974, "step": 273300 }, { "epoch": 0.0343296, "grad_norm": 0.8056603074073792, "learning_rate": 4.140306072416798e-06, "loss": 2.7922, "step": 273310 }, { "epoch": 0.0343552, "grad_norm": 0.9510876536369324, "learning_rate": 4.139653547544681e-06, "loss": 2.8057, "step": 273320 }, { "epoch": 0.0343808, "grad_norm": 0.7075402140617371, "learning_rate": 4.1390010606753535e-06, "loss": 2.9092, "step": 273330 }, { "epoch": 0.0344064, "grad_norm": 0.8091506361961365, "learning_rate": 4.1383486118130475e-06, "loss": 2.8904, "step": 273340 }, { "epoch": 0.034432, "grad_norm": 0.858709990978241, "learning_rate": 4.137696200961992e-06, "loss": 2.9388, "step": 273350 }, { "epoch": 0.0344576, "grad_norm": 0.9466115236282349, "learning_rate": 4.13704382812642e-06, "loss": 2.8416, "step": 273360 }, { "epoch": 0.0344832, "grad_norm": 0.8403854966163635, "learning_rate": 4.136391493310563e-06, "loss": 2.8454, "step": 273370 }, { "epoch": 0.0345088, "grad_norm": 0.8817979693412781, "learning_rate": 4.135739196518645e-06, "loss": 3.1356, "step": 273380 }, { "epoch": 0.0345344, "grad_norm": 0.8426916003227234, "learning_rate": 4.135086937754901e-06, "loss": 3.2771, "step": 273390 }, { "epoch": 0.03456, "grad_norm": 0.7664378881454468, "learning_rate": 4.134434717023559e-06, "loss": 2.9062, "step": 273400 }, { "epoch": 0.0345856, "grad_norm": 0.7470726370811462, "learning_rate": 4.133782534328848e-06, "loss": 2.8333, "step": 273410 }, { "epoch": 0.0346112, "grad_norm": 0.819907546043396, "learning_rate": 4.133130389675e-06, "loss": 2.8173, "step": 273420 }, { "epoch": 0.0346368, "grad_norm": 0.885006308555603, "learning_rate": 4.1324782830662404e-06, "loss": 2.7525, "step": 273430 }, { "epoch": 0.0346624, "grad_norm": 0.9587236642837524, "learning_rate": 4.1318262145068e-06, "loss": 2.8993, "step": 273440 }, { "epoch": 0.034688, "grad_norm": 1.628888487815857, "learning_rate": 4.131174184000908e-06, "loss": 2.8439, "step": 273450 }, { "epoch": 0.0347136, "grad_norm": 0.8095308542251587, "learning_rate": 4.130522191552792e-06, "loss": 2.912, "step": 273460 }, { "epoch": 0.0347392, "grad_norm": 0.7286016941070557, "learning_rate": 4.129870237166678e-06, "loss": 2.7371, "step": 273470 }, { "epoch": 0.0347648, "grad_norm": 0.8730716705322266, "learning_rate": 4.129218320846796e-06, "loss": 2.8348, "step": 273480 }, { "epoch": 0.0347904, "grad_norm": 0.8426666855812073, "learning_rate": 4.128566442597374e-06, "loss": 2.9061, "step": 273490 }, { "epoch": 0.034816, "grad_norm": 0.899684488773346, "learning_rate": 4.12791460242264e-06, "loss": 2.96, "step": 273500 }, { "epoch": 0.0348416, "grad_norm": 0.7294618487358093, "learning_rate": 4.127262800326817e-06, "loss": 2.8316, "step": 273510 }, { "epoch": 0.0348672, "grad_norm": 1.0057473182678223, "learning_rate": 4.126611036314133e-06, "loss": 2.8944, "step": 273520 }, { "epoch": 0.0348928, "grad_norm": 0.9073140621185303, "learning_rate": 4.125959310388816e-06, "loss": 2.9361, "step": 273530 }, { "epoch": 0.0349184, "grad_norm": 1.6518861055374146, "learning_rate": 4.125307622555091e-06, "loss": 2.9417, "step": 273540 }, { "epoch": 0.034944, "grad_norm": 0.9164655804634094, "learning_rate": 4.1246559728171865e-06, "loss": 2.9655, "step": 273550 }, { "epoch": 0.0349696, "grad_norm": 0.7740375399589539, "learning_rate": 4.124004361179326e-06, "loss": 2.6945, "step": 273560 }, { "epoch": 0.0349952, "grad_norm": 0.9210798144340515, "learning_rate": 4.123352787645738e-06, "loss": 3.258, "step": 273570 }, { "epoch": 0.0350208, "grad_norm": 0.8539106249809265, "learning_rate": 4.122701252220642e-06, "loss": 2.7733, "step": 273580 }, { "epoch": 0.0350464, "grad_norm": 0.8673703670501709, "learning_rate": 4.122049754908267e-06, "loss": 2.87, "step": 273590 }, { "epoch": 0.035072, "grad_norm": 0.9975975751876831, "learning_rate": 4.121398295712837e-06, "loss": 2.9398, "step": 273600 }, { "epoch": 0.0350976, "grad_norm": 0.9552422761917114, "learning_rate": 4.120746874638574e-06, "loss": 2.7714, "step": 273610 }, { "epoch": 0.0351232, "grad_norm": 0.829624593257904, "learning_rate": 4.120095491689707e-06, "loss": 3.0491, "step": 273620 }, { "epoch": 0.0351488, "grad_norm": 0.892141580581665, "learning_rate": 4.1194441468704625e-06, "loss": 2.6633, "step": 273630 }, { "epoch": 0.0351744, "grad_norm": 1.0368587970733643, "learning_rate": 4.118792840185056e-06, "loss": 2.8244, "step": 273640 }, { "epoch": 0.0352, "grad_norm": 0.8314436674118042, "learning_rate": 4.118141571637715e-06, "loss": 2.8914, "step": 273650 }, { "epoch": 0.0352256, "grad_norm": 0.8287574052810669, "learning_rate": 4.1174903412326626e-06, "loss": 2.8022, "step": 273660 }, { "epoch": 0.0352512, "grad_norm": 0.8071659207344055, "learning_rate": 4.116839148974122e-06, "loss": 2.9177, "step": 273670 }, { "epoch": 0.0352768, "grad_norm": 0.9109936952590942, "learning_rate": 4.1161879948663164e-06, "loss": 2.9386, "step": 273680 }, { "epoch": 0.0353024, "grad_norm": 1.3642559051513672, "learning_rate": 4.1155368789134666e-06, "loss": 3.0154, "step": 273690 }, { "epoch": 0.035328, "grad_norm": 0.8384470343589783, "learning_rate": 4.114885801119801e-06, "loss": 2.9043, "step": 273700 }, { "epoch": 0.0353536, "grad_norm": 0.9099428653717041, "learning_rate": 4.114234761489533e-06, "loss": 2.8876, "step": 273710 }, { "epoch": 0.0353792, "grad_norm": 1.1184732913970947, "learning_rate": 4.113583760026888e-06, "loss": 2.9123, "step": 273720 }, { "epoch": 0.0354048, "grad_norm": 0.8857976198196411, "learning_rate": 4.1129327967360886e-06, "loss": 2.7039, "step": 273730 }, { "epoch": 0.0354304, "grad_norm": 0.8447704911231995, "learning_rate": 4.112281871621355e-06, "loss": 2.6734, "step": 273740 }, { "epoch": 0.035456, "grad_norm": 0.7373231053352356, "learning_rate": 4.1116309846869085e-06, "loss": 2.8933, "step": 273750 }, { "epoch": 0.0354816, "grad_norm": 0.7548036575317383, "learning_rate": 4.110980135936971e-06, "loss": 2.794, "step": 273760 }, { "epoch": 0.0355072, "grad_norm": 0.9389304518699646, "learning_rate": 4.110329325375765e-06, "loss": 2.8319, "step": 273770 }, { "epoch": 0.0355328, "grad_norm": 0.867570161819458, "learning_rate": 4.1096785530075016e-06, "loss": 2.8384, "step": 273780 }, { "epoch": 0.0355584, "grad_norm": 0.805254340171814, "learning_rate": 4.109027818836409e-06, "loss": 2.8868, "step": 273790 }, { "epoch": 0.035584, "grad_norm": 1.0419775247573853, "learning_rate": 4.108377122866708e-06, "loss": 2.8899, "step": 273800 }, { "epoch": 0.0356096, "grad_norm": 0.7813316583633423, "learning_rate": 4.107726465102614e-06, "loss": 2.84, "step": 273810 }, { "epoch": 0.0356352, "grad_norm": 0.7841144800186157, "learning_rate": 4.107075845548348e-06, "loss": 2.6347, "step": 273820 }, { "epoch": 0.0356608, "grad_norm": 3.5548110008239746, "learning_rate": 4.1064252642081324e-06, "loss": 2.6415, "step": 273830 }, { "epoch": 0.0356864, "grad_norm": 0.9490878582000732, "learning_rate": 4.1057747210861795e-06, "loss": 2.9921, "step": 273840 }, { "epoch": 0.035712, "grad_norm": 0.7902402281761169, "learning_rate": 4.105124216186711e-06, "loss": 2.8036, "step": 273850 }, { "epoch": 0.0357376, "grad_norm": 0.7573291659355164, "learning_rate": 4.1044737495139455e-06, "loss": 2.7639, "step": 273860 }, { "epoch": 0.0357632, "grad_norm": 0.9465230703353882, "learning_rate": 4.1038233210721e-06, "loss": 3.0642, "step": 273870 }, { "epoch": 0.0357888, "grad_norm": 0.770544707775116, "learning_rate": 4.103172930865395e-06, "loss": 2.7082, "step": 273880 }, { "epoch": 0.0358144, "grad_norm": 0.8298226594924927, "learning_rate": 4.1025225788980434e-06, "loss": 2.8708, "step": 273890 }, { "epoch": 0.03584, "grad_norm": 0.8198435306549072, "learning_rate": 4.101872265174271e-06, "loss": 2.6618, "step": 273900 }, { "epoch": 0.0358656, "grad_norm": 0.9432844519615173, "learning_rate": 4.101221989698286e-06, "loss": 2.8004, "step": 273910 }, { "epoch": 0.0358912, "grad_norm": 0.7976024746894836, "learning_rate": 4.100571752474309e-06, "loss": 2.8573, "step": 273920 }, { "epoch": 0.0359168, "grad_norm": 0.7439562678337097, "learning_rate": 4.099921553506556e-06, "loss": 2.8947, "step": 273930 }, { "epoch": 0.0359424, "grad_norm": 0.8603683114051819, "learning_rate": 4.099271392799244e-06, "loss": 3.0003, "step": 273940 }, { "epoch": 0.035968, "grad_norm": 1.5432404279708862, "learning_rate": 4.098621270356588e-06, "loss": 2.8492, "step": 273950 }, { "epoch": 0.0359936, "grad_norm": 0.7841318249702454, "learning_rate": 4.097971186182804e-06, "loss": 2.9943, "step": 273960 }, { "epoch": 0.0360192, "grad_norm": 0.8826979398727417, "learning_rate": 4.097321140282109e-06, "loss": 2.7782, "step": 273970 }, { "epoch": 0.0360448, "grad_norm": 0.8086159825325012, "learning_rate": 4.096671132658718e-06, "loss": 2.8501, "step": 273980 }, { "epoch": 0.0360704, "grad_norm": 0.787869930267334, "learning_rate": 4.0960211633168454e-06, "loss": 2.7894, "step": 273990 }, { "epoch": 0.036096, "grad_norm": 0.9059522747993469, "learning_rate": 4.095371232260708e-06, "loss": 3.0353, "step": 274000 }, { "epoch": 0.0361216, "grad_norm": 0.8613285422325134, "learning_rate": 4.094721339494516e-06, "loss": 2.911, "step": 274010 }, { "epoch": 0.0361472, "grad_norm": 0.8220321536064148, "learning_rate": 4.09407148502249e-06, "loss": 2.858, "step": 274020 }, { "epoch": 0.0361728, "grad_norm": 0.8213481307029724, "learning_rate": 4.093421668848841e-06, "loss": 2.9327, "step": 274030 }, { "epoch": 0.0361984, "grad_norm": 0.8979451656341553, "learning_rate": 4.092771890977781e-06, "loss": 3.147, "step": 274040 }, { "epoch": 0.036224, "grad_norm": 0.7457184791564941, "learning_rate": 4.0921221514135255e-06, "loss": 2.9221, "step": 274050 }, { "epoch": 0.0362496, "grad_norm": 1.097783088684082, "learning_rate": 4.091472450160288e-06, "loss": 2.9981, "step": 274060 }, { "epoch": 0.0362752, "grad_norm": 0.9017205238342285, "learning_rate": 4.09082278722228e-06, "loss": 3.0259, "step": 274070 }, { "epoch": 0.0363008, "grad_norm": 0.8936958909034729, "learning_rate": 4.090173162603717e-06, "loss": 2.9311, "step": 274080 }, { "epoch": 0.0363264, "grad_norm": 1.3431291580200195, "learning_rate": 4.089523576308809e-06, "loss": 2.8962, "step": 274090 }, { "epoch": 0.036352, "grad_norm": 0.8084173202514648, "learning_rate": 4.088874028341774e-06, "loss": 3.0099, "step": 274100 }, { "epoch": 0.0363776, "grad_norm": 1.015114188194275, "learning_rate": 4.088224518706817e-06, "loss": 2.8852, "step": 274110 }, { "epoch": 0.0364032, "grad_norm": 0.8207240700721741, "learning_rate": 4.087575047408152e-06, "loss": 2.8522, "step": 274120 }, { "epoch": 0.0364288, "grad_norm": 0.8033041954040527, "learning_rate": 4.086925614449988e-06, "loss": 2.98, "step": 274130 }, { "epoch": 0.0364544, "grad_norm": 0.7785809636116028, "learning_rate": 4.0862762198365445e-06, "loss": 2.8051, "step": 274140 }, { "epoch": 0.03648, "grad_norm": 0.826079785823822, "learning_rate": 4.085626863572031e-06, "loss": 2.8504, "step": 274150 }, { "epoch": 0.0365056, "grad_norm": 0.905900239944458, "learning_rate": 4.084977545660651e-06, "loss": 2.8599, "step": 274160 }, { "epoch": 0.0365312, "grad_norm": 0.8419578671455383, "learning_rate": 4.084328266106621e-06, "loss": 3.0832, "step": 274170 }, { "epoch": 0.0365568, "grad_norm": 0.8943098187446594, "learning_rate": 4.083679024914149e-06, "loss": 2.726, "step": 274180 }, { "epoch": 0.0365824, "grad_norm": 0.8481705784797668, "learning_rate": 4.0830298220874465e-06, "loss": 2.9091, "step": 274190 }, { "epoch": 0.036608, "grad_norm": 0.8021446466445923, "learning_rate": 4.082380657630723e-06, "loss": 2.9682, "step": 274200 }, { "epoch": 0.0366336, "grad_norm": 0.9035523533821106, "learning_rate": 4.081731531548189e-06, "loss": 2.9742, "step": 274210 }, { "epoch": 0.0366592, "grad_norm": 0.9557926058769226, "learning_rate": 4.081082443844052e-06, "loss": 3.0839, "step": 274220 }, { "epoch": 0.0366848, "grad_norm": 0.8185689449310303, "learning_rate": 4.080433394522526e-06, "loss": 2.818, "step": 274230 }, { "epoch": 0.0367104, "grad_norm": 0.9377174377441406, "learning_rate": 4.079784383587813e-06, "loss": 2.9241, "step": 274240 }, { "epoch": 0.036736, "grad_norm": 0.7833027243614197, "learning_rate": 4.079135411044125e-06, "loss": 3.0559, "step": 274250 }, { "epoch": 0.0367616, "grad_norm": 0.7488181591033936, "learning_rate": 4.07848647689567e-06, "loss": 2.7404, "step": 274260 }, { "epoch": 0.0367872, "grad_norm": 0.7908713817596436, "learning_rate": 4.077837581146656e-06, "loss": 2.9551, "step": 274270 }, { "epoch": 0.0368128, "grad_norm": 0.8772180676460266, "learning_rate": 4.077188723801292e-06, "loss": 2.8496, "step": 274280 }, { "epoch": 0.0368384, "grad_norm": 0.8574503660202026, "learning_rate": 4.0765399048637845e-06, "loss": 2.9254, "step": 274290 }, { "epoch": 0.036864, "grad_norm": 0.7982151508331299, "learning_rate": 4.075891124338346e-06, "loss": 2.888, "step": 274300 }, { "epoch": 0.0368896, "grad_norm": 0.8195638060569763, "learning_rate": 4.075242382229171e-06, "loss": 2.7298, "step": 274310 }, { "epoch": 0.0369152, "grad_norm": 0.8195134401321411, "learning_rate": 4.074593678540478e-06, "loss": 2.8816, "step": 274320 }, { "epoch": 0.0369408, "grad_norm": 1.077857255935669, "learning_rate": 4.073945013276471e-06, "loss": 3.035, "step": 274330 }, { "epoch": 0.0369664, "grad_norm": 0.7184845805168152, "learning_rate": 4.073296386441356e-06, "loss": 2.926, "step": 274340 }, { "epoch": 0.036992, "grad_norm": 0.9235416650772095, "learning_rate": 4.072647798039341e-06, "loss": 2.8492, "step": 274350 }, { "epoch": 0.0370176, "grad_norm": 0.9116580486297607, "learning_rate": 4.071999248074627e-06, "loss": 2.8365, "step": 274360 }, { "epoch": 0.0370432, "grad_norm": 0.765673816204071, "learning_rate": 4.071350736551422e-06, "loss": 2.9448, "step": 274370 }, { "epoch": 0.0370688, "grad_norm": 0.7921739816665649, "learning_rate": 4.070702263473932e-06, "loss": 2.8538, "step": 274380 }, { "epoch": 0.0370944, "grad_norm": 0.7975950837135315, "learning_rate": 4.070053828846362e-06, "loss": 2.8775, "step": 274390 }, { "epoch": 0.03712, "grad_norm": 1.0372287034988403, "learning_rate": 4.0694054326729175e-06, "loss": 3.003, "step": 274400 }, { "epoch": 0.0371456, "grad_norm": 0.7718266844749451, "learning_rate": 4.068757074957802e-06, "loss": 2.7812, "step": 274410 }, { "epoch": 0.0371712, "grad_norm": 0.916907012462616, "learning_rate": 4.068108755705222e-06, "loss": 2.9286, "step": 274420 }, { "epoch": 0.0371968, "grad_norm": 0.8003231883049011, "learning_rate": 4.067460474919381e-06, "loss": 2.8443, "step": 274430 }, { "epoch": 0.0372224, "grad_norm": 1.0759648084640503, "learning_rate": 4.06681223260448e-06, "loss": 2.9467, "step": 274440 }, { "epoch": 0.037248, "grad_norm": 0.8940799832344055, "learning_rate": 4.0661640287647256e-06, "loss": 2.9624, "step": 274450 }, { "epoch": 0.0372736, "grad_norm": 0.7624768018722534, "learning_rate": 4.065515863404319e-06, "loss": 2.9073, "step": 274460 }, { "epoch": 0.0372992, "grad_norm": 0.7442333102226257, "learning_rate": 4.064867736527466e-06, "loss": 2.9528, "step": 274470 }, { "epoch": 0.0373248, "grad_norm": 0.9125546216964722, "learning_rate": 4.064219648138368e-06, "loss": 2.813, "step": 274480 }, { "epoch": 0.0373504, "grad_norm": 0.8692967891693115, "learning_rate": 4.063571598241227e-06, "loss": 2.9294, "step": 274490 }, { "epoch": 0.037376, "grad_norm": 0.9279845952987671, "learning_rate": 4.062923586840247e-06, "loss": 2.8996, "step": 274500 }, { "epoch": 0.0374016, "grad_norm": 0.9624271392822266, "learning_rate": 4.06227561393963e-06, "loss": 2.91, "step": 274510 }, { "epoch": 0.0374272, "grad_norm": 0.7972977757453918, "learning_rate": 4.061627679543579e-06, "loss": 2.8113, "step": 274520 }, { "epoch": 0.0374528, "grad_norm": 0.7571088671684265, "learning_rate": 4.060979783656293e-06, "loss": 2.7066, "step": 274530 }, { "epoch": 0.0374784, "grad_norm": 0.8327368497848511, "learning_rate": 4.060331926281974e-06, "loss": 2.8014, "step": 274540 }, { "epoch": 0.037504, "grad_norm": 0.8124174475669861, "learning_rate": 4.059684107424828e-06, "loss": 2.8245, "step": 274550 }, { "epoch": 0.0375296, "grad_norm": 0.9319597482681274, "learning_rate": 4.0590363270890475e-06, "loss": 2.8848, "step": 274560 }, { "epoch": 0.0375552, "grad_norm": 0.7816774845123291, "learning_rate": 4.0583885852788385e-06, "loss": 2.8212, "step": 274570 }, { "epoch": 0.0375808, "grad_norm": 0.8333502411842346, "learning_rate": 4.057740881998401e-06, "loss": 2.7629, "step": 274580 }, { "epoch": 0.0376064, "grad_norm": 0.9877787828445435, "learning_rate": 4.057093217251934e-06, "loss": 2.9089, "step": 274590 }, { "epoch": 0.037632, "grad_norm": 0.7724836468696594, "learning_rate": 4.056445591043637e-06, "loss": 2.9517, "step": 274600 }, { "epoch": 0.0376576, "grad_norm": 0.7834147214889526, "learning_rate": 4.055798003377712e-06, "loss": 2.5869, "step": 274610 }, { "epoch": 0.0376832, "grad_norm": 0.9426229596138, "learning_rate": 4.05515045425836e-06, "loss": 2.8813, "step": 274620 }, { "epoch": 0.0377088, "grad_norm": 0.8410841822624207, "learning_rate": 4.054502943689775e-06, "loss": 2.9783, "step": 274630 }, { "epoch": 0.0377344, "grad_norm": 0.7692101001739502, "learning_rate": 4.053855471676157e-06, "loss": 2.7915, "step": 274640 }, { "epoch": 0.03776, "grad_norm": 0.8180186748504639, "learning_rate": 4.053208038221703e-06, "loss": 2.8421, "step": 274650 }, { "epoch": 0.0377856, "grad_norm": 0.8201648592948914, "learning_rate": 4.052560643330618e-06, "loss": 3.0469, "step": 274660 }, { "epoch": 0.0378112, "grad_norm": 0.9480323195457458, "learning_rate": 4.051913287007095e-06, "loss": 2.6875, "step": 274670 }, { "epoch": 0.0378368, "grad_norm": 0.8841412663459778, "learning_rate": 4.051265969255339e-06, "loss": 2.8417, "step": 274680 }, { "epoch": 0.0378624, "grad_norm": 0.8361046314239502, "learning_rate": 4.0506186900795386e-06, "loss": 3.0576, "step": 274690 }, { "epoch": 0.037888, "grad_norm": 0.97096186876297, "learning_rate": 4.049971449483894e-06, "loss": 3.1428, "step": 274700 }, { "epoch": 0.0379136, "grad_norm": 0.9296486377716064, "learning_rate": 4.049324247472604e-06, "loss": 2.8101, "step": 274710 }, { "epoch": 0.0379392, "grad_norm": 0.7501494288444519, "learning_rate": 4.048677084049865e-06, "loss": 2.9111, "step": 274720 }, { "epoch": 0.0379648, "grad_norm": 0.7605747580528259, "learning_rate": 4.048029959219873e-06, "loss": 2.8134, "step": 274730 }, { "epoch": 0.0379904, "grad_norm": 0.7611253261566162, "learning_rate": 4.047382872986824e-06, "loss": 3.055, "step": 274740 }, { "epoch": 0.038016, "grad_norm": 0.7539830803871155, "learning_rate": 4.046735825354918e-06, "loss": 2.7118, "step": 274750 }, { "epoch": 0.0380416, "grad_norm": 0.8515286445617676, "learning_rate": 4.046088816328347e-06, "loss": 3.0036, "step": 274760 }, { "epoch": 0.0380672, "grad_norm": 0.9558116793632507, "learning_rate": 4.045441845911305e-06, "loss": 2.8606, "step": 274770 }, { "epoch": 0.0380928, "grad_norm": 0.8509762287139893, "learning_rate": 4.044794914107992e-06, "loss": 2.6248, "step": 274780 }, { "epoch": 0.0381184, "grad_norm": 1.0161901712417603, "learning_rate": 4.044148020922601e-06, "loss": 2.813, "step": 274790 }, { "epoch": 0.038144, "grad_norm": 0.8521443009376526, "learning_rate": 4.043501166359327e-06, "loss": 2.7989, "step": 274800 }, { "epoch": 0.0381696, "grad_norm": 0.7814096808433533, "learning_rate": 4.042854350422365e-06, "loss": 2.7812, "step": 274810 }, { "epoch": 0.0381952, "grad_norm": 0.8012984991073608, "learning_rate": 4.0422075731159125e-06, "loss": 2.5887, "step": 274820 }, { "epoch": 0.0382208, "grad_norm": 0.8576967120170593, "learning_rate": 4.041560834444154e-06, "loss": 2.8251, "step": 274830 }, { "epoch": 0.0382464, "grad_norm": 0.7757047414779663, "learning_rate": 4.0409141344112925e-06, "loss": 2.9404, "step": 274840 }, { "epoch": 0.038272, "grad_norm": 0.8123729825019836, "learning_rate": 4.04026747302152e-06, "loss": 2.8674, "step": 274850 }, { "epoch": 0.0382976, "grad_norm": 0.9671359062194824, "learning_rate": 4.039620850279028e-06, "loss": 2.8751, "step": 274860 }, { "epoch": 0.0383232, "grad_norm": 0.9416031837463379, "learning_rate": 4.038974266188012e-06, "loss": 2.8738, "step": 274870 }, { "epoch": 0.0383488, "grad_norm": 0.8182072639465332, "learning_rate": 4.038327720752666e-06, "loss": 2.8612, "step": 274880 }, { "epoch": 0.0383744, "grad_norm": 0.8276184797286987, "learning_rate": 4.0376812139771785e-06, "loss": 2.8897, "step": 274890 }, { "epoch": 0.0384, "grad_norm": 0.8073115944862366, "learning_rate": 4.037034745865742e-06, "loss": 2.8166, "step": 274900 }, { "epoch": 0.0384256, "grad_norm": 0.7366288304328918, "learning_rate": 4.03638831642255e-06, "loss": 2.6379, "step": 274910 }, { "epoch": 0.0384512, "grad_norm": 0.9087705016136169, "learning_rate": 4.035741925651796e-06, "loss": 2.9938, "step": 274920 }, { "epoch": 0.0384768, "grad_norm": 0.8195438385009766, "learning_rate": 4.035095573557672e-06, "loss": 2.8454, "step": 274930 }, { "epoch": 0.0385024, "grad_norm": 1.0031499862670898, "learning_rate": 4.034449260144365e-06, "loss": 2.7012, "step": 274940 }, { "epoch": 0.038528, "grad_norm": 0.9567277431488037, "learning_rate": 4.033802985416074e-06, "loss": 2.6974, "step": 274950 }, { "epoch": 0.0385536, "grad_norm": 0.8457300662994385, "learning_rate": 4.033156749376981e-06, "loss": 2.918, "step": 274960 }, { "epoch": 0.0385792, "grad_norm": 0.9057359099388123, "learning_rate": 4.032510552031281e-06, "loss": 2.887, "step": 274970 }, { "epoch": 0.0386048, "grad_norm": 0.8358859419822693, "learning_rate": 4.031864393383165e-06, "loss": 2.9469, "step": 274980 }, { "epoch": 0.0386304, "grad_norm": 0.8031188249588013, "learning_rate": 4.031218273436821e-06, "loss": 2.8464, "step": 274990 }, { "epoch": 0.038656, "grad_norm": 0.8501421213150024, "learning_rate": 4.030572192196437e-06, "loss": 2.5184, "step": 275000 }, { "epoch": 0.0386816, "grad_norm": 0.7367379069328308, "learning_rate": 4.029926149666213e-06, "loss": 2.6809, "step": 275010 }, { "epoch": 0.0387072, "grad_norm": 1.0471147298812866, "learning_rate": 4.029280145850327e-06, "loss": 2.5415, "step": 275020 }, { "epoch": 0.0387328, "grad_norm": 0.8165977597236633, "learning_rate": 4.028634180752974e-06, "loss": 2.5723, "step": 275030 }, { "epoch": 0.0387584, "grad_norm": 0.8635575175285339, "learning_rate": 4.02798825437834e-06, "loss": 2.3467, "step": 275040 }, { "epoch": 0.038784, "grad_norm": 0.7628697752952576, "learning_rate": 4.027342366730615e-06, "loss": 2.6663, "step": 275050 }, { "epoch": 0.0388096, "grad_norm": 0.9750018119812012, "learning_rate": 4.026696517813988e-06, "loss": 2.8491, "step": 275060 }, { "epoch": 0.0388352, "grad_norm": 0.9987844824790955, "learning_rate": 4.0260507076326475e-06, "loss": 2.4338, "step": 275070 }, { "epoch": 0.0388608, "grad_norm": 0.8950113654136658, "learning_rate": 4.025404936190782e-06, "loss": 2.666, "step": 275080 }, { "epoch": 0.0388864, "grad_norm": 0.7598345279693604, "learning_rate": 4.024759203492575e-06, "loss": 2.3033, "step": 275090 }, { "epoch": 0.038912, "grad_norm": 0.8434642553329468, "learning_rate": 4.024113509542217e-06, "loss": 2.8174, "step": 275100 }, { "epoch": 0.0389376, "grad_norm": 0.7983855605125427, "learning_rate": 4.023467854343895e-06, "loss": 2.65, "step": 275110 }, { "epoch": 0.0389632, "grad_norm": 0.8651891946792603, "learning_rate": 4.022822237901794e-06, "loss": 2.8209, "step": 275120 }, { "epoch": 0.0389888, "grad_norm": 0.8415257334709167, "learning_rate": 4.022176660220103e-06, "loss": 2.8199, "step": 275130 }, { "epoch": 0.0390144, "grad_norm": 0.8552854657173157, "learning_rate": 4.021531121303008e-06, "loss": 2.7874, "step": 275140 }, { "epoch": 0.03904, "grad_norm": 0.8638049364089966, "learning_rate": 4.020885621154698e-06, "loss": 2.6138, "step": 275150 }, { "epoch": 0.0390656, "grad_norm": 0.8139866590499878, "learning_rate": 4.020240159779353e-06, "loss": 2.6538, "step": 275160 }, { "epoch": 0.0390912, "grad_norm": 0.8046845197677612, "learning_rate": 4.0195947371811595e-06, "loss": 2.5842, "step": 275170 }, { "epoch": 0.0391168, "grad_norm": 0.8234555125236511, "learning_rate": 4.018949353364303e-06, "loss": 2.5804, "step": 275180 }, { "epoch": 0.0391424, "grad_norm": 0.8546410202980042, "learning_rate": 4.018304008332973e-06, "loss": 2.3913, "step": 275190 }, { "epoch": 0.039168, "grad_norm": 0.9247785806655884, "learning_rate": 4.017658702091352e-06, "loss": 2.5659, "step": 275200 }, { "epoch": 0.0391936, "grad_norm": 0.87806636095047, "learning_rate": 4.017013434643627e-06, "loss": 2.9166, "step": 275210 }, { "epoch": 0.0392192, "grad_norm": 0.7458896636962891, "learning_rate": 4.016368205993978e-06, "loss": 2.7006, "step": 275220 }, { "epoch": 0.0392448, "grad_norm": 0.8817204236984253, "learning_rate": 4.01572301614659e-06, "loss": 3.0556, "step": 275230 }, { "epoch": 0.0392704, "grad_norm": 0.8019108176231384, "learning_rate": 4.015077865105647e-06, "loss": 2.6903, "step": 275240 }, { "epoch": 0.039296, "grad_norm": 0.8824719190597534, "learning_rate": 4.014432752875334e-06, "loss": 2.7929, "step": 275250 }, { "epoch": 0.0393216, "grad_norm": 0.7674468755722046, "learning_rate": 4.013787679459835e-06, "loss": 2.6012, "step": 275260 }, { "epoch": 0.0393472, "grad_norm": 0.8220663070678711, "learning_rate": 4.013142644863331e-06, "loss": 2.7294, "step": 275270 }, { "epoch": 0.0393728, "grad_norm": 0.7583694458007812, "learning_rate": 4.0124976490900095e-06, "loss": 2.585, "step": 275280 }, { "epoch": 0.0393984, "grad_norm": 0.8156632781028748, "learning_rate": 4.011852692144046e-06, "loss": 2.8301, "step": 275290 }, { "epoch": 0.039424, "grad_norm": 0.7565398812294006, "learning_rate": 4.011207774029626e-06, "loss": 2.6824, "step": 275300 }, { "epoch": 0.0394496, "grad_norm": 0.7522592544555664, "learning_rate": 4.0105628947509325e-06, "loss": 2.5915, "step": 275310 }, { "epoch": 0.0394752, "grad_norm": 0.7727938294410706, "learning_rate": 4.009918054312146e-06, "loss": 2.2037, "step": 275320 }, { "epoch": 0.0395008, "grad_norm": 1.1253695487976074, "learning_rate": 4.0092732527174495e-06, "loss": 3.3006, "step": 275330 }, { "epoch": 0.0395264, "grad_norm": 1.344876766204834, "learning_rate": 4.008628489971024e-06, "loss": 2.7959, "step": 275340 }, { "epoch": 0.039552, "grad_norm": 0.8109654784202576, "learning_rate": 4.0079837660770505e-06, "loss": 2.4673, "step": 275350 }, { "epoch": 0.0395776, "grad_norm": 0.801170825958252, "learning_rate": 4.007339081039709e-06, "loss": 2.5428, "step": 275360 }, { "epoch": 0.0396032, "grad_norm": 1.0034679174423218, "learning_rate": 4.006694434863181e-06, "loss": 2.6711, "step": 275370 }, { "epoch": 0.0396288, "grad_norm": 0.9134883880615234, "learning_rate": 4.006049827551648e-06, "loss": 2.6304, "step": 275380 }, { "epoch": 0.0396544, "grad_norm": 0.846109926700592, "learning_rate": 4.005405259109287e-06, "loss": 2.9174, "step": 275390 }, { "epoch": 0.03968, "grad_norm": 0.8798345923423767, "learning_rate": 4.004760729540281e-06, "loss": 2.6147, "step": 275400 }, { "epoch": 0.0397056, "grad_norm": 0.7198163270950317, "learning_rate": 4.004116238848812e-06, "loss": 2.5688, "step": 275410 }, { "epoch": 0.0397312, "grad_norm": 0.8745529055595398, "learning_rate": 4.003471787039052e-06, "loss": 2.5191, "step": 275420 }, { "epoch": 0.0397568, "grad_norm": 0.8501038551330566, "learning_rate": 4.0028273741151825e-06, "loss": 3.0179, "step": 275430 }, { "epoch": 0.0397824, "grad_norm": 0.79860520362854, "learning_rate": 4.002183000081385e-06, "loss": 2.7743, "step": 275440 }, { "epoch": 0.039808, "grad_norm": 0.9057707786560059, "learning_rate": 4.0015386649418365e-06, "loss": 2.7977, "step": 275450 }, { "epoch": 0.0398336, "grad_norm": 0.9120529890060425, "learning_rate": 4.000894368700715e-06, "loss": 2.9561, "step": 275460 }, { "epoch": 0.0398592, "grad_norm": 0.787708580493927, "learning_rate": 4.000250111362199e-06, "loss": 2.7638, "step": 275470 }, { "epoch": 0.0398848, "grad_norm": 0.8079213500022888, "learning_rate": 3.999605892930471e-06, "loss": 2.6096, "step": 275480 }, { "epoch": 0.0399104, "grad_norm": 0.909640371799469, "learning_rate": 3.9989617134097e-06, "loss": 2.4794, "step": 275490 }, { "epoch": 0.039936, "grad_norm": 0.9515799880027771, "learning_rate": 3.998317572804068e-06, "loss": 2.8487, "step": 275500 }, { "epoch": 0.0399616, "grad_norm": 0.8339130878448486, "learning_rate": 3.997673471117751e-06, "loss": 2.4751, "step": 275510 }, { "epoch": 0.0399872, "grad_norm": 0.8515024185180664, "learning_rate": 3.997029408354922e-06, "loss": 3.0064, "step": 275520 }, { "epoch": 0.0400128, "grad_norm": 0.8107501268386841, "learning_rate": 3.996385384519766e-06, "loss": 2.8182, "step": 275530 }, { "epoch": 0.0400384, "grad_norm": 2.044398069381714, "learning_rate": 3.995741399616459e-06, "loss": 2.3605, "step": 275540 }, { "epoch": 0.040064, "grad_norm": 0.8932648301124573, "learning_rate": 3.99509745364917e-06, "loss": 2.5907, "step": 275550 }, { "epoch": 0.0400896, "grad_norm": 0.8873615860939026, "learning_rate": 3.994453546622078e-06, "loss": 2.825, "step": 275560 }, { "epoch": 0.0401152, "grad_norm": 0.8436872959136963, "learning_rate": 3.993809678539359e-06, "loss": 2.2467, "step": 275570 }, { "epoch": 0.0401408, "grad_norm": 0.8689242601394653, "learning_rate": 3.993165849405187e-06, "loss": 2.7428, "step": 275580 }, { "epoch": 0.0401664, "grad_norm": 0.8310116529464722, "learning_rate": 3.992522059223738e-06, "loss": 2.6159, "step": 275590 }, { "epoch": 0.040192, "grad_norm": 0.7734315395355225, "learning_rate": 3.991878307999187e-06, "loss": 2.9139, "step": 275600 }, { "epoch": 0.0402176, "grad_norm": 0.7967441082000732, "learning_rate": 3.991234595735713e-06, "loss": 2.816, "step": 275610 }, { "epoch": 0.0402432, "grad_norm": 3.3299920558929443, "learning_rate": 3.990590922437482e-06, "loss": 2.6341, "step": 275620 }, { "epoch": 0.0402688, "grad_norm": 1.0290535688400269, "learning_rate": 3.989947288108671e-06, "loss": 2.6656, "step": 275630 }, { "epoch": 0.0402944, "grad_norm": 0.8203932046890259, "learning_rate": 3.989303692753455e-06, "loss": 2.6142, "step": 275640 }, { "epoch": 0.04032, "grad_norm": 0.7386081218719482, "learning_rate": 3.988660136376005e-06, "loss": 2.5208, "step": 275650 }, { "epoch": 0.0403456, "grad_norm": 0.7350621819496155, "learning_rate": 3.988016618980499e-06, "loss": 2.6233, "step": 275660 }, { "epoch": 0.0403712, "grad_norm": 0.8597016334533691, "learning_rate": 3.9873731405711066e-06, "loss": 2.8741, "step": 275670 }, { "epoch": 0.0403968, "grad_norm": 0.9910752773284912, "learning_rate": 3.9867297011520055e-06, "loss": 2.4301, "step": 275680 }, { "epoch": 0.0404224, "grad_norm": 0.7730739116668701, "learning_rate": 3.98608630072736e-06, "loss": 2.6752, "step": 275690 }, { "epoch": 0.040448, "grad_norm": 0.739309549331665, "learning_rate": 3.985442939301343e-06, "loss": 2.4746, "step": 275700 }, { "epoch": 0.0404736, "grad_norm": 0.8673328757286072, "learning_rate": 3.984799616878133e-06, "loss": 2.7144, "step": 275710 }, { "epoch": 0.0404992, "grad_norm": 0.858634889125824, "learning_rate": 3.9841563334619e-06, "loss": 2.8129, "step": 275720 }, { "epoch": 0.0405248, "grad_norm": 0.7846753597259521, "learning_rate": 3.983513089056814e-06, "loss": 2.7949, "step": 275730 }, { "epoch": 0.0405504, "grad_norm": 0.764284610748291, "learning_rate": 3.98286988366705e-06, "loss": 2.7554, "step": 275740 }, { "epoch": 0.040576, "grad_norm": 0.7991724014282227, "learning_rate": 3.982226717296772e-06, "loss": 2.6522, "step": 275750 }, { "epoch": 0.0406016, "grad_norm": 1.1979808807373047, "learning_rate": 3.981583589950156e-06, "loss": 2.5442, "step": 275760 }, { "epoch": 0.0406272, "grad_norm": 0.7624194025993347, "learning_rate": 3.980940501631369e-06, "loss": 2.7375, "step": 275770 }, { "epoch": 0.0406528, "grad_norm": 0.7450792789459229, "learning_rate": 3.9802974523445835e-06, "loss": 2.6347, "step": 275780 }, { "epoch": 0.0406784, "grad_norm": 0.7637826204299927, "learning_rate": 3.979654442093969e-06, "loss": 2.6636, "step": 275790 }, { "epoch": 0.040704, "grad_norm": 0.8807222843170166, "learning_rate": 3.979011470883696e-06, "loss": 2.801, "step": 275800 }, { "epoch": 0.0407296, "grad_norm": 0.909952700138092, "learning_rate": 3.978368538717937e-06, "loss": 2.9027, "step": 275810 }, { "epoch": 0.0407552, "grad_norm": 0.8280407190322876, "learning_rate": 3.977725645600854e-06, "loss": 2.835, "step": 275820 }, { "epoch": 0.0407808, "grad_norm": 0.8239502310752869, "learning_rate": 3.9770827915366195e-06, "loss": 2.8299, "step": 275830 }, { "epoch": 0.0408064, "grad_norm": 0.8747493624687195, "learning_rate": 3.9764399765294014e-06, "loss": 2.9501, "step": 275840 }, { "epoch": 0.040832, "grad_norm": 0.9324370622634888, "learning_rate": 3.975797200583371e-06, "loss": 2.9017, "step": 275850 }, { "epoch": 0.0408576, "grad_norm": 0.8219139575958252, "learning_rate": 3.975154463702694e-06, "loss": 2.5552, "step": 275860 }, { "epoch": 0.0408832, "grad_norm": 0.9587632417678833, "learning_rate": 3.974511765891539e-06, "loss": 2.8812, "step": 275870 }, { "epoch": 0.0409088, "grad_norm": 0.8540681004524231, "learning_rate": 3.973869107154074e-06, "loss": 2.6496, "step": 275880 }, { "epoch": 0.0409344, "grad_norm": 0.8830975294113159, "learning_rate": 3.973226487494466e-06, "loss": 2.9301, "step": 275890 }, { "epoch": 0.04096, "grad_norm": 0.6970778703689575, "learning_rate": 3.972583906916882e-06, "loss": 2.6901, "step": 275900 }, { "epoch": 0.0409856, "grad_norm": 0.7429263591766357, "learning_rate": 3.97194136542549e-06, "loss": 2.6545, "step": 275910 }, { "epoch": 0.0410112, "grad_norm": 0.7802576422691345, "learning_rate": 3.971298863024457e-06, "loss": 2.6759, "step": 275920 }, { "epoch": 0.0410368, "grad_norm": 0.7755171656608582, "learning_rate": 3.970656399717952e-06, "loss": 2.7628, "step": 275930 }, { "epoch": 0.0410624, "grad_norm": 0.8205166459083557, "learning_rate": 3.970013975510133e-06, "loss": 2.9128, "step": 275940 }, { "epoch": 0.041088, "grad_norm": 0.8316431641578674, "learning_rate": 3.969371590405171e-06, "loss": 2.7669, "step": 275950 }, { "epoch": 0.0411136, "grad_norm": 0.872796893119812, "learning_rate": 3.968729244407233e-06, "loss": 2.8277, "step": 275960 }, { "epoch": 0.0411392, "grad_norm": 0.8514000773429871, "learning_rate": 3.968086937520481e-06, "loss": 2.7571, "step": 275970 }, { "epoch": 0.0411648, "grad_norm": 0.7998419404029846, "learning_rate": 3.967444669749083e-06, "loss": 2.8876, "step": 275980 }, { "epoch": 0.0411904, "grad_norm": 0.9346917271614075, "learning_rate": 3.9668024410972035e-06, "loss": 2.8471, "step": 275990 }, { "epoch": 0.041216, "grad_norm": 0.8203355073928833, "learning_rate": 3.966160251569006e-06, "loss": 2.7871, "step": 276000 }, { "epoch": 0.0412416, "grad_norm": 0.798287570476532, "learning_rate": 3.96551810116866e-06, "loss": 2.9261, "step": 276010 }, { "epoch": 0.0412672, "grad_norm": 0.8674581050872803, "learning_rate": 3.964875989900322e-06, "loss": 2.7917, "step": 276020 }, { "epoch": 0.0412928, "grad_norm": 0.8441579341888428, "learning_rate": 3.9642339177681574e-06, "loss": 2.9227, "step": 276030 }, { "epoch": 0.0413184, "grad_norm": 0.8298338055610657, "learning_rate": 3.963591884776333e-06, "loss": 2.9193, "step": 276040 }, { "epoch": 0.041344, "grad_norm": 0.9025363922119141, "learning_rate": 3.962949890929007e-06, "loss": 2.8801, "step": 276050 }, { "epoch": 0.0413696, "grad_norm": 0.7635959982872009, "learning_rate": 3.962307936230355e-06, "loss": 2.9278, "step": 276060 }, { "epoch": 0.0413952, "grad_norm": 0.7665325403213501, "learning_rate": 3.961666020684527e-06, "loss": 2.8103, "step": 276070 }, { "epoch": 0.0414208, "grad_norm": 0.8055325746536255, "learning_rate": 3.96102414429569e-06, "loss": 2.8604, "step": 276080 }, { "epoch": 0.0414464, "grad_norm": 0.7224733233451843, "learning_rate": 3.960382307068006e-06, "loss": 2.9166, "step": 276090 }, { "epoch": 0.041472, "grad_norm": 0.7652872204780579, "learning_rate": 3.9597405090056385e-06, "loss": 2.7089, "step": 276100 }, { "epoch": 0.0414976, "grad_norm": 0.8363157510757446, "learning_rate": 3.9590987501127475e-06, "loss": 2.7399, "step": 276110 }, { "epoch": 0.0415232, "grad_norm": 0.8654050827026367, "learning_rate": 3.958457030393496e-06, "loss": 3.1146, "step": 276120 }, { "epoch": 0.0415488, "grad_norm": 0.848377525806427, "learning_rate": 3.957815349852049e-06, "loss": 2.9302, "step": 276130 }, { "epoch": 0.0415744, "grad_norm": 0.8263378739356995, "learning_rate": 3.957173708492559e-06, "loss": 2.7773, "step": 276140 }, { "epoch": 0.0416, "grad_norm": 0.8524620532989502, "learning_rate": 3.956532106319192e-06, "loss": 2.8629, "step": 276150 }, { "epoch": 0.0416256, "grad_norm": 0.8272261619567871, "learning_rate": 3.955890543336108e-06, "loss": 2.7594, "step": 276160 }, { "epoch": 0.0416512, "grad_norm": 0.7921437621116638, "learning_rate": 3.9552490195474695e-06, "loss": 2.7849, "step": 276170 }, { "epoch": 0.0416768, "grad_norm": 0.896596372127533, "learning_rate": 3.954607534957432e-06, "loss": 3.0352, "step": 276180 }, { "epoch": 0.0417024, "grad_norm": 0.8965384364128113, "learning_rate": 3.953966089570159e-06, "loss": 2.7786, "step": 276190 }, { "epoch": 0.041728, "grad_norm": 0.739355206489563, "learning_rate": 3.9533246833898086e-06, "loss": 3.0189, "step": 276200 }, { "epoch": 0.0417536, "grad_norm": 0.8059762716293335, "learning_rate": 3.9526833164205446e-06, "loss": 2.772, "step": 276210 }, { "epoch": 0.0417792, "grad_norm": 0.7603344321250916, "learning_rate": 3.952041988666515e-06, "loss": 2.9153, "step": 276220 }, { "epoch": 0.0418048, "grad_norm": 0.842374861240387, "learning_rate": 3.95140070013189e-06, "loss": 2.6657, "step": 276230 }, { "epoch": 0.0418304, "grad_norm": 1.0154520273208618, "learning_rate": 3.950759450820823e-06, "loss": 3.0004, "step": 276240 }, { "epoch": 0.041856, "grad_norm": 1.547268033027649, "learning_rate": 3.950118240737472e-06, "loss": 2.8708, "step": 276250 }, { "epoch": 0.0418816, "grad_norm": 0.9899912476539612, "learning_rate": 3.949477069886002e-06, "loss": 2.7807, "step": 276260 }, { "epoch": 0.0419072, "grad_norm": 0.8841814398765564, "learning_rate": 3.948835938270561e-06, "loss": 2.8133, "step": 276270 }, { "epoch": 0.0419328, "grad_norm": 1.102465271949768, "learning_rate": 3.948194845895311e-06, "loss": 2.9186, "step": 276280 }, { "epoch": 0.0419584, "grad_norm": 0.7748079299926758, "learning_rate": 3.947553792764409e-06, "loss": 2.7568, "step": 276290 }, { "epoch": 0.041984, "grad_norm": 0.8078336715698242, "learning_rate": 3.946912778882012e-06, "loss": 2.9506, "step": 276300 }, { "epoch": 0.0420096, "grad_norm": 0.7944186925888062, "learning_rate": 3.9462718042522765e-06, "loss": 2.8181, "step": 276310 }, { "epoch": 0.0420352, "grad_norm": 0.8000310063362122, "learning_rate": 3.94563086887936e-06, "loss": 2.8451, "step": 276320 }, { "epoch": 0.0420608, "grad_norm": 0.8488495349884033, "learning_rate": 3.944989972767421e-06, "loss": 2.7922, "step": 276330 }, { "epoch": 0.0420864, "grad_norm": 0.7764689922332764, "learning_rate": 3.94434911592061e-06, "loss": 2.9969, "step": 276340 }, { "epoch": 0.042112, "grad_norm": 0.7380080819129944, "learning_rate": 3.943708298343085e-06, "loss": 2.7754, "step": 276350 }, { "epoch": 0.0421376, "grad_norm": 0.7461369037628174, "learning_rate": 3.943067520039002e-06, "loss": 2.7448, "step": 276360 }, { "epoch": 0.0421632, "grad_norm": 0.7430951595306396, "learning_rate": 3.942426781012517e-06, "loss": 2.6652, "step": 276370 }, { "epoch": 0.0421888, "grad_norm": 1.1010559797286987, "learning_rate": 3.941786081267784e-06, "loss": 3.0344, "step": 276380 }, { "epoch": 0.0422144, "grad_norm": 0.8134261965751648, "learning_rate": 3.941145420808959e-06, "loss": 2.7987, "step": 276390 }, { "epoch": 0.04224, "grad_norm": 1.2969671487808228, "learning_rate": 3.940504799640196e-06, "loss": 2.8952, "step": 276400 }, { "epoch": 0.0422656, "grad_norm": 0.835209846496582, "learning_rate": 3.939864217765648e-06, "loss": 2.9928, "step": 276410 }, { "epoch": 0.0422912, "grad_norm": 0.9074357748031616, "learning_rate": 3.939223675189469e-06, "loss": 3.0504, "step": 276420 }, { "epoch": 0.0423168, "grad_norm": 0.821867048740387, "learning_rate": 3.938583171915815e-06, "loss": 2.6233, "step": 276430 }, { "epoch": 0.0423424, "grad_norm": 0.8178092837333679, "learning_rate": 3.937942707948839e-06, "loss": 2.6784, "step": 276440 }, { "epoch": 0.042368, "grad_norm": 1.108669638633728, "learning_rate": 3.937302283292691e-06, "loss": 2.7991, "step": 276450 }, { "epoch": 0.0423936, "grad_norm": 1.0170286893844604, "learning_rate": 3.9366618979515315e-06, "loss": 2.9525, "step": 276460 }, { "epoch": 0.0424192, "grad_norm": 0.968953013420105, "learning_rate": 3.936021551929505e-06, "loss": 2.6857, "step": 276470 }, { "epoch": 0.0424448, "grad_norm": 0.9315524697303772, "learning_rate": 3.935381245230766e-06, "loss": 2.7971, "step": 276480 }, { "epoch": 0.0424704, "grad_norm": 0.8558468818664551, "learning_rate": 3.93474097785947e-06, "loss": 2.826, "step": 276490 }, { "epoch": 0.042496, "grad_norm": 1.2791416645050049, "learning_rate": 3.934100749819764e-06, "loss": 2.9578, "step": 276500 }, { "epoch": 0.0425216, "grad_norm": 0.8210378885269165, "learning_rate": 3.933460561115804e-06, "loss": 3.0664, "step": 276510 }, { "epoch": 0.0425472, "grad_norm": 0.9031515121459961, "learning_rate": 3.9328204117517395e-06, "loss": 2.9576, "step": 276520 }, { "epoch": 0.0425728, "grad_norm": 0.8630966544151306, "learning_rate": 3.932180301731726e-06, "loss": 2.7187, "step": 276530 }, { "epoch": 0.0425984, "grad_norm": 0.8982163071632385, "learning_rate": 3.931540231059907e-06, "loss": 2.8602, "step": 276540 }, { "epoch": 0.042624, "grad_norm": 0.7817816138267517, "learning_rate": 3.9309001997404376e-06, "loss": 2.8224, "step": 276550 }, { "epoch": 0.0426496, "grad_norm": 0.8284989595413208, "learning_rate": 3.930260207777466e-06, "loss": 2.7341, "step": 276560 }, { "epoch": 0.0426752, "grad_norm": 0.8182055950164795, "learning_rate": 3.929620255175141e-06, "loss": 2.8948, "step": 276570 }, { "epoch": 0.0427008, "grad_norm": 0.8458719253540039, "learning_rate": 3.928980341937619e-06, "loss": 2.7452, "step": 276580 }, { "epoch": 0.0427264, "grad_norm": 0.7627066969871521, "learning_rate": 3.928340468069048e-06, "loss": 2.6903, "step": 276590 }, { "epoch": 0.042752, "grad_norm": 1.0732815265655518, "learning_rate": 3.927700633573573e-06, "loss": 2.5702, "step": 276600 }, { "epoch": 0.0427776, "grad_norm": 0.8323730826377869, "learning_rate": 3.927060838455347e-06, "loss": 2.7963, "step": 276610 }, { "epoch": 0.0428032, "grad_norm": 1.4105442762374878, "learning_rate": 3.9264210827185154e-06, "loss": 2.8704, "step": 276620 }, { "epoch": 0.0428288, "grad_norm": 0.7603666186332703, "learning_rate": 3.92578136636723e-06, "loss": 3.0294, "step": 276630 }, { "epoch": 0.0428544, "grad_norm": 0.7768315672874451, "learning_rate": 3.925141689405638e-06, "loss": 2.9243, "step": 276640 }, { "epoch": 0.04288, "grad_norm": 0.8065463900566101, "learning_rate": 3.924502051837888e-06, "loss": 2.7169, "step": 276650 }, { "epoch": 0.0429056, "grad_norm": 0.7316620945930481, "learning_rate": 3.92386245366813e-06, "loss": 2.7949, "step": 276660 }, { "epoch": 0.0429312, "grad_norm": 0.8568074107170105, "learning_rate": 3.9232228949005055e-06, "loss": 2.8513, "step": 276670 }, { "epoch": 0.0429568, "grad_norm": 0.759859025478363, "learning_rate": 3.922583375539168e-06, "loss": 2.7574, "step": 276680 }, { "epoch": 0.0429824, "grad_norm": 0.9026045799255371, "learning_rate": 3.9219438955882595e-06, "loss": 2.6723, "step": 276690 }, { "epoch": 0.043008, "grad_norm": 0.7805119752883911, "learning_rate": 3.92130445505193e-06, "loss": 2.9586, "step": 276700 }, { "epoch": 0.0430336, "grad_norm": 1.1581830978393555, "learning_rate": 3.920665053934326e-06, "loss": 2.9722, "step": 276710 }, { "epoch": 0.0430592, "grad_norm": 0.7597025632858276, "learning_rate": 3.920025692239593e-06, "loss": 2.9137, "step": 276720 }, { "epoch": 0.0430848, "grad_norm": 0.8390324711799622, "learning_rate": 3.9193863699718815e-06, "loss": 2.9948, "step": 276730 }, { "epoch": 0.0431104, "grad_norm": 0.7409444451332092, "learning_rate": 3.918747087135326e-06, "loss": 2.8284, "step": 276740 }, { "epoch": 0.043136, "grad_norm": 0.7658196091651917, "learning_rate": 3.918107843734082e-06, "loss": 2.8116, "step": 276750 }, { "epoch": 0.0431616, "grad_norm": 0.9297093152999878, "learning_rate": 3.917468639772294e-06, "loss": 2.9564, "step": 276760 }, { "epoch": 0.0431872, "grad_norm": 0.8373451232910156, "learning_rate": 3.916829475254105e-06, "loss": 2.8036, "step": 276770 }, { "epoch": 0.0432128, "grad_norm": 0.808035671710968, "learning_rate": 3.9161903501836594e-06, "loss": 2.7577, "step": 276780 }, { "epoch": 0.0432384, "grad_norm": 0.7807614207267761, "learning_rate": 3.915551264565107e-06, "loss": 2.8923, "step": 276790 }, { "epoch": 0.043264, "grad_norm": 0.9973729848861694, "learning_rate": 3.914912218402583e-06, "loss": 2.905, "step": 276800 }, { "epoch": 0.0432896, "grad_norm": 1.4383682012557983, "learning_rate": 3.9142732117002365e-06, "loss": 3.1369, "step": 276810 }, { "epoch": 0.0433152, "grad_norm": 0.7787821292877197, "learning_rate": 3.913634244462211e-06, "loss": 3.1003, "step": 276820 }, { "epoch": 0.0433408, "grad_norm": 0.9071584343910217, "learning_rate": 3.91299531669265e-06, "loss": 3.1211, "step": 276830 }, { "epoch": 0.0433664, "grad_norm": 0.8863264918327332, "learning_rate": 3.9123564283956965e-06, "loss": 3.0066, "step": 276840 }, { "epoch": 0.043392, "grad_norm": 0.8002884984016418, "learning_rate": 3.911717579575493e-06, "loss": 3.009, "step": 276850 }, { "epoch": 0.0434176, "grad_norm": 0.754246711730957, "learning_rate": 3.911078770236187e-06, "loss": 2.8477, "step": 276860 }, { "epoch": 0.0434432, "grad_norm": 0.9102635979652405, "learning_rate": 3.9104400003819146e-06, "loss": 2.8899, "step": 276870 }, { "epoch": 0.0434688, "grad_norm": 0.8755835890769958, "learning_rate": 3.909801270016819e-06, "loss": 2.7759, "step": 276880 }, { "epoch": 0.0434944, "grad_norm": 0.9333806037902832, "learning_rate": 3.909162579145044e-06, "loss": 2.8428, "step": 276890 }, { "epoch": 0.04352, "grad_norm": 0.7689518928527832, "learning_rate": 3.90852392777073e-06, "loss": 2.9159, "step": 276900 }, { "epoch": 0.0435456, "grad_norm": 0.7585752010345459, "learning_rate": 3.90788531589802e-06, "loss": 2.7091, "step": 276910 }, { "epoch": 0.0435712, "grad_norm": 0.737254798412323, "learning_rate": 3.907246743531054e-06, "loss": 2.7316, "step": 276920 }, { "epoch": 0.0435968, "grad_norm": 0.7661866545677185, "learning_rate": 3.9066082106739756e-06, "loss": 3.0742, "step": 276930 }, { "epoch": 0.0436224, "grad_norm": 0.870448112487793, "learning_rate": 3.905969717330922e-06, "loss": 2.7928, "step": 276940 }, { "epoch": 0.043648, "grad_norm": 0.8410549759864807, "learning_rate": 3.905331263506036e-06, "loss": 2.8337, "step": 276950 }, { "epoch": 0.0436736, "grad_norm": 0.8616862297058105, "learning_rate": 3.9046928492034555e-06, "loss": 3.0638, "step": 276960 }, { "epoch": 0.0436992, "grad_norm": 0.7672930359840393, "learning_rate": 3.904054474427323e-06, "loss": 2.8057, "step": 276970 }, { "epoch": 0.0437248, "grad_norm": 0.8396576642990112, "learning_rate": 3.9034161391817775e-06, "loss": 2.8238, "step": 276980 }, { "epoch": 0.0437504, "grad_norm": 1.2619049549102783, "learning_rate": 3.902777843470961e-06, "loss": 2.9152, "step": 276990 }, { "epoch": 0.043776, "grad_norm": 0.7728023529052734, "learning_rate": 3.902139587299006e-06, "loss": 2.7455, "step": 277000 }, { "epoch": 0.0438016, "grad_norm": 0.7103636264801025, "learning_rate": 3.901501370670056e-06, "loss": 2.8743, "step": 277010 }, { "epoch": 0.0438272, "grad_norm": 0.8089278340339661, "learning_rate": 3.900863193588248e-06, "loss": 2.8779, "step": 277020 }, { "epoch": 0.0438528, "grad_norm": 0.8557242155075073, "learning_rate": 3.900225056057722e-06, "loss": 2.7606, "step": 277030 }, { "epoch": 0.0438784, "grad_norm": 0.7441329956054688, "learning_rate": 3.899586958082614e-06, "loss": 2.7214, "step": 277040 }, { "epoch": 0.043904, "grad_norm": 0.8823068737983704, "learning_rate": 3.898948899667064e-06, "loss": 2.8231, "step": 277050 }, { "epoch": 0.0439296, "grad_norm": 1.0327155590057373, "learning_rate": 3.8983108808152126e-06, "loss": 2.9493, "step": 277060 }, { "epoch": 0.0439552, "grad_norm": 0.8039857745170593, "learning_rate": 3.897672901531191e-06, "loss": 2.6999, "step": 277070 }, { "epoch": 0.0439808, "grad_norm": 0.7477666139602661, "learning_rate": 3.897034961819137e-06, "loss": 2.7446, "step": 277080 }, { "epoch": 0.0440064, "grad_norm": 0.9399629831314087, "learning_rate": 3.896397061683187e-06, "loss": 2.8219, "step": 277090 }, { "epoch": 0.044032, "grad_norm": 1.0358319282531738, "learning_rate": 3.8957592011274824e-06, "loss": 2.6827, "step": 277100 }, { "epoch": 0.0440576, "grad_norm": 0.8023210763931274, "learning_rate": 3.895121380156158e-06, "loss": 2.7765, "step": 277110 }, { "epoch": 0.0440832, "grad_norm": 0.8856915235519409, "learning_rate": 3.894483598773352e-06, "loss": 2.8137, "step": 277120 }, { "epoch": 0.0441088, "grad_norm": 0.7756412625312805, "learning_rate": 3.8938458569831946e-06, "loss": 2.7524, "step": 277130 }, { "epoch": 0.0441344, "grad_norm": 1.1706552505493164, "learning_rate": 3.893208154789823e-06, "loss": 2.7995, "step": 277140 }, { "epoch": 0.04416, "grad_norm": 1.065322756767273, "learning_rate": 3.892570492197374e-06, "loss": 2.7906, "step": 277150 }, { "epoch": 0.0441856, "grad_norm": 0.9104822874069214, "learning_rate": 3.8919328692099825e-06, "loss": 3.0395, "step": 277160 }, { "epoch": 0.0442112, "grad_norm": 1.0159168243408203, "learning_rate": 3.891295285831782e-06, "loss": 2.7015, "step": 277170 }, { "epoch": 0.0442368, "grad_norm": 0.7533564567565918, "learning_rate": 3.890657742066909e-06, "loss": 2.976, "step": 277180 }, { "epoch": 0.0442624, "grad_norm": 0.7997167110443115, "learning_rate": 3.8900202379195e-06, "loss": 2.806, "step": 277190 }, { "epoch": 0.044288, "grad_norm": 0.8192152976989746, "learning_rate": 3.889382773393682e-06, "loss": 2.8956, "step": 277200 }, { "epoch": 0.0443136, "grad_norm": 0.8703069686889648, "learning_rate": 3.8887453484935935e-06, "loss": 2.7613, "step": 277210 }, { "epoch": 0.0443392, "grad_norm": 0.7988027930259705, "learning_rate": 3.888107963223366e-06, "loss": 2.8236, "step": 277220 }, { "epoch": 0.0443648, "grad_norm": 0.7741541862487793, "learning_rate": 3.887470617587135e-06, "loss": 2.7048, "step": 277230 }, { "epoch": 0.0443904, "grad_norm": 0.7647241950035095, "learning_rate": 3.886833311589032e-06, "loss": 2.8393, "step": 277240 }, { "epoch": 0.044416, "grad_norm": 0.8868099451065063, "learning_rate": 3.886196045233192e-06, "loss": 2.8974, "step": 277250 }, { "epoch": 0.0444416, "grad_norm": 0.7735319137573242, "learning_rate": 3.885558818523747e-06, "loss": 2.5462, "step": 277260 }, { "epoch": 0.0444672, "grad_norm": 0.7825947999954224, "learning_rate": 3.884921631464822e-06, "loss": 2.8833, "step": 277270 }, { "epoch": 0.0444928, "grad_norm": 0.8117233514785767, "learning_rate": 3.884284484060559e-06, "loss": 2.8725, "step": 277280 }, { "epoch": 0.0445184, "grad_norm": 0.8794825673103333, "learning_rate": 3.883647376315085e-06, "loss": 2.7982, "step": 277290 }, { "epoch": 0.044544, "grad_norm": 0.8126317262649536, "learning_rate": 3.883010308232532e-06, "loss": 2.6141, "step": 277300 }, { "epoch": 0.0445696, "grad_norm": 0.8276943564414978, "learning_rate": 3.882373279817032e-06, "loss": 2.8626, "step": 277310 }, { "epoch": 0.0445952, "grad_norm": 0.7790105938911438, "learning_rate": 3.8817362910727195e-06, "loss": 2.8958, "step": 277320 }, { "epoch": 0.0446208, "grad_norm": 0.9073827266693115, "learning_rate": 3.881099342003717e-06, "loss": 2.7071, "step": 277330 }, { "epoch": 0.0446464, "grad_norm": 0.8161038756370544, "learning_rate": 3.8804624326141595e-06, "loss": 2.9502, "step": 277340 }, { "epoch": 0.044672, "grad_norm": 0.8560212254524231, "learning_rate": 3.879825562908176e-06, "loss": 2.4864, "step": 277350 }, { "epoch": 0.0446976, "grad_norm": 0.7402295470237732, "learning_rate": 3.879188732889897e-06, "loss": 2.5677, "step": 277360 }, { "epoch": 0.0447232, "grad_norm": 0.766404926776886, "learning_rate": 3.878551942563453e-06, "loss": 2.6528, "step": 277370 }, { "epoch": 0.0447488, "grad_norm": 0.8590070605278015, "learning_rate": 3.877915191932972e-06, "loss": 2.852, "step": 277380 }, { "epoch": 0.0447744, "grad_norm": 0.8059275150299072, "learning_rate": 3.877278481002589e-06, "loss": 2.86, "step": 277390 }, { "epoch": 0.0448, "grad_norm": 0.8905022144317627, "learning_rate": 3.876641809776423e-06, "loss": 2.8431, "step": 277400 }, { "epoch": 0.0448256, "grad_norm": 0.7875933647155762, "learning_rate": 3.876005178258608e-06, "loss": 2.8525, "step": 277410 }, { "epoch": 0.0448512, "grad_norm": 0.7562359571456909, "learning_rate": 3.875368586453272e-06, "loss": 2.8558, "step": 277420 }, { "epoch": 0.0448768, "grad_norm": 0.7011483907699585, "learning_rate": 3.874732034364542e-06, "loss": 2.7777, "step": 277430 }, { "epoch": 0.0449024, "grad_norm": 0.8107632398605347, "learning_rate": 3.874095521996548e-06, "loss": 2.8479, "step": 277440 }, { "epoch": 0.044928, "grad_norm": 0.7532062530517578, "learning_rate": 3.8734590493534154e-06, "loss": 2.802, "step": 277450 }, { "epoch": 0.0449536, "grad_norm": 0.8128983378410339, "learning_rate": 3.872822616439273e-06, "loss": 2.8415, "step": 277460 }, { "epoch": 0.0449792, "grad_norm": 0.9684584736824036, "learning_rate": 3.872186223258249e-06, "loss": 2.8476, "step": 277470 }, { "epoch": 0.0450048, "grad_norm": 1.8617528676986694, "learning_rate": 3.871549869814467e-06, "loss": 2.8485, "step": 277480 }, { "epoch": 0.0450304, "grad_norm": 0.7944595217704773, "learning_rate": 3.870913556112056e-06, "loss": 2.8011, "step": 277490 }, { "epoch": 0.045056, "grad_norm": 1.0910062789916992, "learning_rate": 3.870277282155141e-06, "loss": 2.7065, "step": 277500 }, { "epoch": 0.0450816, "grad_norm": 1.1424037218093872, "learning_rate": 3.86964104794785e-06, "loss": 2.8265, "step": 277510 }, { "epoch": 0.0451072, "grad_norm": 0.7785648703575134, "learning_rate": 3.869004853494309e-06, "loss": 2.8236, "step": 277520 }, { "epoch": 0.0451328, "grad_norm": 0.8536086678504944, "learning_rate": 3.86836869879864e-06, "loss": 2.8272, "step": 277530 }, { "epoch": 0.0451584, "grad_norm": 0.782331109046936, "learning_rate": 3.86773258386497e-06, "loss": 2.8742, "step": 277540 }, { "epoch": 0.045184, "grad_norm": 0.9487164616584778, "learning_rate": 3.867096508697425e-06, "loss": 2.7279, "step": 277550 }, { "epoch": 0.0452096, "grad_norm": 0.9537992477416992, "learning_rate": 3.86646047330013e-06, "loss": 2.7622, "step": 277560 }, { "epoch": 0.0452352, "grad_norm": 0.8435332179069519, "learning_rate": 3.865824477677207e-06, "loss": 2.8187, "step": 277570 }, { "epoch": 0.0452608, "grad_norm": 0.7824506163597107, "learning_rate": 3.865188521832783e-06, "loss": 2.8172, "step": 277580 }, { "epoch": 0.0452864, "grad_norm": 0.8799380660057068, "learning_rate": 3.864552605770984e-06, "loss": 2.7207, "step": 277590 }, { "epoch": 0.045312, "grad_norm": 0.9493439793586731, "learning_rate": 3.863916729495928e-06, "loss": 2.8935, "step": 277600 }, { "epoch": 0.0453376, "grad_norm": 0.8108438849449158, "learning_rate": 3.863280893011737e-06, "loss": 3.0876, "step": 277610 }, { "epoch": 0.0453632, "grad_norm": 0.7304884195327759, "learning_rate": 3.862645096322542e-06, "loss": 2.9356, "step": 277620 }, { "epoch": 0.0453888, "grad_norm": 0.8515063524246216, "learning_rate": 3.862009339432462e-06, "loss": 2.95, "step": 277630 }, { "epoch": 0.0454144, "grad_norm": 0.848662257194519, "learning_rate": 3.861373622345624e-06, "loss": 2.8849, "step": 277640 }, { "epoch": 0.04544, "grad_norm": 0.8776323795318604, "learning_rate": 3.860737945066144e-06, "loss": 2.8115, "step": 277650 }, { "epoch": 0.0454656, "grad_norm": 0.8888182640075684, "learning_rate": 3.8601023075981456e-06, "loss": 2.6915, "step": 277660 }, { "epoch": 0.0454912, "grad_norm": 0.7338719367980957, "learning_rate": 3.859466709945752e-06, "loss": 2.7033, "step": 277670 }, { "epoch": 0.0455168, "grad_norm": 0.7963383793830872, "learning_rate": 3.858831152113086e-06, "loss": 2.7163, "step": 277680 }, { "epoch": 0.0455424, "grad_norm": 0.7395588755607605, "learning_rate": 3.858195634104267e-06, "loss": 2.8086, "step": 277690 }, { "epoch": 0.045568, "grad_norm": 0.8297552466392517, "learning_rate": 3.857560155923418e-06, "loss": 2.6815, "step": 277700 }, { "epoch": 0.0455936, "grad_norm": 0.7423298954963684, "learning_rate": 3.856924717574657e-06, "loss": 2.8203, "step": 277710 }, { "epoch": 0.0456192, "grad_norm": 0.8143066763877869, "learning_rate": 3.856289319062112e-06, "loss": 2.7725, "step": 277720 }, { "epoch": 0.0456448, "grad_norm": 0.7706140279769897, "learning_rate": 3.8556539603898936e-06, "loss": 2.797, "step": 277730 }, { "epoch": 0.0456704, "grad_norm": 0.8523167967796326, "learning_rate": 3.855018641562126e-06, "loss": 2.8063, "step": 277740 }, { "epoch": 0.045696, "grad_norm": 1.0067362785339355, "learning_rate": 3.85438336258293e-06, "loss": 2.8328, "step": 277750 }, { "epoch": 0.0457216, "grad_norm": 0.9220812916755676, "learning_rate": 3.853748123456425e-06, "loss": 2.7138, "step": 277760 }, { "epoch": 0.0457472, "grad_norm": 0.8011825084686279, "learning_rate": 3.853112924186728e-06, "loss": 2.6379, "step": 277770 }, { "epoch": 0.0457728, "grad_norm": 0.8402513861656189, "learning_rate": 3.852477764777961e-06, "loss": 2.8606, "step": 277780 }, { "epoch": 0.0457984, "grad_norm": 0.7849716544151306, "learning_rate": 3.8518426452342424e-06, "loss": 2.9098, "step": 277790 }, { "epoch": 0.045824, "grad_norm": 0.9162849187850952, "learning_rate": 3.8512075655596905e-06, "loss": 2.8654, "step": 277800 }, { "epoch": 0.0458496, "grad_norm": 0.8086736798286438, "learning_rate": 3.8505725257584225e-06, "loss": 2.8271, "step": 277810 }, { "epoch": 0.0458752, "grad_norm": 0.8186283111572266, "learning_rate": 3.849937525834558e-06, "loss": 2.8776, "step": 277820 }, { "epoch": 0.0459008, "grad_norm": 0.7821094393730164, "learning_rate": 3.849302565792213e-06, "loss": 2.999, "step": 277830 }, { "epoch": 0.0459264, "grad_norm": 0.7596017718315125, "learning_rate": 3.848667645635511e-06, "loss": 2.9751, "step": 277840 }, { "epoch": 0.045952, "grad_norm": 0.8888336420059204, "learning_rate": 3.848032765368561e-06, "loss": 2.8643, "step": 277850 }, { "epoch": 0.0459776, "grad_norm": 0.9289155602455139, "learning_rate": 3.8473979249954826e-06, "loss": 2.9469, "step": 277860 }, { "epoch": 0.0460032, "grad_norm": 0.8834463357925415, "learning_rate": 3.846763124520394e-06, "loss": 2.8223, "step": 277870 }, { "epoch": 0.0460288, "grad_norm": 0.8387191891670227, "learning_rate": 3.846128363947411e-06, "loss": 2.8601, "step": 277880 }, { "epoch": 0.0460544, "grad_norm": 0.9398642778396606, "learning_rate": 3.8454936432806514e-06, "loss": 2.797, "step": 277890 }, { "epoch": 0.04608, "grad_norm": 2.1403744220733643, "learning_rate": 3.844858962524228e-06, "loss": 2.8459, "step": 277900 }, { "epoch": 0.0461056, "grad_norm": 0.8628337979316711, "learning_rate": 3.844224321682263e-06, "loss": 2.9049, "step": 277910 }, { "epoch": 0.0461312, "grad_norm": 1.0621607303619385, "learning_rate": 3.843589720758863e-06, "loss": 2.8039, "step": 277920 }, { "epoch": 0.0461568, "grad_norm": 0.7685983777046204, "learning_rate": 3.842955159758147e-06, "loss": 3.0551, "step": 277930 }, { "epoch": 0.0461824, "grad_norm": 0.7520484924316406, "learning_rate": 3.842320638684232e-06, "loss": 2.6751, "step": 277940 }, { "epoch": 0.046208, "grad_norm": 0.7879745364189148, "learning_rate": 3.84168615754123e-06, "loss": 2.7708, "step": 277950 }, { "epoch": 0.0462336, "grad_norm": 0.9176813960075378, "learning_rate": 3.841051716333254e-06, "loss": 2.9215, "step": 277960 }, { "epoch": 0.0462592, "grad_norm": 0.9278104305267334, "learning_rate": 3.840417315064428e-06, "loss": 3.0613, "step": 277970 }, { "epoch": 0.0462848, "grad_norm": 0.8854594230651855, "learning_rate": 3.839782953738853e-06, "loss": 2.9225, "step": 277980 }, { "epoch": 0.0463104, "grad_norm": 0.8773389458656311, "learning_rate": 3.839148632360651e-06, "loss": 2.8441, "step": 277990 }, { "epoch": 0.046336, "grad_norm": 0.8840153813362122, "learning_rate": 3.83851435093393e-06, "loss": 2.7635, "step": 278000 }, { "epoch": 0.0463616, "grad_norm": 0.8524991869926453, "learning_rate": 3.8378801094628085e-06, "loss": 3.0225, "step": 278010 }, { "epoch": 0.0463872, "grad_norm": 0.8338685631752014, "learning_rate": 3.837245907951396e-06, "loss": 2.8361, "step": 278020 }, { "epoch": 0.0464128, "grad_norm": 0.858820378780365, "learning_rate": 3.836611746403806e-06, "loss": 2.7992, "step": 278030 }, { "epoch": 0.0464384, "grad_norm": 0.7586305737495422, "learning_rate": 3.835977624824153e-06, "loss": 2.851, "step": 278040 }, { "epoch": 0.046464, "grad_norm": 0.8049401640892029, "learning_rate": 3.835343543216544e-06, "loss": 3.1131, "step": 278050 }, { "epoch": 0.0464896, "grad_norm": 0.8002474904060364, "learning_rate": 3.834709501585094e-06, "loss": 2.8452, "step": 278060 }, { "epoch": 0.0465152, "grad_norm": 0.840754508972168, "learning_rate": 3.834075499933916e-06, "loss": 3.0847, "step": 278070 }, { "epoch": 0.0465408, "grad_norm": 0.8008524179458618, "learning_rate": 3.833441538267118e-06, "loss": 2.8222, "step": 278080 }, { "epoch": 0.0465664, "grad_norm": 1.9698392152786255, "learning_rate": 3.832807616588813e-06, "loss": 3.0749, "step": 278090 }, { "epoch": 0.046592, "grad_norm": 0.7289434671401978, "learning_rate": 3.832173734903111e-06, "loss": 2.9218, "step": 278100 }, { "epoch": 0.0466176, "grad_norm": 0.7677487134933472, "learning_rate": 3.831539893214127e-06, "loss": 2.8889, "step": 278110 }, { "epoch": 0.0466432, "grad_norm": 0.7678927779197693, "learning_rate": 3.8309060915259645e-06, "loss": 2.7959, "step": 278120 }, { "epoch": 0.0466688, "grad_norm": 0.7572709321975708, "learning_rate": 3.830272329842735e-06, "loss": 2.8217, "step": 278130 }, { "epoch": 0.0466944, "grad_norm": 0.9491515159606934, "learning_rate": 3.829638608168548e-06, "loss": 3.0117, "step": 278140 }, { "epoch": 0.04672, "grad_norm": 0.8217942714691162, "learning_rate": 3.829004926507517e-06, "loss": 2.8162, "step": 278150 }, { "epoch": 0.0467456, "grad_norm": 0.9098007678985596, "learning_rate": 3.828371284863749e-06, "loss": 2.9099, "step": 278160 }, { "epoch": 0.0467712, "grad_norm": 0.8468597531318665, "learning_rate": 3.827737683241357e-06, "loss": 2.9262, "step": 278170 }, { "epoch": 0.0467968, "grad_norm": 0.8384430408477783, "learning_rate": 3.827104121644442e-06, "loss": 2.8628, "step": 278180 }, { "epoch": 0.0468224, "grad_norm": 1.026811122894287, "learning_rate": 3.826470600077117e-06, "loss": 2.8724, "step": 278190 }, { "epoch": 0.046848, "grad_norm": 0.8419800400733948, "learning_rate": 3.8258371185434886e-06, "loss": 2.7374, "step": 278200 }, { "epoch": 0.0468736, "grad_norm": 0.779054582118988, "learning_rate": 3.825203677047665e-06, "loss": 2.9051, "step": 278210 }, { "epoch": 0.0468992, "grad_norm": 0.7745800018310547, "learning_rate": 3.824570275593756e-06, "loss": 2.7845, "step": 278220 }, { "epoch": 0.0469248, "grad_norm": 0.7830432653427124, "learning_rate": 3.823936914185868e-06, "loss": 2.9028, "step": 278230 }, { "epoch": 0.0469504, "grad_norm": 0.7869971394538879, "learning_rate": 3.823303592828109e-06, "loss": 2.7604, "step": 278240 }, { "epoch": 0.046976, "grad_norm": 0.8732497096061707, "learning_rate": 3.822670311524582e-06, "loss": 2.7742, "step": 278250 }, { "epoch": 0.0470016, "grad_norm": 0.801040530204773, "learning_rate": 3.822037070279396e-06, "loss": 2.7293, "step": 278260 }, { "epoch": 0.0470272, "grad_norm": 1.0837918519973755, "learning_rate": 3.821403869096658e-06, "loss": 2.8873, "step": 278270 }, { "epoch": 0.0470528, "grad_norm": 0.8197852969169617, "learning_rate": 3.820770707980474e-06, "loss": 2.8317, "step": 278280 }, { "epoch": 0.0470784, "grad_norm": 0.8450965881347656, "learning_rate": 3.82013758693495e-06, "loss": 3.0889, "step": 278290 }, { "epoch": 0.047104, "grad_norm": 0.8226400017738342, "learning_rate": 3.819504505964191e-06, "loss": 2.6504, "step": 278300 }, { "epoch": 0.0471296, "grad_norm": 0.8029442429542542, "learning_rate": 3.818871465072303e-06, "loss": 2.9754, "step": 278310 }, { "epoch": 0.0471552, "grad_norm": 0.8653092384338379, "learning_rate": 3.818238464263391e-06, "loss": 3.0344, "step": 278320 }, { "epoch": 0.0471808, "grad_norm": 0.8074533343315125, "learning_rate": 3.817605503541558e-06, "loss": 2.8604, "step": 278330 }, { "epoch": 0.0472064, "grad_norm": 1.0609972476959229, "learning_rate": 3.816972582910912e-06, "loss": 2.8084, "step": 278340 }, { "epoch": 0.047232, "grad_norm": 0.7867618203163147, "learning_rate": 3.8163397023755545e-06, "loss": 2.8894, "step": 278350 }, { "epoch": 0.0472576, "grad_norm": 0.8033934831619263, "learning_rate": 3.815706861939591e-06, "loss": 2.8603, "step": 278360 }, { "epoch": 0.0472832, "grad_norm": 0.8676564693450928, "learning_rate": 3.8150740616071294e-06, "loss": 2.8828, "step": 278370 }, { "epoch": 0.0473088, "grad_norm": 0.7455241084098816, "learning_rate": 3.8144413013822635e-06, "loss": 2.8311, "step": 278380 }, { "epoch": 0.0473344, "grad_norm": 0.7843145132064819, "learning_rate": 3.8138085812691027e-06, "loss": 2.7919, "step": 278390 }, { "epoch": 0.04736, "grad_norm": 1.3068904876708984, "learning_rate": 3.8131759012717497e-06, "loss": 2.692, "step": 278400 }, { "epoch": 0.0473856, "grad_norm": 0.8415486216545105, "learning_rate": 3.8125432613943057e-06, "loss": 2.7549, "step": 278410 }, { "epoch": 0.0474112, "grad_norm": 0.7644022107124329, "learning_rate": 3.811910661640875e-06, "loss": 2.7899, "step": 278420 }, { "epoch": 0.0474368, "grad_norm": 0.7886379361152649, "learning_rate": 3.8112781020155585e-06, "loss": 2.7134, "step": 278430 }, { "epoch": 0.0474624, "grad_norm": 0.8464845418930054, "learning_rate": 3.810645582522462e-06, "loss": 2.646, "step": 278440 }, { "epoch": 0.047488, "grad_norm": 0.7721912264823914, "learning_rate": 3.810013103165682e-06, "loss": 2.6756, "step": 278450 }, { "epoch": 0.0475136, "grad_norm": 0.8614579439163208, "learning_rate": 3.8093806639493204e-06, "loss": 2.8738, "step": 278460 }, { "epoch": 0.0475392, "grad_norm": 0.7613319158554077, "learning_rate": 3.8087482648774798e-06, "loss": 2.7946, "step": 278470 }, { "epoch": 0.0475648, "grad_norm": 0.8137136101722717, "learning_rate": 3.8081159059542584e-06, "loss": 2.8967, "step": 278480 }, { "epoch": 0.0475904, "grad_norm": 0.885894775390625, "learning_rate": 3.8074835871837636e-06, "loss": 2.849, "step": 278490 }, { "epoch": 0.047616, "grad_norm": 1.0264692306518555, "learning_rate": 3.8068513085700954e-06, "loss": 2.9339, "step": 278500 }, { "epoch": 0.0476416, "grad_norm": 0.7553460597991943, "learning_rate": 3.8062190701173463e-06, "loss": 3.0274, "step": 278510 }, { "epoch": 0.0476672, "grad_norm": 0.8141540884971619, "learning_rate": 3.805586871829621e-06, "loss": 2.9512, "step": 278520 }, { "epoch": 0.0476928, "grad_norm": 0.8765444755554199, "learning_rate": 3.8049547137110176e-06, "loss": 2.8767, "step": 278530 }, { "epoch": 0.0477184, "grad_norm": 0.8035599589347839, "learning_rate": 3.804322595765637e-06, "loss": 2.9487, "step": 278540 }, { "epoch": 0.047744, "grad_norm": 0.7781983613967896, "learning_rate": 3.803690517997578e-06, "loss": 2.7734, "step": 278550 }, { "epoch": 0.0477696, "grad_norm": 0.8037853240966797, "learning_rate": 3.8030584804109395e-06, "loss": 2.7803, "step": 278560 }, { "epoch": 0.0477952, "grad_norm": 0.7728770971298218, "learning_rate": 3.8024264830098233e-06, "loss": 2.7336, "step": 278570 }, { "epoch": 0.0478208, "grad_norm": 0.8699479699134827, "learning_rate": 3.80179452579832e-06, "loss": 2.9822, "step": 278580 }, { "epoch": 0.0478464, "grad_norm": 0.8571211099624634, "learning_rate": 3.801162608780532e-06, "loss": 2.9075, "step": 278590 }, { "epoch": 0.047872, "grad_norm": 0.8351954221725464, "learning_rate": 3.800530731960558e-06, "loss": 2.9673, "step": 278600 }, { "epoch": 0.0478976, "grad_norm": 0.8309428691864014, "learning_rate": 3.7998988953424933e-06, "loss": 2.7364, "step": 278610 }, { "epoch": 0.0479232, "grad_norm": 0.8123878836631775, "learning_rate": 3.799267098930437e-06, "loss": 2.7689, "step": 278620 }, { "epoch": 0.0479488, "grad_norm": 0.8624955415725708, "learning_rate": 3.798635342728486e-06, "loss": 2.7359, "step": 278630 }, { "epoch": 0.0479744, "grad_norm": 1.2220426797866821, "learning_rate": 3.7980036267407393e-06, "loss": 2.7951, "step": 278640 }, { "epoch": 0.048, "grad_norm": 0.7597370147705078, "learning_rate": 3.797371950971288e-06, "loss": 2.7863, "step": 278650 }, { "epoch": 0.0480256, "grad_norm": 1.2018368244171143, "learning_rate": 3.796740315424228e-06, "loss": 2.7436, "step": 278660 }, { "epoch": 0.0480512, "grad_norm": 0.8882431983947754, "learning_rate": 3.796108720103662e-06, "loss": 2.6375, "step": 278670 }, { "epoch": 0.0480768, "grad_norm": 0.8807348012924194, "learning_rate": 3.7954771650136813e-06, "loss": 3.0435, "step": 278680 }, { "epoch": 0.0481024, "grad_norm": 0.8232803344726562, "learning_rate": 3.794845650158382e-06, "loss": 2.8812, "step": 278690 }, { "epoch": 0.048128, "grad_norm": 0.8781935572624207, "learning_rate": 3.794214175541864e-06, "loss": 2.8194, "step": 278700 }, { "epoch": 0.0481536, "grad_norm": 0.8663098216056824, "learning_rate": 3.7935827411682135e-06, "loss": 2.6744, "step": 278710 }, { "epoch": 0.0481792, "grad_norm": 1.1148535013198853, "learning_rate": 3.792951347041529e-06, "loss": 2.8298, "step": 278720 }, { "epoch": 0.0482048, "grad_norm": 0.8076782822608948, "learning_rate": 3.792319993165907e-06, "loss": 2.9756, "step": 278730 }, { "epoch": 0.0482304, "grad_norm": 0.9433309435844421, "learning_rate": 3.791688679545439e-06, "loss": 2.8918, "step": 278740 }, { "epoch": 0.048256, "grad_norm": 0.8373391628265381, "learning_rate": 3.79105740618422e-06, "loss": 2.8584, "step": 278750 }, { "epoch": 0.0482816, "grad_norm": 0.8043674230575562, "learning_rate": 3.7904261730863433e-06, "loss": 2.9367, "step": 278760 }, { "epoch": 0.0483072, "grad_norm": 0.7907927632331848, "learning_rate": 3.7897949802559066e-06, "loss": 2.761, "step": 278770 }, { "epoch": 0.0483328, "grad_norm": 0.9506669044494629, "learning_rate": 3.7891638276969957e-06, "loss": 2.9361, "step": 278780 }, { "epoch": 0.0483584, "grad_norm": 0.7901151180267334, "learning_rate": 3.7885327154137066e-06, "loss": 2.6405, "step": 278790 }, { "epoch": 0.048384, "grad_norm": 0.8285606503486633, "learning_rate": 3.787901643410131e-06, "loss": 3.0172, "step": 278800 }, { "epoch": 0.0484096, "grad_norm": 0.9170075058937073, "learning_rate": 3.7872706116903636e-06, "loss": 3.0209, "step": 278810 }, { "epoch": 0.0484352, "grad_norm": 0.7812553644180298, "learning_rate": 3.786639620258494e-06, "loss": 2.7152, "step": 278820 }, { "epoch": 0.0484608, "grad_norm": 0.8383054733276367, "learning_rate": 3.786008669118616e-06, "loss": 2.884, "step": 278830 }, { "epoch": 0.0484864, "grad_norm": 0.8880220055580139, "learning_rate": 3.7853777582748196e-06, "loss": 2.8019, "step": 278840 }, { "epoch": 0.048512, "grad_norm": 0.831481397151947, "learning_rate": 3.7847468877311977e-06, "loss": 2.852, "step": 278850 }, { "epoch": 0.0485376, "grad_norm": 0.7776331305503845, "learning_rate": 3.78411605749184e-06, "loss": 2.8928, "step": 278860 }, { "epoch": 0.0485632, "grad_norm": 0.9463975429534912, "learning_rate": 3.783485267560838e-06, "loss": 2.7333, "step": 278870 }, { "epoch": 0.0485888, "grad_norm": 0.8594571948051453, "learning_rate": 3.7828545179422817e-06, "loss": 2.8622, "step": 278880 }, { "epoch": 0.0486144, "grad_norm": 0.9931235313415527, "learning_rate": 3.7822238086402607e-06, "loss": 2.8993, "step": 278890 }, { "epoch": 0.04864, "grad_norm": 1.050861120223999, "learning_rate": 3.7815931396588702e-06, "loss": 2.7793, "step": 278900 }, { "epoch": 0.0486656, "grad_norm": 0.7936334609985352, "learning_rate": 3.780962511002192e-06, "loss": 2.9927, "step": 278910 }, { "epoch": 0.0486912, "grad_norm": 0.8128682374954224, "learning_rate": 3.7803319226743183e-06, "loss": 2.8269, "step": 278920 }, { "epoch": 0.0487168, "grad_norm": 0.8030577301979065, "learning_rate": 3.77970137467934e-06, "loss": 2.9388, "step": 278930 }, { "epoch": 0.0487424, "grad_norm": 0.9679458737373352, "learning_rate": 3.779070867021345e-06, "loss": 2.6487, "step": 278940 }, { "epoch": 0.048768, "grad_norm": 0.7546759843826294, "learning_rate": 3.7784403997044216e-06, "loss": 2.7543, "step": 278950 }, { "epoch": 0.0487936, "grad_norm": 0.8193309307098389, "learning_rate": 3.7778099727326588e-06, "loss": 2.9394, "step": 278960 }, { "epoch": 0.0488192, "grad_norm": 0.9032337069511414, "learning_rate": 3.777179586110149e-06, "loss": 3.1048, "step": 278970 }, { "epoch": 0.0488448, "grad_norm": 0.8221540451049805, "learning_rate": 3.776549239840971e-06, "loss": 2.8318, "step": 278980 }, { "epoch": 0.0488704, "grad_norm": 0.7604391574859619, "learning_rate": 3.775918933929218e-06, "loss": 2.7414, "step": 278990 }, { "epoch": 0.048896, "grad_norm": 1.8026881217956543, "learning_rate": 3.775288668378977e-06, "loss": 2.8036, "step": 279000 }, { "epoch": 0.0489216, "grad_norm": 0.8360291719436646, "learning_rate": 3.77465844319433e-06, "loss": 2.7339, "step": 279010 }, { "epoch": 0.0489472, "grad_norm": 0.8003894686698914, "learning_rate": 3.7740282583793763e-06, "loss": 2.8303, "step": 279020 }, { "epoch": 0.0489728, "grad_norm": 0.8895532488822937, "learning_rate": 3.77339811393819e-06, "loss": 2.9405, "step": 279030 }, { "epoch": 0.0489984, "grad_norm": 1.0841078758239746, "learning_rate": 3.7727680098748643e-06, "loss": 3.1208, "step": 279040 }, { "epoch": 0.049024, "grad_norm": 0.9296964406967163, "learning_rate": 3.7721379461934815e-06, "loss": 3.1481, "step": 279050 }, { "epoch": 0.0490496, "grad_norm": 0.9631197452545166, "learning_rate": 3.771507922898129e-06, "loss": 2.85, "step": 279060 }, { "epoch": 0.0490752, "grad_norm": 0.7842550277709961, "learning_rate": 3.770877939992893e-06, "loss": 2.8796, "step": 279070 }, { "epoch": 0.0491008, "grad_norm": 0.7586265206336975, "learning_rate": 3.7702479974818574e-06, "loss": 2.7675, "step": 279080 }, { "epoch": 0.0491264, "grad_norm": 1.05777907371521, "learning_rate": 3.7696180953691077e-06, "loss": 2.8201, "step": 279090 }, { "epoch": 0.049152, "grad_norm": 0.8393561244010925, "learning_rate": 3.7689882336587323e-06, "loss": 3.0327, "step": 279100 }, { "epoch": 0.0491776, "grad_norm": 0.9026904702186584, "learning_rate": 3.768358412354809e-06, "loss": 2.7538, "step": 279110 }, { "epoch": 0.0492032, "grad_norm": 0.8294845223426819, "learning_rate": 3.767728631461426e-06, "loss": 2.8944, "step": 279120 }, { "epoch": 0.0492288, "grad_norm": 1.0013529062271118, "learning_rate": 3.7670988909826657e-06, "loss": 2.8047, "step": 279130 }, { "epoch": 0.0492544, "grad_norm": 0.9724600315093994, "learning_rate": 3.7664691909226125e-06, "loss": 2.8577, "step": 279140 }, { "epoch": 0.04928, "grad_norm": 0.7985894680023193, "learning_rate": 3.76583953128535e-06, "loss": 2.7595, "step": 279150 }, { "epoch": 0.0493056, "grad_norm": 0.8039364814758301, "learning_rate": 3.7652099120749606e-06, "loss": 2.883, "step": 279160 }, { "epoch": 0.0493312, "grad_norm": 0.8175240755081177, "learning_rate": 3.7645803332955332e-06, "loss": 3.0155, "step": 279170 }, { "epoch": 0.0493568, "grad_norm": 0.8281113505363464, "learning_rate": 3.7639507949511366e-06, "loss": 2.8394, "step": 279180 }, { "epoch": 0.0493824, "grad_norm": 0.7840535640716553, "learning_rate": 3.7633212970458657e-06, "loss": 2.7477, "step": 279190 }, { "epoch": 0.049408, "grad_norm": 0.8459386825561523, "learning_rate": 3.7626918395837985e-06, "loss": 2.7243, "step": 279200 }, { "epoch": 0.0494336, "grad_norm": 0.7661861181259155, "learning_rate": 3.7620624225690183e-06, "loss": 2.7479, "step": 279210 }, { "epoch": 0.0494592, "grad_norm": 0.8099448084831238, "learning_rate": 3.761433046005607e-06, "loss": 2.8841, "step": 279220 }, { "epoch": 0.0494848, "grad_norm": 0.811100423336029, "learning_rate": 3.7608037098976426e-06, "loss": 2.8554, "step": 279230 }, { "epoch": 0.0495104, "grad_norm": 0.8278551697731018, "learning_rate": 3.760174414249207e-06, "loss": 2.7764, "step": 279240 }, { "epoch": 0.049536, "grad_norm": 0.8546517491340637, "learning_rate": 3.7595451590643815e-06, "loss": 2.9896, "step": 279250 }, { "epoch": 0.0495616, "grad_norm": 0.77596116065979, "learning_rate": 3.758915944347248e-06, "loss": 3.0258, "step": 279260 }, { "epoch": 0.0495872, "grad_norm": 0.8689912557601929, "learning_rate": 3.758286770101887e-06, "loss": 2.904, "step": 279270 }, { "epoch": 0.0496128, "grad_norm": 1.04209303855896, "learning_rate": 3.757657636332376e-06, "loss": 2.9994, "step": 279280 }, { "epoch": 0.0496384, "grad_norm": 0.7809316515922546, "learning_rate": 3.7570285430427956e-06, "loss": 2.7841, "step": 279290 }, { "epoch": 0.049664, "grad_norm": 0.8269070386886597, "learning_rate": 3.7563994902372304e-06, "loss": 2.9165, "step": 279300 }, { "epoch": 0.0496896, "grad_norm": 0.837556004524231, "learning_rate": 3.755770477919751e-06, "loss": 2.8377, "step": 279310 }, { "epoch": 0.0497152, "grad_norm": 1.115097165107727, "learning_rate": 3.755141506094442e-06, "loss": 2.8322, "step": 279320 }, { "epoch": 0.0497408, "grad_norm": 0.8222944736480713, "learning_rate": 3.754512574765379e-06, "loss": 2.7619, "step": 279330 }, { "epoch": 0.0497664, "grad_norm": 0.796183168888092, "learning_rate": 3.7538836839366422e-06, "loss": 2.9343, "step": 279340 }, { "epoch": 0.049792, "grad_norm": 1.0271865129470825, "learning_rate": 3.753254833612311e-06, "loss": 3.0332, "step": 279350 }, { "epoch": 0.0498176, "grad_norm": 0.8929396271705627, "learning_rate": 3.7526260237964608e-06, "loss": 2.9864, "step": 279360 }, { "epoch": 0.0498432, "grad_norm": 1.101130723953247, "learning_rate": 3.7519972544931716e-06, "loss": 2.9662, "step": 279370 }, { "epoch": 0.0498688, "grad_norm": 0.7385777831077576, "learning_rate": 3.751368525706518e-06, "loss": 2.8315, "step": 279380 }, { "epoch": 0.0498944, "grad_norm": 0.9036399722099304, "learning_rate": 3.7507398374405803e-06, "loss": 2.8178, "step": 279390 }, { "epoch": 0.04992, "grad_norm": 1.0478469133377075, "learning_rate": 3.7501111896994334e-06, "loss": 2.7211, "step": 279400 }, { "epoch": 0.0499456, "grad_norm": 1.6395111083984375, "learning_rate": 3.749482582487155e-06, "loss": 2.727, "step": 279410 }, { "epoch": 0.0499712, "grad_norm": 0.8800098299980164, "learning_rate": 3.7488540158078233e-06, "loss": 2.9616, "step": 279420 }, { "epoch": 0.0499968, "grad_norm": 0.832374095916748, "learning_rate": 3.748225489665509e-06, "loss": 3.0012, "step": 279430 }, { "epoch": 0.0500224, "grad_norm": 0.8549783825874329, "learning_rate": 3.7475970040642907e-06, "loss": 2.6408, "step": 279440 }, { "epoch": 0.050048, "grad_norm": 0.9708766341209412, "learning_rate": 3.746968559008245e-06, "loss": 2.9405, "step": 279450 }, { "epoch": 0.0500736, "grad_norm": 0.7804952263832092, "learning_rate": 3.7463401545014454e-06, "loss": 2.935, "step": 279460 }, { "epoch": 0.0500992, "grad_norm": 0.8516182899475098, "learning_rate": 3.745711790547968e-06, "loss": 2.6991, "step": 279470 }, { "epoch": 0.0501248, "grad_norm": 0.78197181224823, "learning_rate": 3.7450834671518876e-06, "loss": 2.6599, "step": 279480 }, { "epoch": 0.0501504, "grad_norm": 0.9432385563850403, "learning_rate": 3.744455184317278e-06, "loss": 2.9809, "step": 279490 }, { "epoch": 0.050176, "grad_norm": 0.8117458820343018, "learning_rate": 3.7438269420482188e-06, "loss": 2.8354, "step": 279500 }, { "epoch": 0.0502016, "grad_norm": 1.1446998119354248, "learning_rate": 3.743198740348775e-06, "loss": 2.8845, "step": 279510 }, { "epoch": 0.0502272, "grad_norm": 0.7702919840812683, "learning_rate": 3.742570579223024e-06, "loss": 2.8303, "step": 279520 }, { "epoch": 0.0502528, "grad_norm": 1.070642352104187, "learning_rate": 3.7419424586750363e-06, "loss": 2.7696, "step": 279530 }, { "epoch": 0.0502784, "grad_norm": 0.9236319661140442, "learning_rate": 3.7413143787088924e-06, "loss": 2.6179, "step": 279540 }, { "epoch": 0.050304, "grad_norm": 0.7309450507164001, "learning_rate": 3.740686339328665e-06, "loss": 2.8649, "step": 279550 }, { "epoch": 0.0503296, "grad_norm": 0.7910405397415161, "learning_rate": 3.7400583405384195e-06, "loss": 2.8158, "step": 279560 }, { "epoch": 0.0503552, "grad_norm": 1.1434941291809082, "learning_rate": 3.739430382342232e-06, "loss": 2.8979, "step": 279570 }, { "epoch": 0.0503808, "grad_norm": 0.7721723914146423, "learning_rate": 3.7388024647441745e-06, "loss": 2.7797, "step": 279580 }, { "epoch": 0.0504064, "grad_norm": 0.7930700778961182, "learning_rate": 3.73817458774832e-06, "loss": 2.8729, "step": 279590 }, { "epoch": 0.050432, "grad_norm": 0.8062036037445068, "learning_rate": 3.7375467513587384e-06, "loss": 2.6833, "step": 279600 }, { "epoch": 0.0504576, "grad_norm": 0.8653727769851685, "learning_rate": 3.7369189555795014e-06, "loss": 2.9623, "step": 279610 }, { "epoch": 0.0504832, "grad_norm": 0.8647180795669556, "learning_rate": 3.7362912004146836e-06, "loss": 2.8802, "step": 279620 }, { "epoch": 0.0505088, "grad_norm": 0.9226280450820923, "learning_rate": 3.7356634858683503e-06, "loss": 2.9039, "step": 279630 }, { "epoch": 0.0505344, "grad_norm": 1.1732721328735352, "learning_rate": 3.735035811944574e-06, "loss": 2.7848, "step": 279640 }, { "epoch": 0.05056, "grad_norm": 1.1165152788162231, "learning_rate": 3.734408178647425e-06, "loss": 2.7698, "step": 279650 }, { "epoch": 0.0505856, "grad_norm": 0.7414426803588867, "learning_rate": 3.7337805859809738e-06, "loss": 2.6775, "step": 279660 }, { "epoch": 0.0506112, "grad_norm": 0.7984407544136047, "learning_rate": 3.7331530339492905e-06, "loss": 2.7439, "step": 279670 }, { "epoch": 0.0506368, "grad_norm": 0.9530912041664124, "learning_rate": 3.7325255225564438e-06, "loss": 2.8718, "step": 279680 }, { "epoch": 0.0506624, "grad_norm": 0.7672803997993469, "learning_rate": 3.731898051806504e-06, "loss": 2.6406, "step": 279690 }, { "epoch": 0.050688, "grad_norm": 0.8254499435424805, "learning_rate": 3.73127062170354e-06, "loss": 2.7164, "step": 279700 }, { "epoch": 0.0507136, "grad_norm": 0.797993540763855, "learning_rate": 3.7306432322516184e-06, "loss": 2.7192, "step": 279710 }, { "epoch": 0.0507392, "grad_norm": 0.8267179131507874, "learning_rate": 3.7300158834548105e-06, "loss": 2.7082, "step": 279720 }, { "epoch": 0.0507648, "grad_norm": 1.0452934503555298, "learning_rate": 3.729388575317182e-06, "loss": 2.968, "step": 279730 }, { "epoch": 0.0507904, "grad_norm": 0.7974790334701538, "learning_rate": 3.728761307842803e-06, "loss": 2.8346, "step": 279740 }, { "epoch": 0.050816, "grad_norm": 0.7918692827224731, "learning_rate": 3.7281340810357437e-06, "loss": 2.8824, "step": 279750 }, { "epoch": 0.0508416, "grad_norm": 0.9372215867042542, "learning_rate": 3.727506894900065e-06, "loss": 2.7128, "step": 279760 }, { "epoch": 0.0508672, "grad_norm": 0.7641686797142029, "learning_rate": 3.726879749439836e-06, "loss": 2.8544, "step": 279770 }, { "epoch": 0.0508928, "grad_norm": 1.0567383766174316, "learning_rate": 3.726252644659125e-06, "loss": 2.8087, "step": 279780 }, { "epoch": 0.0509184, "grad_norm": 0.8989893794059753, "learning_rate": 3.725625580561999e-06, "loss": 2.9847, "step": 279790 }, { "epoch": 0.050944, "grad_norm": 1.0343977212905884, "learning_rate": 3.7249985571525228e-06, "loss": 2.9001, "step": 279800 }, { "epoch": 0.0509696, "grad_norm": 0.7568866610527039, "learning_rate": 3.7243715744347633e-06, "loss": 2.7571, "step": 279810 }, { "epoch": 0.0509952, "grad_norm": 0.865878701210022, "learning_rate": 3.723744632412789e-06, "loss": 2.9998, "step": 279820 }, { "epoch": 0.0510208, "grad_norm": 1.0619394779205322, "learning_rate": 3.723117731090661e-06, "loss": 3.0559, "step": 279830 }, { "epoch": 0.0510464, "grad_norm": 0.8922590017318726, "learning_rate": 3.722490870472445e-06, "loss": 3.0981, "step": 279840 }, { "epoch": 0.051072, "grad_norm": 1.0956975221633911, "learning_rate": 3.7218640505622072e-06, "loss": 2.9391, "step": 279850 }, { "epoch": 0.0510976, "grad_norm": 0.7587553858757019, "learning_rate": 3.721237271364013e-06, "loss": 2.736, "step": 279860 }, { "epoch": 0.0511232, "grad_norm": 0.8129920363426208, "learning_rate": 3.7206105328819274e-06, "loss": 2.8094, "step": 279870 }, { "epoch": 0.0511488, "grad_norm": 0.9301201105117798, "learning_rate": 3.7199838351200126e-06, "loss": 3.0021, "step": 279880 }, { "epoch": 0.0511744, "grad_norm": 0.8269095420837402, "learning_rate": 3.7193571780823335e-06, "loss": 2.9407, "step": 279890 }, { "epoch": 0.0512, "grad_norm": 0.9663445949554443, "learning_rate": 3.7187305617729543e-06, "loss": 2.839, "step": 279900 }, { "epoch": 0.0512256, "grad_norm": 0.7809277176856995, "learning_rate": 3.7181039861959377e-06, "loss": 2.7132, "step": 279910 }, { "epoch": 0.0512512, "grad_norm": 0.8182414770126343, "learning_rate": 3.717477451355347e-06, "loss": 2.8582, "step": 279920 }, { "epoch": 0.0512768, "grad_norm": 0.7953876256942749, "learning_rate": 3.716850957255246e-06, "loss": 2.9622, "step": 279930 }, { "epoch": 0.0513024, "grad_norm": 1.6814969778060913, "learning_rate": 3.7162245038996957e-06, "loss": 2.5895, "step": 279940 }, { "epoch": 0.051328, "grad_norm": 0.7497251033782959, "learning_rate": 3.715598091292764e-06, "loss": 2.7931, "step": 279950 }, { "epoch": 0.0513536, "grad_norm": 0.793358325958252, "learning_rate": 3.7149717194385047e-06, "loss": 2.7932, "step": 279960 }, { "epoch": 0.0513792, "grad_norm": 0.8737236857414246, "learning_rate": 3.714345388340983e-06, "loss": 2.7956, "step": 279970 }, { "epoch": 0.0514048, "grad_norm": 0.788611650466919, "learning_rate": 3.7137190980042616e-06, "loss": 3.0307, "step": 279980 }, { "epoch": 0.0514304, "grad_norm": 0.8429844975471497, "learning_rate": 3.7130928484324e-06, "loss": 2.9306, "step": 279990 }, { "epoch": 0.051456, "grad_norm": 0.7538207173347473, "learning_rate": 3.7124666396294618e-06, "loss": 2.8775, "step": 280000 }, { "epoch": 0.0514816, "grad_norm": 0.7907274961471558, "learning_rate": 3.711840471599506e-06, "loss": 2.8232, "step": 280010 }, { "epoch": 0.0515072, "grad_norm": 0.837586522102356, "learning_rate": 3.7112143443465966e-06, "loss": 2.6555, "step": 280020 }, { "epoch": 0.0515328, "grad_norm": 0.8412137031555176, "learning_rate": 3.7105882578747874e-06, "loss": 2.8413, "step": 280030 }, { "epoch": 0.0515584, "grad_norm": 1.070715308189392, "learning_rate": 3.7099622121881427e-06, "loss": 2.6337, "step": 280040 }, { "epoch": 0.051584, "grad_norm": 0.8275434970855713, "learning_rate": 3.709336207290717e-06, "loss": 2.7891, "step": 280050 }, { "epoch": 0.0516096, "grad_norm": 0.8139044046401978, "learning_rate": 3.7087102431865786e-06, "loss": 2.829, "step": 280060 }, { "epoch": 0.0516352, "grad_norm": 0.7834919691085815, "learning_rate": 3.708084319879781e-06, "loss": 3.0324, "step": 280070 }, { "epoch": 0.0516608, "grad_norm": 0.9230082035064697, "learning_rate": 3.707458437374388e-06, "loss": 3.0325, "step": 280080 }, { "epoch": 0.0516864, "grad_norm": 0.8334706425666809, "learning_rate": 3.706832595674452e-06, "loss": 2.6636, "step": 280090 }, { "epoch": 0.051712, "grad_norm": 0.7940629124641418, "learning_rate": 3.706206794784034e-06, "loss": 2.7062, "step": 280100 }, { "epoch": 0.0517376, "grad_norm": 0.8392516374588013, "learning_rate": 3.705581034707192e-06, "loss": 2.7211, "step": 280110 }, { "epoch": 0.0517632, "grad_norm": 0.8171446323394775, "learning_rate": 3.7049553154479833e-06, "loss": 2.8252, "step": 280120 }, { "epoch": 0.0517888, "grad_norm": 0.9102767109870911, "learning_rate": 3.7043296370104677e-06, "loss": 2.8373, "step": 280130 }, { "epoch": 0.0518144, "grad_norm": 0.7601456046104431, "learning_rate": 3.7037039993987e-06, "loss": 2.9196, "step": 280140 }, { "epoch": 0.05184, "grad_norm": 0.818381667137146, "learning_rate": 3.7030784026167433e-06, "loss": 2.7446, "step": 280150 }, { "epoch": 0.0518656, "grad_norm": 0.8102425336837769, "learning_rate": 3.702452846668645e-06, "loss": 2.8005, "step": 280160 }, { "epoch": 0.0518912, "grad_norm": 0.8184072375297546, "learning_rate": 3.701827331558466e-06, "loss": 2.6373, "step": 280170 }, { "epoch": 0.0519168, "grad_norm": 0.7829461097717285, "learning_rate": 3.7012018572902643e-06, "loss": 2.5244, "step": 280180 }, { "epoch": 0.0519424, "grad_norm": 0.8332270979881287, "learning_rate": 3.7005764238680943e-06, "loss": 2.8976, "step": 280190 }, { "epoch": 0.051968, "grad_norm": 0.7538360953330994, "learning_rate": 3.699951031296012e-06, "loss": 2.8756, "step": 280200 }, { "epoch": 0.0519936, "grad_norm": 0.8515781760215759, "learning_rate": 3.699325679578072e-06, "loss": 2.7293, "step": 280210 }, { "epoch": 0.0520192, "grad_norm": 1.1156624555587769, "learning_rate": 3.698700368718331e-06, "loss": 2.7694, "step": 280220 }, { "epoch": 0.0520448, "grad_norm": 1.071297526359558, "learning_rate": 3.6980750987208436e-06, "loss": 3.0589, "step": 280230 }, { "epoch": 0.0520704, "grad_norm": 0.7768795490264893, "learning_rate": 3.6974498695896644e-06, "loss": 2.7642, "step": 280240 }, { "epoch": 0.052096, "grad_norm": 0.7657027840614319, "learning_rate": 3.6968246813288465e-06, "loss": 2.865, "step": 280250 }, { "epoch": 0.0521216, "grad_norm": 0.8363035321235657, "learning_rate": 3.696199533942447e-06, "loss": 2.9095, "step": 280260 }, { "epoch": 0.0521472, "grad_norm": 0.8458229303359985, "learning_rate": 3.6955744274345172e-06, "loss": 2.9483, "step": 280270 }, { "epoch": 0.0521728, "grad_norm": 1.4566333293914795, "learning_rate": 3.6949493618091157e-06, "loss": 3.1637, "step": 280280 }, { "epoch": 0.0521984, "grad_norm": 0.8183542490005493, "learning_rate": 3.6943243370702885e-06, "loss": 2.8403, "step": 280290 }, { "epoch": 0.052224, "grad_norm": 0.7925370335578918, "learning_rate": 3.693699353222091e-06, "loss": 2.7393, "step": 280300 }, { "epoch": 0.0522496, "grad_norm": 1.9028594493865967, "learning_rate": 3.693074410268579e-06, "loss": 2.839, "step": 280310 }, { "epoch": 0.0522752, "grad_norm": 1.026271104812622, "learning_rate": 3.692449508213802e-06, "loss": 2.9772, "step": 280320 }, { "epoch": 0.0523008, "grad_norm": 0.8428704142570496, "learning_rate": 3.6918246470618136e-06, "loss": 2.8184, "step": 280330 }, { "epoch": 0.0523264, "grad_norm": 0.9289534687995911, "learning_rate": 3.691199826816666e-06, "loss": 2.809, "step": 280340 }, { "epoch": 0.052352, "grad_norm": 0.833239734172821, "learning_rate": 3.6905750474824154e-06, "loss": 2.8708, "step": 280350 }, { "epoch": 0.0523776, "grad_norm": 0.9645955562591553, "learning_rate": 3.689950309063104e-06, "loss": 2.7449, "step": 280360 }, { "epoch": 0.0524032, "grad_norm": 0.878304660320282, "learning_rate": 3.6893256115627884e-06, "loss": 2.8238, "step": 280370 }, { "epoch": 0.0524288, "grad_norm": 0.7924956679344177, "learning_rate": 3.6887009549855192e-06, "loss": 2.8172, "step": 280380 }, { "epoch": 0.0524544, "grad_norm": 0.9056559801101685, "learning_rate": 3.688076339335346e-06, "loss": 3.0542, "step": 280390 }, { "epoch": 0.05248, "grad_norm": 1.2874388694763184, "learning_rate": 3.687451764616318e-06, "loss": 2.8083, "step": 280400 }, { "epoch": 0.0525056, "grad_norm": 0.794967532157898, "learning_rate": 3.6868272308324935e-06, "loss": 2.7641, "step": 280410 }, { "epoch": 0.0525312, "grad_norm": 0.8217763304710388, "learning_rate": 3.686202737987914e-06, "loss": 2.9174, "step": 280420 }, { "epoch": 0.0525568, "grad_norm": 0.7291660308837891, "learning_rate": 3.685578286086631e-06, "loss": 2.6389, "step": 280430 }, { "epoch": 0.0525824, "grad_norm": 0.808752715587616, "learning_rate": 3.6849538751326953e-06, "loss": 2.8805, "step": 280440 }, { "epoch": 0.052608, "grad_norm": 1.0556718111038208, "learning_rate": 3.684329505130154e-06, "loss": 2.979, "step": 280450 }, { "epoch": 0.0526336, "grad_norm": 0.7600454092025757, "learning_rate": 3.683705176083059e-06, "loss": 2.809, "step": 280460 }, { "epoch": 0.0526592, "grad_norm": 0.7872987389564514, "learning_rate": 3.683080887995457e-06, "loss": 2.7869, "step": 280470 }, { "epoch": 0.0526848, "grad_norm": 0.7743575572967529, "learning_rate": 3.6824566408713993e-06, "loss": 2.9248, "step": 280480 }, { "epoch": 0.0527104, "grad_norm": 0.795922577381134, "learning_rate": 3.681832434714928e-06, "loss": 2.6714, "step": 280490 }, { "epoch": 0.052736, "grad_norm": 0.8830820322036743, "learning_rate": 3.6812082695300942e-06, "loss": 2.7357, "step": 280500 }, { "epoch": 0.0527616, "grad_norm": 0.8285354375839233, "learning_rate": 3.6805841453209467e-06, "loss": 2.9231, "step": 280510 }, { "epoch": 0.0527872, "grad_norm": 0.8518951535224915, "learning_rate": 3.67996006209153e-06, "loss": 2.4098, "step": 280520 }, { "epoch": 0.0528128, "grad_norm": 0.741959810256958, "learning_rate": 3.679336019845893e-06, "loss": 2.8796, "step": 280530 }, { "epoch": 0.0528384, "grad_norm": 0.7890147566795349, "learning_rate": 3.678712018588082e-06, "loss": 2.7255, "step": 280540 }, { "epoch": 0.052864, "grad_norm": 0.789514422416687, "learning_rate": 3.678088058322148e-06, "loss": 2.7234, "step": 280550 }, { "epoch": 0.0528896, "grad_norm": 0.8627002239227295, "learning_rate": 3.677464139052128e-06, "loss": 3.0197, "step": 280560 }, { "epoch": 0.0529152, "grad_norm": 0.8908616900444031, "learning_rate": 3.6768402607820698e-06, "loss": 2.8373, "step": 280570 }, { "epoch": 0.0529408, "grad_norm": 0.8507333397865295, "learning_rate": 3.6762164235160256e-06, "loss": 2.8089, "step": 280580 }, { "epoch": 0.0529664, "grad_norm": 0.7297568917274475, "learning_rate": 3.6755926272580368e-06, "loss": 2.8912, "step": 280590 }, { "epoch": 0.052992, "grad_norm": 0.7463636994361877, "learning_rate": 3.67496887201215e-06, "loss": 3.0062, "step": 280600 }, { "epoch": 0.0530176, "grad_norm": 1.086410403251648, "learning_rate": 3.674345157782412e-06, "loss": 2.6783, "step": 280610 }, { "epoch": 0.0530432, "grad_norm": 0.7837464809417725, "learning_rate": 3.6737214845728608e-06, "loss": 2.8374, "step": 280620 }, { "epoch": 0.0530688, "grad_norm": 0.8401869535446167, "learning_rate": 3.6730978523875437e-06, "loss": 2.9162, "step": 280630 }, { "epoch": 0.0530944, "grad_norm": 0.7533463835716248, "learning_rate": 3.6724742612305066e-06, "loss": 2.8037, "step": 280640 }, { "epoch": 0.05312, "grad_norm": 0.844382643699646, "learning_rate": 3.6718507111057923e-06, "loss": 2.8113, "step": 280650 }, { "epoch": 0.0531456, "grad_norm": 0.7183166742324829, "learning_rate": 3.6712272020174445e-06, "loss": 2.7068, "step": 280660 }, { "epoch": 0.0531712, "grad_norm": 0.9783536791801453, "learning_rate": 3.6706037339695054e-06, "loss": 2.8968, "step": 280670 }, { "epoch": 0.0531968, "grad_norm": 1.2679959535598755, "learning_rate": 3.6699803069660224e-06, "loss": 2.9336, "step": 280680 }, { "epoch": 0.0532224, "grad_norm": 0.7760734558105469, "learning_rate": 3.6693569210110325e-06, "loss": 2.7437, "step": 280690 }, { "epoch": 0.053248, "grad_norm": 0.8818215727806091, "learning_rate": 3.6687335761085794e-06, "loss": 2.7748, "step": 280700 }, { "epoch": 0.0532736, "grad_norm": 0.9415162801742554, "learning_rate": 3.668110272262707e-06, "loss": 2.8238, "step": 280710 }, { "epoch": 0.0532992, "grad_norm": 1.0152589082717896, "learning_rate": 3.667487009477456e-06, "loss": 2.726, "step": 280720 }, { "epoch": 0.0533248, "grad_norm": 0.7530731558799744, "learning_rate": 3.6668637877568693e-06, "loss": 2.9278, "step": 280730 }, { "epoch": 0.0533504, "grad_norm": 0.7456937432289124, "learning_rate": 3.6662406071049874e-06, "loss": 2.9786, "step": 280740 }, { "epoch": 0.053376, "grad_norm": 0.7757372260093689, "learning_rate": 3.6656174675258525e-06, "loss": 2.8381, "step": 280750 }, { "epoch": 0.0534016, "grad_norm": 0.9594228863716125, "learning_rate": 3.6649943690235046e-06, "loss": 2.9509, "step": 280760 }, { "epoch": 0.0534272, "grad_norm": 1.2940433025360107, "learning_rate": 3.664371311601984e-06, "loss": 3.0613, "step": 280770 }, { "epoch": 0.0534528, "grad_norm": 0.8399789929389954, "learning_rate": 3.6637482952653324e-06, "loss": 3.0807, "step": 280780 }, { "epoch": 0.0534784, "grad_norm": 0.7404793500900269, "learning_rate": 3.6631253200175887e-06, "loss": 2.7041, "step": 280790 }, { "epoch": 0.053504, "grad_norm": 1.0325454473495483, "learning_rate": 3.662502385862794e-06, "loss": 2.9884, "step": 280800 }, { "epoch": 0.0535296, "grad_norm": 0.7938246726989746, "learning_rate": 3.6618794928049894e-06, "loss": 2.7567, "step": 280810 }, { "epoch": 0.0535552, "grad_norm": 0.8003106117248535, "learning_rate": 3.6612566408482085e-06, "loss": 2.7689, "step": 280820 }, { "epoch": 0.0535808, "grad_norm": 0.9003434181213379, "learning_rate": 3.660633829996494e-06, "loss": 2.8328, "step": 280830 }, { "epoch": 0.0536064, "grad_norm": 0.8801031112670898, "learning_rate": 3.660011060253884e-06, "loss": 2.871, "step": 280840 }, { "epoch": 0.053632, "grad_norm": 0.8252796530723572, "learning_rate": 3.659388331624417e-06, "loss": 2.7747, "step": 280850 }, { "epoch": 0.0536576, "grad_norm": 0.7679160237312317, "learning_rate": 3.6587656441121322e-06, "loss": 2.8072, "step": 280860 }, { "epoch": 0.0536832, "grad_norm": 0.8565882444381714, "learning_rate": 3.658142997721067e-06, "loss": 2.7226, "step": 280870 }, { "epoch": 0.0537088, "grad_norm": 0.9428129196166992, "learning_rate": 3.657520392455263e-06, "loss": 2.8878, "step": 280880 }, { "epoch": 0.0537344, "grad_norm": 0.7724279761314392, "learning_rate": 3.6568978283187485e-06, "loss": 2.7902, "step": 280890 }, { "epoch": 0.05376, "grad_norm": 0.8213465213775635, "learning_rate": 3.6562753053155665e-06, "loss": 3.0183, "step": 280900 }, { "epoch": 0.0537856, "grad_norm": 1.0791690349578857, "learning_rate": 3.6556528234497534e-06, "loss": 2.9933, "step": 280910 }, { "epoch": 0.0538112, "grad_norm": 0.851448118686676, "learning_rate": 3.655030382725342e-06, "loss": 2.8608, "step": 280920 }, { "epoch": 0.0538368, "grad_norm": 0.7495488524436951, "learning_rate": 3.6544079831463795e-06, "loss": 2.8197, "step": 280930 }, { "epoch": 0.0538624, "grad_norm": 0.7418445944786072, "learning_rate": 3.6537856247168914e-06, "loss": 2.92, "step": 280940 }, { "epoch": 0.053888, "grad_norm": 0.9972439408302307, "learning_rate": 3.6531633074409167e-06, "loss": 2.7541, "step": 280950 }, { "epoch": 0.0539136, "grad_norm": 0.7816610932350159, "learning_rate": 3.652541031322491e-06, "loss": 2.814, "step": 280960 }, { "epoch": 0.0539392, "grad_norm": 0.8644272685050964, "learning_rate": 3.65191879636565e-06, "loss": 3.0771, "step": 280970 }, { "epoch": 0.0539648, "grad_norm": 0.7967699766159058, "learning_rate": 3.651296602574428e-06, "loss": 2.6464, "step": 280980 }, { "epoch": 0.0539904, "grad_norm": 0.7622295022010803, "learning_rate": 3.65067444995286e-06, "loss": 2.8035, "step": 280990 }, { "epoch": 0.054016, "grad_norm": 0.7753183841705322, "learning_rate": 3.6500523385049846e-06, "loss": 2.9159, "step": 281000 }, { "epoch": 0.0540416, "grad_norm": 0.9856364727020264, "learning_rate": 3.649430268234829e-06, "loss": 2.62, "step": 281010 }, { "epoch": 0.0540672, "grad_norm": 1.1626336574554443, "learning_rate": 3.648808239146431e-06, "loss": 2.7821, "step": 281020 }, { "epoch": 0.0540928, "grad_norm": 0.8966653943061829, "learning_rate": 3.6481862512438225e-06, "loss": 2.7634, "step": 281030 }, { "epoch": 0.0541184, "grad_norm": 0.7253828048706055, "learning_rate": 3.647564304531038e-06, "loss": 3.036, "step": 281040 }, { "epoch": 0.054144, "grad_norm": 1.270037293434143, "learning_rate": 3.6469423990121112e-06, "loss": 2.9153, "step": 281050 }, { "epoch": 0.0541696, "grad_norm": 0.7767655849456787, "learning_rate": 3.6463205346910746e-06, "loss": 2.9362, "step": 281060 }, { "epoch": 0.0541952, "grad_norm": 0.7773063778877258, "learning_rate": 3.6456987115719613e-06, "loss": 2.7586, "step": 281070 }, { "epoch": 0.0542208, "grad_norm": 1.3524823188781738, "learning_rate": 3.645076929658805e-06, "loss": 2.681, "step": 281080 }, { "epoch": 0.0542464, "grad_norm": 1.0739959478378296, "learning_rate": 3.6444551889556303e-06, "loss": 2.7216, "step": 281090 }, { "epoch": 0.054272, "grad_norm": 0.8952800631523132, "learning_rate": 3.643833489466477e-06, "loss": 2.7637, "step": 281100 }, { "epoch": 0.0542976, "grad_norm": 1.1347551345825195, "learning_rate": 3.6432118311953744e-06, "loss": 2.904, "step": 281110 }, { "epoch": 0.0543232, "grad_norm": 0.7935363054275513, "learning_rate": 3.642590214146354e-06, "loss": 2.8935, "step": 281120 }, { "epoch": 0.0543488, "grad_norm": 0.9239153265953064, "learning_rate": 3.6419686383234497e-06, "loss": 2.7777, "step": 281130 }, { "epoch": 0.0543744, "grad_norm": 0.7872231602668762, "learning_rate": 3.641347103730686e-06, "loss": 2.9319, "step": 281140 }, { "epoch": 0.0544, "grad_norm": 1.0719106197357178, "learning_rate": 3.6407256103720966e-06, "loss": 2.9006, "step": 281150 }, { "epoch": 0.0544256, "grad_norm": 0.8853344917297363, "learning_rate": 3.6401041582517104e-06, "loss": 2.8312, "step": 281160 }, { "epoch": 0.0544512, "grad_norm": 0.73929762840271, "learning_rate": 3.6394827473735596e-06, "loss": 2.7776, "step": 281170 }, { "epoch": 0.0544768, "grad_norm": 1.1041626930236816, "learning_rate": 3.6388613777416726e-06, "loss": 2.8523, "step": 281180 }, { "epoch": 0.0545024, "grad_norm": 0.7650001645088196, "learning_rate": 3.638240049360079e-06, "loss": 2.6708, "step": 281190 }, { "epoch": 0.054528, "grad_norm": 0.8399608731269836, "learning_rate": 3.6376187622328106e-06, "loss": 2.9218, "step": 281200 }, { "epoch": 0.0545536, "grad_norm": 0.7880059480667114, "learning_rate": 3.6369975163638904e-06, "loss": 2.7528, "step": 281210 }, { "epoch": 0.0545792, "grad_norm": 0.8264554738998413, "learning_rate": 3.63637631175735e-06, "loss": 2.9111, "step": 281220 }, { "epoch": 0.0546048, "grad_norm": 0.8204026818275452, "learning_rate": 3.635755148417219e-06, "loss": 3.1078, "step": 281230 }, { "epoch": 0.0546304, "grad_norm": 0.8507864475250244, "learning_rate": 3.635134026347524e-06, "loss": 2.8937, "step": 281240 }, { "epoch": 0.054656, "grad_norm": 0.7159416079521179, "learning_rate": 3.6345129455522934e-06, "loss": 2.8854, "step": 281250 }, { "epoch": 0.0546816, "grad_norm": 0.8049778938293457, "learning_rate": 3.633891906035554e-06, "loss": 3.0988, "step": 281260 }, { "epoch": 0.0547072, "grad_norm": 0.9214119911193848, "learning_rate": 3.633270907801334e-06, "loss": 2.7857, "step": 281270 }, { "epoch": 0.0547328, "grad_norm": 0.7979275584220886, "learning_rate": 3.6326499508536604e-06, "loss": 2.6892, "step": 281280 }, { "epoch": 0.0547584, "grad_norm": 0.9213555455207825, "learning_rate": 3.632029035196559e-06, "loss": 2.7976, "step": 281290 }, { "epoch": 0.054784, "grad_norm": 0.7478232979774475, "learning_rate": 3.631408160834057e-06, "loss": 2.8232, "step": 281300 }, { "epoch": 0.0548096, "grad_norm": 1.7645807266235352, "learning_rate": 3.630787327770181e-06, "loss": 2.7243, "step": 281310 }, { "epoch": 0.0548352, "grad_norm": 1.1596723794937134, "learning_rate": 3.6301665360089567e-06, "loss": 2.7887, "step": 281320 }, { "epoch": 0.0548608, "grad_norm": 0.8980315327644348, "learning_rate": 3.6295457855544115e-06, "loss": 2.767, "step": 281330 }, { "epoch": 0.0548864, "grad_norm": 0.8039773106575012, "learning_rate": 3.6289250764105666e-06, "loss": 2.8586, "step": 281340 }, { "epoch": 0.054912, "grad_norm": 0.7852105498313904, "learning_rate": 3.628304408581449e-06, "loss": 2.752, "step": 281350 }, { "epoch": 0.0549376, "grad_norm": 1.0583970546722412, "learning_rate": 3.627683782071084e-06, "loss": 2.7684, "step": 281360 }, { "epoch": 0.0549632, "grad_norm": 0.7825512886047363, "learning_rate": 3.6270631968834967e-06, "loss": 2.6343, "step": 281370 }, { "epoch": 0.0549888, "grad_norm": 1.0265783071517944, "learning_rate": 3.6264426530227104e-06, "loss": 2.9405, "step": 281380 }, { "epoch": 0.0550144, "grad_norm": 0.7991588115692139, "learning_rate": 3.6258221504927493e-06, "loss": 2.8682, "step": 281390 }, { "epoch": 0.05504, "grad_norm": 0.9631965756416321, "learning_rate": 3.6252016892976403e-06, "loss": 2.9556, "step": 281400 }, { "epoch": 0.0550656, "grad_norm": 0.9332606792449951, "learning_rate": 3.6245812694414017e-06, "loss": 2.7616, "step": 281410 }, { "epoch": 0.0550912, "grad_norm": 0.7530226707458496, "learning_rate": 3.623960890928059e-06, "loss": 2.8089, "step": 281420 }, { "epoch": 0.0551168, "grad_norm": 0.9201672077178955, "learning_rate": 3.6233405537616352e-06, "loss": 3.0305, "step": 281430 }, { "epoch": 0.0551424, "grad_norm": 0.8251107335090637, "learning_rate": 3.622720257946151e-06, "loss": 2.7094, "step": 281440 }, { "epoch": 0.055168, "grad_norm": 0.7660237550735474, "learning_rate": 3.6221000034856325e-06, "loss": 2.7943, "step": 281450 }, { "epoch": 0.0551936, "grad_norm": 0.7876573801040649, "learning_rate": 3.6214797903841046e-06, "loss": 2.6906, "step": 281460 }, { "epoch": 0.0552192, "grad_norm": 0.8168686032295227, "learning_rate": 3.6208596186455824e-06, "loss": 2.813, "step": 281470 }, { "epoch": 0.0552448, "grad_norm": 0.7658928632736206, "learning_rate": 3.6202394882740886e-06, "loss": 2.7566, "step": 281480 }, { "epoch": 0.0552704, "grad_norm": 0.941246509552002, "learning_rate": 3.619619399273648e-06, "loss": 2.9644, "step": 281490 }, { "epoch": 0.055296, "grad_norm": 0.8947834372520447, "learning_rate": 3.618999351648279e-06, "loss": 2.8162, "step": 281500 }, { "epoch": 0.0553216, "grad_norm": 0.8744909763336182, "learning_rate": 3.6183793454020034e-06, "loss": 3.0205, "step": 281510 }, { "epoch": 0.0553472, "grad_norm": 0.8333958983421326, "learning_rate": 3.6177593805388412e-06, "loss": 2.8692, "step": 281520 }, { "epoch": 0.0553728, "grad_norm": 0.8910235166549683, "learning_rate": 3.617139457062817e-06, "loss": 2.8888, "step": 281530 }, { "epoch": 0.0553984, "grad_norm": 0.7957921624183655, "learning_rate": 3.6165195749779445e-06, "loss": 2.9832, "step": 281540 }, { "epoch": 0.055424, "grad_norm": 0.8889681696891785, "learning_rate": 3.615899734288245e-06, "loss": 2.988, "step": 281550 }, { "epoch": 0.0554496, "grad_norm": 0.8056368827819824, "learning_rate": 3.6152799349977398e-06, "loss": 2.8115, "step": 281560 }, { "epoch": 0.0554752, "grad_norm": 0.813881516456604, "learning_rate": 3.614660177110446e-06, "loss": 2.7494, "step": 281570 }, { "epoch": 0.0555008, "grad_norm": 0.7650827765464783, "learning_rate": 3.614040460630386e-06, "loss": 2.8138, "step": 281580 }, { "epoch": 0.0555264, "grad_norm": 1.2579259872436523, "learning_rate": 3.6134207855615743e-06, "loss": 2.8183, "step": 281590 }, { "epoch": 0.055552, "grad_norm": 0.8731691837310791, "learning_rate": 3.6128011519080363e-06, "loss": 2.8798, "step": 281600 }, { "epoch": 0.0555776, "grad_norm": 0.7724884748458862, "learning_rate": 3.612181559673782e-06, "loss": 2.8739, "step": 281610 }, { "epoch": 0.0556032, "grad_norm": 1.079256534576416, "learning_rate": 3.6115620088628277e-06, "loss": 2.9776, "step": 281620 }, { "epoch": 0.0556288, "grad_norm": 0.8613798022270203, "learning_rate": 3.610942499479201e-06, "loss": 3.085, "step": 281630 }, { "epoch": 0.0556544, "grad_norm": 0.960761547088623, "learning_rate": 3.610323031526913e-06, "loss": 2.5777, "step": 281640 }, { "epoch": 0.05568, "grad_norm": 1.1446219682693481, "learning_rate": 3.609703605009982e-06, "loss": 2.8496, "step": 281650 }, { "epoch": 0.0557056, "grad_norm": 0.9231471419334412, "learning_rate": 3.6090842199324295e-06, "loss": 2.9292, "step": 281660 }, { "epoch": 0.0557312, "grad_norm": 0.820159912109375, "learning_rate": 3.608464876298263e-06, "loss": 2.8121, "step": 281670 }, { "epoch": 0.0557568, "grad_norm": 0.759782075881958, "learning_rate": 3.6078455741115025e-06, "loss": 2.7957, "step": 281680 }, { "epoch": 0.0557824, "grad_norm": 0.7986273765563965, "learning_rate": 3.6072263133761652e-06, "loss": 2.8389, "step": 281690 }, { "epoch": 0.055808, "grad_norm": 1.9350050687789917, "learning_rate": 3.606607094096267e-06, "loss": 2.9903, "step": 281700 }, { "epoch": 0.0558336, "grad_norm": 0.803634524345398, "learning_rate": 3.6059879162758215e-06, "loss": 2.6019, "step": 281710 }, { "epoch": 0.0558592, "grad_norm": 0.7571471333503723, "learning_rate": 3.6053687799188463e-06, "loss": 2.8171, "step": 281720 }, { "epoch": 0.0558848, "grad_norm": 0.7616685628890991, "learning_rate": 3.6047496850293574e-06, "loss": 2.9311, "step": 281730 }, { "epoch": 0.0559104, "grad_norm": 0.9451366066932678, "learning_rate": 3.6041306316113644e-06, "loss": 2.9695, "step": 281740 }, { "epoch": 0.055936, "grad_norm": 0.9416189193725586, "learning_rate": 3.6035116196688835e-06, "loss": 2.9675, "step": 281750 }, { "epoch": 0.0559616, "grad_norm": 0.8463675379753113, "learning_rate": 3.60289264920593e-06, "loss": 2.753, "step": 281760 }, { "epoch": 0.0559872, "grad_norm": 0.8538432121276855, "learning_rate": 3.602273720226518e-06, "loss": 3.0191, "step": 281770 }, { "epoch": 0.0560128, "grad_norm": 0.8005919456481934, "learning_rate": 3.601654832734661e-06, "loss": 2.9337, "step": 281780 }, { "epoch": 0.0560384, "grad_norm": 0.9426860213279724, "learning_rate": 3.6010359867343714e-06, "loss": 2.7216, "step": 281790 }, { "epoch": 0.056064, "grad_norm": 0.8960503935813904, "learning_rate": 3.600417182229663e-06, "loss": 2.9741, "step": 281800 }, { "epoch": 0.0560896, "grad_norm": 0.8907759785652161, "learning_rate": 3.5997984192245473e-06, "loss": 2.8601, "step": 281810 }, { "epoch": 0.0561152, "grad_norm": 0.8337543606758118, "learning_rate": 3.5991796977230385e-06, "loss": 2.8023, "step": 281820 }, { "epoch": 0.0561408, "grad_norm": 0.8722425103187561, "learning_rate": 3.5985610177291485e-06, "loss": 2.844, "step": 281830 }, { "epoch": 0.0561664, "grad_norm": 0.9128798842430115, "learning_rate": 3.5979423792468893e-06, "loss": 2.904, "step": 281840 }, { "epoch": 0.056192, "grad_norm": 0.9012911319732666, "learning_rate": 3.597323782280272e-06, "loss": 2.9195, "step": 281850 }, { "epoch": 0.0562176, "grad_norm": 0.8129566311836243, "learning_rate": 3.5967052268333113e-06, "loss": 2.6545, "step": 281860 }, { "epoch": 0.0562432, "grad_norm": 0.7904760241508484, "learning_rate": 3.596086712910013e-06, "loss": 3.0129, "step": 281870 }, { "epoch": 0.0562688, "grad_norm": 0.8414548635482788, "learning_rate": 3.5954682405143905e-06, "loss": 2.9604, "step": 281880 }, { "epoch": 0.0562944, "grad_norm": 1.0515245199203491, "learning_rate": 3.5948498096504523e-06, "loss": 3.044, "step": 281890 }, { "epoch": 0.05632, "grad_norm": 1.0331685543060303, "learning_rate": 3.594231420322213e-06, "loss": 2.9467, "step": 281900 }, { "epoch": 0.0563456, "grad_norm": 0.9753516316413879, "learning_rate": 3.5936130725336792e-06, "loss": 2.8956, "step": 281910 }, { "epoch": 0.0563712, "grad_norm": 1.0122655630111694, "learning_rate": 3.5929947662888632e-06, "loss": 2.8724, "step": 281920 }, { "epoch": 0.0563968, "grad_norm": 0.9742740988731384, "learning_rate": 3.5923765015917756e-06, "loss": 2.9238, "step": 281930 }, { "epoch": 0.0564224, "grad_norm": 0.9964593648910522, "learning_rate": 3.5917582784464212e-06, "loss": 2.9194, "step": 281940 }, { "epoch": 0.056448, "grad_norm": 1.16258704662323, "learning_rate": 3.59114009685681e-06, "loss": 2.9431, "step": 281950 }, { "epoch": 0.0564736, "grad_norm": 0.8894971609115601, "learning_rate": 3.5905219568269487e-06, "loss": 2.7688, "step": 281960 }, { "epoch": 0.0564992, "grad_norm": 0.8539565205574036, "learning_rate": 3.5899038583608538e-06, "loss": 2.7882, "step": 281970 }, { "epoch": 0.0565248, "grad_norm": 0.8761526346206665, "learning_rate": 3.5892858014625276e-06, "loss": 2.6471, "step": 281980 }, { "epoch": 0.0565504, "grad_norm": 0.8034496307373047, "learning_rate": 3.588667786135983e-06, "loss": 2.9799, "step": 281990 }, { "epoch": 0.056576, "grad_norm": 0.9359822869300842, "learning_rate": 3.5880498123852205e-06, "loss": 2.7857, "step": 282000 }, { "epoch": 0.0566016, "grad_norm": 0.8860377073287964, "learning_rate": 3.5874318802142504e-06, "loss": 2.8121, "step": 282010 }, { "epoch": 0.0566272, "grad_norm": 0.7668333649635315, "learning_rate": 3.5868139896270804e-06, "loss": 2.9164, "step": 282020 }, { "epoch": 0.0566528, "grad_norm": 0.8131315112113953, "learning_rate": 3.586196140627718e-06, "loss": 2.8572, "step": 282030 }, { "epoch": 0.0566784, "grad_norm": 0.7545493245124817, "learning_rate": 3.585578333220169e-06, "loss": 2.8295, "step": 282040 }, { "epoch": 0.056704, "grad_norm": 0.9014966487884521, "learning_rate": 3.5849605674084386e-06, "loss": 2.9174, "step": 282050 }, { "epoch": 0.0567296, "grad_norm": 0.9605920910835266, "learning_rate": 3.5843428431965377e-06, "loss": 2.5578, "step": 282060 }, { "epoch": 0.0567552, "grad_norm": 0.9561027884483337, "learning_rate": 3.583725160588466e-06, "loss": 2.6751, "step": 282070 }, { "epoch": 0.0567808, "grad_norm": 1.1362677812576294, "learning_rate": 3.5831075195882304e-06, "loss": 2.9611, "step": 282080 }, { "epoch": 0.0568064, "grad_norm": 0.7616156339645386, "learning_rate": 3.5824899201998374e-06, "loss": 2.789, "step": 282090 }, { "epoch": 0.056832, "grad_norm": 0.7869749665260315, "learning_rate": 3.581872362427292e-06, "loss": 2.8735, "step": 282100 }, { "epoch": 0.0568576, "grad_norm": 0.8021798133850098, "learning_rate": 3.581254846274599e-06, "loss": 2.7435, "step": 282110 }, { "epoch": 0.0568832, "grad_norm": 1.0407538414001465, "learning_rate": 3.580637371745761e-06, "loss": 2.835, "step": 282120 }, { "epoch": 0.0569088, "grad_norm": 0.8048579096794128, "learning_rate": 3.580019938844789e-06, "loss": 2.9075, "step": 282130 }, { "epoch": 0.0569344, "grad_norm": 0.7773354649543762, "learning_rate": 3.5794025475756732e-06, "loss": 2.9048, "step": 282140 }, { "epoch": 0.05696, "grad_norm": 0.8670147061347961, "learning_rate": 3.578785197942429e-06, "loss": 2.9075, "step": 282150 }, { "epoch": 0.0569856, "grad_norm": 0.890913724899292, "learning_rate": 3.578167889949057e-06, "loss": 2.7465, "step": 282160 }, { "epoch": 0.0570112, "grad_norm": 0.788293719291687, "learning_rate": 3.5775506235995583e-06, "loss": 2.597, "step": 282170 }, { "epoch": 0.0570368, "grad_norm": 1.011080026626587, "learning_rate": 3.576933398897937e-06, "loss": 2.8574, "step": 282180 }, { "epoch": 0.0570624, "grad_norm": 0.7888473868370056, "learning_rate": 3.5763162158482e-06, "loss": 2.9173, "step": 282190 }, { "epoch": 0.057088, "grad_norm": 0.8873864412307739, "learning_rate": 3.575699074454342e-06, "loss": 2.796, "step": 282200 }, { "epoch": 0.0571136, "grad_norm": 0.8616052269935608, "learning_rate": 3.5750819747203667e-06, "loss": 2.6776, "step": 282210 }, { "epoch": 0.0571392, "grad_norm": 0.7729833722114563, "learning_rate": 3.574464916650278e-06, "loss": 2.7357, "step": 282220 }, { "epoch": 0.0571648, "grad_norm": 0.8175713419914246, "learning_rate": 3.5738479002480766e-06, "loss": 2.7628, "step": 282230 }, { "epoch": 0.0571904, "grad_norm": 0.8298690915107727, "learning_rate": 3.5732309255177635e-06, "loss": 2.9284, "step": 282240 }, { "epoch": 0.057216, "grad_norm": 0.8982298374176025, "learning_rate": 3.5726139924633396e-06, "loss": 2.8134, "step": 282250 }, { "epoch": 0.0572416, "grad_norm": 1.0212881565093994, "learning_rate": 3.57199710108881e-06, "loss": 2.8583, "step": 282260 }, { "epoch": 0.0572672, "grad_norm": 0.8038955330848694, "learning_rate": 3.5713802513981667e-06, "loss": 2.9043, "step": 282270 }, { "epoch": 0.0572928, "grad_norm": 0.8020465970039368, "learning_rate": 3.5707634433954143e-06, "loss": 2.835, "step": 282280 }, { "epoch": 0.0573184, "grad_norm": 0.8063422441482544, "learning_rate": 3.5701466770845527e-06, "loss": 2.926, "step": 282290 }, { "epoch": 0.057344, "grad_norm": 0.9190803170204163, "learning_rate": 3.5695299524695803e-06, "loss": 2.8699, "step": 282300 }, { "epoch": 0.0573696, "grad_norm": 0.9445845484733582, "learning_rate": 3.568913269554497e-06, "loss": 2.67, "step": 282310 }, { "epoch": 0.0573952, "grad_norm": 0.7686469554901123, "learning_rate": 3.568296628343303e-06, "loss": 2.9632, "step": 282320 }, { "epoch": 0.0574208, "grad_norm": 0.7713804244995117, "learning_rate": 3.5676800288399947e-06, "loss": 2.7893, "step": 282330 }, { "epoch": 0.0574464, "grad_norm": 0.7783017158508301, "learning_rate": 3.567063471048573e-06, "loss": 2.8878, "step": 282340 }, { "epoch": 0.057472, "grad_norm": 0.8067165613174438, "learning_rate": 3.566446954973035e-06, "loss": 2.6799, "step": 282350 }, { "epoch": 0.0574976, "grad_norm": 0.8024999499320984, "learning_rate": 3.5658304806173784e-06, "loss": 2.9366, "step": 282360 }, { "epoch": 0.0575232, "grad_norm": 0.741175651550293, "learning_rate": 3.5652140479856e-06, "loss": 2.9185, "step": 282370 }, { "epoch": 0.0575488, "grad_norm": 0.7884333729743958, "learning_rate": 3.5645976570816997e-06, "loss": 3.1534, "step": 282380 }, { "epoch": 0.0575744, "grad_norm": 0.8318665623664856, "learning_rate": 3.5639813079096763e-06, "loss": 2.8137, "step": 282390 }, { "epoch": 0.0576, "grad_norm": 0.8691872358322144, "learning_rate": 3.5633650004735208e-06, "loss": 2.8447, "step": 282400 }, { "epoch": 0.0576256, "grad_norm": 0.7700433135032654, "learning_rate": 3.5627487347772317e-06, "loss": 2.8743, "step": 282410 }, { "epoch": 0.0576512, "grad_norm": 0.9903349280357361, "learning_rate": 3.5621325108248074e-06, "loss": 2.8837, "step": 282420 }, { "epoch": 0.0576768, "grad_norm": 0.8060488104820251, "learning_rate": 3.5615163286202416e-06, "loss": 2.7645, "step": 282430 }, { "epoch": 0.0577024, "grad_norm": 0.8963585495948792, "learning_rate": 3.5609001881675307e-06, "loss": 2.9192, "step": 282440 }, { "epoch": 0.057728, "grad_norm": 0.7756354212760925, "learning_rate": 3.5602840894706723e-06, "loss": 2.7876, "step": 282450 }, { "epoch": 0.0577536, "grad_norm": 0.7602532505989075, "learning_rate": 3.559668032533663e-06, "loss": 2.8572, "step": 282460 }, { "epoch": 0.0577792, "grad_norm": 0.8691879510879517, "learning_rate": 3.5590520173604916e-06, "loss": 2.7671, "step": 282470 }, { "epoch": 0.0578048, "grad_norm": 0.7795130014419556, "learning_rate": 3.558436043955156e-06, "loss": 3.0375, "step": 282480 }, { "epoch": 0.0578304, "grad_norm": 1.0209444761276245, "learning_rate": 3.557820112321647e-06, "loss": 2.8371, "step": 282490 }, { "epoch": 0.057856, "grad_norm": 0.7998336553573608, "learning_rate": 3.557204222463966e-06, "loss": 2.8177, "step": 282500 }, { "epoch": 0.0578816, "grad_norm": 0.80689936876297, "learning_rate": 3.556588374386107e-06, "loss": 3.0445, "step": 282510 }, { "epoch": 0.0579072, "grad_norm": 0.8653448224067688, "learning_rate": 3.5559725680920563e-06, "loss": 2.9062, "step": 282520 }, { "epoch": 0.0579328, "grad_norm": 0.8561112284660339, "learning_rate": 3.555356803585811e-06, "loss": 2.8906, "step": 282530 }, { "epoch": 0.0579584, "grad_norm": 0.863034188747406, "learning_rate": 3.5547410808713635e-06, "loss": 2.9242, "step": 282540 }, { "epoch": 0.057984, "grad_norm": 1.8672435283660889, "learning_rate": 3.5541253999527083e-06, "loss": 2.7733, "step": 282550 }, { "epoch": 0.0580096, "grad_norm": 0.8469155430793762, "learning_rate": 3.5535097608338355e-06, "loss": 2.3949, "step": 282560 }, { "epoch": 0.0580352, "grad_norm": 0.8242157101631165, "learning_rate": 3.5528941635187387e-06, "loss": 2.6473, "step": 282570 }, { "epoch": 0.0580608, "grad_norm": 0.9562524557113647, "learning_rate": 3.55227860801141e-06, "loss": 2.6242, "step": 282580 }, { "epoch": 0.0580864, "grad_norm": 0.7940802574157715, "learning_rate": 3.5516630943158446e-06, "loss": 2.6345, "step": 282590 }, { "epoch": 0.058112, "grad_norm": 0.7710585594177246, "learning_rate": 3.551047622436028e-06, "loss": 2.7688, "step": 282600 }, { "epoch": 0.0581376, "grad_norm": 0.9093503952026367, "learning_rate": 3.550432192375952e-06, "loss": 2.4879, "step": 282610 }, { "epoch": 0.0581632, "grad_norm": 0.8822441697120667, "learning_rate": 3.5498168041396095e-06, "loss": 2.3683, "step": 282620 }, { "epoch": 0.0581888, "grad_norm": 0.7875648736953735, "learning_rate": 3.5492014577309906e-06, "loss": 2.5552, "step": 282630 }, { "epoch": 0.0582144, "grad_norm": 0.8544639348983765, "learning_rate": 3.5485861531540866e-06, "loss": 2.6071, "step": 282640 }, { "epoch": 0.05824, "grad_norm": 0.8425102233886719, "learning_rate": 3.547970890412886e-06, "loss": 2.9535, "step": 282650 }, { "epoch": 0.0582656, "grad_norm": 0.8645474314689636, "learning_rate": 3.547355669511381e-06, "loss": 2.6222, "step": 282660 }, { "epoch": 0.0582912, "grad_norm": 1.0724979639053345, "learning_rate": 3.5467404904535573e-06, "loss": 2.5945, "step": 282670 }, { "epoch": 0.0583168, "grad_norm": 0.8770096302032471, "learning_rate": 3.5461253532434082e-06, "loss": 2.5916, "step": 282680 }, { "epoch": 0.0583424, "grad_norm": 0.8343448042869568, "learning_rate": 3.54551025788492e-06, "loss": 2.6443, "step": 282690 }, { "epoch": 0.058368, "grad_norm": 0.8614189624786377, "learning_rate": 3.5448952043820827e-06, "loss": 2.5663, "step": 282700 }, { "epoch": 0.0583936, "grad_norm": 0.8366994261741638, "learning_rate": 3.5442801927388883e-06, "loss": 2.727, "step": 282710 }, { "epoch": 0.0584192, "grad_norm": 0.7708069086074829, "learning_rate": 3.5436652229593172e-06, "loss": 2.3278, "step": 282720 }, { "epoch": 0.0584448, "grad_norm": 0.8071258664131165, "learning_rate": 3.5430502950473612e-06, "loss": 2.669, "step": 282730 }, { "epoch": 0.0584704, "grad_norm": 0.9208940863609314, "learning_rate": 3.5424354090070077e-06, "loss": 2.5975, "step": 282740 }, { "epoch": 0.058496, "grad_norm": 0.9240843057632446, "learning_rate": 3.5418205648422444e-06, "loss": 2.6669, "step": 282750 }, { "epoch": 0.0585216, "grad_norm": 0.6998133659362793, "learning_rate": 3.5412057625570597e-06, "loss": 2.5862, "step": 282760 }, { "epoch": 0.0585472, "grad_norm": 0.8821069598197937, "learning_rate": 3.540591002155438e-06, "loss": 2.9359, "step": 282770 }, { "epoch": 0.0585728, "grad_norm": 0.8785794377326965, "learning_rate": 3.5399762836413666e-06, "loss": 2.9525, "step": 282780 }, { "epoch": 0.0585984, "grad_norm": 0.8046777844429016, "learning_rate": 3.5393616070188364e-06, "loss": 2.8871, "step": 282790 }, { "epoch": 0.058624, "grad_norm": 0.9389383792877197, "learning_rate": 3.5387469722918243e-06, "loss": 2.7013, "step": 282800 }, { "epoch": 0.0586496, "grad_norm": 0.7786878347396851, "learning_rate": 3.5381323794643217e-06, "loss": 2.8231, "step": 282810 }, { "epoch": 0.0586752, "grad_norm": 0.8045656681060791, "learning_rate": 3.5375178285403146e-06, "loss": 2.5576, "step": 282820 }, { "epoch": 0.0587008, "grad_norm": 0.7759900093078613, "learning_rate": 3.536903319523782e-06, "loss": 2.721, "step": 282830 }, { "epoch": 0.0587264, "grad_norm": 0.7430704832077026, "learning_rate": 3.536288852418721e-06, "loss": 2.4121, "step": 282840 }, { "epoch": 0.058752, "grad_norm": 0.7677242755889893, "learning_rate": 3.535674427229104e-06, "loss": 2.6752, "step": 282850 }, { "epoch": 0.0587776, "grad_norm": 0.7913395762443542, "learning_rate": 3.535060043958922e-06, "loss": 2.6654, "step": 282860 }, { "epoch": 0.0588032, "grad_norm": 0.8860854506492615, "learning_rate": 3.5344457026121583e-06, "loss": 2.6157, "step": 282870 }, { "epoch": 0.0588288, "grad_norm": 0.8565835952758789, "learning_rate": 3.5338314031927945e-06, "loss": 2.5771, "step": 282880 }, { "epoch": 0.0588544, "grad_norm": 1.086011528968811, "learning_rate": 3.5332171457048146e-06, "loss": 2.5819, "step": 282890 }, { "epoch": 0.05888, "grad_norm": 0.7449562549591064, "learning_rate": 3.5326029301522047e-06, "loss": 2.8815, "step": 282900 }, { "epoch": 0.0589056, "grad_norm": 0.9979103207588196, "learning_rate": 3.531988756538949e-06, "loss": 2.6666, "step": 282910 }, { "epoch": 0.0589312, "grad_norm": 0.8397099375724792, "learning_rate": 3.531374624869023e-06, "loss": 2.5893, "step": 282920 }, { "epoch": 0.0589568, "grad_norm": 1.076920986175537, "learning_rate": 3.5307605351464137e-06, "loss": 2.5981, "step": 282930 }, { "epoch": 0.0589824, "grad_norm": 0.7912556529045105, "learning_rate": 3.5301464873751035e-06, "loss": 2.6792, "step": 282940 }, { "epoch": 0.059008, "grad_norm": 0.7816200256347656, "learning_rate": 3.5295324815590737e-06, "loss": 2.5026, "step": 282950 }, { "epoch": 0.0590336, "grad_norm": 0.9007536768913269, "learning_rate": 3.5289185177023054e-06, "loss": 2.6019, "step": 282960 }, { "epoch": 0.0590592, "grad_norm": 0.8043314218521118, "learning_rate": 3.5283045958087814e-06, "loss": 2.6779, "step": 282970 }, { "epoch": 0.0590848, "grad_norm": 0.8299691677093506, "learning_rate": 3.527690715882486e-06, "loss": 2.575, "step": 282980 }, { "epoch": 0.0591104, "grad_norm": 0.8313766121864319, "learning_rate": 3.5270768779273913e-06, "loss": 2.8902, "step": 282990 }, { "epoch": 0.059136, "grad_norm": 0.857509195804596, "learning_rate": 3.526463081947484e-06, "loss": 2.6485, "step": 283000 }, { "epoch": 0.0591616, "grad_norm": 0.8592374324798584, "learning_rate": 3.52584932794674e-06, "loss": 2.7248, "step": 283010 }, { "epoch": 0.0591872, "grad_norm": 0.7581596970558167, "learning_rate": 3.5252356159291446e-06, "loss": 2.7699, "step": 283020 }, { "epoch": 0.0592128, "grad_norm": 0.8172523975372314, "learning_rate": 3.5246219458986765e-06, "loss": 2.4348, "step": 283030 }, { "epoch": 0.0592384, "grad_norm": 0.7621378302574158, "learning_rate": 3.5240083178593167e-06, "loss": 2.5636, "step": 283040 }, { "epoch": 0.059264, "grad_norm": 0.8612263798713684, "learning_rate": 3.5233947318150385e-06, "loss": 2.6971, "step": 283050 }, { "epoch": 0.0592896, "grad_norm": 1.0953137874603271, "learning_rate": 3.5227811877698246e-06, "loss": 2.6012, "step": 283060 }, { "epoch": 0.0593152, "grad_norm": 0.8069667816162109, "learning_rate": 3.5221676857276533e-06, "loss": 2.8009, "step": 283070 }, { "epoch": 0.0593408, "grad_norm": 1.012966275215149, "learning_rate": 3.521554225692503e-06, "loss": 2.5668, "step": 283080 }, { "epoch": 0.0593664, "grad_norm": 0.8272461891174316, "learning_rate": 3.5209408076683517e-06, "loss": 2.6417, "step": 283090 }, { "epoch": 0.059392, "grad_norm": 0.9153042435646057, "learning_rate": 3.520327431659177e-06, "loss": 2.7748, "step": 283100 }, { "epoch": 0.0594176, "grad_norm": 0.816961407661438, "learning_rate": 3.5197140976689605e-06, "loss": 2.5725, "step": 283110 }, { "epoch": 0.0594432, "grad_norm": 1.1033607721328735, "learning_rate": 3.519100805701673e-06, "loss": 2.2197, "step": 283120 }, { "epoch": 0.0594688, "grad_norm": 1.1165705919265747, "learning_rate": 3.518487555761293e-06, "loss": 2.1812, "step": 283130 }, { "epoch": 0.0594944, "grad_norm": 0.8478395938873291, "learning_rate": 3.5178743478517997e-06, "loss": 2.7102, "step": 283140 }, { "epoch": 0.05952, "grad_norm": 0.8554633855819702, "learning_rate": 3.5172611819771684e-06, "loss": 2.5755, "step": 283150 }, { "epoch": 0.0595456, "grad_norm": 0.860910952091217, "learning_rate": 3.5166480581413753e-06, "loss": 2.7172, "step": 283160 }, { "epoch": 0.0595712, "grad_norm": 0.9218149185180664, "learning_rate": 3.5160349763483958e-06, "loss": 2.6793, "step": 283170 }, { "epoch": 0.0595968, "grad_norm": 0.8347735404968262, "learning_rate": 3.5154219366022068e-06, "loss": 2.8447, "step": 283180 }, { "epoch": 0.0596224, "grad_norm": 0.8495067358016968, "learning_rate": 3.514808938906784e-06, "loss": 2.7694, "step": 283190 }, { "epoch": 0.059648, "grad_norm": 0.7471202611923218, "learning_rate": 3.514195983266101e-06, "loss": 2.6371, "step": 283200 }, { "epoch": 2.56e-05, "grad_norm": 0.7950068116188049, "learning_rate": 3.5135830696841333e-06, "loss": 1.7747, "step": 283210 }, { "epoch": 5.12e-05, "grad_norm": 0.8738780617713928, "learning_rate": 3.5129701981648544e-06, "loss": 2.8628, "step": 283220 }, { "epoch": 7.68e-05, "grad_norm": 0.8152276873588562, "learning_rate": 3.5123573687122416e-06, "loss": 2.6721, "step": 283230 }, { "epoch": 0.0001024, "grad_norm": 0.8456998467445374, "learning_rate": 3.5117445813302696e-06, "loss": 2.5292, "step": 283240 }, { "epoch": 0.000128, "grad_norm": 0.8213587403297424, "learning_rate": 3.5111318360229063e-06, "loss": 2.7648, "step": 283250 }, { "epoch": 0.0001536, "grad_norm": 0.788155734539032, "learning_rate": 3.5105191327941277e-06, "loss": 2.4318, "step": 283260 }, { "epoch": 0.0001792, "grad_norm": 0.8690712451934814, "learning_rate": 3.5099064716479094e-06, "loss": 2.9005, "step": 283270 }, { "epoch": 0.0002048, "grad_norm": 0.8414232730865479, "learning_rate": 3.5092938525882213e-06, "loss": 2.7059, "step": 283280 }, { "epoch": 0.0002304, "grad_norm": 1.1028679609298706, "learning_rate": 3.508681275619038e-06, "loss": 2.6872, "step": 283290 }, { "epoch": 0.000256, "grad_norm": 0.8388330936431885, "learning_rate": 3.5080687407443315e-06, "loss": 2.5978, "step": 283300 }, { "epoch": 0.0002816, "grad_norm": 0.8191434741020203, "learning_rate": 3.5074562479680774e-06, "loss": 2.5036, "step": 283310 }, { "epoch": 0.0003072, "grad_norm": 0.8457283973693848, "learning_rate": 3.5068437972942406e-06, "loss": 2.6306, "step": 283320 }, { "epoch": 0.0003328, "grad_norm": 0.8684128522872925, "learning_rate": 3.506231388726796e-06, "loss": 2.6373, "step": 283330 }, { "epoch": 0.0003584, "grad_norm": 0.9154194593429565, "learning_rate": 3.505619022269715e-06, "loss": 2.7714, "step": 283340 }, { "epoch": 0.000384, "grad_norm": 0.873955488204956, "learning_rate": 3.505006697926969e-06, "loss": 2.5991, "step": 283350 }, { "epoch": 0.0004096, "grad_norm": 0.8618406057357788, "learning_rate": 3.5043944157025245e-06, "loss": 2.6139, "step": 283360 }, { "epoch": 0.0004352, "grad_norm": 0.8346862196922302, "learning_rate": 3.503782175600363e-06, "loss": 2.7772, "step": 283370 }, { "epoch": 0.0004608, "grad_norm": 0.8263989090919495, "learning_rate": 3.503169977624444e-06, "loss": 2.6398, "step": 283380 }, { "epoch": 0.0004864, "grad_norm": 0.9356706738471985, "learning_rate": 3.5025578217787417e-06, "loss": 2.758, "step": 283390 }, { "epoch": 0.000512, "grad_norm": 0.895528256893158, "learning_rate": 3.501945708067225e-06, "loss": 2.5595, "step": 283400 }, { "epoch": 0.0005376, "grad_norm": 0.8082800507545471, "learning_rate": 3.501333636493862e-06, "loss": 2.4339, "step": 283410 }, { "epoch": 0.0005632, "grad_norm": 1.010962963104248, "learning_rate": 3.500721607062625e-06, "loss": 2.735, "step": 283420 }, { "epoch": 0.0005888, "grad_norm": 0.8266632556915283, "learning_rate": 3.5001096197774797e-06, "loss": 2.5789, "step": 283430 }, { "epoch": 0.0006144, "grad_norm": 0.8075805902481079, "learning_rate": 3.4994976746424003e-06, "loss": 2.4646, "step": 283440 }, { "epoch": 0.00064, "grad_norm": 0.8559695482254028, "learning_rate": 3.498885771661348e-06, "loss": 2.7596, "step": 283450 }, { "epoch": 0.0006656, "grad_norm": 0.8568155765533447, "learning_rate": 3.4982739108382925e-06, "loss": 2.7664, "step": 283460 }, { "epoch": 0.0006912, "grad_norm": 0.7903651595115662, "learning_rate": 3.4976620921772032e-06, "loss": 2.986, "step": 283470 }, { "epoch": 0.0007168, "grad_norm": 0.7821181416511536, "learning_rate": 3.497050315682047e-06, "loss": 2.7047, "step": 283480 }, { "epoch": 0.0007424, "grad_norm": 0.8469326496124268, "learning_rate": 3.4964385813567912e-06, "loss": 2.7856, "step": 283490 }, { "epoch": 0.000768, "grad_norm": 0.7601625323295593, "learning_rate": 3.4958268892054026e-06, "loss": 2.8465, "step": 283500 }, { "epoch": 0.0007936, "grad_norm": 0.7890599966049194, "learning_rate": 3.495215239231852e-06, "loss": 2.8037, "step": 283510 }, { "epoch": 0.0008192, "grad_norm": 0.8635018467903137, "learning_rate": 3.4946036314400977e-06, "loss": 2.8449, "step": 283520 }, { "epoch": 0.0008448, "grad_norm": 0.8453453183174133, "learning_rate": 3.4939920658341075e-06, "loss": 2.6783, "step": 283530 }, { "epoch": 0.0008704, "grad_norm": 1.017195224761963, "learning_rate": 3.4933805424178526e-06, "loss": 2.5767, "step": 283540 }, { "epoch": 0.000896, "grad_norm": 0.9207848310470581, "learning_rate": 3.4927690611952946e-06, "loss": 2.6876, "step": 283550 }, { "epoch": 0.0009216, "grad_norm": 0.8238672018051147, "learning_rate": 3.492157622170401e-06, "loss": 2.5401, "step": 283560 }, { "epoch": 0.0009472, "grad_norm": 0.7404062151908875, "learning_rate": 3.491546225347139e-06, "loss": 2.4979, "step": 283570 }, { "epoch": 0.0009728, "grad_norm": 0.7906031012535095, "learning_rate": 3.4909348707294656e-06, "loss": 2.732, "step": 283580 }, { "epoch": 0.0009984, "grad_norm": 0.9629719853401184, "learning_rate": 3.490323558321351e-06, "loss": 2.7347, "step": 283590 }, { "epoch": 0.001024, "grad_norm": 0.8967549204826355, "learning_rate": 3.4897122881267575e-06, "loss": 2.8749, "step": 283600 }, { "epoch": 0.0010496, "grad_norm": 0.7998476624488831, "learning_rate": 3.489101060149649e-06, "loss": 2.7998, "step": 283610 }, { "epoch": 0.0010752, "grad_norm": 0.8573558926582336, "learning_rate": 3.488489874393991e-06, "loss": 2.5668, "step": 283620 }, { "epoch": 0.0011008, "grad_norm": 0.8348225951194763, "learning_rate": 3.487878730863744e-06, "loss": 2.6932, "step": 283630 }, { "epoch": 0.0011264, "grad_norm": 0.8233124017715454, "learning_rate": 3.487267629562877e-06, "loss": 2.6407, "step": 283640 }, { "epoch": 0.001152, "grad_norm": 0.8998880982398987, "learning_rate": 3.486656570495346e-06, "loss": 2.6285, "step": 283650 }, { "epoch": 0.0011776, "grad_norm": 0.8838226199150085, "learning_rate": 3.486045553665115e-06, "loss": 2.8972, "step": 283660 }, { "epoch": 0.0012032, "grad_norm": 0.9840090870857239, "learning_rate": 3.4854345790761477e-06, "loss": 2.8352, "step": 283670 }, { "epoch": 0.0012288, "grad_norm": 0.8439797163009644, "learning_rate": 3.4848236467324058e-06, "loss": 2.8337, "step": 283680 }, { "epoch": 0.0012544, "grad_norm": 0.7773038744926453, "learning_rate": 3.4842127566378503e-06, "loss": 2.6589, "step": 283690 }, { "epoch": 0.00128, "grad_norm": 0.8276573419570923, "learning_rate": 3.4836019087964445e-06, "loss": 2.5781, "step": 283700 }, { "epoch": 0.0013056, "grad_norm": 0.825692892074585, "learning_rate": 3.482991103212148e-06, "loss": 2.7235, "step": 283710 }, { "epoch": 0.0013312, "grad_norm": 0.9398384690284729, "learning_rate": 3.4823803398889224e-06, "loss": 2.5488, "step": 283720 }, { "epoch": 0.0013568, "grad_norm": 0.8753634691238403, "learning_rate": 3.481769618830727e-06, "loss": 2.4997, "step": 283730 }, { "epoch": 0.0013824, "grad_norm": 0.7915464639663696, "learning_rate": 3.4811589400415245e-06, "loss": 2.6335, "step": 283740 }, { "epoch": 0.001408, "grad_norm": 0.8104538321495056, "learning_rate": 3.480548303525274e-06, "loss": 2.6673, "step": 283750 }, { "epoch": 0.0014336, "grad_norm": 0.8805339336395264, "learning_rate": 3.4799377092859344e-06, "loss": 2.7607, "step": 283760 }, { "epoch": 0.0014592, "grad_norm": 0.7976265549659729, "learning_rate": 3.479327157327469e-06, "loss": 2.6034, "step": 283770 }, { "epoch": 0.0014848, "grad_norm": 0.7701603174209595, "learning_rate": 3.4787166476538313e-06, "loss": 2.4297, "step": 283780 }, { "epoch": 0.0015104, "grad_norm": 1.0694342851638794, "learning_rate": 3.478106180268983e-06, "loss": 2.1824, "step": 283790 }, { "epoch": 0.001536, "grad_norm": 0.9944294691085815, "learning_rate": 3.477495755176883e-06, "loss": 2.5777, "step": 283800 }, { "epoch": 0.0015616, "grad_norm": 0.7870601415634155, "learning_rate": 3.476885372381489e-06, "loss": 2.6481, "step": 283810 }, { "epoch": 0.0015872, "grad_norm": 0.9298821091651917, "learning_rate": 3.47627503188676e-06, "loss": 2.6476, "step": 283820 }, { "epoch": 0.0016128, "grad_norm": 0.8537552356719971, "learning_rate": 3.475664733696653e-06, "loss": 2.7359, "step": 283830 }, { "epoch": 0.0016384, "grad_norm": 0.8459446430206299, "learning_rate": 3.47505447781513e-06, "loss": 2.7512, "step": 283840 }, { "epoch": 0.001664, "grad_norm": 0.8116874098777771, "learning_rate": 3.4744442642461416e-06, "loss": 2.6601, "step": 283850 }, { "epoch": 0.0016896, "grad_norm": 0.7898092865943909, "learning_rate": 3.4738340929936477e-06, "loss": 2.6592, "step": 283860 }, { "epoch": 0.0017152, "grad_norm": 0.8250575065612793, "learning_rate": 3.473223964061605e-06, "loss": 2.5408, "step": 283870 }, { "epoch": 0.0017408, "grad_norm": 0.8807704448699951, "learning_rate": 3.4726138774539663e-06, "loss": 2.8734, "step": 283880 }, { "epoch": 0.0017664, "grad_norm": 1.1005394458770752, "learning_rate": 3.4720038331746965e-06, "loss": 2.9802, "step": 283890 }, { "epoch": 0.001792, "grad_norm": 1.0146350860595703, "learning_rate": 3.4713938312277495e-06, "loss": 2.6761, "step": 283900 }, { "epoch": 2.56e-05, "grad_norm": 1.2175874710083008, "learning_rate": 3.470783871617075e-06, "loss": 1.9035, "step": 283910 }, { "epoch": 5.12e-05, "grad_norm": 0.824678897857666, "learning_rate": 3.470173954346633e-06, "loss": 2.6575, "step": 283920 }, { "epoch": 7.68e-05, "grad_norm": 0.8677675724029541, "learning_rate": 3.4695640794203765e-06, "loss": 2.8308, "step": 283930 }, { "epoch": 0.0001024, "grad_norm": 0.7826007008552551, "learning_rate": 3.468954246842261e-06, "loss": 2.6476, "step": 283940 }, { "epoch": 0.000128, "grad_norm": 0.8487231731414795, "learning_rate": 3.4683444566162417e-06, "loss": 2.8188, "step": 283950 }, { "epoch": 0.0001536, "grad_norm": 0.7565584182739258, "learning_rate": 3.4677347087462733e-06, "loss": 2.6089, "step": 283960 }, { "epoch": 0.0001792, "grad_norm": 0.8885096907615662, "learning_rate": 3.4671250032363124e-06, "loss": 2.7586, "step": 283970 }, { "epoch": 0.0002048, "grad_norm": 0.7965621948242188, "learning_rate": 3.4665153400903063e-06, "loss": 2.641, "step": 283980 }, { "epoch": 0.0002304, "grad_norm": 0.8201695084571838, "learning_rate": 3.4659057193122115e-06, "loss": 2.4866, "step": 283990 }, { "epoch": 0.000256, "grad_norm": 0.9300751686096191, "learning_rate": 3.4652961409059814e-06, "loss": 2.6082, "step": 284000 }, { "epoch": 0.0002816, "grad_norm": 0.7769632339477539, "learning_rate": 3.464686604875569e-06, "loss": 2.6132, "step": 284010 }, { "epoch": 0.0003072, "grad_norm": 0.8103506565093994, "learning_rate": 3.464077111224927e-06, "loss": 2.4948, "step": 284020 }, { "epoch": 0.0003328, "grad_norm": 0.9639072418212891, "learning_rate": 3.4634676599580087e-06, "loss": 2.4345, "step": 284030 }, { "epoch": 0.0003584, "grad_norm": 0.8961217999458313, "learning_rate": 3.4628582510787688e-06, "loss": 2.7743, "step": 284040 }, { "epoch": 0.000384, "grad_norm": 0.8192261457443237, "learning_rate": 3.4622488845911485e-06, "loss": 2.8946, "step": 284050 }, { "epoch": 0.0004096, "grad_norm": 0.9096919298171997, "learning_rate": 3.4616395604991103e-06, "loss": 2.7763, "step": 284060 }, { "epoch": 0.0004352, "grad_norm": 0.7924925088882446, "learning_rate": 3.461030278806602e-06, "loss": 2.7702, "step": 284070 }, { "epoch": 0.0004608, "grad_norm": 0.8635206818580627, "learning_rate": 3.460421039517574e-06, "loss": 2.6086, "step": 284080 }, { "epoch": 0.0004864, "grad_norm": 0.8943058848381042, "learning_rate": 3.4598118426359785e-06, "loss": 2.8014, "step": 284090 }, { "epoch": 0.000512, "grad_norm": 0.7288547158241272, "learning_rate": 3.459202688165767e-06, "loss": 2.6098, "step": 284100 }, { "epoch": 0.0005376, "grad_norm": 0.7589943408966064, "learning_rate": 3.4585935761108857e-06, "loss": 2.6297, "step": 284110 }, { "epoch": 0.0005632, "grad_norm": 0.8411102890968323, "learning_rate": 3.457984506475286e-06, "loss": 2.3604, "step": 284120 }, { "epoch": 0.0005888, "grad_norm": 0.8573785424232483, "learning_rate": 3.4573754792629176e-06, "loss": 2.7726, "step": 284130 }, { "epoch": 0.0006144, "grad_norm": 0.7545106410980225, "learning_rate": 3.456766494477731e-06, "loss": 2.6326, "step": 284140 }, { "epoch": 0.00064, "grad_norm": 0.8620581030845642, "learning_rate": 3.4561575521236747e-06, "loss": 2.8446, "step": 284150 }, { "epoch": 0.0006656, "grad_norm": 0.9055934548377991, "learning_rate": 3.4555486522046966e-06, "loss": 3.1424, "step": 284160 }, { "epoch": 0.0006912, "grad_norm": 0.8288428783416748, "learning_rate": 3.454939794724751e-06, "loss": 2.7281, "step": 284170 }, { "epoch": 0.0007168, "grad_norm": 0.8371344804763794, "learning_rate": 3.4543309796877767e-06, "loss": 2.7612, "step": 284180 }, { "epoch": 0.0007424, "grad_norm": 0.9846145510673523, "learning_rate": 3.453722207097727e-06, "loss": 2.9146, "step": 284190 }, { "epoch": 0.000768, "grad_norm": 0.8130753636360168, "learning_rate": 3.4531134769585493e-06, "loss": 2.5128, "step": 284200 }, { "epoch": 0.0007936, "grad_norm": 0.8429943919181824, "learning_rate": 3.4525047892741904e-06, "loss": 2.747, "step": 284210 }, { "epoch": 0.0008192, "grad_norm": 0.8381312489509583, "learning_rate": 3.4518961440485976e-06, "loss": 1.9032, "step": 284220 }, { "epoch": 0.0008448, "grad_norm": 0.8269469738006592, "learning_rate": 3.451287541285718e-06, "loss": 2.6788, "step": 284230 }, { "epoch": 0.0008704, "grad_norm": 0.8428122401237488, "learning_rate": 3.450678980989499e-06, "loss": 2.5779, "step": 284240 }, { "epoch": 0.000896, "grad_norm": 0.7839991450309753, "learning_rate": 3.450070463163886e-06, "loss": 2.6535, "step": 284250 }, { "epoch": 0.0009216, "grad_norm": 0.8470325469970703, "learning_rate": 3.4494619878128243e-06, "loss": 2.7714, "step": 284260 }, { "epoch": 0.0009472, "grad_norm": 0.7446955442428589, "learning_rate": 3.4488535549402625e-06, "loss": 2.5208, "step": 284270 }, { "epoch": 0.0009728, "grad_norm": 0.8650442957878113, "learning_rate": 3.4482451645501426e-06, "loss": 2.7127, "step": 284280 }, { "epoch": 0.0009984, "grad_norm": 0.9431222677230835, "learning_rate": 3.4476368166464157e-06, "loss": 2.7253, "step": 284290 }, { "epoch": 0.001024, "grad_norm": 0.8702569007873535, "learning_rate": 3.447028511233019e-06, "loss": 2.7038, "step": 284300 }, { "epoch": 0.0010496, "grad_norm": 0.9559741616249084, "learning_rate": 3.446420248313901e-06, "loss": 2.5927, "step": 284310 }, { "epoch": 0.0010752, "grad_norm": 0.7729634046554565, "learning_rate": 3.4458120278930063e-06, "loss": 2.5739, "step": 284320 }, { "epoch": 0.0011008, "grad_norm": 0.7765179872512817, "learning_rate": 3.445203849974278e-06, "loss": 2.7423, "step": 284330 }, { "epoch": 0.0011264, "grad_norm": 0.8300279974937439, "learning_rate": 3.4445957145616615e-06, "loss": 2.478, "step": 284340 }, { "epoch": 0.001152, "grad_norm": 0.8955919146537781, "learning_rate": 3.4439876216590982e-06, "loss": 2.7749, "step": 284350 }, { "epoch": 0.0011776, "grad_norm": 1.118276596069336, "learning_rate": 3.4433795712705343e-06, "loss": 2.9432, "step": 284360 }, { "epoch": 0.0012032, "grad_norm": 0.9436737298965454, "learning_rate": 3.4427715633999137e-06, "loss": 2.464, "step": 284370 }, { "epoch": 0.0012288, "grad_norm": 0.7708812952041626, "learning_rate": 3.4421635980511736e-06, "loss": 2.8932, "step": 284380 }, { "epoch": 0.0012544, "grad_norm": 0.845844566822052, "learning_rate": 3.4415556752282598e-06, "loss": 2.7212, "step": 284390 }, { "epoch": 0.00128, "grad_norm": 0.8724701404571533, "learning_rate": 3.440947794935111e-06, "loss": 2.7019, "step": 284400 }, { "epoch": 0.0013056, "grad_norm": 0.8417163491249084, "learning_rate": 3.4403399571756767e-06, "loss": 2.668, "step": 284410 }, { "epoch": 0.0013312, "grad_norm": 0.8875302672386169, "learning_rate": 3.439732161953897e-06, "loss": 2.6209, "step": 284420 }, { "epoch": 0.0013568, "grad_norm": 0.7940436601638794, "learning_rate": 3.4391244092737074e-06, "loss": 2.7941, "step": 284430 }, { "epoch": 0.0013824, "grad_norm": 0.9629684686660767, "learning_rate": 3.4385166991390526e-06, "loss": 2.6983, "step": 284440 }, { "epoch": 0.001408, "grad_norm": 0.7800756692886353, "learning_rate": 3.437909031553873e-06, "loss": 2.6748, "step": 284450 }, { "epoch": 0.0014336, "grad_norm": 0.9031720161437988, "learning_rate": 3.43730140652211e-06, "loss": 2.7985, "step": 284460 }, { "epoch": 0.0014592, "grad_norm": 0.7452176213264465, "learning_rate": 3.436693824047702e-06, "loss": 2.462, "step": 284470 }, { "epoch": 0.0014848, "grad_norm": 1.0023212432861328, "learning_rate": 3.436086284134591e-06, "loss": 2.2745, "step": 284480 }, { "epoch": 0.0015104, "grad_norm": 0.9499115943908691, "learning_rate": 3.435478786786719e-06, "loss": 2.1973, "step": 284490 }, { "epoch": 0.001536, "grad_norm": 0.8923957347869873, "learning_rate": 3.4348713320080194e-06, "loss": 2.5587, "step": 284500 }, { "epoch": 2.56e-05, "grad_norm": 1.309242844581604, "learning_rate": 3.434263919802434e-06, "loss": 1.3254, "step": 284510 }, { "epoch": 5.12e-05, "grad_norm": 0.8580725193023682, "learning_rate": 3.4336565501739026e-06, "loss": 2.8248, "step": 284520 }, { "epoch": 7.68e-05, "grad_norm": 0.8423535823822021, "learning_rate": 3.4330492231263633e-06, "loss": 2.5956, "step": 284530 }, { "epoch": 0.0001024, "grad_norm": 0.7513912320137024, "learning_rate": 3.4324419386637543e-06, "loss": 2.5411, "step": 284540 }, { "epoch": 0.000128, "grad_norm": 0.8404826521873474, "learning_rate": 3.431834696790014e-06, "loss": 2.8235, "step": 284550 }, { "epoch": 0.0001536, "grad_norm": 0.8728119134902954, "learning_rate": 3.43122749750908e-06, "loss": 2.5403, "step": 284560 }, { "epoch": 0.0001792, "grad_norm": 0.8500999808311462, "learning_rate": 3.4306203408248893e-06, "loss": 2.8833, "step": 284570 }, { "epoch": 0.0002048, "grad_norm": 0.7898193001747131, "learning_rate": 3.430013226741381e-06, "loss": 2.4727, "step": 284580 }, { "epoch": 0.0002304, "grad_norm": 0.9413917064666748, "learning_rate": 3.42940615526249e-06, "loss": 2.6487, "step": 284590 }, { "epoch": 0.000256, "grad_norm": 1.001663327217102, "learning_rate": 3.428799126392154e-06, "loss": 2.5665, "step": 284600 }, { "epoch": 0.0002816, "grad_norm": 0.8171842694282532, "learning_rate": 3.428192140134309e-06, "loss": 2.5206, "step": 284610 }, { "epoch": 0.0003072, "grad_norm": 0.8461750149726868, "learning_rate": 3.4275851964928953e-06, "loss": 2.3617, "step": 284620 }, { "epoch": 0.0003328, "grad_norm": 1.0124980211257935, "learning_rate": 3.426978295471841e-06, "loss": 2.3184, "step": 284630 }, { "epoch": 0.0003584, "grad_norm": 0.9383437633514404, "learning_rate": 3.4263714370750856e-06, "loss": 2.7669, "step": 284640 }, { "epoch": 0.000384, "grad_norm": 0.8627045154571533, "learning_rate": 3.4257646213065643e-06, "loss": 2.6309, "step": 284650 }, { "epoch": 0.0004096, "grad_norm": 1.012940764427185, "learning_rate": 3.425157848170213e-06, "loss": 2.8785, "step": 284660 }, { "epoch": 0.0004352, "grad_norm": 0.921003520488739, "learning_rate": 3.424551117669965e-06, "loss": 2.6263, "step": 284670 }, { "epoch": 0.0004608, "grad_norm": 0.8668383359909058, "learning_rate": 3.4239444298097555e-06, "loss": 2.707, "step": 284680 }, { "epoch": 0.0004864, "grad_norm": 0.7812322974205017, "learning_rate": 3.423337784593522e-06, "loss": 2.6027, "step": 284690 }, { "epoch": 0.000512, "grad_norm": 0.8484175205230713, "learning_rate": 3.4227311820251917e-06, "loss": 2.7295, "step": 284700 }, { "epoch": 0.0005376, "grad_norm": 0.834001898765564, "learning_rate": 3.4221246221087023e-06, "loss": 2.5207, "step": 284710 }, { "epoch": 0.0005632, "grad_norm": 0.8847063183784485, "learning_rate": 3.421518104847986e-06, "loss": 2.5031, "step": 284720 }, { "epoch": 0.0005888, "grad_norm": 0.8322683572769165, "learning_rate": 3.420911630246976e-06, "loss": 2.6179, "step": 284730 }, { "epoch": 0.0006144, "grad_norm": 0.8604039549827576, "learning_rate": 3.420305198309606e-06, "loss": 2.432, "step": 284740 }, { "epoch": 0.00064, "grad_norm": 0.8305638432502747, "learning_rate": 3.4196988090398087e-06, "loss": 2.5569, "step": 284750 }, { "epoch": 0.0006656, "grad_norm": 0.8133777379989624, "learning_rate": 3.4190924624415146e-06, "loss": 2.7605, "step": 284760 }, { "epoch": 0.0006912, "grad_norm": 0.822979748249054, "learning_rate": 3.418486158518658e-06, "loss": 2.782, "step": 284770 }, { "epoch": 0.0007168, "grad_norm": 0.8258178234100342, "learning_rate": 3.4178798972751694e-06, "loss": 2.5825, "step": 284780 }, { "epoch": 0.0007424, "grad_norm": 0.8196871280670166, "learning_rate": 3.4172736787149796e-06, "loss": 3.0399, "step": 284790 }, { "epoch": 0.000768, "grad_norm": 0.8272944092750549, "learning_rate": 3.4166675028420205e-06, "loss": 2.7521, "step": 284800 }, { "epoch": 0.0007936, "grad_norm": 0.8327066898345947, "learning_rate": 3.4160613696602242e-06, "loss": 2.7654, "step": 284810 }, { "epoch": 0.0008192, "grad_norm": 0.8338490128517151, "learning_rate": 3.4154552791735217e-06, "loss": 2.6277, "step": 284820 }, { "epoch": 0.0008448, "grad_norm": 0.847760796546936, "learning_rate": 3.4148492313858395e-06, "loss": 2.646, "step": 284830 }, { "epoch": 0.0008704, "grad_norm": 0.8800761103630066, "learning_rate": 3.4142432263011106e-06, "loss": 2.4967, "step": 284840 }, { "epoch": 0.000896, "grad_norm": 0.8693573474884033, "learning_rate": 3.413637263923263e-06, "loss": 2.5748, "step": 284850 }, { "epoch": 0.0009216, "grad_norm": 0.8384279608726501, "learning_rate": 3.413031344256227e-06, "loss": 2.7754, "step": 284860 }, { "epoch": 0.0009472, "grad_norm": 0.7840909957885742, "learning_rate": 3.4124254673039327e-06, "loss": 2.4739, "step": 284870 }, { "epoch": 0.0009728, "grad_norm": 0.8168819546699524, "learning_rate": 3.411819633070309e-06, "loss": 2.4348, "step": 284880 }, { "epoch": 0.0009984, "grad_norm": 0.9122936129570007, "learning_rate": 3.4112138415592867e-06, "loss": 2.6842, "step": 284890 }, { "epoch": 0.001024, "grad_norm": 1.1181925535202026, "learning_rate": 3.410608092774789e-06, "loss": 2.8242, "step": 284900 }, { "epoch": 0.0010496, "grad_norm": 0.9088864326477051, "learning_rate": 3.4100023867207454e-06, "loss": 2.5638, "step": 284910 }, { "epoch": 0.0010752, "grad_norm": 0.8110350966453552, "learning_rate": 3.4093967234010815e-06, "loss": 2.884, "step": 284920 }, { "epoch": 0.0011008, "grad_norm": 0.7910068035125732, "learning_rate": 3.408791102819733e-06, "loss": 2.8008, "step": 284930 }, { "epoch": 0.0011264, "grad_norm": 0.8648487329483032, "learning_rate": 3.4081855249806217e-06, "loss": 2.6962, "step": 284940 }, { "epoch": 0.001152, "grad_norm": 0.9028014540672302, "learning_rate": 3.407579989887678e-06, "loss": 2.5109, "step": 284950 }, { "epoch": 0.0011776, "grad_norm": 0.8836267590522766, "learning_rate": 3.4069744975448237e-06, "loss": 2.8717, "step": 284960 }, { "epoch": 0.0012032, "grad_norm": 0.961950421333313, "learning_rate": 3.4063690479559874e-06, "loss": 2.9031, "step": 284970 }, { "epoch": 0.0012288, "grad_norm": 0.8221468329429626, "learning_rate": 3.405763641125095e-06, "loss": 2.5842, "step": 284980 }, { "epoch": 0.0012544, "grad_norm": 0.7599897980690002, "learning_rate": 3.4051582770560733e-06, "loss": 2.5223, "step": 284990 }, { "epoch": 0.00128, "grad_norm": 1.0169248580932617, "learning_rate": 3.404552955752847e-06, "loss": 2.7147, "step": 285000 }, { "epoch": 0.0013056, "grad_norm": 0.8767223954200745, "learning_rate": 3.4039476772193423e-06, "loss": 2.7164, "step": 285010 }, { "epoch": 0.0013312, "grad_norm": 1.0004854202270508, "learning_rate": 3.403342441459487e-06, "loss": 2.5944, "step": 285020 }, { "epoch": 0.0013568, "grad_norm": 0.8276439309120178, "learning_rate": 3.4027372484772002e-06, "loss": 2.7938, "step": 285030 }, { "epoch": 0.0013824, "grad_norm": 0.9193512797355652, "learning_rate": 3.402132098276408e-06, "loss": 2.8068, "step": 285040 }, { "epoch": 0.001408, "grad_norm": 0.7913495898246765, "learning_rate": 3.401526990861036e-06, "loss": 2.5834, "step": 285050 }, { "epoch": 0.0014336, "grad_norm": 0.9167509078979492, "learning_rate": 3.400921926235008e-06, "loss": 2.5531, "step": 285060 }, { "epoch": 0.0014592, "grad_norm": 0.8899197578430176, "learning_rate": 3.4003169044022477e-06, "loss": 2.5113, "step": 285070 }, { "epoch": 0.0014848, "grad_norm": 0.7487753033638, "learning_rate": 3.399711925366678e-06, "loss": 2.3821, "step": 285080 }, { "epoch": 0.0015104, "grad_norm": 0.9974222183227539, "learning_rate": 3.399106989132225e-06, "loss": 2.2327, "step": 285090 }, { "epoch": 0.001536, "grad_norm": 1.0164154767990112, "learning_rate": 3.3985020957028025e-06, "loss": 2.683, "step": 285100 }, { "epoch": 0.0015616, "grad_norm": 0.7750421762466431, "learning_rate": 3.3978972450823432e-06, "loss": 2.5269, "step": 285110 }, { "epoch": 0.0015872, "grad_norm": 1.0632905960083008, "learning_rate": 3.397292437274766e-06, "loss": 2.7375, "step": 285120 }, { "epoch": 0.0016128, "grad_norm": 0.9493259787559509, "learning_rate": 3.3966876722839913e-06, "loss": 2.6967, "step": 285130 }, { "epoch": 0.0016384, "grad_norm": 0.9439873695373535, "learning_rate": 3.3960829501139435e-06, "loss": 2.7638, "step": 285140 }, { "epoch": 0.001664, "grad_norm": 0.8575742840766907, "learning_rate": 3.3954782707685453e-06, "loss": 2.681, "step": 285150 }, { "epoch": 0.0016896, "grad_norm": 1.0231815576553345, "learning_rate": 3.394873634251712e-06, "loss": 2.5941, "step": 285160 }, { "epoch": 0.0017152, "grad_norm": 0.8011809587478638, "learning_rate": 3.3942690405673674e-06, "loss": 2.7186, "step": 285170 }, { "epoch": 0.0017408, "grad_norm": 0.8403463959693909, "learning_rate": 3.3936644897194326e-06, "loss": 2.8188, "step": 285180 }, { "epoch": 0.0017664, "grad_norm": 0.8540915846824646, "learning_rate": 3.393059981711828e-06, "loss": 2.7776, "step": 285190 }, { "epoch": 0.001792, "grad_norm": 0.9969847202301025, "learning_rate": 3.392455516548474e-06, "loss": 2.4751, "step": 285200 }, { "epoch": 0.0018176, "grad_norm": 0.9352418184280396, "learning_rate": 3.3918510942332893e-06, "loss": 2.6182, "step": 285210 }, { "epoch": 0.0018432, "grad_norm": 0.8226258754730225, "learning_rate": 3.391246714770198e-06, "loss": 2.5872, "step": 285220 }, { "epoch": 0.0018688, "grad_norm": 0.83817458152771, "learning_rate": 3.3906423781631106e-06, "loss": 2.7615, "step": 285230 }, { "epoch": 0.0018944, "grad_norm": 0.7867002487182617, "learning_rate": 3.390038084415952e-06, "loss": 2.5874, "step": 285240 }, { "epoch": 0.00192, "grad_norm": 0.7938541769981384, "learning_rate": 3.3894338335326392e-06, "loss": 2.781, "step": 285250 }, { "epoch": 0.0019456, "grad_norm": 0.7377727627754211, "learning_rate": 3.3888296255170914e-06, "loss": 2.5504, "step": 285260 }, { "epoch": 0.0019712, "grad_norm": 0.7357239723205566, "learning_rate": 3.3882254603732235e-06, "loss": 2.4733, "step": 285270 }, { "epoch": 0.0019968, "grad_norm": 0.8692814707756042, "learning_rate": 3.3876213381049614e-06, "loss": 2.9052, "step": 285280 }, { "epoch": 0.0020224, "grad_norm": 0.8491354584693909, "learning_rate": 3.3870172587162155e-06, "loss": 2.7973, "step": 285290 }, { "epoch": 0.002048, "grad_norm": 0.8137177228927612, "learning_rate": 3.386413222210906e-06, "loss": 2.9752, "step": 285300 }, { "epoch": 0.0020736, "grad_norm": 0.824548065662384, "learning_rate": 3.385809228592948e-06, "loss": 2.6009, "step": 285310 }, { "epoch": 0.0020992, "grad_norm": 0.7353882193565369, "learning_rate": 3.385205277866259e-06, "loss": 2.6172, "step": 285320 }, { "epoch": 0.0021248, "grad_norm": 0.8058502078056335, "learning_rate": 3.3846013700347557e-06, "loss": 2.7551, "step": 285330 }, { "epoch": 0.0021504, "grad_norm": 0.8585501313209534, "learning_rate": 3.3839975051023553e-06, "loss": 2.8761, "step": 285340 }, { "epoch": 0.002176, "grad_norm": 0.8613325953483582, "learning_rate": 3.3833936830729754e-06, "loss": 2.7745, "step": 285350 }, { "epoch": 0.0022016, "grad_norm": 0.9170626401901245, "learning_rate": 3.382789903950525e-06, "loss": 2.69, "step": 285360 }, { "epoch": 0.0022272, "grad_norm": 0.9396491646766663, "learning_rate": 3.3821861677389243e-06, "loss": 2.8095, "step": 285370 }, { "epoch": 0.0022528, "grad_norm": 0.7763006091117859, "learning_rate": 3.3815824744420878e-06, "loss": 2.6532, "step": 285380 }, { "epoch": 0.0022784, "grad_norm": 0.8541542291641235, "learning_rate": 3.3809788240639296e-06, "loss": 2.7265, "step": 285390 }, { "epoch": 0.002304, "grad_norm": 0.9258832931518555, "learning_rate": 3.3803752166083646e-06, "loss": 2.7912, "step": 285400 }, { "epoch": 0.0023296, "grad_norm": 0.8217238187789917, "learning_rate": 3.3797716520793076e-06, "loss": 2.5876, "step": 285410 }, { "epoch": 0.0023552, "grad_norm": 0.774933397769928, "learning_rate": 3.3791681304806747e-06, "loss": 2.6501, "step": 285420 }, { "epoch": 0.0023808, "grad_norm": 0.8033272624015808, "learning_rate": 3.378564651816374e-06, "loss": 2.5461, "step": 285430 }, { "epoch": 0.0024064, "grad_norm": 0.8776481747627258, "learning_rate": 3.3779612160903174e-06, "loss": 2.7217, "step": 285440 }, { "epoch": 0.002432, "grad_norm": 0.8586705923080444, "learning_rate": 3.3773578233064265e-06, "loss": 2.8247, "step": 285450 }, { "epoch": 0.0024576, "grad_norm": 0.8927343487739563, "learning_rate": 3.37675447346861e-06, "loss": 2.6464, "step": 285460 }, { "epoch": 0.0024832, "grad_norm": 0.9068018198013306, "learning_rate": 3.37615116658078e-06, "loss": 2.9004, "step": 285470 }, { "epoch": 0.0025088, "grad_norm": 0.8184506297111511, "learning_rate": 3.375547902646853e-06, "loss": 2.493, "step": 285480 }, { "epoch": 0.0025344, "grad_norm": 0.881422221660614, "learning_rate": 3.3749446816707343e-06, "loss": 2.8631, "step": 285490 }, { "epoch": 0.00256, "grad_norm": 0.8203944563865662, "learning_rate": 3.3743415036563386e-06, "loss": 2.6881, "step": 285500 }, { "epoch": 0.0025856, "grad_norm": 0.7584049105644226, "learning_rate": 3.3737383686075766e-06, "loss": 2.8488, "step": 285510 }, { "epoch": 0.0026112, "grad_norm": 0.7836775183677673, "learning_rate": 3.373135276528361e-06, "loss": 2.6426, "step": 285520 }, { "epoch": 0.0026368, "grad_norm": 1.0149520635604858, "learning_rate": 3.3725322274226013e-06, "loss": 3.0054, "step": 285530 }, { "epoch": 0.0026624, "grad_norm": 0.8459898829460144, "learning_rate": 3.371929221294209e-06, "loss": 2.892, "step": 285540 }, { "epoch": 0.002688, "grad_norm": 2.097468852996826, "learning_rate": 3.3713262581470974e-06, "loss": 2.7602, "step": 285550 }, { "epoch": 0.0027136, "grad_norm": 0.9276173114776611, "learning_rate": 3.37072333798517e-06, "loss": 2.8403, "step": 285560 }, { "epoch": 0.0027392, "grad_norm": 0.8931220769882202, "learning_rate": 3.3701204608123404e-06, "loss": 2.4242, "step": 285570 }, { "epoch": 0.0027648, "grad_norm": 0.8440992832183838, "learning_rate": 3.369517626632517e-06, "loss": 3.046, "step": 285580 }, { "epoch": 0.0027904, "grad_norm": 0.8467938303947449, "learning_rate": 3.3689148354496104e-06, "loss": 2.9342, "step": 285590 }, { "epoch": 0.002816, "grad_norm": 0.7806288599967957, "learning_rate": 3.368312087267528e-06, "loss": 2.8469, "step": 285600 }, { "epoch": 0.0028416, "grad_norm": 0.7954725027084351, "learning_rate": 3.3677093820901797e-06, "loss": 2.7646, "step": 285610 }, { "epoch": 0.0028672, "grad_norm": 0.81952303647995, "learning_rate": 3.3671067199214726e-06, "loss": 2.8519, "step": 285620 }, { "epoch": 0.0028928, "grad_norm": 0.8053986430168152, "learning_rate": 3.3665041007653166e-06, "loss": 2.8119, "step": 285630 }, { "epoch": 0.0029184, "grad_norm": 0.7687223553657532, "learning_rate": 3.3659015246256175e-06, "loss": 2.9008, "step": 285640 }, { "epoch": 0.002944, "grad_norm": 0.9548282027244568, "learning_rate": 3.365298991506284e-06, "loss": 2.9484, "step": 285650 }, { "epoch": 0.0029696, "grad_norm": 0.8150346279144287, "learning_rate": 3.3646965014112244e-06, "loss": 2.8197, "step": 285660 }, { "epoch": 0.0029952, "grad_norm": 0.8832091689109802, "learning_rate": 3.3640940543443434e-06, "loss": 2.7403, "step": 285670 }, { "epoch": 0.0030208, "grad_norm": 0.7982509732246399, "learning_rate": 3.363491650309553e-06, "loss": 2.8799, "step": 285680 }, { "epoch": 0.0030464, "grad_norm": 0.7518913745880127, "learning_rate": 3.362889289310751e-06, "loss": 2.8043, "step": 285690 }, { "epoch": 0.003072, "grad_norm": 0.7463252544403076, "learning_rate": 3.3622869713518493e-06, "loss": 2.5116, "step": 285700 }, { "epoch": 0.0030976, "grad_norm": 0.8680075407028198, "learning_rate": 3.361684696436752e-06, "loss": 3.0108, "step": 285710 }, { "epoch": 0.0031232, "grad_norm": 0.7714858651161194, "learning_rate": 3.361082464569365e-06, "loss": 2.7952, "step": 285720 }, { "epoch": 0.0031488, "grad_norm": 0.9068840146064758, "learning_rate": 3.3604802757535937e-06, "loss": 2.9506, "step": 285730 }, { "epoch": 0.0031744, "grad_norm": 0.7638298869132996, "learning_rate": 3.3598781299933435e-06, "loss": 2.8017, "step": 285740 }, { "epoch": 0.0032, "grad_norm": 0.8082290291786194, "learning_rate": 3.359276027292523e-06, "loss": 3.0017, "step": 285750 }, { "epoch": 0.0032256, "grad_norm": 0.7977892756462097, "learning_rate": 3.358673967655028e-06, "loss": 2.8591, "step": 285760 }, { "epoch": 0.0032512, "grad_norm": 0.81370609998703, "learning_rate": 3.358071951084768e-06, "loss": 2.7939, "step": 285770 }, { "epoch": 0.0032768, "grad_norm": 0.7771335244178772, "learning_rate": 3.3574699775856457e-06, "loss": 2.9075, "step": 285780 }, { "epoch": 0.0033024, "grad_norm": 0.90948885679245, "learning_rate": 3.356868047161562e-06, "loss": 3.0222, "step": 285790 }, { "epoch": 0.003328, "grad_norm": 0.8108780384063721, "learning_rate": 3.35626615981643e-06, "loss": 2.5783, "step": 285800 }, { "epoch": 0.0033536, "grad_norm": 0.8120585680007935, "learning_rate": 3.3556643155541435e-06, "loss": 2.7451, "step": 285810 }, { "epoch": 0.0033792, "grad_norm": 0.7477201819419861, "learning_rate": 3.3550625143786077e-06, "loss": 2.6364, "step": 285820 }, { "epoch": 0.0034048, "grad_norm": 0.9997864365577698, "learning_rate": 3.354460756293726e-06, "loss": 2.7016, "step": 285830 }, { "epoch": 0.0034304, "grad_norm": 0.9141011238098145, "learning_rate": 3.353859041303399e-06, "loss": 2.7683, "step": 285840 }, { "epoch": 0.003456, "grad_norm": 0.8585475087165833, "learning_rate": 3.353257369411531e-06, "loss": 2.7538, "step": 285850 }, { "epoch": 0.0034816, "grad_norm": 0.9957593679428101, "learning_rate": 3.3526557406220217e-06, "loss": 2.9565, "step": 285860 }, { "epoch": 0.0035072, "grad_norm": 0.842418909072876, "learning_rate": 3.352054154938773e-06, "loss": 2.5736, "step": 285870 }, { "epoch": 0.0035328, "grad_norm": 0.79677414894104, "learning_rate": 3.351452612365691e-06, "loss": 2.7127, "step": 285880 }, { "epoch": 0.0035584, "grad_norm": 1.049203872680664, "learning_rate": 3.3508511129066667e-06, "loss": 2.8848, "step": 285890 }, { "epoch": 0.003584, "grad_norm": 0.832090437412262, "learning_rate": 3.3502496565656074e-06, "loss": 2.8099, "step": 285900 }, { "epoch": 0.0036096, "grad_norm": 0.7346471548080444, "learning_rate": 3.3496482433464105e-06, "loss": 2.5032, "step": 285910 }, { "epoch": 0.0036352, "grad_norm": 0.7732651829719543, "learning_rate": 3.349046873252978e-06, "loss": 2.9671, "step": 285920 }, { "epoch": 0.0036608, "grad_norm": 0.8306416273117065, "learning_rate": 3.3484455462892095e-06, "loss": 2.9287, "step": 285930 }, { "epoch": 0.0036864, "grad_norm": 0.8959001302719116, "learning_rate": 3.3478442624590023e-06, "loss": 2.7952, "step": 285940 }, { "epoch": 0.003712, "grad_norm": 0.8401502966880798, "learning_rate": 3.347243021766262e-06, "loss": 2.8885, "step": 285950 }, { "epoch": 0.0037376, "grad_norm": 0.8025455474853516, "learning_rate": 3.346641824214878e-06, "loss": 2.9934, "step": 285960 }, { "epoch": 0.0037632, "grad_norm": 0.8872770071029663, "learning_rate": 3.346040669808751e-06, "loss": 2.881, "step": 285970 }, { "epoch": 0.0037888, "grad_norm": 0.7755155563354492, "learning_rate": 3.345439558551785e-06, "loss": 2.7822, "step": 285980 }, { "epoch": 0.0038144, "grad_norm": 0.8184744119644165, "learning_rate": 3.344838490447875e-06, "loss": 2.6428, "step": 285990 }, { "epoch": 0.00384, "grad_norm": 0.7959297299385071, "learning_rate": 3.3442374655009213e-06, "loss": 2.6215, "step": 286000 }, { "epoch": 0.0038656, "grad_norm": 0.8617013096809387, "learning_rate": 3.343636483714816e-06, "loss": 2.9426, "step": 286010 }, { "epoch": 0.0038912, "grad_norm": 0.7863832712173462, "learning_rate": 3.343035545093459e-06, "loss": 2.8965, "step": 286020 }, { "epoch": 0.0039168, "grad_norm": 0.7365179061889648, "learning_rate": 3.342434649640748e-06, "loss": 2.7169, "step": 286030 }, { "epoch": 0.0039424, "grad_norm": 0.8296220302581787, "learning_rate": 3.3418337973605786e-06, "loss": 2.93, "step": 286040 }, { "epoch": 0.003968, "grad_norm": 0.9389696717262268, "learning_rate": 3.3412329882568463e-06, "loss": 2.589, "step": 286050 }, { "epoch": 0.0039936, "grad_norm": 0.9611747860908508, "learning_rate": 3.34063222233345e-06, "loss": 2.9248, "step": 286060 }, { "epoch": 0.0040192, "grad_norm": 0.8653222322463989, "learning_rate": 3.3400314995942828e-06, "loss": 2.9101, "step": 286070 }, { "epoch": 0.0040448, "grad_norm": 0.7471975684165955, "learning_rate": 3.3394308200432446e-06, "loss": 2.854, "step": 286080 }, { "epoch": 0.0040704, "grad_norm": 0.8660579323768616, "learning_rate": 3.338830183684224e-06, "loss": 2.8139, "step": 286090 }, { "epoch": 0.004096, "grad_norm": 1.371803641319275, "learning_rate": 3.3382295905211206e-06, "loss": 2.994, "step": 286100 }, { "epoch": 0.0041216, "grad_norm": 2.812638282775879, "learning_rate": 3.337629040557826e-06, "loss": 2.9457, "step": 286110 }, { "epoch": 0.0041472, "grad_norm": 0.8364810347557068, "learning_rate": 3.337028533798238e-06, "loss": 2.7608, "step": 286120 }, { "epoch": 0.0041728, "grad_norm": 0.9596352577209473, "learning_rate": 3.336428070246248e-06, "loss": 2.8606, "step": 286130 }, { "epoch": 0.0041984, "grad_norm": 0.9414204359054565, "learning_rate": 3.335827649905752e-06, "loss": 2.933, "step": 286140 }, { "epoch": 0.004224, "grad_norm": 0.8298652768135071, "learning_rate": 3.335227272780642e-06, "loss": 2.8926, "step": 286150 }, { "epoch": 0.0042496, "grad_norm": 0.8328872919082642, "learning_rate": 3.334626938874812e-06, "loss": 2.7687, "step": 286160 }, { "epoch": 0.0042752, "grad_norm": 0.8836736679077148, "learning_rate": 3.334026648192155e-06, "loss": 2.9738, "step": 286170 }, { "epoch": 0.0043008, "grad_norm": 0.9331100583076477, "learning_rate": 3.3334264007365635e-06, "loss": 2.8881, "step": 286180 }, { "epoch": 0.0043264, "grad_norm": 0.7452616691589355, "learning_rate": 3.332826196511929e-06, "loss": 2.9266, "step": 286190 }, { "epoch": 0.004352, "grad_norm": 1.0156855583190918, "learning_rate": 3.3322260355221494e-06, "loss": 2.9294, "step": 286200 }, { "epoch": 0.0043776, "grad_norm": 0.8303204774856567, "learning_rate": 3.3316259177711073e-06, "loss": 2.772, "step": 286210 }, { "epoch": 0.0044032, "grad_norm": 1.64179265499115, "learning_rate": 3.3310258432627007e-06, "loss": 3.0584, "step": 286220 }, { "epoch": 0.0044288, "grad_norm": 0.7797567844390869, "learning_rate": 3.330425812000818e-06, "loss": 2.7844, "step": 286230 }, { "epoch": 0.0044544, "grad_norm": 0.7993584871292114, "learning_rate": 3.329825823989351e-06, "loss": 2.7674, "step": 286240 }, { "epoch": 0.00448, "grad_norm": 0.7861626744270325, "learning_rate": 3.3292258792321906e-06, "loss": 2.8052, "step": 286250 }, { "epoch": 0.0045056, "grad_norm": 0.9700416922569275, "learning_rate": 3.3286259777332284e-06, "loss": 2.9336, "step": 286260 }, { "epoch": 0.0045312, "grad_norm": 0.8681989908218384, "learning_rate": 3.328026119496357e-06, "loss": 2.6555, "step": 286270 }, { "epoch": 0.0045568, "grad_norm": 0.8344413638114929, "learning_rate": 3.3274263045254597e-06, "loss": 2.7375, "step": 286280 }, { "epoch": 0.0045824, "grad_norm": 0.897700846195221, "learning_rate": 3.3268265328244288e-06, "loss": 2.8936, "step": 286290 }, { "epoch": 0.004608, "grad_norm": 0.911754310131073, "learning_rate": 3.3262268043971545e-06, "loss": 2.7982, "step": 286300 }, { "epoch": 0.0046336, "grad_norm": 0.8652032613754272, "learning_rate": 3.325627119247523e-06, "loss": 3.0605, "step": 286310 }, { "epoch": 0.0046592, "grad_norm": 0.8714678287506104, "learning_rate": 3.325027477379429e-06, "loss": 2.9604, "step": 286320 }, { "epoch": 0.0046848, "grad_norm": 0.7695733308792114, "learning_rate": 3.324427878796761e-06, "loss": 2.7204, "step": 286330 }, { "epoch": 0.0047104, "grad_norm": 0.8630520105361938, "learning_rate": 3.3238283235033995e-06, "loss": 2.8688, "step": 286340 }, { "epoch": 0.004736, "grad_norm": 0.8254885077476501, "learning_rate": 3.3232288115032386e-06, "loss": 2.7043, "step": 286350 }, { "epoch": 0.0047616, "grad_norm": 1.0025615692138672, "learning_rate": 3.322629342800164e-06, "loss": 2.7066, "step": 286360 }, { "epoch": 0.0047872, "grad_norm": 0.8319041728973389, "learning_rate": 3.322029917398063e-06, "loss": 2.8096, "step": 286370 }, { "epoch": 0.0048128, "grad_norm": 0.9845414757728577, "learning_rate": 3.3214305353008237e-06, "loss": 2.8366, "step": 286380 }, { "epoch": 0.0048384, "grad_norm": 0.9912234544754028, "learning_rate": 3.320831196512332e-06, "loss": 2.8117, "step": 286390 }, { "epoch": 0.004864, "grad_norm": 0.8126397132873535, "learning_rate": 3.320231901036479e-06, "loss": 2.8798, "step": 286400 }, { "epoch": 0.0048896, "grad_norm": 0.79722660779953, "learning_rate": 3.3196326488771422e-06, "loss": 2.7519, "step": 286410 }, { "epoch": 0.0049152, "grad_norm": 0.7693269848823547, "learning_rate": 3.319033440038213e-06, "loss": 2.8417, "step": 286420 }, { "epoch": 0.0049408, "grad_norm": 0.8097338676452637, "learning_rate": 3.3184342745235764e-06, "loss": 2.7698, "step": 286430 }, { "epoch": 0.0049664, "grad_norm": 0.7586210370063782, "learning_rate": 3.3178351523371177e-06, "loss": 2.7315, "step": 286440 }, { "epoch": 0.004992, "grad_norm": 0.7799363136291504, "learning_rate": 3.3172360734827225e-06, "loss": 2.722, "step": 286450 }, { "epoch": 0.0050176, "grad_norm": 0.7393247485160828, "learning_rate": 3.3166370379642756e-06, "loss": 2.6145, "step": 286460 }, { "epoch": 0.0050432, "grad_norm": 1.221310019493103, "learning_rate": 3.3160380457856632e-06, "loss": 2.8432, "step": 286470 }, { "epoch": 0.0050688, "grad_norm": 0.7272070050239563, "learning_rate": 3.3154390969507657e-06, "loss": 2.7353, "step": 286480 }, { "epoch": 0.0050944, "grad_norm": 0.8295924067497253, "learning_rate": 3.3148401914634653e-06, "loss": 2.8078, "step": 286490 }, { "epoch": 0.00512, "grad_norm": 0.783488929271698, "learning_rate": 3.314241329327652e-06, "loss": 2.7756, "step": 286500 }, { "epoch": 0.0051456, "grad_norm": 0.8873987197875977, "learning_rate": 3.313642510547208e-06, "loss": 2.7733, "step": 286510 }, { "epoch": 0.0051712, "grad_norm": 0.8049460649490356, "learning_rate": 3.3130437351260146e-06, "loss": 2.7596, "step": 286520 }, { "epoch": 0.0051968, "grad_norm": 0.8186836242675781, "learning_rate": 3.31244500306796e-06, "loss": 2.8482, "step": 286530 }, { "epoch": 0.0052224, "grad_norm": 0.7769724130630493, "learning_rate": 3.3118463143769177e-06, "loss": 2.7634, "step": 286540 }, { "epoch": 0.005248, "grad_norm": 0.7514668107032776, "learning_rate": 3.311247669056775e-06, "loss": 2.6489, "step": 286550 }, { "epoch": 0.0052736, "grad_norm": 0.8817275166511536, "learning_rate": 3.310649067111413e-06, "loss": 2.7805, "step": 286560 }, { "epoch": 0.0052992, "grad_norm": 0.8305990099906921, "learning_rate": 3.3100505085447145e-06, "loss": 2.6897, "step": 286570 }, { "epoch": 0.0053248, "grad_norm": 0.7604116201400757, "learning_rate": 3.3094519933605595e-06, "loss": 2.8008, "step": 286580 }, { "epoch": 0.0053504, "grad_norm": 0.752128541469574, "learning_rate": 3.3088535215628314e-06, "loss": 2.7802, "step": 286590 }, { "epoch": 0.005376, "grad_norm": 0.8355734348297119, "learning_rate": 3.3082550931554125e-06, "loss": 2.8839, "step": 286600 }, { "epoch": 0.0054016, "grad_norm": 0.8923419713973999, "learning_rate": 3.3076567081421785e-06, "loss": 2.9222, "step": 286610 }, { "epoch": 0.0054272, "grad_norm": 0.9856265187263489, "learning_rate": 3.3070583665270116e-06, "loss": 2.7998, "step": 286620 }, { "epoch": 0.0054528, "grad_norm": 0.714671790599823, "learning_rate": 3.3064600683137914e-06, "loss": 2.6475, "step": 286630 }, { "epoch": 0.0054784, "grad_norm": 1.1587519645690918, "learning_rate": 3.3058618135064e-06, "loss": 2.8496, "step": 286640 }, { "epoch": 0.005504, "grad_norm": 0.8023975491523743, "learning_rate": 3.305263602108716e-06, "loss": 2.8684, "step": 286650 }, { "epoch": 0.0055296, "grad_norm": 0.7968068718910217, "learning_rate": 3.3046654341246177e-06, "loss": 2.8283, "step": 286660 }, { "epoch": 0.0055552, "grad_norm": 2.5303847789764404, "learning_rate": 3.3040673095579835e-06, "loss": 2.6742, "step": 286670 }, { "epoch": 0.0055808, "grad_norm": 0.7889405488967896, "learning_rate": 3.3034692284126948e-06, "loss": 2.7292, "step": 286680 }, { "epoch": 0.0056064, "grad_norm": 0.8309851884841919, "learning_rate": 3.3028711906926282e-06, "loss": 2.7783, "step": 286690 }, { "epoch": 0.005632, "grad_norm": 0.7810426950454712, "learning_rate": 3.302273196401661e-06, "loss": 2.7817, "step": 286700 }, { "epoch": 0.0056576, "grad_norm": 0.7839982509613037, "learning_rate": 3.3016752455436727e-06, "loss": 2.8366, "step": 286710 }, { "epoch": 0.0056832, "grad_norm": 0.9414252638816833, "learning_rate": 3.3010773381225405e-06, "loss": 2.826, "step": 286720 }, { "epoch": 0.0057088, "grad_norm": 0.779263973236084, "learning_rate": 3.300479474142145e-06, "loss": 2.8451, "step": 286730 }, { "epoch": 0.0057344, "grad_norm": 0.9186210036277771, "learning_rate": 3.299881653606356e-06, "loss": 2.9145, "step": 286740 }, { "epoch": 0.00576, "grad_norm": 0.7929618954658508, "learning_rate": 3.2992838765190536e-06, "loss": 2.7686, "step": 286750 }, { "epoch": 0.0057856, "grad_norm": 0.9781726002693176, "learning_rate": 3.2986861428841146e-06, "loss": 2.6444, "step": 286760 }, { "epoch": 0.0058112, "grad_norm": 0.8653520941734314, "learning_rate": 3.298088452705415e-06, "loss": 2.7976, "step": 286770 }, { "epoch": 0.0058368, "grad_norm": 0.8696946501731873, "learning_rate": 3.297490805986829e-06, "loss": 2.5306, "step": 286780 }, { "epoch": 0.0058624, "grad_norm": 0.8042900562286377, "learning_rate": 3.2968932027322365e-06, "loss": 2.7805, "step": 286790 }, { "epoch": 0.005888, "grad_norm": 0.8510625958442688, "learning_rate": 3.2962956429455117e-06, "loss": 2.7896, "step": 286800 }, { "epoch": 0.0059136, "grad_norm": 0.7858009338378906, "learning_rate": 3.2956981266305256e-06, "loss": 2.6289, "step": 286810 }, { "epoch": 0.0059392, "grad_norm": 0.7928344011306763, "learning_rate": 3.2951006537911543e-06, "loss": 2.7559, "step": 286820 }, { "epoch": 0.0059648, "grad_norm": 0.7335305213928223, "learning_rate": 3.2945032244312747e-06, "loss": 2.7488, "step": 286830 }, { "epoch": 0.0059904, "grad_norm": 0.9265589714050293, "learning_rate": 3.293905838554755e-06, "loss": 2.85, "step": 286840 }, { "epoch": 0.006016, "grad_norm": 0.8491755127906799, "learning_rate": 3.2933084961654783e-06, "loss": 2.7458, "step": 286850 }, { "epoch": 0.0060416, "grad_norm": 0.983066201210022, "learning_rate": 3.2927111972673152e-06, "loss": 2.6429, "step": 286860 }, { "epoch": 0.0060672, "grad_norm": 0.8793602585792542, "learning_rate": 3.2921139418641357e-06, "loss": 2.8605, "step": 286870 }, { "epoch": 0.0060928, "grad_norm": 0.7762777209281921, "learning_rate": 3.291516729959814e-06, "loss": 2.8912, "step": 286880 }, { "epoch": 0.0061184, "grad_norm": 0.8061335682868958, "learning_rate": 3.2909195615582225e-06, "loss": 2.7763, "step": 286890 }, { "epoch": 0.006144, "grad_norm": 0.8162142634391785, "learning_rate": 3.290322436663236e-06, "loss": 2.7946, "step": 286900 }, { "epoch": 0.0061696, "grad_norm": 1.1184356212615967, "learning_rate": 3.2897253552787244e-06, "loss": 2.7966, "step": 286910 }, { "epoch": 0.0061952, "grad_norm": 0.9131284952163696, "learning_rate": 3.289128317408561e-06, "loss": 2.9248, "step": 286920 }, { "epoch": 0.0062208, "grad_norm": 0.8584364056587219, "learning_rate": 3.2885313230566197e-06, "loss": 2.6957, "step": 286930 }, { "epoch": 0.0062464, "grad_norm": 0.8798779845237732, "learning_rate": 3.287934372226765e-06, "loss": 2.9178, "step": 286940 }, { "epoch": 0.006272, "grad_norm": 0.7953574061393738, "learning_rate": 3.287337464922874e-06, "loss": 2.8359, "step": 286950 }, { "epoch": 0.0062976, "grad_norm": 0.766825795173645, "learning_rate": 3.2867406011488144e-06, "loss": 2.718, "step": 286960 }, { "epoch": 0.0063232, "grad_norm": 0.8914700746536255, "learning_rate": 3.2861437809084583e-06, "loss": 2.7411, "step": 286970 }, { "epoch": 0.0063488, "grad_norm": 0.822521448135376, "learning_rate": 3.285547004205675e-06, "loss": 2.59, "step": 286980 }, { "epoch": 0.0063744, "grad_norm": 0.8131662011146545, "learning_rate": 3.284950271044335e-06, "loss": 2.7076, "step": 286990 }, { "epoch": 0.0064, "grad_norm": 0.9557933211326599, "learning_rate": 3.284353581428311e-06, "loss": 2.7463, "step": 287000 }, { "epoch": 0.0064256, "grad_norm": 0.7797072529792786, "learning_rate": 3.283756935361463e-06, "loss": 2.6464, "step": 287010 }, { "epoch": 0.0064512, "grad_norm": 0.7889363169670105, "learning_rate": 3.283160332847669e-06, "loss": 2.6049, "step": 287020 }, { "epoch": 0.0064768, "grad_norm": 0.8121075630187988, "learning_rate": 3.2825637738907957e-06, "loss": 2.5835, "step": 287030 }, { "epoch": 0.0065024, "grad_norm": 0.7275682091712952, "learning_rate": 3.2819672584947114e-06, "loss": 2.6133, "step": 287040 }, { "epoch": 0.006528, "grad_norm": 0.9275797605514526, "learning_rate": 3.2813707866632837e-06, "loss": 2.743, "step": 287050 }, { "epoch": 0.0065536, "grad_norm": 0.7969613671302795, "learning_rate": 3.2807743584003835e-06, "loss": 2.7011, "step": 287060 }, { "epoch": 0.0065792, "grad_norm": 0.8879275918006897, "learning_rate": 3.2801779737098725e-06, "loss": 2.8959, "step": 287070 }, { "epoch": 0.0066048, "grad_norm": 0.7974385619163513, "learning_rate": 3.279581632595622e-06, "loss": 2.8042, "step": 287080 }, { "epoch": 0.0066304, "grad_norm": 0.7971316576004028, "learning_rate": 3.2789853350614985e-06, "loss": 3.0927, "step": 287090 }, { "epoch": 0.006656, "grad_norm": 0.8216460347175598, "learning_rate": 3.2783890811113695e-06, "loss": 2.5661, "step": 287100 }, { "epoch": 0.0066816, "grad_norm": 0.9465959072113037, "learning_rate": 3.2777928707491002e-06, "loss": 2.7502, "step": 287110 }, { "epoch": 0.0067072, "grad_norm": 0.7954169511795044, "learning_rate": 3.277196703978557e-06, "loss": 2.7679, "step": 287120 }, { "epoch": 0.0067328, "grad_norm": 0.8421711921691895, "learning_rate": 3.2766005808036107e-06, "loss": 2.9062, "step": 287130 }, { "epoch": 0.0067584, "grad_norm": 0.8918294310569763, "learning_rate": 3.2760045012281193e-06, "loss": 2.7062, "step": 287140 }, { "epoch": 0.006784, "grad_norm": 0.7883575558662415, "learning_rate": 3.2754084652559515e-06, "loss": 2.7895, "step": 287150 }, { "epoch": 0.0068096, "grad_norm": 0.7975062727928162, "learning_rate": 3.2748124728909727e-06, "loss": 2.8325, "step": 287160 }, { "epoch": 0.0068352, "grad_norm": 0.9051735401153564, "learning_rate": 3.274216524137047e-06, "loss": 2.7913, "step": 287170 }, { "epoch": 0.0068608, "grad_norm": 0.7417708039283752, "learning_rate": 3.273620618998037e-06, "loss": 2.7356, "step": 287180 }, { "epoch": 0.0068864, "grad_norm": 0.7748892903327942, "learning_rate": 3.273024757477816e-06, "loss": 2.7749, "step": 287190 }, { "epoch": 0.006912, "grad_norm": 0.7646074891090393, "learning_rate": 3.272428939580239e-06, "loss": 2.6486, "step": 287200 }, { "epoch": 0.0069376, "grad_norm": 0.7781855463981628, "learning_rate": 3.2718331653091718e-06, "loss": 2.6745, "step": 287210 }, { "epoch": 0.0069632, "grad_norm": 1.095894694328308, "learning_rate": 3.271237434668478e-06, "loss": 2.8503, "step": 287220 }, { "epoch": 0.0069888, "grad_norm": 0.8154774308204651, "learning_rate": 3.2706417476620223e-06, "loss": 2.8424, "step": 287230 }, { "epoch": 0.0070144, "grad_norm": 0.8597018718719482, "learning_rate": 3.270046104293666e-06, "loss": 2.6464, "step": 287240 }, { "epoch": 0.00704, "grad_norm": 0.785787045955658, "learning_rate": 3.269450504567272e-06, "loss": 2.8399, "step": 287250 }, { "epoch": 0.0070656, "grad_norm": 0.9443941116333008, "learning_rate": 3.268854948486705e-06, "loss": 2.6955, "step": 287260 }, { "epoch": 0.0070912, "grad_norm": 0.7967674136161804, "learning_rate": 3.268259436055823e-06, "loss": 2.6787, "step": 287270 }, { "epoch": 0.0071168, "grad_norm": 0.8113547563552856, "learning_rate": 3.2676639672784882e-06, "loss": 2.8476, "step": 287280 }, { "epoch": 0.0071424, "grad_norm": 0.8834633827209473, "learning_rate": 3.2670685421585646e-06, "loss": 2.7444, "step": 287290 }, { "epoch": 0.007168, "grad_norm": 0.7892204523086548, "learning_rate": 3.266473160699911e-06, "loss": 2.8068, "step": 287300 }, { "epoch": 0.0071936, "grad_norm": 0.7618678212165833, "learning_rate": 3.2658778229063894e-06, "loss": 2.6719, "step": 287310 }, { "epoch": 0.0072192, "grad_norm": 0.8083832263946533, "learning_rate": 3.2652825287818614e-06, "loss": 2.701, "step": 287320 }, { "epoch": 0.0072448, "grad_norm": 0.8313145637512207, "learning_rate": 3.2646872783301888e-06, "loss": 2.9839, "step": 287330 }, { "epoch": 0.0072704, "grad_norm": 0.8801666498184204, "learning_rate": 3.264092071555226e-06, "loss": 2.8867, "step": 287340 }, { "epoch": 0.007296, "grad_norm": 0.8456485867500305, "learning_rate": 3.263496908460836e-06, "loss": 2.8853, "step": 287350 }, { "epoch": 0.0073216, "grad_norm": 1.0747612714767456, "learning_rate": 3.262901789050875e-06, "loss": 2.9476, "step": 287360 }, { "epoch": 0.0073472, "grad_norm": 0.9675946235656738, "learning_rate": 3.2623067133292084e-06, "loss": 2.9487, "step": 287370 }, { "epoch": 0.0073728, "grad_norm": 0.7961512207984924, "learning_rate": 3.2617116812996963e-06, "loss": 2.8167, "step": 287380 }, { "epoch": 0.0073984, "grad_norm": 0.8255829811096191, "learning_rate": 3.2611166929661884e-06, "loss": 2.872, "step": 287390 }, { "epoch": 0.007424, "grad_norm": 0.8116006851196289, "learning_rate": 3.260521748332549e-06, "loss": 2.9926, "step": 287400 }, { "epoch": 0.0074496, "grad_norm": 0.8512586951255798, "learning_rate": 3.2599268474026346e-06, "loss": 2.91, "step": 287410 }, { "epoch": 0.0074752, "grad_norm": 0.9310526251792908, "learning_rate": 3.259331990180302e-06, "loss": 2.7527, "step": 287420 }, { "epoch": 0.0075008, "grad_norm": 0.9301791787147522, "learning_rate": 3.258737176669412e-06, "loss": 2.731, "step": 287430 }, { "epoch": 0.0075264, "grad_norm": 0.8298847079277039, "learning_rate": 3.2581424068738187e-06, "loss": 2.8078, "step": 287440 }, { "epoch": 0.007552, "grad_norm": 0.8042241930961609, "learning_rate": 3.2575476807973806e-06, "loss": 2.7504, "step": 287450 }, { "epoch": 0.0075776, "grad_norm": 0.9825581908226013, "learning_rate": 3.256952998443956e-06, "loss": 2.8936, "step": 287460 }, { "epoch": 0.0076032, "grad_norm": 1.0117719173431396, "learning_rate": 3.2563583598173975e-06, "loss": 2.7452, "step": 287470 }, { "epoch": 0.0076288, "grad_norm": 0.8377805948257446, "learning_rate": 3.255763764921562e-06, "loss": 2.7922, "step": 287480 }, { "epoch": 0.0076544, "grad_norm": 0.9472309947013855, "learning_rate": 3.2551692137603063e-06, "loss": 2.9631, "step": 287490 }, { "epoch": 0.00768, "grad_norm": 0.8433656096458435, "learning_rate": 3.2545747063374855e-06, "loss": 2.7273, "step": 287500 }, { "epoch": 0.0077056, "grad_norm": 0.9037414789199829, "learning_rate": 3.2539802426569543e-06, "loss": 2.7348, "step": 287510 }, { "epoch": 0.0077312, "grad_norm": 0.8282715082168579, "learning_rate": 3.253385822722569e-06, "loss": 3.0832, "step": 287520 }, { "epoch": 0.0077568, "grad_norm": 0.8069460988044739, "learning_rate": 3.2527914465381837e-06, "loss": 3.0023, "step": 287530 }, { "epoch": 0.0077824, "grad_norm": 0.8384937047958374, "learning_rate": 3.2521971141076527e-06, "loss": 2.9883, "step": 287540 }, { "epoch": 0.007808, "grad_norm": 0.7894148230552673, "learning_rate": 3.2516028254348298e-06, "loss": 2.7147, "step": 287550 }, { "epoch": 0.0078336, "grad_norm": 0.7903317213058472, "learning_rate": 3.2510085805235694e-06, "loss": 2.9352, "step": 287560 }, { "epoch": 0.0078592, "grad_norm": 0.8057121634483337, "learning_rate": 3.250414379377724e-06, "loss": 2.6453, "step": 287570 }, { "epoch": 0.0078848, "grad_norm": 1.0461859703063965, "learning_rate": 3.249820222001152e-06, "loss": 3.0115, "step": 287580 }, { "epoch": 0.0079104, "grad_norm": 1.0832773447036743, "learning_rate": 3.2492261083976985e-06, "loss": 2.8413, "step": 287590 }, { "epoch": 0.007936, "grad_norm": 0.9338743686676025, "learning_rate": 3.2486320385712188e-06, "loss": 2.9731, "step": 287600 }, { "epoch": 0.0079616, "grad_norm": 0.7955946922302246, "learning_rate": 3.248038012525566e-06, "loss": 2.7536, "step": 287610 }, { "epoch": 0.0079872, "grad_norm": 0.8369946479797363, "learning_rate": 3.2474440302645926e-06, "loss": 2.764, "step": 287620 }, { "epoch": 0.0080128, "grad_norm": 0.8255676031112671, "learning_rate": 3.2468500917921496e-06, "loss": 2.592, "step": 287630 }, { "epoch": 0.0080384, "grad_norm": 0.8324923515319824, "learning_rate": 3.2462561971120887e-06, "loss": 2.8051, "step": 287640 }, { "epoch": 0.008064, "grad_norm": 0.7868658900260925, "learning_rate": 3.2456623462282624e-06, "loss": 2.8357, "step": 287650 }, { "epoch": 0.0080896, "grad_norm": 0.774640679359436, "learning_rate": 3.2450685391445237e-06, "loss": 2.9009, "step": 287660 }, { "epoch": 0.0081152, "grad_norm": 0.8176813721656799, "learning_rate": 3.2444747758647166e-06, "loss": 2.9696, "step": 287670 }, { "epoch": 0.0081408, "grad_norm": 0.7883301973342896, "learning_rate": 3.2438810563926958e-06, "loss": 2.6577, "step": 287680 }, { "epoch": 0.0081664, "grad_norm": 0.7831227779388428, "learning_rate": 3.2432873807323107e-06, "loss": 2.7553, "step": 287690 }, { "epoch": 0.008192, "grad_norm": 0.774000346660614, "learning_rate": 3.242693748887411e-06, "loss": 2.7635, "step": 287700 }, { "epoch": 0.0082176, "grad_norm": 1.0099968910217285, "learning_rate": 3.2421001608618463e-06, "loss": 2.8223, "step": 287710 }, { "epoch": 0.0082432, "grad_norm": 0.8687630295753479, "learning_rate": 3.2415066166594656e-06, "loss": 2.838, "step": 287720 }, { "epoch": 0.0082688, "grad_norm": 0.7674843072891235, "learning_rate": 3.240913116284119e-06, "loss": 2.8884, "step": 287730 }, { "epoch": 0.0082944, "grad_norm": 0.8243243098258972, "learning_rate": 3.2403196597396546e-06, "loss": 2.7945, "step": 287740 }, { "epoch": 0.00832, "grad_norm": 0.8292272090911865, "learning_rate": 3.2397262470299207e-06, "loss": 2.9108, "step": 287750 }, { "epoch": 0.0083456, "grad_norm": 1.3711823225021362, "learning_rate": 3.239132878158765e-06, "loss": 2.7403, "step": 287760 }, { "epoch": 0.0083712, "grad_norm": 0.8754733800888062, "learning_rate": 3.2385395531300354e-06, "loss": 2.8473, "step": 287770 }, { "epoch": 0.0083968, "grad_norm": 0.7727702856063843, "learning_rate": 3.237946271947584e-06, "loss": 2.935, "step": 287780 }, { "epoch": 0.0084224, "grad_norm": 0.8607861399650574, "learning_rate": 3.2373530346152502e-06, "loss": 2.835, "step": 287790 }, { "epoch": 0.008448, "grad_norm": 0.9685192704200745, "learning_rate": 3.236759841136885e-06, "loss": 2.7313, "step": 287800 }, { "epoch": 0.0084736, "grad_norm": 0.7539849281311035, "learning_rate": 3.2361666915163335e-06, "loss": 2.7017, "step": 287810 }, { "epoch": 0.0084992, "grad_norm": 0.7486064434051514, "learning_rate": 3.2355735857574446e-06, "loss": 2.6881, "step": 287820 }, { "epoch": 0.0085248, "grad_norm": 0.7424217462539673, "learning_rate": 3.234980523864063e-06, "loss": 2.7187, "step": 287830 }, { "epoch": 0.0085504, "grad_norm": 0.9855384230613708, "learning_rate": 3.2343875058400354e-06, "loss": 2.9087, "step": 287840 }, { "epoch": 0.008576, "grad_norm": 0.8728731274604797, "learning_rate": 3.233794531689206e-06, "loss": 2.8186, "step": 287850 }, { "epoch": 0.0086016, "grad_norm": 0.8214504718780518, "learning_rate": 3.2332016014154243e-06, "loss": 2.8909, "step": 287860 }, { "epoch": 0.0086272, "grad_norm": 0.7776926755905151, "learning_rate": 3.2326087150225283e-06, "loss": 2.8712, "step": 287870 }, { "epoch": 0.0086528, "grad_norm": 0.908603310585022, "learning_rate": 3.2320158725143626e-06, "loss": 2.8097, "step": 287880 }, { "epoch": 0.0086784, "grad_norm": 0.8168140649795532, "learning_rate": 3.231423073894779e-06, "loss": 2.9417, "step": 287890 }, { "epoch": 0.008704, "grad_norm": 0.7546183466911316, "learning_rate": 3.230830319167618e-06, "loss": 2.7827, "step": 287900 }, { "epoch": 0.0087296, "grad_norm": 0.8008772730827332, "learning_rate": 3.230237608336727e-06, "loss": 2.8907, "step": 287910 }, { "epoch": 0.0087552, "grad_norm": 0.888032078742981, "learning_rate": 3.2296449414059416e-06, "loss": 3.0323, "step": 287920 }, { "epoch": 0.0087808, "grad_norm": 0.8108900785446167, "learning_rate": 3.22905231837911e-06, "loss": 2.7615, "step": 287930 }, { "epoch": 0.0088064, "grad_norm": 0.811365008354187, "learning_rate": 3.2284597392600745e-06, "loss": 2.768, "step": 287940 }, { "epoch": 0.008832, "grad_norm": 0.9116649031639099, "learning_rate": 3.2278672040526783e-06, "loss": 2.808, "step": 287950 }, { "epoch": 0.0088576, "grad_norm": 0.8297942876815796, "learning_rate": 3.227274712760764e-06, "loss": 3.0438, "step": 287960 }, { "epoch": 0.0088832, "grad_norm": 0.767890989780426, "learning_rate": 3.2266822653881722e-06, "loss": 2.8467, "step": 287970 }, { "epoch": 0.0089088, "grad_norm": 0.8871746063232422, "learning_rate": 3.226089861938749e-06, "loss": 2.6598, "step": 287980 }, { "epoch": 0.0089344, "grad_norm": 0.8221619725227356, "learning_rate": 3.2254975024163293e-06, "loss": 2.8157, "step": 287990 }, { "epoch": 0.00896, "grad_norm": 0.8647187948226929, "learning_rate": 3.2249051868247583e-06, "loss": 2.7193, "step": 288000 }, { "epoch": 0.0089856, "grad_norm": 0.828469455242157, "learning_rate": 3.2243129151678766e-06, "loss": 2.7683, "step": 288010 }, { "epoch": 0.0090112, "grad_norm": 0.8364804983139038, "learning_rate": 3.2237206874495254e-06, "loss": 2.7671, "step": 288020 }, { "epoch": 0.0090368, "grad_norm": 0.7654609680175781, "learning_rate": 3.223128503673544e-06, "loss": 2.6109, "step": 288030 }, { "epoch": 0.0090624, "grad_norm": 0.8577880859375, "learning_rate": 3.2225363638437735e-06, "loss": 2.7714, "step": 288040 }, { "epoch": 0.009088, "grad_norm": 0.8356485962867737, "learning_rate": 3.2219442679640533e-06, "loss": 2.8828, "step": 288050 }, { "epoch": 0.0091136, "grad_norm": 0.8692479133605957, "learning_rate": 3.221352216038223e-06, "loss": 2.7683, "step": 288060 }, { "epoch": 0.0091392, "grad_norm": 0.8218371272087097, "learning_rate": 3.220760208070122e-06, "loss": 2.7197, "step": 288070 }, { "epoch": 0.0091648, "grad_norm": 1.342962384223938, "learning_rate": 3.2201682440635894e-06, "loss": 2.8962, "step": 288080 }, { "epoch": 0.0091904, "grad_norm": 0.9275806546211243, "learning_rate": 3.219576324022463e-06, "loss": 3.2259, "step": 288090 }, { "epoch": 0.009216, "grad_norm": 0.8354640603065491, "learning_rate": 3.218984447950584e-06, "loss": 2.8551, "step": 288100 }, { "epoch": 0.0092416, "grad_norm": 0.9828537702560425, "learning_rate": 3.2183926158517907e-06, "loss": 2.7999, "step": 288110 }, { "epoch": 0.0092672, "grad_norm": 0.8941342234611511, "learning_rate": 3.2178008277299155e-06, "loss": 2.8366, "step": 288120 }, { "epoch": 0.0092928, "grad_norm": 0.8831412196159363, "learning_rate": 3.2172090835888005e-06, "loss": 2.9244, "step": 288130 }, { "epoch": 0.0093184, "grad_norm": 0.9301644563674927, "learning_rate": 3.216617383432281e-06, "loss": 2.8349, "step": 288140 }, { "epoch": 0.009344, "grad_norm": 0.9771103858947754, "learning_rate": 3.2160257272641947e-06, "loss": 2.7544, "step": 288150 }, { "epoch": 0.0093696, "grad_norm": 0.7593784928321838, "learning_rate": 3.2154341150883794e-06, "loss": 2.5917, "step": 288160 }, { "epoch": 0.0093952, "grad_norm": 0.8287583589553833, "learning_rate": 3.2148425469086708e-06, "loss": 2.8244, "step": 288170 }, { "epoch": 0.0094208, "grad_norm": 0.782594621181488, "learning_rate": 3.214251022728908e-06, "loss": 2.8594, "step": 288180 }, { "epoch": 0.0094464, "grad_norm": 0.8527307510375977, "learning_rate": 3.2136595425529205e-06, "loss": 2.9571, "step": 288190 }, { "epoch": 0.009472, "grad_norm": 0.9441353678703308, "learning_rate": 3.213068106384547e-06, "loss": 2.7762, "step": 288200 }, { "epoch": 0.0094976, "grad_norm": 0.7509833574295044, "learning_rate": 3.2124767142276238e-06, "loss": 2.5935, "step": 288210 }, { "epoch": 0.0095232, "grad_norm": 0.795295000076294, "learning_rate": 3.211885366085985e-06, "loss": 2.9243, "step": 288220 }, { "epoch": 0.0095488, "grad_norm": 0.8119003772735596, "learning_rate": 3.2112940619634615e-06, "loss": 2.8017, "step": 288230 }, { "epoch": 0.0095744, "grad_norm": 0.7925642728805542, "learning_rate": 3.2107028018639e-06, "loss": 2.7633, "step": 288240 }, { "epoch": 0.0096, "grad_norm": 0.8036895394325256, "learning_rate": 3.2101115857911215e-06, "loss": 2.9716, "step": 288250 }, { "epoch": 0.0096256, "grad_norm": 0.7834755182266235, "learning_rate": 3.2095204137489654e-06, "loss": 2.8497, "step": 288260 }, { "epoch": 0.0096512, "grad_norm": 0.8032413721084595, "learning_rate": 3.2089292857412645e-06, "loss": 2.9122, "step": 288270 }, { "epoch": 0.0096768, "grad_norm": 0.849061131477356, "learning_rate": 3.2083382017718522e-06, "loss": 3.0044, "step": 288280 }, { "epoch": 0.0097024, "grad_norm": 0.8243955969810486, "learning_rate": 3.2077471618445622e-06, "loss": 2.8372, "step": 288290 }, { "epoch": 0.009728, "grad_norm": 0.9073106646537781, "learning_rate": 3.2071561659632265e-06, "loss": 2.7801, "step": 288300 }, { "epoch": 0.0097536, "grad_norm": 0.8428090810775757, "learning_rate": 3.2065652141316816e-06, "loss": 2.9372, "step": 288310 }, { "epoch": 0.0097792, "grad_norm": 0.8058331608772278, "learning_rate": 3.205974306353752e-06, "loss": 2.7627, "step": 288320 }, { "epoch": 0.0098048, "grad_norm": 0.8687593936920166, "learning_rate": 3.205383442633273e-06, "loss": 2.8388, "step": 288330 }, { "epoch": 0.0098304, "grad_norm": 0.9269878268241882, "learning_rate": 3.2047926229740776e-06, "loss": 2.7914, "step": 288340 }, { "epoch": 0.009856, "grad_norm": 0.7893519401550293, "learning_rate": 3.2042018473799953e-06, "loss": 2.841, "step": 288350 }, { "epoch": 0.0098816, "grad_norm": 0.8216515779495239, "learning_rate": 3.2036111158548577e-06, "loss": 2.8117, "step": 288360 }, { "epoch": 0.0099072, "grad_norm": 0.8365702033042908, "learning_rate": 3.203020428402497e-06, "loss": 2.7547, "step": 288370 }, { "epoch": 0.0099328, "grad_norm": 0.7905628681182861, "learning_rate": 3.2024297850267438e-06, "loss": 2.7366, "step": 288380 }, { "epoch": 0.0099584, "grad_norm": 0.9159860610961914, "learning_rate": 3.2018391857314245e-06, "loss": 2.8088, "step": 288390 }, { "epoch": 0.009984, "grad_norm": 0.7799779772758484, "learning_rate": 3.2012486305203683e-06, "loss": 2.8298, "step": 288400 }, { "epoch": 0.0100096, "grad_norm": 0.8426455855369568, "learning_rate": 3.200658119397411e-06, "loss": 2.6809, "step": 288410 }, { "epoch": 0.0100352, "grad_norm": 0.8995066285133362, "learning_rate": 3.2000676523663786e-06, "loss": 2.9207, "step": 288420 }, { "epoch": 0.0100608, "grad_norm": 0.8319149613380432, "learning_rate": 3.1994772294311007e-06, "loss": 2.8537, "step": 288430 }, { "epoch": 0.0100864, "grad_norm": 0.8117222189903259, "learning_rate": 3.1988868505954086e-06, "loss": 2.8921, "step": 288440 }, { "epoch": 0.010112, "grad_norm": 3.7176432609558105, "learning_rate": 3.198296515863124e-06, "loss": 3.0355, "step": 288450 }, { "epoch": 0.0101376, "grad_norm": 0.8561798334121704, "learning_rate": 3.1977062252380797e-06, "loss": 3.151, "step": 288460 }, { "epoch": 0.0101632, "grad_norm": 0.7917779088020325, "learning_rate": 3.197115978724102e-06, "loss": 2.8664, "step": 288470 }, { "epoch": 0.0101888, "grad_norm": 1.0165231227874756, "learning_rate": 3.1965257763250202e-06, "loss": 2.777, "step": 288480 }, { "epoch": 0.0102144, "grad_norm": 0.7416680455207825, "learning_rate": 3.195935618044659e-06, "loss": 2.7471, "step": 288490 }, { "epoch": 0.01024, "grad_norm": 1.0464507341384888, "learning_rate": 3.1953455038868487e-06, "loss": 2.9925, "step": 288500 }, { "epoch": 0.0102656, "grad_norm": 0.8261627554893494, "learning_rate": 3.1947554338554164e-06, "loss": 2.9447, "step": 288510 }, { "epoch": 0.0102912, "grad_norm": 0.8475132584571838, "learning_rate": 3.194165407954184e-06, "loss": 2.7831, "step": 288520 }, { "epoch": 0.0103168, "grad_norm": 0.752723753452301, "learning_rate": 3.193575426186979e-06, "loss": 2.7044, "step": 288530 }, { "epoch": 0.0103424, "grad_norm": 0.7570744156837463, "learning_rate": 3.1929854885576294e-06, "loss": 2.8368, "step": 288540 }, { "epoch": 0.010368, "grad_norm": 0.7685967683792114, "learning_rate": 3.1923955950699593e-06, "loss": 2.9175, "step": 288550 }, { "epoch": 0.0103936, "grad_norm": 0.7999095320701599, "learning_rate": 3.191805745727795e-06, "loss": 2.8494, "step": 288560 }, { "epoch": 0.0104192, "grad_norm": 0.7655049562454224, "learning_rate": 3.191215940534961e-06, "loss": 2.9818, "step": 288570 }, { "epoch": 0.0104448, "grad_norm": 1.0166425704956055, "learning_rate": 3.190626179495282e-06, "loss": 2.8934, "step": 288580 }, { "epoch": 0.0104704, "grad_norm": 0.783552885055542, "learning_rate": 3.1900364626125826e-06, "loss": 2.8755, "step": 288590 }, { "epoch": 0.010496, "grad_norm": 0.9094949960708618, "learning_rate": 3.1894467898906866e-06, "loss": 2.7624, "step": 288600 }, { "epoch": 0.0105216, "grad_norm": 0.826183021068573, "learning_rate": 3.1888571613334184e-06, "loss": 2.9166, "step": 288610 }, { "epoch": 0.0105472, "grad_norm": 0.9205394387245178, "learning_rate": 3.1882675769446003e-06, "loss": 2.74, "step": 288620 }, { "epoch": 0.0105728, "grad_norm": 0.8034763932228088, "learning_rate": 3.187678036728058e-06, "loss": 2.8498, "step": 288630 }, { "epoch": 0.0105984, "grad_norm": 0.8014287948608398, "learning_rate": 3.187088540687616e-06, "loss": 2.8724, "step": 288640 }, { "epoch": 0.010624, "grad_norm": 0.8275606036186218, "learning_rate": 3.186499088827091e-06, "loss": 2.9039, "step": 288650 }, { "epoch": 0.0106496, "grad_norm": 1.9991459846496582, "learning_rate": 3.185909681150308e-06, "loss": 2.973, "step": 288660 }, { "epoch": 0.0106752, "grad_norm": 2.010373592376709, "learning_rate": 3.1853203176610903e-06, "loss": 2.7504, "step": 288670 }, { "epoch": 0.0107008, "grad_norm": 0.9879314303398132, "learning_rate": 3.184730998363259e-06, "loss": 3.001, "step": 288680 }, { "epoch": 0.0107264, "grad_norm": 0.8550890684127808, "learning_rate": 3.1841417232606363e-06, "loss": 2.9603, "step": 288690 }, { "epoch": 0.010752, "grad_norm": 0.887584924697876, "learning_rate": 3.1835524923570425e-06, "loss": 2.8787, "step": 288700 }, { "epoch": 0.0107776, "grad_norm": 0.9330102801322937, "learning_rate": 3.1829633056563023e-06, "loss": 2.7922, "step": 288710 }, { "epoch": 0.0108032, "grad_norm": 4.463079452514648, "learning_rate": 3.1823741631622307e-06, "loss": 3.2403, "step": 288720 }, { "epoch": 0.0108288, "grad_norm": 1.006919503211975, "learning_rate": 3.1817850648786508e-06, "loss": 2.8166, "step": 288730 }, { "epoch": 0.0108544, "grad_norm": 0.7967254519462585, "learning_rate": 3.181196010809382e-06, "loss": 2.8443, "step": 288740 }, { "epoch": 0.01088, "grad_norm": 0.904525876045227, "learning_rate": 3.1806070009582425e-06, "loss": 2.784, "step": 288750 }, { "epoch": 0.0109056, "grad_norm": 0.8191792368888855, "learning_rate": 3.1800180353290566e-06, "loss": 2.7776, "step": 288760 }, { "epoch": 0.0109312, "grad_norm": 0.8986817002296448, "learning_rate": 3.179429113925645e-06, "loss": 2.8428, "step": 288770 }, { "epoch": 0.0109568, "grad_norm": 0.84711092710495, "learning_rate": 3.178840236751819e-06, "loss": 2.8674, "step": 288780 }, { "epoch": 0.0109824, "grad_norm": 0.8859431743621826, "learning_rate": 3.1782514038114022e-06, "loss": 2.9446, "step": 288790 }, { "epoch": 0.011008, "grad_norm": 0.9117154479026794, "learning_rate": 3.177662615108211e-06, "loss": 2.8059, "step": 288800 }, { "epoch": 0.0110336, "grad_norm": 0.7787767052650452, "learning_rate": 3.1770738706460645e-06, "loss": 2.9011, "step": 288810 }, { "epoch": 0.0110592, "grad_norm": 0.8546193242073059, "learning_rate": 3.1764851704287814e-06, "loss": 3.1229, "step": 288820 }, { "epoch": 0.0110848, "grad_norm": 0.8407701849937439, "learning_rate": 3.1758965144601784e-06, "loss": 2.9045, "step": 288830 }, { "epoch": 0.0111104, "grad_norm": 0.8946781754493713, "learning_rate": 3.1753079027440757e-06, "loss": 2.8144, "step": 288840 }, { "epoch": 0.011136, "grad_norm": 0.8498762249946594, "learning_rate": 3.1747193352842852e-06, "loss": 2.8104, "step": 288850 }, { "epoch": 0.0111616, "grad_norm": 1.0263394117355347, "learning_rate": 3.1741308120846247e-06, "loss": 3.1863, "step": 288860 }, { "epoch": 0.0111872, "grad_norm": 0.8101439476013184, "learning_rate": 3.1735423331489125e-06, "loss": 2.755, "step": 288870 }, { "epoch": 0.0112128, "grad_norm": 0.7717148661613464, "learning_rate": 3.1729538984809637e-06, "loss": 2.753, "step": 288880 }, { "epoch": 0.0112384, "grad_norm": 0.9876390695571899, "learning_rate": 3.172365508084595e-06, "loss": 2.8884, "step": 288890 }, { "epoch": 0.011264, "grad_norm": 0.7840248942375183, "learning_rate": 3.1717771619636206e-06, "loss": 2.8179, "step": 288900 }, { "epoch": 0.0112896, "grad_norm": 0.7474378943443298, "learning_rate": 3.171188860121861e-06, "loss": 2.8692, "step": 288910 }, { "epoch": 0.0113152, "grad_norm": 0.8082808256149292, "learning_rate": 3.17060060256312e-06, "loss": 3.1253, "step": 288920 }, { "epoch": 0.0113408, "grad_norm": 0.8231500387191772, "learning_rate": 3.1700123892912214e-06, "loss": 2.9009, "step": 288930 }, { "epoch": 0.0113664, "grad_norm": 0.7995244264602661, "learning_rate": 3.1694242203099767e-06, "loss": 2.7236, "step": 288940 }, { "epoch": 0.011392, "grad_norm": 0.7916783690452576, "learning_rate": 3.168836095623202e-06, "loss": 2.918, "step": 288950 }, { "epoch": 0.0114176, "grad_norm": 1.0933351516723633, "learning_rate": 3.1682480152347097e-06, "loss": 2.839, "step": 288960 }, { "epoch": 0.0114432, "grad_norm": 0.8670771718025208, "learning_rate": 3.1676599791483154e-06, "loss": 2.8726, "step": 288970 }, { "epoch": 0.0114688, "grad_norm": 0.8453372716903687, "learning_rate": 3.167071987367828e-06, "loss": 2.8666, "step": 288980 }, { "epoch": 0.0114944, "grad_norm": 0.8772324919700623, "learning_rate": 3.1664840398970633e-06, "loss": 2.8433, "step": 288990 }, { "epoch": 0.01152, "grad_norm": 2.3481152057647705, "learning_rate": 3.165896136739832e-06, "loss": 3.2223, "step": 289000 }, { "epoch": 0.0115456, "grad_norm": 0.8373615145683289, "learning_rate": 3.165308277899949e-06, "loss": 2.8406, "step": 289010 }, { "epoch": 0.0115712, "grad_norm": 0.76758873462677, "learning_rate": 3.1647204633812256e-06, "loss": 2.8324, "step": 289020 }, { "epoch": 0.0115968, "grad_norm": 0.8916561007499695, "learning_rate": 3.1641326931874728e-06, "loss": 2.8617, "step": 289030 }, { "epoch": 0.0116224, "grad_norm": 0.889163076877594, "learning_rate": 3.163544967322506e-06, "loss": 2.8203, "step": 289040 }, { "epoch": 0.011648, "grad_norm": 0.845094621181488, "learning_rate": 3.16295728579013e-06, "loss": 2.9829, "step": 289050 }, { "epoch": 0.0116736, "grad_norm": 0.8414230942726135, "learning_rate": 3.162369648594159e-06, "loss": 2.7668, "step": 289060 }, { "epoch": 0.0116992, "grad_norm": 0.8713226318359375, "learning_rate": 3.1617820557384037e-06, "loss": 2.8021, "step": 289070 }, { "epoch": 0.0117248, "grad_norm": 0.9913536310195923, "learning_rate": 3.1611945072266736e-06, "loss": 2.7062, "step": 289080 }, { "epoch": 0.0117504, "grad_norm": 1.0691851377487183, "learning_rate": 3.160607003062781e-06, "loss": 3.0243, "step": 289090 }, { "epoch": 0.011776, "grad_norm": 0.7886620163917542, "learning_rate": 3.1600195432505334e-06, "loss": 2.7903, "step": 289100 }, { "epoch": 0.0118016, "grad_norm": 0.8309621810913086, "learning_rate": 3.1594321277937414e-06, "loss": 3.0228, "step": 289110 }, { "epoch": 0.0118272, "grad_norm": 0.9527907371520996, "learning_rate": 3.158844756696213e-06, "loss": 2.9179, "step": 289120 }, { "epoch": 0.0118528, "grad_norm": 0.7810391187667847, "learning_rate": 3.1582574299617597e-06, "loss": 2.7414, "step": 289130 }, { "epoch": 0.0118784, "grad_norm": 1.002660870552063, "learning_rate": 3.1576701475941886e-06, "loss": 2.9019, "step": 289140 }, { "epoch": 0.011904, "grad_norm": 0.8072208166122437, "learning_rate": 3.1570829095973077e-06, "loss": 2.9093, "step": 289150 }, { "epoch": 0.0119296, "grad_norm": 0.7767460346221924, "learning_rate": 3.1564957159749256e-06, "loss": 2.7376, "step": 289160 }, { "epoch": 0.0119552, "grad_norm": 1.3664040565490723, "learning_rate": 3.155908566730852e-06, "loss": 2.9216, "step": 289170 }, { "epoch": 0.0119808, "grad_norm": 1.2229117155075073, "learning_rate": 3.155321461868891e-06, "loss": 2.7409, "step": 289180 }, { "epoch": 0.0120064, "grad_norm": 0.7877373099327087, "learning_rate": 3.1547344013928503e-06, "loss": 2.741, "step": 289190 }, { "epoch": 0.012032, "grad_norm": 0.7991163730621338, "learning_rate": 3.1541473853065373e-06, "loss": 2.9478, "step": 289200 }, { "epoch": 0.0120576, "grad_norm": 1.0313348770141602, "learning_rate": 3.1535604136137597e-06, "loss": 2.6043, "step": 289210 }, { "epoch": 0.0120832, "grad_norm": 0.8325128555297852, "learning_rate": 3.1529734863183238e-06, "loss": 2.7612, "step": 289220 }, { "epoch": 0.0121088, "grad_norm": 0.7981945276260376, "learning_rate": 3.1523866034240347e-06, "loss": 2.7277, "step": 289230 }, { "epoch": 0.0121344, "grad_norm": 0.8519823551177979, "learning_rate": 3.151799764934702e-06, "loss": 2.7847, "step": 289240 }, { "epoch": 0.01216, "grad_norm": 0.9087983965873718, "learning_rate": 3.1512129708541237e-06, "loss": 2.8775, "step": 289250 }, { "epoch": 0.0121856, "grad_norm": 0.9003401398658752, "learning_rate": 3.1506262211861106e-06, "loss": 2.7137, "step": 289260 }, { "epoch": 0.0122112, "grad_norm": 0.8596645593643188, "learning_rate": 3.150039515934461e-06, "loss": 2.7577, "step": 289270 }, { "epoch": 0.0122368, "grad_norm": 0.8358784317970276, "learning_rate": 3.149452855102989e-06, "loss": 2.8807, "step": 289280 }, { "epoch": 0.0122624, "grad_norm": 0.8370013236999512, "learning_rate": 3.148866238695498e-06, "loss": 2.8414, "step": 289290 }, { "epoch": 0.012288, "grad_norm": 0.8492915034294128, "learning_rate": 3.1482796667157845e-06, "loss": 2.8044, "step": 289300 }, { "epoch": 0.0123136, "grad_norm": 0.9097852110862732, "learning_rate": 3.1476931391676568e-06, "loss": 2.9055, "step": 289310 }, { "epoch": 0.0123392, "grad_norm": 0.8481084108352661, "learning_rate": 3.147106656054919e-06, "loss": 2.788, "step": 289320 }, { "epoch": 0.0123648, "grad_norm": 0.8496371507644653, "learning_rate": 3.1465202173813713e-06, "loss": 2.6809, "step": 289330 }, { "epoch": 0.0123904, "grad_norm": 0.9015195369720459, "learning_rate": 3.1459338231508208e-06, "loss": 2.7854, "step": 289340 }, { "epoch": 0.012416, "grad_norm": 2.1855969429016113, "learning_rate": 3.1453474733670663e-06, "loss": 2.9391, "step": 289350 }, { "epoch": 0.0124416, "grad_norm": 0.8872856497764587, "learning_rate": 3.1447611680339153e-06, "loss": 3.0618, "step": 289360 }, { "epoch": 0.0124672, "grad_norm": 0.9188830852508545, "learning_rate": 3.144174907155163e-06, "loss": 2.906, "step": 289370 }, { "epoch": 0.0124928, "grad_norm": 1.0098234415054321, "learning_rate": 3.1435886907346136e-06, "loss": 2.9792, "step": 289380 }, { "epoch": 0.0125184, "grad_norm": 0.8404697179794312, "learning_rate": 3.1430025187760705e-06, "loss": 2.8844, "step": 289390 }, { "epoch": 0.012544, "grad_norm": 0.8879860639572144, "learning_rate": 3.1424163912833326e-06, "loss": 2.7166, "step": 289400 }, { "epoch": 0.0125696, "grad_norm": 0.986809492111206, "learning_rate": 3.141830308260203e-06, "loss": 2.8692, "step": 289410 }, { "epoch": 0.0125952, "grad_norm": 1.026826024055481, "learning_rate": 3.1412442697104807e-06, "loss": 3.0948, "step": 289420 }, { "epoch": 0.0126208, "grad_norm": 0.8873839974403381, "learning_rate": 3.140658275637967e-06, "loss": 2.8552, "step": 289430 }, { "epoch": 0.0126464, "grad_norm": 0.860868513584137, "learning_rate": 3.140072326046464e-06, "loss": 3.3683, "step": 289440 }, { "epoch": 0.012672, "grad_norm": 1.3116933107376099, "learning_rate": 3.1394864209397623e-06, "loss": 2.5779, "step": 289450 }, { "epoch": 0.0126976, "grad_norm": 0.9136570692062378, "learning_rate": 3.1389005603216714e-06, "loss": 2.7482, "step": 289460 }, { "epoch": 0.0127232, "grad_norm": 0.7992715239524841, "learning_rate": 3.138314744195986e-06, "loss": 2.8021, "step": 289470 }, { "epoch": 0.0127488, "grad_norm": 0.7961689829826355, "learning_rate": 3.137728972566507e-06, "loss": 2.7813, "step": 289480 }, { "epoch": 0.0127744, "grad_norm": 0.9708912968635559, "learning_rate": 3.1371432454370343e-06, "loss": 2.8854, "step": 289490 }, { "epoch": 0.0128, "grad_norm": 0.7755069732666016, "learning_rate": 3.1365575628113608e-06, "loss": 2.9336, "step": 289500 }, { "epoch": 0.0128256, "grad_norm": 0.8567432165145874, "learning_rate": 3.1359719246932883e-06, "loss": 2.6921, "step": 289510 }, { "epoch": 0.0128512, "grad_norm": 1.1249722242355347, "learning_rate": 3.1353863310866127e-06, "loss": 3.0618, "step": 289520 }, { "epoch": 0.0128768, "grad_norm": 0.9226755499839783, "learning_rate": 3.1348007819951322e-06, "loss": 2.7545, "step": 289530 }, { "epoch": 0.0129024, "grad_norm": 0.8818320035934448, "learning_rate": 3.1342152774226443e-06, "loss": 2.7627, "step": 289540 }, { "epoch": 0.012928, "grad_norm": 0.8544482588768005, "learning_rate": 3.133629817372946e-06, "loss": 2.8626, "step": 289550 }, { "epoch": 0.0129536, "grad_norm": 2.955700159072876, "learning_rate": 3.133044401849836e-06, "loss": 3.0332, "step": 289560 }, { "epoch": 0.0129792, "grad_norm": 1.0338466167449951, "learning_rate": 3.1324590308571058e-06, "loss": 2.5457, "step": 289570 }, { "epoch": 0.0130048, "grad_norm": 0.8055727481842041, "learning_rate": 3.1318737043985527e-06, "loss": 2.7547, "step": 289580 }, { "epoch": 0.0130304, "grad_norm": 1.035406231880188, "learning_rate": 3.1312884224779736e-06, "loss": 2.8673, "step": 289590 }, { "epoch": 0.013056, "grad_norm": 0.9746801257133484, "learning_rate": 3.130703185099163e-06, "loss": 2.8383, "step": 289600 }, { "epoch": 0.0130816, "grad_norm": 0.8137537837028503, "learning_rate": 3.1301179922659175e-06, "loss": 2.9286, "step": 289610 }, { "epoch": 0.0131072, "grad_norm": 0.9922377467155457, "learning_rate": 3.1295328439820306e-06, "loss": 2.7581, "step": 289620 }, { "epoch": 0.0131328, "grad_norm": 0.7983953952789307, "learning_rate": 3.128947740251298e-06, "loss": 2.7884, "step": 289630 }, { "epoch": 0.0131584, "grad_norm": 0.8665645718574524, "learning_rate": 3.1283626810775124e-06, "loss": 2.913, "step": 289640 }, { "epoch": 0.013184, "grad_norm": 1.2523043155670166, "learning_rate": 3.127777666464469e-06, "loss": 2.8835, "step": 289650 }, { "epoch": 0.0132096, "grad_norm": 0.7415218353271484, "learning_rate": 3.1271926964159606e-06, "loss": 2.8236, "step": 289660 }, { "epoch": 0.0132352, "grad_norm": 0.8365862965583801, "learning_rate": 3.126607770935781e-06, "loss": 2.9453, "step": 289670 }, { "epoch": 0.0132608, "grad_norm": 0.7796602249145508, "learning_rate": 3.1260228900277234e-06, "loss": 2.8551, "step": 289680 }, { "epoch": 0.0132864, "grad_norm": 1.090336561203003, "learning_rate": 3.125438053695583e-06, "loss": 3.1208, "step": 289690 }, { "epoch": 0.013312, "grad_norm": 0.7288010716438293, "learning_rate": 3.124853261943147e-06, "loss": 2.7609, "step": 289700 }, { "epoch": 0.0133376, "grad_norm": 0.7945446372032166, "learning_rate": 3.1242685147742112e-06, "loss": 2.6127, "step": 289710 }, { "epoch": 0.0133632, "grad_norm": 0.9101090431213379, "learning_rate": 3.123683812192566e-06, "loss": 2.8299, "step": 289720 }, { "epoch": 0.0133888, "grad_norm": 0.9954261779785156, "learning_rate": 3.123099154202004e-06, "loss": 2.9173, "step": 289730 }, { "epoch": 0.0134144, "grad_norm": 1.6250300407409668, "learning_rate": 3.122514540806316e-06, "loss": 2.9223, "step": 289740 }, { "epoch": 0.01344, "grad_norm": 0.8877707719802856, "learning_rate": 3.1219299720092945e-06, "loss": 2.8614, "step": 289750 }, { "epoch": 0.0134656, "grad_norm": 0.8266696929931641, "learning_rate": 3.121345447814731e-06, "loss": 2.8161, "step": 289760 }, { "epoch": 0.0134912, "grad_norm": 1.5570482015609741, "learning_rate": 3.1207609682264116e-06, "loss": 3.3398, "step": 289770 }, { "epoch": 0.0135168, "grad_norm": 0.8663159012794495, "learning_rate": 3.120176533248129e-06, "loss": 2.925, "step": 289780 }, { "epoch": 0.0135424, "grad_norm": 0.8389260172843933, "learning_rate": 3.1195921428836695e-06, "loss": 2.8034, "step": 289790 }, { "epoch": 0.013568, "grad_norm": 1.617092251777649, "learning_rate": 3.1190077971368305e-06, "loss": 2.8736, "step": 289800 }, { "epoch": 0.0135936, "grad_norm": 0.8867846131324768, "learning_rate": 3.1184234960113957e-06, "loss": 2.9371, "step": 289810 }, { "epoch": 0.0136192, "grad_norm": 0.7800809741020203, "learning_rate": 3.11783923951116e-06, "loss": 2.8276, "step": 289820 }, { "epoch": 0.0136448, "grad_norm": 0.7336079478263855, "learning_rate": 3.117255027639905e-06, "loss": 2.7838, "step": 289830 }, { "epoch": 0.0136704, "grad_norm": 0.7574891448020935, "learning_rate": 3.1166708604014207e-06, "loss": 2.6626, "step": 289840 }, { "epoch": 0.013696, "grad_norm": 0.8129856586456299, "learning_rate": 3.1160867377994976e-06, "loss": 2.9322, "step": 289850 }, { "epoch": 0.0137216, "grad_norm": 1.055050015449524, "learning_rate": 3.115502659837921e-06, "loss": 2.8206, "step": 289860 }, { "epoch": 0.0137472, "grad_norm": 0.7867077589035034, "learning_rate": 3.114918626520481e-06, "loss": 2.4502, "step": 289870 }, { "epoch": 0.0137728, "grad_norm": 0.7941999435424805, "learning_rate": 3.114334637850964e-06, "loss": 2.8729, "step": 289880 }, { "epoch": 0.0137984, "grad_norm": 0.7385810613632202, "learning_rate": 3.1137506938331597e-06, "loss": 2.614, "step": 289890 }, { "epoch": 0.013824, "grad_norm": 0.8202453851699829, "learning_rate": 3.113166794470849e-06, "loss": 2.7609, "step": 289900 }, { "epoch": 0.0138496, "grad_norm": 0.8141583800315857, "learning_rate": 3.112582939767821e-06, "loss": 2.7791, "step": 289910 }, { "epoch": 0.0138752, "grad_norm": 0.7263689637184143, "learning_rate": 3.1119991297278628e-06, "loss": 2.9086, "step": 289920 }, { "epoch": 0.0139008, "grad_norm": 0.759699821472168, "learning_rate": 3.1114153643547596e-06, "loss": 2.7501, "step": 289930 }, { "epoch": 0.0139264, "grad_norm": 0.8872852325439453, "learning_rate": 3.1108316436522967e-06, "loss": 2.9941, "step": 289940 }, { "epoch": 0.013952, "grad_norm": 0.8313525915145874, "learning_rate": 3.110247967624259e-06, "loss": 2.9352, "step": 289950 }, { "epoch": 0.0139776, "grad_norm": 0.7464190125465393, "learning_rate": 3.109664336274436e-06, "loss": 2.6038, "step": 289960 }, { "epoch": 0.0140032, "grad_norm": 0.81333988904953, "learning_rate": 3.109080749606602e-06, "loss": 2.713, "step": 289970 }, { "epoch": 0.0140288, "grad_norm": 0.7746862173080444, "learning_rate": 3.1084972076245514e-06, "loss": 2.8834, "step": 289980 }, { "epoch": 0.0140544, "grad_norm": 0.8411521911621094, "learning_rate": 3.1079137103320656e-06, "loss": 2.918, "step": 289990 }, { "epoch": 0.01408, "grad_norm": 0.9895440340042114, "learning_rate": 3.107330257732927e-06, "loss": 2.8347, "step": 290000 }, { "epoch": 0.0141056, "grad_norm": 0.8478321433067322, "learning_rate": 3.1067468498309193e-06, "loss": 2.8427, "step": 290010 }, { "epoch": 0.0141312, "grad_norm": 0.9451937079429626, "learning_rate": 3.1061634866298295e-06, "loss": 2.8992, "step": 290020 }, { "epoch": 0.0141568, "grad_norm": 1.1400090456008911, "learning_rate": 3.105580168133434e-06, "loss": 2.8171, "step": 290030 }, { "epoch": 0.0141824, "grad_norm": 0.7221740484237671, "learning_rate": 3.10499689434552e-06, "loss": 2.7085, "step": 290040 }, { "epoch": 0.014208, "grad_norm": 0.8335179090499878, "learning_rate": 3.1044136652698674e-06, "loss": 2.845, "step": 290050 }, { "epoch": 0.0142336, "grad_norm": 1.0015162229537964, "learning_rate": 3.1038304809102604e-06, "loss": 2.9556, "step": 290060 }, { "epoch": 0.0142592, "grad_norm": 0.8934453129768372, "learning_rate": 3.10324734127048e-06, "loss": 2.8236, "step": 290070 }, { "epoch": 0.0142848, "grad_norm": 0.8399970531463623, "learning_rate": 3.1026642463543067e-06, "loss": 2.816, "step": 290080 }, { "epoch": 0.0143104, "grad_norm": 0.8160063028335571, "learning_rate": 3.102081196165527e-06, "loss": 2.9325, "step": 290090 }, { "epoch": 0.014336, "grad_norm": 0.885333240032196, "learning_rate": 3.101498190707913e-06, "loss": 2.823, "step": 290100 }, { "epoch": 0.0143616, "grad_norm": 0.9128190875053406, "learning_rate": 3.100915229985251e-06, "loss": 2.6918, "step": 290110 }, { "epoch": 0.0143872, "grad_norm": 0.7821437120437622, "learning_rate": 3.1003323140013196e-06, "loss": 2.7436, "step": 290120 }, { "epoch": 0.0144128, "grad_norm": 1.0689939260482788, "learning_rate": 3.0997494427598997e-06, "loss": 2.8038, "step": 290130 }, { "epoch": 0.0144384, "grad_norm": 0.8568644523620605, "learning_rate": 3.099166616264768e-06, "loss": 2.8379, "step": 290140 }, { "epoch": 0.014464, "grad_norm": 0.8145362734794617, "learning_rate": 3.0985838345197126e-06, "loss": 2.7901, "step": 290150 }, { "epoch": 0.0144896, "grad_norm": 0.8234432339668274, "learning_rate": 3.0980010975285035e-06, "loss": 3.0081, "step": 290160 }, { "epoch": 0.0145152, "grad_norm": 0.8389308452606201, "learning_rate": 3.097418405294923e-06, "loss": 2.8385, "step": 290170 }, { "epoch": 0.0145408, "grad_norm": 0.7439853549003601, "learning_rate": 3.0968357578227504e-06, "loss": 2.9725, "step": 290180 }, { "epoch": 0.0145664, "grad_norm": 1.0276057720184326, "learning_rate": 3.0962531551157626e-06, "loss": 2.7511, "step": 290190 }, { "epoch": 0.014592, "grad_norm": 0.885869026184082, "learning_rate": 3.095670597177738e-06, "loss": 2.9656, "step": 290200 }, { "epoch": 0.0146176, "grad_norm": 0.8098225593566895, "learning_rate": 3.0950880840124554e-06, "loss": 2.9815, "step": 290210 }, { "epoch": 0.0146432, "grad_norm": 0.829012393951416, "learning_rate": 3.094505615623694e-06, "loss": 2.7722, "step": 290220 }, { "epoch": 0.0146688, "grad_norm": 1.5131175518035889, "learning_rate": 3.093923192015226e-06, "loss": 3.0061, "step": 290230 }, { "epoch": 0.0146944, "grad_norm": 0.8926732540130615, "learning_rate": 3.0933408131908296e-06, "loss": 2.8721, "step": 290240 }, { "epoch": 0.01472, "grad_norm": 0.8566213250160217, "learning_rate": 3.092758479154283e-06, "loss": 2.8955, "step": 290250 }, { "epoch": 0.0147456, "grad_norm": 0.7910241484642029, "learning_rate": 3.092176189909363e-06, "loss": 2.9953, "step": 290260 }, { "epoch": 0.0147712, "grad_norm": 0.8095777630805969, "learning_rate": 3.0915939454598433e-06, "loss": 2.8923, "step": 290270 }, { "epoch": 0.0147968, "grad_norm": 0.9031580090522766, "learning_rate": 3.0910117458095e-06, "loss": 2.7454, "step": 290280 }, { "epoch": 0.0148224, "grad_norm": 2.076504707336426, "learning_rate": 3.090429590962114e-06, "loss": 2.7373, "step": 290290 }, { "epoch": 0.014848, "grad_norm": 1.0187838077545166, "learning_rate": 3.0898474809214527e-06, "loss": 2.8435, "step": 290300 }, { "epoch": 0.0148736, "grad_norm": 0.9425445199012756, "learning_rate": 3.0892654156912926e-06, "loss": 2.9629, "step": 290310 }, { "epoch": 0.0148992, "grad_norm": 0.7472952008247375, "learning_rate": 3.088683395275408e-06, "loss": 2.8631, "step": 290320 }, { "epoch": 0.0149248, "grad_norm": 0.800089418888092, "learning_rate": 3.088101419677577e-06, "loss": 2.9036, "step": 290330 }, { "epoch": 0.0149504, "grad_norm": 0.8550286293029785, "learning_rate": 3.0875194889015713e-06, "loss": 2.9958, "step": 290340 }, { "epoch": 0.014976, "grad_norm": 0.8869839906692505, "learning_rate": 3.0869376029511687e-06, "loss": 3.0715, "step": 290350 }, { "epoch": 0.0150016, "grad_norm": 0.7836545705795288, "learning_rate": 3.0863557618301344e-06, "loss": 2.6274, "step": 290360 }, { "epoch": 0.0150272, "grad_norm": 0.8320878744125366, "learning_rate": 3.085773965542246e-06, "loss": 2.9285, "step": 290370 }, { "epoch": 0.0150528, "grad_norm": 0.8845234513282776, "learning_rate": 3.085192214091275e-06, "loss": 2.8943, "step": 290380 }, { "epoch": 0.0150784, "grad_norm": 1.0392791032791138, "learning_rate": 3.0846105074809963e-06, "loss": 2.8873, "step": 290390 }, { "epoch": 0.015104, "grad_norm": 0.8351988196372986, "learning_rate": 3.084028845715179e-06, "loss": 2.7205, "step": 290400 }, { "epoch": 0.0151296, "grad_norm": 1.0344637632369995, "learning_rate": 3.083447228797597e-06, "loss": 2.9763, "step": 290410 }, { "epoch": 0.0151552, "grad_norm": 0.7125838994979858, "learning_rate": 3.082865656732026e-06, "loss": 2.6428, "step": 290420 }, { "epoch": 0.0151808, "grad_norm": 0.7848168015480042, "learning_rate": 3.0822841295222285e-06, "loss": 2.6515, "step": 290430 }, { "epoch": 0.0152064, "grad_norm": 1.9469645023345947, "learning_rate": 3.0817026471719803e-06, "loss": 3.0645, "step": 290440 }, { "epoch": 0.015232, "grad_norm": 1.01844322681427, "learning_rate": 3.081121209685051e-06, "loss": 2.7774, "step": 290450 }, { "epoch": 0.0152576, "grad_norm": 0.7803349494934082, "learning_rate": 3.080539817065212e-06, "loss": 2.9732, "step": 290460 }, { "epoch": 0.0152832, "grad_norm": 0.7939924597740173, "learning_rate": 3.0799584693162344e-06, "loss": 3.0169, "step": 290470 }, { "epoch": 0.0153088, "grad_norm": 0.7780846953392029, "learning_rate": 3.0793771664418857e-06, "loss": 2.9479, "step": 290480 }, { "epoch": 0.0153344, "grad_norm": 0.7617605924606323, "learning_rate": 3.0787959084459375e-06, "loss": 2.859, "step": 290490 }, { "epoch": 0.01536, "grad_norm": 0.8162053823471069, "learning_rate": 3.0782146953321577e-06, "loss": 2.9228, "step": 290500 }, { "epoch": 0.0153856, "grad_norm": 0.8565866351127625, "learning_rate": 3.077633527104317e-06, "loss": 2.8931, "step": 290510 }, { "epoch": 0.0154112, "grad_norm": 0.8189916610717773, "learning_rate": 3.077052403766183e-06, "loss": 2.8547, "step": 290520 }, { "epoch": 0.0154368, "grad_norm": 0.771709680557251, "learning_rate": 3.076471325321524e-06, "loss": 2.9326, "step": 290530 }, { "epoch": 0.0154624, "grad_norm": 0.7733650803565979, "learning_rate": 3.0758902917741086e-06, "loss": 2.8296, "step": 290540 }, { "epoch": 0.015488, "grad_norm": 2.194923162460327, "learning_rate": 3.075309303127707e-06, "loss": 2.7797, "step": 290550 }, { "epoch": 0.0155136, "grad_norm": 0.8736094236373901, "learning_rate": 3.074728359386081e-06, "loss": 2.9124, "step": 290560 }, { "epoch": 0.0155392, "grad_norm": 0.8562620282173157, "learning_rate": 3.074147460553002e-06, "loss": 2.7855, "step": 290570 }, { "epoch": 0.0155648, "grad_norm": 0.8061885833740234, "learning_rate": 3.0735666066322354e-06, "loss": 2.9604, "step": 290580 }, { "epoch": 0.0155904, "grad_norm": 2.688748598098755, "learning_rate": 3.072985797627549e-06, "loss": 2.8026, "step": 290590 }, { "epoch": 0.015616, "grad_norm": 0.8127325773239136, "learning_rate": 3.072405033542708e-06, "loss": 2.9543, "step": 290600 }, { "epoch": 0.0156416, "grad_norm": 1.0409985780715942, "learning_rate": 3.0718243143814798e-06, "loss": 2.7709, "step": 290610 }, { "epoch": 0.0156672, "grad_norm": 0.7275015711784363, "learning_rate": 3.071243640147633e-06, "loss": 2.9367, "step": 290620 }, { "epoch": 0.0156928, "grad_norm": 1.1212031841278076, "learning_rate": 3.0706630108449263e-06, "loss": 2.8287, "step": 290630 }, { "epoch": 0.0157184, "grad_norm": 0.9376404881477356, "learning_rate": 3.0700824264771278e-06, "loss": 2.818, "step": 290640 }, { "epoch": 0.015744, "grad_norm": 1.079490065574646, "learning_rate": 3.069501887048004e-06, "loss": 3.0781, "step": 290650 }, { "epoch": 0.0157696, "grad_norm": 0.8149916529655457, "learning_rate": 3.068921392561315e-06, "loss": 2.8885, "step": 290660 }, { "epoch": 0.0157952, "grad_norm": 0.8219143152236938, "learning_rate": 3.0683409430208344e-06, "loss": 2.8554, "step": 290670 }, { "epoch": 0.0158208, "grad_norm": 0.7223184704780579, "learning_rate": 3.0677605384303176e-06, "loss": 2.8604, "step": 290680 }, { "epoch": 0.0158464, "grad_norm": 0.817711353302002, "learning_rate": 3.0671801787935317e-06, "loss": 2.9313, "step": 290690 }, { "epoch": 0.015872, "grad_norm": 0.808319091796875, "learning_rate": 3.06659986411424e-06, "loss": 2.9914, "step": 290700 }, { "epoch": 0.0158976, "grad_norm": 0.7258265018463135, "learning_rate": 3.0660195943962057e-06, "loss": 2.7625, "step": 290710 }, { "epoch": 0.0159232, "grad_norm": 0.8455238938331604, "learning_rate": 3.0654393696431905e-06, "loss": 2.9615, "step": 290720 }, { "epoch": 0.0159488, "grad_norm": 0.882511556148529, "learning_rate": 3.0648591898589598e-06, "loss": 2.8117, "step": 290730 }, { "epoch": 0.0159744, "grad_norm": 0.8531449437141418, "learning_rate": 3.0642790550472725e-06, "loss": 2.6322, "step": 290740 }, { "epoch": 0.016, "grad_norm": 0.8882771134376526, "learning_rate": 3.0636989652118954e-06, "loss": 2.8234, "step": 290750 }, { "epoch": 0.0160256, "grad_norm": 0.7690314054489136, "learning_rate": 3.0631189203565847e-06, "loss": 2.6817, "step": 290760 }, { "epoch": 0.0160512, "grad_norm": 0.8698864579200745, "learning_rate": 3.0625389204851043e-06, "loss": 2.8442, "step": 290770 }, { "epoch": 0.0160768, "grad_norm": 0.9014601111412048, "learning_rate": 3.061958965601214e-06, "loss": 2.7593, "step": 290780 }, { "epoch": 0.0161024, "grad_norm": 0.9893926382064819, "learning_rate": 3.0613790557086764e-06, "loss": 2.7865, "step": 290790 }, { "epoch": 0.016128, "grad_norm": 1.0584909915924072, "learning_rate": 3.0607991908112524e-06, "loss": 3.0995, "step": 290800 }, { "epoch": 0.0161536, "grad_norm": 0.7946230173110962, "learning_rate": 3.0602193709127006e-06, "loss": 3.1613, "step": 290810 }, { "epoch": 0.0161792, "grad_norm": 0.7937042117118835, "learning_rate": 3.0596395960167856e-06, "loss": 2.8196, "step": 290820 }, { "epoch": 0.0162048, "grad_norm": 0.8801640272140503, "learning_rate": 3.05905986612726e-06, "loss": 2.8133, "step": 290830 }, { "epoch": 0.0162304, "grad_norm": 1.4034335613250732, "learning_rate": 3.0584801812478825e-06, "loss": 2.9167, "step": 290840 }, { "epoch": 0.016256, "grad_norm": 0.9137220978736877, "learning_rate": 3.0579005413824203e-06, "loss": 3.1357, "step": 290850 }, { "epoch": 0.0162816, "grad_norm": 0.8351515531539917, "learning_rate": 3.0573209465346276e-06, "loss": 2.8481, "step": 290860 }, { "epoch": 0.0163072, "grad_norm": 0.8762027621269226, "learning_rate": 3.056741396708267e-06, "loss": 2.7243, "step": 290870 }, { "epoch": 0.0163328, "grad_norm": 0.8934824466705322, "learning_rate": 3.056161891907089e-06, "loss": 2.7855, "step": 290880 }, { "epoch": 0.0163584, "grad_norm": 0.8587654829025269, "learning_rate": 3.055582432134857e-06, "loss": 2.8717, "step": 290890 }, { "epoch": 0.016384, "grad_norm": 0.8735856413841248, "learning_rate": 3.055003017395326e-06, "loss": 2.8049, "step": 290900 }, { "epoch": 0.0164096, "grad_norm": 0.818666398525238, "learning_rate": 3.054423647692256e-06, "loss": 2.906, "step": 290910 }, { "epoch": 0.0164352, "grad_norm": 0.9347618222236633, "learning_rate": 3.0538443230294025e-06, "loss": 2.8262, "step": 290920 }, { "epoch": 0.0164608, "grad_norm": 0.8483183979988098, "learning_rate": 3.0532650434105217e-06, "loss": 2.8281, "step": 290930 }, { "epoch": 0.0164864, "grad_norm": 0.7917776107788086, "learning_rate": 3.052685808839371e-06, "loss": 2.9419, "step": 290940 }, { "epoch": 0.016512, "grad_norm": 0.8636927604675293, "learning_rate": 3.05210661931971e-06, "loss": 2.7938, "step": 290950 }, { "epoch": 0.0165376, "grad_norm": 0.8851959109306335, "learning_rate": 3.051527474855288e-06, "loss": 2.8522, "step": 290960 }, { "epoch": 0.0165632, "grad_norm": 0.8355017900466919, "learning_rate": 3.0509483754498636e-06, "loss": 2.8345, "step": 290970 }, { "epoch": 0.0165888, "grad_norm": 1.1390886306762695, "learning_rate": 3.0503693211071916e-06, "loss": 3.0009, "step": 290980 }, { "epoch": 0.0166144, "grad_norm": 1.0924272537231445, "learning_rate": 3.049790311831028e-06, "loss": 2.9145, "step": 290990 }, { "epoch": 0.01664, "grad_norm": 0.829034686088562, "learning_rate": 3.0492113476251274e-06, "loss": 2.7706, "step": 291000 }, { "epoch": 0.0166656, "grad_norm": 1.0138674974441528, "learning_rate": 3.048632428493243e-06, "loss": 2.8556, "step": 291010 }, { "epoch": 0.0166912, "grad_norm": 0.7899671792984009, "learning_rate": 3.0480535544391298e-06, "loss": 2.7475, "step": 291020 }, { "epoch": 0.0167168, "grad_norm": 0.9180470705032349, "learning_rate": 3.0474747254665425e-06, "loss": 2.7951, "step": 291030 }, { "epoch": 0.0167424, "grad_norm": 0.8421278595924377, "learning_rate": 3.046895941579234e-06, "loss": 2.9546, "step": 291040 }, { "epoch": 0.016768, "grad_norm": 0.8751968145370483, "learning_rate": 3.0463172027809575e-06, "loss": 3.0049, "step": 291050 }, { "epoch": 0.0167936, "grad_norm": 0.8426350355148315, "learning_rate": 3.0457385090754644e-06, "loss": 3.1379, "step": 291060 }, { "epoch": 0.0168192, "grad_norm": 1.1913044452667236, "learning_rate": 3.0451598604665133e-06, "loss": 2.8128, "step": 291070 }, { "epoch": 0.0168448, "grad_norm": 0.7507283091545105, "learning_rate": 3.0445812569578492e-06, "loss": 2.7022, "step": 291080 }, { "epoch": 0.0168704, "grad_norm": 1.0621010065078735, "learning_rate": 3.0440026985532267e-06, "loss": 3.3395, "step": 291090 }, { "epoch": 0.016896, "grad_norm": 0.7816777229309082, "learning_rate": 3.043424185256397e-06, "loss": 3.0946, "step": 291100 }, { "epoch": 2.56e-05, "grad_norm": 0.8365731239318848, "learning_rate": 3.0428457170711144e-06, "loss": 1.9722, "step": 291110 }, { "epoch": 5.12e-05, "grad_norm": 0.8084228038787842, "learning_rate": 3.0422672940011268e-06, "loss": 2.7435, "step": 291120 }, { "epoch": 7.68e-05, "grad_norm": 0.8730597496032715, "learning_rate": 3.041688916050187e-06, "loss": 2.8814, "step": 291130 }, { "epoch": 0.0001024, "grad_norm": 0.8506877422332764, "learning_rate": 3.0411105832220467e-06, "loss": 2.8651, "step": 291140 }, { "epoch": 0.000128, "grad_norm": 0.8594673871994019, "learning_rate": 3.0405322955204563e-06, "loss": 2.4028, "step": 291150 }, { "epoch": 0.0001536, "grad_norm": 0.852390706539154, "learning_rate": 3.039954052949162e-06, "loss": 3.065, "step": 291160 }, { "epoch": 0.0001792, "grad_norm": 0.8023937344551086, "learning_rate": 3.0393758555119156e-06, "loss": 2.3148, "step": 291170 }, { "epoch": 0.0002048, "grad_norm": 0.8803454637527466, "learning_rate": 3.038797703212467e-06, "loss": 2.8383, "step": 291180 }, { "epoch": 0.0002304, "grad_norm": 0.7840575575828552, "learning_rate": 3.038219596054561e-06, "loss": 2.5448, "step": 291190 }, { "epoch": 0.000256, "grad_norm": 0.8669385313987732, "learning_rate": 3.0376415340419587e-06, "loss": 2.6704, "step": 291200 }, { "epoch": 0.0002816, "grad_norm": 0.9156582951545715, "learning_rate": 3.037063517178397e-06, "loss": 2.5015, "step": 291210 }, { "epoch": 0.0003072, "grad_norm": 0.7752838134765625, "learning_rate": 3.036485545467628e-06, "loss": 2.4903, "step": 291220 }, { "epoch": 0.0003328, "grad_norm": 0.9329352974891663, "learning_rate": 3.0359076189133996e-06, "loss": 2.7271, "step": 291230 }, { "epoch": 0.0003584, "grad_norm": 0.8734314441680908, "learning_rate": 3.0353297375194592e-06, "loss": 2.779, "step": 291240 }, { "epoch": 0.000384, "grad_norm": 0.757198691368103, "learning_rate": 3.0347519012895545e-06, "loss": 2.6856, "step": 291250 }, { "epoch": 0.0004096, "grad_norm": 0.8898646831512451, "learning_rate": 3.0341741102274313e-06, "loss": 2.7767, "step": 291260 }, { "epoch": 0.0004352, "grad_norm": 0.9805929064750671, "learning_rate": 3.0335963643368426e-06, "loss": 2.5588, "step": 291270 }, { "epoch": 0.0004608, "grad_norm": 0.7660136222839355, "learning_rate": 3.0330186636215266e-06, "loss": 2.7285, "step": 291280 }, { "epoch": 0.0004864, "grad_norm": 0.9242696762084961, "learning_rate": 3.032441008085232e-06, "loss": 2.8175, "step": 291290 }, { "epoch": 0.000512, "grad_norm": 0.8827329277992249, "learning_rate": 3.031863397731707e-06, "loss": 2.5691, "step": 291300 }, { "epoch": 0.0005376, "grad_norm": 0.8435514569282532, "learning_rate": 3.0312858325646956e-06, "loss": 2.6212, "step": 291310 }, { "epoch": 0.0005632, "grad_norm": 0.8493874669075012, "learning_rate": 3.0307083125879422e-06, "loss": 2.48, "step": 291320 }, { "epoch": 0.0005888, "grad_norm": 0.8253639936447144, "learning_rate": 3.0301308378051954e-06, "loss": 2.5333, "step": 291330 }, { "epoch": 0.0006144, "grad_norm": 0.8589767813682556, "learning_rate": 3.0295534082201995e-06, "loss": 2.5998, "step": 291340 }, { "epoch": 0.00064, "grad_norm": 0.7462946176528931, "learning_rate": 3.0289760238366937e-06, "loss": 2.6307, "step": 291350 }, { "epoch": 0.0006656, "grad_norm": 1.0252890586853027, "learning_rate": 3.0283986846584233e-06, "loss": 2.7691, "step": 291360 }, { "epoch": 0.0006912, "grad_norm": 0.7944760918617249, "learning_rate": 3.0278213906891375e-06, "loss": 2.7051, "step": 291370 }, { "epoch": 0.0007168, "grad_norm": 0.7315673828125, "learning_rate": 3.0272441419325773e-06, "loss": 2.4696, "step": 291380 }, { "epoch": 0.0007424, "grad_norm": 0.8891987800598145, "learning_rate": 3.0266669383924853e-06, "loss": 2.6952, "step": 291390 }, { "epoch": 0.000768, "grad_norm": 0.843396782875061, "learning_rate": 3.0260897800726084e-06, "loss": 2.6209, "step": 291400 }, { "epoch": 0.0007936, "grad_norm": 0.7990409135818481, "learning_rate": 3.025512666976682e-06, "loss": 2.6391, "step": 291410 }, { "epoch": 0.0008192, "grad_norm": 0.7645835280418396, "learning_rate": 3.0249355991084538e-06, "loss": 2.3504, "step": 291420 }, { "epoch": 0.0008448, "grad_norm": 0.8901993632316589, "learning_rate": 3.0243585764716634e-06, "loss": 2.7054, "step": 291430 }, { "epoch": 0.0008704, "grad_norm": 1.6074901819229126, "learning_rate": 3.0237815990700535e-06, "loss": 2.6465, "step": 291440 }, { "epoch": 0.000896, "grad_norm": 1.0178388357162476, "learning_rate": 3.0232046669073666e-06, "loss": 2.751, "step": 291450 }, { "epoch": 0.0009216, "grad_norm": 0.8184998035430908, "learning_rate": 3.022627779987343e-06, "loss": 2.6555, "step": 291460 }, { "epoch": 0.0009472, "grad_norm": 0.8120680451393127, "learning_rate": 3.0220509383137266e-06, "loss": 2.6631, "step": 291470 }, { "epoch": 0.0009728, "grad_norm": 0.8008124828338623, "learning_rate": 3.0214741418902527e-06, "loss": 2.8009, "step": 291480 }, { "epoch": 0.0009984, "grad_norm": 0.7894353270530701, "learning_rate": 3.020897390720664e-06, "loss": 2.5968, "step": 291490 }, { "epoch": 0.001024, "grad_norm": 0.8196423053741455, "learning_rate": 3.020320684808702e-06, "loss": 2.7475, "step": 291500 }, { "epoch": 0.0010496, "grad_norm": 0.8754448890686035, "learning_rate": 3.0197440241581043e-06, "loss": 2.6374, "step": 291510 }, { "epoch": 0.0010752, "grad_norm": 0.9098632335662842, "learning_rate": 3.0191674087726107e-06, "loss": 2.6775, "step": 291520 }, { "epoch": 0.0011008, "grad_norm": 0.8202005624771118, "learning_rate": 3.0185908386559624e-06, "loss": 2.5498, "step": 291530 }, { "epoch": 0.0011264, "grad_norm": 0.8022459745407104, "learning_rate": 3.0180143138118966e-06, "loss": 2.7175, "step": 291540 }, { "epoch": 0.001152, "grad_norm": 0.8260623812675476, "learning_rate": 3.0174378342441534e-06, "loss": 2.6214, "step": 291550 }, { "epoch": 0.0011776, "grad_norm": 0.8861662745475769, "learning_rate": 3.016861399956469e-06, "loss": 2.6617, "step": 291560 }, { "epoch": 0.0012032, "grad_norm": 0.9690107703208923, "learning_rate": 3.0162850109525833e-06, "loss": 2.6162, "step": 291570 }, { "epoch": 0.0012288, "grad_norm": 0.8428665995597839, "learning_rate": 3.0157086672362324e-06, "loss": 2.7266, "step": 291580 }, { "epoch": 0.0012544, "grad_norm": 0.8103495240211487, "learning_rate": 3.0151323688111546e-06, "loss": 2.828, "step": 291590 }, { "epoch": 0.00128, "grad_norm": 0.8673279881477356, "learning_rate": 3.0145561156810908e-06, "loss": 2.8206, "step": 291600 }, { "epoch": 0.0013056, "grad_norm": 0.9528937935829163, "learning_rate": 3.01397990784977e-06, "loss": 2.7026, "step": 291610 }, { "epoch": 0.0013312, "grad_norm": 0.8533841967582703, "learning_rate": 3.013403745320934e-06, "loss": 2.6612, "step": 291620 }, { "epoch": 0.0013568, "grad_norm": 1.1032335758209229, "learning_rate": 3.0128276280983173e-06, "loss": 2.7396, "step": 291630 }, { "epoch": 0.0013824, "grad_norm": 1.1761127710342407, "learning_rate": 3.012251556185657e-06, "loss": 2.8376, "step": 291640 }, { "epoch": 0.001408, "grad_norm": 0.970345675945282, "learning_rate": 3.011675529586687e-06, "loss": 2.7222, "step": 291650 }, { "epoch": 0.0014336, "grad_norm": 0.7690395712852478, "learning_rate": 3.0110995483051452e-06, "loss": 2.6173, "step": 291660 }, { "epoch": 0.0014592, "grad_norm": 0.8313828110694885, "learning_rate": 3.0105236123447677e-06, "loss": 2.6298, "step": 291670 }, { "epoch": 0.0014848, "grad_norm": 0.8216176629066467, "learning_rate": 3.009947721709284e-06, "loss": 2.8983, "step": 291680 }, { "epoch": 0.0015104, "grad_norm": 0.8208276629447937, "learning_rate": 3.0093718764024325e-06, "loss": 2.6129, "step": 291690 }, { "epoch": 0.001536, "grad_norm": 0.8372732400894165, "learning_rate": 3.0087960764279457e-06, "loss": 2.3392, "step": 291700 }, { "epoch": 0.0015616, "grad_norm": 0.8984838128089905, "learning_rate": 3.0082203217895556e-06, "loss": 2.4508, "step": 291710 }, { "epoch": 0.0015872, "grad_norm": 0.9156449437141418, "learning_rate": 3.0076446124910007e-06, "loss": 2.7328, "step": 291720 }, { "epoch": 0.0016128, "grad_norm": 0.9060423970222473, "learning_rate": 3.007068948536016e-06, "loss": 2.6179, "step": 291730 }, { "epoch": 0.0016384, "grad_norm": 0.9108005166053772, "learning_rate": 3.006493329928327e-06, "loss": 2.6584, "step": 291740 }, { "epoch": 0.001664, "grad_norm": 0.8894887566566467, "learning_rate": 3.0059177566716712e-06, "loss": 2.6254, "step": 291750 }, { "epoch": 0.0016896, "grad_norm": 0.8910499811172485, "learning_rate": 3.0053422287697786e-06, "loss": 2.8163, "step": 291760 }, { "epoch": 0.0017152, "grad_norm": 0.7760960459709167, "learning_rate": 3.0047667462263843e-06, "loss": 2.6294, "step": 291770 }, { "epoch": 0.0017408, "grad_norm": 0.7650084495544434, "learning_rate": 3.004191309045217e-06, "loss": 2.6699, "step": 291780 }, { "epoch": 0.0017664, "grad_norm": 0.8944633603096008, "learning_rate": 3.0036159172300105e-06, "loss": 2.8201, "step": 291790 }, { "epoch": 0.001792, "grad_norm": 0.8241269588470459, "learning_rate": 3.003040570784499e-06, "loss": 2.6431, "step": 291800 }, { "epoch": 0.0018176, "grad_norm": 0.8248049020767212, "learning_rate": 3.0024652697124057e-06, "loss": 2.4667, "step": 291810 }, { "epoch": 0.0018432, "grad_norm": 0.8302156329154968, "learning_rate": 3.001890014017466e-06, "loss": 2.6372, "step": 291820 }, { "epoch": 0.0018688, "grad_norm": 0.7539719939231873, "learning_rate": 3.001314803703409e-06, "loss": 2.4583, "step": 291830 }, { "epoch": 0.0018944, "grad_norm": 0.9881533980369568, "learning_rate": 3.000739638773966e-06, "loss": 2.6373, "step": 291840 }, { "epoch": 0.00192, "grad_norm": 0.7560387849807739, "learning_rate": 3.0001645192328665e-06, "loss": 2.4361, "step": 291850 }, { "epoch": 0.0019456, "grad_norm": 0.8208013772964478, "learning_rate": 2.999589445083839e-06, "loss": 2.6235, "step": 291860 }, { "epoch": 0.0019712, "grad_norm": 0.7317057251930237, "learning_rate": 2.999014416330618e-06, "loss": 2.8762, "step": 291870 }, { "epoch": 0.0019968, "grad_norm": 0.9228572845458984, "learning_rate": 2.9984394329769206e-06, "loss": 2.4854, "step": 291880 }, { "epoch": 0.0020224, "grad_norm": 0.7851486802101135, "learning_rate": 2.9978644950264856e-06, "loss": 2.5448, "step": 291890 }, { "epoch": 0.002048, "grad_norm": 0.8785938620567322, "learning_rate": 2.997289602483039e-06, "loss": 2.64, "step": 291900 }, { "epoch": 0.0020736, "grad_norm": 1.5083576440811157, "learning_rate": 2.9967147553503073e-06, "loss": 2.7577, "step": 291910 }, { "epoch": 0.0020992, "grad_norm": 0.9578819274902344, "learning_rate": 2.99613995363202e-06, "loss": 2.4768, "step": 291920 }, { "epoch": 0.0021248, "grad_norm": 0.744651734828949, "learning_rate": 2.9955651973319065e-06, "loss": 2.6835, "step": 291930 }, { "epoch": 0.0021504, "grad_norm": 0.7637271285057068, "learning_rate": 2.9949904864536883e-06, "loss": 2.7638, "step": 291940 }, { "epoch": 0.002176, "grad_norm": 0.8219293355941772, "learning_rate": 2.9944158210010954e-06, "loss": 2.8404, "step": 291950 }, { "epoch": 0.0022016, "grad_norm": 0.9027038216590881, "learning_rate": 2.993841200977853e-06, "loss": 2.4802, "step": 291960 }, { "epoch": 0.0022272, "grad_norm": 0.7597752809524536, "learning_rate": 2.9932666263876887e-06, "loss": 2.7372, "step": 291970 }, { "epoch": 0.0022528, "grad_norm": 0.8189398050308228, "learning_rate": 2.992692097234329e-06, "loss": 2.943, "step": 291980 }, { "epoch": 0.0022784, "grad_norm": 1.0709706544876099, "learning_rate": 2.992117613521498e-06, "loss": 2.7013, "step": 291990 }, { "epoch": 0.002304, "grad_norm": 0.8313784599304199, "learning_rate": 2.9915431752529246e-06, "loss": 2.9878, "step": 292000 }, { "epoch": 0.0023296, "grad_norm": 1.1962921619415283, "learning_rate": 2.9909687824323285e-06, "loss": 2.7516, "step": 292010 }, { "epoch": 0.0023552, "grad_norm": 0.895389199256897, "learning_rate": 2.9903944350634373e-06, "loss": 2.673, "step": 292020 }, { "epoch": 0.0023808, "grad_norm": 0.7935490608215332, "learning_rate": 2.9898201331499755e-06, "loss": 2.6667, "step": 292030 }, { "epoch": 0.0024064, "grad_norm": 0.7898461818695068, "learning_rate": 2.9892458766956657e-06, "loss": 2.6895, "step": 292040 }, { "epoch": 0.002432, "grad_norm": 0.8126165866851807, "learning_rate": 2.988671665704235e-06, "loss": 2.5984, "step": 292050 }, { "epoch": 0.0024576, "grad_norm": 0.7846565246582031, "learning_rate": 2.9880975001794033e-06, "loss": 2.6484, "step": 292060 }, { "epoch": 0.0024832, "grad_norm": 0.7998097538948059, "learning_rate": 2.987523380124897e-06, "loss": 2.7229, "step": 292070 }, { "epoch": 0.0025088, "grad_norm": 0.8251866698265076, "learning_rate": 2.986949305544438e-06, "loss": 2.6598, "step": 292080 }, { "epoch": 0.0025344, "grad_norm": 0.8000670075416565, "learning_rate": 2.9863752764417476e-06, "loss": 2.885, "step": 292090 }, { "epoch": 0.00256, "grad_norm": 0.9046807289123535, "learning_rate": 2.9858012928205503e-06, "loss": 2.7209, "step": 292100 }, { "epoch": 0.0025856, "grad_norm": 0.9842608571052551, "learning_rate": 2.985227354684568e-06, "loss": 2.8267, "step": 292110 }, { "epoch": 0.0026112, "grad_norm": 0.851435661315918, "learning_rate": 2.984653462037521e-06, "loss": 2.6134, "step": 292120 }, { "epoch": 0.0026368, "grad_norm": 0.9312133193016052, "learning_rate": 2.984079614883135e-06, "loss": 2.7577, "step": 292130 }, { "epoch": 0.0026624, "grad_norm": 0.9480482935905457, "learning_rate": 2.983505813225126e-06, "loss": 2.9536, "step": 292140 }, { "epoch": 0.002688, "grad_norm": 0.8023737072944641, "learning_rate": 2.982932057067216e-06, "loss": 2.9643, "step": 292150 }, { "epoch": 0.0027136, "grad_norm": 0.8742222189903259, "learning_rate": 2.982358346413128e-06, "loss": 2.936, "step": 292160 }, { "epoch": 0.0027392, "grad_norm": 0.9265894293785095, "learning_rate": 2.9817846812665797e-06, "loss": 2.8081, "step": 292170 }, { "epoch": 0.0027648, "grad_norm": 0.8264776468276978, "learning_rate": 2.981211061631293e-06, "loss": 2.8073, "step": 292180 }, { "epoch": 0.0027904, "grad_norm": 0.7394962906837463, "learning_rate": 2.980637487510988e-06, "loss": 2.7648, "step": 292190 }, { "epoch": 0.002816, "grad_norm": 0.8271269202232361, "learning_rate": 2.980063958909385e-06, "loss": 2.8168, "step": 292200 }, { "epoch": 0.0028416, "grad_norm": 0.8862187266349792, "learning_rate": 2.979490475830199e-06, "loss": 2.9304, "step": 292210 }, { "epoch": 0.0028672, "grad_norm": 0.7498824596405029, "learning_rate": 2.9789170382771515e-06, "loss": 2.6974, "step": 292220 }, { "epoch": 0.0028928, "grad_norm": 0.8338369131088257, "learning_rate": 2.9783436462539583e-06, "loss": 2.906, "step": 292230 }, { "epoch": 0.0029184, "grad_norm": 0.9473158121109009, "learning_rate": 2.977770299764343e-06, "loss": 2.9043, "step": 292240 }, { "epoch": 0.002944, "grad_norm": 0.7520460486412048, "learning_rate": 2.977196998812022e-06, "loss": 2.6544, "step": 292250 }, { "epoch": 0.0029696, "grad_norm": 0.7984793186187744, "learning_rate": 2.9766237434007138e-06, "loss": 2.8192, "step": 292260 }, { "epoch": 0.0029952, "grad_norm": 0.9282498955726624, "learning_rate": 2.9760505335341316e-06, "loss": 2.5495, "step": 292270 }, { "epoch": 0.0030208, "grad_norm": 0.7719498872756958, "learning_rate": 2.9754773692159956e-06, "loss": 2.9091, "step": 292280 }, { "epoch": 0.0030464, "grad_norm": 0.9068325757980347, "learning_rate": 2.97490425045002e-06, "loss": 2.7237, "step": 292290 }, { "epoch": 0.003072, "grad_norm": 0.7411773204803467, "learning_rate": 2.9743311772399254e-06, "loss": 2.7252, "step": 292300 }, { "epoch": 0.0030976, "grad_norm": 0.8092767000198364, "learning_rate": 2.973758149589425e-06, "loss": 2.972, "step": 292310 }, { "epoch": 0.0031232, "grad_norm": 0.8415560722351074, "learning_rate": 2.973185167502236e-06, "loss": 2.8249, "step": 292320 }, { "epoch": 0.0031488, "grad_norm": 0.8962295055389404, "learning_rate": 2.972612230982076e-06, "loss": 2.879, "step": 292330 }, { "epoch": 0.0031744, "grad_norm": 0.8058866262435913, "learning_rate": 2.9720393400326553e-06, "loss": 2.454, "step": 292340 }, { "epoch": 0.0032, "grad_norm": 0.8080031871795654, "learning_rate": 2.971466494657692e-06, "loss": 2.813, "step": 292350 }, { "epoch": 0.0032256, "grad_norm": 0.7312645316123962, "learning_rate": 2.9708936948609e-06, "loss": 2.6844, "step": 292360 }, { "epoch": 0.0032512, "grad_norm": 0.8431812524795532, "learning_rate": 2.9703209406459955e-06, "loss": 3.0071, "step": 292370 }, { "epoch": 0.0032768, "grad_norm": 0.9031839966773987, "learning_rate": 2.9697482320166905e-06, "loss": 2.6921, "step": 292380 }, { "epoch": 0.0033024, "grad_norm": 0.755105197429657, "learning_rate": 2.9691755689766986e-06, "loss": 2.8052, "step": 292390 }, { "epoch": 0.003328, "grad_norm": 0.7591538429260254, "learning_rate": 2.9686029515297366e-06, "loss": 2.6662, "step": 292400 }, { "epoch": 0.0033536, "grad_norm": 0.8668217658996582, "learning_rate": 2.9680303796795152e-06, "loss": 2.7008, "step": 292410 }, { "epoch": 0.0033792, "grad_norm": 0.8889774680137634, "learning_rate": 2.967457853429747e-06, "loss": 2.7345, "step": 292420 }, { "epoch": 0.0034048, "grad_norm": 0.8379674553871155, "learning_rate": 2.966885372784146e-06, "loss": 2.7229, "step": 292430 }, { "epoch": 0.0034304, "grad_norm": 0.7689076662063599, "learning_rate": 2.966312937746425e-06, "loss": 2.764, "step": 292440 }, { "epoch": 0.003456, "grad_norm": 0.7546005845069885, "learning_rate": 2.965740548320294e-06, "loss": 2.6641, "step": 292450 }, { "epoch": 0.0034816, "grad_norm": 0.8539822101593018, "learning_rate": 2.96516820450947e-06, "loss": 3.019, "step": 292460 }, { "epoch": 0.0035072, "grad_norm": 0.7710260152816772, "learning_rate": 2.964595906317658e-06, "loss": 2.7679, "step": 292470 }, { "epoch": 0.0035328, "grad_norm": 0.8158957362174988, "learning_rate": 2.9640236537485714e-06, "loss": 2.8023, "step": 292480 }, { "epoch": 0.0035584, "grad_norm": 0.7787231802940369, "learning_rate": 2.9634514468059204e-06, "loss": 2.6873, "step": 292490 }, { "epoch": 0.003584, "grad_norm": 0.7930185794830322, "learning_rate": 2.9628792854934176e-06, "loss": 2.9, "step": 292500 }, { "epoch": 0.0036096, "grad_norm": 0.9217131733894348, "learning_rate": 2.962307169814772e-06, "loss": 3.0391, "step": 292510 }, { "epoch": 0.0036352, "grad_norm": 0.8656735420227051, "learning_rate": 2.9617350997736938e-06, "loss": 3.1298, "step": 292520 }, { "epoch": 0.0036608, "grad_norm": 0.8697493076324463, "learning_rate": 2.9611630753738974e-06, "loss": 2.8223, "step": 292530 }, { "epoch": 0.0036864, "grad_norm": 0.8185278177261353, "learning_rate": 2.9605910966190832e-06, "loss": 2.8005, "step": 292540 }, { "epoch": 0.003712, "grad_norm": 0.8387826085090637, "learning_rate": 2.960019163512965e-06, "loss": 2.922, "step": 292550 }, { "epoch": 0.0037376, "grad_norm": 0.8800286054611206, "learning_rate": 2.959447276059252e-06, "loss": 2.9749, "step": 292560 }, { "epoch": 0.0037632, "grad_norm": 0.8357695937156677, "learning_rate": 2.958875434261652e-06, "loss": 2.8718, "step": 292570 }, { "epoch": 0.0037888, "grad_norm": 0.8704016804695129, "learning_rate": 2.958303638123874e-06, "loss": 2.9045, "step": 292580 }, { "epoch": 0.0038144, "grad_norm": 0.794206976890564, "learning_rate": 2.9577318876496254e-06, "loss": 2.7034, "step": 292590 }, { "epoch": 0.00384, "grad_norm": 0.8138930797576904, "learning_rate": 2.957160182842613e-06, "loss": 2.7095, "step": 292600 }, { "epoch": 0.0038656, "grad_norm": 0.8322561383247375, "learning_rate": 2.956588523706546e-06, "loss": 2.7598, "step": 292610 }, { "epoch": 0.0038912, "grad_norm": 0.8360911011695862, "learning_rate": 2.9560169102451295e-06, "loss": 3.0563, "step": 292620 }, { "epoch": 0.0039168, "grad_norm": 0.8304941654205322, "learning_rate": 2.955445342462072e-06, "loss": 2.8005, "step": 292630 }, { "epoch": 0.0039424, "grad_norm": 0.8218469023704529, "learning_rate": 2.9548738203610793e-06, "loss": 2.7921, "step": 292640 }, { "epoch": 0.003968, "grad_norm": 0.8406661152839661, "learning_rate": 2.954302343945861e-06, "loss": 2.806, "step": 292650 }, { "epoch": 0.0039936, "grad_norm": 0.9081253409385681, "learning_rate": 2.9537309132201163e-06, "loss": 2.98, "step": 292660 }, { "epoch": 0.0040192, "grad_norm": 0.9629272222518921, "learning_rate": 2.9531595281875535e-06, "loss": 3.0016, "step": 292670 }, { "epoch": 0.0040448, "grad_norm": 0.7416769862174988, "learning_rate": 2.9525881888518782e-06, "loss": 3.0835, "step": 292680 }, { "epoch": 0.0040704, "grad_norm": 0.7631211280822754, "learning_rate": 2.952016895216795e-06, "loss": 2.7704, "step": 292690 }, { "epoch": 0.004096, "grad_norm": 0.8308663964271545, "learning_rate": 2.95144564728601e-06, "loss": 2.9792, "step": 292700 }, { "epoch": 0.0041216, "grad_norm": 0.8415123820304871, "learning_rate": 2.9508744450632267e-06, "loss": 2.8904, "step": 292710 }, { "epoch": 0.0041472, "grad_norm": 0.9032887816429138, "learning_rate": 2.9503032885521486e-06, "loss": 3.0056, "step": 292720 }, { "epoch": 0.0041728, "grad_norm": 0.7966935038566589, "learning_rate": 2.9497321777564835e-06, "loss": 2.7394, "step": 292730 }, { "epoch": 0.0041984, "grad_norm": 0.8136230111122131, "learning_rate": 2.949161112679929e-06, "loss": 2.8694, "step": 292740 }, { "epoch": 0.004224, "grad_norm": 0.827422559261322, "learning_rate": 2.948590093326187e-06, "loss": 2.6693, "step": 292750 }, { "epoch": 0.0042496, "grad_norm": 0.9195379018783569, "learning_rate": 2.9480191196989673e-06, "loss": 2.7931, "step": 292760 }, { "epoch": 0.0042752, "grad_norm": 0.8258553743362427, "learning_rate": 2.9474481918019695e-06, "loss": 2.8555, "step": 292770 }, { "epoch": 0.0043008, "grad_norm": 0.7716012001037598, "learning_rate": 2.946877309638899e-06, "loss": 2.7997, "step": 292780 }, { "epoch": 0.0043264, "grad_norm": 0.8095130324363708, "learning_rate": 2.9463064732134516e-06, "loss": 2.9155, "step": 292790 }, { "epoch": 0.004352, "grad_norm": 0.8822397589683533, "learning_rate": 2.9457356825293326e-06, "loss": 2.8293, "step": 292800 }, { "epoch": 0.0043776, "grad_norm": 0.8542432188987732, "learning_rate": 2.9451649375902413e-06, "loss": 2.9224, "step": 292810 }, { "epoch": 0.0044032, "grad_norm": 0.8612437844276428, "learning_rate": 2.9445942383998815e-06, "loss": 2.847, "step": 292820 }, { "epoch": 0.0044288, "grad_norm": 1.1843864917755127, "learning_rate": 2.944023584961954e-06, "loss": 2.6707, "step": 292830 }, { "epoch": 0.0044544, "grad_norm": 0.9190157651901245, "learning_rate": 2.943452977280157e-06, "loss": 2.8464, "step": 292840 }, { "epoch": 0.00448, "grad_norm": 0.9681028127670288, "learning_rate": 2.9428824153581946e-06, "loss": 2.961, "step": 292850 }, { "epoch": 0.0045056, "grad_norm": 0.8763824701309204, "learning_rate": 2.9423118991997624e-06, "loss": 3.1371, "step": 292860 }, { "epoch": 0.0045312, "grad_norm": 0.7744510173797607, "learning_rate": 2.941741428808561e-06, "loss": 2.6943, "step": 292870 }, { "epoch": 0.0045568, "grad_norm": 0.8777211904525757, "learning_rate": 2.9411710041882914e-06, "loss": 2.8883, "step": 292880 }, { "epoch": 0.0045824, "grad_norm": 2.323176622390747, "learning_rate": 2.940600625342651e-06, "loss": 2.7742, "step": 292890 }, { "epoch": 0.004608, "grad_norm": 0.8160460591316223, "learning_rate": 2.9400302922753386e-06, "loss": 2.7538, "step": 292900 }, { "epoch": 0.0046336, "grad_norm": 0.7886711359024048, "learning_rate": 2.9394600049900546e-06, "loss": 2.8721, "step": 292910 }, { "epoch": 0.0046592, "grad_norm": 1.0516070127487183, "learning_rate": 2.9388897634904966e-06, "loss": 3.2227, "step": 292920 }, { "epoch": 0.0046848, "grad_norm": 0.776724636554718, "learning_rate": 2.938319567780361e-06, "loss": 2.8032, "step": 292930 }, { "epoch": 0.0047104, "grad_norm": 1.0194649696350098, "learning_rate": 2.937749417863346e-06, "loss": 2.8129, "step": 292940 }, { "epoch": 0.004736, "grad_norm": 0.8194969296455383, "learning_rate": 2.9371793137431493e-06, "loss": 2.9212, "step": 292950 }, { "epoch": 0.0047616, "grad_norm": 0.9986925721168518, "learning_rate": 2.9366092554234684e-06, "loss": 2.9409, "step": 292960 }, { "epoch": 0.0047872, "grad_norm": 0.87424236536026, "learning_rate": 2.9360392429079988e-06, "loss": 2.8544, "step": 292970 }, { "epoch": 0.0048128, "grad_norm": 0.8776074647903442, "learning_rate": 2.93546927620044e-06, "loss": 3.0245, "step": 292980 }, { "epoch": 0.0048384, "grad_norm": 0.918717622756958, "learning_rate": 2.9348993553044824e-06, "loss": 2.9324, "step": 292990 }, { "epoch": 0.004864, "grad_norm": 0.8762423992156982, "learning_rate": 2.9343294802238253e-06, "loss": 2.8722, "step": 293000 }, { "epoch": 0.0048896, "grad_norm": 0.8717649579048157, "learning_rate": 2.9337596509621635e-06, "loss": 2.8122, "step": 293010 }, { "epoch": 0.0049152, "grad_norm": 0.7533742785453796, "learning_rate": 2.933189867523193e-06, "loss": 2.9038, "step": 293020 }, { "epoch": 0.0049408, "grad_norm": 0.8257538080215454, "learning_rate": 2.932620129910607e-06, "loss": 3.0399, "step": 293030 }, { "epoch": 0.0049664, "grad_norm": 0.914851725101471, "learning_rate": 2.9320504381281013e-06, "loss": 2.8744, "step": 293040 }, { "epoch": 0.004992, "grad_norm": 0.7944598197937012, "learning_rate": 2.931480792179373e-06, "loss": 2.713, "step": 293050 }, { "epoch": 0.0050176, "grad_norm": 0.7507520914077759, "learning_rate": 2.930911192068111e-06, "loss": 2.6261, "step": 293060 }, { "epoch": 0.0050432, "grad_norm": 0.8276375532150269, "learning_rate": 2.9303416377980108e-06, "loss": 2.7531, "step": 293070 }, { "epoch": 0.0050688, "grad_norm": 0.7321251034736633, "learning_rate": 2.9297721293727664e-06, "loss": 2.6402, "step": 293080 }, { "epoch": 0.0050944, "grad_norm": 0.8289088606834412, "learning_rate": 2.9292026667960703e-06, "loss": 2.8121, "step": 293090 }, { "epoch": 0.00512, "grad_norm": 1.0665459632873535, "learning_rate": 2.928633250071612e-06, "loss": 2.8603, "step": 293100 }, { "epoch": 0.0051456, "grad_norm": 0.9174655675888062, "learning_rate": 2.9280638792030947e-06, "loss": 2.5454, "step": 293110 }, { "epoch": 0.0051712, "grad_norm": 0.7772504091262817, "learning_rate": 2.9274945541942012e-06, "loss": 2.7593, "step": 293120 }, { "epoch": 0.0051968, "grad_norm": 0.8645064830780029, "learning_rate": 2.9269252750486245e-06, "loss": 2.8508, "step": 293130 }, { "epoch": 0.0052224, "grad_norm": 1.4489648342132568, "learning_rate": 2.9263560417700586e-06, "loss": 2.7867, "step": 293140 }, { "epoch": 0.005248, "grad_norm": 0.8641650676727295, "learning_rate": 2.925786854362194e-06, "loss": 2.8237, "step": 293150 }, { "epoch": 0.0052736, "grad_norm": 0.8013227581977844, "learning_rate": 2.9252177128287208e-06, "loss": 2.9079, "step": 293160 }, { "epoch": 0.0052992, "grad_norm": 0.763516366481781, "learning_rate": 2.92464861717333e-06, "loss": 2.7819, "step": 293170 }, { "epoch": 0.0053248, "grad_norm": 0.8156355619430542, "learning_rate": 2.9240795673997167e-06, "loss": 2.7552, "step": 293180 }, { "epoch": 0.0053504, "grad_norm": 0.8069645166397095, "learning_rate": 2.9235105635115634e-06, "loss": 2.7743, "step": 293190 }, { "epoch": 0.005376, "grad_norm": 0.7425135374069214, "learning_rate": 2.922941605512564e-06, "loss": 2.8209, "step": 293200 }, { "epoch": 0.0054016, "grad_norm": 0.8242005109786987, "learning_rate": 2.9223726934064057e-06, "loss": 2.9689, "step": 293210 }, { "epoch": 0.0054272, "grad_norm": 0.776582658290863, "learning_rate": 2.9218038271967807e-06, "loss": 2.8686, "step": 293220 }, { "epoch": 0.0054528, "grad_norm": 0.7316480875015259, "learning_rate": 2.921235006887376e-06, "loss": 2.5772, "step": 293230 }, { "epoch": 0.0054784, "grad_norm": 0.861640453338623, "learning_rate": 2.9206662324818823e-06, "loss": 3.0147, "step": 293240 }, { "epoch": 0.005504, "grad_norm": 0.776487410068512, "learning_rate": 2.9200975039839886e-06, "loss": 2.4776, "step": 293250 }, { "epoch": 0.0055296, "grad_norm": 0.8256711959838867, "learning_rate": 2.9195288213973783e-06, "loss": 2.7528, "step": 293260 }, { "epoch": 0.0055552, "grad_norm": 0.9116725921630859, "learning_rate": 2.9189601847257386e-06, "loss": 2.7891, "step": 293270 }, { "epoch": 0.0055808, "grad_norm": 0.8349756002426147, "learning_rate": 2.9183915939727637e-06, "loss": 2.6505, "step": 293280 }, { "epoch": 0.0056064, "grad_norm": 0.8138648271560669, "learning_rate": 2.9178230491421367e-06, "loss": 2.7625, "step": 293290 }, { "epoch": 0.005632, "grad_norm": 0.806391179561615, "learning_rate": 2.917254550237546e-06, "loss": 2.7887, "step": 293300 }, { "epoch": 0.0056576, "grad_norm": 0.9149456024169922, "learning_rate": 2.9166860972626787e-06, "loss": 2.8291, "step": 293310 }, { "epoch": 0.0056832, "grad_norm": 0.7672438025474548, "learning_rate": 2.9161176902212174e-06, "loss": 2.7487, "step": 293320 }, { "epoch": 0.0057088, "grad_norm": 1.2296260595321655, "learning_rate": 2.9155493291168503e-06, "loss": 2.6239, "step": 293330 }, { "epoch": 0.0057344, "grad_norm": 0.7957618832588196, "learning_rate": 2.9149810139532633e-06, "loss": 2.6961, "step": 293340 }, { "epoch": 0.00576, "grad_norm": 0.8140498995780945, "learning_rate": 2.9144127447341396e-06, "loss": 2.8574, "step": 293350 }, { "epoch": 0.0057856, "grad_norm": 0.8034954071044922, "learning_rate": 2.913844521463168e-06, "loss": 2.898, "step": 293360 }, { "epoch": 0.0058112, "grad_norm": 0.7981682419776917, "learning_rate": 2.91327634414403e-06, "loss": 2.722, "step": 293370 }, { "epoch": 0.0058368, "grad_norm": 0.9105181097984314, "learning_rate": 2.912708212780416e-06, "loss": 2.7487, "step": 293380 }, { "epoch": 0.0058624, "grad_norm": 0.7965788841247559, "learning_rate": 2.9121401273760018e-06, "loss": 2.8069, "step": 293390 }, { "epoch": 0.005888, "grad_norm": 0.8300359845161438, "learning_rate": 2.9115720879344754e-06, "loss": 2.6733, "step": 293400 }, { "epoch": 0.0059136, "grad_norm": 0.9296786785125732, "learning_rate": 2.9110040944595196e-06, "loss": 2.8252, "step": 293410 }, { "epoch": 0.0059392, "grad_norm": 0.8727655410766602, "learning_rate": 2.9104361469548193e-06, "loss": 2.7895, "step": 293420 }, { "epoch": 0.0059648, "grad_norm": 0.8769784569740295, "learning_rate": 2.909868245424056e-06, "loss": 2.7933, "step": 293430 }, { "epoch": 0.0059904, "grad_norm": 0.7799661159515381, "learning_rate": 2.9093003898709128e-06, "loss": 2.7873, "step": 293440 }, { "epoch": 0.006016, "grad_norm": 1.4195667505264282, "learning_rate": 2.908732580299072e-06, "loss": 2.8889, "step": 293450 }, { "epoch": 0.0060416, "grad_norm": 0.8336088061332703, "learning_rate": 2.9081648167122143e-06, "loss": 2.7415, "step": 293460 }, { "epoch": 0.0060672, "grad_norm": 0.7605451345443726, "learning_rate": 2.9075970991140247e-06, "loss": 2.953, "step": 293470 }, { "epoch": 0.0060928, "grad_norm": 0.755750834941864, "learning_rate": 2.9070294275081825e-06, "loss": 2.6645, "step": 293480 }, { "epoch": 0.0061184, "grad_norm": 0.834980309009552, "learning_rate": 2.9064618018983694e-06, "loss": 2.8277, "step": 293490 }, { "epoch": 0.006144, "grad_norm": 0.8356707692146301, "learning_rate": 2.905894222288267e-06, "loss": 2.5749, "step": 293500 }, { "epoch": 0.0061696, "grad_norm": 0.7538474202156067, "learning_rate": 2.905326688681557e-06, "loss": 2.6885, "step": 293510 }, { "epoch": 0.0061952, "grad_norm": 1.3011049032211304, "learning_rate": 2.904759201081916e-06, "loss": 2.8129, "step": 293520 }, { "epoch": 0.0062208, "grad_norm": 0.8003817200660706, "learning_rate": 2.9041917594930248e-06, "loss": 2.7097, "step": 293530 }, { "epoch": 0.0062464, "grad_norm": 0.8528649210929871, "learning_rate": 2.9036243639185645e-06, "loss": 2.6949, "step": 293540 }, { "epoch": 0.006272, "grad_norm": 0.8058011531829834, "learning_rate": 2.903057014362215e-06, "loss": 2.8264, "step": 293550 }, { "epoch": 0.0062976, "grad_norm": 0.8039954304695129, "learning_rate": 2.902489710827654e-06, "loss": 2.9235, "step": 293560 }, { "epoch": 0.0063232, "grad_norm": 0.9897778034210205, "learning_rate": 2.9019224533185608e-06, "loss": 2.8728, "step": 293570 }, { "epoch": 0.0063488, "grad_norm": 0.8521115779876709, "learning_rate": 2.901355241838617e-06, "loss": 2.6921, "step": 293580 }, { "epoch": 0.0063744, "grad_norm": 0.8145236968994141, "learning_rate": 2.9007880763914954e-06, "loss": 2.8144, "step": 293590 }, { "epoch": 0.0064, "grad_norm": 0.8348879218101501, "learning_rate": 2.900220956980876e-06, "loss": 2.771, "step": 293600 }, { "epoch": 0.0064256, "grad_norm": 0.8311923146247864, "learning_rate": 2.8996538836104372e-06, "loss": 2.758, "step": 293610 }, { "epoch": 0.0064512, "grad_norm": 0.7551579475402832, "learning_rate": 2.8990868562838527e-06, "loss": 2.7478, "step": 293620 }, { "epoch": 0.0064768, "grad_norm": 1.058200716972351, "learning_rate": 2.898519875004805e-06, "loss": 2.9589, "step": 293630 }, { "epoch": 0.0065024, "grad_norm": 0.9017800092697144, "learning_rate": 2.8979529397769724e-06, "loss": 2.9831, "step": 293640 }, { "epoch": 0.006528, "grad_norm": 0.7698839902877808, "learning_rate": 2.8973860506040243e-06, "loss": 2.721, "step": 293650 }, { "epoch": 0.0065536, "grad_norm": 0.8077221512794495, "learning_rate": 2.8968192074896395e-06, "loss": 2.807, "step": 293660 }, { "epoch": 0.0065792, "grad_norm": 0.8195781707763672, "learning_rate": 2.8962524104374947e-06, "loss": 2.8774, "step": 293670 }, { "epoch": 0.0066048, "grad_norm": 1.012143850326538, "learning_rate": 2.895685659451265e-06, "loss": 2.886, "step": 293680 }, { "epoch": 0.0066304, "grad_norm": 0.8052921295166016, "learning_rate": 2.895118954534626e-06, "loss": 2.7809, "step": 293690 }, { "epoch": 0.006656, "grad_norm": 0.8210878372192383, "learning_rate": 2.894552295691253e-06, "loss": 2.6653, "step": 293700 }, { "epoch": 0.0066816, "grad_norm": 0.8191426396369934, "learning_rate": 2.8939856829248216e-06, "loss": 2.8476, "step": 293710 }, { "epoch": 0.0067072, "grad_norm": 0.7999492287635803, "learning_rate": 2.893419116239002e-06, "loss": 2.7696, "step": 293720 }, { "epoch": 0.0067328, "grad_norm": 0.7913024425506592, "learning_rate": 2.892852595637471e-06, "loss": 2.7706, "step": 293730 }, { "epoch": 0.0067584, "grad_norm": 0.8580616116523743, "learning_rate": 2.8922861211239016e-06, "loss": 3.0248, "step": 293740 }, { "epoch": 0.006784, "grad_norm": 0.8499749898910522, "learning_rate": 2.8917196927019684e-06, "loss": 2.8934, "step": 293750 }, { "epoch": 0.0068096, "grad_norm": 0.7536463141441345, "learning_rate": 2.891153310375343e-06, "loss": 2.8566, "step": 293760 }, { "epoch": 0.0068352, "grad_norm": 0.8535696864128113, "learning_rate": 2.8905869741477e-06, "loss": 2.915, "step": 293770 }, { "epoch": 0.0068608, "grad_norm": 1.0789507627487183, "learning_rate": 2.890020684022713e-06, "loss": 2.7973, "step": 293780 }, { "epoch": 0.0068864, "grad_norm": 0.8587031960487366, "learning_rate": 2.8894544400040504e-06, "loss": 2.8764, "step": 293790 }, { "epoch": 0.006912, "grad_norm": 0.9235358238220215, "learning_rate": 2.888888242095382e-06, "loss": 2.6778, "step": 293800 }, { "epoch": 0.0069376, "grad_norm": 0.8637205958366394, "learning_rate": 2.888322090300386e-06, "loss": 2.8804, "step": 293810 }, { "epoch": 0.0069632, "grad_norm": 0.9706541895866394, "learning_rate": 2.887755984622732e-06, "loss": 2.9899, "step": 293820 }, { "epoch": 0.0069888, "grad_norm": 0.7488729953765869, "learning_rate": 2.8871899250660904e-06, "loss": 2.8803, "step": 293830 }, { "epoch": 0.0070144, "grad_norm": 0.8705014586448669, "learning_rate": 2.8866239116341355e-06, "loss": 2.716, "step": 293840 }, { "epoch": 0.00704, "grad_norm": 0.8158301115036011, "learning_rate": 2.8860579443305293e-06, "loss": 2.8073, "step": 293850 }, { "epoch": 0.0070656, "grad_norm": 0.8294258117675781, "learning_rate": 2.885492023158948e-06, "loss": 2.6981, "step": 293860 }, { "epoch": 0.0070912, "grad_norm": 1.4842585325241089, "learning_rate": 2.8849261481230596e-06, "loss": 2.5602, "step": 293870 }, { "epoch": 0.0071168, "grad_norm": 0.8429816365242004, "learning_rate": 2.8843603192265347e-06, "loss": 2.8671, "step": 293880 }, { "epoch": 0.0071424, "grad_norm": 0.8087233304977417, "learning_rate": 2.883794536473041e-06, "loss": 2.7061, "step": 293890 }, { "epoch": 0.007168, "grad_norm": 0.9039226174354553, "learning_rate": 2.883228799866249e-06, "loss": 2.9543, "step": 293900 }, { "epoch": 0.0071936, "grad_norm": 0.7878862619400024, "learning_rate": 2.88266310940983e-06, "loss": 2.781, "step": 293910 }, { "epoch": 0.0072192, "grad_norm": 0.919579029083252, "learning_rate": 2.882097465107445e-06, "loss": 3.1297, "step": 293920 }, { "epoch": 0.0072448, "grad_norm": 0.8851045966148376, "learning_rate": 2.8815318669627678e-06, "loss": 2.8533, "step": 293930 }, { "epoch": 0.0072704, "grad_norm": 0.8371787667274475, "learning_rate": 2.880966314979464e-06, "loss": 2.7033, "step": 293940 }, { "epoch": 0.007296, "grad_norm": 0.7335663437843323, "learning_rate": 2.8804008091612012e-06, "loss": 3.0198, "step": 293950 }, { "epoch": 0.0073216, "grad_norm": 0.7653440833091736, "learning_rate": 2.8798353495116483e-06, "loss": 2.945, "step": 293960 }, { "epoch": 0.0073472, "grad_norm": 0.8885301351547241, "learning_rate": 2.879269936034469e-06, "loss": 2.9648, "step": 293970 }, { "epoch": 0.0073728, "grad_norm": 0.8583115935325623, "learning_rate": 2.8787045687333324e-06, "loss": 2.9823, "step": 293980 }, { "epoch": 0.0073984, "grad_norm": 0.7912615537643433, "learning_rate": 2.8781392476119053e-06, "loss": 2.9041, "step": 293990 }, { "epoch": 0.007424, "grad_norm": 0.9084290266036987, "learning_rate": 2.8775739726738505e-06, "loss": 2.7736, "step": 294000 }, { "epoch": 0.0074496, "grad_norm": 0.9100368022918701, "learning_rate": 2.8770087439228367e-06, "loss": 2.9745, "step": 294010 }, { "epoch": 0.0074752, "grad_norm": 0.7801244854927063, "learning_rate": 2.8764435613625285e-06, "loss": 2.9115, "step": 294020 }, { "epoch": 0.0075008, "grad_norm": 0.8449789881706238, "learning_rate": 2.8758784249965897e-06, "loss": 2.8882, "step": 294030 }, { "epoch": 0.0075264, "grad_norm": 0.7964618802070618, "learning_rate": 2.875313334828689e-06, "loss": 2.5799, "step": 294040 }, { "epoch": 0.007552, "grad_norm": 0.8829953074455261, "learning_rate": 2.8747482908624856e-06, "loss": 2.7463, "step": 294050 }, { "epoch": 0.0075776, "grad_norm": 0.9315078854560852, "learning_rate": 2.8741832931016454e-06, "loss": 2.796, "step": 294060 }, { "epoch": 0.0076032, "grad_norm": 0.7558081746101379, "learning_rate": 2.8736183415498318e-06, "loss": 2.8116, "step": 294070 }, { "epoch": 0.0076288, "grad_norm": 0.8703988790512085, "learning_rate": 2.873053436210711e-06, "loss": 2.8018, "step": 294080 }, { "epoch": 0.0076544, "grad_norm": 0.8004171252250671, "learning_rate": 2.8724885770879428e-06, "loss": 2.6653, "step": 294090 }, { "epoch": 0.00768, "grad_norm": 0.8527767658233643, "learning_rate": 2.871923764185194e-06, "loss": 2.7512, "step": 294100 }, { "epoch": 0.0077056, "grad_norm": 0.8198904991149902, "learning_rate": 2.8713589975061263e-06, "loss": 2.6963, "step": 294110 }, { "epoch": 0.0077312, "grad_norm": 0.8362119793891907, "learning_rate": 2.8707942770543986e-06, "loss": 2.9137, "step": 294120 }, { "epoch": 0.0077568, "grad_norm": 0.8214374780654907, "learning_rate": 2.8702296028336753e-06, "loss": 2.888, "step": 294130 }, { "epoch": 0.0077824, "grad_norm": 0.8262403607368469, "learning_rate": 2.8696649748476157e-06, "loss": 2.9008, "step": 294140 }, { "epoch": 0.007808, "grad_norm": 0.7881212830543518, "learning_rate": 2.869100393099886e-06, "loss": 2.8032, "step": 294150 }, { "epoch": 0.0078336, "grad_norm": 0.7737229466438293, "learning_rate": 2.868535857594148e-06, "loss": 2.853, "step": 294160 }, { "epoch": 0.0078592, "grad_norm": 0.7696631550788879, "learning_rate": 2.867971368334057e-06, "loss": 2.8364, "step": 294170 }, { "epoch": 0.0078848, "grad_norm": 0.8659260272979736, "learning_rate": 2.867406925323276e-06, "loss": 3.0333, "step": 294180 }, { "epoch": 0.0079104, "grad_norm": 0.7929141521453857, "learning_rate": 2.8668425285654656e-06, "loss": 3.0199, "step": 294190 }, { "epoch": 0.007936, "grad_norm": 0.8444973230361938, "learning_rate": 2.866278178064286e-06, "loss": 2.7712, "step": 294200 }, { "epoch": 0.0079616, "grad_norm": 0.9303492307662964, "learning_rate": 2.865713873823396e-06, "loss": 2.7446, "step": 294210 }, { "epoch": 0.0079872, "grad_norm": 0.7426602244377136, "learning_rate": 2.8651496158464553e-06, "loss": 2.8513, "step": 294220 }, { "epoch": 0.0080128, "grad_norm": 1.5193941593170166, "learning_rate": 2.8645854041371237e-06, "loss": 3.0545, "step": 294230 }, { "epoch": 0.0080384, "grad_norm": 0.8169777393341064, "learning_rate": 2.864021238699062e-06, "loss": 2.8411, "step": 294240 }, { "epoch": 0.008064, "grad_norm": 0.8555150628089905, "learning_rate": 2.863457119535924e-06, "loss": 2.8905, "step": 294250 }, { "epoch": 0.0080896, "grad_norm": 0.8937460780143738, "learning_rate": 2.862893046651368e-06, "loss": 2.9295, "step": 294260 }, { "epoch": 0.0081152, "grad_norm": 0.8103678822517395, "learning_rate": 2.862329020049055e-06, "loss": 2.8913, "step": 294270 }, { "epoch": 0.0081408, "grad_norm": 0.9006955623626709, "learning_rate": 2.8617650397326413e-06, "loss": 2.8242, "step": 294280 }, { "epoch": 0.0081664, "grad_norm": 0.9009668231010437, "learning_rate": 2.861201105705784e-06, "loss": 2.7227, "step": 294290 }, { "epoch": 0.008192, "grad_norm": 0.7876830697059631, "learning_rate": 2.860637217972141e-06, "loss": 2.7928, "step": 294300 }, { "epoch": 0.0082176, "grad_norm": 0.9076073169708252, "learning_rate": 2.8600733765353705e-06, "loss": 2.8485, "step": 294310 }, { "epoch": 0.0082432, "grad_norm": 0.8467920422554016, "learning_rate": 2.8595095813991205e-06, "loss": 2.6374, "step": 294320 }, { "epoch": 0.0082688, "grad_norm": 0.7712057828903198, "learning_rate": 2.858945832567056e-06, "loss": 2.8551, "step": 294330 }, { "epoch": 0.0082944, "grad_norm": 0.8744303584098816, "learning_rate": 2.8583821300428304e-06, "loss": 2.8674, "step": 294340 }, { "epoch": 0.00832, "grad_norm": 0.8993549346923828, "learning_rate": 2.8578184738300986e-06, "loss": 2.8768, "step": 294350 }, { "epoch": 0.0083456, "grad_norm": 0.8314843773841858, "learning_rate": 2.8572548639325193e-06, "loss": 2.7917, "step": 294360 }, { "epoch": 0.0083712, "grad_norm": 0.7867790460586548, "learning_rate": 2.85669130035374e-06, "loss": 2.9748, "step": 294370 }, { "epoch": 0.0083968, "grad_norm": 0.7643500566482544, "learning_rate": 2.85612778309742e-06, "loss": 2.7092, "step": 294380 }, { "epoch": 0.0084224, "grad_norm": 1.3654545545578003, "learning_rate": 2.8555643121672115e-06, "loss": 2.8026, "step": 294390 }, { "epoch": 0.008448, "grad_norm": 0.926379919052124, "learning_rate": 2.855000887566771e-06, "loss": 2.7304, "step": 294400 }, { "epoch": 0.0084736, "grad_norm": 0.9900377988815308, "learning_rate": 2.85443750929975e-06, "loss": 3.1857, "step": 294410 }, { "epoch": 0.0084992, "grad_norm": 0.7863929271697998, "learning_rate": 2.8538741773698033e-06, "loss": 2.8855, "step": 294420 }, { "epoch": 0.0085248, "grad_norm": 0.849812924861908, "learning_rate": 2.853310891780583e-06, "loss": 2.5902, "step": 294430 }, { "epoch": 0.0085504, "grad_norm": 0.8658655881881714, "learning_rate": 2.8527476525357444e-06, "loss": 2.8813, "step": 294440 }, { "epoch": 0.008576, "grad_norm": 0.774568498134613, "learning_rate": 2.852184459638936e-06, "loss": 2.7644, "step": 294450 }, { "epoch": 0.0086016, "grad_norm": 0.8412672281265259, "learning_rate": 2.8516213130938107e-06, "loss": 2.9874, "step": 294460 }, { "epoch": 0.0086272, "grad_norm": 0.7730079889297485, "learning_rate": 2.851058212904022e-06, "loss": 3.0315, "step": 294470 }, { "epoch": 0.0086528, "grad_norm": 0.7399863600730896, "learning_rate": 2.850495159073221e-06, "loss": 2.7623, "step": 294480 }, { "epoch": 0.0086784, "grad_norm": 0.7971875667572021, "learning_rate": 2.8499321516050584e-06, "loss": 2.9237, "step": 294490 }, { "epoch": 0.008704, "grad_norm": 0.8224398493766785, "learning_rate": 2.8493691905031863e-06, "loss": 2.7109, "step": 294500 }, { "epoch": 0.0087296, "grad_norm": 0.8364611268043518, "learning_rate": 2.8488062757712533e-06, "loss": 2.7956, "step": 294510 }, { "epoch": 0.0087552, "grad_norm": 0.8245827555656433, "learning_rate": 2.8482434074129117e-06, "loss": 2.9203, "step": 294520 }, { "epoch": 0.0087808, "grad_norm": 0.7616608142852783, "learning_rate": 2.847680585431811e-06, "loss": 2.8218, "step": 294530 }, { "epoch": 0.0088064, "grad_norm": 0.7781314253807068, "learning_rate": 2.847117809831601e-06, "loss": 2.8396, "step": 294540 }, { "epoch": 0.008832, "grad_norm": 0.7886893153190613, "learning_rate": 2.8465550806159303e-06, "loss": 2.8513, "step": 294550 }, { "epoch": 0.0088576, "grad_norm": 0.7925297021865845, "learning_rate": 2.845992397788452e-06, "loss": 2.8569, "step": 294560 }, { "epoch": 0.0088832, "grad_norm": 0.7929970026016235, "learning_rate": 2.8454297613528093e-06, "loss": 2.9264, "step": 294570 }, { "epoch": 0.0089088, "grad_norm": 0.883571982383728, "learning_rate": 2.844867171312653e-06, "loss": 2.8929, "step": 294580 }, { "epoch": 0.0089344, "grad_norm": 0.9000180959701538, "learning_rate": 2.8443046276716315e-06, "loss": 2.7741, "step": 294590 }, { "epoch": 0.00896, "grad_norm": 0.7653883099555969, "learning_rate": 2.8437421304333913e-06, "loss": 2.802, "step": 294600 }, { "epoch": 0.0089856, "grad_norm": 0.7835252285003662, "learning_rate": 2.843179679601583e-06, "loss": 2.7401, "step": 294610 }, { "epoch": 0.0090112, "grad_norm": 0.7570295929908752, "learning_rate": 2.842617275179852e-06, "loss": 2.724, "step": 294620 }, { "epoch": 0.0090368, "grad_norm": 0.8000978231430054, "learning_rate": 2.842054917171849e-06, "loss": 2.8593, "step": 294630 }, { "epoch": 0.0090624, "grad_norm": 0.8179818391799927, "learning_rate": 2.841492605581214e-06, "loss": 2.7876, "step": 294640 }, { "epoch": 0.009088, "grad_norm": 0.8079856634140015, "learning_rate": 2.840930340411597e-06, "loss": 2.5774, "step": 294650 }, { "epoch": 0.0091136, "grad_norm": 0.7995225787162781, "learning_rate": 2.8403681216666444e-06, "loss": 2.9802, "step": 294660 }, { "epoch": 0.0091392, "grad_norm": 0.768909752368927, "learning_rate": 2.8398059493499984e-06, "loss": 2.8251, "step": 294670 }, { "epoch": 0.0091648, "grad_norm": 0.9370141625404358, "learning_rate": 2.8392438234653117e-06, "loss": 2.7627, "step": 294680 }, { "epoch": 0.0091904, "grad_norm": 0.9795104265213013, "learning_rate": 2.8386817440162286e-06, "loss": 2.798, "step": 294690 }, { "epoch": 0.009216, "grad_norm": 0.8806750774383545, "learning_rate": 2.8381197110063875e-06, "loss": 2.6784, "step": 294700 }, { "epoch": 0.0092416, "grad_norm": 0.9053001403808594, "learning_rate": 2.837557724439437e-06, "loss": 2.9661, "step": 294710 }, { "epoch": 0.0092672, "grad_norm": 0.7628008127212524, "learning_rate": 2.836995784319021e-06, "loss": 3.1503, "step": 294720 }, { "epoch": 0.0092928, "grad_norm": 0.795235276222229, "learning_rate": 2.836433890648783e-06, "loss": 2.7632, "step": 294730 }, { "epoch": 0.0093184, "grad_norm": 0.7231917977333069, "learning_rate": 2.835872043432368e-06, "loss": 2.9212, "step": 294740 }, { "epoch": 0.009344, "grad_norm": 0.8153671026229858, "learning_rate": 2.835310242673419e-06, "loss": 2.8845, "step": 294750 }, { "epoch": 0.0093696, "grad_norm": 0.8257023692131042, "learning_rate": 2.8347484883755815e-06, "loss": 2.7533, "step": 294760 }, { "epoch": 0.0093952, "grad_norm": 10.609437942504883, "learning_rate": 2.8341867805424926e-06, "loss": 2.82, "step": 294770 }, { "epoch": 0.0094208, "grad_norm": 0.873499870300293, "learning_rate": 2.8336251191777986e-06, "loss": 2.7575, "step": 294780 }, { "epoch": 0.0094464, "grad_norm": 0.7651394009590149, "learning_rate": 2.8330635042851407e-06, "loss": 2.8409, "step": 294790 }, { "epoch": 0.009472, "grad_norm": 0.762244701385498, "learning_rate": 2.8325019358681617e-06, "loss": 2.8733, "step": 294800 }, { "epoch": 0.0094976, "grad_norm": 0.7914513349533081, "learning_rate": 2.831940413930503e-06, "loss": 2.8918, "step": 294810 }, { "epoch": 0.0095232, "grad_norm": 0.8988451957702637, "learning_rate": 2.831378938475805e-06, "loss": 2.9155, "step": 294820 }, { "epoch": 0.0095488, "grad_norm": 0.8159618973731995, "learning_rate": 2.830817509507712e-06, "loss": 2.6802, "step": 294830 }, { "epoch": 0.0095744, "grad_norm": 0.882577121257782, "learning_rate": 2.8302561270298567e-06, "loss": 2.922, "step": 294840 }, { "epoch": 0.0096, "grad_norm": 0.7822532057762146, "learning_rate": 2.8296947910458885e-06, "loss": 2.8126, "step": 294850 }, { "epoch": 0.0096256, "grad_norm": 1.0415946245193481, "learning_rate": 2.8291335015594424e-06, "loss": 2.8168, "step": 294860 }, { "epoch": 0.0096512, "grad_norm": 0.7953366637229919, "learning_rate": 2.8285722585741603e-06, "loss": 2.8509, "step": 294870 }, { "epoch": 0.0096768, "grad_norm": 0.7819271683692932, "learning_rate": 2.828011062093682e-06, "loss": 2.8171, "step": 294880 }, { "epoch": 0.0097024, "grad_norm": 0.8679658770561218, "learning_rate": 2.827449912121648e-06, "loss": 2.9449, "step": 294890 }, { "epoch": 0.009728, "grad_norm": 0.8297010064125061, "learning_rate": 2.826888808661693e-06, "loss": 2.9243, "step": 294900 }, { "epoch": 0.0097536, "grad_norm": 0.7843496799468994, "learning_rate": 2.826327751717457e-06, "loss": 2.7623, "step": 294910 }, { "epoch": 0.0097792, "grad_norm": 0.7437459826469421, "learning_rate": 2.8257667412925792e-06, "loss": 2.7041, "step": 294920 }, { "epoch": 0.0098048, "grad_norm": 0.7822574377059937, "learning_rate": 2.825205777390697e-06, "loss": 2.8273, "step": 294930 }, { "epoch": 0.0098304, "grad_norm": 0.884291410446167, "learning_rate": 2.8246448600154496e-06, "loss": 2.7906, "step": 294940 }, { "epoch": 0.009856, "grad_norm": 0.7633417248725891, "learning_rate": 2.8240839891704718e-06, "loss": 2.9169, "step": 294950 }, { "epoch": 0.0098816, "grad_norm": 0.8349471092224121, "learning_rate": 2.8235231648594063e-06, "loss": 2.7184, "step": 294960 }, { "epoch": 0.0099072, "grad_norm": 0.7771671414375305, "learning_rate": 2.8229623870858823e-06, "loss": 2.5964, "step": 294970 }, { "epoch": 0.0099328, "grad_norm": 0.9024823307991028, "learning_rate": 2.8224016558535407e-06, "loss": 2.8235, "step": 294980 }, { "epoch": 0.0099584, "grad_norm": 0.825839638710022, "learning_rate": 2.8218409711660154e-06, "loss": 2.7001, "step": 294990 }, { "epoch": 0.009984, "grad_norm": 0.9155374765396118, "learning_rate": 2.8212803330269447e-06, "loss": 2.8106, "step": 295000 }, { "epoch": 0.0100096, "grad_norm": 0.9440215229988098, "learning_rate": 2.820719741439959e-06, "loss": 2.7179, "step": 295010 }, { "epoch": 0.0100352, "grad_norm": 0.9285081624984741, "learning_rate": 2.8201591964087047e-06, "loss": 3.1181, "step": 295020 }, { "epoch": 0.0100608, "grad_norm": 0.8596892356872559, "learning_rate": 2.8195986979368073e-06, "loss": 2.8035, "step": 295030 }, { "epoch": 0.0100864, "grad_norm": 0.7790061831474304, "learning_rate": 2.819038246027903e-06, "loss": 2.7783, "step": 295040 }, { "epoch": 0.010112, "grad_norm": 0.8645818829536438, "learning_rate": 2.8184778406856273e-06, "loss": 2.8328, "step": 295050 }, { "epoch": 0.0101376, "grad_norm": 0.789193332195282, "learning_rate": 2.8179174819136148e-06, "loss": 2.7919, "step": 295060 }, { "epoch": 0.0101632, "grad_norm": 0.8878381848335266, "learning_rate": 2.817357169715498e-06, "loss": 2.9893, "step": 295070 }, { "epoch": 0.0101888, "grad_norm": 0.8973753452301025, "learning_rate": 2.816796904094912e-06, "loss": 2.6848, "step": 295080 }, { "epoch": 0.0102144, "grad_norm": 0.8863459825515747, "learning_rate": 2.816236685055491e-06, "loss": 2.6489, "step": 295090 }, { "epoch": 0.01024, "grad_norm": 0.7929941415786743, "learning_rate": 2.8156765126008635e-06, "loss": 2.8712, "step": 295100 }, { "epoch": 0.0102656, "grad_norm": 0.7790953516960144, "learning_rate": 2.815116386734664e-06, "loss": 2.6401, "step": 295110 }, { "epoch": 0.0102912, "grad_norm": 0.7979352474212646, "learning_rate": 2.814556307460525e-06, "loss": 2.8378, "step": 295120 }, { "epoch": 0.0103168, "grad_norm": 0.7903211712837219, "learning_rate": 2.8139962747820794e-06, "loss": 2.7717, "step": 295130 }, { "epoch": 0.0103424, "grad_norm": 0.9013096690177917, "learning_rate": 2.8134362887029577e-06, "loss": 3.0797, "step": 295140 }, { "epoch": 0.010368, "grad_norm": 0.7778114080429077, "learning_rate": 2.8128763492267917e-06, "loss": 2.8147, "step": 295150 }, { "epoch": 0.0103936, "grad_norm": 0.8406596779823303, "learning_rate": 2.812316456357216e-06, "loss": 2.8369, "step": 295160 }, { "epoch": 0.0104192, "grad_norm": 1.017952561378479, "learning_rate": 2.811756610097854e-06, "loss": 2.9229, "step": 295170 }, { "epoch": 0.0104448, "grad_norm": 1.0755383968353271, "learning_rate": 2.8111968104523414e-06, "loss": 2.4236, "step": 295180 }, { "epoch": 0.0104704, "grad_norm": 0.8686784505844116, "learning_rate": 2.8106370574243024e-06, "loss": 2.7692, "step": 295190 }, { "epoch": 0.010496, "grad_norm": 0.8929160833358765, "learning_rate": 2.8100773510173753e-06, "loss": 3.084, "step": 295200 }, { "epoch": 0.0105216, "grad_norm": 0.99664705991745, "learning_rate": 2.8095176912351853e-06, "loss": 2.8636, "step": 295210 }, { "epoch": 0.0105472, "grad_norm": 0.8611671924591064, "learning_rate": 2.808958078081365e-06, "loss": 2.8759, "step": 295220 }, { "epoch": 0.0105728, "grad_norm": 0.8032681941986084, "learning_rate": 2.808398511559538e-06, "loss": 2.8048, "step": 295230 }, { "epoch": 0.0105984, "grad_norm": 0.8222493529319763, "learning_rate": 2.807838991673334e-06, "loss": 2.8932, "step": 295240 }, { "epoch": 0.010624, "grad_norm": 0.7876918911933899, "learning_rate": 2.8072795184263848e-06, "loss": 2.8151, "step": 295250 }, { "epoch": 0.0106496, "grad_norm": 0.7522736191749573, "learning_rate": 2.8067200918223147e-06, "loss": 2.8534, "step": 295260 }, { "epoch": 0.0106752, "grad_norm": 0.9590766429901123, "learning_rate": 2.8061607118647527e-06, "loss": 2.8033, "step": 295270 }, { "epoch": 0.0107008, "grad_norm": 0.7939921617507935, "learning_rate": 2.8056013785573278e-06, "loss": 2.7366, "step": 295280 }, { "epoch": 0.0107264, "grad_norm": 0.8009306788444519, "learning_rate": 2.805042091903669e-06, "loss": 2.8527, "step": 295290 }, { "epoch": 0.010752, "grad_norm": 0.8098236322402954, "learning_rate": 2.8044828519073965e-06, "loss": 2.8029, "step": 295300 }, { "epoch": 0.0107776, "grad_norm": 1.0596610307693481, "learning_rate": 2.80392365857214e-06, "loss": 2.612, "step": 295310 }, { "epoch": 0.0108032, "grad_norm": 0.7880839705467224, "learning_rate": 2.8033645119015262e-06, "loss": 3.3068, "step": 295320 }, { "epoch": 0.0108288, "grad_norm": 0.9934529662132263, "learning_rate": 2.802805411899181e-06, "loss": 2.6565, "step": 295330 }, { "epoch": 0.0108544, "grad_norm": 0.85518479347229, "learning_rate": 2.802246358568731e-06, "loss": 2.7279, "step": 295340 }, { "epoch": 0.01088, "grad_norm": 0.7684697508811951, "learning_rate": 2.801687351913799e-06, "loss": 2.9483, "step": 295350 }, { "epoch": 0.0109056, "grad_norm": 0.7921300530433655, "learning_rate": 2.801128391938012e-06, "loss": 2.9519, "step": 295360 }, { "epoch": 0.0109312, "grad_norm": 0.9161556363105774, "learning_rate": 2.8005694786449944e-06, "loss": 2.8917, "step": 295370 }, { "epoch": 0.0109568, "grad_norm": 0.7716845870018005, "learning_rate": 2.8000106120383696e-06, "loss": 2.9747, "step": 295380 }, { "epoch": 0.0109824, "grad_norm": 0.8298519849777222, "learning_rate": 2.7994517921217636e-06, "loss": 2.9619, "step": 295390 }, { "epoch": 0.011008, "grad_norm": 0.841934323310852, "learning_rate": 2.7988930188987984e-06, "loss": 2.8882, "step": 295400 }, { "epoch": 0.0110336, "grad_norm": 0.8354876637458801, "learning_rate": 2.7983342923730984e-06, "loss": 2.9329, "step": 295410 }, { "epoch": 0.0110592, "grad_norm": 0.8206929564476013, "learning_rate": 2.7977756125482893e-06, "loss": 2.9657, "step": 295420 }, { "epoch": 0.0110848, "grad_norm": 0.9031288027763367, "learning_rate": 2.797216979427988e-06, "loss": 3.0791, "step": 295430 }, { "epoch": 0.0111104, "grad_norm": 0.8685933351516724, "learning_rate": 2.7966583930158196e-06, "loss": 2.8727, "step": 295440 }, { "epoch": 0.011136, "grad_norm": 0.9910812973976135, "learning_rate": 2.7960998533154084e-06, "loss": 2.6625, "step": 295450 }, { "epoch": 0.0111616, "grad_norm": 0.8378008604049683, "learning_rate": 2.795541360330375e-06, "loss": 2.8483, "step": 295460 }, { "epoch": 0.0111872, "grad_norm": 0.7898299694061279, "learning_rate": 2.79498291406434e-06, "loss": 2.9172, "step": 295470 }, { "epoch": 0.0112128, "grad_norm": 0.8240737915039062, "learning_rate": 2.7944245145209283e-06, "loss": 2.8596, "step": 295480 }, { "epoch": 0.0112384, "grad_norm": 0.8305984735488892, "learning_rate": 2.79386616170376e-06, "loss": 2.8348, "step": 295490 }, { "epoch": 0.011264, "grad_norm": 1.825636863708496, "learning_rate": 2.7933078556164505e-06, "loss": 2.9649, "step": 295500 }, { "epoch": 0.0112896, "grad_norm": 0.8447986245155334, "learning_rate": 2.7927495962626263e-06, "loss": 2.8405, "step": 295510 }, { "epoch": 0.0113152, "grad_norm": 0.7788644433021545, "learning_rate": 2.792191383645905e-06, "loss": 2.9186, "step": 295520 }, { "epoch": 0.0113408, "grad_norm": 1.0112297534942627, "learning_rate": 2.7916332177699057e-06, "loss": 2.9226, "step": 295530 }, { "epoch": 0.0113664, "grad_norm": 0.7229516506195068, "learning_rate": 2.7910750986382473e-06, "loss": 2.7749, "step": 295540 }, { "epoch": 0.011392, "grad_norm": 0.7251419425010681, "learning_rate": 2.7905170262545566e-06, "loss": 2.7493, "step": 295550 }, { "epoch": 0.0114176, "grad_norm": 0.8175010085105896, "learning_rate": 2.789959000622444e-06, "loss": 2.7465, "step": 295560 }, { "epoch": 0.0114432, "grad_norm": 0.8205006122589111, "learning_rate": 2.7894010217455314e-06, "loss": 2.8082, "step": 295570 }, { "epoch": 0.0114688, "grad_norm": 0.992384135723114, "learning_rate": 2.7888430896274365e-06, "loss": 2.7657, "step": 295580 }, { "epoch": 0.0114944, "grad_norm": 0.842103898525238, "learning_rate": 2.788285204271778e-06, "loss": 2.7911, "step": 295590 }, { "epoch": 0.01152, "grad_norm": 0.8849474191665649, "learning_rate": 2.7877273656821724e-06, "loss": 2.9437, "step": 295600 }, { "epoch": 0.0115456, "grad_norm": 0.9765430092811584, "learning_rate": 2.7871695738622383e-06, "loss": 2.7991, "step": 295610 }, { "epoch": 0.0115712, "grad_norm": 0.838767409324646, "learning_rate": 2.7866118288155964e-06, "loss": 2.6846, "step": 295620 }, { "epoch": 0.0115968, "grad_norm": 0.876113772392273, "learning_rate": 2.7860541305458566e-06, "loss": 2.9852, "step": 295630 }, { "epoch": 0.0116224, "grad_norm": 0.8070418238639832, "learning_rate": 2.7854964790566384e-06, "loss": 2.8493, "step": 295640 }, { "epoch": 0.011648, "grad_norm": 0.7766674757003784, "learning_rate": 2.784938874351558e-06, "loss": 2.8016, "step": 295650 }, { "epoch": 0.0116736, "grad_norm": 0.7751555442810059, "learning_rate": 2.7843813164342314e-06, "loss": 2.9236, "step": 295660 }, { "epoch": 0.0116992, "grad_norm": 0.7841020226478577, "learning_rate": 2.7838238053082745e-06, "loss": 2.8389, "step": 295670 }, { "epoch": 0.0117248, "grad_norm": 0.7722381949424744, "learning_rate": 2.783266340977302e-06, "loss": 2.8823, "step": 295680 }, { "epoch": 0.0117504, "grad_norm": 0.8479636311531067, "learning_rate": 2.7827089234449324e-06, "loss": 2.7208, "step": 295690 }, { "epoch": 0.011776, "grad_norm": 0.8231874704360962, "learning_rate": 2.7821515527147745e-06, "loss": 2.8186, "step": 295700 }, { "epoch": 0.0118016, "grad_norm": 0.824049711227417, "learning_rate": 2.781594228790442e-06, "loss": 2.7315, "step": 295710 }, { "epoch": 0.0118272, "grad_norm": 1.3040728569030762, "learning_rate": 2.7810369516755562e-06, "loss": 2.7288, "step": 295720 }, { "epoch": 0.0118528, "grad_norm": 0.912056028842926, "learning_rate": 2.7804797213737265e-06, "loss": 2.8136, "step": 295730 }, { "epoch": 0.0118784, "grad_norm": 0.8113899827003479, "learning_rate": 2.7799225378885708e-06, "loss": 2.8186, "step": 295740 }, { "epoch": 0.011904, "grad_norm": 0.8205493688583374, "learning_rate": 2.779365401223695e-06, "loss": 2.6354, "step": 295750 }, { "epoch": 0.0119296, "grad_norm": 1.306997537612915, "learning_rate": 2.778808311382716e-06, "loss": 2.8056, "step": 295760 }, { "epoch": 0.0119552, "grad_norm": 0.98251873254776, "learning_rate": 2.7782512683692454e-06, "loss": 2.7809, "step": 295770 }, { "epoch": 0.0119808, "grad_norm": 0.833771288394928, "learning_rate": 2.777694272186897e-06, "loss": 2.7093, "step": 295780 }, { "epoch": 0.0120064, "grad_norm": 0.7899823784828186, "learning_rate": 2.7771373228392806e-06, "loss": 3.0352, "step": 295790 }, { "epoch": 0.012032, "grad_norm": 0.7851185202598572, "learning_rate": 2.7765804203300095e-06, "loss": 2.3294, "step": 295800 }, { "epoch": 0.0120576, "grad_norm": 1.001326084136963, "learning_rate": 2.7760235646626954e-06, "loss": 2.8474, "step": 295810 }, { "epoch": 0.0120832, "grad_norm": 0.8188337087631226, "learning_rate": 2.775466755840951e-06, "loss": 2.7609, "step": 295820 }, { "epoch": 0.0121088, "grad_norm": 0.9389534592628479, "learning_rate": 2.774909993868381e-06, "loss": 2.7374, "step": 295830 }, { "epoch": 0.0121344, "grad_norm": 1.8675682544708252, "learning_rate": 2.7743532787486003e-06, "loss": 2.8213, "step": 295840 }, { "epoch": 0.01216, "grad_norm": 0.8725161552429199, "learning_rate": 2.7737966104852175e-06, "loss": 2.8144, "step": 295850 }, { "epoch": 0.0121856, "grad_norm": 0.8242159485816956, "learning_rate": 2.773239989081844e-06, "loss": 2.9168, "step": 295860 }, { "epoch": 0.0122112, "grad_norm": 0.8783971667289734, "learning_rate": 2.772683414542088e-06, "loss": 2.9092, "step": 295870 }, { "epoch": 0.0122368, "grad_norm": 3.6583948135375977, "learning_rate": 2.7721268868695594e-06, "loss": 2.7472, "step": 295880 }, { "epoch": 0.0122624, "grad_norm": 0.7539001107215881, "learning_rate": 2.771570406067867e-06, "loss": 2.8333, "step": 295890 }, { "epoch": 0.012288, "grad_norm": 0.851829469203949, "learning_rate": 2.7710139721406203e-06, "loss": 2.9661, "step": 295900 }, { "epoch": 0.0123136, "grad_norm": 0.7966562509536743, "learning_rate": 2.770457585091425e-06, "loss": 2.9535, "step": 295910 }, { "epoch": 0.0123392, "grad_norm": 0.8838567733764648, "learning_rate": 2.7699012449238925e-06, "loss": 2.9392, "step": 295920 }, { "epoch": 0.0123648, "grad_norm": 1.2350953817367554, "learning_rate": 2.7693449516416283e-06, "loss": 2.8221, "step": 295930 }, { "epoch": 0.0123904, "grad_norm": 0.8095512986183167, "learning_rate": 2.768788705248243e-06, "loss": 2.7953, "step": 295940 }, { "epoch": 0.012416, "grad_norm": 0.7939698696136475, "learning_rate": 2.768232505747339e-06, "loss": 2.6522, "step": 295950 }, { "epoch": 0.0124416, "grad_norm": 0.8680924773216248, "learning_rate": 2.767676353142524e-06, "loss": 2.9969, "step": 295960 }, { "epoch": 0.0124672, "grad_norm": 1.0098406076431274, "learning_rate": 2.767120247437407e-06, "loss": 2.8148, "step": 295970 }, { "epoch": 0.0124928, "grad_norm": 0.8839541077613831, "learning_rate": 2.766564188635592e-06, "loss": 2.7609, "step": 295980 }, { "epoch": 0.0125184, "grad_norm": 0.7989532351493835, "learning_rate": 2.7660081767406866e-06, "loss": 2.7465, "step": 295990 }, { "epoch": 0.012544, "grad_norm": 0.8632597327232361, "learning_rate": 2.765452211756294e-06, "loss": 2.8591, "step": 296000 }, { "epoch": 0.0125696, "grad_norm": 0.8396223187446594, "learning_rate": 2.7648962936860234e-06, "loss": 2.8669, "step": 296010 }, { "epoch": 0.0125952, "grad_norm": 1.2012051343917847, "learning_rate": 2.764340422533479e-06, "loss": 2.7641, "step": 296020 }, { "epoch": 0.0126208, "grad_norm": 0.7664854526519775, "learning_rate": 2.7637845983022615e-06, "loss": 2.8246, "step": 296030 }, { "epoch": 0.0126464, "grad_norm": 1.5928771495819092, "learning_rate": 2.7632288209959778e-06, "loss": 2.8488, "step": 296040 }, { "epoch": 0.012672, "grad_norm": 0.8059074282646179, "learning_rate": 2.762673090618231e-06, "loss": 2.9319, "step": 296050 }, { "epoch": 0.0126976, "grad_norm": 0.8255332112312317, "learning_rate": 2.762117407172623e-06, "loss": 2.5275, "step": 296060 }, { "epoch": 0.0127232, "grad_norm": 1.0541722774505615, "learning_rate": 2.7615617706627673e-06, "loss": 2.8757, "step": 296070 }, { "epoch": 0.0127488, "grad_norm": 0.8377218842506409, "learning_rate": 2.7610061810922552e-06, "loss": 2.7611, "step": 296080 }, { "epoch": 0.0127744, "grad_norm": 0.8154325485229492, "learning_rate": 2.7604506384646945e-06, "loss": 2.8592, "step": 296090 }, { "epoch": 0.0128, "grad_norm": 0.828843891620636, "learning_rate": 2.7598951427836884e-06, "loss": 2.8986, "step": 296100 }, { "epoch": 0.0128256, "grad_norm": 0.889610767364502, "learning_rate": 2.7593396940528373e-06, "loss": 2.8434, "step": 296110 }, { "epoch": 0.0128512, "grad_norm": 0.8843364715576172, "learning_rate": 2.7587842922757435e-06, "loss": 2.8992, "step": 296120 }, { "epoch": 0.0128768, "grad_norm": 0.7405228614807129, "learning_rate": 2.75822893745601e-06, "loss": 2.7323, "step": 296130 }, { "epoch": 0.0129024, "grad_norm": 0.8090672492980957, "learning_rate": 2.7576736295972394e-06, "loss": 2.811, "step": 296140 }, { "epoch": 0.012928, "grad_norm": 0.8230181932449341, "learning_rate": 2.757118368703028e-06, "loss": 2.8878, "step": 296150 }, { "epoch": 0.0129536, "grad_norm": 0.8548917770385742, "learning_rate": 2.756563154776979e-06, "loss": 2.933, "step": 296160 }, { "epoch": 0.0129792, "grad_norm": 0.9495748281478882, "learning_rate": 2.7560079878226933e-06, "loss": 2.7632, "step": 296170 }, { "epoch": 0.0130048, "grad_norm": 0.8618828654289246, "learning_rate": 2.7554528678437697e-06, "loss": 2.7731, "step": 296180 }, { "epoch": 0.0130304, "grad_norm": 0.7610411643981934, "learning_rate": 2.7548977948438095e-06, "loss": 3.0388, "step": 296190 }, { "epoch": 0.013056, "grad_norm": 0.9669749736785889, "learning_rate": 2.7543427688264113e-06, "loss": 2.8976, "step": 296200 }, { "epoch": 0.0130816, "grad_norm": 0.9533450603485107, "learning_rate": 2.753787789795175e-06, "loss": 2.4293, "step": 296210 }, { "epoch": 0.0131072, "grad_norm": 0.8195176720619202, "learning_rate": 2.7532328577537016e-06, "loss": 2.9908, "step": 296220 }, { "epoch": 0.0131328, "grad_norm": 0.8027108907699585, "learning_rate": 2.7526779727055806e-06, "loss": 2.9984, "step": 296230 }, { "epoch": 0.0131584, "grad_norm": 0.8016328811645508, "learning_rate": 2.7521231346544208e-06, "loss": 2.9483, "step": 296240 }, { "epoch": 0.013184, "grad_norm": 0.7917109131813049, "learning_rate": 2.7515683436038164e-06, "loss": 2.7971, "step": 296250 }, { "epoch": 0.0132096, "grad_norm": 0.8967556953430176, "learning_rate": 2.751013599557364e-06, "loss": 2.8023, "step": 296260 }, { "epoch": 0.0132352, "grad_norm": 0.797008216381073, "learning_rate": 2.7504589025186658e-06, "loss": 2.7573, "step": 296270 }, { "epoch": 0.0132608, "grad_norm": 1.1884948015213013, "learning_rate": 2.749904252491311e-06, "loss": 3.4551, "step": 296280 }, { "epoch": 0.0132864, "grad_norm": 0.764890730381012, "learning_rate": 2.7493496494789007e-06, "loss": 2.7322, "step": 296290 }, { "epoch": 0.013312, "grad_norm": 0.7548713088035583, "learning_rate": 2.7487950934850315e-06, "loss": 2.8183, "step": 296300 }, { "epoch": 0.0133376, "grad_norm": 0.8290936946868896, "learning_rate": 2.7482405845132974e-06, "loss": 2.5578, "step": 296310 }, { "epoch": 0.0133632, "grad_norm": 0.8594967722892761, "learning_rate": 2.747686122567298e-06, "loss": 2.7215, "step": 296320 }, { "epoch": 0.0133888, "grad_norm": 0.806438148021698, "learning_rate": 2.7471317076506253e-06, "loss": 2.7199, "step": 296330 }, { "epoch": 0.0134144, "grad_norm": 1.0720396041870117, "learning_rate": 2.7465773397668784e-06, "loss": 2.9546, "step": 296340 }, { "epoch": 0.01344, "grad_norm": 1.0690213441848755, "learning_rate": 2.746023018919648e-06, "loss": 2.7545, "step": 296350 }, { "epoch": 0.0134656, "grad_norm": 2.4497838020324707, "learning_rate": 2.7454687451125294e-06, "loss": 3.0793, "step": 296360 }, { "epoch": 0.0134912, "grad_norm": 0.7827104926109314, "learning_rate": 2.7449145183491187e-06, "loss": 2.725, "step": 296370 }, { "epoch": 0.0135168, "grad_norm": 0.7930742502212524, "learning_rate": 2.7443603386330087e-06, "loss": 2.8502, "step": 296380 }, { "epoch": 0.0135424, "grad_norm": 0.7444143295288086, "learning_rate": 2.743806205967794e-06, "loss": 2.7322, "step": 296390 }, { "epoch": 0.013568, "grad_norm": 0.7712159156799316, "learning_rate": 2.743252120357067e-06, "loss": 2.6704, "step": 296400 }, { "epoch": 0.0135936, "grad_norm": 0.9283543229103088, "learning_rate": 2.7426980818044223e-06, "loss": 2.7501, "step": 296410 }, { "epoch": 0.0136192, "grad_norm": 0.8740168809890747, "learning_rate": 2.742144090313451e-06, "loss": 2.7996, "step": 296420 }, { "epoch": 0.0136448, "grad_norm": 1.0154205560684204, "learning_rate": 2.741590145887747e-06, "loss": 2.6666, "step": 296430 }, { "epoch": 0.0136704, "grad_norm": 0.8526554107666016, "learning_rate": 2.7410362485309006e-06, "loss": 2.8685, "step": 296440 }, { "epoch": 0.013696, "grad_norm": 1.1519513130187988, "learning_rate": 2.740482398246506e-06, "loss": 3.3028, "step": 296450 }, { "epoch": 0.0137216, "grad_norm": 0.8820399641990662, "learning_rate": 2.7399285950381536e-06, "loss": 2.9174, "step": 296460 }, { "epoch": 0.0137472, "grad_norm": 1.0744538307189941, "learning_rate": 2.7393748389094376e-06, "loss": 2.8196, "step": 296470 }, { "epoch": 0.0137728, "grad_norm": 0.85306715965271, "learning_rate": 2.7388211298639434e-06, "loss": 2.846, "step": 296480 }, { "epoch": 0.0137984, "grad_norm": 0.8455560803413391, "learning_rate": 2.7382674679052634e-06, "loss": 2.8551, "step": 296490 }, { "epoch": 0.013824, "grad_norm": 0.827064573764801, "learning_rate": 2.7377138530369906e-06, "loss": 3.0254, "step": 296500 }, { "epoch": 0.0138496, "grad_norm": 0.7525220513343811, "learning_rate": 2.737160285262712e-06, "loss": 2.9655, "step": 296510 }, { "epoch": 0.0138752, "grad_norm": 0.8450950980186462, "learning_rate": 2.7366067645860195e-06, "loss": 2.7636, "step": 296520 }, { "epoch": 0.0139008, "grad_norm": 0.7744186520576477, "learning_rate": 2.7360532910105007e-06, "loss": 3.0543, "step": 296530 }, { "epoch": 0.0139264, "grad_norm": 0.9224372506141663, "learning_rate": 2.73549986453975e-06, "loss": 2.8817, "step": 296540 }, { "epoch": 0.013952, "grad_norm": 0.8465234041213989, "learning_rate": 2.734946485177349e-06, "loss": 2.7042, "step": 296550 }, { "epoch": 0.0139776, "grad_norm": 0.7343843579292297, "learning_rate": 2.7343931529268884e-06, "loss": 2.717, "step": 296560 }, { "epoch": 0.0140032, "grad_norm": 0.9407958984375, "learning_rate": 2.7338398677919575e-06, "loss": 2.7736, "step": 296570 }, { "epoch": 0.0140288, "grad_norm": 1.0320817232131958, "learning_rate": 2.7332866297761408e-06, "loss": 2.9296, "step": 296580 }, { "epoch": 0.0140544, "grad_norm": 0.8652480244636536, "learning_rate": 2.7327334388830328e-06, "loss": 2.8984, "step": 296590 }, { "epoch": 0.01408, "grad_norm": 0.8065606951713562, "learning_rate": 2.732180295116219e-06, "loss": 2.5525, "step": 296600 }, { "epoch": 0.0141056, "grad_norm": 0.8255330920219421, "learning_rate": 2.7316271984792817e-06, "loss": 3.0003, "step": 296610 }, { "epoch": 0.0141312, "grad_norm": 0.9181301593780518, "learning_rate": 2.731074148975811e-06, "loss": 2.7754, "step": 296620 }, { "epoch": 0.0141568, "grad_norm": 0.8212506175041199, "learning_rate": 2.7305211466093917e-06, "loss": 2.9576, "step": 296630 }, { "epoch": 0.0141824, "grad_norm": 0.8893017172813416, "learning_rate": 2.729968191383612e-06, "loss": 2.7678, "step": 296640 }, { "epoch": 0.014208, "grad_norm": 0.840181291103363, "learning_rate": 2.729415283302055e-06, "loss": 2.9667, "step": 296650 }, { "epoch": 0.0142336, "grad_norm": 0.8102692365646362, "learning_rate": 2.728862422368308e-06, "loss": 2.6531, "step": 296660 }, { "epoch": 0.0142592, "grad_norm": 0.808357834815979, "learning_rate": 2.7283096085859595e-06, "loss": 2.931, "step": 296670 }, { "epoch": 0.0142848, "grad_norm": 0.7925757765769958, "learning_rate": 2.7277568419585876e-06, "loss": 2.8531, "step": 296680 }, { "epoch": 0.0143104, "grad_norm": 0.8319128155708313, "learning_rate": 2.7272041224897793e-06, "loss": 2.9429, "step": 296690 }, { "epoch": 0.014336, "grad_norm": 0.7616786360740662, "learning_rate": 2.7266514501831187e-06, "loss": 3.0034, "step": 296700 }, { "epoch": 0.0143616, "grad_norm": 0.7497715353965759, "learning_rate": 2.7260988250421915e-06, "loss": 2.8184, "step": 296710 }, { "epoch": 0.0143872, "grad_norm": 0.948677122592926, "learning_rate": 2.7255462470705796e-06, "loss": 3.1682, "step": 296720 }, { "epoch": 0.0144128, "grad_norm": 0.8173273801803589, "learning_rate": 2.7249937162718677e-06, "loss": 2.9122, "step": 296730 }, { "epoch": 0.0144384, "grad_norm": 0.854297399520874, "learning_rate": 2.7244412326496406e-06, "loss": 2.6596, "step": 296740 }, { "epoch": 0.014464, "grad_norm": 0.8372269868850708, "learning_rate": 2.723888796207472e-06, "loss": 2.9595, "step": 296750 }, { "epoch": 0.0144896, "grad_norm": 0.8131276965141296, "learning_rate": 2.7233364069489555e-06, "loss": 2.7374, "step": 296760 }, { "epoch": 0.0145152, "grad_norm": 0.7903869152069092, "learning_rate": 2.722784064877667e-06, "loss": 2.8625, "step": 296770 }, { "epoch": 0.0145408, "grad_norm": 0.7742188572883606, "learning_rate": 2.72223176999719e-06, "loss": 2.6363, "step": 296780 }, { "epoch": 0.0145664, "grad_norm": 1.1308214664459229, "learning_rate": 2.721679522311106e-06, "loss": 3.0061, "step": 296790 }, { "epoch": 0.014592, "grad_norm": 1.062548041343689, "learning_rate": 2.7211273218229984e-06, "loss": 2.6905, "step": 296800 }, { "epoch": 0.0146176, "grad_norm": 0.9474539160728455, "learning_rate": 2.7205751685364433e-06, "loss": 2.7859, "step": 296810 }, { "epoch": 0.0146432, "grad_norm": 0.7325178980827332, "learning_rate": 2.7200230624550238e-06, "loss": 2.8455, "step": 296820 }, { "epoch": 0.0146688, "grad_norm": 0.92795330286026, "learning_rate": 2.719471003582319e-06, "loss": 2.8519, "step": 296830 }, { "epoch": 0.0146944, "grad_norm": 0.831275999546051, "learning_rate": 2.7189189919219105e-06, "loss": 2.6933, "step": 296840 }, { "epoch": 0.01472, "grad_norm": 0.8150894045829773, "learning_rate": 2.7183670274773776e-06, "loss": 3.0126, "step": 296850 }, { "epoch": 0.0147456, "grad_norm": 0.8274803757667542, "learning_rate": 2.7178151102522985e-06, "loss": 2.7499, "step": 296860 }, { "epoch": 0.0147712, "grad_norm": 0.9308618903160095, "learning_rate": 2.717263240250256e-06, "loss": 2.569, "step": 296870 }, { "epoch": 0.0147968, "grad_norm": 0.8847368359565735, "learning_rate": 2.7167114174748234e-06, "loss": 2.8788, "step": 296880 }, { "epoch": 0.0148224, "grad_norm": 0.9438833594322205, "learning_rate": 2.716159641929582e-06, "loss": 2.7407, "step": 296890 }, { "epoch": 0.014848, "grad_norm": 0.7911865711212158, "learning_rate": 2.715607913618109e-06, "loss": 2.9036, "step": 296900 }, { "epoch": 2.56e-05, "grad_norm": 0.7946609854698181, "learning_rate": 2.715056232543982e-06, "loss": 1.9429, "step": 296910 }, { "epoch": 5.12e-05, "grad_norm": 0.8229779005050659, "learning_rate": 2.71450459871078e-06, "loss": 2.8248, "step": 296920 }, { "epoch": 7.68e-05, "grad_norm": 1.5670907497406006, "learning_rate": 2.713953012122079e-06, "loss": 3.2221, "step": 296930 }, { "epoch": 0.0001024, "grad_norm": 0.7497262358665466, "learning_rate": 2.7134014727814573e-06, "loss": 2.638, "step": 296940 }, { "epoch": 0.000128, "grad_norm": 0.7883610725402832, "learning_rate": 2.712849980692489e-06, "loss": 2.8055, "step": 296950 }, { "epoch": 0.0001536, "grad_norm": 0.749722957611084, "learning_rate": 2.7122985358587526e-06, "loss": 2.6221, "step": 296960 }, { "epoch": 0.0001792, "grad_norm": 0.7284587025642395, "learning_rate": 2.7117471382838246e-06, "loss": 2.5641, "step": 296970 }, { "epoch": 0.0002048, "grad_norm": 0.9044354557991028, "learning_rate": 2.711195787971278e-06, "loss": 2.4847, "step": 296980 }, { "epoch": 0.0002304, "grad_norm": 0.8946922421455383, "learning_rate": 2.7106444849246893e-06, "loss": 2.629, "step": 296990 }, { "epoch": 0.000256, "grad_norm": 0.8112311959266663, "learning_rate": 2.710093229147638e-06, "loss": 2.5841, "step": 297000 }, { "epoch": 0.0002816, "grad_norm": 0.846741259098053, "learning_rate": 2.709542020643691e-06, "loss": 2.6403, "step": 297010 }, { "epoch": 0.0003072, "grad_norm": 0.8067474365234375, "learning_rate": 2.708990859416426e-06, "loss": 2.5266, "step": 297020 }, { "epoch": 0.0003328, "grad_norm": 1.0359939336776733, "learning_rate": 2.708439745469418e-06, "loss": 2.3179, "step": 297030 }, { "epoch": 0.0003584, "grad_norm": 0.8951148986816406, "learning_rate": 2.707888678806241e-06, "loss": 2.962, "step": 297040 }, { "epoch": 0.000384, "grad_norm": 0.9703299403190613, "learning_rate": 2.7073376594304666e-06, "loss": 2.6423, "step": 297050 }, { "epoch": 0.0004096, "grad_norm": 0.8720802068710327, "learning_rate": 2.7067866873456705e-06, "loss": 2.8637, "step": 297060 }, { "epoch": 0.0004352, "grad_norm": 0.8221355080604553, "learning_rate": 2.706235762555427e-06, "loss": 2.472, "step": 297070 }, { "epoch": 0.0004608, "grad_norm": 0.8447878956794739, "learning_rate": 2.7056848850633033e-06, "loss": 2.7228, "step": 297080 }, { "epoch": 0.0004864, "grad_norm": 0.8716173768043518, "learning_rate": 2.7051340548728745e-06, "loss": 2.6228, "step": 297090 }, { "epoch": 0.000512, "grad_norm": 0.8170892596244812, "learning_rate": 2.7045832719877098e-06, "loss": 2.6379, "step": 297100 }, { "epoch": 0.0005376, "grad_norm": 0.8283686637878418, "learning_rate": 2.7040325364113863e-06, "loss": 2.573, "step": 297110 }, { "epoch": 0.0005632, "grad_norm": 0.8006325364112854, "learning_rate": 2.703481848147472e-06, "loss": 2.3073, "step": 297120 }, { "epoch": 0.0005888, "grad_norm": 0.7947779893875122, "learning_rate": 2.702931207199543e-06, "loss": 2.6215, "step": 297130 }, { "epoch": 0.0006144, "grad_norm": 0.7747880816459656, "learning_rate": 2.702380613571163e-06, "loss": 2.4384, "step": 297140 }, { "epoch": 0.00064, "grad_norm": 0.8768162727355957, "learning_rate": 2.7018300672659036e-06, "loss": 2.5746, "step": 297150 }, { "epoch": 0.0006656, "grad_norm": 0.9267411828041077, "learning_rate": 2.7012795682873384e-06, "loss": 2.6518, "step": 297160 }, { "epoch": 0.0006912, "grad_norm": 0.8194406628608704, "learning_rate": 2.700729116639034e-06, "loss": 2.6064, "step": 297170 }, { "epoch": 0.0007168, "grad_norm": 1.3788727521896362, "learning_rate": 2.7001787123245617e-06, "loss": 3.232, "step": 297180 }, { "epoch": 0.0007424, "grad_norm": 0.8147701025009155, "learning_rate": 2.699628355347491e-06, "loss": 2.4447, "step": 297190 }, { "epoch": 0.000768, "grad_norm": 0.8465210199356079, "learning_rate": 2.699078045711393e-06, "loss": 2.7653, "step": 297200 }, { "epoch": 0.0007936, "grad_norm": 0.856594443321228, "learning_rate": 2.69852778341983e-06, "loss": 2.7038, "step": 297210 }, { "epoch": 0.0008192, "grad_norm": 0.7490590810775757, "learning_rate": 2.697977568476374e-06, "loss": 2.6156, "step": 297220 }, { "epoch": 0.0008448, "grad_norm": 0.8469620943069458, "learning_rate": 2.6974274008845934e-06, "loss": 2.7332, "step": 297230 }, { "epoch": 0.0008704, "grad_norm": 0.7802176475524902, "learning_rate": 2.696877280648055e-06, "loss": 2.495, "step": 297240 }, { "epoch": 0.000896, "grad_norm": 0.8163298964500427, "learning_rate": 2.696327207770326e-06, "loss": 2.5719, "step": 297250 }, { "epoch": 0.0009216, "grad_norm": 0.6766195893287659, "learning_rate": 2.695777182254975e-06, "loss": 2.5, "step": 297260 }, { "epoch": 0.0009472, "grad_norm": 0.7969729900360107, "learning_rate": 2.6952272041055704e-06, "loss": 2.544, "step": 297270 }, { "epoch": 0.0009728, "grad_norm": 0.8516013026237488, "learning_rate": 2.69467727332567e-06, "loss": 2.4609, "step": 297280 }, { "epoch": 0.0009984, "grad_norm": 0.8517272472381592, "learning_rate": 2.694127389918848e-06, "loss": 2.6938, "step": 297290 }, { "epoch": 0.001024, "grad_norm": 1.161582589149475, "learning_rate": 2.693577553888669e-06, "loss": 2.5278, "step": 297300 }, { "epoch": 0.0010496, "grad_norm": 0.8390074372291565, "learning_rate": 2.6930277652386982e-06, "loss": 2.7846, "step": 297310 }, { "epoch": 0.0010752, "grad_norm": 0.780711829662323, "learning_rate": 2.6924780239725e-06, "loss": 2.4962, "step": 297320 }, { "epoch": 0.0011008, "grad_norm": 0.7707986831665039, "learning_rate": 2.6919283300936425e-06, "loss": 2.7131, "step": 297330 }, { "epoch": 0.0011264, "grad_norm": 0.8037949800491333, "learning_rate": 2.6913786836056845e-06, "loss": 2.6179, "step": 297340 }, { "epoch": 0.001152, "grad_norm": 0.8501172661781311, "learning_rate": 2.690829084512193e-06, "loss": 2.7673, "step": 297350 }, { "epoch": 0.0011776, "grad_norm": 0.7939934134483337, "learning_rate": 2.690279532816733e-06, "loss": 2.86, "step": 297360 }, { "epoch": 0.0012032, "grad_norm": 1.1490811109542847, "learning_rate": 2.689730028522868e-06, "loss": 2.7334, "step": 297370 }, { "epoch": 0.0012288, "grad_norm": 0.8216522336006165, "learning_rate": 2.68918057163416e-06, "loss": 2.6335, "step": 297380 }, { "epoch": 0.0012544, "grad_norm": 0.9342618584632874, "learning_rate": 2.688631162154173e-06, "loss": 2.8119, "step": 297390 }, { "epoch": 0.00128, "grad_norm": 0.8797379732131958, "learning_rate": 2.688081800086473e-06, "loss": 2.7093, "step": 297400 }, { "epoch": 0.0013056, "grad_norm": 0.8691281676292419, "learning_rate": 2.6875324854346165e-06, "loss": 2.5443, "step": 297410 }, { "epoch": 0.0013312, "grad_norm": 0.7558825016021729, "learning_rate": 2.6869832182021683e-06, "loss": 2.6922, "step": 297420 }, { "epoch": 0.0013568, "grad_norm": 0.8480957746505737, "learning_rate": 2.68643399839269e-06, "loss": 2.7718, "step": 297430 }, { "epoch": 0.0013824, "grad_norm": 0.840645968914032, "learning_rate": 2.6858848260097436e-06, "loss": 2.664, "step": 297440 }, { "epoch": 0.001408, "grad_norm": 0.8339208960533142, "learning_rate": 2.6853357010568903e-06, "loss": 2.7261, "step": 297450 }, { "epoch": 0.0014336, "grad_norm": 0.8204389810562134, "learning_rate": 2.684786623537692e-06, "loss": 2.7919, "step": 297460 }, { "epoch": 0.0014592, "grad_norm": 0.7929918766021729, "learning_rate": 2.6842375934557075e-06, "loss": 2.5757, "step": 297470 }, { "epoch": 0.0014848, "grad_norm": 0.8604438304901123, "learning_rate": 2.6836886108144978e-06, "loss": 2.2583, "step": 297480 }, { "epoch": 0.0015104, "grad_norm": 0.9456060528755188, "learning_rate": 2.683139675617623e-06, "loss": 2.6937, "step": 297490 }, { "epoch": 0.001536, "grad_norm": 0.7985236644744873, "learning_rate": 2.682590787868643e-06, "loss": 2.6297, "step": 297500 }, { "epoch": 2.56e-05, "grad_norm": 0.8218485116958618, "learning_rate": 2.6820419475711168e-06, "loss": 1.9702, "step": 297510 }, { "epoch": 5.12e-05, "grad_norm": 0.8912681937217712, "learning_rate": 2.6814931547286038e-06, "loss": 2.6185, "step": 297520 }, { "epoch": 7.68e-05, "grad_norm": 1.0331146717071533, "learning_rate": 2.680944409344666e-06, "loss": 2.959, "step": 297530 }, { "epoch": 0.0001024, "grad_norm": 0.8674059510231018, "learning_rate": 2.680395711422854e-06, "loss": 2.6177, "step": 297540 }, { "epoch": 0.000128, "grad_norm": 0.7885103821754456, "learning_rate": 2.6798470609667316e-06, "loss": 2.6753, "step": 297550 }, { "epoch": 0.0001536, "grad_norm": 0.7778246998786926, "learning_rate": 2.6792984579798563e-06, "loss": 2.6867, "step": 297560 }, { "epoch": 0.0001792, "grad_norm": 0.709479570388794, "learning_rate": 2.6787499024657837e-06, "loss": 2.2123, "step": 297570 }, { "epoch": 0.0002048, "grad_norm": 0.768305242061615, "learning_rate": 2.678201394428073e-06, "loss": 2.5693, "step": 297580 }, { "epoch": 0.0002304, "grad_norm": 0.913050651550293, "learning_rate": 2.6776529338702794e-06, "loss": 2.878, "step": 297590 }, { "epoch": 0.000256, "grad_norm": 0.9516846537590027, "learning_rate": 2.6771045207959645e-06, "loss": 2.7593, "step": 297600 }, { "epoch": 0.0002816, "grad_norm": 0.8470028042793274, "learning_rate": 2.6765561552086774e-06, "loss": 2.5907, "step": 297610 }, { "epoch": 0.0003072, "grad_norm": 0.8978214263916016, "learning_rate": 2.6760078371119733e-06, "loss": 2.4098, "step": 297620 }, { "epoch": 0.0003328, "grad_norm": 0.8487337231636047, "learning_rate": 2.6754595665094163e-06, "loss": 2.9534, "step": 297630 }, { "epoch": 0.0003584, "grad_norm": 0.9116899371147156, "learning_rate": 2.674911343404557e-06, "loss": 2.8143, "step": 297640 }, { "epoch": 0.000384, "grad_norm": 1.0545517206192017, "learning_rate": 2.674363167800954e-06, "loss": 2.5958, "step": 297650 }, { "epoch": 0.0004096, "grad_norm": 0.8251143097877502, "learning_rate": 2.6738150397021557e-06, "loss": 2.6501, "step": 297660 }, { "epoch": 0.0004352, "grad_norm": 0.9613494277000427, "learning_rate": 2.673266959111719e-06, "loss": 2.7296, "step": 297670 }, { "epoch": 0.0004608, "grad_norm": 0.8686197400093079, "learning_rate": 2.6727189260332e-06, "loss": 2.5619, "step": 297680 }, { "epoch": 0.0004864, "grad_norm": 0.8144328594207764, "learning_rate": 2.672170940470151e-06, "loss": 2.8268, "step": 297690 }, { "epoch": 0.000512, "grad_norm": 0.803255558013916, "learning_rate": 2.671623002426126e-06, "loss": 2.4765, "step": 297700 }, { "epoch": 0.0005376, "grad_norm": 0.8463457226753235, "learning_rate": 2.6710751119046775e-06, "loss": 2.784, "step": 297710 }, { "epoch": 0.0005632, "grad_norm": 0.8348860740661621, "learning_rate": 2.670527268909363e-06, "loss": 2.2248, "step": 297720 }, { "epoch": 0.0005888, "grad_norm": 0.9818695783615112, "learning_rate": 2.6699794734437278e-06, "loss": 2.841, "step": 297730 }, { "epoch": 0.0006144, "grad_norm": 0.8149007558822632, "learning_rate": 2.669431725511328e-06, "loss": 2.5221, "step": 297740 }, { "epoch": 0.00064, "grad_norm": 1.0349764823913574, "learning_rate": 2.6688840251157144e-06, "loss": 2.5251, "step": 297750 }, { "epoch": 0.0006656, "grad_norm": 0.9547232985496521, "learning_rate": 2.6683363722604403e-06, "loss": 3.1118, "step": 297760 }, { "epoch": 0.0006912, "grad_norm": 0.8740898966789246, "learning_rate": 2.6677887669490566e-06, "loss": 2.8263, "step": 297770 }, { "epoch": 0.0007168, "grad_norm": 0.9407128095626831, "learning_rate": 2.667241209185113e-06, "loss": 2.5419, "step": 297780 }, { "epoch": 0.0007424, "grad_norm": 0.8344884514808655, "learning_rate": 2.6666936989721626e-06, "loss": 2.7117, "step": 297790 }, { "epoch": 0.000768, "grad_norm": 0.8069198131561279, "learning_rate": 2.6661462363137535e-06, "loss": 2.5188, "step": 297800 }, { "epoch": 0.0007936, "grad_norm": 0.85873943567276, "learning_rate": 2.6655988212134377e-06, "loss": 2.7088, "step": 297810 }, { "epoch": 0.0008192, "grad_norm": 0.8587969541549683, "learning_rate": 2.665051453674764e-06, "loss": 2.5833, "step": 297820 }, { "epoch": 0.0008448, "grad_norm": 0.8577806353569031, "learning_rate": 2.6645041337012824e-06, "loss": 2.6155, "step": 297830 }, { "epoch": 0.0008704, "grad_norm": 0.785578191280365, "learning_rate": 2.6639568612965415e-06, "loss": 2.5735, "step": 297840 }, { "epoch": 0.000896, "grad_norm": 0.896977424621582, "learning_rate": 2.6634096364640936e-06, "loss": 2.4697, "step": 297850 }, { "epoch": 0.0009216, "grad_norm": 0.7500602602958679, "learning_rate": 2.6628624592074814e-06, "loss": 2.4383, "step": 297860 }, { "epoch": 0.0009472, "grad_norm": 0.8720247745513916, "learning_rate": 2.6623153295302574e-06, "loss": 2.8923, "step": 297870 }, { "epoch": 0.0009728, "grad_norm": 0.8414741158485413, "learning_rate": 2.6617682474359664e-06, "loss": 2.394, "step": 297880 }, { "epoch": 0.0009984, "grad_norm": 0.9668756723403931, "learning_rate": 2.66122121292816e-06, "loss": 2.3708, "step": 297890 }, { "epoch": 0.001024, "grad_norm": 0.8544779419898987, "learning_rate": 2.660674226010382e-06, "loss": 2.5408, "step": 297900 }, { "epoch": 0.0010496, "grad_norm": 0.8840266466140747, "learning_rate": 2.6601272866861816e-06, "loss": 2.667, "step": 297910 }, { "epoch": 0.0010752, "grad_norm": 0.8359503149986267, "learning_rate": 2.6595803949591083e-06, "loss": 2.7887, "step": 297920 }, { "epoch": 0.0011008, "grad_norm": 0.9203823208808899, "learning_rate": 2.6590335508327027e-06, "loss": 2.8569, "step": 297930 }, { "epoch": 0.0011264, "grad_norm": 0.8916299343109131, "learning_rate": 2.658486754310513e-06, "loss": 2.6215, "step": 297940 }, { "epoch": 0.001152, "grad_norm": 0.89598149061203, "learning_rate": 2.657940005396086e-06, "loss": 2.692, "step": 297950 }, { "epoch": 0.0011776, "grad_norm": 1.0147409439086914, "learning_rate": 2.6573933040929657e-06, "loss": 2.8463, "step": 297960 }, { "epoch": 0.0012032, "grad_norm": 0.912720263004303, "learning_rate": 2.6568466504046964e-06, "loss": 2.8259, "step": 297970 }, { "epoch": 0.0012288, "grad_norm": 0.7839108109474182, "learning_rate": 2.656300044334831e-06, "loss": 2.7796, "step": 297980 }, { "epoch": 0.0012544, "grad_norm": 0.8770528435707092, "learning_rate": 2.655753485886904e-06, "loss": 2.8325, "step": 297990 }, { "epoch": 0.00128, "grad_norm": 0.8596173524856567, "learning_rate": 2.6552069750644647e-06, "loss": 2.7767, "step": 298000 }, { "epoch": 0.0013056, "grad_norm": 0.810030996799469, "learning_rate": 2.6546605118710554e-06, "loss": 2.8349, "step": 298010 }, { "epoch": 0.0013312, "grad_norm": 0.8849117159843445, "learning_rate": 2.654114096310222e-06, "loss": 2.5056, "step": 298020 }, { "epoch": 0.0013568, "grad_norm": 1.3131152391433716, "learning_rate": 2.6535677283855056e-06, "loss": 2.5497, "step": 298030 }, { "epoch": 0.0013824, "grad_norm": 0.9407961368560791, "learning_rate": 2.65302140810045e-06, "loss": 2.8979, "step": 298040 }, { "epoch": 0.001408, "grad_norm": 0.7541195154190063, "learning_rate": 2.6524751354586007e-06, "loss": 2.5807, "step": 298050 }, { "epoch": 0.0014336, "grad_norm": 0.8973755836486816, "learning_rate": 2.6519289104634948e-06, "loss": 2.9752, "step": 298060 }, { "epoch": 0.0014592, "grad_norm": 0.9236288070678711, "learning_rate": 2.651382733118677e-06, "loss": 2.6411, "step": 298070 }, { "epoch": 0.0014848, "grad_norm": 0.8659576773643494, "learning_rate": 2.650836603427689e-06, "loss": 2.5329, "step": 298080 }, { "epoch": 0.0015104, "grad_norm": 0.8301628828048706, "learning_rate": 2.650290521394072e-06, "loss": 2.1736, "step": 298090 }, { "epoch": 0.001536, "grad_norm": 1.0444831848144531, "learning_rate": 2.6497444870213685e-06, "loss": 2.6652, "step": 298100 }, { "epoch": 0.0015616, "grad_norm": 0.8031703233718872, "learning_rate": 2.649198500313118e-06, "loss": 2.4377, "step": 298110 }, { "epoch": 0.0015872, "grad_norm": 0.9431771636009216, "learning_rate": 2.6486525612728644e-06, "loss": 2.9135, "step": 298120 }, { "epoch": 0.0016128, "grad_norm": 0.8743370771408081, "learning_rate": 2.6481066699041425e-06, "loss": 2.7171, "step": 298130 }, { "epoch": 0.0016384, "grad_norm": 0.8279697299003601, "learning_rate": 2.6475608262104947e-06, "loss": 2.7004, "step": 298140 }, { "epoch": 0.001664, "grad_norm": 0.8525456786155701, "learning_rate": 2.647015030195459e-06, "loss": 2.6022, "step": 298150 }, { "epoch": 0.0016896, "grad_norm": 0.9177297353744507, "learning_rate": 2.646469281862578e-06, "loss": 2.7088, "step": 298160 }, { "epoch": 0.0017152, "grad_norm": 0.7611833810806274, "learning_rate": 2.6459235812153906e-06, "loss": 2.636, "step": 298170 }, { "epoch": 0.0017408, "grad_norm": 0.9148533344268799, "learning_rate": 2.6453779282574367e-06, "loss": 2.6351, "step": 298180 }, { "epoch": 0.0017664, "grad_norm": 0.8788484334945679, "learning_rate": 2.64483232299225e-06, "loss": 2.3348, "step": 298190 }, { "epoch": 0.001792, "grad_norm": 0.8037952780723572, "learning_rate": 2.6442867654233708e-06, "loss": 2.3178, "step": 298200 }, { "epoch": 0.0018176, "grad_norm": 0.9575822353363037, "learning_rate": 2.643741255554336e-06, "loss": 2.5806, "step": 298210 }, { "epoch": 0.0018432, "grad_norm": 0.7869728207588196, "learning_rate": 2.643195793388684e-06, "loss": 2.4601, "step": 298220 }, { "epoch": 0.0018688, "grad_norm": 1.0120986700057983, "learning_rate": 2.6426503789299527e-06, "loss": 2.8478, "step": 298230 }, { "epoch": 0.0018944, "grad_norm": 0.8519919514656067, "learning_rate": 2.6421050121816793e-06, "loss": 2.6872, "step": 298240 }, { "epoch": 0.00192, "grad_norm": 0.8360257148742676, "learning_rate": 2.641559693147402e-06, "loss": 2.9927, "step": 298250 }, { "epoch": 0.0019456, "grad_norm": 0.8644871711730957, "learning_rate": 2.64101442183065e-06, "loss": 2.8659, "step": 298260 }, { "epoch": 0.0019712, "grad_norm": 0.9133459329605103, "learning_rate": 2.640469198234965e-06, "loss": 2.7821, "step": 298270 }, { "epoch": 0.0019968, "grad_norm": 1.1055378913879395, "learning_rate": 2.639924022363881e-06, "loss": 2.656, "step": 298280 }, { "epoch": 0.0020224, "grad_norm": 0.7930586934089661, "learning_rate": 2.639378894220934e-06, "loss": 2.8635, "step": 298290 }, { "epoch": 0.002048, "grad_norm": 0.7537441849708557, "learning_rate": 2.6388338138096582e-06, "loss": 2.4835, "step": 298300 }, { "epoch": 0.0020736, "grad_norm": 0.8688458800315857, "learning_rate": 2.638288781133589e-06, "loss": 2.6619, "step": 298310 }, { "epoch": 0.0020992, "grad_norm": 0.7916542291641235, "learning_rate": 2.6377437961962603e-06, "loss": 2.6449, "step": 298320 }, { "epoch": 0.0021248, "grad_norm": 0.9183592796325684, "learning_rate": 2.6371988590012067e-06, "loss": 2.8703, "step": 298330 }, { "epoch": 0.0021504, "grad_norm": 0.9644977450370789, "learning_rate": 2.6366539695519613e-06, "loss": 2.8447, "step": 298340 }, { "epoch": 0.002176, "grad_norm": 0.8052151799201965, "learning_rate": 2.6361091278520578e-06, "loss": 2.9223, "step": 298350 }, { "epoch": 0.0022016, "grad_norm": 0.8507406711578369, "learning_rate": 2.635564333905031e-06, "loss": 2.8874, "step": 298360 }, { "epoch": 0.0022272, "grad_norm": 0.782537579536438, "learning_rate": 2.6350195877144102e-06, "loss": 2.5979, "step": 298370 }, { "epoch": 0.0022528, "grad_norm": 0.8018877506256104, "learning_rate": 2.6344748892837347e-06, "loss": 2.7271, "step": 298380 }, { "epoch": 0.0022784, "grad_norm": 0.9175357818603516, "learning_rate": 2.633930238616528e-06, "loss": 2.8398, "step": 298390 }, { "epoch": 0.002304, "grad_norm": 0.8253533840179443, "learning_rate": 2.6333856357163258e-06, "loss": 2.6631, "step": 298400 }, { "epoch": 0.0023296, "grad_norm": 0.8188309669494629, "learning_rate": 2.632841080586661e-06, "loss": 2.6927, "step": 298410 }, { "epoch": 0.0023552, "grad_norm": 0.7961538434028625, "learning_rate": 2.6322965732310633e-06, "loss": 2.6851, "step": 298420 }, { "epoch": 0.0023808, "grad_norm": 0.8328834772109985, "learning_rate": 2.6317521136530643e-06, "loss": 2.7788, "step": 298430 }, { "epoch": 0.0024064, "grad_norm": 0.802546501159668, "learning_rate": 2.6312077018561942e-06, "loss": 2.5263, "step": 298440 }, { "epoch": 0.002432, "grad_norm": 0.9853793382644653, "learning_rate": 2.6306633378439874e-06, "loss": 2.8543, "step": 298450 }, { "epoch": 0.0024576, "grad_norm": 0.9550002813339233, "learning_rate": 2.6301190216199666e-06, "loss": 2.5099, "step": 298460 }, { "epoch": 0.0024832, "grad_norm": 0.9299010038375854, "learning_rate": 2.629574753187666e-06, "loss": 2.9651, "step": 298470 }, { "epoch": 0.0025088, "grad_norm": 0.8629811406135559, "learning_rate": 2.6290305325506137e-06, "loss": 2.9127, "step": 298480 }, { "epoch": 0.0025344, "grad_norm": 0.7746670842170715, "learning_rate": 2.6284863597123366e-06, "loss": 2.8122, "step": 298490 }, { "epoch": 0.00256, "grad_norm": 0.8128622174263, "learning_rate": 2.6279422346763693e-06, "loss": 2.5758, "step": 298500 }, { "epoch": 0.0025856, "grad_norm": 0.8503068685531616, "learning_rate": 2.6273981574462403e-06, "loss": 2.6365, "step": 298510 }, { "epoch": 0.0026112, "grad_norm": 0.8538016676902771, "learning_rate": 2.6268541280254722e-06, "loss": 3.0503, "step": 298520 }, { "epoch": 0.0026368, "grad_norm": 0.9883971214294434, "learning_rate": 2.6263101464175956e-06, "loss": 2.844, "step": 298530 }, { "epoch": 0.0026624, "grad_norm": 1.5589712858200073, "learning_rate": 2.6257662126261384e-06, "loss": 2.7926, "step": 298540 }, { "epoch": 0.002688, "grad_norm": 1.0181041955947876, "learning_rate": 2.625222326654626e-06, "loss": 2.7567, "step": 298550 }, { "epoch": 0.0027136, "grad_norm": 0.8654170036315918, "learning_rate": 2.624678488506588e-06, "loss": 2.8588, "step": 298560 }, { "epoch": 0.0027392, "grad_norm": 1.1274093389511108, "learning_rate": 2.6241346981855496e-06, "loss": 3.1156, "step": 298570 }, { "epoch": 0.0027648, "grad_norm": 0.8094833493232727, "learning_rate": 2.62359095569504e-06, "loss": 2.8504, "step": 298580 }, { "epoch": 0.0027904, "grad_norm": 0.7906102538108826, "learning_rate": 2.6230472610385806e-06, "loss": 2.6905, "step": 298590 }, { "epoch": 0.002816, "grad_norm": 0.7699177265167236, "learning_rate": 2.622503614219698e-06, "loss": 2.6625, "step": 298600 }, { "epoch": 0.0028416, "grad_norm": 0.7965071201324463, "learning_rate": 2.62196001524192e-06, "loss": 2.7876, "step": 298610 }, { "epoch": 0.0028672, "grad_norm": 0.833825409412384, "learning_rate": 2.621416464108769e-06, "loss": 2.9446, "step": 298620 }, { "epoch": 0.0028928, "grad_norm": 0.787368655204773, "learning_rate": 2.6208729608237725e-06, "loss": 2.8995, "step": 298630 }, { "epoch": 0.0029184, "grad_norm": 0.8216109275817871, "learning_rate": 2.6203295053904533e-06, "loss": 2.8777, "step": 298640 }, { "epoch": 0.002944, "grad_norm": 1.0031218528747559, "learning_rate": 2.619786097812339e-06, "loss": 2.8172, "step": 298650 }, { "epoch": 0.0029696, "grad_norm": 0.8245217204093933, "learning_rate": 2.6192427380929475e-06, "loss": 2.8145, "step": 298660 }, { "epoch": 0.0029952, "grad_norm": 0.7928630113601685, "learning_rate": 2.6186994262358024e-06, "loss": 2.9484, "step": 298670 }, { "epoch": 0.0030208, "grad_norm": 0.9326948523521423, "learning_rate": 2.618156162244432e-06, "loss": 2.7194, "step": 298680 }, { "epoch": 0.0030464, "grad_norm": 0.8810766339302063, "learning_rate": 2.6176129461223576e-06, "loss": 2.9362, "step": 298690 }, { "epoch": 0.003072, "grad_norm": 0.92030268907547, "learning_rate": 2.6170697778731015e-06, "loss": 2.8098, "step": 298700 }, { "epoch": 0.0030976, "grad_norm": 0.8537315726280212, "learning_rate": 2.616526657500188e-06, "loss": 2.8178, "step": 298710 }, { "epoch": 0.0031232, "grad_norm": 0.8163818120956421, "learning_rate": 2.6159835850071346e-06, "loss": 2.8203, "step": 298720 }, { "epoch": 0.0031488, "grad_norm": 0.8636249899864197, "learning_rate": 2.615440560397464e-06, "loss": 2.8327, "step": 298730 }, { "epoch": 0.0031744, "grad_norm": 0.8108971118927002, "learning_rate": 2.6148975836747002e-06, "loss": 2.7479, "step": 298740 }, { "epoch": 0.0032, "grad_norm": 0.8090289235115051, "learning_rate": 2.6143546548423628e-06, "loss": 2.6177, "step": 298750 }, { "epoch": 0.0032256, "grad_norm": 0.832785964012146, "learning_rate": 2.6138117739039715e-06, "loss": 2.9303, "step": 298760 }, { "epoch": 0.0032512, "grad_norm": 0.8529322147369385, "learning_rate": 2.6132689408630487e-06, "loss": 2.9504, "step": 298770 }, { "epoch": 0.0032768, "grad_norm": 0.8071756362915039, "learning_rate": 2.6127261557231164e-06, "loss": 2.9852, "step": 298780 }, { "epoch": 0.0033024, "grad_norm": 0.7928798198699951, "learning_rate": 2.612183418487688e-06, "loss": 2.8515, "step": 298790 }, { "epoch": 0.003328, "grad_norm": 0.8011922836303711, "learning_rate": 2.6116407291602874e-06, "loss": 3.0791, "step": 298800 }, { "epoch": 0.0033536, "grad_norm": 0.8039498329162598, "learning_rate": 2.6110980877444338e-06, "loss": 2.9044, "step": 298810 }, { "epoch": 0.0033792, "grad_norm": 0.8011322617530823, "learning_rate": 2.610555494243644e-06, "loss": 2.831, "step": 298820 }, { "epoch": 0.0034048, "grad_norm": 0.8218445181846619, "learning_rate": 2.610012948661439e-06, "loss": 2.7149, "step": 298830 }, { "epoch": 0.0034304, "grad_norm": 0.9098562598228455, "learning_rate": 2.6094704510013356e-06, "loss": 2.9693, "step": 298840 }, { "epoch": 0.003456, "grad_norm": 0.7971295118331909, "learning_rate": 2.6089280012668518e-06, "loss": 2.7595, "step": 298850 }, { "epoch": 0.0034816, "grad_norm": 0.8650346994400024, "learning_rate": 2.6083855994615057e-06, "loss": 2.93, "step": 298860 }, { "epoch": 0.0035072, "grad_norm": 0.8723875284194946, "learning_rate": 2.607843245588815e-06, "loss": 2.5999, "step": 298870 }, { "epoch": 0.0035328, "grad_norm": 0.796758770942688, "learning_rate": 2.6073009396522964e-06, "loss": 2.9251, "step": 298880 }, { "epoch": 0.0035584, "grad_norm": 0.7493913173675537, "learning_rate": 2.6067586816554656e-06, "loss": 2.7817, "step": 298890 }, { "epoch": 0.003584, "grad_norm": 0.8016611337661743, "learning_rate": 2.60621647160184e-06, "loss": 2.7889, "step": 298900 }, { "epoch": 0.0036096, "grad_norm": 1.0370423793792725, "learning_rate": 2.6056743094949397e-06, "loss": 3.0006, "step": 298910 }, { "epoch": 0.0036352, "grad_norm": 0.8724368810653687, "learning_rate": 2.6051321953382726e-06, "loss": 2.8543, "step": 298920 }, { "epoch": 0.0036608, "grad_norm": 0.9131735563278198, "learning_rate": 2.6045901291353572e-06, "loss": 2.9159, "step": 298930 }, { "epoch": 0.0036864, "grad_norm": 0.8231111168861389, "learning_rate": 2.6040481108897096e-06, "loss": 3.0205, "step": 298940 }, { "epoch": 0.003712, "grad_norm": 0.9646931290626526, "learning_rate": 2.6035061406048454e-06, "loss": 2.8798, "step": 298950 }, { "epoch": 0.0037376, "grad_norm": 0.7615324854850769, "learning_rate": 2.6029642182842773e-06, "loss": 2.6933, "step": 298960 }, { "epoch": 0.0037632, "grad_norm": 0.923069417476654, "learning_rate": 2.602422343931521e-06, "loss": 2.7021, "step": 298970 }, { "epoch": 0.0037888, "grad_norm": 0.7932442426681519, "learning_rate": 2.6018805175500915e-06, "loss": 2.6899, "step": 298980 }, { "epoch": 0.0038144, "grad_norm": 0.8934822678565979, "learning_rate": 2.6013387391434984e-06, "loss": 2.6606, "step": 298990 }, { "epoch": 0.00384, "grad_norm": 0.8254549503326416, "learning_rate": 2.600797008715258e-06, "loss": 2.8256, "step": 299000 }, { "epoch": 0.0038656, "grad_norm": 1.0146406888961792, "learning_rate": 2.600255326268881e-06, "loss": 3.0216, "step": 299010 }, { "epoch": 0.0038912, "grad_norm": 0.8090090155601501, "learning_rate": 2.59971369180788e-06, "loss": 2.6676, "step": 299020 }, { "epoch": 0.0039168, "grad_norm": 0.8205191493034363, "learning_rate": 2.5991721053357744e-06, "loss": 2.8281, "step": 299030 }, { "epoch": 0.0039424, "grad_norm": 0.8204403519630432, "learning_rate": 2.5986305668560675e-06, "loss": 2.7725, "step": 299040 }, { "epoch": 0.003968, "grad_norm": 0.7754949927330017, "learning_rate": 2.5980890763722743e-06, "loss": 2.796, "step": 299050 }, { "epoch": 0.0039936, "grad_norm": 0.7890117764472961, "learning_rate": 2.597547633887907e-06, "loss": 2.7742, "step": 299060 }, { "epoch": 0.0040192, "grad_norm": 0.9344207048416138, "learning_rate": 2.5970062394064756e-06, "loss": 2.7932, "step": 299070 }, { "epoch": 0.0040448, "grad_norm": 0.8774805665016174, "learning_rate": 2.5964648929314908e-06, "loss": 2.7487, "step": 299080 }, { "epoch": 0.0040704, "grad_norm": 0.8827197551727295, "learning_rate": 2.595923594466464e-06, "loss": 2.8129, "step": 299090 }, { "epoch": 0.004096, "grad_norm": 0.8802890181541443, "learning_rate": 2.595382344014904e-06, "loss": 2.8024, "step": 299100 }, { "epoch": 0.0041216, "grad_norm": 0.8130683898925781, "learning_rate": 2.5948411415803253e-06, "loss": 2.8588, "step": 299110 }, { "epoch": 0.0041472, "grad_norm": 0.9647886157035828, "learning_rate": 2.594299987166231e-06, "loss": 2.8332, "step": 299120 }, { "epoch": 0.0041728, "grad_norm": 0.8744273781776428, "learning_rate": 2.593758880776133e-06, "loss": 2.9002, "step": 299130 }, { "epoch": 0.0041984, "grad_norm": 0.8491162061691284, "learning_rate": 2.5932178224135406e-06, "loss": 2.7633, "step": 299140 }, { "epoch": 0.004224, "grad_norm": 0.8746843934059143, "learning_rate": 2.5926768120819614e-06, "loss": 2.8703, "step": 299150 }, { "epoch": 0.0042496, "grad_norm": 1.1285320520401, "learning_rate": 2.592135849784905e-06, "loss": 2.8727, "step": 299160 }, { "epoch": 0.0042752, "grad_norm": 0.8277060985565186, "learning_rate": 2.5915949355258785e-06, "loss": 2.7986, "step": 299170 }, { "epoch": 0.0043008, "grad_norm": 0.7725759744644165, "learning_rate": 2.591054069308393e-06, "loss": 2.7884, "step": 299180 }, { "epoch": 0.0043264, "grad_norm": 0.8946351408958435, "learning_rate": 2.5905132511359466e-06, "loss": 2.8046, "step": 299190 }, { "epoch": 0.004352, "grad_norm": 0.7898014783859253, "learning_rate": 2.5899724810120565e-06, "loss": 2.8194, "step": 299200 }, { "epoch": 0.0043776, "grad_norm": 0.8251870274543762, "learning_rate": 2.589431758940224e-06, "loss": 2.842, "step": 299210 }, { "epoch": 0.0044032, "grad_norm": 0.7975075244903564, "learning_rate": 2.588891084923958e-06, "loss": 2.9727, "step": 299220 }, { "epoch": 0.0044288, "grad_norm": 0.8084461688995361, "learning_rate": 2.5883504589667675e-06, "loss": 2.7528, "step": 299230 }, { "epoch": 0.0044544, "grad_norm": 0.8054526448249817, "learning_rate": 2.5878098810721496e-06, "loss": 2.7151, "step": 299240 }, { "epoch": 0.00448, "grad_norm": 0.9267657399177551, "learning_rate": 2.5872693512436165e-06, "loss": 2.8224, "step": 299250 }, { "epoch": 0.0045056, "grad_norm": 0.9034633040428162, "learning_rate": 2.5867288694846703e-06, "loss": 2.7608, "step": 299260 }, { "epoch": 0.0045312, "grad_norm": 1.2848964929580688, "learning_rate": 2.5861884357988177e-06, "loss": 2.9059, "step": 299270 }, { "epoch": 0.0045568, "grad_norm": 0.9746647477149963, "learning_rate": 2.585648050189562e-06, "loss": 2.5617, "step": 299280 }, { "epoch": 0.0045824, "grad_norm": 1.116288661956787, "learning_rate": 2.5851077126604096e-06, "loss": 3.1233, "step": 299290 }, { "epoch": 0.004608, "grad_norm": 0.8422269821166992, "learning_rate": 2.584567423214862e-06, "loss": 2.7038, "step": 299300 }, { "epoch": 0.0046336, "grad_norm": 0.801517903804779, "learning_rate": 2.5840271818564265e-06, "loss": 2.6133, "step": 299310 }, { "epoch": 0.0046592, "grad_norm": 2.0187649726867676, "learning_rate": 2.5834869885886016e-06, "loss": 3.1218, "step": 299320 }, { "epoch": 0.0046848, "grad_norm": 0.8146814703941345, "learning_rate": 2.582946843414893e-06, "loss": 2.9328, "step": 299330 }, { "epoch": 0.0047104, "grad_norm": 1.1114555597305298, "learning_rate": 2.5824067463388023e-06, "loss": 2.7206, "step": 299340 }, { "epoch": 0.004736, "grad_norm": 0.7832559943199158, "learning_rate": 2.5818666973638316e-06, "loss": 2.8223, "step": 299350 }, { "epoch": 0.0047616, "grad_norm": 0.9091808795928955, "learning_rate": 2.5813266964934857e-06, "loss": 2.9948, "step": 299360 }, { "epoch": 0.0047872, "grad_norm": 0.9583549499511719, "learning_rate": 2.580786743731263e-06, "loss": 2.7921, "step": 299370 }, { "epoch": 0.0048128, "grad_norm": 0.8837670087814331, "learning_rate": 2.580246839080667e-06, "loss": 2.7636, "step": 299380 }, { "epoch": 0.0048384, "grad_norm": 0.8551021814346313, "learning_rate": 2.579706982545198e-06, "loss": 2.5416, "step": 299390 }, { "epoch": 0.004864, "grad_norm": 0.937035322189331, "learning_rate": 2.5791671741283585e-06, "loss": 2.8523, "step": 299400 }, { "epoch": 0.0048896, "grad_norm": 0.7908353209495544, "learning_rate": 2.5786274138336464e-06, "loss": 2.6429, "step": 299410 }, { "epoch": 0.0049152, "grad_norm": 0.8589228391647339, "learning_rate": 2.5780877016645635e-06, "loss": 2.8332, "step": 299420 }, { "epoch": 0.0049408, "grad_norm": 0.8139799237251282, "learning_rate": 2.5775480376246122e-06, "loss": 2.8438, "step": 299430 }, { "epoch": 0.0049664, "grad_norm": 0.7503485083580017, "learning_rate": 2.5770084217172866e-06, "loss": 2.8133, "step": 299440 }, { "epoch": 0.004992, "grad_norm": 0.7811346650123596, "learning_rate": 2.576468853946088e-06, "loss": 2.9044, "step": 299450 }, { "epoch": 0.0050176, "grad_norm": 0.9738094210624695, "learning_rate": 2.5759293343145153e-06, "loss": 2.8133, "step": 299460 }, { "epoch": 0.0050432, "grad_norm": 0.8666238188743591, "learning_rate": 2.5753898628260687e-06, "loss": 2.821, "step": 299470 }, { "epoch": 0.0050688, "grad_norm": 0.7826719880104065, "learning_rate": 2.5748504394842454e-06, "loss": 2.7597, "step": 299480 }, { "epoch": 0.0050944, "grad_norm": 0.864764928817749, "learning_rate": 2.574311064292544e-06, "loss": 2.7353, "step": 299490 }, { "epoch": 0.00512, "grad_norm": 0.8209632039070129, "learning_rate": 2.5737717372544624e-06, "loss": 2.7757, "step": 299500 }, { "epoch": 0.0051456, "grad_norm": 0.8997071981430054, "learning_rate": 2.573232458373499e-06, "loss": 2.793, "step": 299510 }, { "epoch": 0.0051712, "grad_norm": 0.7817344665527344, "learning_rate": 2.5726932276531467e-06, "loss": 2.7126, "step": 299520 }, { "epoch": 0.0051968, "grad_norm": 0.8323398232460022, "learning_rate": 2.5721540450969063e-06, "loss": 2.7081, "step": 299530 }, { "epoch": 0.0052224, "grad_norm": 0.7606375217437744, "learning_rate": 2.5716149107082688e-06, "loss": 2.7692, "step": 299540 }, { "epoch": 0.005248, "grad_norm": 0.8311121463775635, "learning_rate": 2.5710758244907364e-06, "loss": 2.8915, "step": 299550 }, { "epoch": 0.0052736, "grad_norm": 1.1459070444107056, "learning_rate": 2.570536786447807e-06, "loss": 3.0316, "step": 299560 }, { "epoch": 0.0052992, "grad_norm": 0.8382669687271118, "learning_rate": 2.569997796582969e-06, "loss": 2.7601, "step": 299570 }, { "epoch": 0.0053248, "grad_norm": 0.9143301844596863, "learning_rate": 2.5694588548997203e-06, "loss": 2.8101, "step": 299580 }, { "epoch": 0.0053504, "grad_norm": 0.813877522945404, "learning_rate": 2.568919961401556e-06, "loss": 2.5073, "step": 299590 }, { "epoch": 0.005376, "grad_norm": 0.852277934551239, "learning_rate": 2.568381116091971e-06, "loss": 2.7796, "step": 299600 }, { "epoch": 0.0054016, "grad_norm": 1.1226696968078613, "learning_rate": 2.5678423189744595e-06, "loss": 2.9518, "step": 299610 }, { "epoch": 0.0054272, "grad_norm": 0.8480280637741089, "learning_rate": 2.567303570052516e-06, "loss": 2.6241, "step": 299620 }, { "epoch": 0.0054528, "grad_norm": 0.9600939750671387, "learning_rate": 2.5667648693296352e-06, "loss": 2.791, "step": 299630 }, { "epoch": 0.0054784, "grad_norm": 0.9561883211135864, "learning_rate": 2.5662262168093064e-06, "loss": 2.7732, "step": 299640 }, { "epoch": 0.005504, "grad_norm": 0.8948724865913391, "learning_rate": 2.565687612495025e-06, "loss": 2.7385, "step": 299650 }, { "epoch": 0.0055296, "grad_norm": 0.7701829075813293, "learning_rate": 2.565149056390284e-06, "loss": 2.6423, "step": 299660 }, { "epoch": 0.0055552, "grad_norm": 0.927509069442749, "learning_rate": 2.564610548498574e-06, "loss": 2.6583, "step": 299670 }, { "epoch": 0.0055808, "grad_norm": 0.7728626132011414, "learning_rate": 2.5640720888233894e-06, "loss": 2.8161, "step": 299680 }, { "epoch": 0.0056064, "grad_norm": 0.8396449089050293, "learning_rate": 2.5635336773682207e-06, "loss": 2.7221, "step": 299690 }, { "epoch": 0.005632, "grad_norm": 0.9608518481254578, "learning_rate": 2.5629953141365637e-06, "loss": 2.8615, "step": 299700 }, { "epoch": 0.0056576, "grad_norm": 0.873335599899292, "learning_rate": 2.5624569991318983e-06, "loss": 2.8312, "step": 299710 }, { "epoch": 0.0056832, "grad_norm": 0.7868092060089111, "learning_rate": 2.5619187323577267e-06, "loss": 2.859, "step": 299720 }, { "epoch": 0.0057088, "grad_norm": 0.8798430562019348, "learning_rate": 2.561380513817535e-06, "loss": 2.7738, "step": 299730 }, { "epoch": 0.0057344, "grad_norm": 0.7730625867843628, "learning_rate": 2.560842343514813e-06, "loss": 2.9005, "step": 299740 }, { "epoch": 0.00576, "grad_norm": 0.8744128942489624, "learning_rate": 2.5603042214530515e-06, "loss": 2.9443, "step": 299750 }, { "epoch": 0.0057856, "grad_norm": 0.7982550859451294, "learning_rate": 2.5597661476357437e-06, "loss": 2.7117, "step": 299760 }, { "epoch": 0.0058112, "grad_norm": 0.7522895932197571, "learning_rate": 2.5592281220663715e-06, "loss": 2.6406, "step": 299770 }, { "epoch": 0.0058368, "grad_norm": 0.8226358890533447, "learning_rate": 2.558690144748428e-06, "loss": 2.7989, "step": 299780 }, { "epoch": 0.0058624, "grad_norm": 0.9121454358100891, "learning_rate": 2.5581522156854e-06, "loss": 2.6694, "step": 299790 }, { "epoch": 0.005888, "grad_norm": 0.7933983206748962, "learning_rate": 2.557614334880778e-06, "loss": 2.9072, "step": 299800 }, { "epoch": 0.0059136, "grad_norm": 0.8359168767929077, "learning_rate": 2.5570765023380493e-06, "loss": 2.6048, "step": 299810 }, { "epoch": 0.0059392, "grad_norm": 0.9520846009254456, "learning_rate": 2.5565387180607006e-06, "loss": 2.7511, "step": 299820 }, { "epoch": 0.0059648, "grad_norm": 0.8650608062744141, "learning_rate": 2.5560009820522235e-06, "loss": 2.7234, "step": 299830 }, { "epoch": 0.0059904, "grad_norm": 0.8819663524627686, "learning_rate": 2.5554632943160983e-06, "loss": 2.8006, "step": 299840 }, { "epoch": 0.006016, "grad_norm": 0.8477904796600342, "learning_rate": 2.5549256548558154e-06, "loss": 2.8373, "step": 299850 }, { "epoch": 0.0060416, "grad_norm": 0.9606834650039673, "learning_rate": 2.554388063674862e-06, "loss": 2.7979, "step": 299860 }, { "epoch": 0.0060672, "grad_norm": 0.7668954133987427, "learning_rate": 2.5538505207767215e-06, "loss": 2.6973, "step": 299870 }, { "epoch": 0.0060928, "grad_norm": 0.8300150632858276, "learning_rate": 2.553313026164882e-06, "loss": 2.6588, "step": 299880 }, { "epoch": 0.0061184, "grad_norm": 1.059448003768921, "learning_rate": 2.5527755798428287e-06, "loss": 2.7888, "step": 299890 }, { "epoch": 0.006144, "grad_norm": 0.8127474188804626, "learning_rate": 2.5522381818140463e-06, "loss": 2.7105, "step": 299900 }, { "epoch": 0.0061696, "grad_norm": 0.9239873290061951, "learning_rate": 2.5517008320820204e-06, "loss": 2.8165, "step": 299910 }, { "epoch": 0.0061952, "grad_norm": 0.8719655871391296, "learning_rate": 2.551163530650235e-06, "loss": 2.7645, "step": 299920 }, { "epoch": 0.0062208, "grad_norm": 0.8961420059204102, "learning_rate": 2.5506262775221735e-06, "loss": 2.8283, "step": 299930 }, { "epoch": 0.0062464, "grad_norm": 0.9636145234107971, "learning_rate": 2.550089072701322e-06, "loss": 2.8516, "step": 299940 }, { "epoch": 0.006272, "grad_norm": 0.8086772561073303, "learning_rate": 2.549551916191162e-06, "loss": 2.8776, "step": 299950 }, { "epoch": 0.0062976, "grad_norm": 0.8418802618980408, "learning_rate": 2.5490148079951817e-06, "loss": 3.0549, "step": 299960 }, { "epoch": 0.0063232, "grad_norm": 0.8164690136909485, "learning_rate": 2.5484777481168567e-06, "loss": 2.9804, "step": 299970 }, { "epoch": 0.0063488, "grad_norm": 0.7321439385414124, "learning_rate": 2.547940736559672e-06, "loss": 2.773, "step": 299980 }, { "epoch": 0.0063744, "grad_norm": 0.8703297972679138, "learning_rate": 2.547403773327112e-06, "loss": 2.7221, "step": 299990 }, { "epoch": 0.0064, "grad_norm": 0.8115136027336121, "learning_rate": 2.5468668584226584e-06, "loss": 2.7111, "step": 300000 }, { "epoch": 0.0064256, "grad_norm": 0.7564700841903687, "learning_rate": 2.546329991849792e-06, "loss": 2.7492, "step": 300010 }, { "epoch": 0.0064512, "grad_norm": 0.7812654376029968, "learning_rate": 2.5457931736119944e-06, "loss": 2.7306, "step": 300020 }, { "epoch": 0.0064768, "grad_norm": 0.7921497821807861, "learning_rate": 2.54525640371275e-06, "loss": 2.7849, "step": 300030 }, { "epoch": 0.0065024, "grad_norm": 0.7761502861976624, "learning_rate": 2.5447196821555343e-06, "loss": 2.6622, "step": 300040 }, { "epoch": 0.006528, "grad_norm": 0.8627834916114807, "learning_rate": 2.5441830089438292e-06, "loss": 2.7255, "step": 300050 }, { "epoch": 0.0065536, "grad_norm": 0.8646459579467773, "learning_rate": 2.5436463840811134e-06, "loss": 2.8318, "step": 300060 }, { "epoch": 0.0065792, "grad_norm": 0.8661954998970032, "learning_rate": 2.5431098075708714e-06, "loss": 2.824, "step": 300070 }, { "epoch": 0.0066048, "grad_norm": 0.7928588390350342, "learning_rate": 2.5425732794165816e-06, "loss": 2.7628, "step": 300080 }, { "epoch": 0.0066304, "grad_norm": 0.8205761909484863, "learning_rate": 2.542036799621723e-06, "loss": 2.5924, "step": 300090 }, { "epoch": 0.006656, "grad_norm": 0.7502779960632324, "learning_rate": 2.5415003681897733e-06, "loss": 2.7921, "step": 300100 }, { "epoch": 0.0066816, "grad_norm": 0.8124361038208008, "learning_rate": 2.5409639851242098e-06, "loss": 2.8603, "step": 300110 }, { "epoch": 0.0067072, "grad_norm": 0.8426488637924194, "learning_rate": 2.540427650428513e-06, "loss": 2.5685, "step": 300120 }, { "epoch": 0.0067328, "grad_norm": 0.9458847641944885, "learning_rate": 2.539891364106161e-06, "loss": 2.9308, "step": 300130 }, { "epoch": 0.0067584, "grad_norm": 0.8202309012413025, "learning_rate": 2.53935512616063e-06, "loss": 2.8654, "step": 300140 }, { "epoch": 0.006784, "grad_norm": 0.7772131562232971, "learning_rate": 2.5388189365953988e-06, "loss": 2.7427, "step": 300150 }, { "epoch": 0.0068096, "grad_norm": 0.8583332896232605, "learning_rate": 2.5382827954139457e-06, "loss": 2.8462, "step": 300160 }, { "epoch": 0.0068352, "grad_norm": 0.7735180854797363, "learning_rate": 2.537746702619743e-06, "loss": 2.9253, "step": 300170 }, { "epoch": 0.0068608, "grad_norm": 1.1430193185806274, "learning_rate": 2.5372106582162703e-06, "loss": 2.8456, "step": 300180 }, { "epoch": 0.0068864, "grad_norm": 0.7880129218101501, "learning_rate": 2.5366746622070026e-06, "loss": 2.8069, "step": 300190 }, { "epoch": 0.006912, "grad_norm": 0.8210017681121826, "learning_rate": 2.5361387145954165e-06, "loss": 2.5598, "step": 300200 }, { "epoch": 0.0069376, "grad_norm": 0.8825343251228333, "learning_rate": 2.5356028153849876e-06, "loss": 2.766, "step": 300210 }, { "epoch": 0.0069632, "grad_norm": 0.8419293165206909, "learning_rate": 2.5350669645791902e-06, "loss": 2.7683, "step": 300220 }, { "epoch": 0.0069888, "grad_norm": 0.829944372177124, "learning_rate": 2.5345311621814995e-06, "loss": 2.7004, "step": 300230 }, { "epoch": 0.0070144, "grad_norm": 1.0120881795883179, "learning_rate": 2.53399540819539e-06, "loss": 2.7836, "step": 300240 }, { "epoch": 0.00704, "grad_norm": 0.8121423721313477, "learning_rate": 2.5334597026243367e-06, "loss": 2.8768, "step": 300250 }, { "epoch": 0.0070656, "grad_norm": 0.7817355394363403, "learning_rate": 2.532924045471812e-06, "loss": 2.7048, "step": 300260 }, { "epoch": 0.0070912, "grad_norm": 0.9165147542953491, "learning_rate": 2.532388436741291e-06, "loss": 2.9664, "step": 300270 }, { "epoch": 0.0071168, "grad_norm": 0.8038797378540039, "learning_rate": 2.531852876436246e-06, "loss": 2.6731, "step": 300280 }, { "epoch": 0.0071424, "grad_norm": 0.834350049495697, "learning_rate": 2.531317364560153e-06, "loss": 3.0523, "step": 300290 }, { "epoch": 0.007168, "grad_norm": 0.834199845790863, "learning_rate": 2.5307819011164804e-06, "loss": 2.9232, "step": 300300 }, { "epoch": 0.0071936, "grad_norm": 0.83800208568573, "learning_rate": 2.530246486108701e-06, "loss": 2.8633, "step": 300310 }, { "epoch": 0.0072192, "grad_norm": 0.8350838422775269, "learning_rate": 2.529711119540288e-06, "loss": 2.8328, "step": 300320 }, { "epoch": 0.0072448, "grad_norm": 0.8874048590660095, "learning_rate": 2.529175801414713e-06, "loss": 3.105, "step": 300330 }, { "epoch": 0.0072704, "grad_norm": 0.9955283999443054, "learning_rate": 2.528640531735447e-06, "loss": 2.8826, "step": 300340 }, { "epoch": 0.007296, "grad_norm": 0.9030163288116455, "learning_rate": 2.5281053105059616e-06, "loss": 2.7936, "step": 300350 }, { "epoch": 0.0073216, "grad_norm": 0.829574704170227, "learning_rate": 2.5275701377297313e-06, "loss": 2.658, "step": 300360 }, { "epoch": 0.0073472, "grad_norm": 0.7373389005661011, "learning_rate": 2.5270350134102184e-06, "loss": 2.8185, "step": 300370 }, { "epoch": 0.0073728, "grad_norm": 0.8040632605552673, "learning_rate": 2.526499937550898e-06, "loss": 2.907, "step": 300380 }, { "epoch": 0.0073984, "grad_norm": 0.8782640695571899, "learning_rate": 2.5259649101552396e-06, "loss": 2.5792, "step": 300390 }, { "epoch": 0.007424, "grad_norm": 0.7672430276870728, "learning_rate": 2.525429931226713e-06, "loss": 2.8513, "step": 300400 }, { "epoch": 0.0074496, "grad_norm": 0.8351352214813232, "learning_rate": 2.5248950007687834e-06, "loss": 2.8015, "step": 300410 }, { "epoch": 0.0074752, "grad_norm": 0.7964571118354797, "learning_rate": 2.5243601187849275e-06, "loss": 2.9468, "step": 300420 }, { "epoch": 0.0075008, "grad_norm": 0.7781196236610413, "learning_rate": 2.523825285278608e-06, "loss": 2.9399, "step": 300430 }, { "epoch": 0.0075264, "grad_norm": 0.7861915230751038, "learning_rate": 2.523290500253295e-06, "loss": 2.8386, "step": 300440 }, { "epoch": 0.007552, "grad_norm": 0.8515740036964417, "learning_rate": 2.5227557637124546e-06, "loss": 2.7314, "step": 300450 }, { "epoch": 0.0075776, "grad_norm": 0.8710972666740417, "learning_rate": 2.5222210756595567e-06, "loss": 2.8777, "step": 300460 }, { "epoch": 0.0076032, "grad_norm": 0.8913779258728027, "learning_rate": 2.5216864360980676e-06, "loss": 2.9788, "step": 300470 }, { "epoch": 0.0076288, "grad_norm": 0.8370050191879272, "learning_rate": 2.5211518450314544e-06, "loss": 3.1202, "step": 300480 }, { "epoch": 0.0076544, "grad_norm": 0.7924556732177734, "learning_rate": 2.5206173024631876e-06, "loss": 2.8115, "step": 300490 }, { "epoch": 0.00768, "grad_norm": 0.8125472068786621, "learning_rate": 2.5200828083967254e-06, "loss": 2.8847, "step": 300500 }, { "epoch": 0.0077056, "grad_norm": 0.7845990657806396, "learning_rate": 2.5195483628355388e-06, "loss": 2.8789, "step": 300510 }, { "epoch": 0.0077312, "grad_norm": 0.725914716720581, "learning_rate": 2.5190139657830935e-06, "loss": 2.6008, "step": 300520 }, { "epoch": 0.0077568, "grad_norm": 0.8430520296096802, "learning_rate": 2.518479617242854e-06, "loss": 2.9989, "step": 300530 }, { "epoch": 0.0077824, "grad_norm": 0.8437262177467346, "learning_rate": 2.5179453172182868e-06, "loss": 2.9108, "step": 300540 }, { "epoch": 0.007808, "grad_norm": 0.7893911004066467, "learning_rate": 2.517411065712855e-06, "loss": 2.7529, "step": 300550 }, { "epoch": 0.0078336, "grad_norm": 0.8381913900375366, "learning_rate": 2.5168768627300256e-06, "loss": 2.7189, "step": 300560 }, { "epoch": 0.0078592, "grad_norm": 0.8250235319137573, "learning_rate": 2.5163427082732595e-06, "loss": 2.9693, "step": 300570 }, { "epoch": 0.0078848, "grad_norm": 0.9293398261070251, "learning_rate": 2.5158086023460182e-06, "loss": 2.5447, "step": 300580 }, { "epoch": 0.0079104, "grad_norm": 0.9282770752906799, "learning_rate": 2.5152745449517722e-06, "loss": 2.993, "step": 300590 }, { "epoch": 0.007936, "grad_norm": 0.8190063238143921, "learning_rate": 2.5147405360939825e-06, "loss": 2.7735, "step": 300600 }, { "epoch": 0.0079616, "grad_norm": 0.804230809211731, "learning_rate": 2.5142065757761103e-06, "loss": 2.7706, "step": 300610 }, { "epoch": 0.0079872, "grad_norm": 0.7975907325744629, "learning_rate": 2.5136726640016218e-06, "loss": 2.8136, "step": 300620 }, { "epoch": 0.0080128, "grad_norm": 0.8973631262779236, "learning_rate": 2.5131388007739745e-06, "loss": 2.7357, "step": 300630 }, { "epoch": 0.0080384, "grad_norm": 0.8804994225502014, "learning_rate": 2.5126049860966308e-06, "loss": 2.8005, "step": 300640 }, { "epoch": 0.008064, "grad_norm": 0.7465834617614746, "learning_rate": 2.5120712199730545e-06, "loss": 2.8391, "step": 300650 }, { "epoch": 0.0080896, "grad_norm": 0.9075197577476501, "learning_rate": 2.511537502406707e-06, "loss": 2.8564, "step": 300660 }, { "epoch": 0.0081152, "grad_norm": 0.8143641352653503, "learning_rate": 2.5110038334010477e-06, "loss": 3.0959, "step": 300670 }, { "epoch": 0.0081408, "grad_norm": 0.8930056095123291, "learning_rate": 2.5104702129595383e-06, "loss": 2.71, "step": 300680 }, { "epoch": 0.0081664, "grad_norm": 0.8973559141159058, "learning_rate": 2.509936641085642e-06, "loss": 2.7184, "step": 300690 }, { "epoch": 0.008192, "grad_norm": 0.7934931516647339, "learning_rate": 2.5094031177828137e-06, "loss": 2.9186, "step": 300700 }, { "epoch": 0.0082176, "grad_norm": 0.827933132648468, "learning_rate": 2.5088696430545145e-06, "loss": 2.8197, "step": 300710 }, { "epoch": 0.0082432, "grad_norm": 0.8568034768104553, "learning_rate": 2.508336216904206e-06, "loss": 2.8223, "step": 300720 }, { "epoch": 0.0082688, "grad_norm": 0.924556314945221, "learning_rate": 2.5078028393353448e-06, "loss": 2.9104, "step": 300730 }, { "epoch": 0.0082944, "grad_norm": 0.8704134225845337, "learning_rate": 2.507269510351391e-06, "loss": 2.9544, "step": 300740 }, { "epoch": 0.00832, "grad_norm": 0.7651199102401733, "learning_rate": 2.506736229955804e-06, "loss": 2.751, "step": 300750 }, { "epoch": 0.0083456, "grad_norm": 1.100642204284668, "learning_rate": 2.5062029981520406e-06, "loss": 2.7893, "step": 300760 }, { "epoch": 0.0083712, "grad_norm": 0.7544904351234436, "learning_rate": 2.50566981494356e-06, "loss": 2.8685, "step": 300770 }, { "epoch": 0.0083968, "grad_norm": 0.7273175120353699, "learning_rate": 2.5051366803338174e-06, "loss": 3.0397, "step": 300780 }, { "epoch": 0.0084224, "grad_norm": 0.7475700974464417, "learning_rate": 2.5046035943262727e-06, "loss": 2.7541, "step": 300790 }, { "epoch": 0.008448, "grad_norm": 1.0511350631713867, "learning_rate": 2.5040705569243804e-06, "loss": 2.602, "step": 300800 }, { "epoch": 0.0084736, "grad_norm": 0.8663104772567749, "learning_rate": 2.5035375681316e-06, "loss": 2.8403, "step": 300810 }, { "epoch": 0.0084992, "grad_norm": 0.9156891703605652, "learning_rate": 2.5030046279513874e-06, "loss": 3.0336, "step": 300820 }, { "epoch": 0.0085248, "grad_norm": 0.9275625348091125, "learning_rate": 2.5024717363871954e-06, "loss": 2.9657, "step": 300830 }, { "epoch": 0.0085504, "grad_norm": 0.8756067752838135, "learning_rate": 2.5019388934424816e-06, "loss": 2.8616, "step": 300840 }, { "epoch": 0.008576, "grad_norm": 0.9332289695739746, "learning_rate": 2.501406099120701e-06, "loss": 2.8084, "step": 300850 }, { "epoch": 0.0086016, "grad_norm": 1.15931236743927, "learning_rate": 2.5008733534253095e-06, "loss": 2.9949, "step": 300860 }, { "epoch": 0.0086272, "grad_norm": 0.7631269693374634, "learning_rate": 2.500340656359761e-06, "loss": 2.6654, "step": 300870 }, { "epoch": 0.0086528, "grad_norm": 0.888275682926178, "learning_rate": 2.49980800792751e-06, "loss": 2.7621, "step": 300880 }, { "epoch": 0.0086784, "grad_norm": 0.8296595811843872, "learning_rate": 2.4992754081320136e-06, "loss": 2.7761, "step": 300890 }, { "epoch": 0.008704, "grad_norm": 0.8035411834716797, "learning_rate": 2.4987428569767204e-06, "loss": 2.7775, "step": 300900 }, { "epoch": 0.0087296, "grad_norm": 0.9361587166786194, "learning_rate": 2.4982103544650848e-06, "loss": 2.6226, "step": 300910 }, { "epoch": 0.0087552, "grad_norm": 0.7276634573936462, "learning_rate": 2.4976779006005626e-06, "loss": 2.8628, "step": 300920 }, { "epoch": 0.0087808, "grad_norm": 0.8548344969749451, "learning_rate": 2.4971454953866012e-06, "loss": 2.8493, "step": 300930 }, { "epoch": 0.0088064, "grad_norm": 0.7735708951950073, "learning_rate": 2.4966131388266647e-06, "loss": 2.6014, "step": 300940 }, { "epoch": 0.008832, "grad_norm": 0.8077785968780518, "learning_rate": 2.4960808309241937e-06, "loss": 2.8854, "step": 300950 }, { "epoch": 0.0088576, "grad_norm": 0.7562600374221802, "learning_rate": 2.4955485716826444e-06, "loss": 2.8414, "step": 300960 }, { "epoch": 0.0088832, "grad_norm": 0.7743840217590332, "learning_rate": 2.495016361105469e-06, "loss": 2.6923, "step": 300970 }, { "epoch": 0.0089088, "grad_norm": 0.848353385925293, "learning_rate": 2.4944841991961177e-06, "loss": 2.7442, "step": 300980 }, { "epoch": 0.0089344, "grad_norm": 0.7729135751724243, "learning_rate": 2.493952085958041e-06, "loss": 2.9878, "step": 300990 }, { "epoch": 0.00896, "grad_norm": 0.798248291015625, "learning_rate": 2.49342002139469e-06, "loss": 2.8074, "step": 301000 }, { "epoch": 0.0089856, "grad_norm": 0.7621723413467407, "learning_rate": 2.4928880055095186e-06, "loss": 2.7698, "step": 301010 }, { "epoch": 0.0090112, "grad_norm": 0.8086488842964172, "learning_rate": 2.4923560383059708e-06, "loss": 2.7248, "step": 301020 }, { "epoch": 0.0090368, "grad_norm": 0.7917207479476929, "learning_rate": 2.4918241197874983e-06, "loss": 2.7544, "step": 301030 }, { "epoch": 0.0090624, "grad_norm": 0.8041267991065979, "learning_rate": 2.4912922499575508e-06, "loss": 2.8833, "step": 301040 }, { "epoch": 0.009088, "grad_norm": 0.7960098385810852, "learning_rate": 2.4907604288195776e-06, "loss": 2.8944, "step": 301050 }, { "epoch": 0.0091136, "grad_norm": 0.8381949663162231, "learning_rate": 2.490228656377027e-06, "loss": 2.7191, "step": 301060 }, { "epoch": 0.0091392, "grad_norm": 0.8202736973762512, "learning_rate": 2.4896969326333485e-06, "loss": 2.7799, "step": 301070 }, { "epoch": 0.0091648, "grad_norm": 0.9386478662490845, "learning_rate": 2.489165257591989e-06, "loss": 2.6925, "step": 301080 }, { "epoch": 0.0091904, "grad_norm": 1.1130011081695557, "learning_rate": 2.4886336312563997e-06, "loss": 2.841, "step": 301090 }, { "epoch": 0.009216, "grad_norm": 0.9222050309181213, "learning_rate": 2.488102053630018e-06, "loss": 3.0007, "step": 301100 }, { "epoch": 0.0092416, "grad_norm": 1.0582964420318604, "learning_rate": 2.487570524716303e-06, "loss": 2.7386, "step": 301110 }, { "epoch": 0.0092672, "grad_norm": 0.8638980388641357, "learning_rate": 2.4870390445186953e-06, "loss": 2.883, "step": 301120 }, { "epoch": 0.0092928, "grad_norm": 0.7638611197471619, "learning_rate": 2.486507613040643e-06, "loss": 2.858, "step": 301130 }, { "epoch": 0.0093184, "grad_norm": 0.8560218811035156, "learning_rate": 2.485976230285595e-06, "loss": 2.797, "step": 301140 }, { "epoch": 0.009344, "grad_norm": 0.7914133667945862, "learning_rate": 2.485444896256991e-06, "loss": 2.5276, "step": 301150 }, { "epoch": 0.0093696, "grad_norm": 0.7842732667922974, "learning_rate": 2.4849136109582794e-06, "loss": 2.9833, "step": 301160 }, { "epoch": 0.0093952, "grad_norm": 0.8792722225189209, "learning_rate": 2.4843823743929065e-06, "loss": 2.7985, "step": 301170 }, { "epoch": 0.0094208, "grad_norm": 0.8612983822822571, "learning_rate": 2.4838511865643156e-06, "loss": 2.6888, "step": 301180 }, { "epoch": 0.0094464, "grad_norm": 0.8827971816062927, "learning_rate": 2.4833200474759533e-06, "loss": 2.8748, "step": 301190 }, { "epoch": 0.009472, "grad_norm": 0.8379100561141968, "learning_rate": 2.4827889571312624e-06, "loss": 2.88, "step": 301200 }, { "epoch": 0.0094976, "grad_norm": 0.8117287755012512, "learning_rate": 2.4822579155336903e-06, "loss": 2.9038, "step": 301210 }, { "epoch": 0.0095232, "grad_norm": 1.5092087984085083, "learning_rate": 2.4817269226866734e-06, "loss": 2.8365, "step": 301220 }, { "epoch": 0.0095488, "grad_norm": 0.8079792261123657, "learning_rate": 2.4811959785936613e-06, "loss": 2.8066, "step": 301230 }, { "epoch": 0.0095744, "grad_norm": 0.8122838735580444, "learning_rate": 2.480665083258094e-06, "loss": 2.9261, "step": 301240 }, { "epoch": 0.0096, "grad_norm": 0.8394253253936768, "learning_rate": 2.480134236683415e-06, "loss": 2.8514, "step": 301250 }, { "epoch": 0.0096256, "grad_norm": 0.7971473932266235, "learning_rate": 2.4796034388730673e-06, "loss": 2.8358, "step": 301260 }, { "epoch": 0.0096512, "grad_norm": 0.9298251867294312, "learning_rate": 2.479072689830493e-06, "loss": 2.8869, "step": 301270 }, { "epoch": 0.0096768, "grad_norm": 0.8351430892944336, "learning_rate": 2.478541989559132e-06, "loss": 2.8015, "step": 301280 }, { "epoch": 0.0097024, "grad_norm": 0.907769501209259, "learning_rate": 2.478011338062428e-06, "loss": 2.7473, "step": 301290 }, { "epoch": 0.009728, "grad_norm": 1.5464173555374146, "learning_rate": 2.477480735343822e-06, "loss": 2.6265, "step": 301300 }, { "epoch": 0.0097536, "grad_norm": 0.7915571331977844, "learning_rate": 2.4769501814067543e-06, "loss": 2.7448, "step": 301310 }, { "epoch": 0.0097792, "grad_norm": 0.8726547360420227, "learning_rate": 2.476419676254664e-06, "loss": 2.6838, "step": 301320 }, { "epoch": 0.0098048, "grad_norm": 0.8398445248603821, "learning_rate": 2.4758892198909946e-06, "loss": 2.8246, "step": 301330 }, { "epoch": 0.0098304, "grad_norm": 0.9562625288963318, "learning_rate": 2.475358812319185e-06, "loss": 3.2461, "step": 301340 }, { "epoch": 0.009856, "grad_norm": 0.8861681818962097, "learning_rate": 2.4748284535426713e-06, "loss": 2.6812, "step": 301350 }, { "epoch": 0.0098816, "grad_norm": 0.8467265367507935, "learning_rate": 2.474298143564896e-06, "loss": 2.8306, "step": 301360 }, { "epoch": 0.0099072, "grad_norm": 0.8192682862281799, "learning_rate": 2.4737678823892964e-06, "loss": 2.9155, "step": 301370 }, { "epoch": 0.0099328, "grad_norm": 1.8184388875961304, "learning_rate": 2.473237670019312e-06, "loss": 2.7944, "step": 301380 }, { "epoch": 0.0099584, "grad_norm": 0.8144351840019226, "learning_rate": 2.4727075064583807e-06, "loss": 2.6227, "step": 301390 }, { "epoch": 0.009984, "grad_norm": 0.7854577898979187, "learning_rate": 2.47217739170994e-06, "loss": 2.875, "step": 301400 }, { "epoch": 0.0100096, "grad_norm": 0.8677172660827637, "learning_rate": 2.471647325777432e-06, "loss": 2.682, "step": 301410 }, { "epoch": 0.0100352, "grad_norm": 0.8043450713157654, "learning_rate": 2.471117308664288e-06, "loss": 2.8858, "step": 301420 }, { "epoch": 0.0100608, "grad_norm": 0.8403698801994324, "learning_rate": 2.4705873403739468e-06, "loss": 2.7875, "step": 301430 }, { "epoch": 0.0100864, "grad_norm": 0.7821211814880371, "learning_rate": 2.470057420909845e-06, "loss": 2.8575, "step": 301440 }, { "epoch": 0.010112, "grad_norm": 0.8044882416725159, "learning_rate": 2.4695275502754166e-06, "loss": 2.762, "step": 301450 }, { "epoch": 0.0101376, "grad_norm": 0.76302570104599, "learning_rate": 2.4689977284741053e-06, "loss": 2.7386, "step": 301460 }, { "epoch": 0.0101632, "grad_norm": 0.7715283632278442, "learning_rate": 2.468467955509344e-06, "loss": 2.8202, "step": 301470 }, { "epoch": 0.0101888, "grad_norm": 0.97967129945755, "learning_rate": 2.4679382313845635e-06, "loss": 2.7248, "step": 301480 }, { "epoch": 0.0102144, "grad_norm": 1.057927131652832, "learning_rate": 2.4674085561032012e-06, "loss": 2.858, "step": 301490 }, { "epoch": 0.01024, "grad_norm": 0.931771993637085, "learning_rate": 2.4668789296686933e-06, "loss": 2.8235, "step": 301500 }, { "epoch": 0.0102656, "grad_norm": 0.9624946117401123, "learning_rate": 2.4663493520844728e-06, "loss": 2.9403, "step": 301510 }, { "epoch": 0.0102912, "grad_norm": 0.7457744479179382, "learning_rate": 2.4658198233539753e-06, "loss": 2.8753, "step": 301520 }, { "epoch": 0.0103168, "grad_norm": 0.8615983724594116, "learning_rate": 2.465290343480634e-06, "loss": 2.7636, "step": 301530 }, { "epoch": 0.0103424, "grad_norm": 0.8564039468765259, "learning_rate": 2.464760912467884e-06, "loss": 2.9997, "step": 301540 }, { "epoch": 0.010368, "grad_norm": 0.9308127164840698, "learning_rate": 2.4642315303191555e-06, "loss": 2.5243, "step": 301550 }, { "epoch": 0.0103936, "grad_norm": 0.8336106538772583, "learning_rate": 2.4637021970378815e-06, "loss": 2.8835, "step": 301560 }, { "epoch": 0.0104192, "grad_norm": 0.9738339781761169, "learning_rate": 2.463172912627496e-06, "loss": 3.0129, "step": 301570 }, { "epoch": 0.0104448, "grad_norm": 0.8688638210296631, "learning_rate": 2.462643677091432e-06, "loss": 2.7753, "step": 301580 }, { "epoch": 0.0104704, "grad_norm": 0.8236077427864075, "learning_rate": 2.4621144904331205e-06, "loss": 2.8044, "step": 301590 }, { "epoch": 0.010496, "grad_norm": 0.8817039728164673, "learning_rate": 2.461585352655993e-06, "loss": 2.9499, "step": 301600 }, { "epoch": 0.0105216, "grad_norm": 0.8696447610855103, "learning_rate": 2.4610562637634828e-06, "loss": 3.015, "step": 301610 }, { "epoch": 0.0105472, "grad_norm": 0.9440036416053772, "learning_rate": 2.460527223759017e-06, "loss": 2.7543, "step": 301620 }, { "epoch": 0.0105728, "grad_norm": 0.8294398188591003, "learning_rate": 2.4599982326460258e-06, "loss": 2.7135, "step": 301630 }, { "epoch": 0.0105984, "grad_norm": 0.7781044244766235, "learning_rate": 2.459469290427945e-06, "loss": 2.8469, "step": 301640 }, { "epoch": 0.010624, "grad_norm": 0.9194156527519226, "learning_rate": 2.4589403971082014e-06, "loss": 2.724, "step": 301650 }, { "epoch": 0.0106496, "grad_norm": 0.8029468655586243, "learning_rate": 2.458411552690225e-06, "loss": 2.8929, "step": 301660 }, { "epoch": 0.0106752, "grad_norm": 1.6214599609375, "learning_rate": 2.45788275717745e-06, "loss": 3.1356, "step": 301670 }, { "epoch": 0.0107008, "grad_norm": 0.8164225816726685, "learning_rate": 2.4573540105732963e-06, "loss": 2.7286, "step": 301680 }, { "epoch": 0.0107264, "grad_norm": 0.7676900625228882, "learning_rate": 2.456825312881197e-06, "loss": 2.8738, "step": 301690 }, { "epoch": 0.010752, "grad_norm": 0.8166685700416565, "learning_rate": 2.4562966641045817e-06, "loss": 3.0898, "step": 301700 }, { "epoch": 0.0107776, "grad_norm": 0.7504870891571045, "learning_rate": 2.455768064246877e-06, "loss": 2.8159, "step": 301710 }, { "epoch": 0.0108032, "grad_norm": 0.773739218711853, "learning_rate": 2.4552395133115128e-06, "loss": 2.7648, "step": 301720 }, { "epoch": 0.0108288, "grad_norm": 1.0265926122665405, "learning_rate": 2.4547110113019137e-06, "loss": 2.5614, "step": 301730 }, { "epoch": 0.0108544, "grad_norm": 0.811870276927948, "learning_rate": 2.4541825582215116e-06, "loss": 2.7727, "step": 301740 }, { "epoch": 0.01088, "grad_norm": 0.8842467069625854, "learning_rate": 2.453654154073728e-06, "loss": 2.7924, "step": 301750 }, { "epoch": 0.0109056, "grad_norm": 0.7475479245185852, "learning_rate": 2.4531257988619903e-06, "loss": 2.7587, "step": 301760 }, { "epoch": 0.0109312, "grad_norm": 0.831160306930542, "learning_rate": 2.4525974925897266e-06, "loss": 2.7637, "step": 301770 }, { "epoch": 0.0109568, "grad_norm": 0.8605748414993286, "learning_rate": 2.4520692352603624e-06, "loss": 2.8581, "step": 301780 }, { "epoch": 0.0109824, "grad_norm": 0.9350261688232422, "learning_rate": 2.4515410268773242e-06, "loss": 2.8177, "step": 301790 }, { "epoch": 0.011008, "grad_norm": 0.8467631340026855, "learning_rate": 2.4510128674440347e-06, "loss": 2.8628, "step": 301800 }, { "epoch": 0.0110336, "grad_norm": 0.8126344084739685, "learning_rate": 2.4504847569639212e-06, "loss": 2.7704, "step": 301810 }, { "epoch": 0.0110592, "grad_norm": 0.9786096215248108, "learning_rate": 2.4499566954404084e-06, "loss": 3.0199, "step": 301820 }, { "epoch": 0.0110848, "grad_norm": 0.8879704475402832, "learning_rate": 2.4494286828769187e-06, "loss": 2.8278, "step": 301830 }, { "epoch": 0.0111104, "grad_norm": 0.8663913607597351, "learning_rate": 2.448900719276878e-06, "loss": 2.8822, "step": 301840 }, { "epoch": 0.011136, "grad_norm": 0.8020291328430176, "learning_rate": 2.448372804643709e-06, "loss": 2.9292, "step": 301850 }, { "epoch": 0.0111616, "grad_norm": 0.8859816789627075, "learning_rate": 2.4478449389808356e-06, "loss": 2.9235, "step": 301860 }, { "epoch": 0.0111872, "grad_norm": 0.7747090458869934, "learning_rate": 2.4473171222916825e-06, "loss": 2.9378, "step": 301870 }, { "epoch": 0.0112128, "grad_norm": 0.9478242993354797, "learning_rate": 2.4467893545796697e-06, "loss": 2.7653, "step": 301880 }, { "epoch": 0.0112384, "grad_norm": 0.9412524104118347, "learning_rate": 2.4462616358482184e-06, "loss": 2.6854, "step": 301890 }, { "epoch": 0.011264, "grad_norm": 1.1799323558807373, "learning_rate": 2.445733966100754e-06, "loss": 2.7068, "step": 301900 }, { "epoch": 0.0112896, "grad_norm": 0.9361905455589294, "learning_rate": 2.445206345340697e-06, "loss": 2.6394, "step": 301910 }, { "epoch": 0.0113152, "grad_norm": 0.741963803768158, "learning_rate": 2.4446787735714695e-06, "loss": 2.7246, "step": 301920 }, { "epoch": 0.0113408, "grad_norm": 0.8185503482818604, "learning_rate": 2.4441512507964916e-06, "loss": 2.8064, "step": 301930 }, { "epoch": 0.0113664, "grad_norm": 0.8096637725830078, "learning_rate": 2.4436237770191874e-06, "loss": 3.1136, "step": 301940 }, { "epoch": 0.011392, "grad_norm": 0.762950599193573, "learning_rate": 2.4430963522429717e-06, "loss": 2.7736, "step": 301950 }, { "epoch": 0.0114176, "grad_norm": 0.8675627112388611, "learning_rate": 2.442568976471269e-06, "loss": 2.7672, "step": 301960 }, { "epoch": 0.0114432, "grad_norm": 0.7448605895042419, "learning_rate": 2.442041649707494e-06, "loss": 2.7535, "step": 301970 }, { "epoch": 0.0114688, "grad_norm": 0.8390241265296936, "learning_rate": 2.4415143719550737e-06, "loss": 2.7908, "step": 301980 }, { "epoch": 0.0114944, "grad_norm": 0.8232386112213135, "learning_rate": 2.4409871432174237e-06, "loss": 2.9679, "step": 301990 }, { "epoch": 0.01152, "grad_norm": 0.8103707432746887, "learning_rate": 2.4404599634979655e-06, "loss": 2.8144, "step": 302000 }, { "epoch": 0.0115456, "grad_norm": 0.8516865372657776, "learning_rate": 2.4399328328001126e-06, "loss": 3.0044, "step": 302010 }, { "epoch": 0.0115712, "grad_norm": 0.8052545189857483, "learning_rate": 2.439405751127287e-06, "loss": 2.8796, "step": 302020 }, { "epoch": 0.0115968, "grad_norm": 0.7577453851699829, "learning_rate": 2.4388787184829055e-06, "loss": 2.9411, "step": 302030 }, { "epoch": 0.0116224, "grad_norm": 0.8859206438064575, "learning_rate": 2.4383517348703855e-06, "loss": 2.6453, "step": 302040 }, { "epoch": 0.011648, "grad_norm": 0.8299705982208252, "learning_rate": 2.437824800293145e-06, "loss": 2.7955, "step": 302050 }, { "epoch": 0.0116736, "grad_norm": 0.9118079543113708, "learning_rate": 2.4372979147546007e-06, "loss": 3.0168, "step": 302060 }, { "epoch": 0.0116992, "grad_norm": 0.7925584316253662, "learning_rate": 2.4367710782581723e-06, "loss": 2.8911, "step": 302070 }, { "epoch": 0.0117248, "grad_norm": 0.7438564896583557, "learning_rate": 2.4362442908072703e-06, "loss": 2.9721, "step": 302080 }, { "epoch": 0.0117504, "grad_norm": 0.7324957847595215, "learning_rate": 2.4357175524053145e-06, "loss": 2.689, "step": 302090 }, { "epoch": 0.011776, "grad_norm": 1.0578151941299438, "learning_rate": 2.4351908630557195e-06, "loss": 2.7276, "step": 302100 }, { "epoch": 0.0118016, "grad_norm": 0.9081430435180664, "learning_rate": 2.434664222761902e-06, "loss": 2.8707, "step": 302110 }, { "epoch": 0.0118272, "grad_norm": 0.8526016473770142, "learning_rate": 2.434137631527276e-06, "loss": 2.9164, "step": 302120 }, { "epoch": 0.0118528, "grad_norm": 0.7856393456459045, "learning_rate": 2.433611089355257e-06, "loss": 2.8182, "step": 302130 }, { "epoch": 0.0118784, "grad_norm": 0.7888028025627136, "learning_rate": 2.4330845962492624e-06, "loss": 2.8746, "step": 302140 }, { "epoch": 0.011904, "grad_norm": 1.8212809562683105, "learning_rate": 2.4325581522126964e-06, "loss": 2.7651, "step": 302150 }, { "epoch": 0.0119296, "grad_norm": 0.8051630258560181, "learning_rate": 2.432031757248983e-06, "loss": 2.605, "step": 302160 }, { "epoch": 0.0119552, "grad_norm": 0.8991134762763977, "learning_rate": 2.431505411361531e-06, "loss": 2.9158, "step": 302170 }, { "epoch": 0.0119808, "grad_norm": 0.9178712368011475, "learning_rate": 2.4309791145537566e-06, "loss": 2.7993, "step": 302180 }, { "epoch": 0.0120064, "grad_norm": 0.8112744688987732, "learning_rate": 2.4304528668290705e-06, "loss": 2.6421, "step": 302190 }, { "epoch": 0.012032, "grad_norm": 0.8550772070884705, "learning_rate": 2.4299266681908886e-06, "loss": 2.7567, "step": 302200 }, { "epoch": 0.0120576, "grad_norm": 0.914293110370636, "learning_rate": 2.4294005186426184e-06, "loss": 2.8452, "step": 302210 }, { "epoch": 0.0120832, "grad_norm": 0.9469628930091858, "learning_rate": 2.4288744181876723e-06, "loss": 2.7904, "step": 302220 }, { "epoch": 0.0121088, "grad_norm": 0.8420214056968689, "learning_rate": 2.428348366829464e-06, "loss": 2.784, "step": 302230 }, { "epoch": 0.0121344, "grad_norm": 0.7865527272224426, "learning_rate": 2.427822364571405e-06, "loss": 2.8975, "step": 302240 }, { "epoch": 0.01216, "grad_norm": 0.7838844656944275, "learning_rate": 2.427296411416905e-06, "loss": 3.0112, "step": 302250 }, { "epoch": 0.0121856, "grad_norm": 0.9736966490745544, "learning_rate": 2.4267705073693746e-06, "loss": 2.8137, "step": 302260 }, { "epoch": 0.0122112, "grad_norm": 1.0243803262710571, "learning_rate": 2.426244652432228e-06, "loss": 2.8571, "step": 302270 }, { "epoch": 0.0122368, "grad_norm": 0.7819643616676331, "learning_rate": 2.42571884660887e-06, "loss": 2.8494, "step": 302280 }, { "epoch": 0.0122624, "grad_norm": 0.8665739297866821, "learning_rate": 2.425193089902711e-06, "loss": 2.8371, "step": 302290 }, { "epoch": 0.012288, "grad_norm": 0.9019653797149658, "learning_rate": 2.4246673823171616e-06, "loss": 2.8012, "step": 302300 }, { "epoch": 0.0123136, "grad_norm": 0.8217177391052246, "learning_rate": 2.424141723855632e-06, "loss": 2.8507, "step": 302310 }, { "epoch": 0.0123392, "grad_norm": 0.8179537057876587, "learning_rate": 2.42361611452153e-06, "loss": 2.7416, "step": 302320 }, { "epoch": 0.0123648, "grad_norm": 0.9087947010993958, "learning_rate": 2.4230905543182624e-06, "loss": 2.7566, "step": 302330 }, { "epoch": 0.0123904, "grad_norm": 0.8806582093238831, "learning_rate": 2.4225650432492397e-06, "loss": 2.9674, "step": 302340 }, { "epoch": 0.012416, "grad_norm": 0.9282866716384888, "learning_rate": 2.42203958131787e-06, "loss": 3.024, "step": 302350 }, { "epoch": 0.0124416, "grad_norm": 1.09755539894104, "learning_rate": 2.4215141685275577e-06, "loss": 2.8306, "step": 302360 }, { "epoch": 0.0124672, "grad_norm": 0.8557116389274597, "learning_rate": 2.4209888048817133e-06, "loss": 3.113, "step": 302370 }, { "epoch": 0.0124928, "grad_norm": 1.2925083637237549, "learning_rate": 2.420463490383742e-06, "loss": 3.0244, "step": 302380 }, { "epoch": 0.0125184, "grad_norm": 0.8299949765205383, "learning_rate": 2.4199382250370496e-06, "loss": 3.0802, "step": 302390 }, { "epoch": 0.012544, "grad_norm": 0.8413457274436951, "learning_rate": 2.4194130088450473e-06, "loss": 2.829, "step": 302400 }, { "epoch": 0.0125696, "grad_norm": 0.8377453088760376, "learning_rate": 2.418887841811134e-06, "loss": 2.6317, "step": 302410 }, { "epoch": 0.0125952, "grad_norm": 1.1491963863372803, "learning_rate": 2.4183627239387175e-06, "loss": 2.9111, "step": 302420 }, { "epoch": 0.0126208, "grad_norm": 0.7804598212242126, "learning_rate": 2.4178376552312043e-06, "loss": 2.6129, "step": 302430 }, { "epoch": 0.0126464, "grad_norm": 0.7681390643119812, "learning_rate": 2.4173126356919983e-06, "loss": 2.7438, "step": 302440 }, { "epoch": 0.012672, "grad_norm": 1.3715505599975586, "learning_rate": 2.416787665324505e-06, "loss": 2.7966, "step": 302450 }, { "epoch": 0.0126976, "grad_norm": 1.1814558506011963, "learning_rate": 2.4162627441321283e-06, "loss": 3.0256, "step": 302460 }, { "epoch": 0.0127232, "grad_norm": 0.8660000562667847, "learning_rate": 2.4157378721182743e-06, "loss": 2.8165, "step": 302470 }, { "epoch": 0.0127488, "grad_norm": 0.8275157809257507, "learning_rate": 2.4152130492863433e-06, "loss": 2.8851, "step": 302480 }, { "epoch": 0.0127744, "grad_norm": 1.5452975034713745, "learning_rate": 2.4146882756397384e-06, "loss": 2.8096, "step": 302490 }, { "epoch": 0.0128, "grad_norm": 0.7669238448143005, "learning_rate": 2.414163551181862e-06, "loss": 2.7432, "step": 302500 }, { "epoch": 0.0128256, "grad_norm": 1.4568326473236084, "learning_rate": 2.413638875916121e-06, "loss": 2.6755, "step": 302510 }, { "epoch": 0.0128512, "grad_norm": 0.828601062297821, "learning_rate": 2.4131142498459193e-06, "loss": 2.8139, "step": 302520 }, { "epoch": 0.0128768, "grad_norm": 0.7599974870681763, "learning_rate": 2.4125896729746522e-06, "loss": 2.9034, "step": 302530 }, { "epoch": 0.0129024, "grad_norm": 0.8737576007843018, "learning_rate": 2.4120651453057254e-06, "loss": 2.7547, "step": 302540 }, { "epoch": 0.012928, "grad_norm": 0.8810720443725586, "learning_rate": 2.411540666842539e-06, "loss": 2.8948, "step": 302550 }, { "epoch": 0.0129536, "grad_norm": 0.9249610900878906, "learning_rate": 2.4110162375884937e-06, "loss": 2.8211, "step": 302560 }, { "epoch": 0.0129792, "grad_norm": 1.0248620510101318, "learning_rate": 2.4104918575469915e-06, "loss": 2.9563, "step": 302570 }, { "epoch": 0.0130048, "grad_norm": 0.8092635869979858, "learning_rate": 2.409967526721434e-06, "loss": 2.8931, "step": 302580 }, { "epoch": 0.0130304, "grad_norm": 0.7823086380958557, "learning_rate": 2.4094432451152183e-06, "loss": 2.851, "step": 302590 }, { "epoch": 0.013056, "grad_norm": 0.7539958357810974, "learning_rate": 2.4089190127317495e-06, "loss": 2.7768, "step": 302600 }, { "epoch": 0.0130816, "grad_norm": 0.9550221562385559, "learning_rate": 2.40839482957442e-06, "loss": 2.7231, "step": 302610 }, { "epoch": 0.0131072, "grad_norm": 0.8342217206954956, "learning_rate": 2.4078706956466325e-06, "loss": 2.6812, "step": 302620 }, { "epoch": 0.0131328, "grad_norm": 0.7896385192871094, "learning_rate": 2.407346610951786e-06, "loss": 2.6674, "step": 302630 }, { "epoch": 0.0131584, "grad_norm": 0.8466648459434509, "learning_rate": 2.4068225754932796e-06, "loss": 2.6701, "step": 302640 }, { "epoch": 0.013184, "grad_norm": 0.8741053938865662, "learning_rate": 2.4062985892745094e-06, "loss": 2.7465, "step": 302650 }, { "epoch": 0.0132096, "grad_norm": 0.7405775785446167, "learning_rate": 2.4057746522988754e-06, "loss": 2.8258, "step": 302660 }, { "epoch": 0.0132352, "grad_norm": 0.7460402846336365, "learning_rate": 2.4052507645697743e-06, "loss": 2.6189, "step": 302670 }, { "epoch": 0.0132608, "grad_norm": 0.7881219387054443, "learning_rate": 2.404726926090604e-06, "loss": 2.9915, "step": 302680 }, { "epoch": 0.0132864, "grad_norm": 0.906624436378479, "learning_rate": 2.4042031368647602e-06, "loss": 2.855, "step": 302690 }, { "epoch": 0.013312, "grad_norm": 0.7813445329666138, "learning_rate": 2.4036793968956405e-06, "loss": 2.9605, "step": 302700 }, { "epoch": 0.0133376, "grad_norm": 0.975041925907135, "learning_rate": 2.4031557061866418e-06, "loss": 2.7261, "step": 302710 }, { "epoch": 0.0133632, "grad_norm": 0.8132044672966003, "learning_rate": 2.402632064741162e-06, "loss": 2.9177, "step": 302720 }, { "epoch": 0.0133888, "grad_norm": 0.9284433126449585, "learning_rate": 2.402108472562591e-06, "loss": 2.7425, "step": 302730 }, { "epoch": 0.0134144, "grad_norm": 0.7972506880760193, "learning_rate": 2.401584929654327e-06, "loss": 2.822, "step": 302740 }, { "epoch": 0.01344, "grad_norm": 0.9332258701324463, "learning_rate": 2.401061436019766e-06, "loss": 2.922, "step": 302750 }, { "epoch": 0.0134656, "grad_norm": 0.8122097253799438, "learning_rate": 2.4005379916623006e-06, "loss": 2.8825, "step": 302760 }, { "epoch": 0.0134912, "grad_norm": 0.8793337941169739, "learning_rate": 2.4000145965853285e-06, "loss": 2.7092, "step": 302770 }, { "epoch": 0.0135168, "grad_norm": 0.824438750743866, "learning_rate": 2.399491250792241e-06, "loss": 2.8318, "step": 302780 }, { "epoch": 0.0135424, "grad_norm": 0.7633980512619019, "learning_rate": 2.398967954286432e-06, "loss": 2.7684, "step": 302790 }, { "epoch": 0.013568, "grad_norm": 0.7617872953414917, "learning_rate": 2.3984447070713e-06, "loss": 2.9048, "step": 302800 }, { "epoch": 0.0135936, "grad_norm": 0.8816124796867371, "learning_rate": 2.3979215091502305e-06, "loss": 2.7225, "step": 302810 }, { "epoch": 0.0136192, "grad_norm": 0.7761578559875488, "learning_rate": 2.3973983605266195e-06, "loss": 2.9453, "step": 302820 }, { "epoch": 0.0136448, "grad_norm": 1.4248958826065063, "learning_rate": 2.3968752612038595e-06, "loss": 2.6965, "step": 302830 }, { "epoch": 0.0136704, "grad_norm": 0.9673916697502136, "learning_rate": 2.3963522111853397e-06, "loss": 2.6515, "step": 302840 }, { "epoch": 0.013696, "grad_norm": 0.8741997480392456, "learning_rate": 2.3958292104744607e-06, "loss": 2.8146, "step": 302850 }, { "epoch": 0.0137216, "grad_norm": 0.8626052737236023, "learning_rate": 2.3953062590746055e-06, "loss": 2.8458, "step": 302860 }, { "epoch": 0.0137472, "grad_norm": 0.7584639191627502, "learning_rate": 2.3947833569891686e-06, "loss": 2.6265, "step": 302870 }, { "epoch": 0.0137728, "grad_norm": 0.812524676322937, "learning_rate": 2.3942605042215406e-06, "loss": 2.8072, "step": 302880 }, { "epoch": 0.0137984, "grad_norm": 0.8723512887954712, "learning_rate": 2.393737700775112e-06, "loss": 2.9659, "step": 302890 }, { "epoch": 0.013824, "grad_norm": 0.955718457698822, "learning_rate": 2.3932149466532716e-06, "loss": 2.7603, "step": 302900 }, { "epoch": 0.0138496, "grad_norm": 0.8636563420295715, "learning_rate": 2.392692241859411e-06, "loss": 2.8455, "step": 302910 }, { "epoch": 0.0138752, "grad_norm": 0.8612207174301147, "learning_rate": 2.392169586396923e-06, "loss": 2.894, "step": 302920 }, { "epoch": 0.0139008, "grad_norm": 0.9224891662597656, "learning_rate": 2.39164698026919e-06, "loss": 2.8037, "step": 302930 }, { "epoch": 0.0139264, "grad_norm": 0.9546517133712769, "learning_rate": 2.391124423479605e-06, "loss": 2.8661, "step": 302940 }, { "epoch": 0.013952, "grad_norm": 0.8766918778419495, "learning_rate": 2.390601916031555e-06, "loss": 2.9063, "step": 302950 }, { "epoch": 0.0139776, "grad_norm": 0.812312662601471, "learning_rate": 2.390079457928429e-06, "loss": 2.6121, "step": 302960 }, { "epoch": 0.0140032, "grad_norm": 0.8163338303565979, "learning_rate": 2.3895570491736162e-06, "loss": 2.7137, "step": 302970 }, { "epoch": 0.0140288, "grad_norm": 0.8245619535446167, "learning_rate": 2.389034689770503e-06, "loss": 2.9487, "step": 302980 }, { "epoch": 0.0140544, "grad_norm": 0.7805649042129517, "learning_rate": 2.3885123797224795e-06, "loss": 2.6343, "step": 302990 }, { "epoch": 0.01408, "grad_norm": 0.7817071676254272, "learning_rate": 2.387990119032928e-06, "loss": 2.8722, "step": 303000 }, { "epoch": 0.0141056, "grad_norm": 0.8853601217269897, "learning_rate": 2.3874679077052377e-06, "loss": 2.8394, "step": 303010 }, { "epoch": 0.0141312, "grad_norm": 0.8922497630119324, "learning_rate": 2.3869457457427913e-06, "loss": 2.7894, "step": 303020 }, { "epoch": 0.0141568, "grad_norm": 0.7602832913398743, "learning_rate": 2.386423633148982e-06, "loss": 2.478, "step": 303030 }, { "epoch": 0.0141824, "grad_norm": 0.8054002523422241, "learning_rate": 2.3859015699271915e-06, "loss": 2.7987, "step": 303040 }, { "epoch": 0.014208, "grad_norm": 0.8608772158622742, "learning_rate": 2.385379556080809e-06, "loss": 2.8401, "step": 303050 }, { "epoch": 0.0142336, "grad_norm": 0.8336431384086609, "learning_rate": 2.384857591613213e-06, "loss": 2.8154, "step": 303060 }, { "epoch": 0.0142592, "grad_norm": 0.9089791178703308, "learning_rate": 2.3843356765277913e-06, "loss": 2.9721, "step": 303070 }, { "epoch": 0.0142848, "grad_norm": 0.8645799160003662, "learning_rate": 2.3838138108279297e-06, "loss": 2.8036, "step": 303080 }, { "epoch": 0.0143104, "grad_norm": 0.8529991507530212, "learning_rate": 2.3832919945170108e-06, "loss": 2.9328, "step": 303090 }, { "epoch": 0.014336, "grad_norm": 0.9249227046966553, "learning_rate": 2.38277022759842e-06, "loss": 2.7628, "step": 303100 }, { "epoch": 0.0143616, "grad_norm": 0.9959591627120972, "learning_rate": 2.3822485100755386e-06, "loss": 2.6945, "step": 303110 }, { "epoch": 0.0143872, "grad_norm": 0.7949106097221375, "learning_rate": 2.381726841951755e-06, "loss": 2.7964, "step": 303120 }, { "epoch": 0.0144128, "grad_norm": 1.645092248916626, "learning_rate": 2.381205223230444e-06, "loss": 2.8501, "step": 303130 }, { "epoch": 0.0144384, "grad_norm": 1.8905342817306519, "learning_rate": 2.3806836539149936e-06, "loss": 2.6674, "step": 303140 }, { "epoch": 0.014464, "grad_norm": 0.84709632396698, "learning_rate": 2.3801621340087843e-06, "loss": 2.8127, "step": 303150 }, { "epoch": 0.0144896, "grad_norm": 0.7773228883743286, "learning_rate": 2.379640663515198e-06, "loss": 2.8584, "step": 303160 }, { "epoch": 0.0145152, "grad_norm": 1.0699094533920288, "learning_rate": 2.3791192424376174e-06, "loss": 2.8042, "step": 303170 }, { "epoch": 0.0145408, "grad_norm": 1.9566969871520996, "learning_rate": 2.378597870779422e-06, "loss": 2.7211, "step": 303180 }, { "epoch": 0.0145664, "grad_norm": 1.0217190980911255, "learning_rate": 2.378076548543995e-06, "loss": 3.0011, "step": 303190 }, { "epoch": 0.014592, "grad_norm": 0.7813637852668762, "learning_rate": 2.3775552757347155e-06, "loss": 2.6838, "step": 303200 }, { "epoch": 0.0146176, "grad_norm": 0.991411566734314, "learning_rate": 2.377034052354964e-06, "loss": 3.0047, "step": 303210 }, { "epoch": 0.0146432, "grad_norm": 0.8124387264251709, "learning_rate": 2.3765128784081216e-06, "loss": 2.765, "step": 303220 }, { "epoch": 0.0146688, "grad_norm": 0.81075119972229, "learning_rate": 2.3759917538975663e-06, "loss": 2.776, "step": 303230 }, { "epoch": 0.0146944, "grad_norm": 0.9920710921287537, "learning_rate": 2.375470678826679e-06, "loss": 2.7539, "step": 303240 }, { "epoch": 0.01472, "grad_norm": 0.8038740158081055, "learning_rate": 2.3749496531988405e-06, "loss": 2.8325, "step": 303250 }, { "epoch": 0.0147456, "grad_norm": 0.803492546081543, "learning_rate": 2.374428677017423e-06, "loss": 2.8785, "step": 303260 }, { "epoch": 0.0147712, "grad_norm": 0.783430278301239, "learning_rate": 2.37390775028581e-06, "loss": 2.8023, "step": 303270 }, { "epoch": 0.0147968, "grad_norm": 0.8274337649345398, "learning_rate": 2.3733868730073782e-06, "loss": 2.829, "step": 303280 }, { "epoch": 0.0148224, "grad_norm": 0.8033906817436218, "learning_rate": 2.372866045185506e-06, "loss": 2.8757, "step": 303290 }, { "epoch": 0.014848, "grad_norm": 0.7864561080932617, "learning_rate": 2.3723452668235704e-06, "loss": 2.8057, "step": 303300 }, { "epoch": 0.0148736, "grad_norm": 0.8142267465591431, "learning_rate": 2.3718245379249484e-06, "loss": 2.9191, "step": 303310 }, { "epoch": 0.0148992, "grad_norm": 0.9021164774894714, "learning_rate": 2.371303858493019e-06, "loss": 2.8925, "step": 303320 }, { "epoch": 0.0149248, "grad_norm": 0.8969704508781433, "learning_rate": 2.370783228531155e-06, "loss": 2.9976, "step": 303330 }, { "epoch": 0.0149504, "grad_norm": 0.9768689274787903, "learning_rate": 2.3702626480427337e-06, "loss": 2.8112, "step": 303340 }, { "epoch": 0.014976, "grad_norm": 0.9303774237632751, "learning_rate": 2.369742117031132e-06, "loss": 3.0318, "step": 303350 }, { "epoch": 0.0150016, "grad_norm": 0.7744235992431641, "learning_rate": 2.3692216354997244e-06, "loss": 2.7392, "step": 303360 }, { "epoch": 0.0150272, "grad_norm": 0.9193218946456909, "learning_rate": 2.368701203451883e-06, "loss": 2.8109, "step": 303370 }, { "epoch": 0.0150528, "grad_norm": 0.9599656462669373, "learning_rate": 2.3681808208909918e-06, "loss": 2.7871, "step": 303380 }, { "epoch": 0.0150784, "grad_norm": 0.7916932106018066, "learning_rate": 2.367660487820418e-06, "loss": 2.9655, "step": 303390 }, { "epoch": 0.015104, "grad_norm": 0.8939862847328186, "learning_rate": 2.3671402042435367e-06, "loss": 2.9407, "step": 303400 }, { "epoch": 0.0151296, "grad_norm": 1.1953216791152954, "learning_rate": 2.3666199701637227e-06, "loss": 2.9776, "step": 303410 }, { "epoch": 0.0151552, "grad_norm": 1.2342865467071533, "learning_rate": 2.3660997855843495e-06, "loss": 2.9166, "step": 303420 }, { "epoch": 0.0151808, "grad_norm": 0.8706612586975098, "learning_rate": 2.36557965050879e-06, "loss": 2.912, "step": 303430 }, { "epoch": 0.0152064, "grad_norm": 0.8551040291786194, "learning_rate": 2.3650595649404174e-06, "loss": 3.0048, "step": 303440 }, { "epoch": 0.015232, "grad_norm": 0.8303683996200562, "learning_rate": 2.3645395288826077e-06, "loss": 2.974, "step": 303450 }, { "epoch": 0.0152576, "grad_norm": 0.8567084670066833, "learning_rate": 2.364019542338727e-06, "loss": 2.7892, "step": 303460 }, { "epoch": 0.0152832, "grad_norm": 0.7786106467247009, "learning_rate": 2.3634996053121508e-06, "loss": 2.7017, "step": 303470 }, { "epoch": 0.0153088, "grad_norm": 0.8441909551620483, "learning_rate": 2.3629797178062484e-06, "loss": 2.6033, "step": 303480 }, { "epoch": 0.0153344, "grad_norm": 0.9018298387527466, "learning_rate": 2.362459879824395e-06, "loss": 2.9093, "step": 303490 }, { "epoch": 0.01536, "grad_norm": 0.8466055989265442, "learning_rate": 2.3619400913699574e-06, "loss": 2.9223, "step": 303500 }, { "epoch": 0.0153856, "grad_norm": 0.8296029567718506, "learning_rate": 2.36142035244631e-06, "loss": 2.8413, "step": 303510 }, { "epoch": 0.0154112, "grad_norm": 0.9190402030944824, "learning_rate": 2.3609006630568233e-06, "loss": 2.67, "step": 303520 }, { "epoch": 0.0154368, "grad_norm": 0.8527680039405823, "learning_rate": 2.360381023204863e-06, "loss": 2.7315, "step": 303530 }, { "epoch": 0.0154624, "grad_norm": 0.8470622301101685, "learning_rate": 2.3598614328937984e-06, "loss": 2.8551, "step": 303540 }, { "epoch": 0.015488, "grad_norm": 0.7868828177452087, "learning_rate": 2.359341892127006e-06, "loss": 3.0067, "step": 303550 }, { "epoch": 0.0155136, "grad_norm": 0.8192375898361206, "learning_rate": 2.3588224009078496e-06, "loss": 2.8045, "step": 303560 }, { "epoch": 0.0155392, "grad_norm": 0.8228724002838135, "learning_rate": 2.358302959239699e-06, "loss": 2.7859, "step": 303570 }, { "epoch": 0.0155648, "grad_norm": 0.8998070955276489, "learning_rate": 2.357783567125926e-06, "loss": 2.9924, "step": 303580 }, { "epoch": 0.0155904, "grad_norm": 0.7627139091491699, "learning_rate": 2.3572642245698928e-06, "loss": 2.9626, "step": 303590 }, { "epoch": 0.015616, "grad_norm": 0.9286828637123108, "learning_rate": 2.35674493157497e-06, "loss": 2.7612, "step": 303600 }, { "epoch": 0.0156416, "grad_norm": 0.908798098564148, "learning_rate": 2.3562256881445246e-06, "loss": 2.8915, "step": 303610 }, { "epoch": 0.0156672, "grad_norm": 0.7730007171630859, "learning_rate": 2.355706494281924e-06, "loss": 2.7741, "step": 303620 }, { "epoch": 0.0156928, "grad_norm": 1.7959043979644775, "learning_rate": 2.3551873499905352e-06, "loss": 2.8351, "step": 303630 }, { "epoch": 0.0157184, "grad_norm": 0.7946288585662842, "learning_rate": 2.3546682552737253e-06, "loss": 2.7611, "step": 303640 }, { "epoch": 0.015744, "grad_norm": 1.0201178789138794, "learning_rate": 2.354149210134863e-06, "loss": 2.7938, "step": 303650 }, { "epoch": 0.0157696, "grad_norm": 0.7189042568206787, "learning_rate": 2.353630214577307e-06, "loss": 2.7012, "step": 303660 }, { "epoch": 0.0157952, "grad_norm": 0.8903312683105469, "learning_rate": 2.3531112686044276e-06, "loss": 2.9035, "step": 303670 }, { "epoch": 0.0158208, "grad_norm": 0.8180766701698303, "learning_rate": 2.35259237221959e-06, "loss": 2.7517, "step": 303680 }, { "epoch": 0.0158464, "grad_norm": 0.9208138585090637, "learning_rate": 2.3520735254261574e-06, "loss": 3.0301, "step": 303690 }, { "epoch": 0.015872, "grad_norm": 0.8151917457580566, "learning_rate": 2.3515547282274954e-06, "loss": 2.7661, "step": 303700 }, { "epoch": 0.0158976, "grad_norm": 0.7727264165878296, "learning_rate": 2.351035980626968e-06, "loss": 2.7296, "step": 303710 }, { "epoch": 0.0159232, "grad_norm": 0.7626520991325378, "learning_rate": 2.35051728262794e-06, "loss": 2.9229, "step": 303720 }, { "epoch": 0.0159488, "grad_norm": 0.778992772102356, "learning_rate": 2.349998634233774e-06, "loss": 2.9906, "step": 303730 }, { "epoch": 0.0159744, "grad_norm": 0.807991623878479, "learning_rate": 2.349480035447834e-06, "loss": 2.843, "step": 303740 }, { "epoch": 0.016, "grad_norm": 0.8440603017807007, "learning_rate": 2.3489614862734834e-06, "loss": 3.0886, "step": 303750 }, { "epoch": 0.0160256, "grad_norm": 0.8564655780792236, "learning_rate": 2.3484429867140836e-06, "loss": 2.7593, "step": 303760 }, { "epoch": 0.0160512, "grad_norm": 0.8043455481529236, "learning_rate": 2.3479245367729977e-06, "loss": 3.0112, "step": 303770 }, { "epoch": 0.0160768, "grad_norm": 1.1509612798690796, "learning_rate": 2.3474061364535895e-06, "loss": 2.5148, "step": 303780 }, { "epoch": 0.0161024, "grad_norm": 0.7911426424980164, "learning_rate": 2.3468877857592164e-06, "loss": 2.6581, "step": 303790 }, { "epoch": 0.016128, "grad_norm": 1.1024911403656006, "learning_rate": 2.3463694846932426e-06, "loss": 2.8066, "step": 303800 }, { "epoch": 0.0161536, "grad_norm": 0.7878000140190125, "learning_rate": 2.3458512332590276e-06, "loss": 2.6922, "step": 303810 }, { "epoch": 0.0161792, "grad_norm": 0.9222785830497742, "learning_rate": 2.345333031459934e-06, "loss": 2.7811, "step": 303820 }, { "epoch": 0.0162048, "grad_norm": 1.1181080341339111, "learning_rate": 2.344814879299321e-06, "loss": 2.7367, "step": 303830 }, { "epoch": 0.0162304, "grad_norm": 1.2182143926620483, "learning_rate": 2.3442967767805492e-06, "loss": 2.6973, "step": 303840 }, { "epoch": 0.016256, "grad_norm": 0.8682180643081665, "learning_rate": 2.3437787239069808e-06, "loss": 2.7032, "step": 303850 }, { "epoch": 0.0162816, "grad_norm": 0.9134208559989929, "learning_rate": 2.343260720681969e-06, "loss": 2.7874, "step": 303860 }, { "epoch": 0.0163072, "grad_norm": 0.8027850985527039, "learning_rate": 2.3427427671088775e-06, "loss": 2.9731, "step": 303870 }, { "epoch": 0.0163328, "grad_norm": 0.8985803127288818, "learning_rate": 2.342224863191064e-06, "loss": 2.822, "step": 303880 }, { "epoch": 0.0163584, "grad_norm": 0.7525594830513, "learning_rate": 2.341707008931884e-06, "loss": 2.7831, "step": 303890 }, { "epoch": 0.016384, "grad_norm": 1.0504541397094727, "learning_rate": 2.3411892043347016e-06, "loss": 2.9206, "step": 303900 }, { "epoch": 0.0164096, "grad_norm": 0.9540886282920837, "learning_rate": 2.3406714494028747e-06, "loss": 2.9128, "step": 303910 }, { "epoch": 0.0164352, "grad_norm": 0.8399786353111267, "learning_rate": 2.340153744139755e-06, "loss": 2.8724, "step": 303920 }, { "epoch": 0.0164608, "grad_norm": 0.8008467555046082, "learning_rate": 2.3396360885487025e-06, "loss": 2.9217, "step": 303930 }, { "epoch": 0.0164864, "grad_norm": 0.8134006857872009, "learning_rate": 2.339118482633074e-06, "loss": 2.9569, "step": 303940 }, { "epoch": 0.016512, "grad_norm": 1.099246859550476, "learning_rate": 2.338600926396226e-06, "loss": 2.8745, "step": 303950 }, { "epoch": 0.0165376, "grad_norm": 0.753678560256958, "learning_rate": 2.3380834198415147e-06, "loss": 2.763, "step": 303960 }, { "epoch": 0.0165632, "grad_norm": 0.8235310316085815, "learning_rate": 2.3375659629722967e-06, "loss": 2.7909, "step": 303970 }, { "epoch": 0.0165888, "grad_norm": 1.4958617687225342, "learning_rate": 2.33704855579193e-06, "loss": 2.7904, "step": 303980 }, { "epoch": 0.0166144, "grad_norm": 0.9160061478614807, "learning_rate": 2.3365311983037633e-06, "loss": 2.8617, "step": 303990 }, { "epoch": 0.01664, "grad_norm": 0.7592426538467407, "learning_rate": 2.336013890511155e-06, "loss": 2.8818, "step": 304000 }, { "epoch": 0.0166656, "grad_norm": 0.8750687837600708, "learning_rate": 2.3354966324174597e-06, "loss": 2.7511, "step": 304010 }, { "epoch": 0.0166912, "grad_norm": 0.8718820810317993, "learning_rate": 2.3349794240260327e-06, "loss": 2.6621, "step": 304020 }, { "epoch": 0.0167168, "grad_norm": 0.8266395926475525, "learning_rate": 2.334462265340226e-06, "loss": 2.9312, "step": 304030 }, { "epoch": 0.0167424, "grad_norm": 0.833737850189209, "learning_rate": 2.333945156363395e-06, "loss": 3.0588, "step": 304040 }, { "epoch": 0.016768, "grad_norm": 0.849247395992279, "learning_rate": 2.333428097098894e-06, "loss": 2.8429, "step": 304050 }, { "epoch": 0.0167936, "grad_norm": 0.7888103723526001, "learning_rate": 2.3329110875500706e-06, "loss": 2.9406, "step": 304060 }, { "epoch": 0.0168192, "grad_norm": 1.1501232385635376, "learning_rate": 2.3323941277202824e-06, "loss": 2.7382, "step": 304070 }, { "epoch": 0.0168448, "grad_norm": 0.7948228120803833, "learning_rate": 2.331877217612881e-06, "loss": 2.7961, "step": 304080 }, { "epoch": 0.0168704, "grad_norm": 0.8452805876731873, "learning_rate": 2.331360357231217e-06, "loss": 3.047, "step": 304090 }, { "epoch": 0.016896, "grad_norm": 0.839724063873291, "learning_rate": 2.330843546578647e-06, "loss": 2.7026, "step": 304100 }, { "epoch": 0.0169216, "grad_norm": 0.7964096069335938, "learning_rate": 2.3303267856585156e-06, "loss": 2.9926, "step": 304110 }, { "epoch": 0.0169472, "grad_norm": 0.8288465142250061, "learning_rate": 2.329810074474176e-06, "loss": 2.8195, "step": 304120 }, { "epoch": 0.0169728, "grad_norm": 0.9427013993263245, "learning_rate": 2.329293413028979e-06, "loss": 2.7896, "step": 304130 }, { "epoch": 0.0169984, "grad_norm": 0.8728042244911194, "learning_rate": 2.3287768013262767e-06, "loss": 2.7886, "step": 304140 }, { "epoch": 0.017024, "grad_norm": 0.9106610417366028, "learning_rate": 2.3282602393694174e-06, "loss": 2.8107, "step": 304150 }, { "epoch": 0.0170496, "grad_norm": 0.8078663349151611, "learning_rate": 2.327743727161752e-06, "loss": 2.8389, "step": 304160 }, { "epoch": 0.0170752, "grad_norm": 0.8940949440002441, "learning_rate": 2.3272272647066298e-06, "loss": 2.8302, "step": 304170 }, { "epoch": 0.0171008, "grad_norm": 0.8042044043540955, "learning_rate": 2.3267108520074023e-06, "loss": 2.9373, "step": 304180 }, { "epoch": 0.0171264, "grad_norm": 0.7901605367660522, "learning_rate": 2.3261944890674125e-06, "loss": 2.7652, "step": 304190 }, { "epoch": 0.017152, "grad_norm": 0.9490076303482056, "learning_rate": 2.3256781758900116e-06, "loss": 2.897, "step": 304200 }, { "epoch": 0.0171776, "grad_norm": 0.9854108095169067, "learning_rate": 2.3251619124785483e-06, "loss": 2.7963, "step": 304210 }, { "epoch": 0.0172032, "grad_norm": 0.843534529209137, "learning_rate": 2.3246456988363706e-06, "loss": 2.9077, "step": 304220 }, { "epoch": 0.0172288, "grad_norm": 0.9375280737876892, "learning_rate": 2.324129534966826e-06, "loss": 2.9692, "step": 304230 }, { "epoch": 0.0172544, "grad_norm": 0.7832844853401184, "learning_rate": 2.323613420873261e-06, "loss": 2.8553, "step": 304240 }, { "epoch": 0.01728, "grad_norm": 1.0449762344360352, "learning_rate": 2.323097356559022e-06, "loss": 2.5557, "step": 304250 }, { "epoch": 0.0173056, "grad_norm": 0.7667592763900757, "learning_rate": 2.322581342027457e-06, "loss": 2.8374, "step": 304260 }, { "epoch": 0.0173312, "grad_norm": 0.975049614906311, "learning_rate": 2.3220653772819123e-06, "loss": 2.7955, "step": 304270 }, { "epoch": 0.0173568, "grad_norm": 0.7933111786842346, "learning_rate": 2.321549462325732e-06, "loss": 2.859, "step": 304280 }, { "epoch": 0.0173824, "grad_norm": 0.9371060729026794, "learning_rate": 2.3210335971622632e-06, "loss": 2.8691, "step": 304290 }, { "epoch": 0.017408, "grad_norm": 0.8400282263755798, "learning_rate": 2.3205177817948533e-06, "loss": 2.904, "step": 304300 }, { "epoch": 0.0174336, "grad_norm": 0.8569563627243042, "learning_rate": 2.320002016226842e-06, "loss": 2.9519, "step": 304310 }, { "epoch": 0.0174592, "grad_norm": 0.8434574604034424, "learning_rate": 2.3194863004615765e-06, "loss": 2.8366, "step": 304320 }, { "epoch": 0.0174848, "grad_norm": 1.0517083406448364, "learning_rate": 2.318970634502402e-06, "loss": 2.8806, "step": 304330 }, { "epoch": 0.0175104, "grad_norm": 1.0010807514190674, "learning_rate": 2.31845501835266e-06, "loss": 2.8324, "step": 304340 }, { "epoch": 0.017536, "grad_norm": 0.7937012314796448, "learning_rate": 2.317939452015696e-06, "loss": 2.8894, "step": 304350 }, { "epoch": 0.0175616, "grad_norm": 1.0346581935882568, "learning_rate": 2.3174239354948546e-06, "loss": 2.9245, "step": 304360 }, { "epoch": 0.0175872, "grad_norm": 0.8240731358528137, "learning_rate": 2.316908468793475e-06, "loss": 2.8024, "step": 304370 }, { "epoch": 0.0176128, "grad_norm": 1.6809643507003784, "learning_rate": 2.316393051914907e-06, "loss": 2.981, "step": 304380 }, { "epoch": 0.0176384, "grad_norm": 0.866750955581665, "learning_rate": 2.315877684862483e-06, "loss": 2.8673, "step": 304390 }, { "epoch": 0.017664, "grad_norm": 0.743111789226532, "learning_rate": 2.315362367639552e-06, "loss": 2.9352, "step": 304400 }, { "epoch": 0.0176896, "grad_norm": 0.8018429279327393, "learning_rate": 2.31484710024945e-06, "loss": 2.8509, "step": 304410 }, { "epoch": 0.0177152, "grad_norm": 0.8946618437767029, "learning_rate": 2.314331882695524e-06, "loss": 2.8847, "step": 304420 }, { "epoch": 0.0177408, "grad_norm": 0.7916920781135559, "learning_rate": 2.3138167149811175e-06, "loss": 2.7377, "step": 304430 }, { "epoch": 0.0177664, "grad_norm": 0.8532619476318359, "learning_rate": 2.313301597109563e-06, "loss": 2.9515, "step": 304440 }, { "epoch": 0.017792, "grad_norm": 4.391141414642334, "learning_rate": 2.3127865290842055e-06, "loss": 2.8016, "step": 304450 }, { "epoch": 0.0178176, "grad_norm": 0.8109167218208313, "learning_rate": 2.3122715109083836e-06, "loss": 2.7436, "step": 304460 }, { "epoch": 0.0178432, "grad_norm": 1.4473849534988403, "learning_rate": 2.3117565425854384e-06, "loss": 2.7286, "step": 304470 }, { "epoch": 0.0178688, "grad_norm": 0.9124568104743958, "learning_rate": 2.3112416241187085e-06, "loss": 2.9472, "step": 304480 }, { "epoch": 0.0178944, "grad_norm": 0.735413670539856, "learning_rate": 2.3107267555115333e-06, "loss": 2.8777, "step": 304490 }, { "epoch": 0.01792, "grad_norm": 0.8115752935409546, "learning_rate": 2.3102119367672547e-06, "loss": 2.8995, "step": 304500 }, { "epoch": 0.0179456, "grad_norm": 0.8237410187721252, "learning_rate": 2.3096971678892043e-06, "loss": 2.7515, "step": 304510 }, { "epoch": 0.0179712, "grad_norm": 0.8663998246192932, "learning_rate": 2.3091824488807247e-06, "loss": 2.893, "step": 304520 }, { "epoch": 0.0179968, "grad_norm": 0.9617971777915955, "learning_rate": 2.308667779745153e-06, "loss": 2.8302, "step": 304530 }, { "epoch": 0.0180224, "grad_norm": 0.7958801984786987, "learning_rate": 2.308153160485825e-06, "loss": 2.918, "step": 304540 }, { "epoch": 0.018048, "grad_norm": 0.8280951976776123, "learning_rate": 2.3076385911060816e-06, "loss": 2.7991, "step": 304550 }, { "epoch": 0.0180736, "grad_norm": 0.9061957001686096, "learning_rate": 2.307124071609257e-06, "loss": 2.9558, "step": 304560 }, { "epoch": 0.0180992, "grad_norm": 0.8526367545127869, "learning_rate": 2.3066096019986883e-06, "loss": 2.8167, "step": 304570 }, { "epoch": 0.0181248, "grad_norm": 1.0637283325195312, "learning_rate": 2.306095182277711e-06, "loss": 2.8259, "step": 304580 }, { "epoch": 0.0181504, "grad_norm": 0.7702716588973999, "learning_rate": 2.3055808124496613e-06, "loss": 2.7482, "step": 304590 }, { "epoch": 0.018176, "grad_norm": 0.7851000428199768, "learning_rate": 2.3050664925178765e-06, "loss": 2.7698, "step": 304600 }, { "epoch": 0.0182016, "grad_norm": 0.8438388705253601, "learning_rate": 2.3045522224856896e-06, "loss": 2.9535, "step": 304610 }, { "epoch": 0.0182272, "grad_norm": 0.8938764333724976, "learning_rate": 2.304038002356437e-06, "loss": 2.9778, "step": 304620 }, { "epoch": 0.0182528, "grad_norm": 0.910706639289856, "learning_rate": 2.303523832133454e-06, "loss": 2.6226, "step": 304630 }, { "epoch": 0.0182784, "grad_norm": 0.9164566993713379, "learning_rate": 2.3030097118200723e-06, "loss": 2.8236, "step": 304640 }, { "epoch": 0.018304, "grad_norm": 0.8163110613822937, "learning_rate": 2.3024956414196266e-06, "loss": 2.7958, "step": 304650 }, { "epoch": 0.0183296, "grad_norm": 0.7705755829811096, "learning_rate": 2.3019816209354505e-06, "loss": 2.966, "step": 304660 }, { "epoch": 0.0183552, "grad_norm": 0.7948514819145203, "learning_rate": 2.3014676503708777e-06, "loss": 2.8777, "step": 304670 }, { "epoch": 0.0183808, "grad_norm": 0.8270175457000732, "learning_rate": 2.3009537297292406e-06, "loss": 2.8723, "step": 304680 }, { "epoch": 0.0184064, "grad_norm": 0.7757772207260132, "learning_rate": 2.3004398590138734e-06, "loss": 2.8038, "step": 304690 }, { "epoch": 0.018432, "grad_norm": 0.7889279127120972, "learning_rate": 2.2999260382281098e-06, "loss": 2.8663, "step": 304700 }, { "epoch": 0.0184576, "grad_norm": 0.8533197641372681, "learning_rate": 2.299412267375277e-06, "loss": 2.8188, "step": 304710 }, { "epoch": 0.0184832, "grad_norm": 5.269033908843994, "learning_rate": 2.2988985464587087e-06, "loss": 2.8544, "step": 304720 }, { "epoch": 0.0185088, "grad_norm": 0.8642373085021973, "learning_rate": 2.298384875481736e-06, "loss": 2.9562, "step": 304730 }, { "epoch": 0.0185344, "grad_norm": 0.7685231566429138, "learning_rate": 2.2978712544476922e-06, "loss": 2.9339, "step": 304740 }, { "epoch": 0.01856, "grad_norm": 0.7437686920166016, "learning_rate": 2.2973576833599053e-06, "loss": 2.6504, "step": 304750 }, { "epoch": 0.0185856, "grad_norm": 0.8577421307563782, "learning_rate": 2.2968441622217063e-06, "loss": 2.8561, "step": 304760 }, { "epoch": 0.0186112, "grad_norm": 0.7667102813720703, "learning_rate": 2.296330691036427e-06, "loss": 2.8515, "step": 304770 }, { "epoch": 0.0186368, "grad_norm": 0.8670727610588074, "learning_rate": 2.2958172698073943e-06, "loss": 2.8412, "step": 304780 }, { "epoch": 0.0186624, "grad_norm": 0.8371685743331909, "learning_rate": 2.2953038985379394e-06, "loss": 2.9508, "step": 304790 }, { "epoch": 0.018688, "grad_norm": 0.7771584987640381, "learning_rate": 2.2947905772313917e-06, "loss": 2.984, "step": 304800 }, { "epoch": 0.0187136, "grad_norm": 0.7456433176994324, "learning_rate": 2.294277305891078e-06, "loss": 2.8761, "step": 304810 }, { "epoch": 0.0187392, "grad_norm": 0.7681368589401245, "learning_rate": 2.2937640845203303e-06, "loss": 2.7123, "step": 304820 }, { "epoch": 0.0187648, "grad_norm": 0.8030473589897156, "learning_rate": 2.2932509131224757e-06, "loss": 2.9751, "step": 304830 }, { "epoch": 0.0187904, "grad_norm": 0.8519230484962463, "learning_rate": 2.2927377917008377e-06, "loss": 2.8714, "step": 304840 }, { "epoch": 0.018816, "grad_norm": 0.8423916101455688, "learning_rate": 2.292224720258748e-06, "loss": 2.6556, "step": 304850 }, { "epoch": 0.0188416, "grad_norm": 0.842002272605896, "learning_rate": 2.2917116987995312e-06, "loss": 3.0113, "step": 304860 }, { "epoch": 0.0188672, "grad_norm": 1.3329111337661743, "learning_rate": 2.291198727326516e-06, "loss": 2.9195, "step": 304870 }, { "epoch": 0.0188928, "grad_norm": 0.8343515992164612, "learning_rate": 2.290685805843028e-06, "loss": 2.9189, "step": 304880 }, { "epoch": 0.0189184, "grad_norm": 0.9033978581428528, "learning_rate": 2.2901729343523936e-06, "loss": 2.7401, "step": 304890 }, { "epoch": 0.018944, "grad_norm": 0.9218825101852417, "learning_rate": 2.289660112857942e-06, "loss": 2.8053, "step": 304900 }, { "epoch": 0.0189696, "grad_norm": 0.7876346707344055, "learning_rate": 2.289147341362993e-06, "loss": 2.8151, "step": 304910 }, { "epoch": 0.0189952, "grad_norm": 0.7729331254959106, "learning_rate": 2.288634619870873e-06, "loss": 2.9352, "step": 304920 }, { "epoch": 0.0190208, "grad_norm": 0.9323297739028931, "learning_rate": 2.288121948384906e-06, "loss": 2.8707, "step": 304930 }, { "epoch": 0.0190464, "grad_norm": 0.8339194059371948, "learning_rate": 2.287609326908421e-06, "loss": 2.8351, "step": 304940 }, { "epoch": 0.019072, "grad_norm": 0.7604584097862244, "learning_rate": 2.28709675544474e-06, "loss": 2.763, "step": 304950 }, { "epoch": 0.0190976, "grad_norm": 0.8282135725021362, "learning_rate": 2.28658423399719e-06, "loss": 2.9171, "step": 304960 }, { "epoch": 0.0191232, "grad_norm": 0.9397959113121033, "learning_rate": 2.2860717625690874e-06, "loss": 2.8945, "step": 304970 }, { "epoch": 0.0191488, "grad_norm": 0.7570102214813232, "learning_rate": 2.28555934116376e-06, "loss": 2.8467, "step": 304980 }, { "epoch": 0.0191744, "grad_norm": 0.8340005874633789, "learning_rate": 2.28504696978453e-06, "loss": 2.7411, "step": 304990 }, { "epoch": 0.0192, "grad_norm": 0.8290228843688965, "learning_rate": 2.2845346484347196e-06, "loss": 2.9122, "step": 305000 }, { "epoch": 0.0192256, "grad_norm": 0.8776277303695679, "learning_rate": 2.2840223771176527e-06, "loss": 2.8905, "step": 305010 }, { "epoch": 0.0192512, "grad_norm": 0.8243706226348877, "learning_rate": 2.283510155836649e-06, "loss": 2.8836, "step": 305020 }, { "epoch": 0.0192768, "grad_norm": 0.8888814449310303, "learning_rate": 2.2829979845950346e-06, "loss": 3.1341, "step": 305030 }, { "epoch": 0.0193024, "grad_norm": 1.0910584926605225, "learning_rate": 2.2824858633961232e-06, "loss": 3.0652, "step": 305040 }, { "epoch": 0.019328, "grad_norm": 0.8827381134033203, "learning_rate": 2.2819737922432405e-06, "loss": 3.0093, "step": 305050 }, { "epoch": 0.0193536, "grad_norm": 0.8851006627082825, "learning_rate": 2.281461771139707e-06, "loss": 2.567, "step": 305060 }, { "epoch": 0.0193792, "grad_norm": 0.8123062252998352, "learning_rate": 2.280949800088843e-06, "loss": 2.5797, "step": 305070 }, { "epoch": 0.0194048, "grad_norm": 1.4794970750808716, "learning_rate": 2.2804378790939674e-06, "loss": 2.9241, "step": 305080 }, { "epoch": 0.0194304, "grad_norm": 0.7702028751373291, "learning_rate": 2.2799260081583997e-06, "loss": 2.4708, "step": 305090 }, { "epoch": 0.019456, "grad_norm": 0.9061329364776611, "learning_rate": 2.2794141872854646e-06, "loss": 2.8871, "step": 305100 }, { "epoch": 0.0194816, "grad_norm": 0.7617142200469971, "learning_rate": 2.27890241647847e-06, "loss": 2.5943, "step": 305110 }, { "epoch": 0.0195072, "grad_norm": 0.7477439641952515, "learning_rate": 2.278390695740744e-06, "loss": 2.632, "step": 305120 }, { "epoch": 0.0195328, "grad_norm": 0.7767522931098938, "learning_rate": 2.2778790250756022e-06, "loss": 2.8067, "step": 305130 }, { "epoch": 0.0195584, "grad_norm": 0.8370915055274963, "learning_rate": 2.2773674044863624e-06, "loss": 2.7805, "step": 305140 }, { "epoch": 0.019584, "grad_norm": 0.8295653462409973, "learning_rate": 2.2768558339763435e-06, "loss": 2.5729, "step": 305150 }, { "epoch": 0.0196096, "grad_norm": 0.846396803855896, "learning_rate": 2.276344313548864e-06, "loss": 2.5795, "step": 305160 }, { "epoch": 0.0196352, "grad_norm": 0.8345592021942139, "learning_rate": 2.2758328432072365e-06, "loss": 2.4446, "step": 305170 }, { "epoch": 0.0196608, "grad_norm": 1.0048253536224365, "learning_rate": 2.2753214229547794e-06, "loss": 2.7776, "step": 305180 }, { "epoch": 0.0196864, "grad_norm": 0.7824702858924866, "learning_rate": 2.2748100527948115e-06, "loss": 2.6237, "step": 305190 }, { "epoch": 0.019712, "grad_norm": 0.7813878655433655, "learning_rate": 2.2742987327306456e-06, "loss": 2.4292, "step": 305200 }, { "epoch": 0.0197376, "grad_norm": 0.8628304600715637, "learning_rate": 2.2737874627656e-06, "loss": 2.7911, "step": 305210 }, { "epoch": 0.0197632, "grad_norm": 0.8729003071784973, "learning_rate": 2.2732762429029887e-06, "loss": 2.6132, "step": 305220 }, { "epoch": 0.0197888, "grad_norm": 0.9622942209243774, "learning_rate": 2.2727650731461313e-06, "loss": 2.6673, "step": 305230 }, { "epoch": 0.0198144, "grad_norm": 0.8837701082229614, "learning_rate": 2.2722539534983357e-06, "loss": 2.6299, "step": 305240 }, { "epoch": 0.01984, "grad_norm": 0.8622003793716431, "learning_rate": 2.2717428839629185e-06, "loss": 2.493, "step": 305250 }, { "epoch": 0.0198656, "grad_norm": 0.9734347462654114, "learning_rate": 2.2712318645431954e-06, "loss": 2.7722, "step": 305260 }, { "epoch": 0.0198912, "grad_norm": 0.7921314239501953, "learning_rate": 2.27072089524248e-06, "loss": 2.4179, "step": 305270 }, { "epoch": 0.0199168, "grad_norm": 0.8605177998542786, "learning_rate": 2.2702099760640815e-06, "loss": 2.678, "step": 305280 }, { "epoch": 0.0199424, "grad_norm": 0.9218520522117615, "learning_rate": 2.269699107011324e-06, "loss": 2.6096, "step": 305290 }, { "epoch": 0.019968, "grad_norm": 0.7999281883239746, "learning_rate": 2.269188288087509e-06, "loss": 2.5363, "step": 305300 }, { "epoch": 0.0199936, "grad_norm": 0.8057438731193542, "learning_rate": 2.2686775192959544e-06, "loss": 2.893, "step": 305310 }, { "epoch": 0.0200192, "grad_norm": 0.9843137860298157, "learning_rate": 2.2681668006399702e-06, "loss": 2.6188, "step": 305320 }, { "epoch": 0.0200448, "grad_norm": 0.8975504040718079, "learning_rate": 2.2676561321228706e-06, "loss": 2.6656, "step": 305330 }, { "epoch": 0.0200704, "grad_norm": 0.7740707397460938, "learning_rate": 2.2671455137479658e-06, "loss": 2.6765, "step": 305340 }, { "epoch": 0.020096, "grad_norm": 0.864345133304596, "learning_rate": 2.2666349455185676e-06, "loss": 2.5833, "step": 305350 }, { "epoch": 0.0201216, "grad_norm": 0.7995315194129944, "learning_rate": 2.266124427437988e-06, "loss": 2.6751, "step": 305360 }, { "epoch": 0.0201472, "grad_norm": 0.8845125436782837, "learning_rate": 2.265613959509534e-06, "loss": 2.6027, "step": 305370 }, { "epoch": 0.0201728, "grad_norm": 0.8385533094406128, "learning_rate": 2.2651035417365186e-06, "loss": 2.4908, "step": 305380 }, { "epoch": 0.0201984, "grad_norm": 1.0037826299667358, "learning_rate": 2.2645931741222505e-06, "loss": 2.5893, "step": 305390 }, { "epoch": 0.020224, "grad_norm": 0.8760501742362976, "learning_rate": 2.264082856670039e-06, "loss": 2.5119, "step": 305400 }, { "epoch": 0.0202496, "grad_norm": 0.9040547609329224, "learning_rate": 2.263572589383195e-06, "loss": 2.612, "step": 305410 }, { "epoch": 0.0202752, "grad_norm": 0.8223000764846802, "learning_rate": 2.2630623722650267e-06, "loss": 2.4517, "step": 305420 }, { "epoch": 0.0203008, "grad_norm": 0.8054200410842896, "learning_rate": 2.2625522053188455e-06, "loss": 2.5294, "step": 305430 }, { "epoch": 0.0203264, "grad_norm": 0.9081507325172424, "learning_rate": 2.2620420885479544e-06, "loss": 2.5873, "step": 305440 }, { "epoch": 0.020352, "grad_norm": 0.8818637728691101, "learning_rate": 2.26153202195566e-06, "loss": 2.6546, "step": 305450 }, { "epoch": 0.0203776, "grad_norm": 0.9045278429985046, "learning_rate": 2.2610220055452767e-06, "loss": 2.5965, "step": 305460 }, { "epoch": 0.0204032, "grad_norm": 0.8673487305641174, "learning_rate": 2.2605120393201098e-06, "loss": 2.5655, "step": 305470 }, { "epoch": 0.0204288, "grad_norm": 0.7952571511268616, "learning_rate": 2.2600021232834657e-06, "loss": 2.737, "step": 305480 }, { "epoch": 0.0204544, "grad_norm": 0.8528713583946228, "learning_rate": 2.2594922574386524e-06, "loss": 2.5471, "step": 305490 }, { "epoch": 0.02048, "grad_norm": 0.891934335231781, "learning_rate": 2.2589824417889727e-06, "loss": 2.6799, "step": 305500 }, { "epoch": 0.0205056, "grad_norm": 0.8275895714759827, "learning_rate": 2.258472676337734e-06, "loss": 2.7123, "step": 305510 }, { "epoch": 0.0205312, "grad_norm": 0.7502826452255249, "learning_rate": 2.2579629610882426e-06, "loss": 2.7924, "step": 305520 }, { "epoch": 0.0205568, "grad_norm": 0.8343455791473389, "learning_rate": 2.2574532960438043e-06, "loss": 2.8885, "step": 305530 }, { "epoch": 0.0205824, "grad_norm": 0.8881528973579407, "learning_rate": 2.256943681207723e-06, "loss": 2.5706, "step": 305540 }, { "epoch": 0.020608, "grad_norm": 0.9443485140800476, "learning_rate": 2.2564341165833048e-06, "loss": 2.6416, "step": 305550 }, { "epoch": 0.0206336, "grad_norm": 0.8092785477638245, "learning_rate": 2.2559246021738556e-06, "loss": 2.6745, "step": 305560 }, { "epoch": 0.0206592, "grad_norm": 0.880887508392334, "learning_rate": 2.255415137982674e-06, "loss": 2.7445, "step": 305570 }, { "epoch": 0.0206848, "grad_norm": 0.833032488822937, "learning_rate": 2.254905724013068e-06, "loss": 2.6233, "step": 305580 }, { "epoch": 0.0207104, "grad_norm": 0.8887847065925598, "learning_rate": 2.2543963602683395e-06, "loss": 2.6023, "step": 305590 }, { "epoch": 0.020736, "grad_norm": 0.8150045871734619, "learning_rate": 2.2538870467517917e-06, "loss": 2.7325, "step": 305600 }, { "epoch": 0.0207616, "grad_norm": 0.7754607796669006, "learning_rate": 2.253377783466728e-06, "loss": 2.5819, "step": 305610 }, { "epoch": 0.0207872, "grad_norm": 0.7587032318115234, "learning_rate": 2.252868570416451e-06, "loss": 2.571, "step": 305620 }, { "epoch": 0.0208128, "grad_norm": 0.9710516929626465, "learning_rate": 2.252359407604262e-06, "loss": 2.2146, "step": 305630 }, { "epoch": 0.0208384, "grad_norm": 0.8294857740402222, "learning_rate": 2.251850295033463e-06, "loss": 2.5158, "step": 305640 }, { "epoch": 0.020864, "grad_norm": 0.8426232933998108, "learning_rate": 2.251341232707356e-06, "loss": 2.6272, "step": 305650 }, { "epoch": 0.0208896, "grad_norm": 0.8286690711975098, "learning_rate": 2.2508322206292422e-06, "loss": 2.6077, "step": 305660 }, { "epoch": 0.0209152, "grad_norm": 0.8552141785621643, "learning_rate": 2.2503232588024216e-06, "loss": 2.9064, "step": 305670 }, { "epoch": 0.0209408, "grad_norm": 0.8416687846183777, "learning_rate": 2.249814347230195e-06, "loss": 2.5973, "step": 305680 }, { "epoch": 0.0209664, "grad_norm": 0.7978317737579346, "learning_rate": 2.2493054859158646e-06, "loss": 2.7657, "step": 305690 }, { "epoch": 0.020992, "grad_norm": 1.1323994398117065, "learning_rate": 2.2487966748627255e-06, "loss": 2.7816, "step": 305700 }, { "epoch": 2.56e-05, "grad_norm": 0.7935490608215332, "learning_rate": 2.2482879140740808e-06, "loss": 1.8794, "step": 305710 }, { "epoch": 5.12e-05, "grad_norm": 0.813640296459198, "learning_rate": 2.2477792035532285e-06, "loss": 2.5927, "step": 305720 }, { "epoch": 7.68e-05, "grad_norm": 0.8232895135879517, "learning_rate": 2.2472705433034682e-06, "loss": 2.5949, "step": 305730 }, { "epoch": 0.0001024, "grad_norm": 0.8561505079269409, "learning_rate": 2.2467619333280975e-06, "loss": 2.6601, "step": 305740 }, { "epoch": 0.000128, "grad_norm": 1.3754159212112427, "learning_rate": 2.246253373630415e-06, "loss": 3.0681, "step": 305750 }, { "epoch": 0.0001536, "grad_norm": 0.8403988480567932, "learning_rate": 2.2457448642137213e-06, "loss": 2.6265, "step": 305760 }, { "epoch": 0.0001792, "grad_norm": 0.9354833960533142, "learning_rate": 2.245236405081309e-06, "loss": 2.8441, "step": 305770 }, { "epoch": 0.0002048, "grad_norm": 0.8866755962371826, "learning_rate": 2.244727996236478e-06, "loss": 2.4801, "step": 305780 }, { "epoch": 0.0002304, "grad_norm": 0.9272379875183105, "learning_rate": 2.2442196376825243e-06, "loss": 2.6256, "step": 305790 }, { "epoch": 0.000256, "grad_norm": 1.09549081325531, "learning_rate": 2.243711329422742e-06, "loss": 2.7883, "step": 305800 }, { "epoch": 0.0002816, "grad_norm": 0.8838543891906738, "learning_rate": 2.2432030714604367e-06, "loss": 2.5119, "step": 305810 }, { "epoch": 0.0003072, "grad_norm": 0.7751644253730774, "learning_rate": 2.242694863798894e-06, "loss": 2.3476, "step": 305820 }, { "epoch": 0.0003328, "grad_norm": 0.9334970712661743, "learning_rate": 2.242186706441415e-06, "loss": 2.2655, "step": 305830 }, { "epoch": 0.0003584, "grad_norm": 0.9584482908248901, "learning_rate": 2.2416785993912924e-06, "loss": 3.0022, "step": 305840 }, { "epoch": 0.000384, "grad_norm": 0.839930534362793, "learning_rate": 2.2411705426518227e-06, "loss": 2.6235, "step": 305850 }, { "epoch": 0.0004096, "grad_norm": 0.9537939429283142, "learning_rate": 2.240662536226299e-06, "loss": 2.6464, "step": 305860 }, { "epoch": 0.0004352, "grad_norm": 0.8002469539642334, "learning_rate": 2.240154580118018e-06, "loss": 2.6083, "step": 305870 }, { "epoch": 0.0004608, "grad_norm": 0.7903920412063599, "learning_rate": 2.2396466743302715e-06, "loss": 2.6266, "step": 305880 }, { "epoch": 0.0004864, "grad_norm": 0.8443275690078735, "learning_rate": 2.2391388188663566e-06, "loss": 2.7125, "step": 305890 }, { "epoch": 0.000512, "grad_norm": 0.9104278087615967, "learning_rate": 2.2386310137295607e-06, "loss": 2.6827, "step": 305900 }, { "epoch": 0.0005376, "grad_norm": 0.8677157163619995, "learning_rate": 2.2381232589231805e-06, "loss": 2.7074, "step": 305910 }, { "epoch": 0.0005632, "grad_norm": 0.8322054743766785, "learning_rate": 2.2376155544505075e-06, "loss": 2.8082, "step": 305920 }, { "epoch": 0.0005888, "grad_norm": 0.8671451807022095, "learning_rate": 2.2371079003148354e-06, "loss": 2.4721, "step": 305930 }, { "epoch": 0.0006144, "grad_norm": 0.7313048243522644, "learning_rate": 2.2366002965194545e-06, "loss": 2.5898, "step": 305940 }, { "epoch": 0.00064, "grad_norm": 0.8625590205192566, "learning_rate": 2.236092743067658e-06, "loss": 2.6598, "step": 305950 }, { "epoch": 0.0006656, "grad_norm": 0.9325386881828308, "learning_rate": 2.23558523996274e-06, "loss": 2.793, "step": 305960 }, { "epoch": 0.0006912, "grad_norm": 2.518667697906494, "learning_rate": 2.2350777872079843e-06, "loss": 2.889, "step": 305970 }, { "epoch": 0.0007168, "grad_norm": 0.8360728025436401, "learning_rate": 2.2345703848066836e-06, "loss": 2.5689, "step": 305980 }, { "epoch": 0.0007424, "grad_norm": 1.3147832155227661, "learning_rate": 2.234063032762133e-06, "loss": 2.6927, "step": 305990 }, { "epoch": 0.000768, "grad_norm": 0.7930183410644531, "learning_rate": 2.233555731077619e-06, "loss": 2.7694, "step": 306000 }, { "epoch": 0.0007936, "grad_norm": 0.9580782055854797, "learning_rate": 2.2330484797564354e-06, "loss": 2.6073, "step": 306010 }, { "epoch": 0.0008192, "grad_norm": 0.7586010098457336, "learning_rate": 2.232541278801865e-06, "loss": 2.531, "step": 306020 }, { "epoch": 0.0008448, "grad_norm": 0.7565842866897583, "learning_rate": 2.2320341282171997e-06, "loss": 2.5083, "step": 306030 }, { "epoch": 0.0008704, "grad_norm": 0.7728313207626343, "learning_rate": 2.23152702800573e-06, "loss": 2.6155, "step": 306040 }, { "epoch": 0.000896, "grad_norm": 0.8370195627212524, "learning_rate": 2.231019978170741e-06, "loss": 2.6304, "step": 306050 }, { "epoch": 0.0009216, "grad_norm": 0.8291208148002625, "learning_rate": 2.2305129787155242e-06, "loss": 2.5748, "step": 306060 }, { "epoch": 0.0009472, "grad_norm": 0.7984004020690918, "learning_rate": 2.2300060296433656e-06, "loss": 2.4658, "step": 306070 }, { "epoch": 0.0009728, "grad_norm": 0.8571231961250305, "learning_rate": 2.2294991309575555e-06, "loss": 2.7141, "step": 306080 }, { "epoch": 0.0009984, "grad_norm": 0.8056670427322388, "learning_rate": 2.228992282661375e-06, "loss": 2.7587, "step": 306090 }, { "epoch": 0.001024, "grad_norm": 0.7996174693107605, "learning_rate": 2.228485484758115e-06, "loss": 2.651, "step": 306100 }, { "epoch": 0.0010496, "grad_norm": 0.8085668087005615, "learning_rate": 2.227978737251061e-06, "loss": 2.5678, "step": 306110 }, { "epoch": 0.0010752, "grad_norm": 1.0729118585586548, "learning_rate": 2.227472040143499e-06, "loss": 2.6172, "step": 306120 }, { "epoch": 0.0011008, "grad_norm": 0.7749377489089966, "learning_rate": 2.226965393438716e-06, "loss": 2.5947, "step": 306130 }, { "epoch": 0.0011264, "grad_norm": 0.7499796748161316, "learning_rate": 2.226458797139995e-06, "loss": 2.5695, "step": 306140 }, { "epoch": 0.001152, "grad_norm": 0.7574242353439331, "learning_rate": 2.2259522512506236e-06, "loss": 2.6898, "step": 306150 }, { "epoch": 0.0011776, "grad_norm": 1.0553817749023438, "learning_rate": 2.225445755773886e-06, "loss": 2.7296, "step": 306160 }, { "epoch": 0.0012032, "grad_norm": 0.8378596901893616, "learning_rate": 2.224939310713067e-06, "loss": 2.6619, "step": 306170 }, { "epoch": 0.0012288, "grad_norm": 0.9021861553192139, "learning_rate": 2.2244329160714496e-06, "loss": 2.5023, "step": 306180 }, { "epoch": 0.0012544, "grad_norm": 0.9303819537162781, "learning_rate": 2.2239265718523184e-06, "loss": 3.0564, "step": 306190 }, { "epoch": 0.00128, "grad_norm": 0.8003025054931641, "learning_rate": 2.2234202780589563e-06, "loss": 2.5856, "step": 306200 }, { "epoch": 0.0013056, "grad_norm": 0.7610201835632324, "learning_rate": 2.22291403469465e-06, "loss": 2.5197, "step": 306210 }, { "epoch": 0.0013312, "grad_norm": 0.8552246689796448, "learning_rate": 2.222407841762676e-06, "loss": 2.8256, "step": 306220 }, { "epoch": 0.0013568, "grad_norm": 0.8293060660362244, "learning_rate": 2.2219016992663213e-06, "loss": 2.7762, "step": 306230 }, { "epoch": 0.0013824, "grad_norm": 0.8373464345932007, "learning_rate": 2.221395607208866e-06, "loss": 2.4947, "step": 306240 }, { "epoch": 0.001408, "grad_norm": 0.8404194712638855, "learning_rate": 2.2208895655935926e-06, "loss": 2.667, "step": 306250 }, { "epoch": 0.0014336, "grad_norm": 0.8046707510948181, "learning_rate": 2.2203835744237836e-06, "loss": 2.5923, "step": 306260 }, { "epoch": 0.0014592, "grad_norm": 0.8401327729225159, "learning_rate": 2.2198776337027195e-06, "loss": 2.554, "step": 306270 }, { "epoch": 0.0014848, "grad_norm": 1.0973834991455078, "learning_rate": 2.219371743433684e-06, "loss": 2.6955, "step": 306280 }, { "epoch": 0.0015104, "grad_norm": 0.8685742020606995, "learning_rate": 2.2188659036199524e-06, "loss": 2.5778, "step": 306290 }, { "epoch": 0.001536, "grad_norm": 1.0004173517227173, "learning_rate": 2.218360114264807e-06, "loss": 1.8777, "step": 306300 }, { "epoch": 0.0015616, "grad_norm": 0.9666034579277039, "learning_rate": 2.2178543753715275e-06, "loss": 2.7582, "step": 306310 }, { "epoch": 0.0015872, "grad_norm": 0.8474709391593933, "learning_rate": 2.2173486869433924e-06, "loss": 2.8026, "step": 306320 }, { "epoch": 0.0016128, "grad_norm": 0.9283592104911804, "learning_rate": 2.2168430489836843e-06, "loss": 2.4037, "step": 306330 }, { "epoch": 0.0016384, "grad_norm": 0.9197515249252319, "learning_rate": 2.2163374614956846e-06, "loss": 2.7024, "step": 306340 }, { "epoch": 0.001664, "grad_norm": 0.8562582731246948, "learning_rate": 2.215831924482664e-06, "loss": 2.7111, "step": 306350 }, { "epoch": 0.0016896, "grad_norm": 0.9092533588409424, "learning_rate": 2.2153264379479055e-06, "loss": 2.6325, "step": 306360 }, { "epoch": 0.0017152, "grad_norm": 0.8647202849388123, "learning_rate": 2.214821001894686e-06, "loss": 2.7884, "step": 306370 }, { "epoch": 0.0017408, "grad_norm": 0.8329209685325623, "learning_rate": 2.2143156163262826e-06, "loss": 2.5354, "step": 306380 }, { "epoch": 0.0017664, "grad_norm": 0.8027743697166443, "learning_rate": 2.213810281245974e-06, "loss": 2.6604, "step": 306390 }, { "epoch": 0.001792, "grad_norm": 0.7837244272232056, "learning_rate": 2.2133049966570362e-06, "loss": 2.6051, "step": 306400 }, { "epoch": 2.56e-05, "grad_norm": 0.7972400784492493, "learning_rate": 2.2127997625627484e-06, "loss": 1.9391, "step": 306410 }, { "epoch": 5.12e-05, "grad_norm": 0.8495474457740784, "learning_rate": 2.212294578966382e-06, "loss": 2.6827, "step": 306420 }, { "epoch": 7.68e-05, "grad_norm": 0.8152188658714294, "learning_rate": 2.211789445871215e-06, "loss": 2.653, "step": 306430 }, { "epoch": 0.0001024, "grad_norm": 0.7806181311607361, "learning_rate": 2.2112843632805246e-06, "loss": 2.8504, "step": 306440 }, { "epoch": 0.000128, "grad_norm": 1.044926404953003, "learning_rate": 2.2107793311975846e-06, "loss": 2.6464, "step": 306450 }, { "epoch": 0.0001536, "grad_norm": 0.8322727680206299, "learning_rate": 2.2102743496256705e-06, "loss": 2.7651, "step": 306460 }, { "epoch": 0.0001792, "grad_norm": 0.8090723752975464, "learning_rate": 2.2097694185680574e-06, "loss": 2.7174, "step": 306470 }, { "epoch": 0.0002048, "grad_norm": 0.8458282947540283, "learning_rate": 2.209264538028022e-06, "loss": 2.9646, "step": 306480 }, { "epoch": 0.0002304, "grad_norm": 0.8271595239639282, "learning_rate": 2.208759708008832e-06, "loss": 2.6703, "step": 306490 }, { "epoch": 0.000256, "grad_norm": 0.8209246397018433, "learning_rate": 2.2082549285137624e-06, "loss": 2.6817, "step": 306500 }, { "epoch": 0.0002816, "grad_norm": 0.7946603298187256, "learning_rate": 2.2077501995460916e-06, "loss": 2.5089, "step": 306510 }, { "epoch": 0.0003072, "grad_norm": 0.785112202167511, "learning_rate": 2.2072455211090895e-06, "loss": 2.4813, "step": 306520 }, { "epoch": 0.0003328, "grad_norm": 0.8742390275001526, "learning_rate": 2.2067408932060285e-06, "loss": 2.4156, "step": 306530 }, { "epoch": 0.0003584, "grad_norm": 0.987567126750946, "learning_rate": 2.206236315840186e-06, "loss": 2.881, "step": 306540 }, { "epoch": 0.000384, "grad_norm": 0.8446251749992371, "learning_rate": 2.2057317890148255e-06, "loss": 2.5098, "step": 306550 }, { "epoch": 0.0004096, "grad_norm": 0.9530279040336609, "learning_rate": 2.2052273127332235e-06, "loss": 2.7607, "step": 306560 }, { "epoch": 0.0004352, "grad_norm": 0.8496828675270081, "learning_rate": 2.20472288699865e-06, "loss": 2.704, "step": 306570 }, { "epoch": 0.0004608, "grad_norm": 0.8542318344116211, "learning_rate": 2.204218511814379e-06, "loss": 2.7025, "step": 306580 }, { "epoch": 0.0004864, "grad_norm": 1.067937970161438, "learning_rate": 2.2037141871836766e-06, "loss": 2.7339, "step": 306590 }, { "epoch": 0.000512, "grad_norm": 0.7707007527351379, "learning_rate": 2.203209913109817e-06, "loss": 2.3464, "step": 306600 }, { "epoch": 0.0005376, "grad_norm": 0.8277918100357056, "learning_rate": 2.202705689596072e-06, "loss": 2.6565, "step": 306610 }, { "epoch": 0.0005632, "grad_norm": 0.7879815101623535, "learning_rate": 2.2022015166457045e-06, "loss": 2.6517, "step": 306620 }, { "epoch": 0.0005888, "grad_norm": 0.7456859350204468, "learning_rate": 2.201697394261988e-06, "loss": 2.4061, "step": 306630 }, { "epoch": 0.0006144, "grad_norm": 0.8051726818084717, "learning_rate": 2.2011933224481917e-06, "loss": 2.671, "step": 306640 }, { "epoch": 0.00064, "grad_norm": 0.9553385376930237, "learning_rate": 2.200689301207585e-06, "loss": 2.6751, "step": 306650 }, { "epoch": 0.0006656, "grad_norm": 0.8638625741004944, "learning_rate": 2.200185330543434e-06, "loss": 3.0342, "step": 306660 }, { "epoch": 0.0006912, "grad_norm": 0.8203372359275818, "learning_rate": 2.199681410459008e-06, "loss": 2.7522, "step": 306670 }, { "epoch": 0.0007168, "grad_norm": 1.327022671699524, "learning_rate": 2.1991775409575757e-06, "loss": 2.7035, "step": 306680 }, { "epoch": 0.0007424, "grad_norm": 0.9166933298110962, "learning_rate": 2.1986737220424037e-06, "loss": 2.7156, "step": 306690 }, { "epoch": 0.000768, "grad_norm": 0.8473147749900818, "learning_rate": 2.198169953716759e-06, "loss": 2.6684, "step": 306700 }, { "epoch": 0.0007936, "grad_norm": 0.8445137143135071, "learning_rate": 2.1976662359839086e-06, "loss": 2.6134, "step": 306710 }, { "epoch": 0.0008192, "grad_norm": 0.8519506454467773, "learning_rate": 2.19716256884712e-06, "loss": 2.1802, "step": 306720 }, { "epoch": 0.0008448, "grad_norm": 0.8338072896003723, "learning_rate": 2.1966589523096573e-06, "loss": 2.7911, "step": 306730 }, { "epoch": 0.0008704, "grad_norm": 0.9871999025344849, "learning_rate": 2.196155386374791e-06, "loss": 2.6719, "step": 306740 }, { "epoch": 0.000896, "grad_norm": 0.8610876798629761, "learning_rate": 2.1956518710457796e-06, "loss": 2.673, "step": 306750 }, { "epoch": 0.0009216, "grad_norm": 0.8156455159187317, "learning_rate": 2.195148406325892e-06, "loss": 2.508, "step": 306760 }, { "epoch": 0.0009472, "grad_norm": 0.7522208094596863, "learning_rate": 2.194644992218392e-06, "loss": 2.501, "step": 306770 }, { "epoch": 0.0009728, "grad_norm": 0.8752181529998779, "learning_rate": 2.194141628726546e-06, "loss": 2.6792, "step": 306780 }, { "epoch": 0.0009984, "grad_norm": 0.8978557586669922, "learning_rate": 2.1936383158536167e-06, "loss": 2.6189, "step": 306790 }, { "epoch": 0.001024, "grad_norm": 0.7818478345870972, "learning_rate": 2.1931350536028675e-06, "loss": 2.588, "step": 306800 }, { "epoch": 0.0010496, "grad_norm": 0.7869881987571716, "learning_rate": 2.192631841977566e-06, "loss": 2.6818, "step": 306810 }, { "epoch": 0.0010752, "grad_norm": 0.8079609870910645, "learning_rate": 2.1921286809809696e-06, "loss": 2.7354, "step": 306820 }, { "epoch": 0.0011008, "grad_norm": 0.8611968755722046, "learning_rate": 2.1916255706163424e-06, "loss": 2.6237, "step": 306830 }, { "epoch": 0.0011264, "grad_norm": 0.8547160625457764, "learning_rate": 2.1911225108869496e-06, "loss": 2.8116, "step": 306840 }, { "epoch": 0.001152, "grad_norm": 0.9015092253684998, "learning_rate": 2.1906195017960485e-06, "loss": 2.7589, "step": 306850 }, { "epoch": 0.0011776, "grad_norm": 0.8032556176185608, "learning_rate": 2.1901165433469074e-06, "loss": 2.6314, "step": 306860 }, { "epoch": 0.0012032, "grad_norm": 0.8578383326530457, "learning_rate": 2.1896136355427865e-06, "loss": 2.8927, "step": 306870 }, { "epoch": 0.0012288, "grad_norm": 0.9350579977035522, "learning_rate": 2.189110778386945e-06, "loss": 3.0581, "step": 306880 }, { "epoch": 0.0012544, "grad_norm": 0.8005237579345703, "learning_rate": 2.1886079718826425e-06, "loss": 2.685, "step": 306890 }, { "epoch": 0.00128, "grad_norm": 0.8762912154197693, "learning_rate": 2.1881052160331416e-06, "loss": 2.6252, "step": 306900 }, { "epoch": 0.0013056, "grad_norm": 0.8853867650032043, "learning_rate": 2.187602510841702e-06, "loss": 2.676, "step": 306910 }, { "epoch": 0.0013312, "grad_norm": 0.8284534215927124, "learning_rate": 2.187099856311584e-06, "loss": 2.5498, "step": 306920 }, { "epoch": 0.0013568, "grad_norm": 0.7831792831420898, "learning_rate": 2.1865972524460477e-06, "loss": 2.5201, "step": 306930 }, { "epoch": 0.0013824, "grad_norm": 0.8303688764572144, "learning_rate": 2.1860946992483535e-06, "loss": 2.7235, "step": 306940 }, { "epoch": 0.001408, "grad_norm": 0.8276473879814148, "learning_rate": 2.1855921967217563e-06, "loss": 2.5863, "step": 306950 }, { "epoch": 0.0014336, "grad_norm": 0.8659465312957764, "learning_rate": 2.1850897448695163e-06, "loss": 2.6607, "step": 306960 }, { "epoch": 0.0014592, "grad_norm": 0.8084693551063538, "learning_rate": 2.184587343694893e-06, "loss": 2.559, "step": 306970 }, { "epoch": 0.0014848, "grad_norm": 0.8156123757362366, "learning_rate": 2.1840849932011433e-06, "loss": 2.5708, "step": 306980 }, { "epoch": 0.0015104, "grad_norm": 1.1556283235549927, "learning_rate": 2.183582693391525e-06, "loss": 1.9075, "step": 306990 }, { "epoch": 0.001536, "grad_norm": 0.8305214047431946, "learning_rate": 2.1830804442692965e-06, "loss": 2.5634, "step": 307000 }, { "epoch": 0.0015616, "grad_norm": 0.8753586411476135, "learning_rate": 2.1825782458377154e-06, "loss": 2.7107, "step": 307010 }, { "epoch": 0.0015872, "grad_norm": 0.8453361392021179, "learning_rate": 2.182076098100032e-06, "loss": 2.7881, "step": 307020 }, { "epoch": 0.0016128, "grad_norm": 0.9633994698524475, "learning_rate": 2.18157400105951e-06, "loss": 2.8409, "step": 307030 }, { "epoch": 0.0016384, "grad_norm": 0.898193895816803, "learning_rate": 2.1810719547194027e-06, "loss": 2.7147, "step": 307040 }, { "epoch": 0.001664, "grad_norm": 0.818830132484436, "learning_rate": 2.1805699590829655e-06, "loss": 2.7365, "step": 307050 }, { "epoch": 0.0016896, "grad_norm": 1.108748197555542, "learning_rate": 2.1800680141534547e-06, "loss": 2.8911, "step": 307060 }, { "epoch": 0.0017152, "grad_norm": 0.907090961933136, "learning_rate": 2.179566119934127e-06, "loss": 2.8429, "step": 307070 }, { "epoch": 0.0017408, "grad_norm": 0.8558290600776672, "learning_rate": 2.179064276428231e-06, "loss": 2.8245, "step": 307080 }, { "epoch": 0.0017664, "grad_norm": 0.8843629360198975, "learning_rate": 2.178562483639025e-06, "loss": 2.6354, "step": 307090 }, { "epoch": 0.001792, "grad_norm": 0.9813207983970642, "learning_rate": 2.178060741569763e-06, "loss": 2.5306, "step": 307100 }, { "epoch": 0.0018176, "grad_norm": 0.8471429347991943, "learning_rate": 2.1775590502236976e-06, "loss": 2.5109, "step": 307110 }, { "epoch": 0.0018432, "grad_norm": 0.7912822365760803, "learning_rate": 2.1770574096040843e-06, "loss": 2.5756, "step": 307120 }, { "epoch": 0.0018688, "grad_norm": 0.8573095798492432, "learning_rate": 2.1765558197141733e-06, "loss": 2.5538, "step": 307130 }, { "epoch": 0.0018944, "grad_norm": 0.8799871802330017, "learning_rate": 2.1760542805572214e-06, "loss": 2.7896, "step": 307140 }, { "epoch": 0.00192, "grad_norm": 0.8644168972969055, "learning_rate": 2.175552792136476e-06, "loss": 2.5997, "step": 307150 }, { "epoch": 0.0019456, "grad_norm": 0.7858593463897705, "learning_rate": 2.175051354455191e-06, "loss": 2.6327, "step": 307160 }, { "epoch": 0.0019712, "grad_norm": 0.8081315159797668, "learning_rate": 2.1745499675166183e-06, "loss": 2.8236, "step": 307170 }, { "epoch": 0.0019968, "grad_norm": 1.0301790237426758, "learning_rate": 2.1740486313240085e-06, "loss": 2.7204, "step": 307180 }, { "epoch": 0.0020224, "grad_norm": 0.8936041593551636, "learning_rate": 2.173547345880611e-06, "loss": 2.6069, "step": 307190 }, { "epoch": 0.002048, "grad_norm": 0.8895581364631653, "learning_rate": 2.173046111189685e-06, "loss": 2.7366, "step": 307200 }, { "epoch": 0.0020736, "grad_norm": 0.7395418286323547, "learning_rate": 2.172544927254472e-06, "loss": 2.3859, "step": 307210 }, { "epoch": 0.0020992, "grad_norm": 0.9751519560813904, "learning_rate": 2.1720437940782247e-06, "loss": 2.9066, "step": 307220 }, { "epoch": 0.0021248, "grad_norm": 1.036832571029663, "learning_rate": 2.1715427116641927e-06, "loss": 2.9306, "step": 307230 }, { "epoch": 0.0021504, "grad_norm": 0.9049229621887207, "learning_rate": 2.1710416800156265e-06, "loss": 2.8985, "step": 307240 }, { "epoch": 0.002176, "grad_norm": 0.870226263999939, "learning_rate": 2.1705406991357736e-06, "loss": 2.5685, "step": 307250 }, { "epoch": 0.0022016, "grad_norm": 0.7506048083305359, "learning_rate": 2.170039769027884e-06, "loss": 2.7107, "step": 307260 }, { "epoch": 0.0022272, "grad_norm": 0.8953083753585815, "learning_rate": 2.1695388896952075e-06, "loss": 2.8275, "step": 307270 }, { "epoch": 0.0022528, "grad_norm": 0.9594444036483765, "learning_rate": 2.169038061140989e-06, "loss": 3.0972, "step": 307280 }, { "epoch": 0.0022784, "grad_norm": 0.8684727549552917, "learning_rate": 2.1685372833684757e-06, "loss": 2.5888, "step": 307290 }, { "epoch": 0.002304, "grad_norm": 0.7885283827781677, "learning_rate": 2.168036556380917e-06, "loss": 2.3437, "step": 307300 }, { "epoch": 0.0023296, "grad_norm": 0.9178683757781982, "learning_rate": 2.1675358801815604e-06, "loss": 2.7066, "step": 307310 }, { "epoch": 0.0023552, "grad_norm": 0.8672012686729431, "learning_rate": 2.167035254773652e-06, "loss": 2.5089, "step": 307320 }, { "epoch": 0.0023808, "grad_norm": 0.9136287569999695, "learning_rate": 2.166534680160438e-06, "loss": 2.862, "step": 307330 }, { "epoch": 0.0024064, "grad_norm": 0.7468648552894592, "learning_rate": 2.166034156345167e-06, "loss": 2.6816, "step": 307340 }, { "epoch": 0.002432, "grad_norm": 1.0003620386123657, "learning_rate": 2.1655336833310804e-06, "loss": 2.9633, "step": 307350 }, { "epoch": 0.0024576, "grad_norm": 0.8704414367675781, "learning_rate": 2.1650332611214254e-06, "loss": 2.8852, "step": 307360 }, { "epoch": 0.0024832, "grad_norm": 0.8176807761192322, "learning_rate": 2.164532889719445e-06, "loss": 2.6637, "step": 307370 }, { "epoch": 0.0025088, "grad_norm": 0.8690260648727417, "learning_rate": 2.1640325691283893e-06, "loss": 2.8634, "step": 307380 }, { "epoch": 0.0025344, "grad_norm": 0.7552016973495483, "learning_rate": 2.163532299351502e-06, "loss": 2.7444, "step": 307390 }, { "epoch": 0.00256, "grad_norm": 0.8271613717079163, "learning_rate": 2.1630320803920223e-06, "loss": 2.611, "step": 307400 }, { "epoch": 0.0025856, "grad_norm": 1.0544668436050415, "learning_rate": 2.1625319122531974e-06, "loss": 2.7263, "step": 307410 }, { "epoch": 0.0026112, "grad_norm": 0.9620754718780518, "learning_rate": 2.1620317949382696e-06, "loss": 2.5669, "step": 307420 }, { "epoch": 0.0026368, "grad_norm": 0.7926899194717407, "learning_rate": 2.161531728450482e-06, "loss": 2.7412, "step": 307430 }, { "epoch": 0.0026624, "grad_norm": 1.8824554681777954, "learning_rate": 2.161031712793079e-06, "loss": 2.8962, "step": 307440 }, { "epoch": 0.002688, "grad_norm": 0.8516499400138855, "learning_rate": 2.1605317479693013e-06, "loss": 2.6889, "step": 307450 }, { "epoch": 0.0027136, "grad_norm": 0.9561299085617065, "learning_rate": 2.1600318339823924e-06, "loss": 2.8198, "step": 307460 }, { "epoch": 0.0027392, "grad_norm": 0.801750898361206, "learning_rate": 2.1595319708355954e-06, "loss": 2.8514, "step": 307470 }, { "epoch": 0.0027648, "grad_norm": 0.8452319502830505, "learning_rate": 2.159032158532147e-06, "loss": 2.9979, "step": 307480 }, { "epoch": 0.0027904, "grad_norm": 0.8120666146278381, "learning_rate": 2.1585323970752913e-06, "loss": 2.8509, "step": 307490 }, { "epoch": 0.002816, "grad_norm": 0.8399400115013123, "learning_rate": 2.15803268646827e-06, "loss": 2.8197, "step": 307500 }, { "epoch": 0.0028416, "grad_norm": 0.7612883448600769, "learning_rate": 2.1575330267143213e-06, "loss": 2.6175, "step": 307510 }, { "epoch": 0.0028672, "grad_norm": 0.7905834913253784, "learning_rate": 2.1570334178166874e-06, "loss": 2.7607, "step": 307520 }, { "epoch": 0.0028928, "grad_norm": 0.7721942067146301, "learning_rate": 2.156533859778607e-06, "loss": 2.8615, "step": 307530 }, { "epoch": 0.0029184, "grad_norm": 0.8292202949523926, "learning_rate": 2.1560343526033203e-06, "loss": 2.872, "step": 307540 }, { "epoch": 0.002944, "grad_norm": 0.8652806878089905, "learning_rate": 2.155534896294066e-06, "loss": 2.7861, "step": 307550 }, { "epoch": 0.0029696, "grad_norm": 0.8359740972518921, "learning_rate": 2.1550354908540827e-06, "loss": 3.1727, "step": 307560 }, { "epoch": 0.0029952, "grad_norm": 0.836606502532959, "learning_rate": 2.15453613628661e-06, "loss": 3.0323, "step": 307570 }, { "epoch": 0.0030208, "grad_norm": 0.8408666849136353, "learning_rate": 2.1540368325948846e-06, "loss": 2.799, "step": 307580 }, { "epoch": 0.0030464, "grad_norm": 0.9380550384521484, "learning_rate": 2.153537579782149e-06, "loss": 3.0387, "step": 307590 }, { "epoch": 0.003072, "grad_norm": 0.9616863131523132, "learning_rate": 2.153038377851633e-06, "loss": 2.7531, "step": 307600 }, { "epoch": 0.0030976, "grad_norm": 0.8318946361541748, "learning_rate": 2.152539226806578e-06, "loss": 2.7481, "step": 307610 }, { "epoch": 0.0031232, "grad_norm": 0.8074041604995728, "learning_rate": 2.152040126650221e-06, "loss": 2.6628, "step": 307620 }, { "epoch": 0.0031488, "grad_norm": 0.8333682417869568, "learning_rate": 2.151541077385798e-06, "loss": 2.7724, "step": 307630 }, { "epoch": 0.0031744, "grad_norm": 0.7739410996437073, "learning_rate": 2.1510420790165443e-06, "loss": 2.6347, "step": 307640 }, { "epoch": 0.0032, "grad_norm": 0.9470269083976746, "learning_rate": 2.150543131545697e-06, "loss": 2.8868, "step": 307650 }, { "epoch": 0.0032256, "grad_norm": 0.7849037647247314, "learning_rate": 2.1500442349764917e-06, "loss": 2.9066, "step": 307660 }, { "epoch": 0.0032512, "grad_norm": 0.8298385739326477, "learning_rate": 2.1495453893121665e-06, "loss": 2.8722, "step": 307670 }, { "epoch": 0.0032768, "grad_norm": 0.801356315612793, "learning_rate": 2.149046594555949e-06, "loss": 2.6623, "step": 307680 }, { "epoch": 0.0033024, "grad_norm": 0.9910181760787964, "learning_rate": 2.148547850711079e-06, "loss": 2.4313, "step": 307690 }, { "epoch": 0.003328, "grad_norm": 0.8420060276985168, "learning_rate": 2.1480491577807894e-06, "loss": 2.7581, "step": 307700 }, { "epoch": 0.0033536, "grad_norm": 0.8511375188827515, "learning_rate": 2.147550515768314e-06, "loss": 3.0158, "step": 307710 }, { "epoch": 0.0033792, "grad_norm": 0.8240973949432373, "learning_rate": 2.1470519246768863e-06, "loss": 2.7576, "step": 307720 }, { "epoch": 0.0034048, "grad_norm": 0.8099158406257629, "learning_rate": 2.1465533845097396e-06, "loss": 2.664, "step": 307730 }, { "epoch": 0.0034304, "grad_norm": 0.9133990406990051, "learning_rate": 2.146054895270108e-06, "loss": 2.9276, "step": 307740 }, { "epoch": 0.003456, "grad_norm": 0.8531157970428467, "learning_rate": 2.1455564569612217e-06, "loss": 2.8717, "step": 307750 }, { "epoch": 0.0034816, "grad_norm": 0.9967886209487915, "learning_rate": 2.1450580695863156e-06, "loss": 2.7025, "step": 307760 }, { "epoch": 0.0035072, "grad_norm": 0.7708924412727356, "learning_rate": 2.14455973314862e-06, "loss": 2.7282, "step": 307770 }, { "epoch": 0.0035328, "grad_norm": 0.7634226083755493, "learning_rate": 2.1440614476513666e-06, "loss": 2.5967, "step": 307780 }, { "epoch": 0.0035584, "grad_norm": 0.8586792349815369, "learning_rate": 2.143563213097789e-06, "loss": 2.6495, "step": 307790 }, { "epoch": 0.003584, "grad_norm": 0.7954689264297485, "learning_rate": 2.143065029491114e-06, "loss": 2.8699, "step": 307800 }, { "epoch": 0.0036096, "grad_norm": 0.8706931471824646, "learning_rate": 2.1425668968345737e-06, "loss": 2.7639, "step": 307810 }, { "epoch": 0.0036352, "grad_norm": 0.8436150550842285, "learning_rate": 2.1420688151313996e-06, "loss": 2.9506, "step": 307820 }, { "epoch": 0.0036608, "grad_norm": 0.8434830904006958, "learning_rate": 2.14157078438482e-06, "loss": 2.8682, "step": 307830 }, { "epoch": 0.0036864, "grad_norm": 0.8467372059822083, "learning_rate": 2.1410728045980656e-06, "loss": 2.7778, "step": 307840 }, { "epoch": 0.003712, "grad_norm": 0.8489021062850952, "learning_rate": 2.140574875774366e-06, "loss": 2.7731, "step": 307850 }, { "epoch": 0.0037376, "grad_norm": 0.8053025603294373, "learning_rate": 2.140076997916949e-06, "loss": 2.7906, "step": 307860 }, { "epoch": 0.0037632, "grad_norm": 0.9384624361991882, "learning_rate": 2.139579171029047e-06, "loss": 2.9775, "step": 307870 }, { "epoch": 0.0037888, "grad_norm": 0.9445421695709229, "learning_rate": 2.139081395113881e-06, "loss": 2.7461, "step": 307880 }, { "epoch": 0.0038144, "grad_norm": 0.9102444052696228, "learning_rate": 2.138583670174681e-06, "loss": 2.7128, "step": 307890 }, { "epoch": 0.00384, "grad_norm": 0.8139195442199707, "learning_rate": 2.1380859962146796e-06, "loss": 2.6932, "step": 307900 }, { "epoch": 0.0038656, "grad_norm": 0.7412728071212769, "learning_rate": 2.1375883732371016e-06, "loss": 2.6677, "step": 307910 }, { "epoch": 0.0038912, "grad_norm": 0.9482025504112244, "learning_rate": 2.137090801245175e-06, "loss": 2.7716, "step": 307920 }, { "epoch": 0.0039168, "grad_norm": 0.8049587607383728, "learning_rate": 2.136593280242123e-06, "loss": 3.0515, "step": 307930 }, { "epoch": 0.0039424, "grad_norm": 0.7925137877464294, "learning_rate": 2.1360958102311722e-06, "loss": 2.6287, "step": 307940 }, { "epoch": 0.003968, "grad_norm": 0.7698922753334045, "learning_rate": 2.135598391215552e-06, "loss": 2.7785, "step": 307950 }, { "epoch": 0.0039936, "grad_norm": 0.8221195936203003, "learning_rate": 2.135101023198485e-06, "loss": 2.7517, "step": 307960 }, { "epoch": 0.0040192, "grad_norm": 0.7862655520439148, "learning_rate": 2.134603706183198e-06, "loss": 2.8535, "step": 307970 }, { "epoch": 0.0040448, "grad_norm": 0.8753225803375244, "learning_rate": 2.134106440172915e-06, "loss": 2.6983, "step": 307980 }, { "epoch": 0.0040704, "grad_norm": 1.220206618309021, "learning_rate": 2.1336092251708653e-06, "loss": 2.8713, "step": 307990 }, { "epoch": 0.004096, "grad_norm": 0.7753434777259827, "learning_rate": 2.1331120611802647e-06, "loss": 2.9469, "step": 308000 }, { "epoch": 0.0041216, "grad_norm": 1.2075120210647583, "learning_rate": 2.132614948204342e-06, "loss": 2.8796, "step": 308010 }, { "epoch": 0.0041472, "grad_norm": 0.8491533994674683, "learning_rate": 2.132117886246322e-06, "loss": 2.7489, "step": 308020 }, { "epoch": 0.0041728, "grad_norm": 0.8197646737098694, "learning_rate": 2.1316208753094246e-06, "loss": 2.7302, "step": 308030 }, { "epoch": 0.0041984, "grad_norm": 0.9609388113021851, "learning_rate": 2.1311239153968767e-06, "loss": 2.8147, "step": 308040 }, { "epoch": 0.004224, "grad_norm": 0.7962812185287476, "learning_rate": 2.130627006511897e-06, "loss": 2.7334, "step": 308050 }, { "epoch": 0.0042496, "grad_norm": 0.8159759044647217, "learning_rate": 2.1301301486577107e-06, "loss": 2.8951, "step": 308060 }, { "epoch": 0.0042752, "grad_norm": 0.7691927552223206, "learning_rate": 2.129633341837538e-06, "loss": 2.7053, "step": 308070 }, { "epoch": 0.0043008, "grad_norm": 1.0359530448913574, "learning_rate": 2.129136586054602e-06, "loss": 2.9958, "step": 308080 }, { "epoch": 0.0043264, "grad_norm": 0.9029603600502014, "learning_rate": 2.1286398813121236e-06, "loss": 3.0116, "step": 308090 }, { "epoch": 0.004352, "grad_norm": 0.8205708265304565, "learning_rate": 2.128143227613323e-06, "loss": 2.9197, "step": 308100 }, { "epoch": 0.0043776, "grad_norm": 0.8722935318946838, "learning_rate": 2.1276466249614225e-06, "loss": 2.7036, "step": 308110 }, { "epoch": 0.0044032, "grad_norm": 0.8056377172470093, "learning_rate": 2.1271500733596428e-06, "loss": 2.5544, "step": 308120 }, { "epoch": 0.0044288, "grad_norm": 0.7971950769424438, "learning_rate": 2.1266535728112014e-06, "loss": 2.8166, "step": 308130 }, { "epoch": 0.0044544, "grad_norm": 0.8524977564811707, "learning_rate": 2.126157123319318e-06, "loss": 2.9982, "step": 308140 }, { "epoch": 0.00448, "grad_norm": 0.804439902305603, "learning_rate": 2.125660724887213e-06, "loss": 2.522, "step": 308150 }, { "epoch": 0.0045056, "grad_norm": 0.7903702855110168, "learning_rate": 2.1251643775181053e-06, "loss": 2.8138, "step": 308160 }, { "epoch": 0.0045312, "grad_norm": 0.8314222693443298, "learning_rate": 2.1246680812152145e-06, "loss": 2.8273, "step": 308170 }, { "epoch": 0.0045568, "grad_norm": 0.8967740535736084, "learning_rate": 2.1241718359817577e-06, "loss": 2.8113, "step": 308180 }, { "epoch": 0.0045824, "grad_norm": 0.7546262741088867, "learning_rate": 2.123675641820956e-06, "loss": 2.8413, "step": 308190 }, { "epoch": 0.004608, "grad_norm": 0.8148840665817261, "learning_rate": 2.1231794987360223e-06, "loss": 2.7933, "step": 308200 }, { "epoch": 0.0046336, "grad_norm": 0.8577901721000671, "learning_rate": 2.122683406730176e-06, "loss": 2.8787, "step": 308210 }, { "epoch": 0.0046592, "grad_norm": 0.8243144750595093, "learning_rate": 2.1221873658066337e-06, "loss": 2.7247, "step": 308220 }, { "epoch": 0.0046848, "grad_norm": 0.8719000220298767, "learning_rate": 2.121691375968613e-06, "loss": 2.7868, "step": 308230 }, { "epoch": 0.0047104, "grad_norm": 0.9230839014053345, "learning_rate": 2.1211954372193277e-06, "loss": 2.8321, "step": 308240 }, { "epoch": 0.004736, "grad_norm": 0.7629652619361877, "learning_rate": 2.1206995495620007e-06, "loss": 2.756, "step": 308250 }, { "epoch": 0.0047616, "grad_norm": 0.7970332503318787, "learning_rate": 2.120203712999841e-06, "loss": 2.8443, "step": 308260 }, { "epoch": 0.0047872, "grad_norm": 0.8907939791679382, "learning_rate": 2.1197079275360654e-06, "loss": 2.8241, "step": 308270 }, { "epoch": 0.0048128, "grad_norm": 0.8249110579490662, "learning_rate": 2.1192121931738896e-06, "loss": 2.8521, "step": 308280 }, { "epoch": 0.0048384, "grad_norm": 0.9200559258460999, "learning_rate": 2.1187165099165284e-06, "loss": 2.9564, "step": 308290 }, { "epoch": 0.004864, "grad_norm": 0.7990629076957703, "learning_rate": 2.1182208777671964e-06, "loss": 2.7235, "step": 308300 }, { "epoch": 0.0048896, "grad_norm": 0.742843747138977, "learning_rate": 2.1177252967291073e-06, "loss": 2.8051, "step": 308310 }, { "epoch": 0.0049152, "grad_norm": 0.8652282357215881, "learning_rate": 2.1172297668054763e-06, "loss": 2.843, "step": 308320 }, { "epoch": 0.0049408, "grad_norm": 0.7450393438339233, "learning_rate": 2.1167342879995145e-06, "loss": 2.9271, "step": 308330 }, { "epoch": 0.0049664, "grad_norm": 0.8484225273132324, "learning_rate": 2.1162388603144344e-06, "loss": 2.8211, "step": 308340 }, { "epoch": 0.004992, "grad_norm": 0.7873411178588867, "learning_rate": 2.115743483753451e-06, "loss": 2.6768, "step": 308350 }, { "epoch": 0.0050176, "grad_norm": 0.7985938191413879, "learning_rate": 2.115248158319775e-06, "loss": 2.8323, "step": 308360 }, { "epoch": 0.0050432, "grad_norm": 0.843008279800415, "learning_rate": 2.1147528840166186e-06, "loss": 2.7484, "step": 308370 }, { "epoch": 0.0050688, "grad_norm": 1.0514706373214722, "learning_rate": 2.1142576608471943e-06, "loss": 2.6796, "step": 308380 }, { "epoch": 0.0050944, "grad_norm": 0.8136385679244995, "learning_rate": 2.113762488814717e-06, "loss": 2.9565, "step": 308390 }, { "epoch": 0.00512, "grad_norm": 0.8508817553520203, "learning_rate": 2.11326736792239e-06, "loss": 2.8124, "step": 308400 }, { "epoch": 0.0051456, "grad_norm": 0.8237109184265137, "learning_rate": 2.1127722981734266e-06, "loss": 2.7279, "step": 308410 }, { "epoch": 0.0051712, "grad_norm": 0.9261105060577393, "learning_rate": 2.112277279571041e-06, "loss": 2.8025, "step": 308420 }, { "epoch": 0.0051968, "grad_norm": 0.8102893829345703, "learning_rate": 2.111782312118441e-06, "loss": 2.7649, "step": 308430 }, { "epoch": 0.0052224, "grad_norm": 0.9154549241065979, "learning_rate": 2.111287395818836e-06, "loss": 2.7677, "step": 308440 }, { "epoch": 0.005248, "grad_norm": 0.9227902889251709, "learning_rate": 2.1107925306754385e-06, "loss": 2.9411, "step": 308450 }, { "epoch": 0.0052736, "grad_norm": 0.855682909488678, "learning_rate": 2.110297716691453e-06, "loss": 2.8646, "step": 308460 }, { "epoch": 0.0052992, "grad_norm": 0.8785244822502136, "learning_rate": 2.109802953870088e-06, "loss": 2.7582, "step": 308470 }, { "epoch": 0.0053248, "grad_norm": 0.7806682586669922, "learning_rate": 2.109308242214555e-06, "loss": 2.8211, "step": 308480 }, { "epoch": 0.0053504, "grad_norm": 0.8141112327575684, "learning_rate": 2.108813581728061e-06, "loss": 2.645, "step": 308490 }, { "epoch": 0.005376, "grad_norm": 0.832053542137146, "learning_rate": 2.108318972413814e-06, "loss": 2.8955, "step": 308500 }, { "epoch": 0.0054016, "grad_norm": 0.8280965089797974, "learning_rate": 2.1078244142750204e-06, "loss": 2.5982, "step": 308510 }, { "epoch": 0.0054272, "grad_norm": 0.7742498517036438, "learning_rate": 2.107329907314891e-06, "loss": 2.6267, "step": 308520 }, { "epoch": 0.0054528, "grad_norm": 0.9484339356422424, "learning_rate": 2.1068354515366274e-06, "loss": 2.889, "step": 308530 }, { "epoch": 0.0054784, "grad_norm": 0.7817787528038025, "learning_rate": 2.1063410469434377e-06, "loss": 2.5956, "step": 308540 }, { "epoch": 0.005504, "grad_norm": 0.8344441056251526, "learning_rate": 2.105846693538528e-06, "loss": 2.7945, "step": 308550 }, { "epoch": 0.0055296, "grad_norm": 0.8295695781707764, "learning_rate": 2.105352391325104e-06, "loss": 2.7834, "step": 308560 }, { "epoch": 0.0055552, "grad_norm": 0.8590511679649353, "learning_rate": 2.104858140306373e-06, "loss": 2.7295, "step": 308570 }, { "epoch": 0.0055808, "grad_norm": 1.152230143547058, "learning_rate": 2.1043639404855377e-06, "loss": 2.7182, "step": 308580 }, { "epoch": 0.0056064, "grad_norm": 0.7651719450950623, "learning_rate": 2.1038697918658046e-06, "loss": 2.7571, "step": 308590 }, { "epoch": 0.005632, "grad_norm": 0.8096716403961182, "learning_rate": 2.103375694450377e-06, "loss": 2.8478, "step": 308600 }, { "epoch": 0.0056576, "grad_norm": 0.8367869257926941, "learning_rate": 2.1028816482424587e-06, "loss": 2.7205, "step": 308610 }, { "epoch": 0.0056832, "grad_norm": 0.9659481048583984, "learning_rate": 2.1023876532452546e-06, "loss": 2.6811, "step": 308620 }, { "epoch": 0.0057088, "grad_norm": 0.8870972394943237, "learning_rate": 2.101893709461967e-06, "loss": 2.8459, "step": 308630 }, { "epoch": 0.0057344, "grad_norm": 0.8338413238525391, "learning_rate": 2.101399816895802e-06, "loss": 2.8576, "step": 308640 }, { "epoch": 0.00576, "grad_norm": 0.8588107228279114, "learning_rate": 2.1009059755499605e-06, "loss": 2.8007, "step": 308650 }, { "epoch": 0.0057856, "grad_norm": 0.8862882256507874, "learning_rate": 2.1004121854276414e-06, "loss": 2.7204, "step": 308660 }, { "epoch": 0.0058112, "grad_norm": 0.7919105291366577, "learning_rate": 2.099918446532051e-06, "loss": 2.698, "step": 308670 }, { "epoch": 0.0058368, "grad_norm": 0.8963086605072021, "learning_rate": 2.0994247588663898e-06, "loss": 2.8093, "step": 308680 }, { "epoch": 0.0058624, "grad_norm": 0.9933091998100281, "learning_rate": 2.098931122433859e-06, "loss": 2.4924, "step": 308690 }, { "epoch": 0.005888, "grad_norm": 0.9609891772270203, "learning_rate": 2.09843753723766e-06, "loss": 2.8303, "step": 308700 }, { "epoch": 0.0059136, "grad_norm": 0.9012961387634277, "learning_rate": 2.0979440032809938e-06, "loss": 2.8614, "step": 308710 }, { "epoch": 0.0059392, "grad_norm": 0.8019901514053345, "learning_rate": 2.0974505205670635e-06, "loss": 2.7139, "step": 308720 }, { "epoch": 0.0059648, "grad_norm": 0.7585158944129944, "learning_rate": 2.096957089099063e-06, "loss": 2.6764, "step": 308730 }, { "epoch": 0.0059904, "grad_norm": 0.8469986915588379, "learning_rate": 2.0964637088801954e-06, "loss": 2.6348, "step": 308740 }, { "epoch": 0.006016, "grad_norm": 0.8042635917663574, "learning_rate": 2.095970379913661e-06, "loss": 2.7884, "step": 308750 }, { "epoch": 0.0060416, "grad_norm": 0.8129196166992188, "learning_rate": 2.095477102202654e-06, "loss": 2.9285, "step": 308760 }, { "epoch": 0.0060672, "grad_norm": 0.8648669123649597, "learning_rate": 2.0949838757503817e-06, "loss": 2.6182, "step": 308770 }, { "epoch": 0.0060928, "grad_norm": 0.7619086503982544, "learning_rate": 2.09449070056004e-06, "loss": 2.6318, "step": 308780 }, { "epoch": 0.0061184, "grad_norm": 0.7441226840019226, "learning_rate": 2.093997576634822e-06, "loss": 2.4175, "step": 308790 }, { "epoch": 0.006144, "grad_norm": 0.8741998672485352, "learning_rate": 2.093504503977929e-06, "loss": 2.7098, "step": 308800 }, { "epoch": 0.0061696, "grad_norm": 0.8148482441902161, "learning_rate": 2.093011482592557e-06, "loss": 2.655, "step": 308810 }, { "epoch": 0.0061952, "grad_norm": 0.888494074344635, "learning_rate": 2.0925185124819047e-06, "loss": 2.8019, "step": 308820 }, { "epoch": 0.0062208, "grad_norm": 1.1733791828155518, "learning_rate": 2.0920255936491686e-06, "loss": 2.8752, "step": 308830 }, { "epoch": 0.0062464, "grad_norm": 0.8222320675849915, "learning_rate": 2.091532726097545e-06, "loss": 2.9742, "step": 308840 }, { "epoch": 0.006272, "grad_norm": 0.8355840444564819, "learning_rate": 2.0910399098302327e-06, "loss": 2.8139, "step": 308850 }, { "epoch": 0.0062976, "grad_norm": 0.8500503301620483, "learning_rate": 2.090547144850421e-06, "loss": 2.7586, "step": 308860 }, { "epoch": 0.0063232, "grad_norm": 0.8214004039764404, "learning_rate": 2.090054431161309e-06, "loss": 2.5449, "step": 308870 }, { "epoch": 0.0063488, "grad_norm": 0.8112009763717651, "learning_rate": 2.089561768766093e-06, "loss": 3.0254, "step": 308880 }, { "epoch": 0.0063744, "grad_norm": 0.869697630405426, "learning_rate": 2.0890691576679654e-06, "loss": 2.7964, "step": 308890 }, { "epoch": 0.0064, "grad_norm": 0.8824992179870605, "learning_rate": 2.088576597870122e-06, "loss": 2.6332, "step": 308900 }, { "epoch": 0.0064256, "grad_norm": 0.8077847361564636, "learning_rate": 2.088084089375757e-06, "loss": 2.7647, "step": 308910 }, { "epoch": 0.0064512, "grad_norm": 0.7606468796730042, "learning_rate": 2.0875916321880674e-06, "loss": 3.0557, "step": 308920 }, { "epoch": 0.0064768, "grad_norm": 0.9703739881515503, "learning_rate": 2.0870992263102362e-06, "loss": 2.7597, "step": 308930 }, { "epoch": 0.0065024, "grad_norm": 0.8730241060256958, "learning_rate": 2.086606871745468e-06, "loss": 2.6674, "step": 308940 }, { "epoch": 0.006528, "grad_norm": 5.547852993011475, "learning_rate": 2.0861145684969497e-06, "loss": 2.8012, "step": 308950 }, { "epoch": 0.0065536, "grad_norm": 0.966758668422699, "learning_rate": 2.085622316567876e-06, "loss": 2.6017, "step": 308960 }, { "epoch": 0.0065792, "grad_norm": 0.822376012802124, "learning_rate": 2.0851301159614377e-06, "loss": 2.6639, "step": 308970 }, { "epoch": 0.0066048, "grad_norm": 0.8453991413116455, "learning_rate": 2.084637966680829e-06, "loss": 2.6808, "step": 308980 }, { "epoch": 0.0066304, "grad_norm": 0.8712595105171204, "learning_rate": 2.0841458687292382e-06, "loss": 2.9571, "step": 308990 }, { "epoch": 0.006656, "grad_norm": 0.8797406554222107, "learning_rate": 2.083653822109857e-06, "loss": 2.7463, "step": 309000 }, { "epoch": 0.0066816, "grad_norm": 0.8773547410964966, "learning_rate": 2.083161826825877e-06, "loss": 2.8553, "step": 309010 }, { "epoch": 0.0067072, "grad_norm": 0.8359566330909729, "learning_rate": 2.0826698828804893e-06, "loss": 2.7531, "step": 309020 }, { "epoch": 0.0067328, "grad_norm": 0.9186136722564697, "learning_rate": 2.0821779902768825e-06, "loss": 2.8236, "step": 309030 }, { "epoch": 0.0067584, "grad_norm": 0.8990823030471802, "learning_rate": 2.081686149018246e-06, "loss": 3.0316, "step": 309040 }, { "epoch": 0.006784, "grad_norm": 0.8188052773475647, "learning_rate": 2.0811943591077744e-06, "loss": 2.7653, "step": 309050 }, { "epoch": 0.0068096, "grad_norm": 0.7925952672958374, "learning_rate": 2.0807026205486504e-06, "loss": 2.7637, "step": 309060 }, { "epoch": 0.0068352, "grad_norm": 0.7273144721984863, "learning_rate": 2.0802109333440645e-06, "loss": 2.8665, "step": 309070 }, { "epoch": 0.0068608, "grad_norm": 0.8382094502449036, "learning_rate": 2.079719297497206e-06, "loss": 2.7537, "step": 309080 }, { "epoch": 0.0068864, "grad_norm": 0.8703230023384094, "learning_rate": 2.0792277130112626e-06, "loss": 2.7631, "step": 309090 }, { "epoch": 0.006912, "grad_norm": 0.8793356418609619, "learning_rate": 2.0787361798894224e-06, "loss": 2.807, "step": 309100 }, { "epoch": 0.0069376, "grad_norm": 0.8466686010360718, "learning_rate": 2.0782446981348735e-06, "loss": 2.8535, "step": 309110 }, { "epoch": 0.0069632, "grad_norm": 0.929844856262207, "learning_rate": 2.0777532677508018e-06, "loss": 2.8172, "step": 309120 }, { "epoch": 0.0069888, "grad_norm": 0.7805706262588501, "learning_rate": 2.077261888740395e-06, "loss": 2.8434, "step": 309130 }, { "epoch": 0.0070144, "grad_norm": 0.7519225478172302, "learning_rate": 2.076770561106839e-06, "loss": 2.6053, "step": 309140 }, { "epoch": 0.00704, "grad_norm": 1.220508337020874, "learning_rate": 2.076279284853321e-06, "loss": 2.775, "step": 309150 }, { "epoch": 0.0070656, "grad_norm": 1.13797926902771, "learning_rate": 2.075788059983025e-06, "loss": 2.9965, "step": 309160 }, { "epoch": 0.0070912, "grad_norm": 0.7419585585594177, "learning_rate": 2.0752968864991372e-06, "loss": 2.8605, "step": 309170 }, { "epoch": 0.0071168, "grad_norm": 0.9452444911003113, "learning_rate": 2.0748057644048468e-06, "loss": 2.4828, "step": 309180 }, { "epoch": 0.0071424, "grad_norm": 0.9591494202613831, "learning_rate": 2.074314693703331e-06, "loss": 2.4113, "step": 309190 }, { "epoch": 0.007168, "grad_norm": 0.8207145929336548, "learning_rate": 2.073823674397779e-06, "loss": 2.7093, "step": 309200 }, { "epoch": 0.0071936, "grad_norm": 0.7562278509140015, "learning_rate": 2.073332706491373e-06, "loss": 2.8218, "step": 309210 }, { "epoch": 0.0072192, "grad_norm": 0.809977114200592, "learning_rate": 2.072841789987299e-06, "loss": 2.7005, "step": 309220 }, { "epoch": 0.0072448, "grad_norm": 0.8673093914985657, "learning_rate": 2.0723509248887376e-06, "loss": 2.5171, "step": 309230 }, { "epoch": 0.0072704, "grad_norm": 0.9597198367118835, "learning_rate": 2.0718601111988747e-06, "loss": 3.0341, "step": 309240 }, { "epoch": 0.007296, "grad_norm": 0.8939404487609863, "learning_rate": 2.071369348920894e-06, "loss": 2.8249, "step": 309250 }, { "epoch": 0.0073216, "grad_norm": 0.8884401917457581, "learning_rate": 2.0708786380579736e-06, "loss": 2.8911, "step": 309260 }, { "epoch": 0.0073472, "grad_norm": 0.8061392307281494, "learning_rate": 2.0703879786132975e-06, "loss": 2.8167, "step": 309270 }, { "epoch": 0.0073728, "grad_norm": 0.8055899739265442, "learning_rate": 2.069897370590046e-06, "loss": 2.7524, "step": 309280 }, { "epoch": 0.0073984, "grad_norm": 0.8655830025672913, "learning_rate": 2.0694068139914047e-06, "loss": 2.7638, "step": 309290 }, { "epoch": 0.007424, "grad_norm": 0.8536667227745056, "learning_rate": 2.0689163088205544e-06, "loss": 2.7618, "step": 309300 }, { "epoch": 0.0074496, "grad_norm": 1.2402313947677612, "learning_rate": 2.0684258550806725e-06, "loss": 2.7142, "step": 309310 }, { "epoch": 0.0074752, "grad_norm": 0.8186902403831482, "learning_rate": 2.0679354527749407e-06, "loss": 2.6408, "step": 309320 }, { "epoch": 0.0075008, "grad_norm": 0.9302608370780945, "learning_rate": 2.067445101906539e-06, "loss": 2.7662, "step": 309330 }, { "epoch": 0.0075264, "grad_norm": 1.0095465183258057, "learning_rate": 2.0669548024786477e-06, "loss": 2.9859, "step": 309340 }, { "epoch": 0.007552, "grad_norm": 0.8229436278343201, "learning_rate": 2.066464554494446e-06, "loss": 2.8587, "step": 309350 }, { "epoch": 0.0075776, "grad_norm": 0.8982342481613159, "learning_rate": 2.0659743579571136e-06, "loss": 2.8955, "step": 309360 }, { "epoch": 0.0076032, "grad_norm": 0.8623790144920349, "learning_rate": 2.0654842128698315e-06, "loss": 2.9132, "step": 309370 }, { "epoch": 0.0076288, "grad_norm": 0.7810900807380676, "learning_rate": 2.0649941192357726e-06, "loss": 2.9466, "step": 309380 }, { "epoch": 0.0076544, "grad_norm": 0.7578645944595337, "learning_rate": 2.064504077058118e-06, "loss": 2.9509, "step": 309390 }, { "epoch": 0.00768, "grad_norm": 0.8478977680206299, "learning_rate": 2.064014086340046e-06, "loss": 2.9583, "step": 309400 }, { "epoch": 0.0077056, "grad_norm": 0.9681391716003418, "learning_rate": 2.063524147084733e-06, "loss": 2.8095, "step": 309410 }, { "epoch": 0.0077312, "grad_norm": 0.8072119951248169, "learning_rate": 2.063034259295357e-06, "loss": 2.8825, "step": 309420 }, { "epoch": 0.0077568, "grad_norm": 0.7578076124191284, "learning_rate": 2.0625444229750947e-06, "loss": 2.7935, "step": 309430 }, { "epoch": 0.0077824, "grad_norm": 0.7962637543678284, "learning_rate": 2.062054638127121e-06, "loss": 2.7435, "step": 309440 }, { "epoch": 0.007808, "grad_norm": 0.7680957317352295, "learning_rate": 2.061564904754617e-06, "loss": 2.8612, "step": 309450 }, { "epoch": 0.0078336, "grad_norm": 0.7916638851165771, "learning_rate": 2.0610752228607487e-06, "loss": 2.8949, "step": 309460 }, { "epoch": 0.0078592, "grad_norm": 0.8034176230430603, "learning_rate": 2.0605855924487018e-06, "loss": 2.6546, "step": 309470 }, { "epoch": 0.0078848, "grad_norm": 0.9185273051261902, "learning_rate": 2.0600960135216463e-06, "loss": 2.6387, "step": 309480 }, { "epoch": 0.0079104, "grad_norm": 0.8559918999671936, "learning_rate": 2.059606486082758e-06, "loss": 2.7565, "step": 309490 }, { "epoch": 0.007936, "grad_norm": 0.819653332233429, "learning_rate": 2.0591170101352144e-06, "loss": 2.7349, "step": 309500 }, { "epoch": 0.0079616, "grad_norm": 0.8559736609458923, "learning_rate": 2.0586275856821835e-06, "loss": 2.7575, "step": 309510 }, { "epoch": 0.0079872, "grad_norm": 0.8657306432723999, "learning_rate": 2.058138212726841e-06, "loss": 2.7395, "step": 309520 }, { "epoch": 0.0080128, "grad_norm": 0.8318414688110352, "learning_rate": 2.0576488912723635e-06, "loss": 2.9418, "step": 309530 }, { "epoch": 0.0080384, "grad_norm": 0.8276878595352173, "learning_rate": 2.0571596213219216e-06, "loss": 2.6889, "step": 309540 }, { "epoch": 0.008064, "grad_norm": 0.8743070363998413, "learning_rate": 2.0566704028786876e-06, "loss": 2.8469, "step": 309550 }, { "epoch": 0.0080896, "grad_norm": 0.8658279776573181, "learning_rate": 2.0561812359458356e-06, "loss": 2.8564, "step": 309560 }, { "epoch": 0.0081152, "grad_norm": 0.8060664534568787, "learning_rate": 2.05569212052654e-06, "loss": 2.7243, "step": 309570 }, { "epoch": 0.0081408, "grad_norm": 0.7928948402404785, "learning_rate": 2.0552030566239667e-06, "loss": 2.8076, "step": 309580 }, { "epoch": 0.0081664, "grad_norm": 1.4064420461654663, "learning_rate": 2.0547140442412906e-06, "loss": 2.3817, "step": 309590 }, { "epoch": 0.008192, "grad_norm": 1.0407441854476929, "learning_rate": 2.0542250833816826e-06, "loss": 2.7804, "step": 309600 }, { "epoch": 0.0082176, "grad_norm": 0.809363067150116, "learning_rate": 2.053736174048312e-06, "loss": 2.5712, "step": 309610 }, { "epoch": 0.0082432, "grad_norm": 0.7968881130218506, "learning_rate": 2.053247316244351e-06, "loss": 2.8411, "step": 309620 }, { "epoch": 0.0082688, "grad_norm": 0.8730881810188293, "learning_rate": 2.0527585099729688e-06, "loss": 2.8804, "step": 309630 }, { "epoch": 0.0082944, "grad_norm": 0.9033227562904358, "learning_rate": 2.052269755237336e-06, "loss": 2.8036, "step": 309640 }, { "epoch": 0.00832, "grad_norm": 0.8262837529182434, "learning_rate": 2.051781052040621e-06, "loss": 2.752, "step": 309650 }, { "epoch": 0.0083456, "grad_norm": 0.8838971853256226, "learning_rate": 2.0512924003859947e-06, "loss": 2.9245, "step": 309660 }, { "epoch": 0.0083712, "grad_norm": 0.8749560117721558, "learning_rate": 2.050803800276624e-06, "loss": 2.8785, "step": 309670 }, { "epoch": 0.0083968, "grad_norm": 0.808709979057312, "learning_rate": 2.0503152517156777e-06, "loss": 2.8853, "step": 309680 }, { "epoch": 0.0084224, "grad_norm": 1.009407877922058, "learning_rate": 2.0498267547063244e-06, "loss": 2.8307, "step": 309690 }, { "epoch": 0.008448, "grad_norm": 0.8124697804450989, "learning_rate": 2.049338309251734e-06, "loss": 2.9259, "step": 309700 }, { "epoch": 0.0084736, "grad_norm": 0.832356870174408, "learning_rate": 2.048849915355069e-06, "loss": 2.9482, "step": 309710 }, { "epoch": 0.0084992, "grad_norm": 0.8101840615272522, "learning_rate": 2.0483615730194995e-06, "loss": 2.517, "step": 309720 }, { "epoch": 0.0085248, "grad_norm": 0.8271698951721191, "learning_rate": 2.0478732822481927e-06, "loss": 2.7015, "step": 309730 }, { "epoch": 0.0085504, "grad_norm": 0.8316813707351685, "learning_rate": 2.047385043044313e-06, "loss": 2.7399, "step": 309740 }, { "epoch": 0.008576, "grad_norm": 0.8760319948196411, "learning_rate": 2.046896855411028e-06, "loss": 2.7505, "step": 309750 }, { "epoch": 0.0086016, "grad_norm": 0.7928896546363831, "learning_rate": 2.046408719351504e-06, "loss": 2.8643, "step": 309760 }, { "epoch": 0.0086272, "grad_norm": 0.8647242188453674, "learning_rate": 2.0459206348689074e-06, "loss": 2.7998, "step": 309770 }, { "epoch": 0.0086528, "grad_norm": 0.743791937828064, "learning_rate": 2.0454326019663993e-06, "loss": 2.5669, "step": 309780 }, { "epoch": 0.0086784, "grad_norm": 0.8129785060882568, "learning_rate": 2.044944620647147e-06, "loss": 2.7892, "step": 309790 }, { "epoch": 0.008704, "grad_norm": 0.8066656589508057, "learning_rate": 2.0444566909143114e-06, "loss": 2.9554, "step": 309800 }, { "epoch": 0.0087296, "grad_norm": 0.7797778844833374, "learning_rate": 2.0439688127710623e-06, "loss": 2.7885, "step": 309810 }, { "epoch": 0.0087552, "grad_norm": 0.8379383683204651, "learning_rate": 2.0434809862205617e-06, "loss": 2.816, "step": 309820 }, { "epoch": 0.0087808, "grad_norm": 0.8994923830032349, "learning_rate": 2.0429932112659745e-06, "loss": 2.6495, "step": 309830 }, { "epoch": 0.0088064, "grad_norm": 0.8299852609634399, "learning_rate": 2.042505487910459e-06, "loss": 2.8051, "step": 309840 }, { "epoch": 0.008832, "grad_norm": 0.8846657872200012, "learning_rate": 2.04201781615718e-06, "loss": 2.8113, "step": 309850 }, { "epoch": 0.0088576, "grad_norm": 1.016584873199463, "learning_rate": 2.0415301960093014e-06, "loss": 2.7253, "step": 309860 }, { "epoch": 0.0088832, "grad_norm": 0.8602189421653748, "learning_rate": 2.0410426274699845e-06, "loss": 2.9959, "step": 309870 }, { "epoch": 0.0089088, "grad_norm": 0.8118423819541931, "learning_rate": 2.0405551105423903e-06, "loss": 2.9022, "step": 309880 }, { "epoch": 0.0089344, "grad_norm": 0.870647668838501, "learning_rate": 2.0400676452296806e-06, "loss": 2.7108, "step": 309890 }, { "epoch": 0.00896, "grad_norm": 0.8012136220932007, "learning_rate": 2.0395802315350198e-06, "loss": 2.8711, "step": 309900 }, { "epoch": 0.0089856, "grad_norm": 0.7299701571464539, "learning_rate": 2.039092869461563e-06, "loss": 2.7959, "step": 309910 }, { "epoch": 0.0090112, "grad_norm": 1.076701283454895, "learning_rate": 2.038605559012473e-06, "loss": 3.0223, "step": 309920 }, { "epoch": 0.0090368, "grad_norm": 0.8157925009727478, "learning_rate": 2.038118300190911e-06, "loss": 3.0143, "step": 309930 }, { "epoch": 0.0090624, "grad_norm": 0.794822096824646, "learning_rate": 2.037631093000034e-06, "loss": 2.8208, "step": 309940 }, { "epoch": 0.009088, "grad_norm": 0.8529348373413086, "learning_rate": 2.0371439374430046e-06, "loss": 2.7858, "step": 309950 }, { "epoch": 0.0091136, "grad_norm": 0.9107646346092224, "learning_rate": 2.036656833522981e-06, "loss": 2.8597, "step": 309960 }, { "epoch": 0.0091392, "grad_norm": 0.7853865623474121, "learning_rate": 2.0361697812431227e-06, "loss": 2.7471, "step": 309970 }, { "epoch": 0.0091648, "grad_norm": 0.747601330280304, "learning_rate": 2.0356827806065815e-06, "loss": 2.7035, "step": 309980 }, { "epoch": 0.0091904, "grad_norm": 0.8047707080841064, "learning_rate": 2.035195831616524e-06, "loss": 2.8458, "step": 309990 }, { "epoch": 0.009216, "grad_norm": 1.0130163431167603, "learning_rate": 2.034708934276105e-06, "loss": 2.9898, "step": 310000 }, { "epoch": 0.0092416, "grad_norm": 0.9160052537918091, "learning_rate": 2.0342220885884812e-06, "loss": 2.7287, "step": 310010 }, { "epoch": 0.0092672, "grad_norm": 0.8752490282058716, "learning_rate": 2.033735294556811e-06, "loss": 2.6969, "step": 310020 }, { "epoch": 0.0092928, "grad_norm": 0.8506874442100525, "learning_rate": 2.0332485521842515e-06, "loss": 2.8169, "step": 310030 }, { "epoch": 0.0093184, "grad_norm": 0.8392053842544556, "learning_rate": 2.0327618614739563e-06, "loss": 2.7823, "step": 310040 }, { "epoch": 0.009344, "grad_norm": 0.8276275992393494, "learning_rate": 2.032275222429082e-06, "loss": 2.8713, "step": 310050 }, { "epoch": 0.0093696, "grad_norm": 0.7940829992294312, "learning_rate": 2.031788635052786e-06, "loss": 2.8601, "step": 310060 }, { "epoch": 0.0093952, "grad_norm": 0.7863911986351013, "learning_rate": 2.0313020993482237e-06, "loss": 2.8144, "step": 310070 }, { "epoch": 0.0094208, "grad_norm": 0.8514819145202637, "learning_rate": 2.0308156153185476e-06, "loss": 3.0005, "step": 310080 }, { "epoch": 0.0094464, "grad_norm": 0.9456355571746826, "learning_rate": 2.0303291829669157e-06, "loss": 2.6888, "step": 310090 }, { "epoch": 0.009472, "grad_norm": 0.7805267572402954, "learning_rate": 2.029842802296483e-06, "loss": 2.8508, "step": 310100 }, { "epoch": 0.0094976, "grad_norm": 0.8257302045822144, "learning_rate": 2.0293564733103986e-06, "loss": 2.7403, "step": 310110 }, { "epoch": 0.0095232, "grad_norm": 0.9304378032684326, "learning_rate": 2.028870196011821e-06, "loss": 3.0841, "step": 310120 }, { "epoch": 0.0095488, "grad_norm": 0.8288136720657349, "learning_rate": 2.0283839704039e-06, "loss": 2.6778, "step": 310130 }, { "epoch": 0.0095744, "grad_norm": 0.8189895153045654, "learning_rate": 2.0278977964897906e-06, "loss": 2.8652, "step": 310140 }, { "epoch": 0.0096, "grad_norm": 0.8873264193534851, "learning_rate": 2.0274116742726426e-06, "loss": 2.8102, "step": 310150 }, { "epoch": 0.0096256, "grad_norm": 1.253961443901062, "learning_rate": 2.0269256037556163e-06, "loss": 2.9831, "step": 310160 }, { "epoch": 0.0096512, "grad_norm": 0.821732759475708, "learning_rate": 2.026439584941856e-06, "loss": 2.8603, "step": 310170 }, { "epoch": 0.0096768, "grad_norm": 0.861136257648468, "learning_rate": 2.025953617834515e-06, "loss": 2.8841, "step": 310180 }, { "epoch": 0.0097024, "grad_norm": 0.8086330890655518, "learning_rate": 2.0254677024367466e-06, "loss": 2.7329, "step": 310190 }, { "epoch": 0.009728, "grad_norm": 0.8294306993484497, "learning_rate": 2.0249818387517007e-06, "loss": 2.759, "step": 310200 }, { "epoch": 0.0097536, "grad_norm": 0.8774784207344055, "learning_rate": 2.0244960267825276e-06, "loss": 2.6637, "step": 310210 }, { "epoch": 0.0097792, "grad_norm": 0.8244155645370483, "learning_rate": 2.024010266532379e-06, "loss": 2.7268, "step": 310220 }, { "epoch": 0.0098048, "grad_norm": 0.8772934675216675, "learning_rate": 2.023524558004406e-06, "loss": 2.7309, "step": 310230 }, { "epoch": 0.0098304, "grad_norm": 0.8922543525695801, "learning_rate": 2.023038901201754e-06, "loss": 2.8621, "step": 310240 }, { "epoch": 0.009856, "grad_norm": 0.8494873642921448, "learning_rate": 2.022553296127574e-06, "loss": 2.5922, "step": 310250 }, { "epoch": 0.0098816, "grad_norm": 0.7686871886253357, "learning_rate": 2.0220677427850168e-06, "loss": 2.8896, "step": 310260 }, { "epoch": 0.0099072, "grad_norm": 0.8386342525482178, "learning_rate": 2.021582241177229e-06, "loss": 2.8924, "step": 310270 }, { "epoch": 0.0099328, "grad_norm": 0.849622905254364, "learning_rate": 2.021096791307361e-06, "loss": 2.9222, "step": 310280 }, { "epoch": 0.0099584, "grad_norm": 1.1798961162567139, "learning_rate": 2.0206113931785587e-06, "loss": 2.7856, "step": 310290 }, { "epoch": 0.009984, "grad_norm": 1.1479803323745728, "learning_rate": 2.020126046793974e-06, "loss": 2.897, "step": 310300 }, { "epoch": 0.0100096, "grad_norm": 0.8048954606056213, "learning_rate": 2.0196407521567487e-06, "loss": 2.7748, "step": 310310 }, { "epoch": 0.0100352, "grad_norm": 0.8789328932762146, "learning_rate": 2.019155509270031e-06, "loss": 2.8342, "step": 310320 }, { "epoch": 0.0100608, "grad_norm": 0.907345712184906, "learning_rate": 2.0186703181369673e-06, "loss": 2.8942, "step": 310330 }, { "epoch": 0.0100864, "grad_norm": 0.837740421295166, "learning_rate": 2.018185178760708e-06, "loss": 3.0609, "step": 310340 }, { "epoch": 0.010112, "grad_norm": 0.7851691842079163, "learning_rate": 2.017700091144397e-06, "loss": 2.8452, "step": 310350 }, { "epoch": 0.0101376, "grad_norm": 0.8165971636772156, "learning_rate": 2.0172150552911807e-06, "loss": 2.7248, "step": 310360 }, { "epoch": 0.0101632, "grad_norm": 0.8396106362342834, "learning_rate": 2.0167300712042015e-06, "loss": 2.8606, "step": 310370 }, { "epoch": 0.0101888, "grad_norm": 0.8315606713294983, "learning_rate": 2.0162451388866054e-06, "loss": 2.8575, "step": 310380 }, { "epoch": 0.0102144, "grad_norm": 0.8047202229499817, "learning_rate": 2.015760258341538e-06, "loss": 2.9367, "step": 310390 }, { "epoch": 0.01024, "grad_norm": 0.7634702324867249, "learning_rate": 2.0152754295721434e-06, "loss": 2.6064, "step": 310400 }, { "epoch": 0.0102656, "grad_norm": 0.7760219573974609, "learning_rate": 2.014790652581566e-06, "loss": 2.7647, "step": 310410 }, { "epoch": 0.0102912, "grad_norm": 0.9019415974617004, "learning_rate": 2.0143059273729494e-06, "loss": 2.8713, "step": 310420 }, { "epoch": 0.0103168, "grad_norm": 0.7916609644889832, "learning_rate": 2.0138212539494384e-06, "loss": 2.9809, "step": 310430 }, { "epoch": 0.0103424, "grad_norm": 0.8668674826622009, "learning_rate": 2.013336632314171e-06, "loss": 2.9246, "step": 310440 }, { "epoch": 0.010368, "grad_norm": 0.8645583987236023, "learning_rate": 2.012852062470293e-06, "loss": 2.8141, "step": 310450 }, { "epoch": 0.0103936, "grad_norm": 0.8099820017814636, "learning_rate": 2.012367544420948e-06, "loss": 2.8764, "step": 310460 }, { "epoch": 0.0104192, "grad_norm": 0.8056118488311768, "learning_rate": 2.0118830781692757e-06, "loss": 2.8752, "step": 310470 }, { "epoch": 0.0104448, "grad_norm": 0.823302686214447, "learning_rate": 2.011398663718418e-06, "loss": 2.9681, "step": 310480 }, { "epoch": 0.0104704, "grad_norm": 0.8228491544723511, "learning_rate": 2.010914301071518e-06, "loss": 2.8998, "step": 310490 }, { "epoch": 0.010496, "grad_norm": 0.7909077405929565, "learning_rate": 2.0104299902317148e-06, "loss": 2.8216, "step": 310500 }, { "epoch": 0.0105216, "grad_norm": 0.8436930775642395, "learning_rate": 2.00994573120215e-06, "loss": 2.8487, "step": 310510 }, { "epoch": 0.0105472, "grad_norm": 0.9786382913589478, "learning_rate": 2.009461523985964e-06, "loss": 2.8804, "step": 310520 }, { "epoch": 0.0105728, "grad_norm": 0.7911093235015869, "learning_rate": 2.0089773685862957e-06, "loss": 2.6559, "step": 310530 }, { "epoch": 0.0105984, "grad_norm": 0.7606446146965027, "learning_rate": 2.008493265006286e-06, "loss": 2.8486, "step": 310540 }, { "epoch": 0.010624, "grad_norm": 0.8831766843795776, "learning_rate": 2.0080092132490725e-06, "loss": 2.8144, "step": 310550 }, { "epoch": 0.0106496, "grad_norm": 0.837470531463623, "learning_rate": 2.0075252133177993e-06, "loss": 2.7786, "step": 310560 }, { "epoch": 0.0106752, "grad_norm": 1.0073069334030151, "learning_rate": 2.007041265215598e-06, "loss": 2.7311, "step": 310570 }, { "epoch": 0.0107008, "grad_norm": 0.7620656490325928, "learning_rate": 2.0065573689456084e-06, "loss": 2.9101, "step": 310580 }, { "epoch": 0.0107264, "grad_norm": 0.8376760482788086, "learning_rate": 2.006073524510972e-06, "loss": 3.3953, "step": 310590 }, { "epoch": 0.010752, "grad_norm": 0.7861424684524536, "learning_rate": 2.005589731914823e-06, "loss": 2.8415, "step": 310600 }, { "epoch": 0.0107776, "grad_norm": 1.291809320449829, "learning_rate": 2.0051059911603e-06, "loss": 2.8977, "step": 310610 }, { "epoch": 0.0108032, "grad_norm": 0.7683410048484802, "learning_rate": 2.0046223022505397e-06, "loss": 2.994, "step": 310620 }, { "epoch": 0.0108288, "grad_norm": 0.7793089747428894, "learning_rate": 2.0041386651886816e-06, "loss": 2.8478, "step": 310630 }, { "epoch": 0.0108544, "grad_norm": 0.9120191335678101, "learning_rate": 2.0036550799778574e-06, "loss": 2.8331, "step": 310640 }, { "epoch": 0.01088, "grad_norm": 0.8303346037864685, "learning_rate": 2.0031715466212043e-06, "loss": 2.7262, "step": 310650 }, { "epoch": 0.0109056, "grad_norm": 0.7926900386810303, "learning_rate": 2.0026880651218585e-06, "loss": 2.8821, "step": 310660 }, { "epoch": 0.0109312, "grad_norm": 1.0240966081619263, "learning_rate": 2.0022046354829516e-06, "loss": 2.8501, "step": 310670 }, { "epoch": 0.0109568, "grad_norm": 2.133504629135132, "learning_rate": 2.0017212577076284e-06, "loss": 2.8413, "step": 310680 }, { "epoch": 0.0109824, "grad_norm": 0.7988526821136475, "learning_rate": 2.001237931799014e-06, "loss": 2.8854, "step": 310690 }, { "epoch": 0.011008, "grad_norm": 0.8626271486282349, "learning_rate": 2.0007546577602465e-06, "loss": 2.9371, "step": 310700 }, { "epoch": 0.0110336, "grad_norm": 0.9206946492195129, "learning_rate": 2.0002714355944587e-06, "loss": 2.7542, "step": 310710 }, { "epoch": 0.0110592, "grad_norm": 0.9406287670135498, "learning_rate": 1.999788265304784e-06, "loss": 3.0079, "step": 310720 }, { "epoch": 0.0110848, "grad_norm": 0.8879364132881165, "learning_rate": 1.999305146894357e-06, "loss": 2.9224, "step": 310730 }, { "epoch": 0.0111104, "grad_norm": 0.9453058838844299, "learning_rate": 1.9988220803663094e-06, "loss": 2.7984, "step": 310740 }, { "epoch": 0.011136, "grad_norm": 0.8700287342071533, "learning_rate": 1.9983390657237733e-06, "loss": 2.7303, "step": 310750 }, { "epoch": 0.0111616, "grad_norm": 0.8674055337905884, "learning_rate": 1.9978561029698864e-06, "loss": 3.0642, "step": 310760 }, { "epoch": 0.0111872, "grad_norm": 0.8638098239898682, "learning_rate": 1.997373192107772e-06, "loss": 2.9229, "step": 310770 }, { "epoch": 0.0112128, "grad_norm": 0.9057856202125549, "learning_rate": 1.9968903331405654e-06, "loss": 2.9232, "step": 310780 }, { "epoch": 0.0112384, "grad_norm": 0.8469926118850708, "learning_rate": 1.996407526071399e-06, "loss": 2.8545, "step": 310790 }, { "epoch": 0.011264, "grad_norm": 0.8646856546401978, "learning_rate": 1.9959247709034022e-06, "loss": 2.9763, "step": 310800 }, { "epoch": 0.0112896, "grad_norm": 0.8301340341567993, "learning_rate": 1.9954420676397058e-06, "loss": 2.8149, "step": 310810 }, { "epoch": 0.0113152, "grad_norm": 0.9074700474739075, "learning_rate": 1.9949594162834406e-06, "loss": 2.8989, "step": 310820 }, { "epoch": 0.0113408, "grad_norm": 0.8616265058517456, "learning_rate": 1.9944768168377383e-06, "loss": 3.0364, "step": 310830 }, { "epoch": 0.0113664, "grad_norm": 0.9886346459388733, "learning_rate": 1.9939942693057236e-06, "loss": 2.864, "step": 310840 }, { "epoch": 0.011392, "grad_norm": 0.7764143943786621, "learning_rate": 1.993511773690526e-06, "loss": 2.9587, "step": 310850 }, { "epoch": 0.0114176, "grad_norm": 0.7910263538360596, "learning_rate": 1.9930293299952796e-06, "loss": 2.6118, "step": 310860 }, { "epoch": 0.0114432, "grad_norm": 0.8571826219558716, "learning_rate": 1.99254693822311e-06, "loss": 2.967, "step": 310870 }, { "epoch": 0.0114688, "grad_norm": 0.9523032307624817, "learning_rate": 1.992064598377147e-06, "loss": 2.704, "step": 310880 }, { "epoch": 0.0114944, "grad_norm": 0.9125062227249146, "learning_rate": 1.9915823104605146e-06, "loss": 2.7304, "step": 310890 }, { "epoch": 0.01152, "grad_norm": 0.8059098124504089, "learning_rate": 1.9911000744763433e-06, "loss": 2.6562, "step": 310900 }, { "epoch": 0.0115456, "grad_norm": 0.8841612935066223, "learning_rate": 1.9906178904277584e-06, "loss": 2.759, "step": 310910 }, { "epoch": 0.0115712, "grad_norm": 0.8295233249664307, "learning_rate": 1.9901357583178895e-06, "loss": 2.99, "step": 310920 }, { "epoch": 0.0115968, "grad_norm": 0.8558593988418579, "learning_rate": 1.9896536781498597e-06, "loss": 2.94, "step": 310930 }, { "epoch": 0.0116224, "grad_norm": 0.9285303354263306, "learning_rate": 1.9891716499267975e-06, "loss": 2.8692, "step": 310940 }, { "epoch": 0.011648, "grad_norm": 0.8424031734466553, "learning_rate": 1.988689673651829e-06, "loss": 2.6751, "step": 310950 }, { "epoch": 0.0116736, "grad_norm": 0.8245841264724731, "learning_rate": 1.9882077493280803e-06, "loss": 2.8032, "step": 310960 }, { "epoch": 0.0116992, "grad_norm": 1.010727047920227, "learning_rate": 1.9877258769586726e-06, "loss": 2.8893, "step": 310970 }, { "epoch": 0.0117248, "grad_norm": 1.9044371843338013, "learning_rate": 1.9872440565467335e-06, "loss": 3.3683, "step": 310980 }, { "epoch": 0.0117504, "grad_norm": 0.7851459980010986, "learning_rate": 1.9867622880953875e-06, "loss": 2.8394, "step": 310990 }, { "epoch": 0.011776, "grad_norm": 0.7542982697486877, "learning_rate": 1.986280571607757e-06, "loss": 2.6608, "step": 311000 }, { "epoch": 0.0118016, "grad_norm": 0.9162600040435791, "learning_rate": 1.9857989070869688e-06, "loss": 2.8004, "step": 311010 }, { "epoch": 0.0118272, "grad_norm": 0.8212898373603821, "learning_rate": 1.9853172945361445e-06, "loss": 2.6502, "step": 311020 }, { "epoch": 0.0118528, "grad_norm": 0.883471667766571, "learning_rate": 1.9848357339584067e-06, "loss": 2.8905, "step": 311030 }, { "epoch": 0.0118784, "grad_norm": 0.8495555520057678, "learning_rate": 1.9843542253568794e-06, "loss": 2.7616, "step": 311040 }, { "epoch": 0.011904, "grad_norm": 0.8973902463912964, "learning_rate": 1.9838727687346847e-06, "loss": 2.95, "step": 311050 }, { "epoch": 0.0119296, "grad_norm": 1.1385425329208374, "learning_rate": 1.9833913640949443e-06, "loss": 2.679, "step": 311060 }, { "epoch": 0.0119552, "grad_norm": 1.1218518018722534, "learning_rate": 1.982910011440781e-06, "loss": 2.6905, "step": 311070 }, { "epoch": 0.0119808, "grad_norm": 1.1101144552230835, "learning_rate": 1.9824287107753172e-06, "loss": 2.8233, "step": 311080 }, { "epoch": 0.0120064, "grad_norm": 0.7706053853034973, "learning_rate": 1.981947462101671e-06, "loss": 2.4722, "step": 311090 }, { "epoch": 0.012032, "grad_norm": 0.8269355893135071, "learning_rate": 1.9814662654229633e-06, "loss": 2.8725, "step": 311100 }, { "epoch": 0.0120576, "grad_norm": 0.808530330657959, "learning_rate": 1.980985120742317e-06, "loss": 2.7766, "step": 311110 }, { "epoch": 0.0120832, "grad_norm": 1.0088868141174316, "learning_rate": 1.980504028062851e-06, "loss": 2.761, "step": 311120 }, { "epoch": 0.0121088, "grad_norm": 0.8461239337921143, "learning_rate": 1.980022987387684e-06, "loss": 3.0816, "step": 311130 }, { "epoch": 0.0121344, "grad_norm": 0.822912335395813, "learning_rate": 1.9795419987199373e-06, "loss": 2.9381, "step": 311140 }, { "epoch": 0.01216, "grad_norm": 0.7966574430465698, "learning_rate": 1.979061062062728e-06, "loss": 2.7284, "step": 311150 }, { "epoch": 0.0121856, "grad_norm": 0.8565349578857422, "learning_rate": 1.9785801774191805e-06, "loss": 2.8792, "step": 311160 }, { "epoch": 0.0122112, "grad_norm": 1.5396479368209839, "learning_rate": 1.9780993447924047e-06, "loss": 2.8783, "step": 311170 }, { "epoch": 0.0122368, "grad_norm": 0.827326774597168, "learning_rate": 1.9776185641855226e-06, "loss": 2.9415, "step": 311180 }, { "epoch": 0.0122624, "grad_norm": 0.855793297290802, "learning_rate": 1.9771378356016537e-06, "loss": 2.9805, "step": 311190 }, { "epoch": 0.012288, "grad_norm": 0.9264487624168396, "learning_rate": 1.9766571590439087e-06, "loss": 2.8728, "step": 311200 }, { "epoch": 0.0123136, "grad_norm": 0.7363401055335999, "learning_rate": 1.976176534515416e-06, "loss": 2.9144, "step": 311210 }, { "epoch": 0.0123392, "grad_norm": 0.9516841769218445, "learning_rate": 1.975695962019284e-06, "loss": 2.9148, "step": 311220 }, { "epoch": 0.0123648, "grad_norm": 0.9383062124252319, "learning_rate": 1.975215441558631e-06, "loss": 2.9071, "step": 311230 }, { "epoch": 0.0123904, "grad_norm": 1.0308839082717896, "learning_rate": 1.974734973136572e-06, "loss": 2.7495, "step": 311240 }, { "epoch": 0.012416, "grad_norm": 0.8415061235427856, "learning_rate": 1.9742545567562244e-06, "loss": 2.9554, "step": 311250 }, { "epoch": 0.0124416, "grad_norm": 0.8396081924438477, "learning_rate": 1.973774192420703e-06, "loss": 2.8602, "step": 311260 }, { "epoch": 0.0124672, "grad_norm": 0.8607376217842102, "learning_rate": 1.9732938801331236e-06, "loss": 2.7219, "step": 311270 }, { "epoch": 0.0124928, "grad_norm": 1.2609398365020752, "learning_rate": 1.972813619896602e-06, "loss": 2.6595, "step": 311280 }, { "epoch": 0.0125184, "grad_norm": 0.7861475348472595, "learning_rate": 1.9723334117142477e-06, "loss": 2.852, "step": 311290 }, { "epoch": 0.012544, "grad_norm": 1.0623737573623657, "learning_rate": 1.971853255589179e-06, "loss": 3.1029, "step": 311300 }, { "epoch": 0.0125696, "grad_norm": 0.90892094373703, "learning_rate": 1.971373151524507e-06, "loss": 2.9024, "step": 311310 }, { "epoch": 0.0125952, "grad_norm": 0.8255841732025146, "learning_rate": 1.970893099523348e-06, "loss": 2.9642, "step": 311320 }, { "epoch": 0.0126208, "grad_norm": 0.7424671053886414, "learning_rate": 1.9704130995888116e-06, "loss": 2.9026, "step": 311330 }, { "epoch": 0.0126464, "grad_norm": 0.9935982823371887, "learning_rate": 1.969933151724013e-06, "loss": 3.0759, "step": 311340 }, { "epoch": 0.012672, "grad_norm": 0.762611985206604, "learning_rate": 1.969453255932067e-06, "loss": 2.5975, "step": 311350 }, { "epoch": 0.0126976, "grad_norm": 0.7316794991493225, "learning_rate": 1.9689734122160788e-06, "loss": 2.8234, "step": 311360 }, { "epoch": 0.0127232, "grad_norm": 0.7554928064346313, "learning_rate": 1.9684936205791605e-06, "loss": 2.6471, "step": 311370 }, { "epoch": 0.0127488, "grad_norm": 0.8000566363334656, "learning_rate": 1.9680138810244296e-06, "loss": 2.7801, "step": 311380 }, { "epoch": 0.0127744, "grad_norm": 0.7954751253128052, "learning_rate": 1.967534193554995e-06, "loss": 2.97, "step": 311390 }, { "epoch": 0.0128, "grad_norm": 1.5884740352630615, "learning_rate": 1.9670545581739665e-06, "loss": 3.0359, "step": 311400 }, { "epoch": 0.0128256, "grad_norm": 1.085235595703125, "learning_rate": 1.9665749748844553e-06, "loss": 2.7595, "step": 311410 }, { "epoch": 0.0128512, "grad_norm": 0.8203478455543518, "learning_rate": 1.9660954436895685e-06, "loss": 2.9031, "step": 311420 }, { "epoch": 0.0128768, "grad_norm": 0.8788136839866638, "learning_rate": 1.9656159645924176e-06, "loss": 2.9382, "step": 311430 }, { "epoch": 0.0129024, "grad_norm": 0.9377872943878174, "learning_rate": 1.965136537596113e-06, "loss": 2.8922, "step": 311440 }, { "epoch": 0.012928, "grad_norm": 0.7618908286094666, "learning_rate": 1.9646571627037614e-06, "loss": 2.8909, "step": 311450 }, { "epoch": 0.0129536, "grad_norm": 0.8733471035957336, "learning_rate": 1.964177839918472e-06, "loss": 2.7311, "step": 311460 }, { "epoch": 0.0129792, "grad_norm": 0.783420741558075, "learning_rate": 1.9636985692433553e-06, "loss": 2.9047, "step": 311470 }, { "epoch": 0.0130048, "grad_norm": 1.1644920110702515, "learning_rate": 1.963219350681519e-06, "loss": 2.741, "step": 311480 }, { "epoch": 0.0130304, "grad_norm": 0.784622073173523, "learning_rate": 1.962740184236067e-06, "loss": 2.7953, "step": 311490 }, { "epoch": 0.013056, "grad_norm": 0.8226067423820496, "learning_rate": 1.9622610699101085e-06, "loss": 2.9379, "step": 311500 }, { "epoch": 0.0130816, "grad_norm": 1.000908613204956, "learning_rate": 1.961782007706751e-06, "loss": 2.7878, "step": 311510 }, { "epoch": 0.0131072, "grad_norm": 0.8460748195648193, "learning_rate": 1.961302997629101e-06, "loss": 2.7777, "step": 311520 }, { "epoch": 0.0131328, "grad_norm": 0.7619823217391968, "learning_rate": 1.960824039680265e-06, "loss": 2.8696, "step": 311530 }, { "epoch": 0.0131584, "grad_norm": 0.7908551692962646, "learning_rate": 1.960345133863347e-06, "loss": 2.9671, "step": 311540 }, { "epoch": 0.013184, "grad_norm": 0.8507810831069946, "learning_rate": 1.959866280181456e-06, "loss": 2.7721, "step": 311550 }, { "epoch": 0.0132096, "grad_norm": 0.8105857968330383, "learning_rate": 1.959387478637694e-06, "loss": 2.7577, "step": 311560 }, { "epoch": 0.0132352, "grad_norm": 0.8140703439712524, "learning_rate": 1.9589087292351683e-06, "loss": 2.7948, "step": 311570 }, { "epoch": 0.0132608, "grad_norm": 0.8754380941390991, "learning_rate": 1.958430031976982e-06, "loss": 2.7143, "step": 311580 }, { "epoch": 0.0132864, "grad_norm": 0.8675410747528076, "learning_rate": 1.9579513868662405e-06, "loss": 2.7473, "step": 311590 }, { "epoch": 0.013312, "grad_norm": 0.8161908388137817, "learning_rate": 1.9574727939060455e-06, "loss": 2.8374, "step": 311600 }, { "epoch": 0.0133376, "grad_norm": 0.7722460031509399, "learning_rate": 1.9569942530995055e-06, "loss": 2.9264, "step": 311610 }, { "epoch": 0.0133632, "grad_norm": 0.8226944804191589, "learning_rate": 1.9565157644497167e-06, "loss": 2.8418, "step": 311620 }, { "epoch": 0.0133888, "grad_norm": 0.8734192252159119, "learning_rate": 1.9560373279597866e-06, "loss": 2.7402, "step": 311630 }, { "epoch": 0.0134144, "grad_norm": 1.0436222553253174, "learning_rate": 1.955558943632816e-06, "loss": 3.0187, "step": 311640 }, { "epoch": 0.01344, "grad_norm": 0.7453101873397827, "learning_rate": 1.9550806114719077e-06, "loss": 2.8699, "step": 311650 }, { "epoch": 0.0134656, "grad_norm": 0.8052358627319336, "learning_rate": 1.9546023314801643e-06, "loss": 2.8487, "step": 311660 }, { "epoch": 0.0134912, "grad_norm": 0.799410343170166, "learning_rate": 1.954124103660686e-06, "loss": 2.8015, "step": 311670 }, { "epoch": 0.0135168, "grad_norm": 0.8392194509506226, "learning_rate": 1.9536459280165764e-06, "loss": 2.7808, "step": 311680 }, { "epoch": 0.0135424, "grad_norm": 0.845655620098114, "learning_rate": 1.953167804550933e-06, "loss": 2.8518, "step": 311690 }, { "epoch": 0.013568, "grad_norm": 0.8778889775276184, "learning_rate": 1.952689733266857e-06, "loss": 2.7137, "step": 311700 }, { "epoch": 0.0135936, "grad_norm": 0.8305994272232056, "learning_rate": 1.95221171416745e-06, "loss": 2.8616, "step": 311710 }, { "epoch": 0.0136192, "grad_norm": 0.7648729085922241, "learning_rate": 1.9517337472558086e-06, "loss": 2.7935, "step": 311720 }, { "epoch": 0.0136448, "grad_norm": 0.9986700415611267, "learning_rate": 1.9512558325350374e-06, "loss": 2.7586, "step": 311730 }, { "epoch": 0.0136704, "grad_norm": 1.6158373355865479, "learning_rate": 1.9507779700082352e-06, "loss": 2.9159, "step": 311740 }, { "epoch": 0.013696, "grad_norm": 0.8822401165962219, "learning_rate": 1.9503001596784967e-06, "loss": 2.8057, "step": 311750 }, { "epoch": 0.0137216, "grad_norm": 0.8277402520179749, "learning_rate": 1.9498224015489208e-06, "loss": 2.9694, "step": 311760 }, { "epoch": 0.0137472, "grad_norm": 0.7554692625999451, "learning_rate": 1.949344695622609e-06, "loss": 2.8475, "step": 311770 }, { "epoch": 0.0137728, "grad_norm": 0.9750134348869324, "learning_rate": 1.9488670419026566e-06, "loss": 3.1234, "step": 311780 }, { "epoch": 0.0137984, "grad_norm": 0.8661219477653503, "learning_rate": 1.948389440392161e-06, "loss": 3.0588, "step": 311790 }, { "epoch": 0.013824, "grad_norm": 0.8259621858596802, "learning_rate": 1.9479118910942206e-06, "loss": 3.0243, "step": 311800 }, { "epoch": 2.56e-05, "grad_norm": 0.8670141100883484, "learning_rate": 1.9474343940119345e-06, "loss": 1.8613, "step": 311810 }, { "epoch": 5.12e-05, "grad_norm": 0.8694731593132019, "learning_rate": 1.946956949148393e-06, "loss": 2.6554, "step": 311820 }, { "epoch": 7.68e-05, "grad_norm": 0.8020718693733215, "learning_rate": 1.9464795565066962e-06, "loss": 2.719, "step": 311830 }, { "epoch": 0.0001024, "grad_norm": 0.742582380771637, "learning_rate": 1.9460022160899385e-06, "loss": 2.43, "step": 311840 }, { "epoch": 0.000128, "grad_norm": 0.8371456265449524, "learning_rate": 1.9455249279012157e-06, "loss": 2.6119, "step": 311850 }, { "epoch": 0.0001536, "grad_norm": 0.8961350321769714, "learning_rate": 1.9450476919436234e-06, "loss": 2.4703, "step": 311860 }, { "epoch": 0.0001792, "grad_norm": 0.8790315389633179, "learning_rate": 1.944570508220257e-06, "loss": 2.7279, "step": 311870 }, { "epoch": 0.0002048, "grad_norm": 0.88679039478302, "learning_rate": 1.944093376734212e-06, "loss": 2.4913, "step": 311880 }, { "epoch": 0.0002304, "grad_norm": 0.9221035838127136, "learning_rate": 1.943616297488574e-06, "loss": 2.9056, "step": 311890 }, { "epoch": 0.000256, "grad_norm": 0.8906496167182922, "learning_rate": 1.9431392704864472e-06, "loss": 2.6625, "step": 311900 }, { "epoch": 0.0002816, "grad_norm": 0.8276948928833008, "learning_rate": 1.9426622957309206e-06, "loss": 2.6451, "step": 311910 }, { "epoch": 0.0003072, "grad_norm": 0.8568216562271118, "learning_rate": 1.942185373225087e-06, "loss": 2.4258, "step": 311920 }, { "epoch": 0.0003328, "grad_norm": 0.8615198135375977, "learning_rate": 1.9417085029720407e-06, "loss": 2.3436, "step": 311930 }, { "epoch": 0.0003584, "grad_norm": 1.2077940702438354, "learning_rate": 1.941231684974876e-06, "loss": 2.8997, "step": 311940 }, { "epoch": 0.000384, "grad_norm": 0.8106172680854797, "learning_rate": 1.940754919236678e-06, "loss": 2.4629, "step": 311950 }, { "epoch": 0.0004096, "grad_norm": 1.0737413167953491, "learning_rate": 1.940278205760544e-06, "loss": 2.71, "step": 311960 }, { "epoch": 0.0004352, "grad_norm": 0.7699894309043884, "learning_rate": 1.939801544549563e-06, "loss": 2.7388, "step": 311970 }, { "epoch": 0.0004608, "grad_norm": 0.8182898163795471, "learning_rate": 1.939324935606827e-06, "loss": 2.4747, "step": 311980 }, { "epoch": 0.0004864, "grad_norm": 0.8514600396156311, "learning_rate": 1.9388483789354263e-06, "loss": 2.7756, "step": 311990 }, { "epoch": 0.000512, "grad_norm": 0.8123248815536499, "learning_rate": 1.938371874538452e-06, "loss": 2.8829, "step": 312000 }, { "epoch": 0.0005376, "grad_norm": 0.9165651202201843, "learning_rate": 1.937895422418995e-06, "loss": 2.5225, "step": 312010 }, { "epoch": 0.0005632, "grad_norm": 0.7798178195953369, "learning_rate": 1.9374190225801427e-06, "loss": 2.6835, "step": 312020 }, { "epoch": 0.0005888, "grad_norm": 0.9307343363761902, "learning_rate": 1.936942675024984e-06, "loss": 2.5914, "step": 312030 }, { "epoch": 0.0006144, "grad_norm": 0.8011735081672668, "learning_rate": 1.9364663797566098e-06, "loss": 2.5662, "step": 312040 }, { "epoch": 0.00064, "grad_norm": 0.8712642192840576, "learning_rate": 1.9359901367781077e-06, "loss": 2.8804, "step": 312050 }, { "epoch": 0.0006656, "grad_norm": 0.8728945851325989, "learning_rate": 1.9355139460925654e-06, "loss": 2.9963, "step": 312060 }, { "epoch": 0.0006912, "grad_norm": 0.8519081473350525, "learning_rate": 1.935037807703073e-06, "loss": 2.8381, "step": 312070 }, { "epoch": 0.0007168, "grad_norm": 0.8075392246246338, "learning_rate": 1.934561721612718e-06, "loss": 2.6482, "step": 312080 }, { "epoch": 0.0007424, "grad_norm": 0.9302177429199219, "learning_rate": 1.934085687824585e-06, "loss": 2.7565, "step": 312090 }, { "epoch": 0.000768, "grad_norm": 0.8660354018211365, "learning_rate": 1.9336097063417637e-06, "loss": 2.6443, "step": 312100 }, { "epoch": 0.0007936, "grad_norm": 0.778869092464447, "learning_rate": 1.9331337771673388e-06, "loss": 2.5987, "step": 312110 }, { "epoch": 0.0008192, "grad_norm": 0.8530387282371521, "learning_rate": 1.9326579003043977e-06, "loss": 2.5859, "step": 312120 }, { "epoch": 0.0008448, "grad_norm": 0.770220160484314, "learning_rate": 1.9321820757560262e-06, "loss": 2.5545, "step": 312130 }, { "epoch": 0.0008704, "grad_norm": 0.8739571571350098, "learning_rate": 1.9317063035253126e-06, "loss": 2.5084, "step": 312140 }, { "epoch": 0.000896, "grad_norm": 0.756859540939331, "learning_rate": 1.9312305836153367e-06, "loss": 2.2438, "step": 312150 }, { "epoch": 0.0009216, "grad_norm": 0.9155511856079102, "learning_rate": 1.930754916029186e-06, "loss": 2.6936, "step": 312160 }, { "epoch": 0.0009472, "grad_norm": 0.8002262711524963, "learning_rate": 1.930279300769945e-06, "loss": 2.6589, "step": 312170 }, { "epoch": 0.0009728, "grad_norm": 0.8568936586380005, "learning_rate": 1.9298037378406976e-06, "loss": 2.7186, "step": 312180 }, { "epoch": 0.0009984, "grad_norm": 0.8408070206642151, "learning_rate": 1.929328227244529e-06, "loss": 2.5179, "step": 312190 }, { "epoch": 0.001024, "grad_norm": 0.9252229332923889, "learning_rate": 1.9288527689845217e-06, "loss": 2.5076, "step": 312200 }, { "epoch": 0.0010496, "grad_norm": 0.8515708446502686, "learning_rate": 1.9283773630637614e-06, "loss": 2.526, "step": 312210 }, { "epoch": 0.0010752, "grad_norm": 0.7995741367340088, "learning_rate": 1.9279020094853264e-06, "loss": 2.7905, "step": 312220 }, { "epoch": 0.0011008, "grad_norm": 0.8556550145149231, "learning_rate": 1.9274267082523013e-06, "loss": 2.6297, "step": 312230 }, { "epoch": 0.0011264, "grad_norm": 0.8111200928688049, "learning_rate": 1.926951459367766e-06, "loss": 2.55, "step": 312240 }, { "epoch": 0.001152, "grad_norm": 0.9316439032554626, "learning_rate": 1.926476262834808e-06, "loss": 2.9944, "step": 312250 }, { "epoch": 0.0011776, "grad_norm": 0.7961824536323547, "learning_rate": 1.9260011186565055e-06, "loss": 2.5754, "step": 312260 }, { "epoch": 0.0012032, "grad_norm": 0.798609733581543, "learning_rate": 1.925526026835942e-06, "loss": 2.7507, "step": 312270 }, { "epoch": 0.0012288, "grad_norm": 0.7781262397766113, "learning_rate": 1.9250509873761946e-06, "loss": 2.5709, "step": 312280 }, { "epoch": 0.0012544, "grad_norm": 0.9394285678863525, "learning_rate": 1.924576000280345e-06, "loss": 2.9116, "step": 312290 }, { "epoch": 0.00128, "grad_norm": 0.8836509585380554, "learning_rate": 1.9241010655514734e-06, "loss": 2.7582, "step": 312300 }, { "epoch": 0.0013056, "grad_norm": 0.8277537226676941, "learning_rate": 1.9236261831926605e-06, "loss": 2.5665, "step": 312310 }, { "epoch": 0.0013312, "grad_norm": 0.8271400332450867, "learning_rate": 1.9231513532069857e-06, "loss": 2.8148, "step": 312320 }, { "epoch": 0.0013568, "grad_norm": 0.8246396780014038, "learning_rate": 1.9226765755975276e-06, "loss": 2.5332, "step": 312330 }, { "epoch": 0.0013824, "grad_norm": 0.9131674766540527, "learning_rate": 1.9222018503673676e-06, "loss": 2.5419, "step": 312340 }, { "epoch": 0.001408, "grad_norm": 0.8574308156967163, "learning_rate": 1.921727177519579e-06, "loss": 2.8552, "step": 312350 }, { "epoch": 0.0014336, "grad_norm": 0.9053831696510315, "learning_rate": 1.921252557057243e-06, "loss": 2.5673, "step": 312360 }, { "epoch": 0.0014592, "grad_norm": 0.8851905465126038, "learning_rate": 1.9207779889834375e-06, "loss": 2.4488, "step": 312370 }, { "epoch": 0.0014848, "grad_norm": 0.8677626848220825, "learning_rate": 1.9203034733012395e-06, "loss": 2.6087, "step": 312380 }, { "epoch": 0.0015104, "grad_norm": 1.1060504913330078, "learning_rate": 1.919829010013725e-06, "loss": 2.2253, "step": 312390 }, { "epoch": 0.001536, "grad_norm": 0.7879648804664612, "learning_rate": 1.9193545991239725e-06, "loss": 2.1336, "step": 312400 }, { "epoch": 0.0015616, "grad_norm": 1.227831482887268, "learning_rate": 1.918880240635058e-06, "loss": 2.865, "step": 312410 }, { "epoch": 0.0015872, "grad_norm": 0.8823519945144653, "learning_rate": 1.9184059345500584e-06, "loss": 2.8044, "step": 312420 }, { "epoch": 0.0016128, "grad_norm": 0.8354253172874451, "learning_rate": 1.917931680872047e-06, "loss": 2.785, "step": 312430 }, { "epoch": 0.0016384, "grad_norm": 0.8455601334571838, "learning_rate": 1.9174574796041012e-06, "loss": 2.72, "step": 312440 }, { "epoch": 0.001664, "grad_norm": 0.9125159978866577, "learning_rate": 1.916983330749296e-06, "loss": 2.6994, "step": 312450 }, { "epoch": 0.0016896, "grad_norm": 0.8273206353187561, "learning_rate": 1.9165092343107074e-06, "loss": 2.5907, "step": 312460 }, { "epoch": 0.0017152, "grad_norm": 0.9937936067581177, "learning_rate": 1.9160351902914064e-06, "loss": 2.6338, "step": 312470 }, { "epoch": 0.0017408, "grad_norm": 0.929022490978241, "learning_rate": 1.915561198694468e-06, "loss": 2.6195, "step": 312480 }, { "epoch": 0.0017664, "grad_norm": 1.067919373512268, "learning_rate": 1.9150872595229676e-06, "loss": 2.8158, "step": 312490 }, { "epoch": 0.001792, "grad_norm": 0.8355890512466431, "learning_rate": 1.9146133727799764e-06, "loss": 2.7167, "step": 312500 }, { "epoch": 2.56e-05, "grad_norm": 0.9031528830528259, "learning_rate": 1.914139538468569e-06, "loss": 1.948, "step": 312510 }, { "epoch": 5.12e-05, "grad_norm": 0.8230396509170532, "learning_rate": 1.9136657565918184e-06, "loss": 2.7421, "step": 312520 }, { "epoch": 7.68e-05, "grad_norm": 0.8967155814170837, "learning_rate": 1.913192027152796e-06, "loss": 2.6811, "step": 312530 }, { "epoch": 0.0001024, "grad_norm": 0.7725256681442261, "learning_rate": 1.9127183501545765e-06, "loss": 2.5802, "step": 312540 }, { "epoch": 0.000128, "grad_norm": 0.8419419527053833, "learning_rate": 1.912244725600226e-06, "loss": 2.7679, "step": 312550 }, { "epoch": 0.0001536, "grad_norm": 0.8854449987411499, "learning_rate": 1.9117711534928198e-06, "loss": 2.6631, "step": 312560 }, { "epoch": 0.0001792, "grad_norm": 0.8299830555915833, "learning_rate": 1.911297633835428e-06, "loss": 2.5995, "step": 312570 }, { "epoch": 0.0002048, "grad_norm": 0.8058999180793762, "learning_rate": 1.910824166631121e-06, "loss": 2.635, "step": 312580 }, { "epoch": 0.0002304, "grad_norm": 0.9590141177177429, "learning_rate": 1.91035075188297e-06, "loss": 2.7398, "step": 312590 }, { "epoch": 0.000256, "grad_norm": 0.8301287293434143, "learning_rate": 1.909877389594044e-06, "loss": 2.5698, "step": 312600 }, { "epoch": 0.0002816, "grad_norm": 0.8698163628578186, "learning_rate": 1.909404079767413e-06, "loss": 2.8751, "step": 312610 }, { "epoch": 0.0003072, "grad_norm": 1.67957603931427, "learning_rate": 1.9089308224061454e-06, "loss": 2.4409, "step": 312620 }, { "epoch": 0.0003328, "grad_norm": 0.8647544980049133, "learning_rate": 1.908457617513313e-06, "loss": 2.9916, "step": 312630 }, { "epoch": 0.0003584, "grad_norm": 0.8451924324035645, "learning_rate": 1.907984465091981e-06, "loss": 2.5491, "step": 312640 }, { "epoch": 0.000384, "grad_norm": 0.7822484970092773, "learning_rate": 1.907511365145219e-06, "loss": 2.404, "step": 312650 }, { "epoch": 0.0004096, "grad_norm": 0.9362528920173645, "learning_rate": 1.9070383176760986e-06, "loss": 2.7487, "step": 312660 }, { "epoch": 0.0004352, "grad_norm": 0.7671499848365784, "learning_rate": 1.9065653226876811e-06, "loss": 2.6497, "step": 312670 }, { "epoch": 0.0004608, "grad_norm": 0.9459179043769836, "learning_rate": 1.9060923801830366e-06, "loss": 2.5766, "step": 312680 }, { "epoch": 0.0004864, "grad_norm": 0.864764928817749, "learning_rate": 1.9056194901652313e-06, "loss": 2.469, "step": 312690 }, { "epoch": 0.000512, "grad_norm": 0.8890225291252136, "learning_rate": 1.905146652637333e-06, "loss": 2.5462, "step": 312700 }, { "epoch": 0.0005376, "grad_norm": 0.9182993769645691, "learning_rate": 1.9046738676024068e-06, "loss": 2.4994, "step": 312710 }, { "epoch": 0.0005632, "grad_norm": 0.7801226377487183, "learning_rate": 1.9042011350635193e-06, "loss": 2.4754, "step": 312720 }, { "epoch": 0.0005888, "grad_norm": 0.945050835609436, "learning_rate": 1.9037284550237367e-06, "loss": 2.5787, "step": 312730 }, { "epoch": 0.0006144, "grad_norm": 0.8431473970413208, "learning_rate": 1.9032558274861245e-06, "loss": 2.7798, "step": 312740 }, { "epoch": 0.00064, "grad_norm": 0.9197422862052917, "learning_rate": 1.9027832524537448e-06, "loss": 2.6929, "step": 312750 }, { "epoch": 0.0006656, "grad_norm": 0.8239476680755615, "learning_rate": 1.9023107299296606e-06, "loss": 2.6671, "step": 312760 }, { "epoch": 0.0006912, "grad_norm": 0.9198643565177917, "learning_rate": 1.9018382599169417e-06, "loss": 2.5954, "step": 312770 }, { "epoch": 0.0007168, "grad_norm": 0.949392557144165, "learning_rate": 1.90136584241865e-06, "loss": 2.5881, "step": 312780 }, { "epoch": 0.0007424, "grad_norm": 0.8684922456741333, "learning_rate": 1.9008934774378506e-06, "loss": 2.5937, "step": 312790 }, { "epoch": 0.000768, "grad_norm": 0.8146740794181824, "learning_rate": 1.9004211649776015e-06, "loss": 2.5794, "step": 312800 }, { "epoch": 0.0007936, "grad_norm": 0.8021025657653809, "learning_rate": 1.8999489050409691e-06, "loss": 2.6416, "step": 312810 }, { "epoch": 0.0008192, "grad_norm": 0.8509665131568909, "learning_rate": 1.8994766976310163e-06, "loss": 2.6118, "step": 312820 }, { "epoch": 0.0008448, "grad_norm": 0.7875211238861084, "learning_rate": 1.8990045427508031e-06, "loss": 2.5843, "step": 312830 }, { "epoch": 0.0008704, "grad_norm": 0.9740704298019409, "learning_rate": 1.8985324404033923e-06, "loss": 2.7997, "step": 312840 }, { "epoch": 0.000896, "grad_norm": 0.8266546726226807, "learning_rate": 1.898060390591846e-06, "loss": 2.4498, "step": 312850 }, { "epoch": 0.0009216, "grad_norm": 0.7411977052688599, "learning_rate": 1.8975883933192274e-06, "loss": 2.4271, "step": 312860 }, { "epoch": 0.0009472, "grad_norm": 0.8238570094108582, "learning_rate": 1.8971164485885917e-06, "loss": 2.6533, "step": 312870 }, { "epoch": 0.0009728, "grad_norm": 0.9344737529754639, "learning_rate": 1.8966445564030022e-06, "loss": 2.4942, "step": 312880 }, { "epoch": 0.0009984, "grad_norm": 0.7452187538146973, "learning_rate": 1.8961727167655197e-06, "loss": 2.5368, "step": 312890 }, { "epoch": 0.001024, "grad_norm": 0.9639459848403931, "learning_rate": 1.895700929679204e-06, "loss": 2.6013, "step": 312900 }, { "epoch": 0.0010496, "grad_norm": 0.9737640023231506, "learning_rate": 1.8952291951471125e-06, "loss": 2.9141, "step": 312910 }, { "epoch": 0.0010752, "grad_norm": 0.8501589894294739, "learning_rate": 1.8947575131723062e-06, "loss": 2.5694, "step": 312920 }, { "epoch": 0.0011008, "grad_norm": 0.8811014294624329, "learning_rate": 1.8942858837578437e-06, "loss": 2.6556, "step": 312930 }, { "epoch": 0.0011264, "grad_norm": 0.9127395153045654, "learning_rate": 1.8938143069067826e-06, "loss": 2.617, "step": 312940 }, { "epoch": 0.001152, "grad_norm": 0.8630660176277161, "learning_rate": 1.8933427826221818e-06, "loss": 2.7541, "step": 312950 }, { "epoch": 0.0011776, "grad_norm": 0.965668797492981, "learning_rate": 1.8928713109070984e-06, "loss": 2.7726, "step": 312960 }, { "epoch": 0.0012032, "grad_norm": 0.8709150552749634, "learning_rate": 1.8923998917645891e-06, "loss": 2.4972, "step": 312970 }, { "epoch": 0.0012288, "grad_norm": 0.8688748478889465, "learning_rate": 1.8919285251977126e-06, "loss": 3.0494, "step": 312980 }, { "epoch": 0.0012544, "grad_norm": 0.8384323120117188, "learning_rate": 1.8914572112095275e-06, "loss": 2.521, "step": 312990 }, { "epoch": 0.00128, "grad_norm": 0.8349071145057678, "learning_rate": 1.8909859498030848e-06, "loss": 2.5838, "step": 313000 }, { "epoch": 0.0013056, "grad_norm": 0.879439651966095, "learning_rate": 1.890514740981443e-06, "loss": 2.642, "step": 313010 }, { "epoch": 0.0013312, "grad_norm": 0.9036542177200317, "learning_rate": 1.8900435847476572e-06, "loss": 2.4767, "step": 313020 }, { "epoch": 0.0013568, "grad_norm": 0.9156922698020935, "learning_rate": 1.8895724811047844e-06, "loss": 2.7513, "step": 313030 }, { "epoch": 0.0013824, "grad_norm": 0.8499745726585388, "learning_rate": 1.8891014300558774e-06, "loss": 2.7771, "step": 313040 }, { "epoch": 0.001408, "grad_norm": 0.8046078085899353, "learning_rate": 1.8886304316039927e-06, "loss": 2.5165, "step": 313050 }, { "epoch": 0.0014336, "grad_norm": 0.7928751707077026, "learning_rate": 1.8881594857521867e-06, "loss": 2.7318, "step": 313060 }, { "epoch": 0.0014592, "grad_norm": 0.7813946604728699, "learning_rate": 1.8876885925035072e-06, "loss": 2.4299, "step": 313070 }, { "epoch": 0.0014848, "grad_norm": 0.9192646145820618, "learning_rate": 1.8872177518610124e-06, "loss": 2.7255, "step": 313080 }, { "epoch": 0.0015104, "grad_norm": 0.8907982110977173, "learning_rate": 1.8867469638277535e-06, "loss": 2.1936, "step": 313090 }, { "epoch": 0.001536, "grad_norm": 0.7980818748474121, "learning_rate": 1.8862762284067838e-06, "loss": 2.6783, "step": 313100 }, { "epoch": 2.56e-05, "grad_norm": 0.8386715054512024, "learning_rate": 1.8858055456011549e-06, "loss": 1.8329, "step": 313110 }, { "epoch": 5.12e-05, "grad_norm": 0.8681167960166931, "learning_rate": 1.8853349154139256e-06, "loss": 2.7195, "step": 313120 }, { "epoch": 7.68e-05, "grad_norm": 0.8355708718299866, "learning_rate": 1.884864337848139e-06, "loss": 2.7587, "step": 313130 }, { "epoch": 0.0001024, "grad_norm": 0.8708471059799194, "learning_rate": 1.8843938129068518e-06, "loss": 2.6747, "step": 313140 }, { "epoch": 0.000128, "grad_norm": 0.823701798915863, "learning_rate": 1.8839233405931135e-06, "loss": 2.445, "step": 313150 }, { "epoch": 0.0001536, "grad_norm": 0.8568058013916016, "learning_rate": 1.883452920909975e-06, "loss": 2.581, "step": 313160 }, { "epoch": 0.0001792, "grad_norm": 0.8924545049667358, "learning_rate": 1.8829825538604874e-06, "loss": 2.8241, "step": 313170 }, { "epoch": 0.0002048, "grad_norm": 0.9641633033752441, "learning_rate": 1.8825122394477003e-06, "loss": 2.9219, "step": 313180 }, { "epoch": 0.0002304, "grad_norm": 0.8986167907714844, "learning_rate": 1.8820419776746669e-06, "loss": 2.6924, "step": 313190 }, { "epoch": 0.000256, "grad_norm": 0.9726510643959045, "learning_rate": 1.8815717685444313e-06, "loss": 2.407, "step": 313200 }, { "epoch": 0.0002816, "grad_norm": 0.8969249725341797, "learning_rate": 1.8811016120600445e-06, "loss": 2.6383, "step": 313210 }, { "epoch": 0.0003072, "grad_norm": 0.8300408124923706, "learning_rate": 1.8806315082245563e-06, "loss": 2.534, "step": 313220 }, { "epoch": 0.0003328, "grad_norm": 0.8365843892097473, "learning_rate": 1.8801614570410153e-06, "loss": 2.6402, "step": 313230 }, { "epoch": 0.0003584, "grad_norm": 0.9475532174110413, "learning_rate": 1.8796914585124692e-06, "loss": 2.9449, "step": 313240 }, { "epoch": 0.000384, "grad_norm": 0.8793221116065979, "learning_rate": 1.8792215126419655e-06, "loss": 2.5166, "step": 313250 }, { "epoch": 0.0004096, "grad_norm": 1.1813982725143433, "learning_rate": 1.878751619432555e-06, "loss": 2.7809, "step": 313260 }, { "epoch": 0.0004352, "grad_norm": 0.8290350437164307, "learning_rate": 1.8782817788872798e-06, "loss": 2.5532, "step": 313270 }, { "epoch": 0.0004608, "grad_norm": 0.813283383846283, "learning_rate": 1.8778119910091853e-06, "loss": 2.654, "step": 313280 }, { "epoch": 0.0004864, "grad_norm": 0.9731194972991943, "learning_rate": 1.8773422558013244e-06, "loss": 2.7702, "step": 313290 }, { "epoch": 0.000512, "grad_norm": 1.0179015398025513, "learning_rate": 1.8768725732667403e-06, "loss": 2.6163, "step": 313300 }, { "epoch": 0.0005376, "grad_norm": 0.952828586101532, "learning_rate": 1.8764029434084786e-06, "loss": 2.6345, "step": 313310 }, { "epoch": 0.0005632, "grad_norm": 0.8256471157073975, "learning_rate": 1.8759333662295864e-06, "loss": 2.5611, "step": 313320 }, { "epoch": 0.0005888, "grad_norm": 0.7850743532180786, "learning_rate": 1.875463841733106e-06, "loss": 2.4919, "step": 313330 }, { "epoch": 0.0006144, "grad_norm": 0.8623831272125244, "learning_rate": 1.8749943699220829e-06, "loss": 2.5719, "step": 313340 }, { "epoch": 0.00064, "grad_norm": 0.8999459147453308, "learning_rate": 1.874524950799561e-06, "loss": 2.8514, "step": 313350 }, { "epoch": 0.0006656, "grad_norm": 0.8599634766578674, "learning_rate": 1.874055584368587e-06, "loss": 2.9249, "step": 313360 }, { "epoch": 0.0006912, "grad_norm": 0.8635157346725464, "learning_rate": 1.8735862706322016e-06, "loss": 2.8682, "step": 313370 }, { "epoch": 0.0007168, "grad_norm": 0.8037906885147095, "learning_rate": 1.8731170095934504e-06, "loss": 2.4647, "step": 313380 }, { "epoch": 0.0007424, "grad_norm": 0.8748177886009216, "learning_rate": 1.8726478012553772e-06, "loss": 3.0209, "step": 313390 }, { "epoch": 0.000768, "grad_norm": 0.8343027830123901, "learning_rate": 1.8721786456210212e-06, "loss": 2.8838, "step": 313400 }, { "epoch": 0.0007936, "grad_norm": 0.8479355573654175, "learning_rate": 1.8717095426934262e-06, "loss": 2.5384, "step": 313410 }, { "epoch": 0.0008192, "grad_norm": 0.8194698095321655, "learning_rate": 1.8712404924756345e-06, "loss": 2.6038, "step": 313420 }, { "epoch": 0.0008448, "grad_norm": 0.9899421334266663, "learning_rate": 1.870771494970689e-06, "loss": 2.6346, "step": 313430 }, { "epoch": 0.0008704, "grad_norm": 0.7884153723716736, "learning_rate": 1.8703025501816285e-06, "loss": 2.538, "step": 313440 }, { "epoch": 0.000896, "grad_norm": 0.8578671813011169, "learning_rate": 1.8698336581114962e-06, "loss": 2.7081, "step": 313450 }, { "epoch": 0.0009216, "grad_norm": 0.9087773561477661, "learning_rate": 1.8693648187633308e-06, "loss": 2.6267, "step": 313460 }, { "epoch": 0.0009472, "grad_norm": 0.8993710279464722, "learning_rate": 1.8688960321401738e-06, "loss": 2.7336, "step": 313470 }, { "epoch": 0.0009728, "grad_norm": 0.9041215181350708, "learning_rate": 1.8684272982450658e-06, "loss": 2.4923, "step": 313480 }, { "epoch": 0.0009984, "grad_norm": 1.0397014617919922, "learning_rate": 1.8679586170810459e-06, "loss": 2.7203, "step": 313490 }, { "epoch": 0.001024, "grad_norm": 0.9867084622383118, "learning_rate": 1.8674899886511521e-06, "loss": 2.7817, "step": 313500 }, { "epoch": 0.0010496, "grad_norm": 0.8684403300285339, "learning_rate": 1.8670214129584252e-06, "loss": 2.554, "step": 313510 }, { "epoch": 0.0010752, "grad_norm": 0.8065489530563354, "learning_rate": 1.866552890005905e-06, "loss": 2.5909, "step": 313520 }, { "epoch": 0.0011008, "grad_norm": 0.8496084213256836, "learning_rate": 1.8660844197966255e-06, "loss": 2.5577, "step": 313530 }, { "epoch": 0.0011264, "grad_norm": 0.8058114051818848, "learning_rate": 1.8656160023336268e-06, "loss": 2.465, "step": 313540 }, { "epoch": 0.001152, "grad_norm": 0.8304267525672913, "learning_rate": 1.8651476376199462e-06, "loss": 2.4202, "step": 313550 }, { "epoch": 0.0011776, "grad_norm": 0.9116355180740356, "learning_rate": 1.8646793256586215e-06, "loss": 2.8867, "step": 313560 }, { "epoch": 0.0012032, "grad_norm": 0.8994553089141846, "learning_rate": 1.8642110664526891e-06, "loss": 2.7593, "step": 313570 }, { "epoch": 0.0012288, "grad_norm": 0.8668369650840759, "learning_rate": 1.8637428600051854e-06, "loss": 2.7174, "step": 313580 }, { "epoch": 0.0012544, "grad_norm": 0.8261486291885376, "learning_rate": 1.8632747063191503e-06, "loss": 2.6078, "step": 313590 }, { "epoch": 0.00128, "grad_norm": 0.9219560623168945, "learning_rate": 1.8628066053976124e-06, "loss": 2.6078, "step": 313600 }, { "epoch": 0.0013056, "grad_norm": 0.8818285465240479, "learning_rate": 1.8623385572436125e-06, "loss": 2.7724, "step": 313610 }, { "epoch": 0.0013312, "grad_norm": 0.8777710199356079, "learning_rate": 1.8618705618601829e-06, "loss": 2.5995, "step": 313620 }, { "epoch": 0.0013568, "grad_norm": 1.0114874839782715, "learning_rate": 1.8614026192503587e-06, "loss": 2.7072, "step": 313630 }, { "epoch": 0.0013824, "grad_norm": 1.0770162343978882, "learning_rate": 1.8609347294171776e-06, "loss": 2.8837, "step": 313640 }, { "epoch": 0.001408, "grad_norm": 0.8214699029922485, "learning_rate": 1.8604668923636738e-06, "loss": 2.6813, "step": 313650 }, { "epoch": 0.0014336, "grad_norm": 1.0038880109786987, "learning_rate": 1.8599991080928758e-06, "loss": 2.7308, "step": 313660 }, { "epoch": 0.0014592, "grad_norm": 0.8267102837562561, "learning_rate": 1.8595313766078216e-06, "loss": 2.4702, "step": 313670 }, { "epoch": 0.0014848, "grad_norm": 0.7972059845924377, "learning_rate": 1.8590636979115428e-06, "loss": 2.6129, "step": 313680 }, { "epoch": 0.0015104, "grad_norm": 1.1062313318252563, "learning_rate": 1.8585960720070717e-06, "loss": 2.4785, "step": 313690 }, { "epoch": 0.001536, "grad_norm": 0.8298373222351074, "learning_rate": 1.8581284988974412e-06, "loss": 2.754, "step": 313700 }, { "epoch": 2.56e-05, "grad_norm": 0.8775140643119812, "learning_rate": 1.8576609785856836e-06, "loss": 1.7803, "step": 313710 }, { "epoch": 5.12e-05, "grad_norm": 0.8871397972106934, "learning_rate": 1.857193511074834e-06, "loss": 2.6452, "step": 313720 }, { "epoch": 7.68e-05, "grad_norm": 0.8657525181770325, "learning_rate": 1.8567260963679168e-06, "loss": 2.6918, "step": 313730 }, { "epoch": 0.0001024, "grad_norm": 0.9504308700561523, "learning_rate": 1.8562587344679672e-06, "loss": 2.527, "step": 313740 }, { "epoch": 0.000128, "grad_norm": 0.8875794410705566, "learning_rate": 1.8557914253780151e-06, "loss": 2.604, "step": 313750 }, { "epoch": 0.0001536, "grad_norm": 0.8580014705657959, "learning_rate": 1.8553241691010914e-06, "loss": 2.8598, "step": 313760 }, { "epoch": 0.0001792, "grad_norm": 0.8339418768882751, "learning_rate": 1.854856965640226e-06, "loss": 2.6421, "step": 313770 }, { "epoch": 0.0002048, "grad_norm": 0.9081927537918091, "learning_rate": 1.8543898149984484e-06, "loss": 2.67, "step": 313780 }, { "epoch": 0.0002304, "grad_norm": 0.9200825095176697, "learning_rate": 1.8539227171787899e-06, "loss": 2.6616, "step": 313790 }, { "epoch": 0.000256, "grad_norm": 1.0427550077438354, "learning_rate": 1.8534556721842757e-06, "loss": 2.6642, "step": 313800 }, { "epoch": 0.0002816, "grad_norm": 0.8982747197151184, "learning_rate": 1.8529886800179341e-06, "loss": 2.559, "step": 313810 }, { "epoch": 0.0003072, "grad_norm": 0.7550155520439148, "learning_rate": 1.8525217406827978e-06, "loss": 2.4878, "step": 313820 }, { "epoch": 0.0003328, "grad_norm": 0.7717707753181458, "learning_rate": 1.8520548541818929e-06, "loss": 2.8028, "step": 313830 }, { "epoch": 0.0003584, "grad_norm": 0.8365904688835144, "learning_rate": 1.8515880205182469e-06, "loss": 2.6935, "step": 313840 }, { "epoch": 0.000384, "grad_norm": 0.9797748923301697, "learning_rate": 1.8511212396948885e-06, "loss": 2.5534, "step": 313850 }, { "epoch": 0.0004096, "grad_norm": 0.8116816878318787, "learning_rate": 1.8506545117148411e-06, "loss": 2.7928, "step": 313860 }, { "epoch": 0.0004352, "grad_norm": 0.9254499673843384, "learning_rate": 1.8501878365811343e-06, "loss": 2.855, "step": 313870 }, { "epoch": 0.0004608, "grad_norm": 0.8153303265571594, "learning_rate": 1.8497212142967924e-06, "loss": 2.5792, "step": 313880 }, { "epoch": 0.0004864, "grad_norm": 0.8986954689025879, "learning_rate": 1.849254644864843e-06, "loss": 2.6906, "step": 313890 }, { "epoch": 0.000512, "grad_norm": 0.809910774230957, "learning_rate": 1.8487881282883102e-06, "loss": 2.5573, "step": 313900 }, { "epoch": 0.0005376, "grad_norm": 0.86641526222229, "learning_rate": 1.8483216645702206e-06, "loss": 2.6142, "step": 313910 }, { "epoch": 0.0005632, "grad_norm": 0.8159860372543335, "learning_rate": 1.8478552537136008e-06, "loss": 2.6899, "step": 313920 }, { "epoch": 0.0005888, "grad_norm": 0.773364782333374, "learning_rate": 1.8473888957214715e-06, "loss": 2.6211, "step": 313930 }, { "epoch": 0.0006144, "grad_norm": 0.8749625086784363, "learning_rate": 1.846922590596857e-06, "loss": 2.5559, "step": 313940 }, { "epoch": 0.00064, "grad_norm": 0.8794748783111572, "learning_rate": 1.8464563383427837e-06, "loss": 2.5858, "step": 313950 }, { "epoch": 0.0006656, "grad_norm": 0.8220847249031067, "learning_rate": 1.8459901389622737e-06, "loss": 2.7206, "step": 313960 }, { "epoch": 0.0006912, "grad_norm": 0.8769067525863647, "learning_rate": 1.8455239924583513e-06, "loss": 2.4971, "step": 313970 }, { "epoch": 0.0007168, "grad_norm": 0.9525285363197327, "learning_rate": 1.8450578988340372e-06, "loss": 2.6351, "step": 313980 }, { "epoch": 0.0007424, "grad_norm": 0.8422467112541199, "learning_rate": 1.8445918580923573e-06, "loss": 2.4289, "step": 313990 }, { "epoch": 0.000768, "grad_norm": 0.8602919578552246, "learning_rate": 1.8441258702363297e-06, "loss": 2.7033, "step": 314000 }, { "epoch": 0.0007936, "grad_norm": 0.9189961552619934, "learning_rate": 1.8436599352689798e-06, "loss": 2.777, "step": 314010 }, { "epoch": 0.0008192, "grad_norm": 0.801362156867981, "learning_rate": 1.8431940531933267e-06, "loss": 2.2555, "step": 314020 }, { "epoch": 0.0008448, "grad_norm": 0.8445404171943665, "learning_rate": 1.8427282240123924e-06, "loss": 2.5931, "step": 314030 }, { "epoch": 0.0008704, "grad_norm": 0.8474922776222229, "learning_rate": 1.8422624477291973e-06, "loss": 2.8831, "step": 314040 }, { "epoch": 0.000896, "grad_norm": 0.84443598985672, "learning_rate": 1.841796724346766e-06, "loss": 2.5661, "step": 314050 }, { "epoch": 0.0009216, "grad_norm": 0.8309491276741028, "learning_rate": 1.8413310538681107e-06, "loss": 2.578, "step": 314060 }, { "epoch": 0.0009472, "grad_norm": 0.876924455165863, "learning_rate": 1.8408654362962563e-06, "loss": 2.8728, "step": 314070 }, { "epoch": 0.0009728, "grad_norm": 0.8261405825614929, "learning_rate": 1.8403998716342208e-06, "loss": 2.7017, "step": 314080 }, { "epoch": 0.0009984, "grad_norm": 0.7838614583015442, "learning_rate": 1.8399343598850228e-06, "loss": 2.7066, "step": 314090 }, { "epoch": 0.001024, "grad_norm": 0.7582070231437683, "learning_rate": 1.839468901051683e-06, "loss": 2.5925, "step": 314100 }, { "epoch": 0.0010496, "grad_norm": 0.8053703904151917, "learning_rate": 1.8390034951372172e-06, "loss": 2.609, "step": 314110 }, { "epoch": 0.0010752, "grad_norm": 0.9154788255691528, "learning_rate": 1.838538142144649e-06, "loss": 2.845, "step": 314120 }, { "epoch": 0.0011008, "grad_norm": 0.8135698437690735, "learning_rate": 1.8380728420769879e-06, "loss": 2.6307, "step": 314130 }, { "epoch": 0.0011264, "grad_norm": 0.8745925426483154, "learning_rate": 1.8376075949372564e-06, "loss": 2.6714, "step": 314140 }, { "epoch": 0.001152, "grad_norm": 0.8549879193305969, "learning_rate": 1.8371424007284666e-06, "loss": 2.6296, "step": 314150 }, { "epoch": 0.0011776, "grad_norm": 0.9367484450340271, "learning_rate": 1.836677259453642e-06, "loss": 2.9065, "step": 314160 }, { "epoch": 0.0012032, "grad_norm": 0.8578802943229675, "learning_rate": 1.8362121711157988e-06, "loss": 2.8621, "step": 314170 }, { "epoch": 0.0012288, "grad_norm": 0.8661658763885498, "learning_rate": 1.8357471357179469e-06, "loss": 2.6767, "step": 314180 }, { "epoch": 0.0012544, "grad_norm": 0.9764991998672485, "learning_rate": 1.835282153263105e-06, "loss": 2.6741, "step": 314190 }, { "epoch": 0.00128, "grad_norm": 0.8149077296257019, "learning_rate": 1.8348172237542893e-06, "loss": 2.7337, "step": 314200 }, { "epoch": 0.0013056, "grad_norm": 0.8580435514450073, "learning_rate": 1.8343523471945124e-06, "loss": 2.7627, "step": 314210 }, { "epoch": 0.0013312, "grad_norm": 0.8785942792892456, "learning_rate": 1.8338875235867915e-06, "loss": 2.7475, "step": 314220 }, { "epoch": 0.0013568, "grad_norm": 1.0510112047195435, "learning_rate": 1.8334227529341397e-06, "loss": 2.5586, "step": 314230 }, { "epoch": 0.0013824, "grad_norm": 0.7596799731254578, "learning_rate": 1.8329580352395703e-06, "loss": 2.5767, "step": 314240 }, { "epoch": 0.001408, "grad_norm": 1.1934163570404053, "learning_rate": 1.8324933705061e-06, "loss": 2.7542, "step": 314250 }, { "epoch": 0.0014336, "grad_norm": 0.8394665718078613, "learning_rate": 1.832028758736738e-06, "loss": 2.6398, "step": 314260 }, { "epoch": 0.0014592, "grad_norm": 1.114333987236023, "learning_rate": 1.831564199934498e-06, "loss": 2.5367, "step": 314270 }, { "epoch": 0.0014848, "grad_norm": 0.8035483360290527, "learning_rate": 1.8310996941023928e-06, "loss": 2.4853, "step": 314280 }, { "epoch": 0.0015104, "grad_norm": 0.8892320990562439, "learning_rate": 1.830635241243436e-06, "loss": 2.5945, "step": 314290 }, { "epoch": 0.001536, "grad_norm": 1.0559649467468262, "learning_rate": 1.830170841360638e-06, "loss": 2.1539, "step": 314300 }, { "epoch": 0.0015616, "grad_norm": 0.9081116914749146, "learning_rate": 1.8297064944570108e-06, "loss": 2.6209, "step": 314310 }, { "epoch": 0.0015872, "grad_norm": 1.0377357006072998, "learning_rate": 1.8292422005355681e-06, "loss": 2.6869, "step": 314320 }, { "epoch": 0.0016128, "grad_norm": 1.041603446006775, "learning_rate": 1.8287779595993126e-06, "loss": 2.8478, "step": 314330 }, { "epoch": 0.0016384, "grad_norm": 0.9849061965942383, "learning_rate": 1.8283137716512644e-06, "loss": 2.7868, "step": 314340 }, { "epoch": 0.001664, "grad_norm": 1.030909776687622, "learning_rate": 1.8278496366944287e-06, "loss": 2.4368, "step": 314350 }, { "epoch": 0.0016896, "grad_norm": 0.8791333436965942, "learning_rate": 1.8273855547318165e-06, "loss": 2.6794, "step": 314360 }, { "epoch": 0.0017152, "grad_norm": 0.8720983862876892, "learning_rate": 1.8269215257664385e-06, "loss": 2.8737, "step": 314370 }, { "epoch": 0.0017408, "grad_norm": 0.8853965401649475, "learning_rate": 1.8264575498013004e-06, "loss": 2.7914, "step": 314380 }, { "epoch": 0.0017664, "grad_norm": 0.7943283319473267, "learning_rate": 1.825993626839413e-06, "loss": 2.5021, "step": 314390 }, { "epoch": 0.001792, "grad_norm": 0.7621260285377502, "learning_rate": 1.825529756883785e-06, "loss": 2.584, "step": 314400 }, { "epoch": 2.56e-05, "grad_norm": 0.9097121953964233, "learning_rate": 1.8250659399374227e-06, "loss": 1.8311, "step": 314410 }, { "epoch": 5.12e-05, "grad_norm": 0.8623171448707581, "learning_rate": 1.8246021760033362e-06, "loss": 2.7643, "step": 314420 }, { "epoch": 7.68e-05, "grad_norm": 0.8157538175582886, "learning_rate": 1.8241384650845317e-06, "loss": 2.6729, "step": 314430 }, { "epoch": 0.0001024, "grad_norm": 0.7684631943702698, "learning_rate": 1.8236748071840183e-06, "loss": 2.8342, "step": 314440 }, { "epoch": 0.000128, "grad_norm": 0.8161900639533997, "learning_rate": 1.8232112023047988e-06, "loss": 2.8247, "step": 314450 }, { "epoch": 0.0001536, "grad_norm": 0.9007474780082703, "learning_rate": 1.822747650449882e-06, "loss": 2.5236, "step": 314460 }, { "epoch": 0.0001792, "grad_norm": 0.8850809335708618, "learning_rate": 1.8222841516222722e-06, "loss": 2.6864, "step": 314470 }, { "epoch": 0.0002048, "grad_norm": 0.8792663812637329, "learning_rate": 1.821820705824977e-06, "loss": 2.6283, "step": 314480 }, { "epoch": 0.0002304, "grad_norm": 0.8853747844696045, "learning_rate": 1.8213573130610007e-06, "loss": 2.5692, "step": 314490 }, { "epoch": 0.000256, "grad_norm": 0.8059452176094055, "learning_rate": 1.8208939733333486e-06, "loss": 2.5903, "step": 314500 }, { "epoch": 0.0002816, "grad_norm": 0.9508260488510132, "learning_rate": 1.820430686645026e-06, "loss": 2.5658, "step": 314510 }, { "epoch": 0.0003072, "grad_norm": 0.8222028017044067, "learning_rate": 1.8199674529990362e-06, "loss": 2.4207, "step": 314520 }, { "epoch": 0.0003328, "grad_norm": 0.9353776574134827, "learning_rate": 1.8195042723983835e-06, "loss": 2.3112, "step": 314530 }, { "epoch": 0.0003584, "grad_norm": 0.8688445687294006, "learning_rate": 1.819041144846071e-06, "loss": 2.8968, "step": 314540 }, { "epoch": 0.000384, "grad_norm": 0.9926611185073853, "learning_rate": 1.8185780703451027e-06, "loss": 2.5645, "step": 314550 }, { "epoch": 0.0004096, "grad_norm": 0.912774920463562, "learning_rate": 1.818115048898481e-06, "loss": 2.5186, "step": 314560 }, { "epoch": 0.0004352, "grad_norm": 0.8477755784988403, "learning_rate": 1.8176520805092123e-06, "loss": 2.3985, "step": 314570 }, { "epoch": 0.0004608, "grad_norm": 0.7673788666725159, "learning_rate": 1.817189165180292e-06, "loss": 2.6668, "step": 314580 }, { "epoch": 0.0004864, "grad_norm": 0.9654676914215088, "learning_rate": 1.8167263029147253e-06, "loss": 2.7876, "step": 314590 }, { "epoch": 0.000512, "grad_norm": 0.7958741188049316, "learning_rate": 1.8162634937155132e-06, "loss": 2.4291, "step": 314600 }, { "epoch": 0.0005376, "grad_norm": 0.8749805688858032, "learning_rate": 1.815800737585658e-06, "loss": 2.5563, "step": 314610 }, { "epoch": 0.0005632, "grad_norm": 0.9285788536071777, "learning_rate": 1.81533803452816e-06, "loss": 2.552, "step": 314620 }, { "epoch": 0.0005888, "grad_norm": 0.8421294689178467, "learning_rate": 1.8148753845460187e-06, "loss": 2.544, "step": 314630 }, { "epoch": 0.0006144, "grad_norm": 0.8168529272079468, "learning_rate": 1.814412787642239e-06, "loss": 2.6049, "step": 314640 }, { "epoch": 0.00064, "grad_norm": 0.8374791145324707, "learning_rate": 1.813950243819813e-06, "loss": 2.57, "step": 314650 }, { "epoch": 0.0006656, "grad_norm": 0.9032840132713318, "learning_rate": 1.8134877530817453e-06, "loss": 3.0875, "step": 314660 }, { "epoch": 0.0006912, "grad_norm": 0.9624844789505005, "learning_rate": 1.813025315431033e-06, "loss": 2.5314, "step": 314670 }, { "epoch": 0.0007168, "grad_norm": 0.8683070540428162, "learning_rate": 1.8125629308706739e-06, "loss": 2.6021, "step": 314680 }, { "epoch": 0.0007424, "grad_norm": 0.8119812607765198, "learning_rate": 1.81210059940367e-06, "loss": 2.7522, "step": 314690 }, { "epoch": 0.000768, "grad_norm": 0.8731971979141235, "learning_rate": 1.8116383210330202e-06, "loss": 2.5948, "step": 314700 }, { "epoch": 0.0007936, "grad_norm": 0.7865947484970093, "learning_rate": 1.8111760957617176e-06, "loss": 2.6881, "step": 314710 }, { "epoch": 0.0008192, "grad_norm": 1.0278218984603882, "learning_rate": 1.8107139235927618e-06, "loss": 2.7656, "step": 314720 }, { "epoch": 0.0008448, "grad_norm": 1.2078443765640259, "learning_rate": 1.8102518045291495e-06, "loss": 2.7723, "step": 314730 }, { "epoch": 0.0008704, "grad_norm": 0.9094473123550415, "learning_rate": 1.8097897385738772e-06, "loss": 2.5192, "step": 314740 }, { "epoch": 0.000896, "grad_norm": 0.7580307126045227, "learning_rate": 1.8093277257299424e-06, "loss": 2.2915, "step": 314750 }, { "epoch": 0.0009216, "grad_norm": 0.8036816716194153, "learning_rate": 1.8088657660003406e-06, "loss": 2.5718, "step": 314760 }, { "epoch": 0.0009472, "grad_norm": 0.813298761844635, "learning_rate": 1.8084038593880692e-06, "loss": 2.6439, "step": 314770 }, { "epoch": 0.0009728, "grad_norm": 0.9856316447257996, "learning_rate": 1.8079420058961195e-06, "loss": 2.6509, "step": 314780 }, { "epoch": 0.0009984, "grad_norm": 0.8116061091423035, "learning_rate": 1.8074802055274886e-06, "loss": 2.5861, "step": 314790 }, { "epoch": 0.001024, "grad_norm": 0.8940436244010925, "learning_rate": 1.8070184582851723e-06, "loss": 2.7813, "step": 314800 }, { "epoch": 0.0010496, "grad_norm": 0.9355199933052063, "learning_rate": 1.8065567641721627e-06, "loss": 2.6392, "step": 314810 }, { "epoch": 0.0010752, "grad_norm": 0.8158387541770935, "learning_rate": 1.8060951231914548e-06, "loss": 2.6774, "step": 314820 }, { "epoch": 0.0011008, "grad_norm": 0.7646873593330383, "learning_rate": 1.805633535346043e-06, "loss": 2.5261, "step": 314830 }, { "epoch": 0.0011264, "grad_norm": 0.729728639125824, "learning_rate": 1.8051720006389228e-06, "loss": 2.4076, "step": 314840 }, { "epoch": 0.001152, "grad_norm": 0.8966556191444397, "learning_rate": 1.8047105190730784e-06, "loss": 2.6292, "step": 314850 }, { "epoch": 0.0011776, "grad_norm": 0.9100728034973145, "learning_rate": 1.8042490906515109e-06, "loss": 2.8776, "step": 314860 }, { "epoch": 0.0012032, "grad_norm": 0.9962702989578247, "learning_rate": 1.8037877153772098e-06, "loss": 2.686, "step": 314870 }, { "epoch": 0.0012288, "grad_norm": 0.881554901599884, "learning_rate": 1.8033263932531674e-06, "loss": 2.8085, "step": 314880 }, { "epoch": 0.0012544, "grad_norm": 0.8263056874275208, "learning_rate": 1.8028651242823748e-06, "loss": 2.6674, "step": 314890 }, { "epoch": 0.00128, "grad_norm": 0.9006763696670532, "learning_rate": 1.8024039084678257e-06, "loss": 2.6173, "step": 314900 }, { "epoch": 0.0013056, "grad_norm": 0.9967973828315735, "learning_rate": 1.801942745812506e-06, "loss": 2.7629, "step": 314910 }, { "epoch": 0.0013312, "grad_norm": 0.9680130481719971, "learning_rate": 1.8014816363194088e-06, "loss": 2.589, "step": 314920 }, { "epoch": 0.0013568, "grad_norm": 0.7648345828056335, "learning_rate": 1.8010205799915247e-06, "loss": 2.4187, "step": 314930 }, { "epoch": 0.0013824, "grad_norm": 1.175366759300232, "learning_rate": 1.800559576831843e-06, "loss": 2.8199, "step": 314940 }, { "epoch": 0.001408, "grad_norm": 0.8384942412376404, "learning_rate": 1.8000986268433528e-06, "loss": 2.6033, "step": 314950 }, { "epoch": 0.0014336, "grad_norm": 1.0277079343795776, "learning_rate": 1.7996377300290447e-06, "loss": 2.7642, "step": 314960 }, { "epoch": 0.0014592, "grad_norm": 0.9160208702087402, "learning_rate": 1.7991768863919078e-06, "loss": 2.5862, "step": 314970 }, { "epoch": 0.0014848, "grad_norm": 0.8833788633346558, "learning_rate": 1.7987160959349282e-06, "loss": 2.6449, "step": 314980 }, { "epoch": 0.0015104, "grad_norm": 0.8450907468795776, "learning_rate": 1.798255358661094e-06, "loss": 2.2011, "step": 314990 }, { "epoch": 0.001536, "grad_norm": 0.8093623518943787, "learning_rate": 1.7977946745733942e-06, "loss": 2.6247, "step": 315000 }, { "epoch": 2.56e-05, "grad_norm": 0.8204980492591858, "learning_rate": 1.7973340436748155e-06, "loss": 1.6429, "step": 315010 }, { "epoch": 5.12e-05, "grad_norm": 0.8405603766441345, "learning_rate": 1.796873465968344e-06, "loss": 2.6436, "step": 315020 }, { "epoch": 7.68e-05, "grad_norm": 0.84679114818573, "learning_rate": 1.7964129414569731e-06, "loss": 2.7022, "step": 315030 }, { "epoch": 0.0001024, "grad_norm": 0.7810588479042053, "learning_rate": 1.7959524701436803e-06, "loss": 2.7802, "step": 315040 }, { "epoch": 0.000128, "grad_norm": 0.861516535282135, "learning_rate": 1.7954920520314567e-06, "loss": 2.6377, "step": 315050 }, { "epoch": 0.0001536, "grad_norm": 0.8878049254417419, "learning_rate": 1.7950316871232853e-06, "loss": 2.6354, "step": 315060 }, { "epoch": 0.0001792, "grad_norm": 0.9783087372779846, "learning_rate": 1.7945713754221538e-06, "loss": 2.7974, "step": 315070 }, { "epoch": 0.0002048, "grad_norm": 0.7906530499458313, "learning_rate": 1.7941111169310465e-06, "loss": 2.4826, "step": 315080 }, { "epoch": 0.0002304, "grad_norm": 0.917815625667572, "learning_rate": 1.7936509116529466e-06, "loss": 2.5236, "step": 315090 }, { "epoch": 0.000256, "grad_norm": 0.8230734467506409, "learning_rate": 1.7931907595908426e-06, "loss": 2.6879, "step": 315100 }, { "epoch": 0.0002816, "grad_norm": 0.8537560105323792, "learning_rate": 1.7927306607477135e-06, "loss": 2.7311, "step": 315110 }, { "epoch": 0.0003072, "grad_norm": 0.9802899360656738, "learning_rate": 1.7922706151265446e-06, "loss": 2.4557, "step": 315120 }, { "epoch": 0.0003328, "grad_norm": 0.8540374636650085, "learning_rate": 1.791810622730319e-06, "loss": 2.7035, "step": 315130 }, { "epoch": 0.0003584, "grad_norm": 0.8502830862998962, "learning_rate": 1.7913506835620198e-06, "loss": 2.8288, "step": 315140 }, { "epoch": 0.000384, "grad_norm": 0.9417415857315063, "learning_rate": 1.7908907976246303e-06, "loss": 2.5629, "step": 315150 }, { "epoch": 0.0004096, "grad_norm": 0.8206956386566162, "learning_rate": 1.7904309649211315e-06, "loss": 2.5555, "step": 315160 }, { "epoch": 0.0004352, "grad_norm": 0.8871239423751831, "learning_rate": 1.7899711854545098e-06, "loss": 2.6573, "step": 315170 }, { "epoch": 0.0004608, "grad_norm": 0.8461006879806519, "learning_rate": 1.7895114592277396e-06, "loss": 2.72, "step": 315180 }, { "epoch": 0.0004864, "grad_norm": 0.8548717498779297, "learning_rate": 1.7890517862438061e-06, "loss": 2.487, "step": 315190 }, { "epoch": 0.000512, "grad_norm": 0.8980796337127686, "learning_rate": 1.788592166505686e-06, "loss": 2.6514, "step": 315200 }, { "epoch": 0.0005376, "grad_norm": 0.8532655835151672, "learning_rate": 1.7881326000163668e-06, "loss": 2.5508, "step": 315210 }, { "epoch": 0.0005632, "grad_norm": 0.9567083716392517, "learning_rate": 1.787673086778825e-06, "loss": 2.3162, "step": 315220 }, { "epoch": 0.0005888, "grad_norm": 0.8732489943504333, "learning_rate": 1.7872136267960427e-06, "loss": 2.6464, "step": 315230 }, { "epoch": 0.0006144, "grad_norm": 0.9259816408157349, "learning_rate": 1.7867542200709963e-06, "loss": 2.8365, "step": 315240 }, { "epoch": 0.00064, "grad_norm": 0.7916921973228455, "learning_rate": 1.7862948666066649e-06, "loss": 2.8094, "step": 315250 }, { "epoch": 0.0006656, "grad_norm": 0.9465498924255371, "learning_rate": 1.7858355664060277e-06, "loss": 2.9502, "step": 315260 }, { "epoch": 0.0006912, "grad_norm": 0.8065047860145569, "learning_rate": 1.7853763194720653e-06, "loss": 2.7771, "step": 315270 }, { "epoch": 0.0007168, "grad_norm": 0.8401731848716736, "learning_rate": 1.7849171258077535e-06, "loss": 2.5362, "step": 315280 }, { "epoch": 0.0007424, "grad_norm": 1.3611046075820923, "learning_rate": 1.7844579854160704e-06, "loss": 3.2486, "step": 315290 }, { "epoch": 0.000768, "grad_norm": 0.8040053248405457, "learning_rate": 1.7839988982999966e-06, "loss": 2.6055, "step": 315300 }, { "epoch": 0.0007936, "grad_norm": 0.7994593381881714, "learning_rate": 1.7835398644625047e-06, "loss": 2.3008, "step": 315310 }, { "epoch": 0.0008192, "grad_norm": 0.8033573031425476, "learning_rate": 1.7830808839065717e-06, "loss": 2.7066, "step": 315320 }, { "epoch": 0.0008448, "grad_norm": 0.802854061126709, "learning_rate": 1.7826219566351766e-06, "loss": 2.5884, "step": 315330 }, { "epoch": 0.0008704, "grad_norm": 0.8673737645149231, "learning_rate": 1.7821630826512938e-06, "loss": 2.5721, "step": 315340 }, { "epoch": 0.000896, "grad_norm": 1.090345859527588, "learning_rate": 1.7817042619578995e-06, "loss": 2.4923, "step": 315350 }, { "epoch": 0.0009216, "grad_norm": 0.7260352969169617, "learning_rate": 1.7812454945579682e-06, "loss": 2.5763, "step": 315360 }, { "epoch": 0.0009472, "grad_norm": 0.8616754412651062, "learning_rate": 1.7807867804544764e-06, "loss": 2.2908, "step": 315370 }, { "epoch": 0.0009728, "grad_norm": 0.9603211879730225, "learning_rate": 1.7803281196503974e-06, "loss": 2.748, "step": 315380 }, { "epoch": 0.0009984, "grad_norm": 0.7675516605377197, "learning_rate": 1.7798695121487064e-06, "loss": 2.5078, "step": 315390 }, { "epoch": 0.001024, "grad_norm": 0.9588932991027832, "learning_rate": 1.7794109579523766e-06, "loss": 2.6268, "step": 315400 }, { "epoch": 0.0010496, "grad_norm": 0.794416069984436, "learning_rate": 1.7789524570643824e-06, "loss": 2.6865, "step": 315410 }, { "epoch": 0.0010752, "grad_norm": 0.8547676205635071, "learning_rate": 1.7784940094876957e-06, "loss": 2.5753, "step": 315420 }, { "epoch": 0.0011008, "grad_norm": 0.8342286348342896, "learning_rate": 1.7780356152252941e-06, "loss": 2.5223, "step": 315430 }, { "epoch": 0.0011264, "grad_norm": 0.9749045968055725, "learning_rate": 1.7775772742801433e-06, "loss": 2.7082, "step": 315440 }, { "epoch": 0.001152, "grad_norm": 0.8755494356155396, "learning_rate": 1.7771189866552185e-06, "loss": 2.7634, "step": 315450 }, { "epoch": 0.0011776, "grad_norm": 1.097872257232666, "learning_rate": 1.7766607523534918e-06, "loss": 2.8173, "step": 315460 }, { "epoch": 0.0012032, "grad_norm": 1.0911667346954346, "learning_rate": 1.776202571377934e-06, "loss": 2.6483, "step": 315470 }, { "epoch": 0.0012288, "grad_norm": 0.84125155210495, "learning_rate": 1.7757444437315175e-06, "loss": 2.8159, "step": 315480 }, { "epoch": 0.0012544, "grad_norm": 0.866714596748352, "learning_rate": 1.7752863694172128e-06, "loss": 2.5837, "step": 315490 }, { "epoch": 0.00128, "grad_norm": 0.8952271342277527, "learning_rate": 1.7748283484379925e-06, "loss": 2.5743, "step": 315500 }, { "epoch": 0.0013056, "grad_norm": 1.0169901847839355, "learning_rate": 1.7743703807968215e-06, "loss": 2.4016, "step": 315510 }, { "epoch": 0.0013312, "grad_norm": 0.8536272644996643, "learning_rate": 1.7739124664966734e-06, "loss": 2.7827, "step": 315520 }, { "epoch": 0.0013568, "grad_norm": 0.9756270051002502, "learning_rate": 1.7734546055405155e-06, "loss": 2.6171, "step": 315530 }, { "epoch": 0.0013824, "grad_norm": 0.8385269045829773, "learning_rate": 1.772996797931319e-06, "loss": 2.6143, "step": 315540 }, { "epoch": 0.001408, "grad_norm": 0.9728022217750549, "learning_rate": 1.7725390436720491e-06, "loss": 2.7339, "step": 315550 }, { "epoch": 0.0014336, "grad_norm": 0.7663761377334595, "learning_rate": 1.7720813427656814e-06, "loss": 2.5696, "step": 315560 }, { "epoch": 0.0014592, "grad_norm": 0.8193328380584717, "learning_rate": 1.7716236952151777e-06, "loss": 2.4028, "step": 315570 }, { "epoch": 0.0014848, "grad_norm": 0.795620858669281, "learning_rate": 1.7711661010235071e-06, "loss": 2.2878, "step": 315580 }, { "epoch": 0.0015104, "grad_norm": 0.9367849230766296, "learning_rate": 1.770708560193637e-06, "loss": 2.5836, "step": 315590 }, { "epoch": 0.001536, "grad_norm": 0.8907846212387085, "learning_rate": 1.770251072728536e-06, "loss": 2.666, "step": 315600 }, { "epoch": 0.0015616, "grad_norm": 0.8044188618659973, "learning_rate": 1.7697936386311687e-06, "loss": 2.8585, "step": 315610 }, { "epoch": 0.0015872, "grad_norm": 0.8484624624252319, "learning_rate": 1.7693362579045025e-06, "loss": 2.6054, "step": 315620 }, { "epoch": 0.0016128, "grad_norm": 0.8825715780258179, "learning_rate": 1.7688789305515052e-06, "loss": 2.6239, "step": 315630 }, { "epoch": 0.0016384, "grad_norm": 0.8357958197593689, "learning_rate": 1.7684216565751388e-06, "loss": 2.4965, "step": 315640 }, { "epoch": 0.001664, "grad_norm": 0.8089102506637573, "learning_rate": 1.76796443597837e-06, "loss": 2.7983, "step": 315650 }, { "epoch": 0.0016896, "grad_norm": 0.790692925453186, "learning_rate": 1.7675072687641636e-06, "loss": 2.3436, "step": 315660 }, { "epoch": 0.0017152, "grad_norm": 0.7735154032707214, "learning_rate": 1.7670501549354846e-06, "loss": 2.5604, "step": 315670 }, { "epoch": 0.0017408, "grad_norm": 0.8531178832054138, "learning_rate": 1.766593094495298e-06, "loss": 2.8735, "step": 315680 }, { "epoch": 0.0017664, "grad_norm": 0.9642974734306335, "learning_rate": 1.766136087446566e-06, "loss": 2.3296, "step": 315690 }, { "epoch": 0.001792, "grad_norm": 0.9286531805992126, "learning_rate": 1.7656791337922563e-06, "loss": 2.7042, "step": 315700 }, { "epoch": 0.0018176, "grad_norm": 0.7716075778007507, "learning_rate": 1.7652222335353275e-06, "loss": 2.394, "step": 315710 }, { "epoch": 0.0018432, "grad_norm": 1.0674210786819458, "learning_rate": 1.7647653866787396e-06, "loss": 2.7924, "step": 315720 }, { "epoch": 0.0018688, "grad_norm": 0.8709076046943665, "learning_rate": 1.7643085932254634e-06, "loss": 2.7357, "step": 315730 }, { "epoch": 0.0018944, "grad_norm": 0.7931557893753052, "learning_rate": 1.7638518531784566e-06, "loss": 2.7931, "step": 315740 }, { "epoch": 0.00192, "grad_norm": 0.869568943977356, "learning_rate": 1.7633951665406846e-06, "loss": 2.7855, "step": 315750 }, { "epoch": 0.0019456, "grad_norm": 1.7103935480117798, "learning_rate": 1.7629385333151028e-06, "loss": 2.6124, "step": 315760 }, { "epoch": 0.0019712, "grad_norm": 0.8369108438491821, "learning_rate": 1.7624819535046755e-06, "loss": 2.6728, "step": 315770 }, { "epoch": 0.0019968, "grad_norm": 0.8959128856658936, "learning_rate": 1.7620254271123638e-06, "loss": 2.5199, "step": 315780 }, { "epoch": 0.0020224, "grad_norm": 0.7561815977096558, "learning_rate": 1.7615689541411274e-06, "loss": 2.3753, "step": 315790 }, { "epoch": 0.002048, "grad_norm": 0.7768420577049255, "learning_rate": 1.7611125345939273e-06, "loss": 2.6001, "step": 315800 }, { "epoch": 0.0020736, "grad_norm": 0.8019106984138489, "learning_rate": 1.7606561684737222e-06, "loss": 2.7876, "step": 315810 }, { "epoch": 0.0020992, "grad_norm": 0.8453384637832642, "learning_rate": 1.7601998557834721e-06, "loss": 2.9852, "step": 315820 }, { "epoch": 0.0021248, "grad_norm": 0.9154175519943237, "learning_rate": 1.759743596526139e-06, "loss": 2.5998, "step": 315830 }, { "epoch": 0.0021504, "grad_norm": 0.7841751575469971, "learning_rate": 1.759287390704677e-06, "loss": 2.8486, "step": 315840 }, { "epoch": 0.002176, "grad_norm": 0.8428319096565247, "learning_rate": 1.7588312383220451e-06, "loss": 2.7041, "step": 315850 }, { "epoch": 0.0022016, "grad_norm": 0.7749093174934387, "learning_rate": 1.758375139381202e-06, "loss": 2.5468, "step": 315860 }, { "epoch": 0.0022272, "grad_norm": 0.7729021310806274, "learning_rate": 1.7579190938851065e-06, "loss": 2.7912, "step": 315870 }, { "epoch": 0.0022528, "grad_norm": 0.9342570304870605, "learning_rate": 1.757463101836715e-06, "loss": 2.403, "step": 315880 }, { "epoch": 0.0022784, "grad_norm": 0.9013388752937317, "learning_rate": 1.7570071632389852e-06, "loss": 2.7177, "step": 315890 }, { "epoch": 0.002304, "grad_norm": 0.7891814112663269, "learning_rate": 1.7565512780948723e-06, "loss": 2.6929, "step": 315900 }, { "epoch": 0.0023296, "grad_norm": 0.9297566413879395, "learning_rate": 1.7560954464073333e-06, "loss": 2.6505, "step": 315910 }, { "epoch": 0.0023552, "grad_norm": 0.9005879163742065, "learning_rate": 1.7556396681793252e-06, "loss": 2.7957, "step": 315920 }, { "epoch": 0.0023808, "grad_norm": 0.9077839851379395, "learning_rate": 1.755183943413803e-06, "loss": 2.604, "step": 315930 }, { "epoch": 0.0024064, "grad_norm": 0.772736668586731, "learning_rate": 1.7547282721137205e-06, "loss": 2.5372, "step": 315940 }, { "epoch": 0.002432, "grad_norm": 0.9087857007980347, "learning_rate": 1.7542726542820376e-06, "loss": 2.9415, "step": 315950 }, { "epoch": 0.0024576, "grad_norm": 0.8236186504364014, "learning_rate": 1.7538170899217022e-06, "loss": 2.8795, "step": 315960 }, { "epoch": 0.0024832, "grad_norm": 0.8620833158493042, "learning_rate": 1.7533615790356706e-06, "loss": 2.9882, "step": 315970 }, { "epoch": 0.0025088, "grad_norm": 0.8548820614814758, "learning_rate": 1.7529061216268982e-06, "loss": 2.8204, "step": 315980 }, { "epoch": 0.0025344, "grad_norm": 0.8070648908615112, "learning_rate": 1.7524507176983374e-06, "loss": 2.7061, "step": 315990 }, { "epoch": 0.00256, "grad_norm": 0.8943372368812561, "learning_rate": 1.7519953672529422e-06, "loss": 2.8081, "step": 316000 }, { "epoch": 0.0025856, "grad_norm": 0.7572308778762817, "learning_rate": 1.7515400702936646e-06, "loss": 2.775, "step": 316010 }, { "epoch": 0.0026112, "grad_norm": 0.9064885973930359, "learning_rate": 1.7510848268234582e-06, "loss": 2.8449, "step": 316020 }, { "epoch": 0.0026368, "grad_norm": 0.9208540916442871, "learning_rate": 1.7506296368452768e-06, "loss": 2.6672, "step": 316030 }, { "epoch": 0.0026624, "grad_norm": 0.8243879675865173, "learning_rate": 1.7501745003620662e-06, "loss": 2.6943, "step": 316040 }, { "epoch": 0.002688, "grad_norm": 1.0266538858413696, "learning_rate": 1.7497194173767829e-06, "loss": 2.6679, "step": 316050 }, { "epoch": 0.0027136, "grad_norm": 0.8049506545066833, "learning_rate": 1.7492643878923755e-06, "loss": 2.9123, "step": 316060 }, { "epoch": 0.0027392, "grad_norm": 0.807356595993042, "learning_rate": 1.7488094119117937e-06, "loss": 2.5933, "step": 316070 }, { "epoch": 0.0027648, "grad_norm": 0.9105111360549927, "learning_rate": 1.7483544894379944e-06, "loss": 2.7641, "step": 316080 }, { "epoch": 0.0027904, "grad_norm": 0.8358033895492554, "learning_rate": 1.7478996204739208e-06, "loss": 2.6191, "step": 316090 }, { "epoch": 0.002816, "grad_norm": 0.8352138996124268, "learning_rate": 1.7474448050225256e-06, "loss": 2.9023, "step": 316100 }, { "epoch": 0.0028416, "grad_norm": 0.9849072098731995, "learning_rate": 1.746990043086757e-06, "loss": 2.8974, "step": 316110 }, { "epoch": 0.0028672, "grad_norm": 0.8525090217590332, "learning_rate": 1.7465353346695645e-06, "loss": 2.8509, "step": 316120 }, { "epoch": 0.0028928, "grad_norm": 0.836676299571991, "learning_rate": 1.7460806797738972e-06, "loss": 2.8042, "step": 316130 }, { "epoch": 0.0029184, "grad_norm": 0.7945520281791687, "learning_rate": 1.7456260784027034e-06, "loss": 2.7535, "step": 316140 }, { "epoch": 0.002944, "grad_norm": 0.772551417350769, "learning_rate": 1.7451715305589334e-06, "loss": 2.8493, "step": 316150 }, { "epoch": 0.0029696, "grad_norm": 0.781602680683136, "learning_rate": 1.744717036245529e-06, "loss": 2.9907, "step": 316160 }, { "epoch": 0.0029952, "grad_norm": 0.857123851776123, "learning_rate": 1.7442625954654403e-06, "loss": 2.7863, "step": 316170 }, { "epoch": 0.0030208, "grad_norm": 0.7847733497619629, "learning_rate": 1.7438082082216156e-06, "loss": 2.9013, "step": 316180 }, { "epoch": 0.0030464, "grad_norm": 0.7783173322677612, "learning_rate": 1.7433538745169998e-06, "loss": 2.7739, "step": 316190 }, { "epoch": 0.003072, "grad_norm": 0.9305505752563477, "learning_rate": 1.742899594354539e-06, "loss": 2.7853, "step": 316200 }, { "epoch": 0.0030976, "grad_norm": 0.8272133469581604, "learning_rate": 1.742445367737181e-06, "loss": 2.8569, "step": 316210 }, { "epoch": 0.0031232, "grad_norm": 0.8350465893745422, "learning_rate": 1.7419911946678703e-06, "loss": 2.7541, "step": 316220 }, { "epoch": 0.0031488, "grad_norm": 0.7984208464622498, "learning_rate": 1.7415370751495541e-06, "loss": 2.6312, "step": 316230 }, { "epoch": 0.0031744, "grad_norm": 1.0766935348510742, "learning_rate": 1.7410830091851693e-06, "loss": 2.7651, "step": 316240 }, { "epoch": 0.0032, "grad_norm": 0.851109504699707, "learning_rate": 1.7406289967776701e-06, "loss": 2.8058, "step": 316250 }, { "epoch": 0.0032256, "grad_norm": 0.8685799837112427, "learning_rate": 1.7401750379299965e-06, "loss": 3.0264, "step": 316260 }, { "epoch": 0.0032512, "grad_norm": 0.952113687992096, "learning_rate": 1.7397211326450914e-06, "loss": 3.049, "step": 316270 }, { "epoch": 0.0032768, "grad_norm": 0.8909728527069092, "learning_rate": 1.7392672809259037e-06, "loss": 2.8305, "step": 316280 }, { "epoch": 0.0033024, "grad_norm": 1.06264066696167, "learning_rate": 1.7388134827753688e-06, "loss": 2.865, "step": 316290 }, { "epoch": 0.003328, "grad_norm": 0.7856207489967346, "learning_rate": 1.7383597381964322e-06, "loss": 2.7254, "step": 316300 }, { "epoch": 0.0033536, "grad_norm": 0.8367161750793457, "learning_rate": 1.7379060471920384e-06, "loss": 2.7027, "step": 316310 }, { "epoch": 0.0033792, "grad_norm": 1.1704899072647095, "learning_rate": 1.737452409765127e-06, "loss": 2.9337, "step": 316320 }, { "epoch": 0.0034048, "grad_norm": 0.7711859941482544, "learning_rate": 1.736998825918641e-06, "loss": 3.0667, "step": 316330 }, { "epoch": 0.0034304, "grad_norm": 0.8504761457443237, "learning_rate": 1.736545295655523e-06, "loss": 2.9852, "step": 316340 }, { "epoch": 0.003456, "grad_norm": 0.7542809247970581, "learning_rate": 1.7360918189787135e-06, "loss": 2.692, "step": 316350 }, { "epoch": 0.0034816, "grad_norm": 0.7829387187957764, "learning_rate": 1.7356383958911504e-06, "loss": 2.5892, "step": 316360 }, { "epoch": 0.0035072, "grad_norm": 0.9445963501930237, "learning_rate": 1.735185026395775e-06, "loss": 2.6857, "step": 316370 }, { "epoch": 0.0035328, "grad_norm": 0.9765653014183044, "learning_rate": 1.7347317104955297e-06, "loss": 2.8937, "step": 316380 }, { "epoch": 0.0035584, "grad_norm": 0.860179603099823, "learning_rate": 1.7342784481933516e-06, "loss": 3.0716, "step": 316390 }, { "epoch": 0.003584, "grad_norm": 0.8122544884681702, "learning_rate": 1.7338252394921806e-06, "loss": 3.3218, "step": 316400 }, { "epoch": 0.0036096, "grad_norm": 0.8141880035400391, "learning_rate": 1.7333720843949564e-06, "loss": 2.7537, "step": 316410 }, { "epoch": 0.0036352, "grad_norm": 0.8825031518936157, "learning_rate": 1.7329189829046178e-06, "loss": 2.9289, "step": 316420 }, { "epoch": 0.0036608, "grad_norm": 0.8559740781784058, "learning_rate": 1.7324659350241025e-06, "loss": 2.9649, "step": 316430 }, { "epoch": 0.0036864, "grad_norm": 0.812617301940918, "learning_rate": 1.732012940756348e-06, "loss": 2.8972, "step": 316440 }, { "epoch": 0.003712, "grad_norm": 0.8173671364784241, "learning_rate": 1.731560000104292e-06, "loss": 2.5285, "step": 316450 }, { "epoch": 0.0037376, "grad_norm": 0.8179564476013184, "learning_rate": 1.7311071130708724e-06, "loss": 2.9187, "step": 316460 }, { "epoch": 0.0037632, "grad_norm": 0.9261185526847839, "learning_rate": 1.7306542796590254e-06, "loss": 2.8695, "step": 316470 }, { "epoch": 0.0037888, "grad_norm": 0.8254170417785645, "learning_rate": 1.7302014998716899e-06, "loss": 2.817, "step": 316480 }, { "epoch": 0.0038144, "grad_norm": 0.768588125705719, "learning_rate": 1.7297487737117969e-06, "loss": 2.6777, "step": 316490 }, { "epoch": 0.00384, "grad_norm": 0.910649836063385, "learning_rate": 1.7292961011822862e-06, "loss": 3.0417, "step": 316500 }, { "epoch": 0.0038656, "grad_norm": 0.809489369392395, "learning_rate": 1.7288434822860911e-06, "loss": 2.7938, "step": 316510 }, { "epoch": 0.0038912, "grad_norm": 0.8673794269561768, "learning_rate": 1.7283909170261481e-06, "loss": 2.8331, "step": 316520 }, { "epoch": 0.0039168, "grad_norm": 0.8132680654525757, "learning_rate": 1.7279384054053926e-06, "loss": 2.6245, "step": 316530 }, { "epoch": 0.0039424, "grad_norm": 0.8978151679039001, "learning_rate": 1.727485947426757e-06, "loss": 2.6385, "step": 316540 }, { "epoch": 0.003968, "grad_norm": 0.8537830710411072, "learning_rate": 1.7270335430931795e-06, "loss": 2.7809, "step": 316550 }, { "epoch": 0.0039936, "grad_norm": 0.8852392435073853, "learning_rate": 1.726581192407588e-06, "loss": 2.8788, "step": 316560 }, { "epoch": 0.0040192, "grad_norm": 0.8792966604232788, "learning_rate": 1.7261288953729194e-06, "loss": 2.8199, "step": 316570 }, { "epoch": 0.0040448, "grad_norm": 0.8112488389015198, "learning_rate": 1.7256766519921053e-06, "loss": 2.5484, "step": 316580 }, { "epoch": 0.0040704, "grad_norm": 0.8546149134635925, "learning_rate": 1.7252244622680768e-06, "loss": 2.9408, "step": 316590 }, { "epoch": 0.004096, "grad_norm": 0.8356878757476807, "learning_rate": 1.7247723262037707e-06, "loss": 2.7498, "step": 316600 }, { "epoch": 0.0041216, "grad_norm": 0.7477536797523499, "learning_rate": 1.72432024380212e-06, "loss": 2.7305, "step": 316610 }, { "epoch": 0.0041472, "grad_norm": 0.8587892055511475, "learning_rate": 1.7238682150660503e-06, "loss": 2.8235, "step": 316620 }, { "epoch": 0.0041728, "grad_norm": 0.9031012058258057, "learning_rate": 1.7234162399984943e-06, "loss": 2.8957, "step": 316630 }, { "epoch": 0.0041984, "grad_norm": 0.8092960119247437, "learning_rate": 1.722964318602386e-06, "loss": 2.6624, "step": 316640 }, { "epoch": 0.004224, "grad_norm": 0.9861732721328735, "learning_rate": 1.7225124508806535e-06, "loss": 2.6856, "step": 316650 }, { "epoch": 0.0042496, "grad_norm": 0.822602391242981, "learning_rate": 1.7220606368362281e-06, "loss": 2.8302, "step": 316660 }, { "epoch": 0.0042752, "grad_norm": 0.810854434967041, "learning_rate": 1.7216088764720396e-06, "loss": 2.8628, "step": 316670 }, { "epoch": 0.0043008, "grad_norm": 0.8041620254516602, "learning_rate": 1.72115716979102e-06, "loss": 2.6875, "step": 316680 }, { "epoch": 0.0043264, "grad_norm": 0.8872296214103699, "learning_rate": 1.720705516796094e-06, "loss": 2.858, "step": 316690 }, { "epoch": 0.004352, "grad_norm": 0.8630126714706421, "learning_rate": 1.7202539174901912e-06, "loss": 2.9041, "step": 316700 }, { "epoch": 0.0043776, "grad_norm": 1.1298785209655762, "learning_rate": 1.7198023718762412e-06, "loss": 2.7334, "step": 316710 }, { "epoch": 0.0044032, "grad_norm": 0.8824192881584167, "learning_rate": 1.7193508799571735e-06, "loss": 2.8359, "step": 316720 }, { "epoch": 0.0044288, "grad_norm": 0.883124589920044, "learning_rate": 1.7188994417359128e-06, "loss": 2.7892, "step": 316730 }, { "epoch": 0.0044544, "grad_norm": 0.8196616172790527, "learning_rate": 1.7184480572153893e-06, "loss": 3.0167, "step": 316740 }, { "epoch": 0.00448, "grad_norm": 0.9067347049713135, "learning_rate": 1.717996726398532e-06, "loss": 2.6993, "step": 316750 }, { "epoch": 0.0045056, "grad_norm": 0.8686688542366028, "learning_rate": 1.7175454492882594e-06, "loss": 2.5511, "step": 316760 }, { "epoch": 0.0045312, "grad_norm": 0.9230719208717346, "learning_rate": 1.7170942258875067e-06, "loss": 2.901, "step": 316770 }, { "epoch": 0.0045568, "grad_norm": 0.8786701560020447, "learning_rate": 1.7166430561991954e-06, "loss": 2.6935, "step": 316780 }, { "epoch": 0.0045824, "grad_norm": 0.8158740401268005, "learning_rate": 1.7161919402262528e-06, "loss": 2.8599, "step": 316790 }, { "epoch": 0.004608, "grad_norm": 0.882217526435852, "learning_rate": 1.7157408779716045e-06, "loss": 2.8785, "step": 316800 }, { "epoch": 0.0046336, "grad_norm": 0.9442312121391296, "learning_rate": 1.715289869438177e-06, "loss": 3.1304, "step": 316810 }, { "epoch": 0.0046592, "grad_norm": 0.8299872875213623, "learning_rate": 1.7148389146288902e-06, "loss": 2.8272, "step": 316820 }, { "epoch": 0.0046848, "grad_norm": 1.049215316772461, "learning_rate": 1.7143880135466707e-06, "loss": 2.8133, "step": 316830 }, { "epoch": 0.0047104, "grad_norm": 0.8533144593238831, "learning_rate": 1.7139371661944438e-06, "loss": 2.9457, "step": 316840 }, { "epoch": 0.004736, "grad_norm": 0.8987125754356384, "learning_rate": 1.7134863725751317e-06, "loss": 2.9248, "step": 316850 }, { "epoch": 0.0047616, "grad_norm": 0.8455930352210999, "learning_rate": 1.7130356326916585e-06, "loss": 2.6534, "step": 316860 }, { "epoch": 0.0047872, "grad_norm": 0.855193018913269, "learning_rate": 1.712584946546947e-06, "loss": 2.9091, "step": 316870 }, { "epoch": 0.0048128, "grad_norm": 0.7878455519676208, "learning_rate": 1.7121343141439207e-06, "loss": 2.7038, "step": 316880 }, { "epoch": 0.0048384, "grad_norm": 0.7926329970359802, "learning_rate": 1.7116837354854998e-06, "loss": 2.7998, "step": 316890 }, { "epoch": 0.004864, "grad_norm": 0.754448413848877, "learning_rate": 1.7112332105746066e-06, "loss": 2.7494, "step": 316900 }, { "epoch": 0.0048896, "grad_norm": 0.7540927529335022, "learning_rate": 1.7107827394141629e-06, "loss": 2.7878, "step": 316910 }, { "epoch": 0.0049152, "grad_norm": 0.800641655921936, "learning_rate": 1.71033232200709e-06, "loss": 2.724, "step": 316920 }, { "epoch": 0.0049408, "grad_norm": 1.0809547901153564, "learning_rate": 1.7098819583563098e-06, "loss": 2.9005, "step": 316930 }, { "epoch": 0.0049664, "grad_norm": 0.773322582244873, "learning_rate": 1.709431648464741e-06, "loss": 2.7344, "step": 316940 }, { "epoch": 0.004992, "grad_norm": 0.8239985108375549, "learning_rate": 1.7089813923353049e-06, "loss": 2.783, "step": 316950 }, { "epoch": 0.0050176, "grad_norm": 0.8507715463638306, "learning_rate": 1.7085311899709212e-06, "loss": 2.6577, "step": 316960 }, { "epoch": 0.0050432, "grad_norm": 0.7986282110214233, "learning_rate": 1.7080810413745086e-06, "loss": 2.6767, "step": 316970 }, { "epoch": 0.0050688, "grad_norm": 0.8279131650924683, "learning_rate": 1.707630946548987e-06, "loss": 2.7311, "step": 316980 }, { "epoch": 0.0050944, "grad_norm": 0.8312504887580872, "learning_rate": 1.7071809054972744e-06, "loss": 2.9121, "step": 316990 }, { "epoch": 0.00512, "grad_norm": 0.8106675148010254, "learning_rate": 1.7067309182222903e-06, "loss": 2.746, "step": 317000 }, { "epoch": 0.0051456, "grad_norm": 0.852675199508667, "learning_rate": 1.7062809847269535e-06, "loss": 2.7744, "step": 317010 }, { "epoch": 0.0051712, "grad_norm": 0.8921718597412109, "learning_rate": 1.7058311050141797e-06, "loss": 2.7318, "step": 317020 }, { "epoch": 0.0051968, "grad_norm": 0.9619848728179932, "learning_rate": 1.705381279086885e-06, "loss": 2.9606, "step": 317030 }, { "epoch": 0.0052224, "grad_norm": 0.8308207988739014, "learning_rate": 1.7049315069479888e-06, "loss": 2.7031, "step": 317040 }, { "epoch": 0.005248, "grad_norm": 0.823189914226532, "learning_rate": 1.7044817886004085e-06, "loss": 2.6034, "step": 317050 }, { "epoch": 0.0052736, "grad_norm": 1.0056159496307373, "learning_rate": 1.7040321240470582e-06, "loss": 2.9463, "step": 317060 }, { "epoch": 0.0052992, "grad_norm": 1.1784050464630127, "learning_rate": 1.7035825132908545e-06, "loss": 2.9568, "step": 317070 }, { "epoch": 0.0053248, "grad_norm": 0.8230981826782227, "learning_rate": 1.7031329563347154e-06, "loss": 2.922, "step": 317080 }, { "epoch": 0.0053504, "grad_norm": 0.8152938485145569, "learning_rate": 1.7026834531815529e-06, "loss": 2.7763, "step": 317090 }, { "epoch": 0.005376, "grad_norm": 0.8310491442680359, "learning_rate": 1.7022340038342821e-06, "loss": 2.6578, "step": 317100 }, { "epoch": 0.0054016, "grad_norm": 0.8069709539413452, "learning_rate": 1.7017846082958157e-06, "loss": 2.8162, "step": 317110 }, { "epoch": 0.0054272, "grad_norm": 0.8123441934585571, "learning_rate": 1.7013352665690742e-06, "loss": 2.6748, "step": 317120 }, { "epoch": 0.0054528, "grad_norm": 0.8687770962715149, "learning_rate": 1.7008859786569664e-06, "loss": 2.8536, "step": 317130 }, { "epoch": 0.0054784, "grad_norm": 0.895791232585907, "learning_rate": 1.7004367445624114e-06, "loss": 2.7244, "step": 317140 }, { "epoch": 0.005504, "grad_norm": 0.7636492252349854, "learning_rate": 1.6999875642883156e-06, "loss": 2.8813, "step": 317150 }, { "epoch": 0.0055296, "grad_norm": 1.4175477027893066, "learning_rate": 1.6995384378375944e-06, "loss": 2.8018, "step": 317160 }, { "epoch": 0.0055552, "grad_norm": 0.8338989615440369, "learning_rate": 1.69908936521316e-06, "loss": 2.8042, "step": 317170 }, { "epoch": 0.0055808, "grad_norm": 0.8892509937286377, "learning_rate": 1.6986403464179258e-06, "loss": 2.7921, "step": 317180 }, { "epoch": 0.0056064, "grad_norm": 0.8053849339485168, "learning_rate": 1.6981913814548023e-06, "loss": 2.7315, "step": 317190 }, { "epoch": 0.005632, "grad_norm": 0.8784292340278625, "learning_rate": 1.697742470326701e-06, "loss": 2.6981, "step": 317200 }, { "epoch": 0.0056576, "grad_norm": 0.8555670976638794, "learning_rate": 1.6972936130365358e-06, "loss": 2.9288, "step": 317210 }, { "epoch": 0.0056832, "grad_norm": 0.7654257416725159, "learning_rate": 1.6968448095872126e-06, "loss": 2.6887, "step": 317220 }, { "epoch": 0.0057088, "grad_norm": 0.7922536730766296, "learning_rate": 1.6963960599816442e-06, "loss": 2.7742, "step": 317230 }, { "epoch": 0.0057344, "grad_norm": 0.7997090220451355, "learning_rate": 1.6959473642227409e-06, "loss": 2.6624, "step": 317240 }, { "epoch": 0.00576, "grad_norm": 0.8351221084594727, "learning_rate": 1.6954987223134112e-06, "loss": 2.7336, "step": 317250 }, { "epoch": 0.0057856, "grad_norm": 0.9008411765098572, "learning_rate": 1.695050134256566e-06, "loss": 2.6917, "step": 317260 }, { "epoch": 0.0058112, "grad_norm": 0.8499450087547302, "learning_rate": 1.6946016000551123e-06, "loss": 2.6677, "step": 317270 }, { "epoch": 0.0058368, "grad_norm": 0.7721462249755859, "learning_rate": 1.6941531197119633e-06, "loss": 2.6442, "step": 317280 }, { "epoch": 0.0058624, "grad_norm": 0.8892305493354797, "learning_rate": 1.6937046932300194e-06, "loss": 2.9214, "step": 317290 }, { "epoch": 0.005888, "grad_norm": 0.7970133423805237, "learning_rate": 1.6932563206121944e-06, "loss": 2.6082, "step": 317300 }, { "epoch": 0.0059136, "grad_norm": 0.807345449924469, "learning_rate": 1.6928080018613957e-06, "loss": 2.6726, "step": 317310 }, { "epoch": 0.0059392, "grad_norm": 0.8475432991981506, "learning_rate": 1.6923597369805278e-06, "loss": 2.9192, "step": 317320 }, { "epoch": 0.0059648, "grad_norm": 0.9546248912811279, "learning_rate": 1.6919115259725006e-06, "loss": 2.8745, "step": 317330 }, { "epoch": 0.0059904, "grad_norm": 0.8681376576423645, "learning_rate": 1.6914633688402204e-06, "loss": 2.5848, "step": 317340 }, { "epoch": 0.006016, "grad_norm": 1.0059384107589722, "learning_rate": 1.6910152655865908e-06, "loss": 2.8799, "step": 317350 }, { "epoch": 0.0060416, "grad_norm": 0.8638327121734619, "learning_rate": 1.6905672162145181e-06, "loss": 2.7869, "step": 317360 }, { "epoch": 0.0060672, "grad_norm": 0.8587802648544312, "learning_rate": 1.690119220726909e-06, "loss": 2.9446, "step": 317370 }, { "epoch": 0.0060928, "grad_norm": 0.8742067217826843, "learning_rate": 1.6896712791266678e-06, "loss": 2.7614, "step": 317380 }, { "epoch": 0.0061184, "grad_norm": 0.7625571489334106, "learning_rate": 1.6892233914167e-06, "loss": 2.849, "step": 317390 }, { "epoch": 0.006144, "grad_norm": 0.886175811290741, "learning_rate": 1.6887755575999098e-06, "loss": 2.685, "step": 317400 }, { "epoch": 0.0061696, "grad_norm": 0.8131573796272278, "learning_rate": 1.688327777679204e-06, "loss": 2.5888, "step": 317410 }, { "epoch": 0.0061952, "grad_norm": 0.7541942000389099, "learning_rate": 1.687880051657481e-06, "loss": 2.8566, "step": 317420 }, { "epoch": 0.0062208, "grad_norm": 0.8399441838264465, "learning_rate": 1.6874323795376457e-06, "loss": 2.6849, "step": 317430 }, { "epoch": 0.0062464, "grad_norm": 0.7437549233436584, "learning_rate": 1.686984761322603e-06, "loss": 2.9127, "step": 317440 }, { "epoch": 0.006272, "grad_norm": 2.988546848297119, "learning_rate": 1.6865371970152555e-06, "loss": 2.7239, "step": 317450 }, { "epoch": 0.0062976, "grad_norm": 0.9717298150062561, "learning_rate": 1.686089686618504e-06, "loss": 2.7026, "step": 317460 }, { "epoch": 0.0063232, "grad_norm": 0.8007952570915222, "learning_rate": 1.6856422301352516e-06, "loss": 2.783, "step": 317470 }, { "epoch": 0.0063488, "grad_norm": 0.8126741051673889, "learning_rate": 1.6851948275683993e-06, "loss": 2.6936, "step": 317480 }, { "epoch": 0.0063744, "grad_norm": 0.79729825258255, "learning_rate": 1.6847474789208495e-06, "loss": 2.6325, "step": 317490 }, { "epoch": 0.0064, "grad_norm": 0.8851847052574158, "learning_rate": 1.684300184195502e-06, "loss": 2.8265, "step": 317500 }, { "epoch": 0.0064256, "grad_norm": 0.868283748626709, "learning_rate": 1.6838529433952577e-06, "loss": 2.647, "step": 317510 }, { "epoch": 0.0064512, "grad_norm": 0.8679280877113342, "learning_rate": 1.6834057565230166e-06, "loss": 2.7649, "step": 317520 }, { "epoch": 0.0064768, "grad_norm": 0.7766543030738831, "learning_rate": 1.6829586235816787e-06, "loss": 2.6836, "step": 317530 }, { "epoch": 0.0065024, "grad_norm": 0.8464369773864746, "learning_rate": 1.6825115445741468e-06, "loss": 2.7894, "step": 317540 }, { "epoch": 0.006528, "grad_norm": 0.781156063079834, "learning_rate": 1.6820645195033147e-06, "loss": 3.0537, "step": 317550 }, { "epoch": 0.0065536, "grad_norm": 0.8540031313896179, "learning_rate": 1.6816175483720832e-06, "loss": 2.8873, "step": 317560 }, { "epoch": 0.0065792, "grad_norm": 0.8048498034477234, "learning_rate": 1.6811706311833508e-06, "loss": 2.653, "step": 317570 }, { "epoch": 0.0066048, "grad_norm": 0.8131536841392517, "learning_rate": 1.6807237679400158e-06, "loss": 2.9061, "step": 317580 }, { "epoch": 0.0066304, "grad_norm": 1.0868967771530151, "learning_rate": 1.6802769586449774e-06, "loss": 2.7183, "step": 317590 }, { "epoch": 0.006656, "grad_norm": 0.8379144072532654, "learning_rate": 1.6798302033011303e-06, "loss": 2.782, "step": 317600 }, { "epoch": 0.0066816, "grad_norm": 0.8637049794197083, "learning_rate": 1.6793835019113769e-06, "loss": 2.7894, "step": 317610 }, { "epoch": 0.0067072, "grad_norm": 0.8471007943153381, "learning_rate": 1.6789368544786066e-06, "loss": 2.7395, "step": 317620 }, { "epoch": 0.0067328, "grad_norm": 0.826748788356781, "learning_rate": 1.6784902610057175e-06, "loss": 2.8023, "step": 317630 }, { "epoch": 0.0067584, "grad_norm": 0.8424851894378662, "learning_rate": 1.6780437214956102e-06, "loss": 2.6346, "step": 317640 }, { "epoch": 0.006784, "grad_norm": 0.9111089706420898, "learning_rate": 1.6775972359511772e-06, "loss": 2.8599, "step": 317650 }, { "epoch": 0.0068096, "grad_norm": 0.9504058361053467, "learning_rate": 1.677150804375317e-06, "loss": 2.8632, "step": 317660 }, { "epoch": 0.0068352, "grad_norm": 0.7616405487060547, "learning_rate": 1.6767044267709187e-06, "loss": 2.7826, "step": 317670 }, { "epoch": 0.0068608, "grad_norm": 0.9014577269554138, "learning_rate": 1.676258103140881e-06, "loss": 2.7839, "step": 317680 }, { "epoch": 0.0068864, "grad_norm": 0.8991861939430237, "learning_rate": 1.6758118334880968e-06, "loss": 2.4214, "step": 317690 }, { "epoch": 0.006912, "grad_norm": 0.796294629573822, "learning_rate": 1.6753656178154608e-06, "loss": 2.6461, "step": 317700 }, { "epoch": 0.0069376, "grad_norm": 3.2767980098724365, "learning_rate": 1.6749194561258665e-06, "loss": 2.9349, "step": 317710 }, { "epoch": 0.0069632, "grad_norm": 0.9449244737625122, "learning_rate": 1.6744733484222064e-06, "loss": 2.8049, "step": 317720 }, { "epoch": 0.0069888, "grad_norm": 0.8840083479881287, "learning_rate": 1.6740272947073766e-06, "loss": 2.883, "step": 317730 }, { "epoch": 0.0070144, "grad_norm": 0.8148043751716614, "learning_rate": 1.6735812949842644e-06, "loss": 2.6681, "step": 317740 }, { "epoch": 0.00704, "grad_norm": 0.8815723657608032, "learning_rate": 1.6731353492557655e-06, "loss": 2.7011, "step": 317750 }, { "epoch": 0.0070656, "grad_norm": 0.8278763294219971, "learning_rate": 1.6726894575247698e-06, "loss": 2.9532, "step": 317760 }, { "epoch": 0.0070912, "grad_norm": 0.7597824335098267, "learning_rate": 1.6722436197941705e-06, "loss": 2.8379, "step": 317770 }, { "epoch": 0.0071168, "grad_norm": 0.8582285642623901, "learning_rate": 1.6717978360668584e-06, "loss": 2.7146, "step": 317780 }, { "epoch": 0.0071424, "grad_norm": 0.8354753255844116, "learning_rate": 1.6713521063457228e-06, "loss": 2.6072, "step": 317790 }, { "epoch": 0.007168, "grad_norm": 0.8048455715179443, "learning_rate": 1.6709064306336564e-06, "loss": 2.9157, "step": 317800 }, { "epoch": 0.0071936, "grad_norm": 0.8681208491325378, "learning_rate": 1.6704608089335473e-06, "loss": 2.9758, "step": 317810 }, { "epoch": 0.0072192, "grad_norm": 0.8357909321784973, "learning_rate": 1.6700152412482872e-06, "loss": 2.9193, "step": 317820 }, { "epoch": 0.0072448, "grad_norm": 0.9051859378814697, "learning_rate": 1.6695697275807631e-06, "loss": 2.8279, "step": 317830 }, { "epoch": 0.0072704, "grad_norm": 0.8434496521949768, "learning_rate": 1.669124267933867e-06, "loss": 2.7508, "step": 317840 }, { "epoch": 0.007296, "grad_norm": 0.8511965274810791, "learning_rate": 1.6686788623104844e-06, "loss": 2.6227, "step": 317850 }, { "epoch": 0.0073216, "grad_norm": 0.8238007426261902, "learning_rate": 1.6682335107135083e-06, "loss": 2.8887, "step": 317860 }, { "epoch": 0.0073472, "grad_norm": 0.9619408845901489, "learning_rate": 1.6677882131458212e-06, "loss": 2.8215, "step": 317870 }, { "epoch": 0.0073728, "grad_norm": 1.050014853477478, "learning_rate": 1.6673429696103127e-06, "loss": 2.7354, "step": 317880 }, { "epoch": 0.0073984, "grad_norm": 0.8729065656661987, "learning_rate": 1.6668977801098706e-06, "loss": 2.9459, "step": 317890 }, { "epoch": 0.007424, "grad_norm": 0.8350881338119507, "learning_rate": 1.6664526446473816e-06, "loss": 2.9055, "step": 317900 }, { "epoch": 0.0074496, "grad_norm": 0.7855013608932495, "learning_rate": 1.6660075632257332e-06, "loss": 2.8259, "step": 317910 }, { "epoch": 0.0074752, "grad_norm": 0.9475581049919128, "learning_rate": 1.66556253584781e-06, "loss": 3.101, "step": 317920 }, { "epoch": 0.0075008, "grad_norm": 0.878860592842102, "learning_rate": 1.6651175625165018e-06, "loss": 2.7523, "step": 317930 }, { "epoch": 0.0075264, "grad_norm": 0.8252183794975281, "learning_rate": 1.6646726432346882e-06, "loss": 2.8543, "step": 317940 }, { "epoch": 0.007552, "grad_norm": 0.9223631620407104, "learning_rate": 1.6642277780052574e-06, "loss": 2.7401, "step": 317950 }, { "epoch": 0.0075776, "grad_norm": 0.8634454011917114, "learning_rate": 1.6637829668310934e-06, "loss": 2.7923, "step": 317960 }, { "epoch": 0.0076032, "grad_norm": 0.8400552272796631, "learning_rate": 1.663338209715083e-06, "loss": 2.7845, "step": 317970 }, { "epoch": 0.0076288, "grad_norm": 0.7545714378356934, "learning_rate": 1.6628935066601048e-06, "loss": 2.7775, "step": 317980 }, { "epoch": 0.0076544, "grad_norm": 0.7930981516838074, "learning_rate": 1.6624488576690513e-06, "loss": 2.6504, "step": 317990 }, { "epoch": 0.00768, "grad_norm": 0.751895546913147, "learning_rate": 1.6620042627447996e-06, "loss": 2.9256, "step": 318000 }, { "epoch": 0.0077056, "grad_norm": 0.8499965667724609, "learning_rate": 1.6615597218902335e-06, "loss": 2.8513, "step": 318010 }, { "epoch": 0.0077312, "grad_norm": 0.7972912192344666, "learning_rate": 1.6611152351082372e-06, "loss": 2.7757, "step": 318020 }, { "epoch": 0.0077568, "grad_norm": 0.82438063621521, "learning_rate": 1.6606708024016916e-06, "loss": 3.052, "step": 318030 }, { "epoch": 0.0077824, "grad_norm": 0.8151947855949402, "learning_rate": 1.66022642377348e-06, "loss": 2.9124, "step": 318040 }, { "epoch": 0.007808, "grad_norm": 0.8349981904029846, "learning_rate": 1.6597820992264834e-06, "loss": 2.8571, "step": 318050 }, { "epoch": 0.0078336, "grad_norm": 0.8843709826469421, "learning_rate": 1.6593378287635864e-06, "loss": 2.6464, "step": 318060 }, { "epoch": 0.0078592, "grad_norm": 0.8265356421470642, "learning_rate": 1.6588936123876641e-06, "loss": 2.795, "step": 318070 }, { "epoch": 0.0078848, "grad_norm": 0.7986303567886353, "learning_rate": 1.6584494501015992e-06, "loss": 2.7608, "step": 318080 }, { "epoch": 0.0079104, "grad_norm": 0.8503922820091248, "learning_rate": 1.658005341908273e-06, "loss": 2.8145, "step": 318090 }, { "epoch": 0.007936, "grad_norm": 0.8636364936828613, "learning_rate": 1.6575612878105662e-06, "loss": 2.8357, "step": 318100 }, { "epoch": 0.0079616, "grad_norm": 0.839560866355896, "learning_rate": 1.6571172878113562e-06, "loss": 2.6506, "step": 318110 }, { "epoch": 0.0079872, "grad_norm": 0.792046844959259, "learning_rate": 1.6566733419135239e-06, "loss": 2.7461, "step": 318120 }, { "epoch": 0.0080128, "grad_norm": 0.8861958384513855, "learning_rate": 1.6562294501199494e-06, "loss": 2.8731, "step": 318130 }, { "epoch": 0.0080384, "grad_norm": 0.820886492729187, "learning_rate": 1.655785612433507e-06, "loss": 2.6731, "step": 318140 }, { "epoch": 0.008064, "grad_norm": 0.9120973944664001, "learning_rate": 1.6553418288570777e-06, "loss": 2.8, "step": 318150 }, { "epoch": 0.0080896, "grad_norm": 1.014654517173767, "learning_rate": 1.654898099393536e-06, "loss": 2.882, "step": 318160 }, { "epoch": 0.0081152, "grad_norm": 0.8924068212509155, "learning_rate": 1.654454424045765e-06, "loss": 2.7357, "step": 318170 }, { "epoch": 0.0081408, "grad_norm": 0.8066496253013611, "learning_rate": 1.654010802816639e-06, "loss": 2.905, "step": 318180 }, { "epoch": 0.0081664, "grad_norm": 0.9460764527320862, "learning_rate": 1.6535672357090381e-06, "loss": 2.7838, "step": 318190 }, { "epoch": 0.008192, "grad_norm": 0.9262436628341675, "learning_rate": 1.6531237227258322e-06, "loss": 2.9151, "step": 318200 }, { "epoch": 0.0082176, "grad_norm": 0.8542307019233704, "learning_rate": 1.6526802638699002e-06, "loss": 2.7159, "step": 318210 }, { "epoch": 0.0082432, "grad_norm": 0.8397727012634277, "learning_rate": 1.6522368591441195e-06, "loss": 2.9215, "step": 318220 }, { "epoch": 0.0082688, "grad_norm": 0.8392531871795654, "learning_rate": 1.6517935085513636e-06, "loss": 2.7207, "step": 318230 }, { "epoch": 0.0082944, "grad_norm": 0.7965660095214844, "learning_rate": 1.651350212094508e-06, "loss": 2.8163, "step": 318240 }, { "epoch": 0.00832, "grad_norm": 0.9155621528625488, "learning_rate": 1.6509069697764267e-06, "loss": 2.7354, "step": 318250 }, { "epoch": 0.0083456, "grad_norm": 0.7833179831504822, "learning_rate": 1.650463781599998e-06, "loss": 2.8996, "step": 318260 }, { "epoch": 0.0083712, "grad_norm": 0.8449397087097168, "learning_rate": 1.6500206475680903e-06, "loss": 2.8477, "step": 318270 }, { "epoch": 0.0083968, "grad_norm": 0.8294820785522461, "learning_rate": 1.64957756768358e-06, "loss": 2.825, "step": 318280 }, { "epoch": 0.0084224, "grad_norm": 0.8929729461669922, "learning_rate": 1.6491345419493387e-06, "loss": 2.6422, "step": 318290 }, { "epoch": 0.008448, "grad_norm": 1.2500108480453491, "learning_rate": 1.6486915703682415e-06, "loss": 2.7499, "step": 318300 }, { "epoch": 0.0084736, "grad_norm": 0.8800353407859802, "learning_rate": 1.6482486529431596e-06, "loss": 2.9409, "step": 318310 }, { "epoch": 0.0084992, "grad_norm": 0.8560771346092224, "learning_rate": 1.647805789676965e-06, "loss": 2.8534, "step": 318320 }, { "epoch": 0.0085248, "grad_norm": 0.8973925113677979, "learning_rate": 1.647362980572531e-06, "loss": 2.8222, "step": 318330 }, { "epoch": 0.0085504, "grad_norm": 0.9835719466209412, "learning_rate": 1.6469202256327277e-06, "loss": 2.8066, "step": 318340 }, { "epoch": 0.008576, "grad_norm": 0.8113311529159546, "learning_rate": 1.646477524860427e-06, "loss": 2.6596, "step": 318350 }, { "epoch": 0.0086016, "grad_norm": 0.8012326955795288, "learning_rate": 1.6460348782584989e-06, "loss": 2.7893, "step": 318360 }, { "epoch": 0.0086272, "grad_norm": 0.8417705297470093, "learning_rate": 1.6455922858298158e-06, "loss": 2.9633, "step": 318370 }, { "epoch": 0.0086528, "grad_norm": 0.8315424919128418, "learning_rate": 1.645149747577245e-06, "loss": 2.8974, "step": 318380 }, { "epoch": 0.0086784, "grad_norm": 0.7704655528068542, "learning_rate": 1.6447072635036599e-06, "loss": 2.651, "step": 318390 }, { "epoch": 0.008704, "grad_norm": 0.9495509266853333, "learning_rate": 1.644264833611926e-06, "loss": 2.8668, "step": 318400 }, { "epoch": 0.0087296, "grad_norm": 0.8622226715087891, "learning_rate": 1.6438224579049144e-06, "loss": 2.802, "step": 318410 }, { "epoch": 0.0087552, "grad_norm": 0.8010839819908142, "learning_rate": 1.6433801363854918e-06, "loss": 2.8181, "step": 318420 }, { "epoch": 0.0087808, "grad_norm": 1.2336640357971191, "learning_rate": 1.6429378690565289e-06, "loss": 2.7959, "step": 318430 }, { "epoch": 0.0088064, "grad_norm": 0.7964609861373901, "learning_rate": 1.6424956559208938e-06, "loss": 2.787, "step": 318440 }, { "epoch": 0.008832, "grad_norm": 0.8717268109321594, "learning_rate": 1.6420534969814516e-06, "loss": 2.623, "step": 318450 }, { "epoch": 0.0088576, "grad_norm": 0.8418142199516296, "learning_rate": 1.6416113922410748e-06, "loss": 2.6896, "step": 318460 }, { "epoch": 0.0088832, "grad_norm": 0.8991169333457947, "learning_rate": 1.6411693417026241e-06, "loss": 2.8016, "step": 318470 }, { "epoch": 0.0089088, "grad_norm": 0.8858863115310669, "learning_rate": 1.6407273453689686e-06, "loss": 2.7597, "step": 318480 }, { "epoch": 0.0089344, "grad_norm": 0.7624420523643494, "learning_rate": 1.6402854032429749e-06, "loss": 2.9029, "step": 318490 }, { "epoch": 0.00896, "grad_norm": 0.7983106970787048, "learning_rate": 1.639843515327505e-06, "loss": 2.9727, "step": 318500 }, { "epoch": 0.0089856, "grad_norm": 0.8407912254333496, "learning_rate": 1.6394016816254322e-06, "loss": 2.7511, "step": 318510 }, { "epoch": 0.0090112, "grad_norm": 0.8185452818870544, "learning_rate": 1.6389599021396185e-06, "loss": 3.0046, "step": 318520 }, { "epoch": 0.0090368, "grad_norm": 0.846415638923645, "learning_rate": 1.6385181768729263e-06, "loss": 3.0259, "step": 318530 }, { "epoch": 0.0090624, "grad_norm": 0.7802665829658508, "learning_rate": 1.638076505828221e-06, "loss": 2.7816, "step": 318540 }, { "epoch": 0.009088, "grad_norm": 0.9224340319633484, "learning_rate": 1.637634889008367e-06, "loss": 3.0011, "step": 318550 }, { "epoch": 0.0091136, "grad_norm": 1.2396061420440674, "learning_rate": 1.6371933264162287e-06, "loss": 2.7688, "step": 318560 }, { "epoch": 0.0091392, "grad_norm": 0.9550679326057434, "learning_rate": 1.6367518180546693e-06, "loss": 2.8655, "step": 318570 }, { "epoch": 0.0091648, "grad_norm": 0.767719030380249, "learning_rate": 1.636310363926551e-06, "loss": 2.9195, "step": 318580 }, { "epoch": 0.0091904, "grad_norm": 0.8368779420852661, "learning_rate": 1.6358689640347404e-06, "loss": 2.7555, "step": 318590 }, { "epoch": 0.009216, "grad_norm": 0.8239015936851501, "learning_rate": 1.6354276183820939e-06, "loss": 2.6107, "step": 318600 }, { "epoch": 0.0092416, "grad_norm": 0.8425650596618652, "learning_rate": 1.634986326971476e-06, "loss": 2.7664, "step": 318610 }, { "epoch": 0.0092672, "grad_norm": 0.8886739015579224, "learning_rate": 1.634545089805748e-06, "loss": 3.0067, "step": 318620 }, { "epoch": 0.0092928, "grad_norm": 0.806139349937439, "learning_rate": 1.6341039068877728e-06, "loss": 2.8298, "step": 318630 }, { "epoch": 0.0093184, "grad_norm": 0.910942554473877, "learning_rate": 1.6336627782204107e-06, "loss": 2.8644, "step": 318640 }, { "epoch": 0.009344, "grad_norm": 0.7581536173820496, "learning_rate": 1.6332217038065212e-06, "loss": 2.7132, "step": 318650 }, { "epoch": 0.0093696, "grad_norm": 0.8955023884773254, "learning_rate": 1.6327806836489678e-06, "loss": 2.6764, "step": 318660 }, { "epoch": 0.0093952, "grad_norm": 0.864512026309967, "learning_rate": 1.632339717750606e-06, "loss": 2.9263, "step": 318670 }, { "epoch": 0.0094208, "grad_norm": 0.8332811594009399, "learning_rate": 1.631898806114295e-06, "loss": 2.9238, "step": 318680 }, { "epoch": 0.0094464, "grad_norm": 0.8476729393005371, "learning_rate": 1.6314579487428971e-06, "loss": 2.6649, "step": 318690 }, { "epoch": 0.009472, "grad_norm": 0.8604419827461243, "learning_rate": 1.631017145639272e-06, "loss": 2.9704, "step": 318700 }, { "epoch": 0.0094976, "grad_norm": 0.8951224088668823, "learning_rate": 1.630576396806276e-06, "loss": 2.8425, "step": 318710 }, { "epoch": 0.0095232, "grad_norm": 0.8941230177879333, "learning_rate": 1.6301357022467701e-06, "loss": 2.8948, "step": 318720 }, { "epoch": 0.0095488, "grad_norm": 0.8651641607284546, "learning_rate": 1.6296950619636076e-06, "loss": 2.8652, "step": 318730 }, { "epoch": 0.0095744, "grad_norm": 0.9075303077697754, "learning_rate": 1.6292544759596474e-06, "loss": 2.7585, "step": 318740 }, { "epoch": 0.0096, "grad_norm": 0.7868968844413757, "learning_rate": 1.6288139442377481e-06, "loss": 2.713, "step": 318750 }, { "epoch": 0.0096256, "grad_norm": 0.7703131437301636, "learning_rate": 1.6283734668007644e-06, "loss": 2.8571, "step": 318760 }, { "epoch": 0.0096512, "grad_norm": 0.8344651460647583, "learning_rate": 1.6279330436515551e-06, "loss": 3.0566, "step": 318770 }, { "epoch": 0.0096768, "grad_norm": 0.7949362993240356, "learning_rate": 1.6274926747929742e-06, "loss": 2.8018, "step": 318780 }, { "epoch": 0.0097024, "grad_norm": 0.8051497340202332, "learning_rate": 1.627052360227881e-06, "loss": 2.7541, "step": 318790 }, { "epoch": 0.009728, "grad_norm": 0.8915402293205261, "learning_rate": 1.626612099959125e-06, "loss": 2.8064, "step": 318800 }, { "epoch": 0.0097536, "grad_norm": 0.7831409573554993, "learning_rate": 1.6261718939895655e-06, "loss": 2.7734, "step": 318810 }, { "epoch": 0.0097792, "grad_norm": 0.8447940349578857, "learning_rate": 1.6257317423220543e-06, "loss": 2.838, "step": 318820 }, { "epoch": 0.0098048, "grad_norm": 0.8601551055908203, "learning_rate": 1.625291644959448e-06, "loss": 2.6555, "step": 318830 }, { "epoch": 0.0098304, "grad_norm": 0.8341624140739441, "learning_rate": 1.624851601904599e-06, "loss": 2.988, "step": 318840 }, { "epoch": 0.009856, "grad_norm": 0.8883026838302612, "learning_rate": 1.6244116131603615e-06, "loss": 2.7808, "step": 318850 }, { "epoch": 0.0098816, "grad_norm": 0.9138389825820923, "learning_rate": 1.6239716787295889e-06, "loss": 3.0437, "step": 318860 }, { "epoch": 0.0099072, "grad_norm": 0.8207789659500122, "learning_rate": 1.6235317986151345e-06, "loss": 3.1008, "step": 318870 }, { "epoch": 0.0099328, "grad_norm": 0.8560407161712646, "learning_rate": 1.6230919728198492e-06, "loss": 2.8339, "step": 318880 }, { "epoch": 0.0099584, "grad_norm": 0.9861081838607788, "learning_rate": 1.6226522013465873e-06, "loss": 2.7047, "step": 318890 }, { "epoch": 0.009984, "grad_norm": 0.8839560151100159, "learning_rate": 1.622212484198199e-06, "loss": 2.7871, "step": 318900 }, { "epoch": 0.0100096, "grad_norm": 0.8710094690322876, "learning_rate": 1.6217728213775351e-06, "loss": 2.8925, "step": 318910 }, { "epoch": 0.0100352, "grad_norm": 0.8156853914260864, "learning_rate": 1.6213332128874515e-06, "loss": 2.8983, "step": 318920 }, { "epoch": 0.0100608, "grad_norm": 0.7344606518745422, "learning_rate": 1.6208936587307922e-06, "loss": 2.827, "step": 318930 }, { "epoch": 0.0100864, "grad_norm": 0.830414354801178, "learning_rate": 1.6204541589104106e-06, "loss": 2.9038, "step": 318940 }, { "epoch": 0.010112, "grad_norm": 0.8120173215866089, "learning_rate": 1.6200147134291578e-06, "loss": 2.7983, "step": 318950 }, { "epoch": 0.0101376, "grad_norm": 0.897839367389679, "learning_rate": 1.6195753222898813e-06, "loss": 2.7553, "step": 318960 }, { "epoch": 0.0101632, "grad_norm": 0.9159312844276428, "learning_rate": 1.6191359854954313e-06, "loss": 2.6448, "step": 318970 }, { "epoch": 0.0101888, "grad_norm": 0.7959699630737305, "learning_rate": 1.6186967030486579e-06, "loss": 2.8871, "step": 318980 }, { "epoch": 0.0102144, "grad_norm": 0.782416045665741, "learning_rate": 1.6182574749524116e-06, "loss": 2.7006, "step": 318990 }, { "epoch": 0.01024, "grad_norm": 0.7544852495193481, "learning_rate": 1.617818301209535e-06, "loss": 2.8928, "step": 319000 }, { "epoch": 0.0102656, "grad_norm": 0.8222268223762512, "learning_rate": 1.617379181822879e-06, "loss": 2.8948, "step": 319010 }, { "epoch": 0.0102912, "grad_norm": 0.7999891042709351, "learning_rate": 1.6169401167952915e-06, "loss": 2.9282, "step": 319020 }, { "epoch": 0.0103168, "grad_norm": 0.9221393465995789, "learning_rate": 1.6165011061296165e-06, "loss": 3.0427, "step": 319030 }, { "epoch": 0.0103424, "grad_norm": 0.8600987195968628, "learning_rate": 1.6160621498287088e-06, "loss": 2.719, "step": 319040 }, { "epoch": 0.010368, "grad_norm": 0.8265026211738586, "learning_rate": 1.615623247895407e-06, "loss": 2.8627, "step": 319050 }, { "epoch": 0.0103936, "grad_norm": 1.0680872201919556, "learning_rate": 1.6151844003325612e-06, "loss": 2.7483, "step": 319060 }, { "epoch": 0.0104192, "grad_norm": 0.8659632802009583, "learning_rate": 1.6147456071430157e-06, "loss": 3.0201, "step": 319070 }, { "epoch": 0.0104448, "grad_norm": 0.8426744937896729, "learning_rate": 1.614306868329616e-06, "loss": 2.9285, "step": 319080 }, { "epoch": 0.0104704, "grad_norm": 0.7855110764503479, "learning_rate": 1.6138681838952075e-06, "loss": 2.8973, "step": 319090 }, { "epoch": 0.010496, "grad_norm": 1.1849316358566284, "learning_rate": 1.6134295538426358e-06, "loss": 3.0761, "step": 319100 }, { "epoch": 0.0105216, "grad_norm": 0.7643060684204102, "learning_rate": 1.6129909781747443e-06, "loss": 2.7603, "step": 319110 }, { "epoch": 0.0105472, "grad_norm": 0.9422360062599182, "learning_rate": 1.6125524568943795e-06, "loss": 2.8665, "step": 319120 }, { "epoch": 0.0105728, "grad_norm": 0.9032825827598572, "learning_rate": 1.6121139900043803e-06, "loss": 2.9479, "step": 319130 }, { "epoch": 0.0105984, "grad_norm": 0.7895042300224304, "learning_rate": 1.6116755775075932e-06, "loss": 2.6714, "step": 319140 }, { "epoch": 0.010624, "grad_norm": 0.9441823959350586, "learning_rate": 1.6112372194068604e-06, "loss": 2.9348, "step": 319150 }, { "epoch": 0.0106496, "grad_norm": 0.8621860146522522, "learning_rate": 1.6107989157050252e-06, "loss": 2.8791, "step": 319160 }, { "epoch": 0.0106752, "grad_norm": 0.7890901565551758, "learning_rate": 1.6103606664049287e-06, "loss": 2.8492, "step": 319170 }, { "epoch": 0.0107008, "grad_norm": 0.9467669129371643, "learning_rate": 1.6099224715094142e-06, "loss": 2.9047, "step": 319180 }, { "epoch": 0.0107264, "grad_norm": 0.9374650716781616, "learning_rate": 1.6094843310213237e-06, "loss": 2.6267, "step": 319190 }, { "epoch": 0.010752, "grad_norm": 0.9796484708786011, "learning_rate": 1.6090462449434941e-06, "loss": 2.7271, "step": 319200 }, { "epoch": 0.0107776, "grad_norm": 0.7991676926612854, "learning_rate": 1.6086082132787717e-06, "loss": 2.8741, "step": 319210 }, { "epoch": 0.0108032, "grad_norm": 0.7859036326408386, "learning_rate": 1.6081702360299944e-06, "loss": 3.1096, "step": 319220 }, { "epoch": 0.0108288, "grad_norm": 0.7748362421989441, "learning_rate": 1.6077323132000034e-06, "loss": 2.8109, "step": 319230 }, { "epoch": 0.0108544, "grad_norm": 0.9821614623069763, "learning_rate": 1.6072944447916393e-06, "loss": 2.9942, "step": 319240 }, { "epoch": 0.01088, "grad_norm": 0.7601692080497742, "learning_rate": 1.6068566308077393e-06, "loss": 2.7959, "step": 319250 }, { "epoch": 0.0109056, "grad_norm": 0.813878059387207, "learning_rate": 1.606418871251142e-06, "loss": 2.9525, "step": 319260 }, { "epoch": 0.0109312, "grad_norm": 0.8438072800636292, "learning_rate": 1.6059811661246882e-06, "loss": 2.906, "step": 319270 }, { "epoch": 0.0109568, "grad_norm": 0.9336647987365723, "learning_rate": 1.605543515431216e-06, "loss": 2.8092, "step": 319280 }, { "epoch": 0.0109824, "grad_norm": 0.8800127506256104, "learning_rate": 1.605105919173563e-06, "loss": 2.9097, "step": 319290 }, { "epoch": 0.011008, "grad_norm": 1.0661460161209106, "learning_rate": 1.6046683773545667e-06, "loss": 3.0157, "step": 319300 }, { "epoch": 0.0110336, "grad_norm": 0.8910713791847229, "learning_rate": 1.6042308899770643e-06, "loss": 2.8322, "step": 319310 }, { "epoch": 0.0110592, "grad_norm": 1.0206717252731323, "learning_rate": 1.6037934570438963e-06, "loss": 3.0261, "step": 319320 }, { "epoch": 0.0110848, "grad_norm": 0.7887061834335327, "learning_rate": 1.603356078557894e-06, "loss": 2.7885, "step": 319330 }, { "epoch": 0.0111104, "grad_norm": 0.8432223200798035, "learning_rate": 1.602918754521896e-06, "loss": 2.9367, "step": 319340 }, { "epoch": 0.011136, "grad_norm": 0.812514066696167, "learning_rate": 1.6024814849387383e-06, "loss": 2.7871, "step": 319350 }, { "epoch": 0.0111616, "grad_norm": 0.8744617700576782, "learning_rate": 1.602044269811256e-06, "loss": 2.6617, "step": 319360 }, { "epoch": 0.0111872, "grad_norm": 0.9492151141166687, "learning_rate": 1.6016071091422847e-06, "loss": 2.7108, "step": 319370 }, { "epoch": 0.0112128, "grad_norm": 0.811055600643158, "learning_rate": 1.6011700029346588e-06, "loss": 2.7904, "step": 319380 }, { "epoch": 0.0112384, "grad_norm": 1.143799901008606, "learning_rate": 1.600732951191214e-06, "loss": 3.0297, "step": 319390 }, { "epoch": 0.011264, "grad_norm": 1.0040112733840942, "learning_rate": 1.6002959539147833e-06, "loss": 3.0072, "step": 319400 }, { "epoch": 0.0112896, "grad_norm": 0.864819347858429, "learning_rate": 1.5998590111082013e-06, "loss": 2.9351, "step": 319410 }, { "epoch": 0.0113152, "grad_norm": 0.8909944295883179, "learning_rate": 1.5994221227743013e-06, "loss": 2.7572, "step": 319420 }, { "epoch": 0.0113408, "grad_norm": 0.8682795166969299, "learning_rate": 1.5989852889159163e-06, "loss": 2.7002, "step": 319430 }, { "epoch": 0.0113664, "grad_norm": 0.8949958682060242, "learning_rate": 1.598548509535881e-06, "loss": 3.0248, "step": 319440 }, { "epoch": 0.011392, "grad_norm": 0.8855903744697571, "learning_rate": 1.5981117846370232e-06, "loss": 2.7891, "step": 319450 }, { "epoch": 0.0114176, "grad_norm": 0.9338037371635437, "learning_rate": 1.5976751142221781e-06, "loss": 2.7071, "step": 319460 }, { "epoch": 0.0114432, "grad_norm": 0.9075767397880554, "learning_rate": 1.597238498294177e-06, "loss": 2.9786, "step": 319470 }, { "epoch": 0.0114688, "grad_norm": 0.8554239869117737, "learning_rate": 1.5968019368558496e-06, "loss": 2.6166, "step": 319480 }, { "epoch": 0.0114944, "grad_norm": 0.9642243385314941, "learning_rate": 1.5963654299100295e-06, "loss": 3.3782, "step": 319490 }, { "epoch": 0.01152, "grad_norm": 0.8488925099372864, "learning_rate": 1.5959289774595465e-06, "loss": 2.8046, "step": 319500 }, { "epoch": 0.0115456, "grad_norm": 0.9034964442253113, "learning_rate": 1.5954925795072297e-06, "loss": 2.8847, "step": 319510 }, { "epoch": 0.0115712, "grad_norm": 0.8060333728790283, "learning_rate": 1.595056236055913e-06, "loss": 2.6052, "step": 319520 }, { "epoch": 0.0115968, "grad_norm": 1.2193081378936768, "learning_rate": 1.59461994710842e-06, "loss": 2.8977, "step": 319530 }, { "epoch": 0.0116224, "grad_norm": 0.8639826774597168, "learning_rate": 1.594183712667583e-06, "loss": 2.6494, "step": 319540 }, { "epoch": 0.011648, "grad_norm": 0.8098084330558777, "learning_rate": 1.5937475327362272e-06, "loss": 2.7374, "step": 319550 }, { "epoch": 0.0116736, "grad_norm": 0.8065200448036194, "learning_rate": 1.5933114073171885e-06, "loss": 2.9138, "step": 319560 }, { "epoch": 0.0116992, "grad_norm": 1.0546917915344238, "learning_rate": 1.592875336413292e-06, "loss": 3.0371, "step": 319570 }, { "epoch": 0.0117248, "grad_norm": 0.9723545908927917, "learning_rate": 1.5924393200273635e-06, "loss": 2.7174, "step": 319580 }, { "epoch": 0.0117504, "grad_norm": 0.8992778062820435, "learning_rate": 1.5920033581622308e-06, "loss": 2.9611, "step": 319590 }, { "epoch": 0.011776, "grad_norm": 0.9134630560874939, "learning_rate": 1.5915674508207214e-06, "loss": 2.942, "step": 319600 }, { "epoch": 0.0118016, "grad_norm": 0.8583469390869141, "learning_rate": 1.591131598005663e-06, "loss": 2.8922, "step": 319610 }, { "epoch": 0.0118272, "grad_norm": 0.8137094378471375, "learning_rate": 1.5906957997198812e-06, "loss": 2.9346, "step": 319620 }, { "epoch": 0.0118528, "grad_norm": 0.7852658033370972, "learning_rate": 1.590260055966203e-06, "loss": 2.56, "step": 319630 }, { "epoch": 0.0118784, "grad_norm": 0.8708332180976868, "learning_rate": 1.5898243667474544e-06, "loss": 2.7681, "step": 319640 }, { "epoch": 0.011904, "grad_norm": 0.8316553831100464, "learning_rate": 1.589388732066458e-06, "loss": 2.8975, "step": 319650 }, { "epoch": 0.0119296, "grad_norm": 1.1915942430496216, "learning_rate": 1.5889531519260403e-06, "loss": 2.224, "step": 319660 }, { "epoch": 0.0119552, "grad_norm": 0.7833628058433533, "learning_rate": 1.5885176263290258e-06, "loss": 2.7336, "step": 319670 }, { "epoch": 0.0119808, "grad_norm": 0.8974377512931824, "learning_rate": 1.58808215527824e-06, "loss": 2.4415, "step": 319680 }, { "epoch": 0.0120064, "grad_norm": 0.8451530337333679, "learning_rate": 1.587646738776505e-06, "loss": 2.9608, "step": 319690 }, { "epoch": 0.012032, "grad_norm": 0.8330399394035339, "learning_rate": 1.587211376826644e-06, "loss": 2.9001, "step": 319700 }, { "epoch": 0.0120576, "grad_norm": 0.7905153036117554, "learning_rate": 1.5867760694314848e-06, "loss": 2.4605, "step": 319710 }, { "epoch": 0.0120832, "grad_norm": 0.7910964488983154, "learning_rate": 1.586340816593842e-06, "loss": 2.7279, "step": 319720 }, { "epoch": 0.0121088, "grad_norm": 0.8735989332199097, "learning_rate": 1.5859056183165455e-06, "loss": 2.9683, "step": 319730 }, { "epoch": 0.0121344, "grad_norm": 0.8871212005615234, "learning_rate": 1.5854704746024153e-06, "loss": 2.7909, "step": 319740 }, { "epoch": 0.01216, "grad_norm": 0.7929089665412903, "learning_rate": 1.5850353854542722e-06, "loss": 2.8696, "step": 319750 }, { "epoch": 0.0121856, "grad_norm": 0.9749642610549927, "learning_rate": 1.5846003508749385e-06, "loss": 3.0332, "step": 319760 }, { "epoch": 0.0122112, "grad_norm": 0.976744532585144, "learning_rate": 1.5841653708672366e-06, "loss": 2.7951, "step": 319770 }, { "epoch": 0.0122368, "grad_norm": 0.8432798981666565, "learning_rate": 1.583730445433984e-06, "loss": 2.9556, "step": 319780 }, { "epoch": 0.0122624, "grad_norm": 0.7778620719909668, "learning_rate": 1.5832955745780021e-06, "loss": 2.8269, "step": 319790 }, { "epoch": 0.012288, "grad_norm": 0.9373905658721924, "learning_rate": 1.5828607583021128e-06, "loss": 2.9455, "step": 319800 }, { "epoch": 0.0123136, "grad_norm": 0.8505869507789612, "learning_rate": 1.5824259966091327e-06, "loss": 2.8653, "step": 319810 }, { "epoch": 0.0123392, "grad_norm": 0.789056658744812, "learning_rate": 1.581991289501884e-06, "loss": 2.9514, "step": 319820 }, { "epoch": 0.0123648, "grad_norm": 0.8966996073722839, "learning_rate": 1.5815566369831836e-06, "loss": 2.9686, "step": 319830 }, { "epoch": 0.0123904, "grad_norm": 0.8265936374664307, "learning_rate": 1.5811220390558546e-06, "loss": 2.8282, "step": 319840 }, { "epoch": 0.012416, "grad_norm": 0.8214167356491089, "learning_rate": 1.5806874957227093e-06, "loss": 2.7929, "step": 319850 }, { "epoch": 0.0124416, "grad_norm": 1.0152653455734253, "learning_rate": 1.5802530069865674e-06, "loss": 2.7828, "step": 319860 }, { "epoch": 0.0124672, "grad_norm": 0.7808579802513123, "learning_rate": 1.579818572850247e-06, "loss": 2.8596, "step": 319870 }, { "epoch": 0.0124928, "grad_norm": 1.0226106643676758, "learning_rate": 1.5793841933165654e-06, "loss": 2.8482, "step": 319880 }, { "epoch": 0.0125184, "grad_norm": 0.9124919176101685, "learning_rate": 1.5789498683883397e-06, "loss": 2.866, "step": 319890 }, { "epoch": 0.012544, "grad_norm": 0.9602745175361633, "learning_rate": 1.5785155980683853e-06, "loss": 3.1359, "step": 319900 }, { "epoch": 0.0125696, "grad_norm": 1.290654182434082, "learning_rate": 1.5780813823595197e-06, "loss": 2.5357, "step": 319910 }, { "epoch": 0.0125952, "grad_norm": 0.8189373016357422, "learning_rate": 1.5776472212645576e-06, "loss": 2.4335, "step": 319920 }, { "epoch": 0.0126208, "grad_norm": 0.7849692702293396, "learning_rate": 1.5772131147863158e-06, "loss": 2.4817, "step": 319930 }, { "epoch": 0.0126464, "grad_norm": 0.8437617421150208, "learning_rate": 1.5767790629276082e-06, "loss": 2.7506, "step": 319940 }, { "epoch": 0.012672, "grad_norm": 0.7961561679840088, "learning_rate": 1.5763450656912493e-06, "loss": 2.7433, "step": 319950 }, { "epoch": 0.0126976, "grad_norm": 0.7983205914497375, "learning_rate": 1.575911123080054e-06, "loss": 2.6923, "step": 319960 }, { "epoch": 0.0127232, "grad_norm": 0.8379310965538025, "learning_rate": 1.5754772350968384e-06, "loss": 2.9231, "step": 319970 }, { "epoch": 0.0127488, "grad_norm": 1.1445093154907227, "learning_rate": 1.5750434017444117e-06, "loss": 2.8505, "step": 319980 }, { "epoch": 0.0127744, "grad_norm": 0.8738371133804321, "learning_rate": 1.574609623025589e-06, "loss": 2.9737, "step": 319990 }, { "epoch": 0.0128, "grad_norm": 0.9538337588310242, "learning_rate": 1.5741758989431855e-06, "loss": 2.7717, "step": 320000 }, { "epoch": 0.0128256, "grad_norm": 0.9366641640663147, "learning_rate": 1.5737422295000104e-06, "loss": 2.7071, "step": 320010 }, { "epoch": 0.0128512, "grad_norm": 0.8083219528198242, "learning_rate": 1.5733086146988775e-06, "loss": 2.8072, "step": 320020 }, { "epoch": 0.0128768, "grad_norm": 1.3360133171081543, "learning_rate": 1.5728750545425997e-06, "loss": 2.6931, "step": 320030 }, { "epoch": 0.0129024, "grad_norm": 1.013251781463623, "learning_rate": 1.5724415490339883e-06, "loss": 2.822, "step": 320040 }, { "epoch": 0.012928, "grad_norm": 0.8090710639953613, "learning_rate": 1.5720080981758524e-06, "loss": 2.8274, "step": 320050 }, { "epoch": 0.0129536, "grad_norm": 0.9063915014266968, "learning_rate": 1.5715747019710038e-06, "loss": 2.7284, "step": 320060 }, { "epoch": 0.0129792, "grad_norm": 0.82509845495224, "learning_rate": 1.5711413604222514e-06, "loss": 2.7913, "step": 320070 }, { "epoch": 0.0130048, "grad_norm": 0.8726847171783447, "learning_rate": 1.5707080735324087e-06, "loss": 2.8001, "step": 320080 }, { "epoch": 0.0130304, "grad_norm": 0.8275505900382996, "learning_rate": 1.5702748413042844e-06, "loss": 2.7807, "step": 320090 }, { "epoch": 0.013056, "grad_norm": 0.9607316255569458, "learning_rate": 1.56984166374069e-06, "loss": 2.8952, "step": 320100 }, { "epoch": 0.0130816, "grad_norm": 1.0372166633605957, "learning_rate": 1.5694085408444292e-06, "loss": 3.0264, "step": 320110 }, { "epoch": 0.0131072, "grad_norm": 1.0967165231704712, "learning_rate": 1.5689754726183138e-06, "loss": 2.7493, "step": 320120 }, { "epoch": 0.0131328, "grad_norm": 1.0810762643814087, "learning_rate": 1.568542459065151e-06, "loss": 2.7144, "step": 320130 }, { "epoch": 0.0131584, "grad_norm": 0.8451375365257263, "learning_rate": 1.56810950018775e-06, "loss": 2.8115, "step": 320140 }, { "epoch": 0.013184, "grad_norm": 0.9307593703269958, "learning_rate": 1.5676765959889183e-06, "loss": 3.1583, "step": 320150 }, { "epoch": 0.0132096, "grad_norm": 0.9274953007698059, "learning_rate": 1.5672437464714629e-06, "loss": 2.722, "step": 320160 }, { "epoch": 0.0132352, "grad_norm": 0.8169254660606384, "learning_rate": 1.5668109516381923e-06, "loss": 2.9659, "step": 320170 }, { "epoch": 0.0132608, "grad_norm": 1.0001708269119263, "learning_rate": 1.566378211491909e-06, "loss": 2.8514, "step": 320180 }, { "epoch": 0.0132864, "grad_norm": 0.8092613816261292, "learning_rate": 1.5659455260354228e-06, "loss": 2.8274, "step": 320190 }, { "epoch": 0.013312, "grad_norm": 0.8820048570632935, "learning_rate": 1.5655128952715371e-06, "loss": 2.7807, "step": 320200 }, { "epoch": 0.0133376, "grad_norm": 1.0027424097061157, "learning_rate": 1.5650803192030596e-06, "loss": 3.1016, "step": 320210 }, { "epoch": 0.0133632, "grad_norm": 0.7972676753997803, "learning_rate": 1.5646477978327933e-06, "loss": 2.6875, "step": 320220 }, { "epoch": 0.0133888, "grad_norm": 0.9065211415290833, "learning_rate": 1.5642153311635445e-06, "loss": 2.6731, "step": 320230 }, { "epoch": 0.0134144, "grad_norm": 0.8233668208122253, "learning_rate": 1.563782919198118e-06, "loss": 2.8958, "step": 320240 }, { "epoch": 0.01344, "grad_norm": 0.812251627445221, "learning_rate": 1.5633505619393175e-06, "loss": 2.7649, "step": 320250 }, { "epoch": 0.0134656, "grad_norm": 1.9993213415145874, "learning_rate": 1.562918259389945e-06, "loss": 3.0375, "step": 320260 }, { "epoch": 0.0134912, "grad_norm": 1.4494892358779907, "learning_rate": 1.5624860115528061e-06, "loss": 3.0901, "step": 320270 }, { "epoch": 0.0135168, "grad_norm": 0.8414632678031921, "learning_rate": 1.5620538184307031e-06, "loss": 2.8213, "step": 320280 }, { "epoch": 0.0135424, "grad_norm": 0.9112334251403809, "learning_rate": 1.5616216800264383e-06, "loss": 2.7179, "step": 320290 }, { "epoch": 0.013568, "grad_norm": 1.0238507986068726, "learning_rate": 1.561189596342817e-06, "loss": 2.8199, "step": 320300 }, { "epoch": 0.0135936, "grad_norm": 0.8913611769676208, "learning_rate": 1.5607575673826358e-06, "loss": 2.886, "step": 320310 }, { "epoch": 0.0136192, "grad_norm": 0.8353785276412964, "learning_rate": 1.5603255931486982e-06, "loss": 2.8719, "step": 320320 }, { "epoch": 0.0136448, "grad_norm": 0.8828496336936951, "learning_rate": 1.5598936736438074e-06, "loss": 2.8122, "step": 320330 }, { "epoch": 0.0136704, "grad_norm": 0.8412123918533325, "learning_rate": 1.5594618088707626e-06, "loss": 3.0413, "step": 320340 }, { "epoch": 0.013696, "grad_norm": 0.7785853743553162, "learning_rate": 1.5590299988323642e-06, "loss": 2.8298, "step": 320350 }, { "epoch": 0.0137216, "grad_norm": 0.8924732208251953, "learning_rate": 1.5585982435314129e-06, "loss": 3.0843, "step": 320360 }, { "epoch": 0.0137472, "grad_norm": 0.8858059048652649, "learning_rate": 1.5581665429707115e-06, "loss": 2.8257, "step": 320370 }, { "epoch": 0.0137728, "grad_norm": 0.8134853839874268, "learning_rate": 1.5577348971530538e-06, "loss": 2.9309, "step": 320380 }, { "epoch": 0.0137984, "grad_norm": 0.9781696796417236, "learning_rate": 1.5573033060812426e-06, "loss": 2.9002, "step": 320390 }, { "epoch": 0.013824, "grad_norm": 0.8993415832519531, "learning_rate": 1.5568717697580738e-06, "loss": 2.9325, "step": 320400 }, { "epoch": 0.0138496, "grad_norm": 0.8170048594474792, "learning_rate": 1.5564402881863494e-06, "loss": 2.8411, "step": 320410 }, { "epoch": 0.0138752, "grad_norm": 0.8696295619010925, "learning_rate": 1.5560088613688617e-06, "loss": 2.5613, "step": 320420 }, { "epoch": 0.0139008, "grad_norm": 0.7284035086631775, "learning_rate": 1.5555774893084186e-06, "loss": 2.9412, "step": 320430 }, { "epoch": 0.0139264, "grad_norm": 0.8184749484062195, "learning_rate": 1.5551461720078077e-06, "loss": 2.9007, "step": 320440 }, { "epoch": 0.013952, "grad_norm": 0.8155785202980042, "learning_rate": 1.55471490946983e-06, "loss": 3.0368, "step": 320450 }, { "epoch": 0.0139776, "grad_norm": 0.7708503007888794, "learning_rate": 1.5542837016972823e-06, "loss": 2.9023, "step": 320460 }, { "epoch": 0.0140032, "grad_norm": 0.8339158892631531, "learning_rate": 1.553852548692959e-06, "loss": 2.7655, "step": 320470 }, { "epoch": 0.0140288, "grad_norm": 0.8497856855392456, "learning_rate": 1.5534214504596578e-06, "loss": 2.9155, "step": 320480 }, { "epoch": 0.0140544, "grad_norm": 0.7759959697723389, "learning_rate": 1.5529904070001745e-06, "loss": 2.8794, "step": 320490 }, { "epoch": 0.01408, "grad_norm": 0.8966677188873291, "learning_rate": 1.5525594183173042e-06, "loss": 2.8703, "step": 320500 }, { "epoch": 0.0141056, "grad_norm": 1.1333202123641968, "learning_rate": 1.5521284844138395e-06, "loss": 2.8939, "step": 320510 }, { "epoch": 0.0141312, "grad_norm": 0.9528945088386536, "learning_rate": 1.551697605292577e-06, "loss": 2.9931, "step": 320520 }, { "epoch": 0.0141568, "grad_norm": 0.8970118761062622, "learning_rate": 1.5512667809563087e-06, "loss": 2.7619, "step": 320530 }, { "epoch": 0.0141824, "grad_norm": 0.7949325442314148, "learning_rate": 1.5508360114078315e-06, "loss": 2.8588, "step": 320540 }, { "epoch": 0.014208, "grad_norm": 0.8195101618766785, "learning_rate": 1.5504052966499362e-06, "loss": 2.9059, "step": 320550 }, { "epoch": 0.0142336, "grad_norm": 0.8175750970840454, "learning_rate": 1.5499746366854175e-06, "loss": 2.7678, "step": 320560 }, { "epoch": 0.0142592, "grad_norm": 0.9192161560058594, "learning_rate": 1.5495440315170695e-06, "loss": 2.9165, "step": 320570 }, { "epoch": 0.0142848, "grad_norm": 0.9378548264503479, "learning_rate": 1.5491134811476804e-06, "loss": 2.907, "step": 320580 }, { "epoch": 0.0143104, "grad_norm": 0.8264772891998291, "learning_rate": 1.548682985580041e-06, "loss": 2.8878, "step": 320590 }, { "epoch": 0.014336, "grad_norm": 1.0428282022476196, "learning_rate": 1.5482525448169504e-06, "loss": 2.9296, "step": 320600 }, { "epoch": 0.0143616, "grad_norm": 0.9351959824562073, "learning_rate": 1.5478221588611953e-06, "loss": 2.8376, "step": 320610 }, { "epoch": 0.0143872, "grad_norm": 0.8327926993370056, "learning_rate": 1.5473918277155663e-06, "loss": 2.9041, "step": 320620 }, { "epoch": 0.0144128, "grad_norm": 0.8747662901878357, "learning_rate": 1.5469615513828584e-06, "loss": 2.8465, "step": 320630 }, { "epoch": 0.0144384, "grad_norm": 1.0215468406677246, "learning_rate": 1.5465313298658557e-06, "loss": 2.6896, "step": 320640 }, { "epoch": 0.014464, "grad_norm": 0.8512827754020691, "learning_rate": 1.5461011631673506e-06, "loss": 2.7119, "step": 320650 }, { "epoch": 0.0144896, "grad_norm": 0.8499526381492615, "learning_rate": 1.5456710512901319e-06, "loss": 2.9726, "step": 320660 }, { "epoch": 0.0145152, "grad_norm": 0.8072376847267151, "learning_rate": 1.5452409942369906e-06, "loss": 3.0015, "step": 320670 }, { "epoch": 0.0145408, "grad_norm": 0.7806615233421326, "learning_rate": 1.5448109920107136e-06, "loss": 2.7295, "step": 320680 }, { "epoch": 0.0145664, "grad_norm": 0.964297890663147, "learning_rate": 1.5443810446140905e-06, "loss": 2.8163, "step": 320690 }, { "epoch": 0.014592, "grad_norm": 0.8242721557617188, "learning_rate": 1.5439511520499118e-06, "loss": 2.8999, "step": 320700 }, { "epoch": 0.0146176, "grad_norm": 0.8757292032241821, "learning_rate": 1.5435213143209604e-06, "loss": 3.0005, "step": 320710 }, { "epoch": 0.0146432, "grad_norm": 0.8133063912391663, "learning_rate": 1.5430915314300255e-06, "loss": 2.8639, "step": 320720 }, { "epoch": 0.0146688, "grad_norm": 0.789625346660614, "learning_rate": 1.5426618033798935e-06, "loss": 2.7231, "step": 320730 }, { "epoch": 0.0146944, "grad_norm": 1.1563762426376343, "learning_rate": 1.5422321301733535e-06, "loss": 2.8303, "step": 320740 }, { "epoch": 0.01472, "grad_norm": 1.1298896074295044, "learning_rate": 1.5418025118131896e-06, "loss": 2.653, "step": 320750 }, { "epoch": 0.0147456, "grad_norm": 0.7920631766319275, "learning_rate": 1.5413729483021878e-06, "loss": 2.8227, "step": 320760 }, { "epoch": 0.0147712, "grad_norm": 0.9129799604415894, "learning_rate": 1.5409434396431355e-06, "loss": 2.8685, "step": 320770 }, { "epoch": 0.0147968, "grad_norm": 0.8330540060997009, "learning_rate": 1.5405139858388162e-06, "loss": 2.8177, "step": 320780 }, { "epoch": 0.0148224, "grad_norm": 0.9616000056266785, "learning_rate": 1.5400845868920156e-06, "loss": 2.6649, "step": 320790 }, { "epoch": 0.014848, "grad_norm": 0.7613072395324707, "learning_rate": 1.539655242805519e-06, "loss": 2.7593, "step": 320800 }, { "epoch": 0.0148736, "grad_norm": 0.8287034034729004, "learning_rate": 1.5392259535821086e-06, "loss": 3.0204, "step": 320810 }, { "epoch": 0.0148992, "grad_norm": 0.8675479888916016, "learning_rate": 1.5387967192245711e-06, "loss": 2.541, "step": 320820 }, { "epoch": 0.0149248, "grad_norm": 0.7432018518447876, "learning_rate": 1.5383675397356867e-06, "loss": 2.6567, "step": 320830 }, { "epoch": 0.0149504, "grad_norm": 0.8397000432014465, "learning_rate": 1.5379384151182408e-06, "loss": 2.7896, "step": 320840 }, { "epoch": 0.014976, "grad_norm": 0.8501406908035278, "learning_rate": 1.5375093453750135e-06, "loss": 2.787, "step": 320850 }, { "epoch": 0.0150016, "grad_norm": 0.8981488347053528, "learning_rate": 1.5370803305087901e-06, "loss": 2.8394, "step": 320860 }, { "epoch": 0.0150272, "grad_norm": 0.8234351277351379, "learning_rate": 1.536651370522352e-06, "loss": 2.6923, "step": 320870 }, { "epoch": 0.0150528, "grad_norm": 1.3787330389022827, "learning_rate": 1.5362224654184799e-06, "loss": 2.5034, "step": 320880 }, { "epoch": 0.0150784, "grad_norm": 0.9610555768013, "learning_rate": 1.5357936151999553e-06, "loss": 2.9016, "step": 320890 }, { "epoch": 0.015104, "grad_norm": 0.8024911880493164, "learning_rate": 1.5353648198695625e-06, "loss": 2.7993, "step": 320900 }, { "epoch": 0.0151296, "grad_norm": 1.235075831413269, "learning_rate": 1.5349360794300761e-06, "loss": 2.9475, "step": 320910 }, { "epoch": 0.0151552, "grad_norm": 0.9092020392417908, "learning_rate": 1.5345073938842803e-06, "loss": 2.6981, "step": 320920 }, { "epoch": 0.0151808, "grad_norm": 0.8713021278381348, "learning_rate": 1.534078763234954e-06, "loss": 2.9171, "step": 320930 }, { "epoch": 0.0152064, "grad_norm": 0.7956317663192749, "learning_rate": 1.5336501874848741e-06, "loss": 2.5783, "step": 320940 }, { "epoch": 0.015232, "grad_norm": 0.7930163145065308, "learning_rate": 1.5332216666368282e-06, "loss": 2.8677, "step": 320950 }, { "epoch": 0.0152576, "grad_norm": 0.8647500872612, "learning_rate": 1.5327932006935875e-06, "loss": 2.8207, "step": 320960 }, { "epoch": 0.0152832, "grad_norm": 0.780215859413147, "learning_rate": 1.532364789657932e-06, "loss": 2.8669, "step": 320970 }, { "epoch": 0.0153088, "grad_norm": 0.7802602052688599, "learning_rate": 1.5319364335326404e-06, "loss": 2.7081, "step": 320980 }, { "epoch": 0.0153344, "grad_norm": 1.1615772247314453, "learning_rate": 1.5315081323204895e-06, "loss": 2.7577, "step": 320990 }, { "epoch": 0.01536, "grad_norm": 0.7708740234375, "learning_rate": 1.5310798860242594e-06, "loss": 2.9097, "step": 321000 }, { "epoch": 0.0153856, "grad_norm": 0.791070282459259, "learning_rate": 1.5306516946467242e-06, "loss": 2.7304, "step": 321010 }, { "epoch": 0.0154112, "grad_norm": 0.955266535282135, "learning_rate": 1.5302235581906645e-06, "loss": 2.8894, "step": 321020 }, { "epoch": 0.0154368, "grad_norm": 0.8706053495407104, "learning_rate": 1.529795476658852e-06, "loss": 2.8643, "step": 321030 }, { "epoch": 0.0154624, "grad_norm": 1.1536688804626465, "learning_rate": 1.5293674500540644e-06, "loss": 2.8951, "step": 321040 }, { "epoch": 0.015488, "grad_norm": 0.8323767185211182, "learning_rate": 1.5289394783790779e-06, "loss": 2.7606, "step": 321050 }, { "epoch": 0.0155136, "grad_norm": 0.8296545147895813, "learning_rate": 1.5285115616366674e-06, "loss": 2.7732, "step": 321060 }, { "epoch": 0.0155392, "grad_norm": 0.7829445004463196, "learning_rate": 1.528083699829609e-06, "loss": 3.1577, "step": 321070 }, { "epoch": 0.0155648, "grad_norm": 0.7519069314002991, "learning_rate": 1.5276558929606754e-06, "loss": 2.9915, "step": 321080 }, { "epoch": 0.0155904, "grad_norm": 0.8207312822341919, "learning_rate": 1.5272281410326418e-06, "loss": 2.7636, "step": 321090 }, { "epoch": 0.015616, "grad_norm": 0.8026759624481201, "learning_rate": 1.526800444048283e-06, "loss": 2.7888, "step": 321100 }, { "epoch": 0.0156416, "grad_norm": 0.8686150312423706, "learning_rate": 1.5263728020103685e-06, "loss": 2.7288, "step": 321110 }, { "epoch": 0.0156672, "grad_norm": 0.7955247163772583, "learning_rate": 1.5259452149216758e-06, "loss": 2.6991, "step": 321120 }, { "epoch": 0.0156928, "grad_norm": 0.8450497984886169, "learning_rate": 1.525517682784976e-06, "loss": 2.72, "step": 321130 }, { "epoch": 0.0157184, "grad_norm": 0.89143967628479, "learning_rate": 1.5250902056030425e-06, "loss": 2.9712, "step": 321140 }, { "epoch": 0.015744, "grad_norm": 0.7925034165382385, "learning_rate": 1.5246627833786487e-06, "loss": 2.724, "step": 321150 }, { "epoch": 0.0157696, "grad_norm": 0.7542381286621094, "learning_rate": 1.5242354161145624e-06, "loss": 2.7345, "step": 321160 }, { "epoch": 0.0157952, "grad_norm": 0.7928304076194763, "learning_rate": 1.5238081038135565e-06, "loss": 2.7295, "step": 321170 }, { "epoch": 0.0158208, "grad_norm": 0.8327104449272156, "learning_rate": 1.5233808464784017e-06, "loss": 2.6904, "step": 321180 }, { "epoch": 0.0158464, "grad_norm": 0.8346956372261047, "learning_rate": 1.5229536441118698e-06, "loss": 2.8316, "step": 321190 }, { "epoch": 0.015872, "grad_norm": 0.9387426376342773, "learning_rate": 1.5225264967167307e-06, "loss": 2.7484, "step": 321200 }, { "epoch": 0.0158976, "grad_norm": 0.7987387180328369, "learning_rate": 1.5220994042957538e-06, "loss": 2.6068, "step": 321210 }, { "epoch": 0.0159232, "grad_norm": 0.85788494348526, "learning_rate": 1.5216723668517119e-06, "loss": 2.7866, "step": 321220 }, { "epoch": 0.0159488, "grad_norm": 1.4376603364944458, "learning_rate": 1.5212453843873698e-06, "loss": 2.4516, "step": 321230 }, { "epoch": 0.0159744, "grad_norm": 0.854022741317749, "learning_rate": 1.5208184569054962e-06, "loss": 2.9806, "step": 321240 }, { "epoch": 0.016, "grad_norm": 0.8472647070884705, "learning_rate": 1.5203915844088624e-06, "loss": 2.7673, "step": 321250 }, { "epoch": 0.0160256, "grad_norm": 0.8502015471458435, "learning_rate": 1.519964766900236e-06, "loss": 2.7908, "step": 321260 }, { "epoch": 0.0160512, "grad_norm": 0.807176411151886, "learning_rate": 1.5195380043823837e-06, "loss": 2.826, "step": 321270 }, { "epoch": 0.0160768, "grad_norm": 0.8379207253456116, "learning_rate": 1.5191112968580745e-06, "loss": 2.7933, "step": 321280 }, { "epoch": 0.0161024, "grad_norm": 0.8458442687988281, "learning_rate": 1.518684644330074e-06, "loss": 2.7182, "step": 321290 }, { "epoch": 0.016128, "grad_norm": 0.8919006586074829, "learning_rate": 1.5182580468011487e-06, "loss": 2.7069, "step": 321300 }, { "epoch": 0.0161536, "grad_norm": 0.8129432201385498, "learning_rate": 1.5178315042740667e-06, "loss": 2.7617, "step": 321310 }, { "epoch": 0.0161792, "grad_norm": 0.8450589179992676, "learning_rate": 1.517405016751593e-06, "loss": 2.7169, "step": 321320 }, { "epoch": 0.0162048, "grad_norm": 0.9472866058349609, "learning_rate": 1.5169785842364938e-06, "loss": 2.8202, "step": 321330 }, { "epoch": 0.0162304, "grad_norm": 0.821946918964386, "learning_rate": 1.516552206731533e-06, "loss": 2.8853, "step": 321340 }, { "epoch": 0.016256, "grad_norm": 0.8578867316246033, "learning_rate": 1.5161258842394787e-06, "loss": 2.8749, "step": 321350 }, { "epoch": 0.0162816, "grad_norm": 0.9264203906059265, "learning_rate": 1.5156996167630922e-06, "loss": 2.8246, "step": 321360 }, { "epoch": 0.0163072, "grad_norm": 0.8868758082389832, "learning_rate": 1.5152734043051375e-06, "loss": 2.9609, "step": 321370 }, { "epoch": 0.0163328, "grad_norm": 0.8166624307632446, "learning_rate": 1.5148472468683807e-06, "loss": 3.0247, "step": 321380 }, { "epoch": 0.0163584, "grad_norm": 0.7855727076530457, "learning_rate": 1.5144211444555846e-06, "loss": 2.7598, "step": 321390 }, { "epoch": 0.016384, "grad_norm": 0.9400821924209595, "learning_rate": 1.5139950970695117e-06, "loss": 2.8745, "step": 321400 }, { "epoch": 2.56e-05, "grad_norm": 0.8854219913482666, "learning_rate": 1.5135691047129253e-06, "loss": 1.8358, "step": 321410 }, { "epoch": 5.12e-05, "grad_norm": 0.8610938191413879, "learning_rate": 1.513143167388591e-06, "loss": 2.6454, "step": 321420 }, { "epoch": 7.68e-05, "grad_norm": 1.127235770225525, "learning_rate": 1.5127172850992645e-06, "loss": 2.7936, "step": 321430 }, { "epoch": 0.0001024, "grad_norm": 0.7362218499183655, "learning_rate": 1.512291457847711e-06, "loss": 2.449, "step": 321440 }, { "epoch": 0.000128, "grad_norm": 0.8100122809410095, "learning_rate": 1.511865685636691e-06, "loss": 2.5703, "step": 321450 }, { "epoch": 0.0001536, "grad_norm": 0.7797660231590271, "learning_rate": 1.511439968468964e-06, "loss": 2.556, "step": 321460 }, { "epoch": 0.0001792, "grad_norm": 0.876829981803894, "learning_rate": 1.511014306347296e-06, "loss": 2.7453, "step": 321470 }, { "epoch": 0.0002048, "grad_norm": 1.3262587785720825, "learning_rate": 1.5105886992744456e-06, "loss": 2.4486, "step": 321480 }, { "epoch": 0.0002304, "grad_norm": 0.9946447014808655, "learning_rate": 1.5101631472531697e-06, "loss": 2.7497, "step": 321490 }, { "epoch": 0.000256, "grad_norm": 0.7956300377845764, "learning_rate": 1.5097376502862293e-06, "loss": 2.4959, "step": 321500 }, { "epoch": 0.0002816, "grad_norm": 0.8522949814796448, "learning_rate": 1.5093122083763835e-06, "loss": 2.6463, "step": 321510 }, { "epoch": 0.0003072, "grad_norm": 0.8700830936431885, "learning_rate": 1.5088868215263908e-06, "loss": 2.4755, "step": 321520 }, { "epoch": 0.0003328, "grad_norm": 0.8941854238510132, "learning_rate": 1.5084614897390115e-06, "loss": 2.714, "step": 321530 }, { "epoch": 0.0003584, "grad_norm": 0.9121322631835938, "learning_rate": 1.5080362130170023e-06, "loss": 2.7594, "step": 321540 }, { "epoch": 0.000384, "grad_norm": 0.8299432396888733, "learning_rate": 1.5076109913631232e-06, "loss": 2.5829, "step": 321550 }, { "epoch": 0.0004096, "grad_norm": 0.8958306312561035, "learning_rate": 1.5071858247801285e-06, "loss": 2.5864, "step": 321560 }, { "epoch": 0.0004352, "grad_norm": 0.7907649874687195, "learning_rate": 1.5067607132707761e-06, "loss": 2.7659, "step": 321570 }, { "epoch": 0.0004608, "grad_norm": 0.8403810262680054, "learning_rate": 1.506335656837823e-06, "loss": 2.3256, "step": 321580 }, { "epoch": 0.0004864, "grad_norm": 0.8505673408508301, "learning_rate": 1.5059106554840265e-06, "loss": 2.5779, "step": 321590 }, { "epoch": 0.000512, "grad_norm": 0.7590788006782532, "learning_rate": 1.505485709212141e-06, "loss": 2.7271, "step": 321600 }, { "epoch": 0.0005376, "grad_norm": 0.8120777010917664, "learning_rate": 1.5050608180249237e-06, "loss": 2.4719, "step": 321610 }, { "epoch": 0.0005632, "grad_norm": 0.7669692635536194, "learning_rate": 1.504635981925131e-06, "loss": 2.3621, "step": 321620 }, { "epoch": 0.0005888, "grad_norm": 0.9468953013420105, "learning_rate": 1.5042112009155152e-06, "loss": 2.6526, "step": 321630 }, { "epoch": 0.0006144, "grad_norm": 0.8486085534095764, "learning_rate": 1.5037864749988284e-06, "loss": 2.526, "step": 321640 }, { "epoch": 0.00064, "grad_norm": 1.1732362508773804, "learning_rate": 1.5033618041778308e-06, "loss": 3.1225, "step": 321650 }, { "epoch": 0.0006656, "grad_norm": 0.8834819197654724, "learning_rate": 1.5029371884552735e-06, "loss": 2.901, "step": 321660 }, { "epoch": 0.0006912, "grad_norm": 0.8927028179168701, "learning_rate": 1.502512627833911e-06, "loss": 2.8162, "step": 321670 }, { "epoch": 0.0007168, "grad_norm": 0.8613739609718323, "learning_rate": 1.5020881223164973e-06, "loss": 2.6369, "step": 321680 }, { "epoch": 0.0007424, "grad_norm": 0.9959481954574585, "learning_rate": 1.501663671905782e-06, "loss": 2.7027, "step": 321690 }, { "epoch": 0.000768, "grad_norm": 0.8292495012283325, "learning_rate": 1.5012392766045192e-06, "loss": 2.6038, "step": 321700 }, { "epoch": 0.0007936, "grad_norm": 0.8588194847106934, "learning_rate": 1.50081493641546e-06, "loss": 2.6593, "step": 321710 }, { "epoch": 0.0008192, "grad_norm": 0.8053372502326965, "learning_rate": 1.500390651341358e-06, "loss": 2.6624, "step": 321720 }, { "epoch": 0.0008448, "grad_norm": 0.8296688199043274, "learning_rate": 1.499966421384964e-06, "loss": 2.6931, "step": 321730 }, { "epoch": 0.0008704, "grad_norm": 0.8054256439208984, "learning_rate": 1.4995422465490283e-06, "loss": 2.646, "step": 321740 }, { "epoch": 0.000896, "grad_norm": 1.4215199947357178, "learning_rate": 1.4991181268363043e-06, "loss": 2.676, "step": 321750 }, { "epoch": 0.0009216, "grad_norm": 0.8644999861717224, "learning_rate": 1.4986940622495372e-06, "loss": 2.4642, "step": 321760 }, { "epoch": 0.0009472, "grad_norm": 0.7085437774658203, "learning_rate": 1.4982700527914806e-06, "loss": 2.5468, "step": 321770 }, { "epoch": 0.0009728, "grad_norm": 0.8204348683357239, "learning_rate": 1.4978460984648824e-06, "loss": 2.5725, "step": 321780 }, { "epoch": 0.0009984, "grad_norm": 0.9438749551773071, "learning_rate": 1.4974221992724935e-06, "loss": 2.7161, "step": 321790 }, { "epoch": 0.001024, "grad_norm": 0.7544596195220947, "learning_rate": 1.4969983552170608e-06, "loss": 2.7367, "step": 321800 }, { "epoch": 0.0010496, "grad_norm": 0.98482346534729, "learning_rate": 1.4965745663013341e-06, "loss": 2.4557, "step": 321810 }, { "epoch": 0.0010752, "grad_norm": 0.8052130937576294, "learning_rate": 1.4961508325280627e-06, "loss": 2.6369, "step": 321820 }, { "epoch": 0.0011008, "grad_norm": 0.892658531665802, "learning_rate": 1.4957271538999918e-06, "loss": 2.5315, "step": 321830 }, { "epoch": 0.0011264, "grad_norm": 0.8720229864120483, "learning_rate": 1.4953035304198715e-06, "loss": 2.4607, "step": 321840 }, { "epoch": 0.001152, "grad_norm": 0.8705998063087463, "learning_rate": 1.4948799620904464e-06, "loss": 2.7423, "step": 321850 }, { "epoch": 0.0011776, "grad_norm": 0.9467238783836365, "learning_rate": 1.4944564489144642e-06, "loss": 2.8916, "step": 321860 }, { "epoch": 0.0012032, "grad_norm": 0.892967700958252, "learning_rate": 1.4940329908946716e-06, "loss": 2.5875, "step": 321870 }, { "epoch": 0.0012288, "grad_norm": 0.9135511517524719, "learning_rate": 1.4936095880338174e-06, "loss": 3.0268, "step": 321880 }, { "epoch": 0.0012544, "grad_norm": 0.8309637904167175, "learning_rate": 1.493186240334642e-06, "loss": 2.9058, "step": 321890 }, { "epoch": 0.00128, "grad_norm": 0.9494144916534424, "learning_rate": 1.4927629477998916e-06, "loss": 2.5461, "step": 321900 }, { "epoch": 0.0013056, "grad_norm": 0.8849063515663147, "learning_rate": 1.4923397104323145e-06, "loss": 2.6076, "step": 321910 }, { "epoch": 0.0013312, "grad_norm": 0.9042947292327881, "learning_rate": 1.4919165282346515e-06, "loss": 2.9275, "step": 321920 }, { "epoch": 0.0013568, "grad_norm": 0.75406813621521, "learning_rate": 1.4914934012096493e-06, "loss": 2.6346, "step": 321930 }, { "epoch": 0.0013824, "grad_norm": 0.8465492129325867, "learning_rate": 1.4910703293600515e-06, "loss": 2.6376, "step": 321940 }, { "epoch": 0.001408, "grad_norm": 0.8248223066329956, "learning_rate": 1.4906473126886033e-06, "loss": 2.6964, "step": 321950 }, { "epoch": 0.0014336, "grad_norm": 0.9099382758140564, "learning_rate": 1.490224351198044e-06, "loss": 2.7264, "step": 321960 }, { "epoch": 0.0014592, "grad_norm": 0.8632528185844421, "learning_rate": 1.4898014448911168e-06, "loss": 2.6188, "step": 321970 }, { "epoch": 0.0014848, "grad_norm": 0.8281935453414917, "learning_rate": 1.4893785937705662e-06, "loss": 2.3552, "step": 321980 }, { "epoch": 0.0015104, "grad_norm": 0.8145989179611206, "learning_rate": 1.4889557978391322e-06, "loss": 2.1682, "step": 321990 }, { "epoch": 0.001536, "grad_norm": 0.8293755054473877, "learning_rate": 1.4885330570995594e-06, "loss": 2.6168, "step": 322000 }, { "epoch": 2.56e-05, "grad_norm": 1.065846562385559, "learning_rate": 1.4881103715545898e-06, "loss": 1.7689, "step": 322010 }, { "epoch": 5.12e-05, "grad_norm": 0.8842432498931885, "learning_rate": 1.487687741206959e-06, "loss": 2.6284, "step": 322020 }, { "epoch": 7.68e-05, "grad_norm": 0.7858076691627502, "learning_rate": 1.4872651660594128e-06, "loss": 2.5821, "step": 322030 }, { "epoch": 0.0001024, "grad_norm": 0.7869444489479065, "learning_rate": 1.4868426461146878e-06, "loss": 2.5662, "step": 322040 }, { "epoch": 0.000128, "grad_norm": 0.824336588382721, "learning_rate": 1.4864201813755275e-06, "loss": 2.6275, "step": 322050 }, { "epoch": 0.0001536, "grad_norm": 0.9811485409736633, "learning_rate": 1.4859977718446684e-06, "loss": 2.4788, "step": 322060 }, { "epoch": 0.0001792, "grad_norm": 0.9659757614135742, "learning_rate": 1.4855754175248516e-06, "loss": 2.6224, "step": 322070 }, { "epoch": 0.0002048, "grad_norm": 0.8346571326255798, "learning_rate": 1.4851531184188173e-06, "loss": 2.75, "step": 322080 }, { "epoch": 0.0002304, "grad_norm": 0.840732753276825, "learning_rate": 1.4847308745292998e-06, "loss": 3.0699, "step": 322090 }, { "epoch": 0.000256, "grad_norm": 0.8594624996185303, "learning_rate": 1.4843086858590394e-06, "loss": 2.4972, "step": 322100 }, { "epoch": 0.0002816, "grad_norm": 0.8075671195983887, "learning_rate": 1.4838865524107748e-06, "loss": 2.3917, "step": 322110 }, { "epoch": 0.0003072, "grad_norm": 0.8542338609695435, "learning_rate": 1.4834644741872417e-06, "loss": 2.4758, "step": 322120 }, { "epoch": 0.0003328, "grad_norm": 0.8911715745925903, "learning_rate": 1.4830424511911778e-06, "loss": 2.6437, "step": 322130 }, { "epoch": 0.0003584, "grad_norm": 0.8797551989555359, "learning_rate": 1.4826204834253212e-06, "loss": 2.5458, "step": 322140 }, { "epoch": 0.000384, "grad_norm": 0.8643655180931091, "learning_rate": 1.4821985708924092e-06, "loss": 2.7214, "step": 322150 }, { "epoch": 0.0004096, "grad_norm": 0.8626958131790161, "learning_rate": 1.4817767135951699e-06, "loss": 2.654, "step": 322160 }, { "epoch": 0.0004352, "grad_norm": 0.8707854151725769, "learning_rate": 1.481354911536348e-06, "loss": 2.671, "step": 322170 }, { "epoch": 0.0004608, "grad_norm": 0.9334884881973267, "learning_rate": 1.4809331647186754e-06, "loss": 2.5396, "step": 322180 }, { "epoch": 0.0004864, "grad_norm": 0.8224497437477112, "learning_rate": 1.4805114731448865e-06, "loss": 2.5711, "step": 322190 }, { "epoch": 0.000512, "grad_norm": 0.8343178629875183, "learning_rate": 1.4800898368177174e-06, "loss": 2.4071, "step": 322200 }, { "epoch": 0.0005376, "grad_norm": 0.9174846410751343, "learning_rate": 1.4796682557399033e-06, "loss": 2.4528, "step": 322210 }, { "epoch": 0.0005632, "grad_norm": 0.9573124647140503, "learning_rate": 1.4792467299141745e-06, "loss": 2.4242, "step": 322220 }, { "epoch": 0.0005888, "grad_norm": 0.9196113348007202, "learning_rate": 1.478825259343265e-06, "loss": 2.5478, "step": 322230 }, { "epoch": 0.0006144, "grad_norm": 0.8654990196228027, "learning_rate": 1.4784038440299099e-06, "loss": 2.4162, "step": 322240 }, { "epoch": 0.00064, "grad_norm": 0.8748064041137695, "learning_rate": 1.477982483976841e-06, "loss": 3.045, "step": 322250 }, { "epoch": 0.0006656, "grad_norm": 0.9078973531723022, "learning_rate": 1.4775611791867906e-06, "loss": 3.0023, "step": 322260 }, { "epoch": 0.0006912, "grad_norm": 0.8825674057006836, "learning_rate": 1.477139929662491e-06, "loss": 2.4093, "step": 322270 }, { "epoch": 0.0007168, "grad_norm": 0.8354582786560059, "learning_rate": 1.4767187354066758e-06, "loss": 2.6565, "step": 322280 }, { "epoch": 0.0007424, "grad_norm": 0.9481065273284912, "learning_rate": 1.4762975964220726e-06, "loss": 2.4184, "step": 322290 }, { "epoch": 0.000768, "grad_norm": 0.8581199049949646, "learning_rate": 1.475876512711414e-06, "loss": 2.653, "step": 322300 }, { "epoch": 2.56e-05, "grad_norm": 0.7991110682487488, "learning_rate": 1.4754554842774304e-06, "loss": 1.745, "step": 322310 }, { "epoch": 5.12e-05, "grad_norm": 0.9888393878936768, "learning_rate": 1.4750345111228526e-06, "loss": 2.7818, "step": 322320 }, { "epoch": 7.68e-05, "grad_norm": 0.8949400782585144, "learning_rate": 1.4746135932504102e-06, "loss": 2.7497, "step": 322330 }, { "epoch": 0.0001024, "grad_norm": 0.827796220779419, "learning_rate": 1.4741927306628335e-06, "loss": 2.3972, "step": 322340 }, { "epoch": 0.000128, "grad_norm": 0.9531009197235107, "learning_rate": 1.47377192336285e-06, "loss": 2.717, "step": 322350 }, { "epoch": 0.0001536, "grad_norm": 0.8958378434181213, "learning_rate": 1.473351171353191e-06, "loss": 2.5011, "step": 322360 }, { "epoch": 0.0001792, "grad_norm": 0.7666257619857788, "learning_rate": 1.4729304746365824e-06, "loss": 2.6323, "step": 322370 }, { "epoch": 0.0002048, "grad_norm": 0.9320635795593262, "learning_rate": 1.472509833215754e-06, "loss": 2.849, "step": 322380 }, { "epoch": 0.0002304, "grad_norm": 0.9403603076934814, "learning_rate": 1.4720892470934334e-06, "loss": 2.6149, "step": 322390 }, { "epoch": 0.000256, "grad_norm": 0.7870754599571228, "learning_rate": 1.4716687162723463e-06, "loss": 2.6123, "step": 322400 }, { "epoch": 0.0002816, "grad_norm": 0.8566220998764038, "learning_rate": 1.4712482407552242e-06, "loss": 2.5648, "step": 322410 }, { "epoch": 0.0003072, "grad_norm": 0.8519991040229797, "learning_rate": 1.470827820544788e-06, "loss": 2.4823, "step": 322420 }, { "epoch": 0.0003328, "grad_norm": 0.8331223726272583, "learning_rate": 1.4704074556437675e-06, "loss": 2.6025, "step": 322430 }, { "epoch": 0.0003584, "grad_norm": 0.910211980342865, "learning_rate": 1.4699871460548865e-06, "loss": 2.465, "step": 322440 }, { "epoch": 0.000384, "grad_norm": 1.0863332748413086, "learning_rate": 1.4695668917808726e-06, "loss": 2.5732, "step": 322450 }, { "epoch": 0.0004096, "grad_norm": 0.8396347761154175, "learning_rate": 1.4691466928244502e-06, "loss": 2.6937, "step": 322460 }, { "epoch": 0.0004352, "grad_norm": 0.819551408290863, "learning_rate": 1.468726549188344e-06, "loss": 2.7735, "step": 322470 }, { "epoch": 0.0004608, "grad_norm": 0.892474889755249, "learning_rate": 1.4683064608752806e-06, "loss": 2.6036, "step": 322480 }, { "epoch": 0.0004864, "grad_norm": 0.7501943111419678, "learning_rate": 1.4678864278879812e-06, "loss": 2.4248, "step": 322490 }, { "epoch": 0.000512, "grad_norm": 0.8021207451820374, "learning_rate": 1.4674664502291703e-06, "loss": 2.7361, "step": 322500 }, { "epoch": 0.0005376, "grad_norm": 0.7561236023902893, "learning_rate": 1.467046527901569e-06, "loss": 2.7372, "step": 322510 }, { "epoch": 0.0005632, "grad_norm": 0.882474422454834, "learning_rate": 1.466626660907906e-06, "loss": 2.7289, "step": 322520 }, { "epoch": 0.0005888, "grad_norm": 0.8604881167411804, "learning_rate": 1.466206849250903e-06, "loss": 2.3991, "step": 322530 }, { "epoch": 0.0006144, "grad_norm": 0.8845801949501038, "learning_rate": 1.4657870929332785e-06, "loss": 2.637, "step": 322540 }, { "epoch": 0.00064, "grad_norm": 0.8743137121200562, "learning_rate": 1.465367391957757e-06, "loss": 2.6192, "step": 322550 }, { "epoch": 0.0006656, "grad_norm": 0.8945808410644531, "learning_rate": 1.46494774632706e-06, "loss": 2.9383, "step": 322560 }, { "epoch": 0.0006912, "grad_norm": 0.8875822424888611, "learning_rate": 1.4645281560439072e-06, "loss": 2.7224, "step": 322570 }, { "epoch": 0.0007168, "grad_norm": 1.0193350315093994, "learning_rate": 1.4641086211110222e-06, "loss": 2.6361, "step": 322580 }, { "epoch": 0.0007424, "grad_norm": 0.8239039182662964, "learning_rate": 1.463689141531124e-06, "loss": 2.5894, "step": 322590 }, { "epoch": 0.000768, "grad_norm": 0.8570990562438965, "learning_rate": 1.4632697173069321e-06, "loss": 2.4385, "step": 322600 }, { "epoch": 0.0007936, "grad_norm": 0.8401039242744446, "learning_rate": 1.4628503484411704e-06, "loss": 2.4798, "step": 322610 }, { "epoch": 0.0008192, "grad_norm": 0.8059440851211548, "learning_rate": 1.4624310349365524e-06, "loss": 2.5641, "step": 322620 }, { "epoch": 0.0008448, "grad_norm": 0.8534198999404907, "learning_rate": 1.4620117767958009e-06, "loss": 2.6641, "step": 322630 }, { "epoch": 0.0008704, "grad_norm": 0.812639594078064, "learning_rate": 1.461592574021633e-06, "loss": 2.6795, "step": 322640 }, { "epoch": 0.000896, "grad_norm": 0.8351761698722839, "learning_rate": 1.4611734266167688e-06, "loss": 2.4172, "step": 322650 }, { "epoch": 0.0009216, "grad_norm": 0.787489652633667, "learning_rate": 1.4607543345839248e-06, "loss": 2.5255, "step": 322660 }, { "epoch": 0.0009472, "grad_norm": 0.8193091750144958, "learning_rate": 1.46033529792582e-06, "loss": 2.7939, "step": 322670 }, { "epoch": 0.0009728, "grad_norm": 0.8522488474845886, "learning_rate": 1.4599163166451713e-06, "loss": 2.433, "step": 322680 }, { "epoch": 0.0009984, "grad_norm": 0.8342142701148987, "learning_rate": 1.4594973907446953e-06, "loss": 2.3586, "step": 322690 }, { "epoch": 0.001024, "grad_norm": 0.8540018200874329, "learning_rate": 1.4590785202271096e-06, "loss": 2.5863, "step": 322700 }, { "epoch": 0.0010496, "grad_norm": 0.8480803966522217, "learning_rate": 1.458659705095129e-06, "loss": 2.8011, "step": 322710 }, { "epoch": 0.0010752, "grad_norm": 0.8832823038101196, "learning_rate": 1.4582409453514713e-06, "loss": 2.607, "step": 322720 }, { "epoch": 0.0011008, "grad_norm": 0.8566228747367859, "learning_rate": 1.4578222409988518e-06, "loss": 2.3847, "step": 322730 }, { "epoch": 0.0011264, "grad_norm": 0.9247416257858276, "learning_rate": 1.4574035920399833e-06, "loss": 2.8078, "step": 322740 }, { "epoch": 0.001152, "grad_norm": 0.9721366167068481, "learning_rate": 1.456984998477582e-06, "loss": 2.8987, "step": 322750 }, { "epoch": 0.0011776, "grad_norm": 0.970325231552124, "learning_rate": 1.4565664603143626e-06, "loss": 2.8594, "step": 322760 }, { "epoch": 0.0012032, "grad_norm": 0.8841304183006287, "learning_rate": 1.4561479775530397e-06, "loss": 2.4926, "step": 322770 }, { "epoch": 0.0012288, "grad_norm": 0.8974308967590332, "learning_rate": 1.4557295501963254e-06, "loss": 2.7787, "step": 322780 }, { "epoch": 0.0012544, "grad_norm": 0.9248987436294556, "learning_rate": 1.4553111782469342e-06, "loss": 2.6298, "step": 322790 }, { "epoch": 0.00128, "grad_norm": 0.9145703911781311, "learning_rate": 1.454892861707582e-06, "loss": 2.5322, "step": 322800 }, { "epoch": 0.0013056, "grad_norm": 0.8870411515235901, "learning_rate": 1.4544746005809763e-06, "loss": 2.6478, "step": 322810 }, { "epoch": 0.0013312, "grad_norm": 0.7974299788475037, "learning_rate": 1.4540563948698317e-06, "loss": 2.5504, "step": 322820 }, { "epoch": 0.0013568, "grad_norm": 0.8634669780731201, "learning_rate": 1.4536382445768593e-06, "loss": 3.0558, "step": 322830 }, { "epoch": 0.0013824, "grad_norm": 0.8205018043518066, "learning_rate": 1.4532201497047728e-06, "loss": 2.635, "step": 322840 }, { "epoch": 0.001408, "grad_norm": 0.8433655500411987, "learning_rate": 1.4528021102562818e-06, "loss": 2.7249, "step": 322850 }, { "epoch": 0.0014336, "grad_norm": 0.9701104164123535, "learning_rate": 1.4523841262340965e-06, "loss": 2.7285, "step": 322860 }, { "epoch": 0.0014592, "grad_norm": 0.8526414036750793, "learning_rate": 1.4519661976409294e-06, "loss": 2.4863, "step": 322870 }, { "epoch": 0.0014848, "grad_norm": 1.0099695920944214, "learning_rate": 1.4515483244794903e-06, "loss": 1.9086, "step": 322880 }, { "epoch": 0.0015104, "grad_norm": 1.3552556037902832, "learning_rate": 1.4511305067524872e-06, "loss": 2.6607, "step": 322890 }, { "epoch": 0.001536, "grad_norm": 0.8666316866874695, "learning_rate": 1.4507127444626312e-06, "loss": 2.7395, "step": 322900 }, { "epoch": 0.0015616, "grad_norm": 0.8105942010879517, "learning_rate": 1.4502950376126312e-06, "loss": 2.7288, "step": 322910 }, { "epoch": 0.0015872, "grad_norm": 0.830518901348114, "learning_rate": 1.4498773862051952e-06, "loss": 2.5759, "step": 322920 }, { "epoch": 0.0016128, "grad_norm": 0.8837395310401917, "learning_rate": 1.449459790243034e-06, "loss": 2.65, "step": 322930 }, { "epoch": 0.0016384, "grad_norm": 0.8728108406066895, "learning_rate": 1.4490422497288527e-06, "loss": 2.8458, "step": 322940 }, { "epoch": 0.001664, "grad_norm": 0.9424389004707336, "learning_rate": 1.4486247646653584e-06, "loss": 2.7356, "step": 322950 }, { "epoch": 0.0016896, "grad_norm": 0.8410218954086304, "learning_rate": 1.4482073350552606e-06, "loss": 2.4765, "step": 322960 }, { "epoch": 0.0017152, "grad_norm": 0.9175885915756226, "learning_rate": 1.4477899609012646e-06, "loss": 2.7195, "step": 322970 }, { "epoch": 0.0017408, "grad_norm": 0.9517049193382263, "learning_rate": 1.4473726422060786e-06, "loss": 3.027, "step": 322980 }, { "epoch": 0.0017664, "grad_norm": 0.8142020106315613, "learning_rate": 1.4469553789724078e-06, "loss": 2.4341, "step": 322990 }, { "epoch": 0.001792, "grad_norm": 0.7797682881355286, "learning_rate": 1.4465381712029602e-06, "loss": 2.8141, "step": 323000 }, { "epoch": 0.0018176, "grad_norm": 0.7688355445861816, "learning_rate": 1.4461210189004382e-06, "loss": 2.4641, "step": 323010 }, { "epoch": 0.0018432, "grad_norm": 0.9788914918899536, "learning_rate": 1.4457039220675473e-06, "loss": 2.7715, "step": 323020 }, { "epoch": 0.0018688, "grad_norm": 1.0301531553268433, "learning_rate": 1.4452868807069898e-06, "loss": 2.8401, "step": 323030 }, { "epoch": 0.0018944, "grad_norm": 0.8028956055641174, "learning_rate": 1.4448698948214767e-06, "loss": 2.7782, "step": 323040 }, { "epoch": 0.00192, "grad_norm": 0.8046851754188538, "learning_rate": 1.4444529644137084e-06, "loss": 2.8791, "step": 323050 }, { "epoch": 0.0019456, "grad_norm": 0.8932695388793945, "learning_rate": 1.4440360894863913e-06, "loss": 2.1256, "step": 323060 }, { "epoch": 0.0019712, "grad_norm": 1.0552564859390259, "learning_rate": 1.4436192700422236e-06, "loss": 2.5145, "step": 323070 }, { "epoch": 0.0019968, "grad_norm": 0.8494528532028198, "learning_rate": 1.4432025060839105e-06, "loss": 2.7761, "step": 323080 }, { "epoch": 0.0020224, "grad_norm": 0.932649552822113, "learning_rate": 1.4427857976141557e-06, "loss": 2.4804, "step": 323090 }, { "epoch": 0.002048, "grad_norm": 0.8635276556015015, "learning_rate": 1.4423691446356614e-06, "loss": 2.6071, "step": 323100 }, { "epoch": 0.0020736, "grad_norm": 1.0139414072036743, "learning_rate": 1.4419525471511275e-06, "loss": 2.8828, "step": 323110 }, { "epoch": 0.0020992, "grad_norm": 0.8664255142211914, "learning_rate": 1.4415360051632577e-06, "loss": 2.7846, "step": 323120 }, { "epoch": 0.0021248, "grad_norm": 0.8088088035583496, "learning_rate": 1.441119518674754e-06, "loss": 2.739, "step": 323130 }, { "epoch": 0.0021504, "grad_norm": 0.8034974932670593, "learning_rate": 1.4407030876883143e-06, "loss": 2.7207, "step": 323140 }, { "epoch": 0.002176, "grad_norm": 0.8560853004455566, "learning_rate": 1.4402867122066388e-06, "loss": 2.7583, "step": 323150 }, { "epoch": 0.0022016, "grad_norm": 0.75776207447052, "learning_rate": 1.4398703922324309e-06, "loss": 2.532, "step": 323160 }, { "epoch": 0.0022272, "grad_norm": 0.8207771182060242, "learning_rate": 1.439454127768387e-06, "loss": 2.7489, "step": 323170 }, { "epoch": 0.0022528, "grad_norm": 1.1669766902923584, "learning_rate": 1.4390379188172087e-06, "loss": 2.7603, "step": 323180 }, { "epoch": 0.0022784, "grad_norm": 0.8414551019668579, "learning_rate": 1.4386217653815938e-06, "loss": 2.7818, "step": 323190 }, { "epoch": 0.002304, "grad_norm": 0.8808318972587585, "learning_rate": 1.438205667464242e-06, "loss": 2.7425, "step": 323200 }, { "epoch": 0.0023296, "grad_norm": 0.8154032230377197, "learning_rate": 1.4377896250678503e-06, "loss": 2.4232, "step": 323210 }, { "epoch": 0.0023552, "grad_norm": 0.9303374886512756, "learning_rate": 1.4373736381951175e-06, "loss": 2.6478, "step": 323220 }, { "epoch": 0.0023808, "grad_norm": 0.8185147047042847, "learning_rate": 1.4369577068487417e-06, "loss": 2.9156, "step": 323230 }, { "epoch": 0.0024064, "grad_norm": 0.7705767154693604, "learning_rate": 1.4365418310314183e-06, "loss": 2.674, "step": 323240 }, { "epoch": 0.002432, "grad_norm": 0.8885040879249573, "learning_rate": 1.4361260107458464e-06, "loss": 2.7555, "step": 323250 }, { "epoch": 0.0024576, "grad_norm": 0.8483799695968628, "learning_rate": 1.4357102459947225e-06, "loss": 2.8936, "step": 323260 }, { "epoch": 0.0024832, "grad_norm": 0.9021895527839661, "learning_rate": 1.4352945367807403e-06, "loss": 2.7598, "step": 323270 }, { "epoch": 0.0025088, "grad_norm": 0.9369493722915649, "learning_rate": 1.4348788831065963e-06, "loss": 2.686, "step": 323280 }, { "epoch": 0.0025344, "grad_norm": 0.8963421583175659, "learning_rate": 1.4344632849749862e-06, "loss": 2.9725, "step": 323290 }, { "epoch": 0.00256, "grad_norm": 0.9114791750907898, "learning_rate": 1.4340477423886047e-06, "loss": 2.5981, "step": 323300 }, { "epoch": 0.0025856, "grad_norm": 0.8909094929695129, "learning_rate": 1.433632255350148e-06, "loss": 2.7545, "step": 323310 }, { "epoch": 0.0026112, "grad_norm": 0.8349252939224243, "learning_rate": 1.433216823862309e-06, "loss": 2.7865, "step": 323320 }, { "epoch": 0.0026368, "grad_norm": 0.9393442869186401, "learning_rate": 1.4328014479277852e-06, "loss": 2.8764, "step": 323330 }, { "epoch": 0.0026624, "grad_norm": 0.9665202498435974, "learning_rate": 1.4323861275492634e-06, "loss": 2.6874, "step": 323340 }, { "epoch": 0.002688, "grad_norm": 0.8529030084609985, "learning_rate": 1.4319708627294415e-06, "loss": 2.8581, "step": 323350 }, { "epoch": 0.0027136, "grad_norm": 0.8599699139595032, "learning_rate": 1.4315556534710117e-06, "loss": 2.7419, "step": 323360 }, { "epoch": 0.0027392, "grad_norm": 0.8554162979125977, "learning_rate": 1.4311404997766653e-06, "loss": 2.884, "step": 323370 }, { "epoch": 0.0027648, "grad_norm": 0.8407399654388428, "learning_rate": 1.4307254016490945e-06, "loss": 2.6752, "step": 323380 }, { "epoch": 0.0027904, "grad_norm": 0.8051616549491882, "learning_rate": 1.430310359090995e-06, "loss": 2.7961, "step": 323390 }, { "epoch": 0.002816, "grad_norm": 0.9310437440872192, "learning_rate": 1.4298953721050545e-06, "loss": 3.0128, "step": 323400 }, { "epoch": 0.0028416, "grad_norm": 1.0458537340164185, "learning_rate": 1.4294804406939644e-06, "loss": 2.8655, "step": 323410 }, { "epoch": 0.0028672, "grad_norm": 0.9171623587608337, "learning_rate": 1.4290655648604157e-06, "loss": 2.8407, "step": 323420 }, { "epoch": 0.0028928, "grad_norm": 0.8102253675460815, "learning_rate": 1.4286507446070986e-06, "loss": 2.7126, "step": 323430 }, { "epoch": 0.0029184, "grad_norm": 0.8237887024879456, "learning_rate": 1.428235979936704e-06, "loss": 2.5599, "step": 323440 }, { "epoch": 0.002944, "grad_norm": 0.8268877267837524, "learning_rate": 1.42782127085192e-06, "loss": 2.882, "step": 323450 }, { "epoch": 0.0029696, "grad_norm": 0.7918088436126709, "learning_rate": 1.42740661735544e-06, "loss": 2.901, "step": 323460 }, { "epoch": 0.0029952, "grad_norm": 0.938911497592926, "learning_rate": 1.4269920194499475e-06, "loss": 2.9, "step": 323470 }, { "epoch": 0.0030208, "grad_norm": 0.8516919612884521, "learning_rate": 1.4265774771381324e-06, "loss": 2.9207, "step": 323480 }, { "epoch": 0.0030464, "grad_norm": 0.8440628051757812, "learning_rate": 1.426162990422685e-06, "loss": 2.7543, "step": 323490 }, { "epoch": 0.003072, "grad_norm": 0.8720751404762268, "learning_rate": 1.4257485593062914e-06, "loss": 2.6606, "step": 323500 }, { "epoch": 0.0030976, "grad_norm": 0.8243182301521301, "learning_rate": 1.4253341837916401e-06, "loss": 2.7861, "step": 323510 }, { "epoch": 0.0031232, "grad_norm": 0.8009012937545776, "learning_rate": 1.4249198638814166e-06, "loss": 2.5473, "step": 323520 }, { "epoch": 0.0031488, "grad_norm": 0.8601261973381042, "learning_rate": 1.424505599578312e-06, "loss": 2.7969, "step": 323530 }, { "epoch": 0.0031744, "grad_norm": 0.9366481900215149, "learning_rate": 1.4240913908850073e-06, "loss": 2.7292, "step": 323540 }, { "epoch": 0.0032, "grad_norm": 0.7710068225860596, "learning_rate": 1.4236772378041885e-06, "loss": 2.8948, "step": 323550 }, { "epoch": 0.0032256, "grad_norm": 0.8126192688941956, "learning_rate": 1.4232631403385454e-06, "loss": 2.6744, "step": 323560 }, { "epoch": 0.0032512, "grad_norm": 0.801437258720398, "learning_rate": 1.4228490984907618e-06, "loss": 2.7263, "step": 323570 }, { "epoch": 0.0032768, "grad_norm": 0.9079595804214478, "learning_rate": 1.4224351122635215e-06, "loss": 2.7863, "step": 323580 }, { "epoch": 0.0033024, "grad_norm": 1.1502981185913086, "learning_rate": 1.4220211816595132e-06, "loss": 2.8355, "step": 323590 }, { "epoch": 0.003328, "grad_norm": 0.8263130784034729, "learning_rate": 1.421607306681414e-06, "loss": 2.855, "step": 323600 }, { "epoch": 0.0033536, "grad_norm": 0.7535345554351807, "learning_rate": 1.4211934873319122e-06, "loss": 2.644, "step": 323610 }, { "epoch": 0.0033792, "grad_norm": 0.8180426359176636, "learning_rate": 1.420779723613691e-06, "loss": 2.8283, "step": 323620 }, { "epoch": 0.0034048, "grad_norm": 0.7554221749305725, "learning_rate": 1.420366015529433e-06, "loss": 2.8836, "step": 323630 }, { "epoch": 0.0034304, "grad_norm": 0.7986737489700317, "learning_rate": 1.4199523630818213e-06, "loss": 2.7704, "step": 323640 }, { "epoch": 0.003456, "grad_norm": 0.8165021538734436, "learning_rate": 1.4195387662735382e-06, "loss": 2.6804, "step": 323650 }, { "epoch": 0.0034816, "grad_norm": 0.8898965120315552, "learning_rate": 1.4191252251072685e-06, "loss": 2.6935, "step": 323660 }, { "epoch": 0.0035072, "grad_norm": 0.9167472720146179, "learning_rate": 1.4187117395856885e-06, "loss": 2.7171, "step": 323670 }, { "epoch": 0.0035328, "grad_norm": 0.79378741979599, "learning_rate": 1.4182983097114822e-06, "loss": 2.8034, "step": 323680 }, { "epoch": 0.0035584, "grad_norm": 0.8596922755241394, "learning_rate": 1.4178849354873303e-06, "loss": 2.9496, "step": 323690 }, { "epoch": 0.003584, "grad_norm": 0.817466139793396, "learning_rate": 1.4174716169159142e-06, "loss": 2.8622, "step": 323700 }, { "epoch": 0.0036096, "grad_norm": 0.8103392720222473, "learning_rate": 1.4170583539999139e-06, "loss": 2.9768, "step": 323710 }, { "epoch": 0.0036352, "grad_norm": 0.7946828007698059, "learning_rate": 1.4166451467420085e-06, "loss": 3.0304, "step": 323720 }, { "epoch": 0.0036608, "grad_norm": 0.8366990089416504, "learning_rate": 1.4162319951448789e-06, "loss": 2.9165, "step": 323730 }, { "epoch": 0.0036864, "grad_norm": 1.0396394729614258, "learning_rate": 1.4158188992112031e-06, "loss": 2.8321, "step": 323740 }, { "epoch": 0.003712, "grad_norm": 0.9675722718238831, "learning_rate": 1.4154058589436602e-06, "loss": 2.7976, "step": 323750 }, { "epoch": 0.0037376, "grad_norm": 0.8253816366195679, "learning_rate": 1.4149928743449292e-06, "loss": 2.7614, "step": 323760 }, { "epoch": 0.0037632, "grad_norm": 0.9390541315078735, "learning_rate": 1.4145799454176867e-06, "loss": 2.8313, "step": 323770 }, { "epoch": 0.0037888, "grad_norm": 0.8432518243789673, "learning_rate": 1.4141670721646127e-06, "loss": 2.9242, "step": 323780 }, { "epoch": 0.0038144, "grad_norm": 0.8720490336418152, "learning_rate": 1.4137542545883853e-06, "loss": 2.7471, "step": 323790 }, { "epoch": 0.00384, "grad_norm": 0.9119452834129333, "learning_rate": 1.4133414926916778e-06, "loss": 2.9567, "step": 323800 }, { "epoch": 0.0038656, "grad_norm": 0.9718103408813477, "learning_rate": 1.4129287864771678e-06, "loss": 2.6468, "step": 323810 }, { "epoch": 0.0038912, "grad_norm": 0.8108771443367004, "learning_rate": 1.4125161359475326e-06, "loss": 2.7987, "step": 323820 }, { "epoch": 0.0039168, "grad_norm": 0.7982513904571533, "learning_rate": 1.4121035411054484e-06, "loss": 2.926, "step": 323830 }, { "epoch": 0.0039424, "grad_norm": 0.802226185798645, "learning_rate": 1.4116910019535902e-06, "loss": 2.8336, "step": 323840 }, { "epoch": 0.003968, "grad_norm": 0.95619136095047, "learning_rate": 1.411278518494633e-06, "loss": 2.8679, "step": 323850 }, { "epoch": 0.0039936, "grad_norm": 0.9406300783157349, "learning_rate": 1.4108660907312543e-06, "loss": 2.888, "step": 323860 }, { "epoch": 0.0040192, "grad_norm": 0.8062988519668579, "learning_rate": 1.4104537186661248e-06, "loss": 2.8157, "step": 323870 }, { "epoch": 0.0040448, "grad_norm": 0.8463011980056763, "learning_rate": 1.4100414023019193e-06, "loss": 2.7748, "step": 323880 }, { "epoch": 0.0040704, "grad_norm": 0.8706021904945374, "learning_rate": 1.4096291416413132e-06, "loss": 2.8583, "step": 323890 }, { "epoch": 0.004096, "grad_norm": 0.8546932935714722, "learning_rate": 1.4092169366869757e-06, "loss": 2.6324, "step": 323900 }, { "epoch": 0.0041216, "grad_norm": 0.8543886542320251, "learning_rate": 1.4088047874415888e-06, "loss": 2.7163, "step": 323910 }, { "epoch": 0.0041472, "grad_norm": 0.8718075156211853, "learning_rate": 1.4083926939078162e-06, "loss": 2.9511, "step": 323920 }, { "epoch": 0.0041728, "grad_norm": 0.8853990435600281, "learning_rate": 1.4079806560883348e-06, "loss": 2.9044, "step": 323930 }, { "epoch": 0.0041984, "grad_norm": 0.8114095330238342, "learning_rate": 1.407568673985814e-06, "loss": 2.7562, "step": 323940 }, { "epoch": 0.004224, "grad_norm": 0.8380603194236755, "learning_rate": 1.407156747602928e-06, "loss": 2.6312, "step": 323950 }, { "epoch": 0.0042496, "grad_norm": 0.8580474257469177, "learning_rate": 1.4067448769423464e-06, "loss": 2.8856, "step": 323960 }, { "epoch": 0.0042752, "grad_norm": 0.8383574485778809, "learning_rate": 1.4063330620067405e-06, "loss": 2.8738, "step": 323970 }, { "epoch": 0.0043008, "grad_norm": 0.8981095552444458, "learning_rate": 1.4059213027987806e-06, "loss": 2.8534, "step": 323980 }, { "epoch": 0.0043264, "grad_norm": 0.8108329772949219, "learning_rate": 1.4055095993211387e-06, "loss": 2.832, "step": 323990 }, { "epoch": 0.004352, "grad_norm": 0.8863660097122192, "learning_rate": 1.4050979515764817e-06, "loss": 2.9169, "step": 324000 }, { "epoch": 0.0043776, "grad_norm": 0.8410173654556274, "learning_rate": 1.4046863595674797e-06, "loss": 2.8837, "step": 324010 }, { "epoch": 0.0044032, "grad_norm": 0.8488134741783142, "learning_rate": 1.4042748232968028e-06, "loss": 2.825, "step": 324020 }, { "epoch": 0.0044288, "grad_norm": 0.893744170665741, "learning_rate": 1.4038633427671188e-06, "loss": 2.8586, "step": 324030 }, { "epoch": 0.0044544, "grad_norm": 0.8626787066459656, "learning_rate": 1.4034519179810958e-06, "loss": 2.5899, "step": 324040 }, { "epoch": 0.00448, "grad_norm": 0.8178374767303467, "learning_rate": 1.4030405489414035e-06, "loss": 2.8179, "step": 324050 }, { "epoch": 0.0045056, "grad_norm": 0.8814774751663208, "learning_rate": 1.4026292356507098e-06, "loss": 2.7432, "step": 324060 }, { "epoch": 0.0045312, "grad_norm": 1.017553448677063, "learning_rate": 1.4022179781116774e-06, "loss": 2.8746, "step": 324070 }, { "epoch": 0.0045568, "grad_norm": 0.9177351593971252, "learning_rate": 1.4018067763269773e-06, "loss": 2.9171, "step": 324080 }, { "epoch": 0.0045824, "grad_norm": 0.9644754528999329, "learning_rate": 1.4013956302992759e-06, "loss": 2.9866, "step": 324090 }, { "epoch": 0.004608, "grad_norm": 0.866961658000946, "learning_rate": 1.4009845400312382e-06, "loss": 2.9329, "step": 324100 }, { "epoch": 0.0046336, "grad_norm": 0.8904508352279663, "learning_rate": 1.4005735055255332e-06, "loss": 2.9137, "step": 324110 }, { "epoch": 0.0046592, "grad_norm": 0.8072302341461182, "learning_rate": 1.4001625267848207e-06, "loss": 2.9554, "step": 324120 }, { "epoch": 0.0046848, "grad_norm": 0.9572417736053467, "learning_rate": 1.3997516038117687e-06, "loss": 2.8571, "step": 324130 }, { "epoch": 0.0047104, "grad_norm": 0.7862038612365723, "learning_rate": 1.3993407366090417e-06, "loss": 2.7141, "step": 324140 }, { "epoch": 0.004736, "grad_norm": 0.8916844129562378, "learning_rate": 1.3989299251793053e-06, "loss": 2.8042, "step": 324150 }, { "epoch": 0.0047616, "grad_norm": 0.9160369634628296, "learning_rate": 1.398519169525221e-06, "loss": 2.878, "step": 324160 }, { "epoch": 0.0047872, "grad_norm": 0.9014955163002014, "learning_rate": 1.3981084696494552e-06, "loss": 3.1112, "step": 324170 }, { "epoch": 0.0048128, "grad_norm": 0.8232826590538025, "learning_rate": 1.3976978255546692e-06, "loss": 2.6845, "step": 324180 }, { "epoch": 0.0048384, "grad_norm": 0.8260430097579956, "learning_rate": 1.3972872372435287e-06, "loss": 2.6672, "step": 324190 }, { "epoch": 0.004864, "grad_norm": 0.8572366237640381, "learning_rate": 1.3968767047186925e-06, "loss": 2.7503, "step": 324200 }, { "epoch": 0.0048896, "grad_norm": 0.7981343865394592, "learning_rate": 1.3964662279828234e-06, "loss": 2.8202, "step": 324210 }, { "epoch": 0.0049152, "grad_norm": 0.8190656900405884, "learning_rate": 1.3960558070385855e-06, "loss": 2.8772, "step": 324220 }, { "epoch": 0.0049408, "grad_norm": 0.8866692185401917, "learning_rate": 1.3956454418886377e-06, "loss": 2.849, "step": 324230 }, { "epoch": 0.0049664, "grad_norm": 0.9254968762397766, "learning_rate": 1.395235132535644e-06, "loss": 2.9472, "step": 324240 }, { "epoch": 0.004992, "grad_norm": 0.8722209930419922, "learning_rate": 1.3948248789822626e-06, "loss": 2.8114, "step": 324250 }, { "epoch": 0.0050176, "grad_norm": 0.8277920484542847, "learning_rate": 1.3944146812311554e-06, "loss": 2.7327, "step": 324260 }, { "epoch": 0.0050432, "grad_norm": 1.0418015718460083, "learning_rate": 1.394004539284981e-06, "loss": 2.703, "step": 324270 }, { "epoch": 0.0050688, "grad_norm": 0.9511978030204773, "learning_rate": 1.3935944531464007e-06, "loss": 2.6642, "step": 324280 }, { "epoch": 0.0050944, "grad_norm": 1.0348687171936035, "learning_rate": 1.3931844228180723e-06, "loss": 2.8824, "step": 324290 }, { "epoch": 0.00512, "grad_norm": 0.8136072158813477, "learning_rate": 1.3927744483026572e-06, "loss": 2.7185, "step": 324300 }, { "epoch": 0.0051456, "grad_norm": 0.8397671580314636, "learning_rate": 1.3923645296028132e-06, "loss": 2.8325, "step": 324310 }, { "epoch": 0.0051712, "grad_norm": 0.7561966180801392, "learning_rate": 1.3919546667211959e-06, "loss": 2.476, "step": 324320 }, { "epoch": 0.0051968, "grad_norm": 0.8542883992195129, "learning_rate": 1.3915448596604641e-06, "loss": 2.7593, "step": 324330 }, { "epoch": 0.0052224, "grad_norm": 0.8200961351394653, "learning_rate": 1.3911351084232761e-06, "loss": 2.9152, "step": 324340 }, { "epoch": 0.005248, "grad_norm": 0.792849063873291, "learning_rate": 1.3907254130122894e-06, "loss": 2.8719, "step": 324350 }, { "epoch": 0.0052736, "grad_norm": 0.7954121828079224, "learning_rate": 1.3903157734301597e-06, "loss": 2.6857, "step": 324360 }, { "epoch": 0.0052992, "grad_norm": 0.850476086139679, "learning_rate": 1.3899061896795452e-06, "loss": 2.9746, "step": 324370 }, { "epoch": 0.0053248, "grad_norm": 0.7848626375198364, "learning_rate": 1.3894966617631e-06, "loss": 2.7855, "step": 324380 }, { "epoch": 0.0053504, "grad_norm": 0.8349761366844177, "learning_rate": 1.3890871896834834e-06, "loss": 2.8209, "step": 324390 }, { "epoch": 0.005376, "grad_norm": 0.979160726070404, "learning_rate": 1.3886777734433443e-06, "loss": 2.6945, "step": 324400 }, { "epoch": 0.0054016, "grad_norm": 0.7936136722564697, "learning_rate": 1.3882684130453427e-06, "loss": 2.7468, "step": 324410 }, { "epoch": 0.0054272, "grad_norm": 0.802020251750946, "learning_rate": 1.3878591084921289e-06, "loss": 2.9285, "step": 324420 }, { "epoch": 0.0054528, "grad_norm": 0.8623080253601074, "learning_rate": 1.3874498597863616e-06, "loss": 2.6952, "step": 324430 }, { "epoch": 0.0054784, "grad_norm": 0.7782000303268433, "learning_rate": 1.3870406669306958e-06, "loss": 2.7775, "step": 324440 }, { "epoch": 0.005504, "grad_norm": 0.8438670039176941, "learning_rate": 1.3866315299277799e-06, "loss": 2.9042, "step": 324450 }, { "epoch": 0.0055296, "grad_norm": 0.9489086270332336, "learning_rate": 1.38622244878027e-06, "loss": 2.7911, "step": 324460 }, { "epoch": 0.0055552, "grad_norm": 0.87049400806427, "learning_rate": 1.385813423490817e-06, "loss": 2.9215, "step": 324470 }, { "epoch": 0.0055808, "grad_norm": 0.8552728295326233, "learning_rate": 1.3854044540620758e-06, "loss": 2.7883, "step": 324480 }, { "epoch": 0.0056064, "grad_norm": 0.8657987117767334, "learning_rate": 1.3849955404966963e-06, "loss": 2.6618, "step": 324490 }, { "epoch": 0.005632, "grad_norm": 0.8227230310440063, "learning_rate": 1.384586682797332e-06, "loss": 2.9632, "step": 324500 }, { "epoch": 0.0056576, "grad_norm": 0.7600973844528198, "learning_rate": 1.384177880966635e-06, "loss": 2.9349, "step": 324510 }, { "epoch": 0.0056832, "grad_norm": 0.8143467903137207, "learning_rate": 1.3837691350072524e-06, "loss": 2.8369, "step": 324520 }, { "epoch": 0.0057088, "grad_norm": 0.8652995824813843, "learning_rate": 1.3833604449218374e-06, "loss": 2.667, "step": 324530 }, { "epoch": 0.0057344, "grad_norm": 0.8483012318611145, "learning_rate": 1.3829518107130391e-06, "loss": 2.5595, "step": 324540 }, { "epoch": 0.00576, "grad_norm": 0.8283141255378723, "learning_rate": 1.3825432323835086e-06, "loss": 2.8269, "step": 324550 }, { "epoch": 0.0057856, "grad_norm": 0.798669159412384, "learning_rate": 1.3821347099358961e-06, "loss": 2.811, "step": 324560 }, { "epoch": 0.0058112, "grad_norm": 0.9199190139770508, "learning_rate": 1.3817262433728485e-06, "loss": 2.5967, "step": 324570 }, { "epoch": 0.0058368, "grad_norm": 0.8633939027786255, "learning_rate": 1.3813178326970155e-06, "loss": 2.6769, "step": 324580 }, { "epoch": 0.0058624, "grad_norm": 0.8556143641471863, "learning_rate": 1.380909477911049e-06, "loss": 2.7033, "step": 324590 }, { "epoch": 0.005888, "grad_norm": 0.9632239937782288, "learning_rate": 1.380501179017588e-06, "loss": 2.5635, "step": 324600 }, { "epoch": 0.0059136, "grad_norm": 0.797455906867981, "learning_rate": 1.3800929360192894e-06, "loss": 2.9375, "step": 324610 }, { "epoch": 0.0059392, "grad_norm": 0.8811426758766174, "learning_rate": 1.379684748918797e-06, "loss": 2.9495, "step": 324620 }, { "epoch": 0.0059648, "grad_norm": 0.9875431656837463, "learning_rate": 1.3792766177187577e-06, "loss": 2.9751, "step": 324630 }, { "epoch": 0.0059904, "grad_norm": 0.7706343531608582, "learning_rate": 1.3788685424218206e-06, "loss": 2.8142, "step": 324640 }, { "epoch": 0.006016, "grad_norm": 0.7444301247596741, "learning_rate": 1.3784605230306281e-06, "loss": 2.6677, "step": 324650 }, { "epoch": 0.0060416, "grad_norm": 0.8423840999603271, "learning_rate": 1.3780525595478278e-06, "loss": 2.6198, "step": 324660 }, { "epoch": 0.0060672, "grad_norm": 0.8472949862480164, "learning_rate": 1.3776446519760645e-06, "loss": 2.7351, "step": 324670 }, { "epoch": 0.0060928, "grad_norm": 0.7953062057495117, "learning_rate": 1.3772368003179849e-06, "loss": 2.6519, "step": 324680 }, { "epoch": 0.0061184, "grad_norm": 1.104465365409851, "learning_rate": 1.3768290045762323e-06, "loss": 2.5409, "step": 324690 }, { "epoch": 0.006144, "grad_norm": 0.8101106882095337, "learning_rate": 1.3764212647534524e-06, "loss": 2.5805, "step": 324700 }, { "epoch": 0.0061696, "grad_norm": 0.8365070223808289, "learning_rate": 1.37601358085229e-06, "loss": 2.9462, "step": 324710 }, { "epoch": 0.0061952, "grad_norm": 0.8837631940841675, "learning_rate": 1.3756059528753863e-06, "loss": 2.753, "step": 324720 }, { "epoch": 0.0062208, "grad_norm": 0.837248682975769, "learning_rate": 1.3751983808253856e-06, "loss": 2.9116, "step": 324730 }, { "epoch": 0.0062464, "grad_norm": 1.04175865650177, "learning_rate": 1.3747908647049314e-06, "loss": 2.7488, "step": 324740 }, { "epoch": 0.006272, "grad_norm": 0.84650719165802, "learning_rate": 1.374383404516666e-06, "loss": 3.6206, "step": 324750 }, { "epoch": 0.0062976, "grad_norm": 0.8212724328041077, "learning_rate": 1.3739760002632319e-06, "loss": 2.7682, "step": 324760 }, { "epoch": 0.0063232, "grad_norm": 0.8270899057388306, "learning_rate": 1.3735686519472702e-06, "loss": 2.9709, "step": 324770 }, { "epoch": 0.0063488, "grad_norm": 0.9028719067573547, "learning_rate": 1.3731613595714245e-06, "loss": 2.7023, "step": 324780 }, { "epoch": 0.0063744, "grad_norm": 0.8937983512878418, "learning_rate": 1.3727541231383335e-06, "loss": 2.9642, "step": 324790 }, { "epoch": 0.0064, "grad_norm": 0.9106534719467163, "learning_rate": 1.3723469426506397e-06, "loss": 2.6483, "step": 324800 }, { "epoch": 0.0064256, "grad_norm": 0.8575599789619446, "learning_rate": 1.3719398181109833e-06, "loss": 2.7619, "step": 324810 }, { "epoch": 0.0064512, "grad_norm": 0.8583718538284302, "learning_rate": 1.3715327495220031e-06, "loss": 2.7835, "step": 324820 }, { "epoch": 0.0064768, "grad_norm": 0.8890973925590515, "learning_rate": 1.3711257368863406e-06, "loss": 2.8238, "step": 324830 }, { "epoch": 0.0065024, "grad_norm": 0.9043317437171936, "learning_rate": 1.3707187802066369e-06, "loss": 2.7084, "step": 324840 }, { "epoch": 0.006528, "grad_norm": 0.851713240146637, "learning_rate": 1.3703118794855252e-06, "loss": 2.8971, "step": 324850 }, { "epoch": 0.0065536, "grad_norm": 0.8222434520721436, "learning_rate": 1.369905034725648e-06, "loss": 2.724, "step": 324860 }, { "epoch": 0.0065792, "grad_norm": 0.7922113537788391, "learning_rate": 1.3694982459296435e-06, "loss": 2.7475, "step": 324870 }, { "epoch": 0.0066048, "grad_norm": 0.833876371383667, "learning_rate": 1.3690915131001491e-06, "loss": 2.7483, "step": 324880 }, { "epoch": 0.0066304, "grad_norm": 0.937047004699707, "learning_rate": 1.368684836239802e-06, "loss": 2.9459, "step": 324890 }, { "epoch": 0.006656, "grad_norm": 0.8704445362091064, "learning_rate": 1.3682782153512397e-06, "loss": 2.8606, "step": 324900 }, { "epoch": 0.0066816, "grad_norm": 0.7820241451263428, "learning_rate": 1.3678716504371015e-06, "loss": 3.0796, "step": 324910 }, { "epoch": 0.0067072, "grad_norm": 0.8052331209182739, "learning_rate": 1.3674651415000195e-06, "loss": 2.6424, "step": 324920 }, { "epoch": 0.0067328, "grad_norm": 0.8177790641784668, "learning_rate": 1.3670586885426318e-06, "loss": 2.9292, "step": 324930 }, { "epoch": 0.0067584, "grad_norm": 0.7729071974754333, "learning_rate": 1.3666522915675718e-06, "loss": 2.8931, "step": 324940 }, { "epoch": 0.006784, "grad_norm": 1.411969542503357, "learning_rate": 1.3662459505774795e-06, "loss": 2.8189, "step": 324950 }, { "epoch": 0.0068096, "grad_norm": 0.813027024269104, "learning_rate": 1.3658396655749873e-06, "loss": 2.7983, "step": 324960 }, { "epoch": 0.0068352, "grad_norm": 0.8439133167266846, "learning_rate": 1.3654334365627332e-06, "loss": 2.7535, "step": 324970 }, { "epoch": 0.0068608, "grad_norm": 0.792928159236908, "learning_rate": 1.365027263543346e-06, "loss": 2.6751, "step": 324980 }, { "epoch": 0.0068864, "grad_norm": 0.8981227874755859, "learning_rate": 1.3646211465194615e-06, "loss": 2.782, "step": 324990 }, { "epoch": 0.006912, "grad_norm": 0.9066131711006165, "learning_rate": 1.364215085493714e-06, "loss": 2.644, "step": 325000 }, { "epoch": 0.0069376, "grad_norm": 0.7946786880493164, "learning_rate": 1.3638090804687365e-06, "loss": 2.7789, "step": 325010 }, { "epoch": 0.0069632, "grad_norm": 1.1277655363082886, "learning_rate": 1.363403131447163e-06, "loss": 2.8171, "step": 325020 }, { "epoch": 0.0069888, "grad_norm": 0.8314410448074341, "learning_rate": 1.3629972384316237e-06, "loss": 2.7226, "step": 325030 }, { "epoch": 0.0070144, "grad_norm": 0.9429675936698914, "learning_rate": 1.3625914014247543e-06, "loss": 2.7584, "step": 325040 }, { "epoch": 0.00704, "grad_norm": 0.8144358396530151, "learning_rate": 1.3621856204291816e-06, "loss": 2.8046, "step": 325050 }, { "epoch": 0.0070656, "grad_norm": 0.8959235548973083, "learning_rate": 1.3617798954475403e-06, "loss": 3.0066, "step": 325060 }, { "epoch": 0.0070912, "grad_norm": 0.8537620902061462, "learning_rate": 1.3613742264824603e-06, "loss": 2.9027, "step": 325070 }, { "epoch": 0.0071168, "grad_norm": 0.7781017422676086, "learning_rate": 1.3609686135365729e-06, "loss": 2.6835, "step": 325080 }, { "epoch": 0.0071424, "grad_norm": 0.8156406283378601, "learning_rate": 1.3605630566125073e-06, "loss": 2.9121, "step": 325090 }, { "epoch": 0.007168, "grad_norm": 0.8838884234428406, "learning_rate": 1.3601575557128943e-06, "loss": 2.9857, "step": 325100 }, { "epoch": 0.0071936, "grad_norm": 0.8439202904701233, "learning_rate": 1.3597521108403655e-06, "loss": 2.6184, "step": 325110 }, { "epoch": 0.0072192, "grad_norm": 0.8828737139701843, "learning_rate": 1.3593467219975442e-06, "loss": 2.7778, "step": 325120 }, { "epoch": 0.0072448, "grad_norm": 0.8213648200035095, "learning_rate": 1.358941389187065e-06, "loss": 2.8014, "step": 325130 }, { "epoch": 0.0072704, "grad_norm": 0.7352481484413147, "learning_rate": 1.3585361124115537e-06, "loss": 2.8534, "step": 325140 }, { "epoch": 0.007296, "grad_norm": 0.8698357343673706, "learning_rate": 1.3581308916736401e-06, "loss": 2.7853, "step": 325150 }, { "epoch": 0.0073216, "grad_norm": 0.7982431650161743, "learning_rate": 1.3577257269759504e-06, "loss": 2.7337, "step": 325160 }, { "epoch": 0.0073472, "grad_norm": 0.7879088521003723, "learning_rate": 1.3573206183211152e-06, "loss": 2.6889, "step": 325170 }, { "epoch": 0.0073728, "grad_norm": 0.8263987898826599, "learning_rate": 1.3569155657117562e-06, "loss": 2.7399, "step": 325180 }, { "epoch": 0.0073984, "grad_norm": 1.0666252374649048, "learning_rate": 1.356510569150502e-06, "loss": 2.7639, "step": 325190 }, { "epoch": 0.007424, "grad_norm": 0.8370980024337769, "learning_rate": 1.356105628639981e-06, "loss": 2.7125, "step": 325200 }, { "epoch": 0.0074496, "grad_norm": 0.7843188643455505, "learning_rate": 1.3557007441828163e-06, "loss": 2.9094, "step": 325210 }, { "epoch": 0.0074752, "grad_norm": 0.9880906939506531, "learning_rate": 1.355295915781636e-06, "loss": 3.0957, "step": 325220 }, { "epoch": 0.0075008, "grad_norm": 0.9984161257743835, "learning_rate": 1.3548911434390632e-06, "loss": 2.7365, "step": 325230 }, { "epoch": 0.0075264, "grad_norm": 0.905357301235199, "learning_rate": 1.3544864271577274e-06, "loss": 2.8528, "step": 325240 }, { "epoch": 0.007552, "grad_norm": 0.866849422454834, "learning_rate": 1.3540817669402462e-06, "loss": 2.9019, "step": 325250 }, { "epoch": 0.0075776, "grad_norm": 0.7902205586433411, "learning_rate": 1.3536771627892465e-06, "loss": 2.8283, "step": 325260 }, { "epoch": 0.0076032, "grad_norm": 0.8917824029922485, "learning_rate": 1.3532726147073538e-06, "loss": 2.818, "step": 325270 }, { "epoch": 0.0076288, "grad_norm": 0.7987084984779358, "learning_rate": 1.3528681226971896e-06, "loss": 2.7912, "step": 325280 }, { "epoch": 0.0076544, "grad_norm": 0.792971670627594, "learning_rate": 1.352463686761375e-06, "loss": 2.6681, "step": 325290 }, { "epoch": 0.00768, "grad_norm": 0.7546751499176025, "learning_rate": 1.352059306902539e-06, "loss": 2.8262, "step": 325300 }, { "epoch": 0.0077056, "grad_norm": 0.9206273555755615, "learning_rate": 1.3516549831232983e-06, "loss": 2.7948, "step": 325310 }, { "epoch": 0.0077312, "grad_norm": 0.8310527801513672, "learning_rate": 1.3512507154262777e-06, "loss": 2.7782, "step": 325320 }, { "epoch": 0.0077568, "grad_norm": 0.9102990627288818, "learning_rate": 1.3508465038140961e-06, "loss": 2.8066, "step": 325330 }, { "epoch": 0.0077824, "grad_norm": 0.9196507334709167, "learning_rate": 1.3504423482893769e-06, "loss": 2.9384, "step": 325340 }, { "epoch": 0.007808, "grad_norm": 0.9177842140197754, "learning_rate": 1.3500382488547403e-06, "loss": 2.5041, "step": 325350 }, { "epoch": 0.0078336, "grad_norm": 0.8754046559333801, "learning_rate": 1.3496342055128076e-06, "loss": 2.8612, "step": 325360 }, { "epoch": 0.0078592, "grad_norm": 0.8339248299598694, "learning_rate": 1.3492302182661988e-06, "loss": 2.5505, "step": 325370 }, { "epoch": 0.0078848, "grad_norm": 0.8130718469619751, "learning_rate": 1.3488262871175318e-06, "loss": 2.7825, "step": 325380 }, { "epoch": 0.0079104, "grad_norm": 0.8533381819725037, "learning_rate": 1.3484224120694267e-06, "loss": 2.9999, "step": 325390 }, { "epoch": 0.007936, "grad_norm": 0.8201419711112976, "learning_rate": 1.3480185931245027e-06, "loss": 2.804, "step": 325400 }, { "epoch": 0.0079616, "grad_norm": 0.9159952402114868, "learning_rate": 1.3476148302853786e-06, "loss": 2.6627, "step": 325410 }, { "epoch": 0.0079872, "grad_norm": 0.8627892136573792, "learning_rate": 1.3472111235546736e-06, "loss": 2.6831, "step": 325420 }, { "epoch": 0.0080128, "grad_norm": 0.911316990852356, "learning_rate": 1.3468074729350045e-06, "loss": 2.7949, "step": 325430 }, { "epoch": 0.0080384, "grad_norm": 0.839026153087616, "learning_rate": 1.3464038784289913e-06, "loss": 2.7458, "step": 325440 }, { "epoch": 0.008064, "grad_norm": 0.8639446496963501, "learning_rate": 1.3460003400392473e-06, "loss": 2.7256, "step": 325450 }, { "epoch": 0.0080896, "grad_norm": 0.8214468359947205, "learning_rate": 1.3455968577683908e-06, "loss": 2.7711, "step": 325460 }, { "epoch": 0.0081152, "grad_norm": 0.8875526785850525, "learning_rate": 1.3451934316190363e-06, "loss": 2.7383, "step": 325470 }, { "epoch": 0.0081408, "grad_norm": 0.7910845875740051, "learning_rate": 1.3447900615938036e-06, "loss": 2.8499, "step": 325480 }, { "epoch": 0.0081664, "grad_norm": 0.9108045697212219, "learning_rate": 1.3443867476953087e-06, "loss": 2.8424, "step": 325490 }, { "epoch": 0.008192, "grad_norm": 0.8331871628761292, "learning_rate": 1.343983489926166e-06, "loss": 2.5301, "step": 325500 }, { "epoch": 0.0082176, "grad_norm": 0.8568456768989563, "learning_rate": 1.3435802882889892e-06, "loss": 2.7577, "step": 325510 }, { "epoch": 0.0082432, "grad_norm": 0.8914285898208618, "learning_rate": 1.3431771427863926e-06, "loss": 2.796, "step": 325520 }, { "epoch": 0.0082688, "grad_norm": 1.0518759489059448, "learning_rate": 1.3427740534209922e-06, "loss": 2.9128, "step": 325530 }, { "epoch": 0.0082944, "grad_norm": 0.8394895195960999, "learning_rate": 1.342371020195401e-06, "loss": 2.8572, "step": 325540 }, { "epoch": 0.00832, "grad_norm": 0.8554434776306152, "learning_rate": 1.3419680431122328e-06, "loss": 2.7116, "step": 325550 }, { "epoch": 0.0083456, "grad_norm": 0.8320954442024231, "learning_rate": 1.3415651221741e-06, "loss": 2.889, "step": 325560 }, { "epoch": 0.0083712, "grad_norm": 0.7681460976600647, "learning_rate": 1.341162257383619e-06, "loss": 2.505, "step": 325570 }, { "epoch": 0.0083968, "grad_norm": 0.8316847681999207, "learning_rate": 1.3407594487433983e-06, "loss": 2.8509, "step": 325580 }, { "epoch": 0.0084224, "grad_norm": 1.1590052843093872, "learning_rate": 1.34035669625605e-06, "loss": 2.7452, "step": 325590 }, { "epoch": 0.008448, "grad_norm": 0.8839588165283203, "learning_rate": 1.3399539999241873e-06, "loss": 2.9595, "step": 325600 }, { "epoch": 0.0084736, "grad_norm": 0.8507599830627441, "learning_rate": 1.3395513597504218e-06, "loss": 2.864, "step": 325610 }, { "epoch": 0.0084992, "grad_norm": 0.8236241936683655, "learning_rate": 1.3391487757373634e-06, "loss": 3.0415, "step": 325620 }, { "epoch": 0.0085248, "grad_norm": 0.8530426025390625, "learning_rate": 1.3387462478876233e-06, "loss": 2.73, "step": 325630 }, { "epoch": 0.0085504, "grad_norm": 0.8240275979042053, "learning_rate": 1.3383437762038132e-06, "loss": 2.7979, "step": 325640 }, { "epoch": 0.008576, "grad_norm": 0.7665932774543762, "learning_rate": 1.3379413606885406e-06, "loss": 2.6658, "step": 325650 }, { "epoch": 0.0086016, "grad_norm": 0.8472528457641602, "learning_rate": 1.3375390013444168e-06, "loss": 2.9595, "step": 325660 }, { "epoch": 0.0086272, "grad_norm": 0.8448962569236755, "learning_rate": 1.3371366981740496e-06, "loss": 2.5775, "step": 325670 }, { "epoch": 0.0086528, "grad_norm": 0.9567878246307373, "learning_rate": 1.3367344511800495e-06, "loss": 2.6711, "step": 325680 }, { "epoch": 0.0086784, "grad_norm": 1.0101970434188843, "learning_rate": 1.336332260365023e-06, "loss": 2.7434, "step": 325690 }, { "epoch": 0.008704, "grad_norm": 0.8058547377586365, "learning_rate": 1.3359301257315827e-06, "loss": 2.7928, "step": 325700 }, { "epoch": 0.0087296, "grad_norm": 0.8132641315460205, "learning_rate": 1.3355280472823296e-06, "loss": 2.7837, "step": 325710 }, { "epoch": 0.0087552, "grad_norm": 0.8059866428375244, "learning_rate": 1.335126025019875e-06, "loss": 2.7392, "step": 325720 }, { "epoch": 0.0087808, "grad_norm": 0.9244834780693054, "learning_rate": 1.3347240589468246e-06, "loss": 2.8478, "step": 325730 }, { "epoch": 0.0088064, "grad_norm": 1.1324846744537354, "learning_rate": 1.3343221490657866e-06, "loss": 2.8125, "step": 325740 }, { "epoch": 0.008832, "grad_norm": 0.908682644367218, "learning_rate": 1.3339202953793662e-06, "loss": 2.8202, "step": 325750 }, { "epoch": 0.0088576, "grad_norm": 0.987186849117279, "learning_rate": 1.3335184978901695e-06, "loss": 2.7761, "step": 325760 }, { "epoch": 0.0088832, "grad_norm": 1.0267279148101807, "learning_rate": 1.333116756600804e-06, "loss": 3.0286, "step": 325770 }, { "epoch": 0.0089088, "grad_norm": 0.7755351066589355, "learning_rate": 1.3327150715138704e-06, "loss": 2.8556, "step": 325780 }, { "epoch": 0.0089344, "grad_norm": 1.0498456954956055, "learning_rate": 1.3323134426319772e-06, "loss": 2.7029, "step": 325790 }, { "epoch": 0.00896, "grad_norm": 0.7940844297409058, "learning_rate": 1.331911869957727e-06, "loss": 2.6497, "step": 325800 }, { "epoch": 2.56e-05, "grad_norm": 0.8362104296684265, "learning_rate": 1.3315103534937224e-06, "loss": 1.7126, "step": 325810 }, { "epoch": 5.12e-05, "grad_norm": 1.2133101224899292, "learning_rate": 1.3311088932425742e-06, "loss": 2.5156, "step": 325820 }, { "epoch": 7.68e-05, "grad_norm": 0.8743240833282471, "learning_rate": 1.3307074892068782e-06, "loss": 2.6285, "step": 325830 }, { "epoch": 0.0001024, "grad_norm": 0.7214495539665222, "learning_rate": 1.3303061413892404e-06, "loss": 2.5888, "step": 325840 }, { "epoch": 0.000128, "grad_norm": 0.8579339385032654, "learning_rate": 1.3299048497922628e-06, "loss": 2.8004, "step": 325850 }, { "epoch": 0.0001536, "grad_norm": 0.8180981278419495, "learning_rate": 1.3295036144185481e-06, "loss": 3.1101, "step": 325860 }, { "epoch": 0.0001792, "grad_norm": 0.8637613654136658, "learning_rate": 1.3291024352706983e-06, "loss": 2.5363, "step": 325870 }, { "epoch": 0.0002048, "grad_norm": 0.7548940777778625, "learning_rate": 1.328701312351315e-06, "loss": 2.5667, "step": 325880 }, { "epoch": 0.0002304, "grad_norm": 0.7878061532974243, "learning_rate": 1.328300245663e-06, "loss": 2.5524, "step": 325890 }, { "epoch": 0.000256, "grad_norm": 0.8327662944793701, "learning_rate": 1.327899235208353e-06, "loss": 2.5574, "step": 325900 }, { "epoch": 0.0002816, "grad_norm": 0.8446125984191895, "learning_rate": 1.3274982809899739e-06, "loss": 2.4579, "step": 325910 }, { "epoch": 0.0003072, "grad_norm": 0.8966789245605469, "learning_rate": 1.3270973830104627e-06, "loss": 2.6198, "step": 325920 }, { "epoch": 0.0003328, "grad_norm": 1.2133128643035889, "learning_rate": 1.3266965412724208e-06, "loss": 2.4431, "step": 325930 }, { "epoch": 0.0003584, "grad_norm": 0.9345282912254333, "learning_rate": 1.3262957557784473e-06, "loss": 3.0184, "step": 325940 }, { "epoch": 0.000384, "grad_norm": 0.8209258317947388, "learning_rate": 1.3258950265311399e-06, "loss": 2.6682, "step": 325950 }, { "epoch": 0.0004096, "grad_norm": 0.8477882742881775, "learning_rate": 1.3254943535330978e-06, "loss": 2.7171, "step": 325960 }, { "epoch": 0.0004352, "grad_norm": 0.8091217875480652, "learning_rate": 1.3250937367869222e-06, "loss": 2.7682, "step": 325970 }, { "epoch": 0.0004608, "grad_norm": 0.8062877655029297, "learning_rate": 1.3246931762952054e-06, "loss": 2.7477, "step": 325980 }, { "epoch": 0.0004864, "grad_norm": 0.8545608520507812, "learning_rate": 1.3242926720605453e-06, "loss": 2.5074, "step": 325990 }, { "epoch": 0.000512, "grad_norm": 0.8476845026016235, "learning_rate": 1.3238922240855445e-06, "loss": 2.4255, "step": 326000 }, { "epoch": 0.0005376, "grad_norm": 0.9041802287101746, "learning_rate": 1.3234918323727963e-06, "loss": 2.6188, "step": 326010 }, { "epoch": 0.0005632, "grad_norm": 0.8703715801239014, "learning_rate": 1.3230914969249008e-06, "loss": 2.5898, "step": 326020 }, { "epoch": 0.0005888, "grad_norm": 0.825666069984436, "learning_rate": 1.3226912177444474e-06, "loss": 2.2733, "step": 326030 }, { "epoch": 0.0006144, "grad_norm": 0.83741694688797, "learning_rate": 1.3222909948340356e-06, "loss": 2.519, "step": 326040 }, { "epoch": 0.00064, "grad_norm": 0.793516218662262, "learning_rate": 1.3218908281962607e-06, "loss": 2.4702, "step": 326050 }, { "epoch": 0.0006656, "grad_norm": 0.9176777601242065, "learning_rate": 1.3214907178337177e-06, "loss": 2.6142, "step": 326060 }, { "epoch": 0.0006912, "grad_norm": 0.9112191200256348, "learning_rate": 1.3210906637490006e-06, "loss": 2.8647, "step": 326070 }, { "epoch": 0.0007168, "grad_norm": 0.9322320818901062, "learning_rate": 1.320690665944704e-06, "loss": 2.8258, "step": 326080 }, { "epoch": 0.0007424, "grad_norm": 0.8634607195854187, "learning_rate": 1.3202907244234243e-06, "loss": 2.5053, "step": 326090 }, { "epoch": 0.000768, "grad_norm": 0.9865230917930603, "learning_rate": 1.3198908391877497e-06, "loss": 2.892, "step": 326100 }, { "epoch": 0.0007936, "grad_norm": 0.9570887088775635, "learning_rate": 1.3194910102402757e-06, "loss": 2.7588, "step": 326110 }, { "epoch": 0.0008192, "grad_norm": 0.8031737804412842, "learning_rate": 1.3190912375835963e-06, "loss": 2.6588, "step": 326120 }, { "epoch": 0.0008448, "grad_norm": 0.9193813800811768, "learning_rate": 1.3186915212203033e-06, "loss": 2.6157, "step": 326130 }, { "epoch": 0.0008704, "grad_norm": 1.3056697845458984, "learning_rate": 1.3182918611529883e-06, "loss": 2.7743, "step": 326140 }, { "epoch": 0.000896, "grad_norm": 0.8267562985420227, "learning_rate": 1.3178922573842424e-06, "loss": 2.5575, "step": 326150 }, { "epoch": 0.0009216, "grad_norm": 0.8156375288963318, "learning_rate": 1.317492709916659e-06, "loss": 2.6663, "step": 326160 }, { "epoch": 0.0009472, "grad_norm": 0.8114175796508789, "learning_rate": 1.3170932187528274e-06, "loss": 2.4025, "step": 326170 }, { "epoch": 0.0009728, "grad_norm": 0.8959010243415833, "learning_rate": 1.3166937838953386e-06, "loss": 2.6831, "step": 326180 }, { "epoch": 0.0009984, "grad_norm": 0.8240697383880615, "learning_rate": 1.316294405346783e-06, "loss": 2.7172, "step": 326190 }, { "epoch": 0.001024, "grad_norm": 0.8001580834388733, "learning_rate": 1.3158950831097495e-06, "loss": 2.5337, "step": 326200 }, { "epoch": 0.0010496, "grad_norm": 0.8952430486679077, "learning_rate": 1.3154958171868293e-06, "loss": 2.7167, "step": 326210 }, { "epoch": 0.0010752, "grad_norm": 0.8884713649749756, "learning_rate": 1.3150966075806127e-06, "loss": 2.8343, "step": 326220 }, { "epoch": 0.0011008, "grad_norm": 1.0986766815185547, "learning_rate": 1.314697454293684e-06, "loss": 2.7106, "step": 326230 }, { "epoch": 0.0011264, "grad_norm": 0.852038562297821, "learning_rate": 1.314298357328635e-06, "loss": 2.792, "step": 326240 }, { "epoch": 0.001152, "grad_norm": 0.8164520859718323, "learning_rate": 1.3138993166880509e-06, "loss": 2.4303, "step": 326250 }, { "epoch": 0.0011776, "grad_norm": 0.9028987288475037, "learning_rate": 1.3135003323745232e-06, "loss": 2.8432, "step": 326260 }, { "epoch": 0.0012032, "grad_norm": 0.9099523425102234, "learning_rate": 1.3131014043906365e-06, "loss": 2.8095, "step": 326270 }, { "epoch": 0.0012288, "grad_norm": 0.9675198793411255, "learning_rate": 1.3127025327389776e-06, "loss": 2.8825, "step": 326280 }, { "epoch": 0.0012544, "grad_norm": 0.8179372549057007, "learning_rate": 1.3123037174221377e-06, "loss": 2.9221, "step": 326290 }, { "epoch": 0.00128, "grad_norm": 0.765418529510498, "learning_rate": 1.3119049584426957e-06, "loss": 2.6287, "step": 326300 }, { "epoch": 0.0013056, "grad_norm": 1.0501891374588013, "learning_rate": 1.3115062558032422e-06, "loss": 2.5258, "step": 326310 }, { "epoch": 0.0013312, "grad_norm": 0.8572341799736023, "learning_rate": 1.3111076095063601e-06, "loss": 2.7414, "step": 326320 }, { "epoch": 0.0013568, "grad_norm": 0.885490894317627, "learning_rate": 1.3107090195546378e-06, "loss": 2.5437, "step": 326330 }, { "epoch": 0.0013824, "grad_norm": 0.9342340230941772, "learning_rate": 1.3103104859506543e-06, "loss": 2.7065, "step": 326340 }, { "epoch": 0.001408, "grad_norm": 0.9545117616653442, "learning_rate": 1.309912008697003e-06, "loss": 2.6611, "step": 326350 }, { "epoch": 0.0014336, "grad_norm": 0.8499059677124023, "learning_rate": 1.3095135877962605e-06, "loss": 2.8422, "step": 326360 }, { "epoch": 0.0014592, "grad_norm": 0.7874073386192322, "learning_rate": 1.3091152232510129e-06, "loss": 2.5275, "step": 326370 }, { "epoch": 0.0014848, "grad_norm": 0.862197995185852, "learning_rate": 1.3087169150638435e-06, "loss": 2.5941, "step": 326380 }, { "epoch": 0.0015104, "grad_norm": 0.9348903298377991, "learning_rate": 1.3083186632373347e-06, "loss": 2.7486, "step": 326390 }, { "epoch": 0.001536, "grad_norm": 0.8135658502578735, "learning_rate": 1.3079204677740687e-06, "loss": 2.2203, "step": 326400 }, { "epoch": 0.0015616, "grad_norm": 0.8995003700256348, "learning_rate": 1.3075223286766293e-06, "loss": 2.7443, "step": 326410 }, { "epoch": 0.0015872, "grad_norm": 0.7997111082077026, "learning_rate": 1.3071242459475996e-06, "loss": 2.5388, "step": 326420 }, { "epoch": 0.0016128, "grad_norm": 0.85710608959198, "learning_rate": 1.3067262195895557e-06, "loss": 2.7292, "step": 326430 }, { "epoch": 0.0016384, "grad_norm": 0.9133060574531555, "learning_rate": 1.3063282496050833e-06, "loss": 2.3581, "step": 326440 }, { "epoch": 0.001664, "grad_norm": 0.7913486957550049, "learning_rate": 1.3059303359967612e-06, "loss": 2.6424, "step": 326450 }, { "epoch": 0.0016896, "grad_norm": 1.075103998184204, "learning_rate": 1.3055324787671697e-06, "loss": 2.5712, "step": 326460 }, { "epoch": 0.0017152, "grad_norm": 0.9380591511726379, "learning_rate": 1.30513467791889e-06, "loss": 2.6775, "step": 326470 }, { "epoch": 0.0017408, "grad_norm": 0.9542899131774902, "learning_rate": 1.3047369334545e-06, "loss": 2.7195, "step": 326480 }, { "epoch": 0.0017664, "grad_norm": 0.7690218687057495, "learning_rate": 1.3043392453765835e-06, "loss": 2.6675, "step": 326490 }, { "epoch": 0.001792, "grad_norm": 1.0157825946807861, "learning_rate": 1.3039416136877137e-06, "loss": 2.6827, "step": 326500 }, { "epoch": 0.0018176, "grad_norm": 0.8965070843696594, "learning_rate": 1.3035440383904686e-06, "loss": 2.4902, "step": 326510 }, { "epoch": 0.0018432, "grad_norm": 0.8724486231803894, "learning_rate": 1.3031465194874316e-06, "loss": 2.5394, "step": 326520 }, { "epoch": 0.0018688, "grad_norm": 0.8368196487426758, "learning_rate": 1.3027490569811785e-06, "loss": 2.4103, "step": 326530 }, { "epoch": 0.0018944, "grad_norm": 1.0587528944015503, "learning_rate": 1.3023516508742862e-06, "loss": 2.9106, "step": 326540 }, { "epoch": 0.00192, "grad_norm": 0.836105465888977, "learning_rate": 1.3019543011693337e-06, "loss": 2.1615, "step": 326550 }, { "epoch": 0.0019456, "grad_norm": 0.7474097013473511, "learning_rate": 1.3015570078688933e-06, "loss": 2.7732, "step": 326560 }, { "epoch": 0.0019712, "grad_norm": 0.8331567049026489, "learning_rate": 1.301159770975544e-06, "loss": 2.8378, "step": 326570 }, { "epoch": 0.0019968, "grad_norm": 1.4155484437942505, "learning_rate": 1.300762590491863e-06, "loss": 2.6003, "step": 326580 }, { "epoch": 0.0020224, "grad_norm": 0.8012579083442688, "learning_rate": 1.3003654664204235e-06, "loss": 2.782, "step": 326590 }, { "epoch": 0.002048, "grad_norm": 0.9955385327339172, "learning_rate": 1.2999683987638024e-06, "loss": 2.5084, "step": 326600 }, { "epoch": 0.0020736, "grad_norm": 1.0990909337997437, "learning_rate": 1.299571387524573e-06, "loss": 2.3559, "step": 326610 }, { "epoch": 0.0020992, "grad_norm": 0.8821748495101929, "learning_rate": 1.2991744327053146e-06, "loss": 2.9867, "step": 326620 }, { "epoch": 0.0021248, "grad_norm": 0.8332223892211914, "learning_rate": 1.298777534308594e-06, "loss": 2.8385, "step": 326630 }, { "epoch": 0.0021504, "grad_norm": 0.8522856831550598, "learning_rate": 1.2983806923369901e-06, "loss": 2.9246, "step": 326640 }, { "epoch": 0.002176, "grad_norm": 0.8285135626792908, "learning_rate": 1.2979839067930733e-06, "loss": 2.8531, "step": 326650 }, { "epoch": 0.0022016, "grad_norm": 0.784846842288971, "learning_rate": 1.2975871776794191e-06, "loss": 2.8976, "step": 326660 }, { "epoch": 0.0022272, "grad_norm": 0.8932733535766602, "learning_rate": 1.2971905049985988e-06, "loss": 2.6053, "step": 326670 }, { "epoch": 0.0022528, "grad_norm": 0.9313924908638, "learning_rate": 1.296793888753185e-06, "loss": 2.5519, "step": 326680 }, { "epoch": 0.0022784, "grad_norm": 0.7903584241867065, "learning_rate": 1.2963973289457498e-06, "loss": 2.8725, "step": 326690 }, { "epoch": 0.002304, "grad_norm": 0.9440039992332458, "learning_rate": 1.2960008255788658e-06, "loss": 2.5364, "step": 326700 }, { "epoch": 0.0023296, "grad_norm": 0.8565203547477722, "learning_rate": 1.295604378655102e-06, "loss": 2.7901, "step": 326710 }, { "epoch": 0.0023552, "grad_norm": 0.8290298581123352, "learning_rate": 1.2952079881770308e-06, "loss": 2.4249, "step": 326720 }, { "epoch": 0.0023808, "grad_norm": 0.8003944158554077, "learning_rate": 1.2948116541472222e-06, "loss": 2.6409, "step": 326730 }, { "epoch": 0.0024064, "grad_norm": 1.1128684282302856, "learning_rate": 1.2944153765682466e-06, "loss": 2.7973, "step": 326740 }, { "epoch": 0.002432, "grad_norm": 0.8906512260437012, "learning_rate": 1.2940191554426763e-06, "loss": 2.524, "step": 326750 }, { "epoch": 0.0024576, "grad_norm": 0.8390956521034241, "learning_rate": 1.2936229907730757e-06, "loss": 2.8187, "step": 326760 }, { "epoch": 0.0024832, "grad_norm": 0.8378747701644897, "learning_rate": 1.2932268825620153e-06, "loss": 2.9187, "step": 326770 }, { "epoch": 0.0025088, "grad_norm": 1.3054797649383545, "learning_rate": 1.2928308308120653e-06, "loss": 2.922, "step": 326780 }, { "epoch": 0.0025344, "grad_norm": 0.7943570613861084, "learning_rate": 1.2924348355257931e-06, "loss": 2.9283, "step": 326790 }, { "epoch": 0.00256, "grad_norm": 0.8234783411026001, "learning_rate": 1.2920388967057662e-06, "loss": 2.6206, "step": 326800 }, { "epoch": 0.0025856, "grad_norm": 0.8553141355514526, "learning_rate": 1.2916430143545534e-06, "loss": 2.853, "step": 326810 }, { "epoch": 0.0026112, "grad_norm": 0.8213635683059692, "learning_rate": 1.2912471884747235e-06, "loss": 2.6544, "step": 326820 }, { "epoch": 0.0026368, "grad_norm": 0.8964807987213135, "learning_rate": 1.2908514190688393e-06, "loss": 2.8134, "step": 326830 }, { "epoch": 0.0026624, "grad_norm": 0.8687035441398621, "learning_rate": 1.2904557061394684e-06, "loss": 2.9434, "step": 326840 }, { "epoch": 0.002688, "grad_norm": 0.8487896919250488, "learning_rate": 1.2900600496891769e-06, "loss": 2.7298, "step": 326850 }, { "epoch": 0.0027136, "grad_norm": 0.8346304297447205, "learning_rate": 1.28966444972053e-06, "loss": 2.7676, "step": 326860 }, { "epoch": 0.0027392, "grad_norm": 0.9750543236732483, "learning_rate": 1.2892689062360964e-06, "loss": 2.8709, "step": 326870 }, { "epoch": 0.0027648, "grad_norm": 0.8736129403114319, "learning_rate": 1.2888734192384411e-06, "loss": 3.0234, "step": 326880 }, { "epoch": 0.0027904, "grad_norm": 0.8076476454734802, "learning_rate": 1.2884779887301236e-06, "loss": 2.8896, "step": 326890 }, { "epoch": 0.002816, "grad_norm": 0.9491422772407532, "learning_rate": 1.2880826147137115e-06, "loss": 2.756, "step": 326900 }, { "epoch": 0.0028416, "grad_norm": 0.8210769295692444, "learning_rate": 1.2876872971917686e-06, "loss": 2.7615, "step": 326910 }, { "epoch": 0.0028672, "grad_norm": 0.8409467935562134, "learning_rate": 1.2872920361668573e-06, "loss": 2.9336, "step": 326920 }, { "epoch": 0.0028928, "grad_norm": 1.0096962451934814, "learning_rate": 1.286896831641542e-06, "loss": 2.7367, "step": 326930 }, { "epoch": 0.0029184, "grad_norm": 0.940328061580658, "learning_rate": 1.2865016836183852e-06, "loss": 2.7914, "step": 326940 }, { "epoch": 0.002944, "grad_norm": 0.8147222995758057, "learning_rate": 1.2861065920999516e-06, "loss": 2.7218, "step": 326950 }, { "epoch": 0.0029696, "grad_norm": 0.8975940942764282, "learning_rate": 1.285711557088799e-06, "loss": 2.6785, "step": 326960 }, { "epoch": 0.0029952, "grad_norm": 0.8025913238525391, "learning_rate": 1.28531657858749e-06, "loss": 2.8357, "step": 326970 }, { "epoch": 0.0030208, "grad_norm": 0.8016673922538757, "learning_rate": 1.2849216565985878e-06, "loss": 2.6635, "step": 326980 }, { "epoch": 0.0030464, "grad_norm": 0.8662099242210388, "learning_rate": 1.2845267911246517e-06, "loss": 2.5784, "step": 326990 }, { "epoch": 0.003072, "grad_norm": 0.9536392688751221, "learning_rate": 1.284131982168243e-06, "loss": 2.8856, "step": 327000 }, { "epoch": 0.0030976, "grad_norm": 0.7984493374824524, "learning_rate": 1.2837372297319228e-06, "loss": 2.7105, "step": 327010 }, { "epoch": 0.0031232, "grad_norm": 1.119134545326233, "learning_rate": 1.2833425338182515e-06, "loss": 2.7724, "step": 327020 }, { "epoch": 0.0031488, "grad_norm": 0.8825649619102478, "learning_rate": 1.2829478944297836e-06, "loss": 2.8263, "step": 327030 }, { "epoch": 0.0031744, "grad_norm": 0.894766092300415, "learning_rate": 1.2825533115690835e-06, "loss": 2.8167, "step": 327040 }, { "epoch": 0.0032, "grad_norm": 0.9536925554275513, "learning_rate": 1.2821587852387086e-06, "loss": 3.0143, "step": 327050 }, { "epoch": 0.0032256, "grad_norm": 0.853448212146759, "learning_rate": 1.2817643154412173e-06, "loss": 2.7056, "step": 327060 }, { "epoch": 0.0032512, "grad_norm": 0.9817862510681152, "learning_rate": 1.281369902179167e-06, "loss": 2.9026, "step": 327070 }, { "epoch": 0.0032768, "grad_norm": 0.8402519822120667, "learning_rate": 1.2809755454551186e-06, "loss": 2.9926, "step": 327080 }, { "epoch": 0.0033024, "grad_norm": 0.933967113494873, "learning_rate": 1.2805812452716237e-06, "loss": 2.7982, "step": 327090 }, { "epoch": 0.003328, "grad_norm": 0.8823518753051758, "learning_rate": 1.2801870016312424e-06, "loss": 2.7533, "step": 327100 }, { "epoch": 0.0033536, "grad_norm": 0.7800216674804688, "learning_rate": 1.2797928145365312e-06, "loss": 2.9119, "step": 327110 }, { "epoch": 0.0033792, "grad_norm": 0.8574318289756775, "learning_rate": 1.2793986839900451e-06, "loss": 2.8854, "step": 327120 }, { "epoch": 0.0034048, "grad_norm": 0.8558619618415833, "learning_rate": 1.279004609994342e-06, "loss": 2.7051, "step": 327130 }, { "epoch": 0.0034304, "grad_norm": 0.9450064897537231, "learning_rate": 1.2786105925519754e-06, "loss": 3.119, "step": 327140 }, { "epoch": 0.003456, "grad_norm": 0.8238637447357178, "learning_rate": 1.2782166316655032e-06, "loss": 2.9029, "step": 327150 }, { "epoch": 0.0034816, "grad_norm": 0.97004634141922, "learning_rate": 1.2778227273374766e-06, "loss": 2.6741, "step": 327160 }, { "epoch": 0.0035072, "grad_norm": 1.2129985094070435, "learning_rate": 1.2774288795704516e-06, "loss": 2.7441, "step": 327170 }, { "epoch": 0.0035328, "grad_norm": 0.8026474714279175, "learning_rate": 1.2770350883669823e-06, "loss": 2.7676, "step": 327180 }, { "epoch": 0.0035584, "grad_norm": 0.7859315872192383, "learning_rate": 1.2766413537296219e-06, "loss": 2.7764, "step": 327190 }, { "epoch": 0.003584, "grad_norm": 0.8023171424865723, "learning_rate": 1.2762476756609233e-06, "loss": 2.8566, "step": 327200 }, { "epoch": 0.0036096, "grad_norm": 0.837863028049469, "learning_rate": 1.2758540541634413e-06, "loss": 2.865, "step": 327210 }, { "epoch": 0.0036352, "grad_norm": 0.9568873643875122, "learning_rate": 1.275460489239726e-06, "loss": 2.8041, "step": 327220 }, { "epoch": 0.0036608, "grad_norm": 0.9311797618865967, "learning_rate": 1.2750669808923312e-06, "loss": 2.8161, "step": 327230 }, { "epoch": 0.0036864, "grad_norm": 0.7726151347160339, "learning_rate": 1.2746735291238077e-06, "loss": 2.8546, "step": 327240 }, { "epoch": 0.003712, "grad_norm": 0.8621149063110352, "learning_rate": 1.2742801339367084e-06, "loss": 2.7857, "step": 327250 }, { "epoch": 0.0037376, "grad_norm": 0.9180018901824951, "learning_rate": 1.2738867953335832e-06, "loss": 2.9961, "step": 327260 }, { "epoch": 0.0037632, "grad_norm": 0.8224981427192688, "learning_rate": 1.2734935133169824e-06, "loss": 2.8045, "step": 327270 }, { "epoch": 0.0037888, "grad_norm": 0.8242180347442627, "learning_rate": 1.2731002878894594e-06, "loss": 2.8857, "step": 327280 }, { "epoch": 0.0038144, "grad_norm": 0.8388217091560364, "learning_rate": 1.27270711905356e-06, "loss": 2.843, "step": 327290 }, { "epoch": 0.00384, "grad_norm": 0.8264674544334412, "learning_rate": 1.2723140068118346e-06, "loss": 2.8871, "step": 327300 }, { "epoch": 0.0038656, "grad_norm": 0.9450434446334839, "learning_rate": 1.271920951166834e-06, "loss": 2.8862, "step": 327310 }, { "epoch": 0.0038912, "grad_norm": 0.857071042060852, "learning_rate": 1.2715279521211066e-06, "loss": 2.8379, "step": 327320 }, { "epoch": 0.0039168, "grad_norm": 0.850509762763977, "learning_rate": 1.2711350096772001e-06, "loss": 2.9095, "step": 327330 }, { "epoch": 0.0039424, "grad_norm": 0.7973949313163757, "learning_rate": 1.2707421238376626e-06, "loss": 2.7486, "step": 327340 }, { "epoch": 0.003968, "grad_norm": 0.8044846057891846, "learning_rate": 1.2703492946050456e-06, "loss": 2.7402, "step": 327350 }, { "epoch": 0.0039936, "grad_norm": 0.7906673550605774, "learning_rate": 1.269956521981891e-06, "loss": 2.7982, "step": 327360 }, { "epoch": 0.0040192, "grad_norm": 0.7836939096450806, "learning_rate": 1.2695638059707482e-06, "loss": 2.7747, "step": 327370 }, { "epoch": 0.0040448, "grad_norm": 1.1124327182769775, "learning_rate": 1.269171146574162e-06, "loss": 2.935, "step": 327380 }, { "epoch": 0.0040704, "grad_norm": 0.8501496315002441, "learning_rate": 1.268778543794682e-06, "loss": 2.7426, "step": 327390 }, { "epoch": 0.004096, "grad_norm": 0.9052197933197021, "learning_rate": 1.2683859976348556e-06, "loss": 2.8375, "step": 327400 }, { "epoch": 0.0041216, "grad_norm": 0.9943901300430298, "learning_rate": 1.2679935080972228e-06, "loss": 2.9253, "step": 327410 }, { "epoch": 0.0041472, "grad_norm": 0.7656209468841553, "learning_rate": 1.2676010751843314e-06, "loss": 2.786, "step": 327420 }, { "epoch": 0.0041728, "grad_norm": 0.8352909088134766, "learning_rate": 1.2672086988987264e-06, "loss": 2.8281, "step": 327430 }, { "epoch": 0.0041984, "grad_norm": 0.9125028848648071, "learning_rate": 1.2668163792429522e-06, "loss": 2.8392, "step": 327440 }, { "epoch": 0.004224, "grad_norm": 0.755190908908844, "learning_rate": 1.2664241162195523e-06, "loss": 2.8115, "step": 327450 }, { "epoch": 0.0042496, "grad_norm": 1.090585708618164, "learning_rate": 1.2660319098310713e-06, "loss": 3.0021, "step": 327460 }, { "epoch": 0.0042752, "grad_norm": 0.8932942152023315, "learning_rate": 1.2656397600800518e-06, "loss": 2.9796, "step": 327470 }, { "epoch": 0.0043008, "grad_norm": 0.8269563913345337, "learning_rate": 1.2652476669690394e-06, "loss": 2.9824, "step": 327480 }, { "epoch": 0.0043264, "grad_norm": 0.8105273842811584, "learning_rate": 1.2648556305005733e-06, "loss": 2.7074, "step": 327490 }, { "epoch": 0.004352, "grad_norm": 0.8184722661972046, "learning_rate": 1.2644636506771956e-06, "loss": 2.8064, "step": 327500 }, { "epoch": 0.0043776, "grad_norm": 0.9169996380805969, "learning_rate": 1.2640717275014502e-06, "loss": 2.6761, "step": 327510 }, { "epoch": 0.0044032, "grad_norm": 0.7990601062774658, "learning_rate": 1.2636798609758771e-06, "loss": 2.6867, "step": 327520 }, { "epoch": 0.0044288, "grad_norm": 0.8188983201980591, "learning_rate": 1.2632880511030187e-06, "loss": 2.9722, "step": 327530 }, { "epoch": 0.0044544, "grad_norm": 0.810746967792511, "learning_rate": 1.2628962978854165e-06, "loss": 2.71, "step": 327540 }, { "epoch": 0.00448, "grad_norm": 0.8677083253860474, "learning_rate": 1.2625046013256082e-06, "loss": 2.9334, "step": 327550 }, { "epoch": 0.0045056, "grad_norm": 0.8912279605865479, "learning_rate": 1.2621129614261362e-06, "loss": 2.8152, "step": 327560 }, { "epoch": 0.0045312, "grad_norm": 0.9763869047164917, "learning_rate": 1.2617213781895388e-06, "loss": 2.4996, "step": 327570 }, { "epoch": 0.0045568, "grad_norm": 0.8755355477333069, "learning_rate": 1.261329851618357e-06, "loss": 2.7806, "step": 327580 }, { "epoch": 0.0045824, "grad_norm": 0.7855064272880554, "learning_rate": 1.2609383817151278e-06, "loss": 2.5947, "step": 327590 }, { "epoch": 0.004608, "grad_norm": 0.8491199612617493, "learning_rate": 1.2605469684823924e-06, "loss": 3.0241, "step": 327600 }, { "epoch": 0.0046336, "grad_norm": 0.8178635239601135, "learning_rate": 1.2601556119226855e-06, "loss": 2.852, "step": 327610 }, { "epoch": 0.0046592, "grad_norm": 0.9074611663818359, "learning_rate": 1.2597643120385471e-06, "loss": 3.1418, "step": 327620 }, { "epoch": 0.0046848, "grad_norm": 0.7886260747909546, "learning_rate": 1.2593730688325134e-06, "loss": 2.9013, "step": 327630 }, { "epoch": 0.0047104, "grad_norm": 0.925348162651062, "learning_rate": 1.2589818823071232e-06, "loss": 3.108, "step": 327640 }, { "epoch": 0.004736, "grad_norm": 0.8227410912513733, "learning_rate": 1.258590752464912e-06, "loss": 2.7488, "step": 327650 }, { "epoch": 0.0047616, "grad_norm": 0.7813758254051208, "learning_rate": 1.2581996793084173e-06, "loss": 2.9088, "step": 327660 }, { "epoch": 0.0047872, "grad_norm": 0.7846332788467407, "learning_rate": 1.2578086628401732e-06, "loss": 2.8224, "step": 327670 }, { "epoch": 0.0048128, "grad_norm": 0.8091239929199219, "learning_rate": 1.25741770306272e-06, "loss": 2.7753, "step": 327680 }, { "epoch": 0.0048384, "grad_norm": 0.9668617844581604, "learning_rate": 1.2570267999785857e-06, "loss": 2.8144, "step": 327690 }, { "epoch": 0.004864, "grad_norm": 0.8155282735824585, "learning_rate": 1.2566359535903106e-06, "loss": 2.789, "step": 327700 }, { "epoch": 0.0048896, "grad_norm": 0.8015909194946289, "learning_rate": 1.256245163900427e-06, "loss": 2.7151, "step": 327710 }, { "epoch": 0.0049152, "grad_norm": 0.7805622220039368, "learning_rate": 1.2558544309114695e-06, "loss": 2.58, "step": 327720 }, { "epoch": 0.0049408, "grad_norm": 0.9852456450462341, "learning_rate": 1.2554637546259719e-06, "loss": 2.8633, "step": 327730 }, { "epoch": 0.0049664, "grad_norm": 0.9919077754020691, "learning_rate": 1.2550731350464684e-06, "loss": 2.6594, "step": 327740 }, { "epoch": 0.004992, "grad_norm": 0.8083094954490662, "learning_rate": 1.2546825721754907e-06, "loss": 2.8198, "step": 327750 }, { "epoch": 0.0050176, "grad_norm": 0.77829909324646, "learning_rate": 1.2542920660155734e-06, "loss": 2.7583, "step": 327760 }, { "epoch": 0.0050432, "grad_norm": 0.8319878578186035, "learning_rate": 1.2539016165692475e-06, "loss": 2.7863, "step": 327770 }, { "epoch": 0.0050688, "grad_norm": 0.8324732184410095, "learning_rate": 1.2535112238390445e-06, "loss": 2.8801, "step": 327780 }, { "epoch": 0.0050944, "grad_norm": 0.8710058331489563, "learning_rate": 1.253120887827497e-06, "loss": 2.9949, "step": 327790 }, { "epoch": 0.00512, "grad_norm": 0.898145318031311, "learning_rate": 1.2527306085371383e-06, "loss": 3.0747, "step": 327800 }, { "epoch": 0.0051456, "grad_norm": 0.8913342356681824, "learning_rate": 1.252340385970494e-06, "loss": 2.811, "step": 327810 }, { "epoch": 0.0051712, "grad_norm": 0.9101749062538147, "learning_rate": 1.2519502201300982e-06, "loss": 2.7993, "step": 327820 }, { "epoch": 0.0051968, "grad_norm": 0.8762945532798767, "learning_rate": 1.2515601110184793e-06, "loss": 2.6196, "step": 327830 }, { "epoch": 0.0052224, "grad_norm": 0.8227319121360779, "learning_rate": 1.2511700586381692e-06, "loss": 2.776, "step": 327840 }, { "epoch": 0.005248, "grad_norm": 0.8391592502593994, "learning_rate": 1.2507800629916954e-06, "loss": 2.7384, "step": 327850 }, { "epoch": 0.0052736, "grad_norm": 0.8749176859855652, "learning_rate": 1.2503901240815886e-06, "loss": 2.9474, "step": 327860 }, { "epoch": 0.0052992, "grad_norm": 0.8200588822364807, "learning_rate": 1.2500002419103775e-06, "loss": 2.7621, "step": 327870 }, { "epoch": 0.0053248, "grad_norm": 0.8244143724441528, "learning_rate": 1.2496104164805878e-06, "loss": 2.8172, "step": 327880 }, { "epoch": 0.0053504, "grad_norm": 1.0557937622070312, "learning_rate": 1.2492206477947489e-06, "loss": 2.7919, "step": 327890 }, { "epoch": 0.005376, "grad_norm": 0.7897970676422119, "learning_rate": 1.2488309358553863e-06, "loss": 2.5248, "step": 327900 }, { "epoch": 0.0054016, "grad_norm": 1.1849123239517212, "learning_rate": 1.2484412806650304e-06, "loss": 2.9518, "step": 327910 }, { "epoch": 0.0054272, "grad_norm": 0.7644347548484802, "learning_rate": 1.2480516822262078e-06, "loss": 2.5615, "step": 327920 }, { "epoch": 0.0054528, "grad_norm": 0.8030368685722351, "learning_rate": 1.2476621405414468e-06, "loss": 2.6877, "step": 327930 }, { "epoch": 0.0054784, "grad_norm": 0.8123141527175903, "learning_rate": 1.2472726556132674e-06, "loss": 2.4819, "step": 327940 }, { "epoch": 0.005504, "grad_norm": 0.7945491671562195, "learning_rate": 1.2468832274442e-06, "loss": 2.6235, "step": 327950 }, { "epoch": 0.0055296, "grad_norm": 0.8448686003684998, "learning_rate": 1.2464938560367678e-06, "loss": 2.6757, "step": 327960 }, { "epoch": 0.0055552, "grad_norm": 0.8530932068824768, "learning_rate": 1.2461045413934958e-06, "loss": 2.7559, "step": 327970 }, { "epoch": 0.0055808, "grad_norm": 0.9563354849815369, "learning_rate": 1.2457152835169106e-06, "loss": 2.7147, "step": 327980 }, { "epoch": 0.0056064, "grad_norm": 0.967740535736084, "learning_rate": 1.2453260824095348e-06, "loss": 2.8592, "step": 327990 }, { "epoch": 0.005632, "grad_norm": 0.8588660955429077, "learning_rate": 1.244936938073894e-06, "loss": 2.7608, "step": 328000 }, { "epoch": 0.0056576, "grad_norm": 0.8185910582542419, "learning_rate": 1.2445478505125096e-06, "loss": 2.6667, "step": 328010 }, { "epoch": 0.0056832, "grad_norm": 0.8339497447013855, "learning_rate": 1.2441588197279042e-06, "loss": 2.8458, "step": 328020 }, { "epoch": 0.0057088, "grad_norm": 0.7771837115287781, "learning_rate": 1.2437698457226022e-06, "loss": 2.9427, "step": 328030 }, { "epoch": 0.0057344, "grad_norm": 0.8272473216056824, "learning_rate": 1.2433809284991261e-06, "loss": 2.7678, "step": 328040 }, { "epoch": 0.00576, "grad_norm": 0.8348110318183899, "learning_rate": 1.2429920680599973e-06, "loss": 2.6224, "step": 328050 }, { "epoch": 0.0057856, "grad_norm": 1.4131252765655518, "learning_rate": 1.2426032644077368e-06, "loss": 2.6581, "step": 328060 }, { "epoch": 0.0058112, "grad_norm": 0.8177957534790039, "learning_rate": 1.2422145175448696e-06, "loss": 2.7596, "step": 328070 }, { "epoch": 0.0058368, "grad_norm": 0.8092009425163269, "learning_rate": 1.2418258274739093e-06, "loss": 2.7781, "step": 328080 }, { "epoch": 0.0058624, "grad_norm": 0.8310366272926331, "learning_rate": 1.2414371941973836e-06, "loss": 2.7963, "step": 328090 }, { "epoch": 0.005888, "grad_norm": 0.7691031694412231, "learning_rate": 1.2410486177178082e-06, "loss": 2.6415, "step": 328100 }, { "epoch": 0.0059136, "grad_norm": 0.8556651473045349, "learning_rate": 1.240660098037706e-06, "loss": 2.8043, "step": 328110 }, { "epoch": 0.0059392, "grad_norm": 0.8306974172592163, "learning_rate": 1.2402716351595955e-06, "loss": 2.8235, "step": 328120 }, { "epoch": 0.0059648, "grad_norm": 0.7670564651489258, "learning_rate": 1.2398832290859963e-06, "loss": 2.7862, "step": 328130 }, { "epoch": 0.0059904, "grad_norm": 0.8510322570800781, "learning_rate": 1.239494879819425e-06, "loss": 2.775, "step": 328140 }, { "epoch": 0.006016, "grad_norm": 0.8294538259506226, "learning_rate": 1.239106587362401e-06, "loss": 2.7962, "step": 328150 }, { "epoch": 0.0060416, "grad_norm": 0.766947329044342, "learning_rate": 1.2387183517174416e-06, "loss": 2.7713, "step": 328160 }, { "epoch": 0.0060672, "grad_norm": 0.8140003085136414, "learning_rate": 1.2383301728870667e-06, "loss": 2.8021, "step": 328170 }, { "epoch": 0.0060928, "grad_norm": 0.7707648277282715, "learning_rate": 1.2379420508737904e-06, "loss": 2.6526, "step": 328180 }, { "epoch": 0.0061184, "grad_norm": 0.8780421018600464, "learning_rate": 1.2375539856801322e-06, "loss": 2.4559, "step": 328190 }, { "epoch": 0.006144, "grad_norm": 0.757397472858429, "learning_rate": 1.2371659773086098e-06, "loss": 2.7091, "step": 328200 }, { "epoch": 0.0061696, "grad_norm": 0.8640891313552856, "learning_rate": 1.2367780257617346e-06, "loss": 2.7491, "step": 328210 }, { "epoch": 0.0061952, "grad_norm": 0.7833463549613953, "learning_rate": 1.2363901310420256e-06, "loss": 2.662, "step": 328220 }, { "epoch": 0.0062208, "grad_norm": 1.0117799043655396, "learning_rate": 1.2360022931519965e-06, "loss": 2.724, "step": 328230 }, { "epoch": 0.0062464, "grad_norm": 0.7826461791992188, "learning_rate": 1.2356145120941642e-06, "loss": 3.1068, "step": 328240 }, { "epoch": 0.006272, "grad_norm": 0.8787962198257446, "learning_rate": 1.2352267878710401e-06, "loss": 2.7388, "step": 328250 }, { "epoch": 0.0062976, "grad_norm": 1.12174654006958, "learning_rate": 1.2348391204851452e-06, "loss": 2.6889, "step": 328260 }, { "epoch": 0.0063232, "grad_norm": 0.8253781199455261, "learning_rate": 1.2344515099389864e-06, "loss": 2.8864, "step": 328270 }, { "epoch": 0.0063488, "grad_norm": 0.9102185964584351, "learning_rate": 1.234063956235081e-06, "loss": 2.8627, "step": 328280 }, { "epoch": 0.0063744, "grad_norm": 0.9543753862380981, "learning_rate": 1.233676459375941e-06, "loss": 2.8286, "step": 328290 }, { "epoch": 0.0064, "grad_norm": 1.0038931369781494, "learning_rate": 1.2332890193640778e-06, "loss": 2.7667, "step": 328300 }, { "epoch": 0.0064256, "grad_norm": 0.7931613922119141, "learning_rate": 1.2329016362020075e-06, "loss": 2.7516, "step": 328310 }, { "epoch": 0.0064512, "grad_norm": 0.9302113056182861, "learning_rate": 1.2325143098922386e-06, "loss": 2.8354, "step": 328320 }, { "epoch": 0.0064768, "grad_norm": 0.9101918339729309, "learning_rate": 1.2321270404372865e-06, "loss": 2.6948, "step": 328330 }, { "epoch": 0.0065024, "grad_norm": 0.8642350435256958, "learning_rate": 1.2317398278396586e-06, "loss": 2.6843, "step": 328340 }, { "epoch": 0.006528, "grad_norm": 0.9262207746505737, "learning_rate": 1.2313526721018676e-06, "loss": 2.6952, "step": 328350 }, { "epoch": 0.0065536, "grad_norm": 0.8722464442253113, "learning_rate": 1.2309655732264247e-06, "loss": 2.7891, "step": 328360 }, { "epoch": 0.0065792, "grad_norm": 0.8020289540290833, "learning_rate": 1.230578531215839e-06, "loss": 2.7221, "step": 328370 }, { "epoch": 0.0066048, "grad_norm": 0.9010424613952637, "learning_rate": 1.2301915460726211e-06, "loss": 2.7516, "step": 328380 }, { "epoch": 0.0066304, "grad_norm": 0.8247590065002441, "learning_rate": 1.2298046177992795e-06, "loss": 2.7545, "step": 328390 }, { "epoch": 0.006656, "grad_norm": 0.9199473261833191, "learning_rate": 1.229417746398327e-06, "loss": 2.9176, "step": 328400 }, { "epoch": 0.0066816, "grad_norm": 1.005042552947998, "learning_rate": 1.229030931872267e-06, "loss": 3.0282, "step": 328410 }, { "epoch": 0.0067072, "grad_norm": 0.9072540402412415, "learning_rate": 1.2286441742236077e-06, "loss": 2.745, "step": 328420 }, { "epoch": 0.0067328, "grad_norm": 0.8448885679244995, "learning_rate": 1.2282574734548624e-06, "loss": 2.7723, "step": 328430 }, { "epoch": 0.0067584, "grad_norm": 0.9278451204299927, "learning_rate": 1.227870829568537e-06, "loss": 2.6969, "step": 328440 }, { "epoch": 0.006784, "grad_norm": 0.8635758757591248, "learning_rate": 1.2274842425671374e-06, "loss": 2.7342, "step": 328450 }, { "epoch": 0.0068096, "grad_norm": 0.8900203108787537, "learning_rate": 1.2270977124531724e-06, "loss": 2.6614, "step": 328460 }, { "epoch": 0.0068352, "grad_norm": 1.31891667842865, "learning_rate": 1.2267112392291458e-06, "loss": 2.9938, "step": 328470 }, { "epoch": 0.0068608, "grad_norm": 0.8096121549606323, "learning_rate": 1.2263248228975645e-06, "loss": 2.7942, "step": 328480 }, { "epoch": 0.0068864, "grad_norm": 0.8344250917434692, "learning_rate": 1.225938463460935e-06, "loss": 2.9958, "step": 328490 }, { "epoch": 0.006912, "grad_norm": 0.8177393674850464, "learning_rate": 1.2255521609217636e-06, "loss": 2.8515, "step": 328500 }, { "epoch": 0.0069376, "grad_norm": 0.8184390664100647, "learning_rate": 1.2251659152825535e-06, "loss": 2.7944, "step": 328510 }, { "epoch": 0.0069632, "grad_norm": 0.8884753584861755, "learning_rate": 1.2247797265458118e-06, "loss": 2.675, "step": 328520 }, { "epoch": 0.0069888, "grad_norm": 0.7718799710273743, "learning_rate": 1.2243935947140417e-06, "loss": 2.6825, "step": 328530 }, { "epoch": 0.0070144, "grad_norm": 1.508011817932129, "learning_rate": 1.2240075197897461e-06, "loss": 2.5258, "step": 328540 }, { "epoch": 0.00704, "grad_norm": 0.8240785002708435, "learning_rate": 1.2236215017754284e-06, "loss": 2.729, "step": 328550 }, { "epoch": 0.0070656, "grad_norm": 0.8590908646583557, "learning_rate": 1.223235540673594e-06, "loss": 2.7622, "step": 328560 }, { "epoch": 0.0070912, "grad_norm": 1.0084493160247803, "learning_rate": 1.2228496364867437e-06, "loss": 2.7101, "step": 328570 }, { "epoch": 0.0071168, "grad_norm": 0.8451863527297974, "learning_rate": 1.2224637892173808e-06, "loss": 2.54, "step": 328580 }, { "epoch": 0.0071424, "grad_norm": 0.8603525161743164, "learning_rate": 1.2220779988680075e-06, "loss": 2.8171, "step": 328590 }, { "epoch": 0.007168, "grad_norm": 0.8814780116081238, "learning_rate": 1.2216922654411255e-06, "loss": 2.8677, "step": 328600 }, { "epoch": 0.0071936, "grad_norm": 0.8864664435386658, "learning_rate": 1.2213065889392361e-06, "loss": 2.8153, "step": 328610 }, { "epoch": 0.0072192, "grad_norm": 0.8928013443946838, "learning_rate": 1.2209209693648406e-06, "loss": 2.8455, "step": 328620 }, { "epoch": 0.0072448, "grad_norm": 0.9994096159934998, "learning_rate": 1.220535406720439e-06, "loss": 2.9083, "step": 328630 }, { "epoch": 0.0072704, "grad_norm": 0.8665103316307068, "learning_rate": 1.2201499010085316e-06, "loss": 2.7753, "step": 328640 }, { "epoch": 0.007296, "grad_norm": 0.7622321844100952, "learning_rate": 1.2197644522316199e-06, "loss": 2.7073, "step": 328650 }, { "epoch": 0.0073216, "grad_norm": 0.8499035835266113, "learning_rate": 1.219379060392204e-06, "loss": 2.8502, "step": 328660 }, { "epoch": 0.0073472, "grad_norm": 0.851860761642456, "learning_rate": 1.2189937254927787e-06, "loss": 2.6674, "step": 328670 }, { "epoch": 0.0073728, "grad_norm": 0.9681347608566284, "learning_rate": 1.218608447535845e-06, "loss": 2.8921, "step": 328680 }, { "epoch": 0.0073984, "grad_norm": 0.8567972779273987, "learning_rate": 1.2182232265239024e-06, "loss": 2.8597, "step": 328690 }, { "epoch": 0.007424, "grad_norm": 0.7896486520767212, "learning_rate": 1.217838062459449e-06, "loss": 2.5468, "step": 328700 }, { "epoch": 0.0074496, "grad_norm": 0.9634888172149658, "learning_rate": 1.217452955344981e-06, "loss": 3.0818, "step": 328710 }, { "epoch": 0.0074752, "grad_norm": 0.7790334820747375, "learning_rate": 1.2170679051829959e-06, "loss": 2.8474, "step": 328720 }, { "epoch": 0.0075008, "grad_norm": 0.8279937505722046, "learning_rate": 1.2166829119759938e-06, "loss": 3.0625, "step": 328730 }, { "epoch": 0.0075264, "grad_norm": 0.8446343541145325, "learning_rate": 1.216297975726467e-06, "loss": 2.8992, "step": 328740 }, { "epoch": 0.007552, "grad_norm": 0.8548524379730225, "learning_rate": 1.2159130964369127e-06, "loss": 2.6924, "step": 328750 }, { "epoch": 0.0075776, "grad_norm": 0.7848942279815674, "learning_rate": 1.2155282741098273e-06, "loss": 2.4984, "step": 328760 }, { "epoch": 0.0076032, "grad_norm": 0.7970472574234009, "learning_rate": 1.2151435087477049e-06, "loss": 2.8758, "step": 328770 }, { "epoch": 0.0076288, "grad_norm": 0.8558732271194458, "learning_rate": 1.2147588003530442e-06, "loss": 2.9877, "step": 328780 }, { "epoch": 0.0076544, "grad_norm": 0.8655338287353516, "learning_rate": 1.21437414892834e-06, "loss": 2.9301, "step": 328790 }, { "epoch": 0.00768, "grad_norm": 0.8596073389053345, "learning_rate": 1.2139895544760816e-06, "loss": 2.9293, "step": 328800 }, { "epoch": 0.0077056, "grad_norm": 1.0371533632278442, "learning_rate": 1.2136050169987667e-06, "loss": 2.8522, "step": 328810 }, { "epoch": 0.0077312, "grad_norm": 1.0072516202926636, "learning_rate": 1.213220536498888e-06, "loss": 2.7454, "step": 328820 }, { "epoch": 0.0077568, "grad_norm": 0.914344310760498, "learning_rate": 1.212836112978939e-06, "loss": 3.0231, "step": 328830 }, { "epoch": 0.0077824, "grad_norm": 0.9373375177383423, "learning_rate": 1.212451746441412e-06, "loss": 2.725, "step": 328840 }, { "epoch": 0.007808, "grad_norm": 0.9168649911880493, "learning_rate": 1.2120674368887998e-06, "loss": 2.8157, "step": 328850 }, { "epoch": 0.0078336, "grad_norm": 0.9124041199684143, "learning_rate": 1.2116831843235977e-06, "loss": 2.7923, "step": 328860 }, { "epoch": 0.0078592, "grad_norm": 1.0865041017532349, "learning_rate": 1.2112989887482919e-06, "loss": 2.5912, "step": 328870 }, { "epoch": 0.0078848, "grad_norm": 0.7897180318832397, "learning_rate": 1.2109148501653766e-06, "loss": 2.8355, "step": 328880 }, { "epoch": 0.0079104, "grad_norm": 0.7941802144050598, "learning_rate": 1.2105307685773426e-06, "loss": 2.9262, "step": 328890 }, { "epoch": 0.007936, "grad_norm": 0.7605833411216736, "learning_rate": 1.2101467439866798e-06, "loss": 2.7722, "step": 328900 }, { "epoch": 0.0079616, "grad_norm": 0.9430522322654724, "learning_rate": 1.2097627763958796e-06, "loss": 2.8603, "step": 328910 }, { "epoch": 0.0079872, "grad_norm": 0.8692484498023987, "learning_rate": 1.2093788658074324e-06, "loss": 2.666, "step": 328920 }, { "epoch": 0.0080128, "grad_norm": 0.7846148610115051, "learning_rate": 1.2089950122238292e-06, "loss": 2.7349, "step": 328930 }, { "epoch": 0.0080384, "grad_norm": 0.8956530094146729, "learning_rate": 1.2086112156475538e-06, "loss": 2.9123, "step": 328940 }, { "epoch": 0.008064, "grad_norm": 0.8489809632301331, "learning_rate": 1.2082274760810963e-06, "loss": 2.9507, "step": 328950 }, { "epoch": 0.0080896, "grad_norm": 0.8788822293281555, "learning_rate": 1.2078437935269505e-06, "loss": 2.8051, "step": 328960 }, { "epoch": 0.0081152, "grad_norm": 0.8963940143585205, "learning_rate": 1.2074601679875997e-06, "loss": 2.9327, "step": 328970 }, { "epoch": 0.0081408, "grad_norm": 1.0467487573623657, "learning_rate": 1.207076599465533e-06, "loss": 2.8633, "step": 328980 }, { "epoch": 0.0081664, "grad_norm": 0.888003408908844, "learning_rate": 1.20669308796324e-06, "loss": 2.5344, "step": 328990 }, { "epoch": 0.008192, "grad_norm": 0.8626018762588501, "learning_rate": 1.2063096334832036e-06, "loss": 2.7, "step": 329000 }, { "epoch": 0.0082176, "grad_norm": 0.88724285364151, "learning_rate": 1.2059262360279123e-06, "loss": 2.8307, "step": 329010 }, { "epoch": 0.0082432, "grad_norm": 1.0803219079971313, "learning_rate": 1.2055428955998516e-06, "loss": 2.7638, "step": 329020 }, { "epoch": 0.0082688, "grad_norm": 0.9231474995613098, "learning_rate": 1.2051596122015074e-06, "loss": 3.0142, "step": 329030 }, { "epoch": 0.0082944, "grad_norm": 0.878800094127655, "learning_rate": 1.2047763858353666e-06, "loss": 2.8887, "step": 329040 }, { "epoch": 0.00832, "grad_norm": 0.9364063143730164, "learning_rate": 1.204393216503913e-06, "loss": 2.7907, "step": 329050 }, { "epoch": 0.0083456, "grad_norm": 1.0062973499298096, "learning_rate": 1.2040101042096341e-06, "loss": 2.6938, "step": 329060 }, { "epoch": 0.0083712, "grad_norm": 1.333996057510376, "learning_rate": 1.2036270489550094e-06, "loss": 2.8308, "step": 329070 }, { "epoch": 0.0083968, "grad_norm": 0.7939886450767517, "learning_rate": 1.203244050742527e-06, "loss": 2.6772, "step": 329080 }, { "epoch": 0.0084224, "grad_norm": 0.775580108165741, "learning_rate": 1.2028611095746678e-06, "loss": 2.7089, "step": 329090 }, { "epoch": 0.008448, "grad_norm": 0.7616184949874878, "learning_rate": 1.2024782254539159e-06, "loss": 2.79, "step": 329100 }, { "epoch": 0.0084736, "grad_norm": 0.8585727214813232, "learning_rate": 1.2020953983827555e-06, "loss": 2.9828, "step": 329110 }, { "epoch": 0.0084992, "grad_norm": 1.578240990638733, "learning_rate": 1.2017126283636682e-06, "loss": 2.8544, "step": 329120 }, { "epoch": 0.0085248, "grad_norm": 1.0738821029663086, "learning_rate": 1.2013299153991353e-06, "loss": 2.6671, "step": 329130 }, { "epoch": 0.0085504, "grad_norm": 0.7951866388320923, "learning_rate": 1.2009472594916404e-06, "loss": 2.8133, "step": 329140 }, { "epoch": 0.008576, "grad_norm": 0.799802303314209, "learning_rate": 1.2005646606436638e-06, "loss": 2.7089, "step": 329150 }, { "epoch": 0.0086016, "grad_norm": 0.9923467040061951, "learning_rate": 1.2001821188576868e-06, "loss": 2.8921, "step": 329160 }, { "epoch": 0.0086272, "grad_norm": 0.8092100024223328, "learning_rate": 1.1997996341361907e-06, "loss": 2.9947, "step": 329170 }, { "epoch": 0.0086528, "grad_norm": 0.8607772588729858, "learning_rate": 1.1994172064816556e-06, "loss": 2.9047, "step": 329180 }, { "epoch": 0.0086784, "grad_norm": 0.7780535817146301, "learning_rate": 1.1990348358965597e-06, "loss": 2.7598, "step": 329190 }, { "epoch": 0.008704, "grad_norm": 0.7925164103507996, "learning_rate": 1.1986525223833845e-06, "loss": 2.7234, "step": 329200 }, { "epoch": 0.0087296, "grad_norm": 0.908281683921814, "learning_rate": 1.1982702659446078e-06, "loss": 2.7032, "step": 329210 }, { "epoch": 0.0087552, "grad_norm": 0.8410993814468384, "learning_rate": 1.1978880665827097e-06, "loss": 2.7109, "step": 329220 }, { "epoch": 0.0087808, "grad_norm": 0.8575204014778137, "learning_rate": 1.1975059243001686e-06, "loss": 2.7857, "step": 329230 }, { "epoch": 0.0088064, "grad_norm": 0.908356785774231, "learning_rate": 1.197123839099461e-06, "loss": 2.6919, "step": 329240 }, { "epoch": 0.008832, "grad_norm": 0.9934182167053223, "learning_rate": 1.1967418109830664e-06, "loss": 2.8196, "step": 329250 }, { "epoch": 0.0088576, "grad_norm": 0.8458428978919983, "learning_rate": 1.1963598399534637e-06, "loss": 2.808, "step": 329260 }, { "epoch": 0.0088832, "grad_norm": 0.7756670713424683, "learning_rate": 1.1959779260131255e-06, "loss": 2.7013, "step": 329270 }, { "epoch": 0.0089088, "grad_norm": 1.8834539651870728, "learning_rate": 1.1955960691645307e-06, "loss": 2.9749, "step": 329280 }, { "epoch": 0.0089344, "grad_norm": 0.8695350885391235, "learning_rate": 1.1952142694101533e-06, "loss": 2.7451, "step": 329290 }, { "epoch": 0.00896, "grad_norm": 0.7732271552085876, "learning_rate": 1.194832526752473e-06, "loss": 2.8215, "step": 329300 }, { "epoch": 0.0089856, "grad_norm": 0.8194432258605957, "learning_rate": 1.194450841193966e-06, "loss": 2.8736, "step": 329310 }, { "epoch": 0.0090112, "grad_norm": 0.90571528673172, "learning_rate": 1.1940692127371034e-06, "loss": 2.8001, "step": 329320 }, { "epoch": 0.0090368, "grad_norm": 0.9790211915969849, "learning_rate": 1.1936876413843612e-06, "loss": 2.9116, "step": 329330 }, { "epoch": 0.0090624, "grad_norm": 0.9487351775169373, "learning_rate": 1.193306127138214e-06, "loss": 2.7906, "step": 329340 }, { "epoch": 0.009088, "grad_norm": 0.7782284617424011, "learning_rate": 1.1929246700011365e-06, "loss": 2.8577, "step": 329350 }, { "epoch": 0.0091136, "grad_norm": 0.9643221497535706, "learning_rate": 1.1925432699756013e-06, "loss": 2.9061, "step": 329360 }, { "epoch": 0.0091392, "grad_norm": 0.8269772529602051, "learning_rate": 1.1921619270640828e-06, "loss": 2.7588, "step": 329370 }, { "epoch": 0.0091648, "grad_norm": 0.8354275822639465, "learning_rate": 1.1917806412690557e-06, "loss": 2.7818, "step": 329380 }, { "epoch": 0.0091904, "grad_norm": 0.7445493936538696, "learning_rate": 1.1913994125929874e-06, "loss": 2.9137, "step": 329390 }, { "epoch": 0.009216, "grad_norm": 0.8906376957893372, "learning_rate": 1.1910182410383531e-06, "loss": 2.638, "step": 329400 }, { "epoch": 0.0092416, "grad_norm": 0.9737378358840942, "learning_rate": 1.1906371266076245e-06, "loss": 2.9095, "step": 329410 }, { "epoch": 0.0092672, "grad_norm": 0.7975066304206848, "learning_rate": 1.1902560693032728e-06, "loss": 2.85, "step": 329420 }, { "epoch": 0.0092928, "grad_norm": 0.8595029711723328, "learning_rate": 1.1898750691277693e-06, "loss": 2.7865, "step": 329430 }, { "epoch": 0.0093184, "grad_norm": 0.9029676914215088, "learning_rate": 1.1894941260835836e-06, "loss": 2.7999, "step": 329440 }, { "epoch": 0.009344, "grad_norm": 1.0101336240768433, "learning_rate": 1.1891132401731887e-06, "loss": 2.9655, "step": 329450 }, { "epoch": 0.0093696, "grad_norm": 0.7376312017440796, "learning_rate": 1.1887324113990528e-06, "loss": 2.7897, "step": 329460 }, { "epoch": 0.0093952, "grad_norm": 0.8385326266288757, "learning_rate": 1.188351639763643e-06, "loss": 2.8133, "step": 329470 }, { "epoch": 0.0094208, "grad_norm": 0.8583812713623047, "learning_rate": 1.1879709252694327e-06, "loss": 2.8969, "step": 329480 }, { "epoch": 0.0094464, "grad_norm": 0.9874745011329651, "learning_rate": 1.1875902679188878e-06, "loss": 2.9389, "step": 329490 }, { "epoch": 0.009472, "grad_norm": 0.81268310546875, "learning_rate": 1.1872096677144796e-06, "loss": 2.8015, "step": 329500 }, { "epoch": 0.0094976, "grad_norm": 0.9834898114204407, "learning_rate": 1.186829124658676e-06, "loss": 2.7495, "step": 329510 }, { "epoch": 0.0095232, "grad_norm": 0.8004642724990845, "learning_rate": 1.1864486387539409e-06, "loss": 2.8192, "step": 329520 }, { "epoch": 0.0095488, "grad_norm": 0.940470814704895, "learning_rate": 1.1860682100027442e-06, "loss": 2.8208, "step": 329530 }, { "epoch": 0.0095744, "grad_norm": 0.8145956993103027, "learning_rate": 1.185687838407552e-06, "loss": 2.9312, "step": 329540 }, { "epoch": 0.0096, "grad_norm": 0.9028474688529968, "learning_rate": 1.1853075239708323e-06, "loss": 2.9588, "step": 329550 }, { "epoch": 0.0096256, "grad_norm": 1.0560225248336792, "learning_rate": 1.1849272666950506e-06, "loss": 2.868, "step": 329560 }, { "epoch": 0.0096512, "grad_norm": 0.8559850454330444, "learning_rate": 1.184547066582673e-06, "loss": 2.8774, "step": 329570 }, { "epoch": 0.0096768, "grad_norm": 0.7852931618690491, "learning_rate": 1.1841669236361675e-06, "loss": 2.8222, "step": 329580 }, { "epoch": 0.0097024, "grad_norm": 0.8773791193962097, "learning_rate": 1.1837868378579931e-06, "loss": 2.851, "step": 329590 }, { "epoch": 0.009728, "grad_norm": 0.8702939748764038, "learning_rate": 1.183406809250619e-06, "loss": 2.8247, "step": 329600 }, { "epoch": 0.0097536, "grad_norm": 0.8033151626586914, "learning_rate": 1.1830268378165078e-06, "loss": 2.9579, "step": 329610 }, { "epoch": 0.0097792, "grad_norm": 0.8410279154777527, "learning_rate": 1.182646923558125e-06, "loss": 2.7825, "step": 329620 }, { "epoch": 0.0098048, "grad_norm": 0.8377613425254822, "learning_rate": 1.1822670664779323e-06, "loss": 3.0952, "step": 329630 }, { "epoch": 0.0098304, "grad_norm": 0.9317438006401062, "learning_rate": 1.1818872665783954e-06, "loss": 2.7383, "step": 329640 }, { "epoch": 0.009856, "grad_norm": 0.9597910642623901, "learning_rate": 1.1815075238619756e-06, "loss": 2.7144, "step": 329650 }, { "epoch": 0.0098816, "grad_norm": 0.9303643703460693, "learning_rate": 1.1811278383311365e-06, "loss": 2.9216, "step": 329660 }, { "epoch": 0.0099072, "grad_norm": 0.9104980826377869, "learning_rate": 1.1807482099883382e-06, "loss": 2.8167, "step": 329670 }, { "epoch": 0.0099328, "grad_norm": 0.8481098413467407, "learning_rate": 1.1803686388360447e-06, "loss": 2.8775, "step": 329680 }, { "epoch": 0.0099584, "grad_norm": 0.8395742177963257, "learning_rate": 1.179989124876717e-06, "loss": 2.5789, "step": 329690 }, { "epoch": 0.009984, "grad_norm": 1.1911392211914062, "learning_rate": 1.1796096681128155e-06, "loss": 2.804, "step": 329700 }, { "epoch": 0.0100096, "grad_norm": 0.7635690569877625, "learning_rate": 1.1792302685468027e-06, "loss": 2.9784, "step": 329710 }, { "epoch": 0.0100352, "grad_norm": 0.9283748865127563, "learning_rate": 1.1788509261811354e-06, "loss": 2.9015, "step": 329720 }, { "epoch": 0.0100608, "grad_norm": 0.9192398190498352, "learning_rate": 1.1784716410182762e-06, "loss": 2.9408, "step": 329730 }, { "epoch": 0.0100864, "grad_norm": 0.7622689604759216, "learning_rate": 1.178092413060683e-06, "loss": 2.7158, "step": 329740 }, { "epoch": 0.010112, "grad_norm": 1.0522152185440063, "learning_rate": 1.1777132423108161e-06, "loss": 2.7906, "step": 329750 }, { "epoch": 0.0101376, "grad_norm": 1.171635389328003, "learning_rate": 1.1773341287711348e-06, "loss": 2.843, "step": 329760 }, { "epoch": 0.0101632, "grad_norm": 1.0708742141723633, "learning_rate": 1.176955072444096e-06, "loss": 2.8949, "step": 329770 }, { "epoch": 0.0101888, "grad_norm": 0.7755547761917114, "learning_rate": 1.1765760733321608e-06, "loss": 2.6631, "step": 329780 }, { "epoch": 0.0102144, "grad_norm": 0.9197796583175659, "learning_rate": 1.176197131437783e-06, "loss": 2.9826, "step": 329790 }, { "epoch": 0.01024, "grad_norm": 0.9239306449890137, "learning_rate": 1.1758182467634227e-06, "loss": 3.0752, "step": 329800 }, { "epoch": 0.0102656, "grad_norm": 0.8285276293754578, "learning_rate": 1.1754394193115349e-06, "loss": 2.8193, "step": 329810 }, { "epoch": 0.0102912, "grad_norm": 0.859909176826477, "learning_rate": 1.175060649084575e-06, "loss": 2.8811, "step": 329820 }, { "epoch": 0.0103168, "grad_norm": 1.1517380475997925, "learning_rate": 1.1746819360850036e-06, "loss": 2.9738, "step": 329830 }, { "epoch": 0.0103424, "grad_norm": 0.8854092359542847, "learning_rate": 1.1743032803152764e-06, "loss": 3.0828, "step": 329840 }, { "epoch": 0.010368, "grad_norm": 0.8630201816558838, "learning_rate": 1.173924681777845e-06, "loss": 3.0403, "step": 329850 }, { "epoch": 0.0103936, "grad_norm": 0.8184930086135864, "learning_rate": 1.1735461404751657e-06, "loss": 2.8588, "step": 329860 }, { "epoch": 0.0104192, "grad_norm": 0.7836662530899048, "learning_rate": 1.173167656409695e-06, "loss": 2.9961, "step": 329870 }, { "epoch": 0.0104448, "grad_norm": 1.1433701515197754, "learning_rate": 1.172789229583885e-06, "loss": 2.6695, "step": 329880 }, { "epoch": 0.0104704, "grad_norm": 0.8170470595359802, "learning_rate": 1.1724108600001916e-06, "loss": 2.8303, "step": 329890 }, { "epoch": 0.010496, "grad_norm": 0.8603324294090271, "learning_rate": 1.1720325476610673e-06, "loss": 2.9774, "step": 329900 }, { "epoch": 0.0105216, "grad_norm": 0.7624418139457703, "learning_rate": 1.1716542925689679e-06, "loss": 2.7688, "step": 329910 }, { "epoch": 0.0105472, "grad_norm": 0.8556450009346008, "learning_rate": 1.1712760947263424e-06, "loss": 2.8411, "step": 329920 }, { "epoch": 0.0105728, "grad_norm": 1.730546474456787, "learning_rate": 1.1708979541356446e-06, "loss": 2.64, "step": 329930 }, { "epoch": 0.0105984, "grad_norm": 0.791110634803772, "learning_rate": 1.1705198707993271e-06, "loss": 2.7475, "step": 329940 }, { "epoch": 0.010624, "grad_norm": 0.8715051412582397, "learning_rate": 1.170141844719842e-06, "loss": 2.7757, "step": 329950 }, { "epoch": 0.0106496, "grad_norm": 1.4671348333358765, "learning_rate": 1.16976387589964e-06, "loss": 3.1339, "step": 329960 }, { "epoch": 0.0106752, "grad_norm": 0.8370621800422668, "learning_rate": 1.169385964341172e-06, "loss": 2.8536, "step": 329970 }, { "epoch": 0.0107008, "grad_norm": 0.835284948348999, "learning_rate": 1.1690081100468919e-06, "loss": 3.2435, "step": 329980 }, { "epoch": 0.0107264, "grad_norm": 2.53303861618042, "learning_rate": 1.1686303130192434e-06, "loss": 2.8392, "step": 329990 }, { "epoch": 0.010752, "grad_norm": 0.8463605046272278, "learning_rate": 1.168252573260682e-06, "loss": 2.8458, "step": 330000 }, { "epoch": 0.0107776, "grad_norm": 0.9002472162246704, "learning_rate": 1.1678748907736547e-06, "loss": 3.0085, "step": 330010 }, { "epoch": 0.0108032, "grad_norm": 1.2517374753952026, "learning_rate": 1.1674972655606121e-06, "loss": 2.7608, "step": 330020 }, { "epoch": 0.0108288, "grad_norm": 0.8093563318252563, "learning_rate": 1.1671196976240028e-06, "loss": 2.9849, "step": 330030 }, { "epoch": 0.0108544, "grad_norm": 1.1335362195968628, "learning_rate": 1.1667421869662765e-06, "loss": 2.89, "step": 330040 }, { "epoch": 0.01088, "grad_norm": 0.7538942098617554, "learning_rate": 1.1663647335898765e-06, "loss": 2.7214, "step": 330050 }, { "epoch": 0.0109056, "grad_norm": 0.887220561504364, "learning_rate": 1.1659873374972541e-06, "loss": 2.692, "step": 330060 }, { "epoch": 0.0109312, "grad_norm": 0.8263095617294312, "learning_rate": 1.1656099986908565e-06, "loss": 2.9632, "step": 330070 }, { "epoch": 0.0109568, "grad_norm": 0.8838387727737427, "learning_rate": 1.1652327171731304e-06, "loss": 2.8016, "step": 330080 }, { "epoch": 0.0109824, "grad_norm": 0.8176586031913757, "learning_rate": 1.1648554929465216e-06, "loss": 2.7982, "step": 330090 }, { "epoch": 0.011008, "grad_norm": 0.7834455966949463, "learning_rate": 1.1644783260134763e-06, "loss": 2.8096, "step": 330100 }, { "epoch": 0.0110336, "grad_norm": 0.9187942147254944, "learning_rate": 1.1641012163764432e-06, "loss": 2.8957, "step": 330110 }, { "epoch": 0.0110592, "grad_norm": 0.7923718690872192, "learning_rate": 1.1637241640378638e-06, "loss": 2.6992, "step": 330120 }, { "epoch": 0.0110848, "grad_norm": 0.8202451467514038, "learning_rate": 1.163347169000184e-06, "loss": 2.7278, "step": 330130 }, { "epoch": 0.0111104, "grad_norm": 0.7916537523269653, "learning_rate": 1.1629702312658509e-06, "loss": 2.8215, "step": 330140 }, { "epoch": 0.011136, "grad_norm": 0.849260151386261, "learning_rate": 1.1625933508373055e-06, "loss": 2.904, "step": 330150 }, { "epoch": 0.0111616, "grad_norm": 0.8531169295310974, "learning_rate": 1.1622165277169928e-06, "loss": 2.7934, "step": 330160 }, { "epoch": 0.0111872, "grad_norm": 0.8305139541625977, "learning_rate": 1.1618397619073607e-06, "loss": 2.8531, "step": 330170 }, { "epoch": 0.0112128, "grad_norm": 0.7889319062232971, "learning_rate": 1.1614630534108463e-06, "loss": 2.7909, "step": 330180 }, { "epoch": 0.0112384, "grad_norm": 0.7884899377822876, "learning_rate": 1.1610864022298951e-06, "loss": 2.7996, "step": 330190 }, { "epoch": 0.011264, "grad_norm": 1.3211662769317627, "learning_rate": 1.1607098083669499e-06, "loss": 2.5005, "step": 330200 }, { "epoch": 0.0112896, "grad_norm": 0.9207274913787842, "learning_rate": 1.160333271824452e-06, "loss": 2.8912, "step": 330210 }, { "epoch": 0.0113152, "grad_norm": 0.805442750453949, "learning_rate": 1.1599567926048439e-06, "loss": 2.9319, "step": 330220 }, { "epoch": 0.0113408, "grad_norm": 0.866060733795166, "learning_rate": 1.1595803707105669e-06, "loss": 2.8599, "step": 330230 }, { "epoch": 0.0113664, "grad_norm": 0.8094910383224487, "learning_rate": 1.1592040061440636e-06, "loss": 2.9934, "step": 330240 }, { "epoch": 0.011392, "grad_norm": 0.8291721343994141, "learning_rate": 1.1588276989077696e-06, "loss": 2.799, "step": 330250 }, { "epoch": 0.0114176, "grad_norm": 1.0221902132034302, "learning_rate": 1.1584514490041299e-06, "loss": 2.8099, "step": 330260 }, { "epoch": 0.0114432, "grad_norm": 0.8417811989784241, "learning_rate": 1.1580752564355812e-06, "loss": 2.9126, "step": 330270 }, { "epoch": 0.0114688, "grad_norm": 0.9884907603263855, "learning_rate": 1.1576991212045653e-06, "loss": 2.7792, "step": 330280 }, { "epoch": 0.0114944, "grad_norm": 0.9058910608291626, "learning_rate": 1.1573230433135207e-06, "loss": 2.8889, "step": 330290 }, { "epoch": 0.01152, "grad_norm": 1.1608659029006958, "learning_rate": 1.156947022764886e-06, "loss": 2.9139, "step": 330300 }, { "epoch": 0.0115456, "grad_norm": 0.8630930781364441, "learning_rate": 1.1565710595611013e-06, "loss": 2.8307, "step": 330310 }, { "epoch": 0.0115712, "grad_norm": 0.8714512586593628, "learning_rate": 1.1561951537046013e-06, "loss": 2.9208, "step": 330320 }, { "epoch": 0.0115968, "grad_norm": 0.7845183610916138, "learning_rate": 1.155819305197826e-06, "loss": 2.7859, "step": 330330 }, { "epoch": 0.0116224, "grad_norm": 0.845212996006012, "learning_rate": 1.1554435140432085e-06, "loss": 2.832, "step": 330340 }, { "epoch": 0.011648, "grad_norm": 0.8159676790237427, "learning_rate": 1.155067780243192e-06, "loss": 2.7908, "step": 330350 }, { "epoch": 0.0116736, "grad_norm": 1.0962954759597778, "learning_rate": 1.1546921038002102e-06, "loss": 2.7089, "step": 330360 }, { "epoch": 0.0116992, "grad_norm": 0.8436718583106995, "learning_rate": 1.154316484716702e-06, "loss": 2.8655, "step": 330370 }, { "epoch": 0.0117248, "grad_norm": 0.9161363244056702, "learning_rate": 1.1539409229950982e-06, "loss": 2.7414, "step": 330380 }, { "epoch": 0.0117504, "grad_norm": 0.8543878793716431, "learning_rate": 1.1535654186378353e-06, "loss": 3.003, "step": 330390 }, { "epoch": 0.011776, "grad_norm": 0.77803635597229, "learning_rate": 1.1531899716473504e-06, "loss": 2.6777, "step": 330400 }, { "epoch": 0.0118016, "grad_norm": 0.8120158910751343, "learning_rate": 1.1528145820260771e-06, "loss": 3.0277, "step": 330410 }, { "epoch": 0.0118272, "grad_norm": 0.8108019232749939, "learning_rate": 1.1524392497764513e-06, "loss": 2.8718, "step": 330420 }, { "epoch": 0.0118528, "grad_norm": 0.9020136594772339, "learning_rate": 1.1520639749009044e-06, "loss": 2.6671, "step": 330430 }, { "epoch": 0.0118784, "grad_norm": 0.8385202884674072, "learning_rate": 1.1516887574018732e-06, "loss": 2.9367, "step": 330440 }, { "epoch": 0.011904, "grad_norm": 0.8531653881072998, "learning_rate": 1.1513135972817869e-06, "loss": 2.7897, "step": 330450 }, { "epoch": 0.0119296, "grad_norm": 0.8273791670799255, "learning_rate": 1.1509384945430802e-06, "loss": 2.9253, "step": 330460 }, { "epoch": 0.0119552, "grad_norm": 0.7299277782440186, "learning_rate": 1.1505634491881846e-06, "loss": 2.8112, "step": 330470 }, { "epoch": 0.0119808, "grad_norm": 0.8949188590049744, "learning_rate": 1.1501884612195346e-06, "loss": 2.8389, "step": 330480 }, { "epoch": 0.0120064, "grad_norm": 0.7539597749710083, "learning_rate": 1.1498135306395598e-06, "loss": 2.5659, "step": 330490 }, { "epoch": 0.012032, "grad_norm": 0.7717450261116028, "learning_rate": 1.1494386574506921e-06, "loss": 2.719, "step": 330500 }, { "epoch": 0.0120576, "grad_norm": 0.7624507546424866, "learning_rate": 1.1490638416553623e-06, "loss": 2.8043, "step": 330510 }, { "epoch": 0.0120832, "grad_norm": 0.7684388756752014, "learning_rate": 1.1486890832560015e-06, "loss": 2.5739, "step": 330520 }, { "epoch": 0.0121088, "grad_norm": 0.8061183094978333, "learning_rate": 1.1483143822550403e-06, "loss": 2.8352, "step": 330530 }, { "epoch": 0.0121344, "grad_norm": 0.8289473652839661, "learning_rate": 1.1479397386549062e-06, "loss": 2.873, "step": 330540 }, { "epoch": 0.01216, "grad_norm": 0.8007058501243591, "learning_rate": 1.1475651524580323e-06, "loss": 2.7372, "step": 330550 }, { "epoch": 0.0121856, "grad_norm": 0.9378123879432678, "learning_rate": 1.147190623666844e-06, "loss": 2.884, "step": 330560 }, { "epoch": 0.0122112, "grad_norm": 0.9024885892868042, "learning_rate": 1.146816152283774e-06, "loss": 2.8608, "step": 330570 }, { "epoch": 0.0122368, "grad_norm": 0.8551216721534729, "learning_rate": 1.1464417383112469e-06, "loss": 2.7851, "step": 330580 }, { "epoch": 0.0122624, "grad_norm": 0.7845830321311951, "learning_rate": 1.146067381751692e-06, "loss": 2.9973, "step": 330590 }, { "epoch": 0.012288, "grad_norm": 0.870467483997345, "learning_rate": 1.1456930826075373e-06, "loss": 3.0647, "step": 330600 }, { "epoch": 0.0123136, "grad_norm": 0.7955492734909058, "learning_rate": 1.1453188408812088e-06, "loss": 2.786, "step": 330610 }, { "epoch": 0.0123392, "grad_norm": 0.9731240272521973, "learning_rate": 1.1449446565751354e-06, "loss": 2.8617, "step": 330620 }, { "epoch": 0.0123648, "grad_norm": 0.8798343539237976, "learning_rate": 1.144570529691742e-06, "loss": 2.8154, "step": 330630 }, { "epoch": 0.0123904, "grad_norm": 0.8613734245300293, "learning_rate": 1.1441964602334576e-06, "loss": 2.7908, "step": 330640 }, { "epoch": 0.012416, "grad_norm": 0.9080227017402649, "learning_rate": 1.1438224482027027e-06, "loss": 2.8164, "step": 330650 }, { "epoch": 0.0124416, "grad_norm": 0.8800355792045593, "learning_rate": 1.1434484936019065e-06, "loss": 2.9216, "step": 330660 }, { "epoch": 0.0124672, "grad_norm": 0.8683156967163086, "learning_rate": 1.1430745964334923e-06, "loss": 2.6499, "step": 330670 }, { "epoch": 0.0124928, "grad_norm": 0.8262485861778259, "learning_rate": 1.142700756699886e-06, "loss": 2.8211, "step": 330680 }, { "epoch": 0.0125184, "grad_norm": 0.9830707907676697, "learning_rate": 1.1423269744035115e-06, "loss": 2.9385, "step": 330690 }, { "epoch": 0.012544, "grad_norm": 0.7904119491577148, "learning_rate": 1.1419532495467921e-06, "loss": 2.8542, "step": 330700 }, { "epoch": 0.0125696, "grad_norm": 0.7862988710403442, "learning_rate": 1.1415795821321507e-06, "loss": 2.8923, "step": 330710 }, { "epoch": 0.0125952, "grad_norm": 0.8198416829109192, "learning_rate": 1.1412059721620129e-06, "loss": 2.6141, "step": 330720 }, { "epoch": 0.0126208, "grad_norm": 0.8944567441940308, "learning_rate": 1.1408324196388e-06, "loss": 2.7921, "step": 330730 }, { "epoch": 0.0126464, "grad_norm": 0.7779932618141174, "learning_rate": 1.1404589245649333e-06, "loss": 2.7599, "step": 330740 }, { "epoch": 0.012672, "grad_norm": 0.7660470008850098, "learning_rate": 1.1400854869428368e-06, "loss": 2.7536, "step": 330750 }, { "epoch": 0.0126976, "grad_norm": 0.7995249629020691, "learning_rate": 1.1397121067749307e-06, "loss": 2.8491, "step": 330760 }, { "epoch": 0.0127232, "grad_norm": 1.0068796873092651, "learning_rate": 1.1393387840636383e-06, "loss": 2.8307, "step": 330770 }, { "epoch": 0.0127488, "grad_norm": 0.7909981608390808, "learning_rate": 1.138965518811378e-06, "loss": 2.8398, "step": 330780 }, { "epoch": 0.0127744, "grad_norm": 0.9649229645729065, "learning_rate": 1.138592311020571e-06, "loss": 2.856, "step": 330790 }, { "epoch": 0.0128, "grad_norm": 0.9318748712539673, "learning_rate": 1.1382191606936376e-06, "loss": 2.8377, "step": 330800 }, { "epoch": 0.0128256, "grad_norm": 0.7859213352203369, "learning_rate": 1.1378460678329983e-06, "loss": 2.7916, "step": 330810 }, { "epoch": 0.0128512, "grad_norm": 1.041015386581421, "learning_rate": 1.1374730324410722e-06, "loss": 2.6596, "step": 330820 }, { "epoch": 0.0128768, "grad_norm": 0.7620820999145508, "learning_rate": 1.1371000545202771e-06, "loss": 2.7823, "step": 330830 }, { "epoch": 0.0129024, "grad_norm": 0.91440349817276, "learning_rate": 1.1367271340730346e-06, "loss": 2.7882, "step": 330840 }, { "epoch": 0.012928, "grad_norm": 1.0761454105377197, "learning_rate": 1.1363542711017594e-06, "loss": 2.6645, "step": 330850 }, { "epoch": 0.0129536, "grad_norm": 0.7627410888671875, "learning_rate": 1.1359814656088697e-06, "loss": 2.5955, "step": 330860 }, { "epoch": 0.0129792, "grad_norm": 0.8717470169067383, "learning_rate": 1.1356087175967857e-06, "loss": 2.9016, "step": 330870 }, { "epoch": 0.0130048, "grad_norm": 0.8662941455841064, "learning_rate": 1.135236027067924e-06, "loss": 2.9172, "step": 330880 }, { "epoch": 0.0130304, "grad_norm": 0.8469918370246887, "learning_rate": 1.1348633940247021e-06, "loss": 2.8621, "step": 330890 }, { "epoch": 0.013056, "grad_norm": 0.8451079726219177, "learning_rate": 1.1344908184695335e-06, "loss": 2.9821, "step": 330900 }, { "epoch": 0.0130816, "grad_norm": 0.8995653390884399, "learning_rate": 1.1341183004048351e-06, "loss": 2.8498, "step": 330910 }, { "epoch": 0.0131072, "grad_norm": 0.7415646314620972, "learning_rate": 1.1337458398330236e-06, "loss": 2.8207, "step": 330920 }, { "epoch": 0.0131328, "grad_norm": 1.0482572317123413, "learning_rate": 1.1333734367565141e-06, "loss": 2.8997, "step": 330930 }, { "epoch": 0.0131584, "grad_norm": 1.544735074043274, "learning_rate": 1.133001091177721e-06, "loss": 2.7922, "step": 330940 }, { "epoch": 0.013184, "grad_norm": 0.7957931160926819, "learning_rate": 1.1326288030990596e-06, "loss": 2.7376, "step": 330950 }, { "epoch": 0.0132096, "grad_norm": 0.7890328764915466, "learning_rate": 1.132256572522944e-06, "loss": 2.7902, "step": 330960 }, { "epoch": 0.0132352, "grad_norm": 0.812735915184021, "learning_rate": 1.1318843994517891e-06, "loss": 2.8308, "step": 330970 }, { "epoch": 0.0132608, "grad_norm": 0.963962733745575, "learning_rate": 1.1315122838880055e-06, "loss": 3.0801, "step": 330980 }, { "epoch": 0.0132864, "grad_norm": 0.8211332559585571, "learning_rate": 1.1311402258340077e-06, "loss": 2.8079, "step": 330990 }, { "epoch": 0.013312, "grad_norm": 0.7817907333374023, "learning_rate": 1.130768225292208e-06, "loss": 2.4976, "step": 331000 }, { "epoch": 0.0133376, "grad_norm": 0.8764006495475769, "learning_rate": 1.1303962822650193e-06, "loss": 2.7753, "step": 331010 }, { "epoch": 0.0133632, "grad_norm": 0.8490604758262634, "learning_rate": 1.1300243967548541e-06, "loss": 2.8132, "step": 331020 }, { "epoch": 0.0133888, "grad_norm": 1.2526394128799438, "learning_rate": 1.1296525687641225e-06, "loss": 2.6243, "step": 331030 }, { "epoch": 0.0134144, "grad_norm": 0.8257054090499878, "learning_rate": 1.1292807982952369e-06, "loss": 2.7965, "step": 331040 }, { "epoch": 0.01344, "grad_norm": 2.404508590698242, "learning_rate": 1.128909085350608e-06, "loss": 3.1124, "step": 331050 }, { "epoch": 0.0134656, "grad_norm": 1.0382128953933716, "learning_rate": 1.1285374299326458e-06, "loss": 2.8718, "step": 331060 }, { "epoch": 0.0134912, "grad_norm": 0.7972152829170227, "learning_rate": 1.1281658320437605e-06, "loss": 2.7941, "step": 331070 }, { "epoch": 0.0135168, "grad_norm": 0.9159799814224243, "learning_rate": 1.1277942916863626e-06, "loss": 2.8104, "step": 331080 }, { "epoch": 0.0135424, "grad_norm": 0.9709067344665527, "learning_rate": 1.1274228088628624e-06, "loss": 2.85, "step": 331090 }, { "epoch": 0.013568, "grad_norm": 0.8682013750076294, "learning_rate": 1.1270513835756647e-06, "loss": 2.7123, "step": 331100 }, { "epoch": 0.0135936, "grad_norm": 0.8260974884033203, "learning_rate": 1.1266800158271819e-06, "loss": 2.6925, "step": 331110 }, { "epoch": 0.0136192, "grad_norm": 0.7887282967567444, "learning_rate": 1.1263087056198196e-06, "loss": 2.7473, "step": 331120 }, { "epoch": 0.0136448, "grad_norm": 0.7700785398483276, "learning_rate": 1.1259374529559874e-06, "loss": 2.6467, "step": 331130 }, { "epoch": 0.0136704, "grad_norm": 0.9567588567733765, "learning_rate": 1.125566257838092e-06, "loss": 2.8546, "step": 331140 }, { "epoch": 0.013696, "grad_norm": 0.8027729988098145, "learning_rate": 1.1251951202685418e-06, "loss": 3.2998, "step": 331150 }, { "epoch": 0.0137216, "grad_norm": 0.8117069005966187, "learning_rate": 1.1248240402497445e-06, "loss": 3.0162, "step": 331160 }, { "epoch": 0.0137472, "grad_norm": 0.8054379224777222, "learning_rate": 1.1244530177841017e-06, "loss": 2.7581, "step": 331170 }, { "epoch": 0.0137728, "grad_norm": 1.6469578742980957, "learning_rate": 1.1240820528740227e-06, "loss": 2.8497, "step": 331180 }, { "epoch": 0.0137984, "grad_norm": 0.828661322593689, "learning_rate": 1.123711145521913e-06, "loss": 2.8388, "step": 331190 }, { "epoch": 0.013824, "grad_norm": 0.9062658548355103, "learning_rate": 1.1233402957301776e-06, "loss": 2.8056, "step": 331200 }, { "epoch": 0.0138496, "grad_norm": 1.062879204750061, "learning_rate": 1.122969503501219e-06, "loss": 2.7163, "step": 331210 }, { "epoch": 0.0138752, "grad_norm": 0.7946019172668457, "learning_rate": 1.1225987688374473e-06, "loss": 2.9744, "step": 331220 }, { "epoch": 0.0139008, "grad_norm": 0.9640782475471497, "learning_rate": 1.122228091741262e-06, "loss": 2.9673, "step": 331230 }, { "epoch": 0.0139264, "grad_norm": 0.7920578122138977, "learning_rate": 1.1218574722150677e-06, "loss": 2.5468, "step": 331240 }, { "epoch": 0.013952, "grad_norm": 0.8144012689590454, "learning_rate": 1.121486910261268e-06, "loss": 2.7513, "step": 331250 }, { "epoch": 0.0139776, "grad_norm": 0.8767343163490295, "learning_rate": 1.1211164058822666e-06, "loss": 2.9778, "step": 331260 }, { "epoch": 0.0140032, "grad_norm": 0.7685823440551758, "learning_rate": 1.1207459590804659e-06, "loss": 2.8193, "step": 331270 }, { "epoch": 0.0140288, "grad_norm": 0.8031609058380127, "learning_rate": 1.1203755698582664e-06, "loss": 2.781, "step": 331280 }, { "epoch": 0.0140544, "grad_norm": 0.8158103823661804, "learning_rate": 1.1200052382180748e-06, "loss": 2.752, "step": 331290 }, { "epoch": 0.01408, "grad_norm": 0.909130871295929, "learning_rate": 1.119634964162286e-06, "loss": 2.4972, "step": 331300 }, { "epoch": 0.0141056, "grad_norm": 0.845534086227417, "learning_rate": 1.1192647476933061e-06, "loss": 2.7235, "step": 331310 }, { "epoch": 0.0141312, "grad_norm": 0.8721428513526917, "learning_rate": 1.1188945888135328e-06, "loss": 2.8652, "step": 331320 }, { "epoch": 0.0141568, "grad_norm": 0.9903003573417664, "learning_rate": 1.1185244875253675e-06, "loss": 2.92, "step": 331330 }, { "epoch": 0.0141824, "grad_norm": 1.1775810718536377, "learning_rate": 1.1181544438312108e-06, "loss": 2.7489, "step": 331340 }, { "epoch": 0.014208, "grad_norm": 0.9116854071617126, "learning_rate": 1.117784457733463e-06, "loss": 2.8763, "step": 331350 }, { "epoch": 0.0142336, "grad_norm": 0.8381370902061462, "learning_rate": 1.1174145292345228e-06, "loss": 2.7391, "step": 331360 }, { "epoch": 0.0142592, "grad_norm": 1.0514017343521118, "learning_rate": 1.1170446583367878e-06, "loss": 2.6889, "step": 331370 }, { "epoch": 0.0142848, "grad_norm": 0.8492468595504761, "learning_rate": 1.1166748450426557e-06, "loss": 3.0346, "step": 331380 }, { "epoch": 0.0143104, "grad_norm": 0.9969774484634399, "learning_rate": 1.1163050893545269e-06, "loss": 2.6761, "step": 331390 }, { "epoch": 0.014336, "grad_norm": 0.8507728576660156, "learning_rate": 1.1159353912747994e-06, "loss": 2.7007, "step": 331400 }, { "epoch": 0.0143616, "grad_norm": 0.9687671065330505, "learning_rate": 1.1155657508058703e-06, "loss": 2.9468, "step": 331410 }, { "epoch": 0.0143872, "grad_norm": 0.881523072719574, "learning_rate": 1.1151961679501377e-06, "loss": 3.0254, "step": 331420 }, { "epoch": 0.0144128, "grad_norm": 0.844662070274353, "learning_rate": 1.114826642709994e-06, "loss": 2.9188, "step": 331430 }, { "epoch": 0.0144384, "grad_norm": 1.0990724563598633, "learning_rate": 1.1144571750878387e-06, "loss": 2.9209, "step": 331440 }, { "epoch": 0.014464, "grad_norm": 0.838254451751709, "learning_rate": 1.114087765086067e-06, "loss": 2.7568, "step": 331450 }, { "epoch": 0.0144896, "grad_norm": 0.8202803730964661, "learning_rate": 1.1137184127070744e-06, "loss": 2.584, "step": 331460 }, { "epoch": 0.0145152, "grad_norm": 0.7775232195854187, "learning_rate": 1.1133491179532564e-06, "loss": 2.7605, "step": 331470 }, { "epoch": 0.0145408, "grad_norm": 0.8334406018257141, "learning_rate": 1.1129798808270077e-06, "loss": 3.0169, "step": 331480 }, { "epoch": 0.0145664, "grad_norm": 0.7947962880134583, "learning_rate": 1.1126107013307252e-06, "loss": 2.7254, "step": 331490 }, { "epoch": 0.014592, "grad_norm": 0.8482659459114075, "learning_rate": 1.1122415794667974e-06, "loss": 2.7784, "step": 331500 }, { "epoch": 0.0146176, "grad_norm": 0.8130854964256287, "learning_rate": 1.111872515237621e-06, "loss": 2.6995, "step": 331510 }, { "epoch": 0.0146432, "grad_norm": 1.0878058671951294, "learning_rate": 1.1115035086455893e-06, "loss": 2.9171, "step": 331520 }, { "epoch": 0.0146688, "grad_norm": 1.0476422309875488, "learning_rate": 1.1111345596930944e-06, "loss": 2.787, "step": 331530 }, { "epoch": 0.0146944, "grad_norm": 0.8784618377685547, "learning_rate": 1.1107656683825307e-06, "loss": 2.8485, "step": 331540 }, { "epoch": 0.01472, "grad_norm": 0.7619056701660156, "learning_rate": 1.1103968347162885e-06, "loss": 2.5809, "step": 331550 }, { "epoch": 0.0147456, "grad_norm": 0.8974700570106506, "learning_rate": 1.1100280586967605e-06, "loss": 3.0018, "step": 331560 }, { "epoch": 0.0147712, "grad_norm": 0.8862720131874084, "learning_rate": 1.109659340326338e-06, "loss": 2.96, "step": 331570 }, { "epoch": 0.0147968, "grad_norm": 0.8348074555397034, "learning_rate": 1.109290679607411e-06, "loss": 2.8858, "step": 331580 }, { "epoch": 0.0148224, "grad_norm": 1.021723985671997, "learning_rate": 1.1089220765423714e-06, "loss": 2.8771, "step": 331590 }, { "epoch": 0.014848, "grad_norm": 0.8790994882583618, "learning_rate": 1.1085535311336092e-06, "loss": 2.7719, "step": 331600 }, { "epoch": 0.0148736, "grad_norm": 0.8349576592445374, "learning_rate": 1.108185043383515e-06, "loss": 2.6032, "step": 331610 }, { "epoch": 0.0148992, "grad_norm": 0.9362851977348328, "learning_rate": 1.107816613294479e-06, "loss": 3.0661, "step": 331620 }, { "epoch": 0.0149248, "grad_norm": 0.9059191346168518, "learning_rate": 1.1074482408688869e-06, "loss": 2.8476, "step": 331630 }, { "epoch": 0.0149504, "grad_norm": 0.9357216358184814, "learning_rate": 1.1070799261091302e-06, "loss": 2.866, "step": 331640 }, { "epoch": 0.014976, "grad_norm": 0.8082342147827148, "learning_rate": 1.106711669017596e-06, "loss": 2.7836, "step": 331650 }, { "epoch": 0.0150016, "grad_norm": 0.7559632658958435, "learning_rate": 1.1063434695966734e-06, "loss": 2.8365, "step": 331660 }, { "epoch": 0.0150272, "grad_norm": 0.814292311668396, "learning_rate": 1.1059753278487506e-06, "loss": 2.7082, "step": 331670 }, { "epoch": 0.0150528, "grad_norm": 0.7672348022460938, "learning_rate": 1.1056072437762133e-06, "loss": 2.7759, "step": 331680 }, { "epoch": 0.0150784, "grad_norm": 0.7762858271598816, "learning_rate": 1.1052392173814508e-06, "loss": 2.7626, "step": 331690 }, { "epoch": 0.015104, "grad_norm": 1.386572241783142, "learning_rate": 1.1048712486668467e-06, "loss": 2.7495, "step": 331700 }, { "epoch": 0.0151296, "grad_norm": 0.8376753926277161, "learning_rate": 1.104503337634788e-06, "loss": 2.7178, "step": 331710 }, { "epoch": 0.0151552, "grad_norm": 0.7991976737976074, "learning_rate": 1.1041354842876616e-06, "loss": 2.8606, "step": 331720 }, { "epoch": 0.0151808, "grad_norm": 0.7949508428573608, "learning_rate": 1.1037676886278492e-06, "loss": 2.6186, "step": 331730 }, { "epoch": 0.0152064, "grad_norm": 0.9976188540458679, "learning_rate": 1.103399950657742e-06, "loss": 2.746, "step": 331740 }, { "epoch": 0.015232, "grad_norm": 0.8162703514099121, "learning_rate": 1.1030322703797236e-06, "loss": 2.7074, "step": 331750 }, { "epoch": 0.0152576, "grad_norm": 0.8796396255493164, "learning_rate": 1.1026646477961744e-06, "loss": 3.0004, "step": 331760 }, { "epoch": 0.0152832, "grad_norm": 0.766542375087738, "learning_rate": 1.1022970829094803e-06, "loss": 3.0895, "step": 331770 }, { "epoch": 0.0153088, "grad_norm": 0.7992466688156128, "learning_rate": 1.1019295757220238e-06, "loss": 2.833, "step": 331780 }, { "epoch": 0.0153344, "grad_norm": 1.0237611532211304, "learning_rate": 1.101562126236191e-06, "loss": 2.6493, "step": 331790 }, { "epoch": 0.01536, "grad_norm": 0.958900511264801, "learning_rate": 1.1011947344543617e-06, "loss": 2.7188, "step": 331800 }, { "epoch": 2.56e-05, "grad_norm": 0.8311470150947571, "learning_rate": 1.10082740037892e-06, "loss": 1.7855, "step": 331810 }, { "epoch": 5.12e-05, "grad_norm": 0.7944139838218689, "learning_rate": 1.1004601240122492e-06, "loss": 2.7048, "step": 331820 }, { "epoch": 7.68e-05, "grad_norm": 0.7418505549430847, "learning_rate": 1.1000929053567278e-06, "loss": 2.5664, "step": 331830 }, { "epoch": 0.0001024, "grad_norm": 0.8464035987854004, "learning_rate": 1.099725744414738e-06, "loss": 2.5553, "step": 331840 }, { "epoch": 0.000128, "grad_norm": 0.9061664342880249, "learning_rate": 1.0993586411886614e-06, "loss": 2.6832, "step": 331850 }, { "epoch": 0.0001536, "grad_norm": 0.8853175044059753, "learning_rate": 1.0989915956808784e-06, "loss": 2.6022, "step": 331860 }, { "epoch": 0.0001792, "grad_norm": 0.806704580783844, "learning_rate": 1.09862460789377e-06, "loss": 2.5112, "step": 331870 }, { "epoch": 0.0002048, "grad_norm": 0.9358189105987549, "learning_rate": 1.098257677829715e-06, "loss": 2.78, "step": 331880 }, { "epoch": 0.0002304, "grad_norm": 0.8188964128494263, "learning_rate": 1.0978908054910952e-06, "loss": 2.8066, "step": 331890 }, { "epoch": 0.000256, "grad_norm": 0.7866132259368896, "learning_rate": 1.0975239908802827e-06, "loss": 2.7099, "step": 331900 }, { "epoch": 0.0002816, "grad_norm": 0.778681218624115, "learning_rate": 1.0971572339996639e-06, "loss": 2.4933, "step": 331910 }, { "epoch": 0.0003072, "grad_norm": 1.565395474433899, "learning_rate": 1.096790534851615e-06, "loss": 2.2537, "step": 331920 }, { "epoch": 0.0003328, "grad_norm": 0.8344776630401611, "learning_rate": 1.0964238934385119e-06, "loss": 2.591, "step": 331930 }, { "epoch": 0.0003584, "grad_norm": 0.8942021727561951, "learning_rate": 1.0960573097627348e-06, "loss": 2.6656, "step": 331940 }, { "epoch": 0.000384, "grad_norm": 0.8559070229530334, "learning_rate": 1.0956907838266606e-06, "loss": 2.6943, "step": 331950 }, { "epoch": 0.0004096, "grad_norm": 0.8644915223121643, "learning_rate": 1.0953243156326643e-06, "loss": 2.8208, "step": 331960 }, { "epoch": 0.0004352, "grad_norm": 0.7716090679168701, "learning_rate": 1.0949579051831228e-06, "loss": 2.7241, "step": 331970 }, { "epoch": 0.0004608, "grad_norm": 1.052997350692749, "learning_rate": 1.0945915524804118e-06, "loss": 2.5555, "step": 331980 }, { "epoch": 0.0004864, "grad_norm": 0.8278565406799316, "learning_rate": 1.0942252575269086e-06, "loss": 2.5283, "step": 331990 }, { "epoch": 0.000512, "grad_norm": 1.0613718032836914, "learning_rate": 1.0938590203249877e-06, "loss": 2.4913, "step": 332000 }, { "epoch": 0.0005376, "grad_norm": 0.8595008254051208, "learning_rate": 1.093492840877024e-06, "loss": 2.6047, "step": 332010 }, { "epoch": 0.0005632, "grad_norm": 0.8125053644180298, "learning_rate": 1.0931267191853946e-06, "loss": 2.6409, "step": 332020 }, { "epoch": 0.0005888, "grad_norm": 0.843475341796875, "learning_rate": 1.0927606552524684e-06, "loss": 2.714, "step": 332030 }, { "epoch": 0.0006144, "grad_norm": 0.9372300505638123, "learning_rate": 1.0923946490806226e-06, "loss": 2.7012, "step": 332040 }, { "epoch": 0.00064, "grad_norm": 0.9294028878211975, "learning_rate": 1.0920287006722308e-06, "loss": 2.782, "step": 332050 }, { "epoch": 0.0006656, "grad_norm": 0.8201891779899597, "learning_rate": 1.0916628100296644e-06, "loss": 2.9011, "step": 332060 }, { "epoch": 0.0006912, "grad_norm": 0.763891875743866, "learning_rate": 1.0912969771552973e-06, "loss": 2.5143, "step": 332070 }, { "epoch": 0.0007168, "grad_norm": 0.8729631304740906, "learning_rate": 1.0909312020515016e-06, "loss": 2.7957, "step": 332080 }, { "epoch": 0.0007424, "grad_norm": 0.820128858089447, "learning_rate": 1.0905654847206492e-06, "loss": 2.6084, "step": 332090 }, { "epoch": 0.000768, "grad_norm": 0.8342142701148987, "learning_rate": 1.0901998251651124e-06, "loss": 2.6271, "step": 332100 }, { "epoch": 0.0007936, "grad_norm": 0.9790160655975342, "learning_rate": 1.0898342233872616e-06, "loss": 2.5748, "step": 332110 }, { "epoch": 0.0008192, "grad_norm": 0.8761657476425171, "learning_rate": 1.0894686793894671e-06, "loss": 2.7285, "step": 332120 }, { "epoch": 0.0008448, "grad_norm": 0.8725764155387878, "learning_rate": 1.0891031931741014e-06, "loss": 2.5292, "step": 332130 }, { "epoch": 0.0008704, "grad_norm": 0.8563762307167053, "learning_rate": 1.0887377647435328e-06, "loss": 2.4535, "step": 332140 }, { "epoch": 0.000896, "grad_norm": 0.9307699203491211, "learning_rate": 1.0883723941001323e-06, "loss": 2.5803, "step": 332150 }, { "epoch": 0.0009216, "grad_norm": 0.8331977128982544, "learning_rate": 1.0880070812462674e-06, "loss": 2.8634, "step": 332160 }, { "epoch": 0.0009472, "grad_norm": 0.9375569820404053, "learning_rate": 1.087641826184308e-06, "loss": 2.4425, "step": 332170 }, { "epoch": 0.0009728, "grad_norm": 0.8553615212440491, "learning_rate": 1.0872766289166225e-06, "loss": 2.6683, "step": 332180 }, { "epoch": 0.0009984, "grad_norm": 1.019819736480713, "learning_rate": 1.0869114894455802e-06, "loss": 2.4926, "step": 332190 }, { "epoch": 0.001024, "grad_norm": 0.9371776580810547, "learning_rate": 1.0865464077735467e-06, "loss": 2.7282, "step": 332200 }, { "epoch": 0.0010496, "grad_norm": 0.8373585939407349, "learning_rate": 1.0861813839028922e-06, "loss": 2.7203, "step": 332210 }, { "epoch": 0.0010752, "grad_norm": 0.8931773900985718, "learning_rate": 1.0858164178359832e-06, "loss": 2.6785, "step": 332220 }, { "epoch": 0.0011008, "grad_norm": 0.8332291841506958, "learning_rate": 1.0854515095751839e-06, "loss": 2.6857, "step": 332230 }, { "epoch": 0.0011264, "grad_norm": 0.8278331160545349, "learning_rate": 1.0850866591228625e-06, "loss": 2.5549, "step": 332240 }, { "epoch": 0.001152, "grad_norm": 0.8958258032798767, "learning_rate": 1.084721866481383e-06, "loss": 2.6937, "step": 332250 }, { "epoch": 0.0011776, "grad_norm": 0.8201335668563843, "learning_rate": 1.0843571316531153e-06, "loss": 2.7704, "step": 332260 }, { "epoch": 0.0012032, "grad_norm": 0.7796531319618225, "learning_rate": 1.0839924546404235e-06, "loss": 2.3385, "step": 332270 }, { "epoch": 0.0012288, "grad_norm": 0.8327628374099731, "learning_rate": 1.0836278354456687e-06, "loss": 2.6988, "step": 332280 }, { "epoch": 0.0012544, "grad_norm": 0.9127781391143799, "learning_rate": 1.083263274071218e-06, "loss": 2.8869, "step": 332290 }, { "epoch": 0.00128, "grad_norm": 0.8552094101905823, "learning_rate": 1.082898770519436e-06, "loss": 2.8655, "step": 332300 }, { "epoch": 0.0013056, "grad_norm": 0.9145452380180359, "learning_rate": 1.0825343247926846e-06, "loss": 2.581, "step": 332310 }, { "epoch": 0.0013312, "grad_norm": 1.2899575233459473, "learning_rate": 1.0821699368933282e-06, "loss": 2.7438, "step": 332320 }, { "epoch": 0.0013568, "grad_norm": 0.8878595232963562, "learning_rate": 1.0818056068237304e-06, "loss": 2.7435, "step": 332330 }, { "epoch": 0.0013824, "grad_norm": 0.8082001805305481, "learning_rate": 1.0814413345862518e-06, "loss": 2.5149, "step": 332340 }, { "epoch": 0.001408, "grad_norm": 0.9290775060653687, "learning_rate": 1.0810771201832593e-06, "loss": 2.6405, "step": 332350 }, { "epoch": 0.0014336, "grad_norm": 0.9592680931091309, "learning_rate": 1.0807129636171075e-06, "loss": 2.7724, "step": 332360 }, { "epoch": 0.0014592, "grad_norm": 0.7865813374519348, "learning_rate": 1.080348864890164e-06, "loss": 2.5881, "step": 332370 }, { "epoch": 0.0014848, "grad_norm": 1.1996577978134155, "learning_rate": 1.0799848240047862e-06, "loss": 2.0401, "step": 332380 }, { "epoch": 0.0015104, "grad_norm": 0.7853342294692993, "learning_rate": 1.0796208409633357e-06, "loss": 2.5395, "step": 332390 }, { "epoch": 0.001536, "grad_norm": 0.8237375617027283, "learning_rate": 1.0792569157681743e-06, "loss": 2.5314, "step": 332400 }, { "epoch": 0.0015616, "grad_norm": 0.8752478361129761, "learning_rate": 1.0788930484216608e-06, "loss": 2.5196, "step": 332410 }, { "epoch": 0.0015872, "grad_norm": 0.8772560358047485, "learning_rate": 1.0785292389261559e-06, "loss": 2.6584, "step": 332420 }, { "epoch": 0.0016128, "grad_norm": 0.7892122864723206, "learning_rate": 1.0781654872840152e-06, "loss": 2.6733, "step": 332430 }, { "epoch": 0.0016384, "grad_norm": 0.8745300769805908, "learning_rate": 1.0778017934976015e-06, "loss": 2.8489, "step": 332440 }, { "epoch": 0.001664, "grad_norm": 0.8807716369628906, "learning_rate": 1.0774381575692717e-06, "loss": 2.7357, "step": 332450 }, { "epoch": 0.0016896, "grad_norm": 0.8235594034194946, "learning_rate": 1.0770745795013849e-06, "loss": 2.6059, "step": 332460 }, { "epoch": 0.0017152, "grad_norm": 0.9098677635192871, "learning_rate": 1.0767110592962993e-06, "loss": 2.7212, "step": 332470 }, { "epoch": 0.0017408, "grad_norm": 0.8737910985946655, "learning_rate": 1.0763475969563697e-06, "loss": 2.6323, "step": 332480 }, { "epoch": 0.0017664, "grad_norm": 0.8311406970024109, "learning_rate": 1.0759841924839531e-06, "loss": 2.8247, "step": 332490 }, { "epoch": 0.001792, "grad_norm": 0.8808715343475342, "learning_rate": 1.0756208458814088e-06, "loss": 2.5237, "step": 332500 }, { "epoch": 0.0018176, "grad_norm": 0.8294239044189453, "learning_rate": 1.0752575571510903e-06, "loss": 2.7622, "step": 332510 }, { "epoch": 0.0018432, "grad_norm": 0.8753010630607605, "learning_rate": 1.074894326295355e-06, "loss": 2.478, "step": 332520 }, { "epoch": 0.0018688, "grad_norm": 0.9614617824554443, "learning_rate": 1.074531153316558e-06, "loss": 2.9339, "step": 332530 }, { "epoch": 0.0018944, "grad_norm": 0.9161588549613953, "learning_rate": 1.0741680382170549e-06, "loss": 2.7528, "step": 332540 }, { "epoch": 0.00192, "grad_norm": 0.8442064523696899, "learning_rate": 1.0738049809992012e-06, "loss": 2.7346, "step": 332550 }, { "epoch": 0.0019456, "grad_norm": 0.9943075776100159, "learning_rate": 1.0734419816653474e-06, "loss": 2.7608, "step": 332560 }, { "epoch": 0.0019712, "grad_norm": 0.7339341640472412, "learning_rate": 1.0730790402178515e-06, "loss": 2.7362, "step": 332570 }, { "epoch": 0.0019968, "grad_norm": 1.0130213499069214, "learning_rate": 1.072716156659064e-06, "loss": 2.5259, "step": 332580 }, { "epoch": 0.0020224, "grad_norm": 0.8360907435417175, "learning_rate": 1.0723533309913403e-06, "loss": 2.66, "step": 332590 }, { "epoch": 0.002048, "grad_norm": 1.1560418605804443, "learning_rate": 1.0719905632170325e-06, "loss": 2.4651, "step": 332600 }, { "epoch": 0.0020736, "grad_norm": 0.8416376709938049, "learning_rate": 1.0716278533384938e-06, "loss": 2.6703, "step": 332610 }, { "epoch": 0.0020992, "grad_norm": 0.9290509223937988, "learning_rate": 1.0712652013580748e-06, "loss": 2.7992, "step": 332620 }, { "epoch": 0.0021248, "grad_norm": 0.9045975804328918, "learning_rate": 1.0709026072781282e-06, "loss": 2.6793, "step": 332630 }, { "epoch": 0.0021504, "grad_norm": 0.912328839302063, "learning_rate": 1.070540071101005e-06, "loss": 3.0173, "step": 332640 }, { "epoch": 0.002176, "grad_norm": 0.9022132158279419, "learning_rate": 1.070177592829056e-06, "loss": 2.7971, "step": 332650 }, { "epoch": 0.0022016, "grad_norm": 0.9231812357902527, "learning_rate": 1.0698151724646322e-06, "loss": 2.9612, "step": 332660 }, { "epoch": 0.0022272, "grad_norm": 0.9874371290206909, "learning_rate": 1.0694528100100865e-06, "loss": 2.5843, "step": 332670 }, { "epoch": 0.0022528, "grad_norm": 0.8082192540168762, "learning_rate": 1.0690905054677636e-06, "loss": 2.724, "step": 332680 }, { "epoch": 0.0022784, "grad_norm": 0.8147331476211548, "learning_rate": 1.0687282588400138e-06, "loss": 2.3412, "step": 332690 }, { "epoch": 0.002304, "grad_norm": 1.1327563524246216, "learning_rate": 1.0683660701291888e-06, "loss": 2.3102, "step": 332700 }, { "epoch": 0.0023296, "grad_norm": 0.9633175134658813, "learning_rate": 1.0680039393376362e-06, "loss": 2.6603, "step": 332710 }, { "epoch": 0.0023552, "grad_norm": 0.8042020797729492, "learning_rate": 1.0676418664677034e-06, "loss": 2.625, "step": 332720 }, { "epoch": 0.0023808, "grad_norm": 0.8719680309295654, "learning_rate": 1.067279851521741e-06, "loss": 2.464, "step": 332730 }, { "epoch": 0.0024064, "grad_norm": 0.8199965357780457, "learning_rate": 1.066917894502093e-06, "loss": 2.7788, "step": 332740 }, { "epoch": 0.002432, "grad_norm": 1.009618878364563, "learning_rate": 1.0665559954111104e-06, "loss": 2.9311, "step": 332750 }, { "epoch": 0.0024576, "grad_norm": 0.7882267236709595, "learning_rate": 1.0661941542511366e-06, "loss": 2.5617, "step": 332760 }, { "epoch": 0.0024832, "grad_norm": 0.7735333442687988, "learning_rate": 1.0658323710245166e-06, "loss": 2.5609, "step": 332770 }, { "epoch": 0.0025088, "grad_norm": 0.9090549349784851, "learning_rate": 1.0654706457336016e-06, "loss": 2.7986, "step": 332780 }, { "epoch": 0.0025344, "grad_norm": 0.9489734172821045, "learning_rate": 1.0651089783807355e-06, "loss": 2.5537, "step": 332790 }, { "epoch": 0.00256, "grad_norm": 1.1082682609558105, "learning_rate": 1.0647473689682652e-06, "loss": 2.8647, "step": 332800 }, { "epoch": 0.0025856, "grad_norm": 0.8717849850654602, "learning_rate": 1.064385817498531e-06, "loss": 2.7373, "step": 332810 }, { "epoch": 0.0026112, "grad_norm": 1.2139769792556763, "learning_rate": 1.06402432397388e-06, "loss": 2.784, "step": 332820 }, { "epoch": 0.0026368, "grad_norm": 0.8870424032211304, "learning_rate": 1.063662888396656e-06, "loss": 2.8824, "step": 332830 }, { "epoch": 0.0026624, "grad_norm": 0.8430402278900146, "learning_rate": 1.0633015107692036e-06, "loss": 2.6823, "step": 332840 }, { "epoch": 0.002688, "grad_norm": 0.874394416809082, "learning_rate": 1.0629401910938652e-06, "loss": 2.9415, "step": 332850 }, { "epoch": 0.0027136, "grad_norm": 0.8221712112426758, "learning_rate": 1.062578929372985e-06, "loss": 2.8472, "step": 332860 }, { "epoch": 0.0027392, "grad_norm": 0.9047221541404724, "learning_rate": 1.0622177256089072e-06, "loss": 2.8352, "step": 332870 }, { "epoch": 0.0027648, "grad_norm": 0.812212347984314, "learning_rate": 1.0618565798039703e-06, "loss": 2.98, "step": 332880 }, { "epoch": 0.0027904, "grad_norm": 0.8236969709396362, "learning_rate": 1.061495491960518e-06, "loss": 2.7965, "step": 332890 }, { "epoch": 0.002816, "grad_norm": 0.8074231743812561, "learning_rate": 1.0611344620808916e-06, "loss": 2.7019, "step": 332900 }, { "epoch": 0.0028416, "grad_norm": 0.9613242149353027, "learning_rate": 1.0607734901674316e-06, "loss": 2.7394, "step": 332910 }, { "epoch": 0.0028672, "grad_norm": 0.8458696007728577, "learning_rate": 1.0604125762224816e-06, "loss": 2.8457, "step": 332920 }, { "epoch": 0.0028928, "grad_norm": 0.8487041592597961, "learning_rate": 1.0600517202483784e-06, "loss": 2.9038, "step": 332930 }, { "epoch": 0.0029184, "grad_norm": 0.8487605452537537, "learning_rate": 1.059690922247465e-06, "loss": 2.7954, "step": 332940 }, { "epoch": 0.002944, "grad_norm": 0.8443219661712646, "learning_rate": 1.0593301822220804e-06, "loss": 2.852, "step": 332950 }, { "epoch": 0.0029696, "grad_norm": 0.7875337600708008, "learning_rate": 1.0589695001745625e-06, "loss": 2.7379, "step": 332960 }, { "epoch": 0.0029952, "grad_norm": 0.8151273131370544, "learning_rate": 1.058608876107251e-06, "loss": 3.0197, "step": 332970 }, { "epoch": 0.0030208, "grad_norm": 0.8916094899177551, "learning_rate": 1.058248310022485e-06, "loss": 2.6422, "step": 332980 }, { "epoch": 0.0030464, "grad_norm": 0.7784891128540039, "learning_rate": 1.0578878019226024e-06, "loss": 2.8392, "step": 332990 }, { "epoch": 0.003072, "grad_norm": 0.9343422055244446, "learning_rate": 1.0575273518099427e-06, "loss": 3.0185, "step": 333000 }, { "epoch": 0.0030976, "grad_norm": 0.8463826775550842, "learning_rate": 1.0571669596868394e-06, "loss": 2.7885, "step": 333010 }, { "epoch": 0.0031232, "grad_norm": 0.81303870677948, "learning_rate": 1.056806625555631e-06, "loss": 2.7307, "step": 333020 }, { "epoch": 0.0031488, "grad_norm": 0.9081551432609558, "learning_rate": 1.056446349418656e-06, "loss": 2.7008, "step": 333030 }, { "epoch": 0.0031744, "grad_norm": 0.9741219878196716, "learning_rate": 1.056086131278249e-06, "loss": 2.8856, "step": 333040 }, { "epoch": 0.0032, "grad_norm": 0.9165816307067871, "learning_rate": 1.0557259711367463e-06, "loss": 3.1232, "step": 333050 }, { "epoch": 0.0032256, "grad_norm": 0.7866579294204712, "learning_rate": 1.055365868996484e-06, "loss": 2.8711, "step": 333060 }, { "epoch": 0.0032512, "grad_norm": 1.157456636428833, "learning_rate": 1.0550058248597984e-06, "loss": 2.842, "step": 333070 }, { "epoch": 0.0032768, "grad_norm": 0.923152506351471, "learning_rate": 1.0546458387290204e-06, "loss": 2.8835, "step": 333080 }, { "epoch": 0.0033024, "grad_norm": 0.8501176834106445, "learning_rate": 1.0542859106064874e-06, "loss": 2.8424, "step": 333090 }, { "epoch": 0.003328, "grad_norm": 1.0156797170639038, "learning_rate": 1.0539260404945328e-06, "loss": 2.9212, "step": 333100 }, { "epoch": 0.0033536, "grad_norm": 0.9121742844581604, "learning_rate": 1.0535662283954905e-06, "loss": 2.8837, "step": 333110 }, { "epoch": 0.0033792, "grad_norm": 0.8294546604156494, "learning_rate": 1.0532064743116909e-06, "loss": 2.804, "step": 333120 }, { "epoch": 0.0034048, "grad_norm": 0.7821899056434631, "learning_rate": 1.0528467782454733e-06, "loss": 2.815, "step": 333130 }, { "epoch": 0.0034304, "grad_norm": 0.8751494884490967, "learning_rate": 1.0524871401991654e-06, "loss": 2.738, "step": 333140 }, { "epoch": 0.003456, "grad_norm": 0.8235014081001282, "learning_rate": 1.0521275601751002e-06, "loss": 2.4554, "step": 333150 }, { "epoch": 0.0034816, "grad_norm": 0.8096911311149597, "learning_rate": 1.051768038175609e-06, "loss": 2.7882, "step": 333160 }, { "epoch": 0.0035072, "grad_norm": 0.8170681595802307, "learning_rate": 1.0514085742030245e-06, "loss": 2.7019, "step": 333170 }, { "epoch": 0.0035328, "grad_norm": 0.8471071124076843, "learning_rate": 1.0510491682596768e-06, "loss": 2.7098, "step": 333180 }, { "epoch": 0.0035584, "grad_norm": 0.8454200625419617, "learning_rate": 1.0506898203478977e-06, "loss": 2.8025, "step": 333190 }, { "epoch": 0.003584, "grad_norm": 0.8014265298843384, "learning_rate": 1.0503305304700172e-06, "loss": 2.874, "step": 333200 }, { "epoch": 0.0036096, "grad_norm": 0.842605710029602, "learning_rate": 1.049971298628364e-06, "loss": 2.4649, "step": 333210 }, { "epoch": 0.0036352, "grad_norm": 0.8515046834945679, "learning_rate": 1.0496121248252678e-06, "loss": 2.8825, "step": 333220 }, { "epoch": 0.0036608, "grad_norm": 0.8419577479362488, "learning_rate": 1.0492530090630582e-06, "loss": 2.8107, "step": 333230 }, { "epoch": 0.0036864, "grad_norm": 0.8539602756500244, "learning_rate": 1.0488939513440644e-06, "loss": 2.7351, "step": 333240 }, { "epoch": 0.003712, "grad_norm": 0.8349481225013733, "learning_rate": 1.0485349516706145e-06, "loss": 2.6589, "step": 333250 }, { "epoch": 0.0037376, "grad_norm": 0.8619201183319092, "learning_rate": 1.0481760100450355e-06, "loss": 2.836, "step": 333260 }, { "epoch": 0.0037632, "grad_norm": 0.933935821056366, "learning_rate": 1.047817126469659e-06, "loss": 2.7025, "step": 333270 }, { "epoch": 0.0037888, "grad_norm": 0.778460681438446, "learning_rate": 1.0474583009468065e-06, "loss": 2.6491, "step": 333280 }, { "epoch": 0.0038144, "grad_norm": 0.8299728035926819, "learning_rate": 1.047099533478808e-06, "loss": 2.9786, "step": 333290 }, { "epoch": 0.00384, "grad_norm": 0.8237646222114563, "learning_rate": 1.0467408240679877e-06, "loss": 2.8713, "step": 333300 }, { "epoch": 0.0038656, "grad_norm": 0.8070095777511597, "learning_rate": 1.0463821727166756e-06, "loss": 2.7759, "step": 333310 }, { "epoch": 0.0038912, "grad_norm": 1.0972570180892944, "learning_rate": 1.0460235794271957e-06, "loss": 2.7725, "step": 333320 }, { "epoch": 0.0039168, "grad_norm": 0.7913302183151245, "learning_rate": 1.0456650442018745e-06, "loss": 2.8262, "step": 333330 }, { "epoch": 0.0039424, "grad_norm": 0.9261825680732727, "learning_rate": 1.0453065670430341e-06, "loss": 3.0471, "step": 333340 }, { "epoch": 0.003968, "grad_norm": 0.8626664876937866, "learning_rate": 1.044948147953001e-06, "loss": 2.8276, "step": 333350 }, { "epoch": 0.0039936, "grad_norm": 1.0397286415100098, "learning_rate": 1.0445897869340981e-06, "loss": 2.7649, "step": 333360 }, { "epoch": 0.0040192, "grad_norm": 0.8750208616256714, "learning_rate": 1.0442314839886514e-06, "loss": 2.974, "step": 333370 }, { "epoch": 0.0040448, "grad_norm": 1.2150940895080566, "learning_rate": 1.0438732391189832e-06, "loss": 2.9546, "step": 333380 }, { "epoch": 0.0040704, "grad_norm": 0.7587742209434509, "learning_rate": 1.0435150523274174e-06, "loss": 2.9151, "step": 333390 }, { "epoch": 0.004096, "grad_norm": 0.7504233121871948, "learning_rate": 1.0431569236162776e-06, "loss": 2.8611, "step": 333400 }, { "epoch": 0.0041216, "grad_norm": 0.9419805407524109, "learning_rate": 1.0427988529878819e-06, "loss": 2.8139, "step": 333410 }, { "epoch": 0.0041472, "grad_norm": 0.7749132513999939, "learning_rate": 1.0424408404445562e-06, "loss": 2.6772, "step": 333420 }, { "epoch": 0.0041728, "grad_norm": 0.8669754266738892, "learning_rate": 1.04208288598862e-06, "loss": 2.6541, "step": 333430 }, { "epoch": 0.0041984, "grad_norm": 1.1903092861175537, "learning_rate": 1.0417249896223969e-06, "loss": 2.7131, "step": 333440 }, { "epoch": 0.004224, "grad_norm": 0.8030692338943481, "learning_rate": 1.041367151348205e-06, "loss": 2.837, "step": 333450 }, { "epoch": 0.0042496, "grad_norm": 0.9578144550323486, "learning_rate": 1.0410093711683678e-06, "loss": 2.9428, "step": 333460 }, { "epoch": 0.0042752, "grad_norm": 0.760387659072876, "learning_rate": 1.0406516490852025e-06, "loss": 2.6534, "step": 333470 }, { "epoch": 0.0043008, "grad_norm": 0.86076420545578, "learning_rate": 1.0402939851010308e-06, "loss": 2.9362, "step": 333480 }, { "epoch": 0.0043264, "grad_norm": 0.8730945587158203, "learning_rate": 1.0399363792181704e-06, "loss": 2.7954, "step": 333490 }, { "epoch": 0.004352, "grad_norm": 0.8807933926582336, "learning_rate": 1.0395788314389422e-06, "loss": 2.8872, "step": 333500 }, { "epoch": 0.0043776, "grad_norm": 0.9104208946228027, "learning_rate": 1.0392213417656637e-06, "loss": 2.9376, "step": 333510 }, { "epoch": 0.0044032, "grad_norm": 0.8250205516815186, "learning_rate": 1.0388639102006537e-06, "loss": 2.832, "step": 333520 }, { "epoch": 0.0044288, "grad_norm": 0.8628438711166382, "learning_rate": 1.038506536746231e-06, "loss": 2.968, "step": 333530 }, { "epoch": 0.0044544, "grad_norm": 0.8769908547401428, "learning_rate": 1.0381492214047106e-06, "loss": 2.6176, "step": 333540 }, { "epoch": 0.00448, "grad_norm": 0.808664083480835, "learning_rate": 1.0377919641784096e-06, "loss": 2.7916, "step": 333550 }, { "epoch": 0.0045056, "grad_norm": 0.7662606239318848, "learning_rate": 1.037434765069647e-06, "loss": 2.7233, "step": 333560 }, { "epoch": 0.0045312, "grad_norm": 0.9204140305519104, "learning_rate": 1.0370776240807368e-06, "loss": 2.8645, "step": 333570 }, { "epoch": 0.0045568, "grad_norm": 0.7951756715774536, "learning_rate": 1.036720541213997e-06, "loss": 2.8858, "step": 333580 }, { "epoch": 0.0045824, "grad_norm": 0.8489330410957336, "learning_rate": 1.0363635164717422e-06, "loss": 2.9075, "step": 333590 }, { "epoch": 0.004608, "grad_norm": 0.9051831364631653, "learning_rate": 1.0360065498562898e-06, "loss": 2.9756, "step": 333600 }, { "epoch": 0.0046336, "grad_norm": 0.8551836609840393, "learning_rate": 1.0356496413699502e-06, "loss": 2.896, "step": 333610 }, { "epoch": 0.0046592, "grad_norm": 0.8623046875, "learning_rate": 1.0352927910150402e-06, "loss": 2.9213, "step": 333620 }, { "epoch": 0.0046848, "grad_norm": 0.940963089466095, "learning_rate": 1.0349359987938746e-06, "loss": 2.9911, "step": 333630 }, { "epoch": 0.0047104, "grad_norm": 0.7946330904960632, "learning_rate": 1.0345792647087639e-06, "loss": 2.7529, "step": 333640 }, { "epoch": 0.004736, "grad_norm": 0.7571150660514832, "learning_rate": 1.0342225887620262e-06, "loss": 2.7673, "step": 333650 }, { "epoch": 0.0047616, "grad_norm": 0.841296911239624, "learning_rate": 1.033865970955974e-06, "loss": 2.7896, "step": 333660 }, { "epoch": 0.0047872, "grad_norm": 0.893017053604126, "learning_rate": 1.0335094112929155e-06, "loss": 2.8163, "step": 333670 }, { "epoch": 0.0048128, "grad_norm": 1.0011016130447388, "learning_rate": 1.0331529097751659e-06, "loss": 2.8624, "step": 333680 }, { "epoch": 0.0048384, "grad_norm": 0.8524145483970642, "learning_rate": 1.0327964664050372e-06, "loss": 2.9088, "step": 333690 }, { "epoch": 0.004864, "grad_norm": 0.8206572532653809, "learning_rate": 1.03244008118484e-06, "loss": 3.0904, "step": 333700 }, { "epoch": 0.0048896, "grad_norm": 0.8925011157989502, "learning_rate": 1.0320837541168848e-06, "loss": 2.6581, "step": 333710 }, { "epoch": 0.0049152, "grad_norm": 0.8688550591468811, "learning_rate": 1.031727485203483e-06, "loss": 2.7831, "step": 333720 }, { "epoch": 0.0049408, "grad_norm": 0.7722882032394409, "learning_rate": 1.0313712744469483e-06, "loss": 2.6374, "step": 333730 }, { "epoch": 0.0049664, "grad_norm": 0.9359403848648071, "learning_rate": 1.0310151218495845e-06, "loss": 2.9338, "step": 333740 }, { "epoch": 0.004992, "grad_norm": 0.8111939430236816, "learning_rate": 1.030659027413704e-06, "loss": 2.7981, "step": 333750 }, { "epoch": 0.0050176, "grad_norm": 0.7834702134132385, "learning_rate": 1.0303029911416163e-06, "loss": 2.9064, "step": 333760 }, { "epoch": 0.0050432, "grad_norm": 0.8092560172080994, "learning_rate": 1.0299470130356292e-06, "loss": 2.6338, "step": 333770 }, { "epoch": 0.0050688, "grad_norm": 0.9332627058029175, "learning_rate": 1.0295910930980524e-06, "loss": 2.5726, "step": 333780 }, { "epoch": 0.0050944, "grad_norm": 0.9821760654449463, "learning_rate": 1.0292352313311927e-06, "loss": 2.6978, "step": 333790 }, { "epoch": 0.00512, "grad_norm": 0.8178145885467529, "learning_rate": 1.0288794277373604e-06, "loss": 2.8459, "step": 333800 }, { "epoch": 0.0051456, "grad_norm": 0.8899049162864685, "learning_rate": 1.0285236823188593e-06, "loss": 2.7744, "step": 333810 }, { "epoch": 0.0051712, "grad_norm": 0.738628625869751, "learning_rate": 1.0281679950779954e-06, "loss": 2.6944, "step": 333820 }, { "epoch": 0.0051968, "grad_norm": 1.1300342082977295, "learning_rate": 1.0278123660170803e-06, "loss": 2.6715, "step": 333830 }, { "epoch": 0.0052224, "grad_norm": 0.7696925401687622, "learning_rate": 1.0274567951384162e-06, "loss": 2.7799, "step": 333840 }, { "epoch": 0.005248, "grad_norm": 0.8917068243026733, "learning_rate": 1.0271012824443116e-06, "loss": 2.7565, "step": 333850 }, { "epoch": 0.0052736, "grad_norm": 0.9727521538734436, "learning_rate": 1.0267458279370712e-06, "loss": 2.9599, "step": 333860 }, { "epoch": 0.0052992, "grad_norm": 0.8350344896316528, "learning_rate": 1.0263904316189976e-06, "loss": 2.6834, "step": 333870 }, { "epoch": 0.0053248, "grad_norm": 0.937065601348877, "learning_rate": 1.026035093492398e-06, "loss": 2.876, "step": 333880 }, { "epoch": 0.0053504, "grad_norm": 0.8407915830612183, "learning_rate": 1.0256798135595747e-06, "loss": 2.7719, "step": 333890 }, { "epoch": 0.005376, "grad_norm": 0.7655319571495056, "learning_rate": 1.0253245918228337e-06, "loss": 2.7597, "step": 333900 }, { "epoch": 0.0054016, "grad_norm": 0.7728853821754456, "learning_rate": 1.0249694282844769e-06, "loss": 2.6919, "step": 333910 }, { "epoch": 0.0054272, "grad_norm": 1.0905137062072754, "learning_rate": 1.0246143229468085e-06, "loss": 2.7001, "step": 333920 }, { "epoch": 0.0054528, "grad_norm": 0.7919524312019348, "learning_rate": 1.0242592758121327e-06, "loss": 2.8329, "step": 333930 }, { "epoch": 0.0054784, "grad_norm": 0.8805187940597534, "learning_rate": 1.0239042868827476e-06, "loss": 2.9204, "step": 333940 }, { "epoch": 0.005504, "grad_norm": 0.9301093816757202, "learning_rate": 1.0235493561609588e-06, "loss": 2.598, "step": 333950 }, { "epoch": 0.0055296, "grad_norm": 0.816862165927887, "learning_rate": 1.0231944836490659e-06, "loss": 2.7342, "step": 333960 }, { "epoch": 0.0055552, "grad_norm": 0.8189600110054016, "learning_rate": 1.0228396693493704e-06, "loss": 2.6382, "step": 333970 }, { "epoch": 0.0055808, "grad_norm": 0.781032383441925, "learning_rate": 1.0224849132641745e-06, "loss": 2.82, "step": 333980 }, { "epoch": 0.0056064, "grad_norm": 0.7706202864646912, "learning_rate": 1.022130215395778e-06, "loss": 2.7648, "step": 333990 }, { "epoch": 0.005632, "grad_norm": 0.8056879043579102, "learning_rate": 1.021775575746481e-06, "loss": 2.5375, "step": 334000 }, { "epoch": 0.0056576, "grad_norm": 0.8036518096923828, "learning_rate": 1.0214209943185837e-06, "loss": 2.5999, "step": 334010 }, { "epoch": 0.0056832, "grad_norm": 0.8400499820709229, "learning_rate": 1.0210664711143847e-06, "loss": 2.8569, "step": 334020 }, { "epoch": 0.0057088, "grad_norm": 0.7736915946006775, "learning_rate": 1.020712006136183e-06, "loss": 2.6732, "step": 334030 }, { "epoch": 0.0057344, "grad_norm": 0.807576060295105, "learning_rate": 1.020357599386278e-06, "loss": 2.8557, "step": 334040 }, { "epoch": 0.00576, "grad_norm": 0.8888845443725586, "learning_rate": 1.0200032508669678e-06, "loss": 2.7098, "step": 334050 }, { "epoch": 0.0057856, "grad_norm": 0.8136208057403564, "learning_rate": 1.0196489605805503e-06, "loss": 2.7388, "step": 334060 }, { "epoch": 0.0058112, "grad_norm": 0.792357325553894, "learning_rate": 1.019294728529322e-06, "loss": 2.7943, "step": 334070 }, { "epoch": 0.0058368, "grad_norm": 0.8331392407417297, "learning_rate": 1.0189405547155807e-06, "loss": 2.8283, "step": 334080 }, { "epoch": 0.0058624, "grad_norm": 0.8787257075309753, "learning_rate": 1.0185864391416223e-06, "loss": 2.6104, "step": 334090 }, { "epoch": 0.005888, "grad_norm": 0.9527376890182495, "learning_rate": 1.018232381809744e-06, "loss": 2.5205, "step": 334100 }, { "epoch": 0.0059136, "grad_norm": 0.9213262796401978, "learning_rate": 1.0178783827222417e-06, "loss": 2.7922, "step": 334110 }, { "epoch": 0.0059392, "grad_norm": 0.8787238001823425, "learning_rate": 1.0175244418814113e-06, "loss": 2.6339, "step": 334120 }, { "epoch": 0.0059648, "grad_norm": 0.9704062342643738, "learning_rate": 1.0171705592895498e-06, "loss": 2.7954, "step": 334130 }, { "epoch": 0.0059904, "grad_norm": 0.8170166611671448, "learning_rate": 1.0168167349489466e-06, "loss": 2.8413, "step": 334140 }, { "epoch": 0.006016, "grad_norm": 0.8376578092575073, "learning_rate": 1.016462968861901e-06, "loss": 3.0042, "step": 334150 }, { "epoch": 0.0060416, "grad_norm": 0.784715473651886, "learning_rate": 1.0161092610307044e-06, "loss": 2.4447, "step": 334160 }, { "epoch": 0.0060672, "grad_norm": 0.7872272729873657, "learning_rate": 1.0157556114576505e-06, "loss": 2.6923, "step": 334170 }, { "epoch": 0.0060928, "grad_norm": 0.7824360132217407, "learning_rate": 1.0154020201450366e-06, "loss": 2.7565, "step": 334180 }, { "epoch": 0.0061184, "grad_norm": 0.7585499286651611, "learning_rate": 1.0150484870951515e-06, "loss": 2.6297, "step": 334190 }, { "epoch": 0.006144, "grad_norm": 0.8032971024513245, "learning_rate": 1.0146950123102895e-06, "loss": 2.7569, "step": 334200 }, { "epoch": 0.0061696, "grad_norm": 0.99727463722229, "learning_rate": 1.0143415957927415e-06, "loss": 2.7337, "step": 334210 }, { "epoch": 0.0061952, "grad_norm": 0.8260914087295532, "learning_rate": 1.0139882375448007e-06, "loss": 2.9052, "step": 334220 }, { "epoch": 0.0062208, "grad_norm": 0.8012580871582031, "learning_rate": 1.0136349375687571e-06, "loss": 2.6699, "step": 334230 }, { "epoch": 0.0062464, "grad_norm": 0.8072203397750854, "learning_rate": 1.0132816958669034e-06, "loss": 2.5401, "step": 334240 }, { "epoch": 0.006272, "grad_norm": 0.8385012745857239, "learning_rate": 1.0129285124415311e-06, "loss": 2.5381, "step": 334250 }, { "epoch": 0.0062976, "grad_norm": 1.0377693176269531, "learning_rate": 1.0125753872949274e-06, "loss": 2.9636, "step": 334260 }, { "epoch": 0.0063232, "grad_norm": 1.2612767219543457, "learning_rate": 1.0122223204293835e-06, "loss": 2.8991, "step": 334270 }, { "epoch": 0.0063488, "grad_norm": 0.8670058250427246, "learning_rate": 1.011869311847189e-06, "loss": 2.7929, "step": 334280 }, { "epoch": 0.0063744, "grad_norm": 0.8085561394691467, "learning_rate": 1.011516361550634e-06, "loss": 2.6819, "step": 334290 }, { "epoch": 0.0064, "grad_norm": 0.8788697123527527, "learning_rate": 1.011163469542006e-06, "loss": 2.6688, "step": 334300 }, { "epoch": 0.0064256, "grad_norm": 0.8737044334411621, "learning_rate": 1.0108106358235937e-06, "loss": 2.7956, "step": 334310 }, { "epoch": 0.0064512, "grad_norm": 0.785605788230896, "learning_rate": 1.0104578603976855e-06, "loss": 2.6754, "step": 334320 }, { "epoch": 0.0064768, "grad_norm": 0.80882328748703, "learning_rate": 1.010105143266571e-06, "loss": 2.8299, "step": 334330 }, { "epoch": 0.0065024, "grad_norm": 0.8940168619155884, "learning_rate": 1.0097524844325313e-06, "loss": 2.7728, "step": 334340 }, { "epoch": 0.006528, "grad_norm": 0.8982833027839661, "learning_rate": 1.0093998838978603e-06, "loss": 2.7871, "step": 334350 }, { "epoch": 0.0065536, "grad_norm": 0.8121665716171265, "learning_rate": 1.0090473416648405e-06, "loss": 2.8379, "step": 334360 }, { "epoch": 0.0065792, "grad_norm": 0.874006986618042, "learning_rate": 1.0086948577357602e-06, "loss": 2.715, "step": 334370 }, { "epoch": 0.0066048, "grad_norm": 0.7757297158241272, "learning_rate": 1.0083424321129054e-06, "loss": 2.749, "step": 334380 }, { "epoch": 0.0066304, "grad_norm": 0.7893951535224915, "learning_rate": 1.0079900647985586e-06, "loss": 2.7141, "step": 334390 }, { "epoch": 0.006656, "grad_norm": 0.8225516080856323, "learning_rate": 1.0076377557950056e-06, "loss": 2.6947, "step": 334400 }, { "epoch": 0.0066816, "grad_norm": 0.8768882751464844, "learning_rate": 1.0072855051045327e-06, "loss": 2.9135, "step": 334410 }, { "epoch": 0.0067072, "grad_norm": 0.8394656777381897, "learning_rate": 1.0069333127294235e-06, "loss": 2.7153, "step": 334420 }, { "epoch": 0.0067328, "grad_norm": 0.8372594714164734, "learning_rate": 1.0065811786719615e-06, "loss": 2.7012, "step": 334430 }, { "epoch": 0.0067584, "grad_norm": 0.8843534588813782, "learning_rate": 1.0062291029344295e-06, "loss": 2.5724, "step": 334440 }, { "epoch": 0.006784, "grad_norm": 0.9448235630989075, "learning_rate": 1.0058770855191147e-06, "loss": 2.6949, "step": 334450 }, { "epoch": 0.0068096, "grad_norm": 0.8452430367469788, "learning_rate": 1.005525126428294e-06, "loss": 2.8258, "step": 334460 }, { "epoch": 0.0068352, "grad_norm": 0.8531296253204346, "learning_rate": 1.0051732256642533e-06, "loss": 2.8771, "step": 334470 }, { "epoch": 0.0068608, "grad_norm": 0.8799534440040588, "learning_rate": 1.004821383229273e-06, "loss": 2.8651, "step": 334480 }, { "epoch": 0.0068864, "grad_norm": 0.9239771366119385, "learning_rate": 1.0044695991256347e-06, "loss": 2.981, "step": 334490 }, { "epoch": 0.006912, "grad_norm": 0.9734703898429871, "learning_rate": 1.0041178733556213e-06, "loss": 2.8698, "step": 334500 }, { "epoch": 0.0069376, "grad_norm": 0.8435357809066772, "learning_rate": 1.0037662059215125e-06, "loss": 2.6748, "step": 334510 }, { "epoch": 0.0069632, "grad_norm": 1.1717536449432373, "learning_rate": 1.0034145968255882e-06, "loss": 2.9682, "step": 334520 }, { "epoch": 0.0069888, "grad_norm": 0.7781186699867249, "learning_rate": 1.0030630460701307e-06, "loss": 2.8542, "step": 334530 }, { "epoch": 0.0070144, "grad_norm": 1.094977855682373, "learning_rate": 1.0027115536574172e-06, "loss": 2.8267, "step": 334540 }, { "epoch": 0.00704, "grad_norm": 0.9871181845664978, "learning_rate": 1.002360119589728e-06, "loss": 2.8939, "step": 334550 }, { "epoch": 0.0070656, "grad_norm": 0.8837922215461731, "learning_rate": 1.0020087438693427e-06, "loss": 2.7061, "step": 334560 }, { "epoch": 0.0070912, "grad_norm": 1.083343744277954, "learning_rate": 1.001657426498539e-06, "loss": 2.623, "step": 334570 }, { "epoch": 0.0071168, "grad_norm": 0.9050017595291138, "learning_rate": 1.0013061674795975e-06, "loss": 3.0634, "step": 334580 }, { "epoch": 0.0071424, "grad_norm": 0.8759593963623047, "learning_rate": 1.0009549668147923e-06, "loss": 2.895, "step": 334590 }, { "epoch": 0.007168, "grad_norm": 0.869428813457489, "learning_rate": 1.000603824506402e-06, "loss": 2.8401, "step": 334600 }, { "epoch": 0.0071936, "grad_norm": 0.8755108714103699, "learning_rate": 1.0002527405567042e-06, "loss": 2.4601, "step": 334610 }, { "epoch": 0.0072192, "grad_norm": 0.8866882920265198, "learning_rate": 9.999017149679757e-07, "loss": 2.9772, "step": 334620 }, { "epoch": 0.0072448, "grad_norm": 0.8223859071731567, "learning_rate": 9.995507477424927e-07, "loss": 2.7341, "step": 334630 }, { "epoch": 0.0072704, "grad_norm": 0.7669513821601868, "learning_rate": 9.991998388825308e-07, "loss": 2.8055, "step": 334640 }, { "epoch": 0.007296, "grad_norm": 0.8726632595062256, "learning_rate": 9.988489883903673e-07, "loss": 2.7372, "step": 334650 }, { "epoch": 0.0073216, "grad_norm": 0.8822106719017029, "learning_rate": 9.984981962682749e-07, "loss": 2.7474, "step": 334660 }, { "epoch": 0.0073472, "grad_norm": 0.7486990690231323, "learning_rate": 9.981474625185284e-07, "loss": 2.8563, "step": 334670 }, { "epoch": 0.0073728, "grad_norm": 0.7922030091285706, "learning_rate": 9.977967871434036e-07, "loss": 2.7877, "step": 334680 }, { "epoch": 0.0073984, "grad_norm": 0.8647270798683167, "learning_rate": 9.974461701451731e-07, "loss": 2.8451, "step": 334690 }, { "epoch": 0.007424, "grad_norm": 0.7957850098609924, "learning_rate": 9.97095611526112e-07, "loss": 2.7451, "step": 334700 }, { "epoch": 0.0074496, "grad_norm": 0.7589166760444641, "learning_rate": 9.967451112884962e-07, "loss": 2.716, "step": 334710 }, { "epoch": 0.0074752, "grad_norm": 0.8838574290275574, "learning_rate": 9.96394669434594e-07, "loss": 2.7633, "step": 334720 }, { "epoch": 0.0075008, "grad_norm": 0.9142937064170837, "learning_rate": 9.960442859666786e-07, "loss": 2.8829, "step": 334730 }, { "epoch": 0.0075264, "grad_norm": 0.9243715405464172, "learning_rate": 9.95693960887023e-07, "loss": 2.8453, "step": 334740 }, { "epoch": 0.007552, "grad_norm": 0.923089861869812, "learning_rate": 9.953436941978988e-07, "loss": 2.7476, "step": 334750 }, { "epoch": 0.0075776, "grad_norm": 1.0061123371124268, "learning_rate": 9.949934859015787e-07, "loss": 2.8294, "step": 334760 }, { "epoch": 0.0076032, "grad_norm": 0.8461161255836487, "learning_rate": 9.946433360003317e-07, "loss": 2.7744, "step": 334770 }, { "epoch": 0.0076288, "grad_norm": 0.8438428044319153, "learning_rate": 9.942932444964314e-07, "loss": 2.8269, "step": 334780 }, { "epoch": 0.0076544, "grad_norm": 0.8600029349327087, "learning_rate": 9.939432113921444e-07, "loss": 2.9337, "step": 334790 }, { "epoch": 0.00768, "grad_norm": 0.8746490478515625, "learning_rate": 9.935932366897427e-07, "loss": 2.9673, "step": 334800 }, { "epoch": 0.0077056, "grad_norm": 0.8896478414535522, "learning_rate": 9.932433203914936e-07, "loss": 2.6869, "step": 334810 }, { "epoch": 0.0077312, "grad_norm": 0.8352142572402954, "learning_rate": 9.928934624996688e-07, "loss": 2.8202, "step": 334820 }, { "epoch": 0.0077568, "grad_norm": 0.8255509734153748, "learning_rate": 9.925436630165363e-07, "loss": 2.8327, "step": 334830 }, { "epoch": 0.0077824, "grad_norm": 0.8547870516777039, "learning_rate": 9.921939219443643e-07, "loss": 2.6709, "step": 334840 }, { "epoch": 0.007808, "grad_norm": 0.8670350909233093, "learning_rate": 9.918442392854222e-07, "loss": 2.7923, "step": 334850 }, { "epoch": 0.0078336, "grad_norm": 0.8447041511535645, "learning_rate": 9.914946150419724e-07, "loss": 2.9281, "step": 334860 }, { "epoch": 0.0078592, "grad_norm": 2.1758179664611816, "learning_rate": 9.911450492162877e-07, "loss": 2.829, "step": 334870 }, { "epoch": 0.0078848, "grad_norm": 0.8154467940330505, "learning_rate": 9.90795541810633e-07, "loss": 2.7781, "step": 334880 }, { "epoch": 0.0079104, "grad_norm": 0.8299452662467957, "learning_rate": 9.904460928272753e-07, "loss": 2.6726, "step": 334890 }, { "epoch": 0.007936, "grad_norm": 0.7837664484977722, "learning_rate": 9.900967022684804e-07, "loss": 3.0173, "step": 334900 }, { "epoch": 0.0079616, "grad_norm": 0.8706325888633728, "learning_rate": 9.897473701365145e-07, "loss": 2.8025, "step": 334910 }, { "epoch": 0.0079872, "grad_norm": 0.7649639248847961, "learning_rate": 9.893980964336414e-07, "loss": 2.7139, "step": 334920 }, { "epoch": 0.0080128, "grad_norm": 0.9178501963615417, "learning_rate": 9.89048881162127e-07, "loss": 2.9202, "step": 334930 }, { "epoch": 0.0080384, "grad_norm": 1.0119045972824097, "learning_rate": 9.886997243242358e-07, "loss": 2.9697, "step": 334940 }, { "epoch": 0.008064, "grad_norm": 0.8759546279907227, "learning_rate": 9.88350625922231e-07, "loss": 2.7849, "step": 334950 }, { "epoch": 0.0080896, "grad_norm": 0.9765670299530029, "learning_rate": 9.88001585958378e-07, "loss": 2.6886, "step": 334960 }, { "epoch": 0.0081152, "grad_norm": 0.7959959506988525, "learning_rate": 9.876526044349399e-07, "loss": 2.8274, "step": 334970 }, { "epoch": 0.0081408, "grad_norm": 0.9681950211524963, "learning_rate": 9.873036813541815e-07, "loss": 2.7613, "step": 334980 }, { "epoch": 0.0081664, "grad_norm": 1.0424038171768188, "learning_rate": 9.869548167183606e-07, "loss": 2.7398, "step": 334990 }, { "epoch": 0.008192, "grad_norm": 0.83339923620224, "learning_rate": 9.866060105297437e-07, "loss": 2.8436, "step": 335000 }, { "epoch": 0.0082176, "grad_norm": 0.9160872101783752, "learning_rate": 9.862572627905898e-07, "loss": 2.8319, "step": 335010 }, { "epoch": 0.0082432, "grad_norm": 0.8842529654502869, "learning_rate": 9.859085735031626e-07, "loss": 2.8585, "step": 335020 }, { "epoch": 0.0082688, "grad_norm": 0.8761703372001648, "learning_rate": 9.855599426697226e-07, "loss": 2.7043, "step": 335030 }, { "epoch": 0.0082944, "grad_norm": 0.9330940842628479, "learning_rate": 9.852113702925314e-07, "loss": 2.8747, "step": 335040 }, { "epoch": 0.00832, "grad_norm": 0.8086165189743042, "learning_rate": 9.848628563738482e-07, "loss": 2.751, "step": 335050 }, { "epoch": 0.0083456, "grad_norm": 0.9981823563575745, "learning_rate": 9.845144009159325e-07, "loss": 2.9291, "step": 335060 }, { "epoch": 0.0083712, "grad_norm": 0.7835702896118164, "learning_rate": 9.841660039210465e-07, "loss": 2.8718, "step": 335070 }, { "epoch": 0.0083968, "grad_norm": 0.8520086407661438, "learning_rate": 9.838176653914466e-07, "loss": 2.7801, "step": 335080 }, { "epoch": 0.0084224, "grad_norm": 0.8102923035621643, "learning_rate": 9.834693853293931e-07, "loss": 2.7053, "step": 335090 }, { "epoch": 0.008448, "grad_norm": 0.908860445022583, "learning_rate": 9.831211637371452e-07, "loss": 2.8867, "step": 335100 }, { "epoch": 0.0084736, "grad_norm": 0.7997217774391174, "learning_rate": 9.82773000616961e-07, "loss": 2.5805, "step": 335110 }, { "epoch": 0.0084992, "grad_norm": 0.833793580532074, "learning_rate": 9.824248959710959e-07, "loss": 2.6894, "step": 335120 }, { "epoch": 0.0085248, "grad_norm": 0.7781314253807068, "learning_rate": 9.820768498018085e-07, "loss": 2.8067, "step": 335130 }, { "epoch": 0.0085504, "grad_norm": 0.8135252594947815, "learning_rate": 9.817288621113564e-07, "loss": 2.818, "step": 335140 }, { "epoch": 0.008576, "grad_norm": 0.8228139877319336, "learning_rate": 9.813809329019951e-07, "loss": 2.9977, "step": 335150 }, { "epoch": 0.0086016, "grad_norm": 0.758689820766449, "learning_rate": 9.810330621759835e-07, "loss": 2.8058, "step": 335160 }, { "epoch": 0.0086272, "grad_norm": 0.9241468906402588, "learning_rate": 9.806852499355735e-07, "loss": 2.9029, "step": 335170 }, { "epoch": 0.0086528, "grad_norm": 0.8523780107498169, "learning_rate": 9.803374961830258e-07, "loss": 2.8738, "step": 335180 }, { "epoch": 0.0086784, "grad_norm": 1.0044506788253784, "learning_rate": 9.7998980092059e-07, "loss": 2.7779, "step": 335190 }, { "epoch": 0.008704, "grad_norm": 0.7622063159942627, "learning_rate": 9.796421641505228e-07, "loss": 2.7198, "step": 335200 }, { "epoch": 0.0087296, "grad_norm": 0.8704033493995667, "learning_rate": 9.792945858750768e-07, "loss": 2.6677, "step": 335210 }, { "epoch": 0.0087552, "grad_norm": 0.863507091999054, "learning_rate": 9.789470660965106e-07, "loss": 2.6404, "step": 335220 }, { "epoch": 0.0087808, "grad_norm": 0.9472583532333374, "learning_rate": 9.785996048170743e-07, "loss": 2.9051, "step": 335230 }, { "epoch": 0.0088064, "grad_norm": 0.9694522619247437, "learning_rate": 9.78252202039025e-07, "loss": 2.7021, "step": 335240 }, { "epoch": 0.008832, "grad_norm": 0.7717681527137756, "learning_rate": 9.7790485776461e-07, "loss": 2.7641, "step": 335250 }, { "epoch": 0.0088576, "grad_norm": 0.9800170063972473, "learning_rate": 9.77557571996084e-07, "loss": 2.7303, "step": 335260 }, { "epoch": 0.0088832, "grad_norm": 1.0657161474227905, "learning_rate": 9.772103447356995e-07, "loss": 2.7191, "step": 335270 }, { "epoch": 0.0089088, "grad_norm": 0.8685394525527954, "learning_rate": 9.768631759857073e-07, "loss": 2.7971, "step": 335280 }, { "epoch": 0.0089344, "grad_norm": 0.9303746819496155, "learning_rate": 9.765160657483596e-07, "loss": 2.7197, "step": 335290 }, { "epoch": 0.00896, "grad_norm": 0.9432135820388794, "learning_rate": 9.761690140259083e-07, "loss": 2.7615, "step": 335300 }, { "epoch": 0.0089856, "grad_norm": 0.7528128623962402, "learning_rate": 9.758220208206037e-07, "loss": 2.7249, "step": 335310 }, { "epoch": 0.0090112, "grad_norm": 0.7652180194854736, "learning_rate": 9.75475086134693e-07, "loss": 2.7691, "step": 335320 }, { "epoch": 0.0090368, "grad_norm": 1.2747902870178223, "learning_rate": 9.751282099704274e-07, "loss": 3.0101, "step": 335330 }, { "epoch": 0.0090624, "grad_norm": 0.8070959448814392, "learning_rate": 9.747813923300576e-07, "loss": 2.687, "step": 335340 }, { "epoch": 0.009088, "grad_norm": 0.9451107382774353, "learning_rate": 9.744346332158317e-07, "loss": 2.9604, "step": 335350 }, { "epoch": 0.0091136, "grad_norm": 0.9892629384994507, "learning_rate": 9.740879326299979e-07, "loss": 2.9417, "step": 335360 }, { "epoch": 0.0091392, "grad_norm": 0.7953073382377625, "learning_rate": 9.737412905748056e-07, "loss": 2.8878, "step": 335370 }, { "epoch": 0.0091648, "grad_norm": 0.8691958785057068, "learning_rate": 9.73394707052503e-07, "loss": 2.6861, "step": 335380 }, { "epoch": 0.0091904, "grad_norm": 0.9576036930084229, "learning_rate": 9.730481820653347e-07, "loss": 2.9091, "step": 335390 }, { "epoch": 0.009216, "grad_norm": 0.8661149144172668, "learning_rate": 9.727017156155515e-07, "loss": 2.7415, "step": 335400 }, { "epoch": 0.0092416, "grad_norm": 0.8509852886199951, "learning_rate": 9.723553077053982e-07, "loss": 2.87, "step": 335410 }, { "epoch": 0.0092672, "grad_norm": 0.8261774778366089, "learning_rate": 9.720089583371217e-07, "loss": 2.7125, "step": 335420 }, { "epoch": 0.0092928, "grad_norm": 0.9641462564468384, "learning_rate": 9.71662667512967e-07, "loss": 2.9676, "step": 335430 }, { "epoch": 0.0093184, "grad_norm": 0.7956542372703552, "learning_rate": 9.713164352351822e-07, "loss": 2.8507, "step": 335440 }, { "epoch": 0.009344, "grad_norm": 1.0527042150497437, "learning_rate": 9.709702615060102e-07, "loss": 2.5276, "step": 335450 }, { "epoch": 0.0093696, "grad_norm": 0.85378098487854, "learning_rate": 9.706241463276954e-07, "loss": 2.671, "step": 335460 }, { "epoch": 0.0093952, "grad_norm": 0.8694398999214172, "learning_rate": 9.702780897024833e-07, "loss": 3.0146, "step": 335470 }, { "epoch": 0.0094208, "grad_norm": 0.7912400960922241, "learning_rate": 9.699320916326183e-07, "loss": 2.7971, "step": 335480 }, { "epoch": 0.0094464, "grad_norm": 0.8209524750709534, "learning_rate": 9.695861521203443e-07, "loss": 2.8455, "step": 335490 }, { "epoch": 0.009472, "grad_norm": 0.9628614187240601, "learning_rate": 9.692402711679027e-07, "loss": 2.8629, "step": 335500 }, { "epoch": 2.56e-05, "grad_norm": 0.8636911511421204, "learning_rate": 9.688944487775408e-07, "loss": 1.8788, "step": 335510 }, { "epoch": 5.12e-05, "grad_norm": 0.801515519618988, "learning_rate": 9.685486849514957e-07, "loss": 2.7425, "step": 335520 }, { "epoch": 7.68e-05, "grad_norm": 0.838252604007721, "learning_rate": 9.682029796920124e-07, "loss": 2.7392, "step": 335530 }, { "epoch": 0.0001024, "grad_norm": 0.8391468524932861, "learning_rate": 9.67857333001333e-07, "loss": 2.5141, "step": 335540 }, { "epoch": 0.000128, "grad_norm": 0.8809888958930969, "learning_rate": 9.675117448816972e-07, "loss": 2.5321, "step": 335550 }, { "epoch": 0.0001536, "grad_norm": 0.9265435338020325, "learning_rate": 9.671662153353477e-07, "loss": 2.6427, "step": 335560 }, { "epoch": 0.0001792, "grad_norm": 0.8936185240745544, "learning_rate": 9.66820744364525e-07, "loss": 2.4757, "step": 335570 }, { "epoch": 0.0002048, "grad_norm": 0.9377357959747314, "learning_rate": 9.664753319714692e-07, "loss": 2.7843, "step": 335580 }, { "epoch": 0.0002304, "grad_norm": 0.7903067469596863, "learning_rate": 9.6612997815842e-07, "loss": 2.4978, "step": 335590 }, { "epoch": 0.000256, "grad_norm": 0.8653144240379333, "learning_rate": 9.657846829276163e-07, "loss": 2.8654, "step": 335600 }, { "epoch": 0.0002816, "grad_norm": 0.8153936266899109, "learning_rate": 9.654394462812978e-07, "loss": 2.4429, "step": 335610 }, { "epoch": 0.0003072, "grad_norm": 0.8504683971405029, "learning_rate": 9.650942682217046e-07, "loss": 2.5305, "step": 335620 }, { "epoch": 0.0003328, "grad_norm": 1.3507426977157593, "learning_rate": 9.64749148751073e-07, "loss": 2.2533, "step": 335630 }, { "epoch": 0.0003584, "grad_norm": 0.8309908509254456, "learning_rate": 9.644040878716432e-07, "loss": 2.5604, "step": 335640 }, { "epoch": 0.000384, "grad_norm": 0.8605023622512817, "learning_rate": 9.640590855856502e-07, "loss": 2.9286, "step": 335650 }, { "epoch": 0.0004096, "grad_norm": 0.7818520069122314, "learning_rate": 9.637141418953332e-07, "loss": 2.5905, "step": 335660 }, { "epoch": 0.0004352, "grad_norm": 0.8158280849456787, "learning_rate": 9.633692568029273e-07, "loss": 2.6832, "step": 335670 }, { "epoch": 0.0004608, "grad_norm": 0.7678672671318054, "learning_rate": 9.630244303106718e-07, "loss": 2.6965, "step": 335680 }, { "epoch": 0.0004864, "grad_norm": 0.9699530601501465, "learning_rate": 9.626796624208e-07, "loss": 2.5064, "step": 335690 }, { "epoch": 0.000512, "grad_norm": 0.8619903922080994, "learning_rate": 9.623349531355485e-07, "loss": 2.6828, "step": 335700 }, { "epoch": 0.0005376, "grad_norm": 1.1208473443984985, "learning_rate": 9.619903024571554e-07, "loss": 2.5039, "step": 335710 }, { "epoch": 0.0005632, "grad_norm": 0.7847955822944641, "learning_rate": 9.616457103878507e-07, "loss": 2.5395, "step": 335720 }, { "epoch": 0.0005888, "grad_norm": 0.7786576747894287, "learning_rate": 9.613011769298696e-07, "loss": 2.4981, "step": 335730 }, { "epoch": 0.0006144, "grad_norm": 0.7719151377677917, "learning_rate": 9.609567020854504e-07, "loss": 2.526, "step": 335740 }, { "epoch": 0.00064, "grad_norm": 0.8571634888648987, "learning_rate": 9.606122858568246e-07, "loss": 2.9116, "step": 335750 }, { "epoch": 0.0006656, "grad_norm": 0.9282042384147644, "learning_rate": 9.602679282462279e-07, "loss": 2.7807, "step": 335760 }, { "epoch": 0.0006912, "grad_norm": 0.9395170211791992, "learning_rate": 9.599236292558888e-07, "loss": 3.0064, "step": 335770 }, { "epoch": 0.0007168, "grad_norm": 0.7276087999343872, "learning_rate": 9.595793888880434e-07, "loss": 2.5491, "step": 335780 }, { "epoch": 0.0007424, "grad_norm": 0.9187979698181152, "learning_rate": 9.592352071449218e-07, "loss": 2.6512, "step": 335790 }, { "epoch": 0.000768, "grad_norm": 0.8635950088500977, "learning_rate": 9.58891084028758e-07, "loss": 2.7586, "step": 335800 }, { "epoch": 0.0007936, "grad_norm": 0.8481006622314453, "learning_rate": 9.585470195417824e-07, "loss": 2.5352, "step": 335810 }, { "epoch": 0.0008192, "grad_norm": 1.2951366901397705, "learning_rate": 9.582030136862263e-07, "loss": 2.6177, "step": 335820 }, { "epoch": 0.0008448, "grad_norm": 0.8200516104698181, "learning_rate": 9.578590664643205e-07, "loss": 2.4076, "step": 335830 }, { "epoch": 0.0008704, "grad_norm": 1.0071823596954346, "learning_rate": 9.575151778782976e-07, "loss": 2.6599, "step": 335840 }, { "epoch": 0.000896, "grad_norm": 0.8155321478843689, "learning_rate": 9.571713479303834e-07, "loss": 2.4014, "step": 335850 }, { "epoch": 0.0009216, "grad_norm": 0.7861737012863159, "learning_rate": 9.568275766228097e-07, "loss": 2.4029, "step": 335860 }, { "epoch": 0.0009472, "grad_norm": 0.9294604659080505, "learning_rate": 9.564838639578067e-07, "loss": 2.9764, "step": 335870 }, { "epoch": 0.0009728, "grad_norm": 0.8296534419059753, "learning_rate": 9.561402099376016e-07, "loss": 2.7691, "step": 335880 }, { "epoch": 0.0009984, "grad_norm": 0.9203979969024658, "learning_rate": 9.557966145644238e-07, "loss": 2.5723, "step": 335890 }, { "epoch": 0.001024, "grad_norm": 0.9741610288619995, "learning_rate": 9.554530778405014e-07, "loss": 2.7218, "step": 335900 }, { "epoch": 0.0010496, "grad_norm": 1.0648390054702759, "learning_rate": 9.551095997680615e-07, "loss": 2.6139, "step": 335910 }, { "epoch": 0.0010752, "grad_norm": 0.9526680707931519, "learning_rate": 9.547661803493324e-07, "loss": 2.5705, "step": 335920 }, { "epoch": 0.0011008, "grad_norm": 0.8945907950401306, "learning_rate": 9.544228195865413e-07, "loss": 2.7964, "step": 335930 }, { "epoch": 0.0011264, "grad_norm": 0.931379497051239, "learning_rate": 9.54079517481914e-07, "loss": 2.9484, "step": 335940 }, { "epoch": 0.001152, "grad_norm": 0.7913095951080322, "learning_rate": 9.537362740376766e-07, "loss": 2.6549, "step": 335950 }, { "epoch": 0.0011776, "grad_norm": 0.8508655428886414, "learning_rate": 9.533930892560572e-07, "loss": 2.5489, "step": 335960 }, { "epoch": 0.0012032, "grad_norm": 0.8732805252075195, "learning_rate": 9.530499631392765e-07, "loss": 2.8368, "step": 335970 }, { "epoch": 0.0012288, "grad_norm": 0.7992335557937622, "learning_rate": 9.527068956895635e-07, "loss": 3.143, "step": 335980 }, { "epoch": 0.0012544, "grad_norm": 0.8322445750236511, "learning_rate": 9.523638869091412e-07, "loss": 2.7754, "step": 335990 }, { "epoch": 0.00128, "grad_norm": 0.8648340702056885, "learning_rate": 9.520209368002353e-07, "loss": 2.8129, "step": 336000 }, { "epoch": 0.0013056, "grad_norm": 0.8428200483322144, "learning_rate": 9.516780453650687e-07, "loss": 2.7473, "step": 336010 }, { "epoch": 0.0013312, "grad_norm": 0.8116957545280457, "learning_rate": 9.51335212605865e-07, "loss": 2.5802, "step": 336020 }, { "epoch": 0.0013568, "grad_norm": 0.8892996907234192, "learning_rate": 9.50992438524847e-07, "loss": 2.5087, "step": 336030 }, { "epoch": 0.0013824, "grad_norm": 0.8188561797142029, "learning_rate": 9.506497231242407e-07, "loss": 2.6308, "step": 336040 }, { "epoch": 0.001408, "grad_norm": 0.849945068359375, "learning_rate": 9.503070664062641e-07, "loss": 2.5262, "step": 336050 }, { "epoch": 0.0014336, "grad_norm": 0.838270366191864, "learning_rate": 9.499644683731413e-07, "loss": 2.5624, "step": 336060 }, { "epoch": 0.0014592, "grad_norm": 0.9662643074989319, "learning_rate": 9.496219290270925e-07, "loss": 2.5755, "step": 336070 }, { "epoch": 0.0014848, "grad_norm": 0.8591448664665222, "learning_rate": 9.492794483703394e-07, "loss": 2.4074, "step": 336080 }, { "epoch": 0.0015104, "grad_norm": 0.9128295183181763, "learning_rate": 9.489370264051067e-07, "loss": 2.7756, "step": 336090 }, { "epoch": 0.001536, "grad_norm": 0.8242028951644897, "learning_rate": 9.485946631336107e-07, "loss": 2.2949, "step": 336100 }, { "epoch": 0.0015616, "grad_norm": 1.1304962635040283, "learning_rate": 9.482523585580727e-07, "loss": 3.0214, "step": 336110 }, { "epoch": 0.0015872, "grad_norm": 0.8541518449783325, "learning_rate": 9.479101126807122e-07, "loss": 2.6668, "step": 336120 }, { "epoch": 0.0016128, "grad_norm": 0.9066085815429688, "learning_rate": 9.475679255037495e-07, "loss": 2.8081, "step": 336130 }, { "epoch": 0.0016384, "grad_norm": 1.0178251266479492, "learning_rate": 9.472257970294019e-07, "loss": 2.5888, "step": 336140 }, { "epoch": 0.001664, "grad_norm": 0.8348689675331116, "learning_rate": 9.468837272598897e-07, "loss": 2.8373, "step": 336150 }, { "epoch": 0.0016896, "grad_norm": 0.9378101825714111, "learning_rate": 9.465417161974321e-07, "loss": 2.88, "step": 336160 }, { "epoch": 0.0017152, "grad_norm": 0.9559800624847412, "learning_rate": 9.461997638442433e-07, "loss": 2.691, "step": 336170 }, { "epoch": 0.0017408, "grad_norm": 0.9162042140960693, "learning_rate": 9.458578702025433e-07, "loss": 2.7943, "step": 336180 }, { "epoch": 0.0017664, "grad_norm": 0.815635621547699, "learning_rate": 9.455160352745485e-07, "loss": 2.5548, "step": 336190 }, { "epoch": 0.001792, "grad_norm": 0.8350225687026978, "learning_rate": 9.451742590624758e-07, "loss": 2.6847, "step": 336200 }, { "epoch": 0.0018176, "grad_norm": 0.7887085676193237, "learning_rate": 9.448325415685422e-07, "loss": 2.4751, "step": 336210 }, { "epoch": 0.0018432, "grad_norm": 0.8846010565757751, "learning_rate": 9.444908827949628e-07, "loss": 2.4939, "step": 336220 }, { "epoch": 0.0018688, "grad_norm": 0.8888088464736938, "learning_rate": 9.441492827439547e-07, "loss": 2.4829, "step": 336230 }, { "epoch": 0.0018944, "grad_norm": 0.9405959844589233, "learning_rate": 9.438077414177293e-07, "loss": 2.6285, "step": 336240 }, { "epoch": 0.00192, "grad_norm": 0.8252403736114502, "learning_rate": 9.434662588185028e-07, "loss": 2.606, "step": 336250 }, { "epoch": 0.0019456, "grad_norm": 0.8181023001670837, "learning_rate": 9.431248349484923e-07, "loss": 2.7685, "step": 336260 }, { "epoch": 0.0019712, "grad_norm": 0.8649290204048157, "learning_rate": 9.427834698099103e-07, "loss": 2.6714, "step": 336270 }, { "epoch": 0.0019968, "grad_norm": 2.213848352432251, "learning_rate": 9.424421634049707e-07, "loss": 2.3846, "step": 336280 }, { "epoch": 0.0020224, "grad_norm": 1.6140196323394775, "learning_rate": 9.421009157358884e-07, "loss": 2.8789, "step": 336290 }, { "epoch": 0.002048, "grad_norm": 0.7987664341926575, "learning_rate": 9.417597268048717e-07, "loss": 2.59, "step": 336300 }, { "epoch": 0.0020736, "grad_norm": 0.7665171027183533, "learning_rate": 9.414185966141365e-07, "loss": 2.3529, "step": 336310 }, { "epoch": 0.0020992, "grad_norm": 0.8136526346206665, "learning_rate": 9.410775251658943e-07, "loss": 2.5275, "step": 336320 }, { "epoch": 0.0021248, "grad_norm": 0.97820645570755, "learning_rate": 9.407365124623557e-07, "loss": 2.7748, "step": 336330 }, { "epoch": 0.0021504, "grad_norm": 0.8156145811080933, "learning_rate": 9.403955585057345e-07, "loss": 2.8612, "step": 336340 }, { "epoch": 0.002176, "grad_norm": 0.9117639064788818, "learning_rate": 9.4005466329824e-07, "loss": 2.9269, "step": 336350 }, { "epoch": 0.0022016, "grad_norm": 0.8309614658355713, "learning_rate": 9.397138268420846e-07, "loss": 2.9216, "step": 336360 }, { "epoch": 0.0022272, "grad_norm": 0.8508558869361877, "learning_rate": 9.393730491394748e-07, "loss": 2.7492, "step": 336370 }, { "epoch": 0.0022528, "grad_norm": 0.8589122891426086, "learning_rate": 9.390323301926241e-07, "loss": 2.6447, "step": 336380 }, { "epoch": 0.0022784, "grad_norm": 0.8439503908157349, "learning_rate": 9.386916700037396e-07, "loss": 3.1515, "step": 336390 }, { "epoch": 0.002304, "grad_norm": 0.7851011753082275, "learning_rate": 9.383510685750319e-07, "loss": 2.3801, "step": 336400 }, { "epoch": 0.0023296, "grad_norm": 1.1476714611053467, "learning_rate": 9.38010525908708e-07, "loss": 2.3383, "step": 336410 }, { "epoch": 0.0023552, "grad_norm": 0.8901497721672058, "learning_rate": 9.376700420069784e-07, "loss": 2.8815, "step": 336420 }, { "epoch": 0.0023808, "grad_norm": 0.8500980138778687, "learning_rate": 9.373296168720503e-07, "loss": 2.6023, "step": 336430 }, { "epoch": 0.0024064, "grad_norm": 1.0112073421478271, "learning_rate": 9.369892505061318e-07, "loss": 2.7964, "step": 336440 }, { "epoch": 0.002432, "grad_norm": 0.7474319934844971, "learning_rate": 9.366489429114278e-07, "loss": 2.6997, "step": 336450 }, { "epoch": 0.0024576, "grad_norm": 0.7898051142692566, "learning_rate": 9.363086940901478e-07, "loss": 2.8109, "step": 336460 }, { "epoch": 0.0024832, "grad_norm": 0.7812632918357849, "learning_rate": 9.359685040444966e-07, "loss": 2.6698, "step": 336470 }, { "epoch": 0.0025088, "grad_norm": 0.7746760845184326, "learning_rate": 9.356283727766802e-07, "loss": 2.58, "step": 336480 }, { "epoch": 0.0025344, "grad_norm": 0.8187030553817749, "learning_rate": 9.352883002889068e-07, "loss": 2.9188, "step": 336490 }, { "epoch": 0.00256, "grad_norm": 0.8478170037269592, "learning_rate": 9.349482865833792e-07, "loss": 2.7564, "step": 336500 }, { "epoch": 0.0025856, "grad_norm": 0.8959540128707886, "learning_rate": 9.346083316623012e-07, "loss": 2.4244, "step": 336510 }, { "epoch": 0.0026112, "grad_norm": 0.887397050857544, "learning_rate": 9.342684355278797e-07, "loss": 2.8381, "step": 336520 }, { "epoch": 0.0026368, "grad_norm": 0.879111111164093, "learning_rate": 9.339285981823176e-07, "loss": 2.7577, "step": 336530 }, { "epoch": 0.0026624, "grad_norm": 0.8721051812171936, "learning_rate": 9.335888196278186e-07, "loss": 2.8249, "step": 336540 }, { "epoch": 0.002688, "grad_norm": 0.8331382870674133, "learning_rate": 9.332490998665877e-07, "loss": 2.8197, "step": 336550 }, { "epoch": 0.0027136, "grad_norm": 0.9316394925117493, "learning_rate": 9.329094389008275e-07, "loss": 2.8422, "step": 336560 }, { "epoch": 0.0027392, "grad_norm": 0.7997321486473083, "learning_rate": 9.325698367327385e-07, "loss": 2.8384, "step": 336570 }, { "epoch": 0.0027648, "grad_norm": 1.0421295166015625, "learning_rate": 9.322302933645244e-07, "loss": 2.7644, "step": 336580 }, { "epoch": 0.0027904, "grad_norm": 0.8146153092384338, "learning_rate": 9.318908087983869e-07, "loss": 2.8725, "step": 336590 }, { "epoch": 0.002816, "grad_norm": 1.2681916952133179, "learning_rate": 9.315513830365253e-07, "loss": 3.0795, "step": 336600 }, { "epoch": 0.0028416, "grad_norm": 0.7911128997802734, "learning_rate": 9.312120160811455e-07, "loss": 2.757, "step": 336610 }, { "epoch": 0.0028672, "grad_norm": 0.8161876201629639, "learning_rate": 9.308727079344471e-07, "loss": 2.8239, "step": 336620 }, { "epoch": 0.0028928, "grad_norm": 0.7567076683044434, "learning_rate": 9.305334585986259e-07, "loss": 2.6791, "step": 336630 }, { "epoch": 0.0029184, "grad_norm": 0.789221465587616, "learning_rate": 9.301942680758858e-07, "loss": 2.8374, "step": 336640 }, { "epoch": 0.002944, "grad_norm": 0.8784273266792297, "learning_rate": 9.298551363684261e-07, "loss": 2.7473, "step": 336650 }, { "epoch": 0.0029696, "grad_norm": 0.8614183664321899, "learning_rate": 9.295160634784439e-07, "loss": 2.7414, "step": 336660 }, { "epoch": 0.0029952, "grad_norm": 0.839576005935669, "learning_rate": 9.291770494081409e-07, "loss": 2.9691, "step": 336670 }, { "epoch": 0.0030208, "grad_norm": 0.8140537142753601, "learning_rate": 9.288380941597131e-07, "loss": 2.7041, "step": 336680 }, { "epoch": 0.0030464, "grad_norm": 0.8268940448760986, "learning_rate": 9.284991977353608e-07, "loss": 2.8822, "step": 336690 }, { "epoch": 0.003072, "grad_norm": 1.1345857381820679, "learning_rate": 9.281603601372791e-07, "loss": 2.9828, "step": 336700 }, { "epoch": 0.0030976, "grad_norm": 0.8354672193527222, "learning_rate": 9.278215813676661e-07, "loss": 2.7957, "step": 336710 }, { "epoch": 0.0031232, "grad_norm": 0.784733235836029, "learning_rate": 9.274828614287201e-07, "loss": 2.6415, "step": 336720 }, { "epoch": 0.0031488, "grad_norm": 0.8425770401954651, "learning_rate": 9.27144200322635e-07, "loss": 2.7539, "step": 336730 }, { "epoch": 0.0031744, "grad_norm": 0.913815438747406, "learning_rate": 9.2680559805161e-07, "loss": 2.858, "step": 336740 }, { "epoch": 0.0032, "grad_norm": 0.8885003328323364, "learning_rate": 9.264670546178378e-07, "loss": 2.9025, "step": 336750 }, { "epoch": 0.0032256, "grad_norm": 0.8011424541473389, "learning_rate": 9.261285700235179e-07, "loss": 2.7655, "step": 336760 }, { "epoch": 0.0032512, "grad_norm": 0.7786352634429932, "learning_rate": 9.257901442708417e-07, "loss": 2.8213, "step": 336770 }, { "epoch": 0.0032768, "grad_norm": 0.9168210029602051, "learning_rate": 9.254517773620019e-07, "loss": 2.6197, "step": 336780 }, { "epoch": 0.0033024, "grad_norm": 0.8719581365585327, "learning_rate": 9.25113469299197e-07, "loss": 2.6436, "step": 336790 }, { "epoch": 0.003328, "grad_norm": 0.8768317699432373, "learning_rate": 9.247752200846205e-07, "loss": 2.6411, "step": 336800 }, { "epoch": 0.0033536, "grad_norm": 0.8226779699325562, "learning_rate": 9.244370297204642e-07, "loss": 2.781, "step": 336810 }, { "epoch": 0.0033792, "grad_norm": 0.9205424785614014, "learning_rate": 9.240988982089238e-07, "loss": 2.6066, "step": 336820 }, { "epoch": 0.0034048, "grad_norm": 0.7801161408424377, "learning_rate": 9.237608255521879e-07, "loss": 2.5793, "step": 336830 }, { "epoch": 0.0034304, "grad_norm": 0.8639826774597168, "learning_rate": 9.234228117524513e-07, "loss": 2.7794, "step": 336840 }, { "epoch": 0.003456, "grad_norm": 0.932092547416687, "learning_rate": 9.230848568119056e-07, "loss": 2.999, "step": 336850 }, { "epoch": 0.0034816, "grad_norm": 0.8284237384796143, "learning_rate": 9.227469607327411e-07, "loss": 2.8062, "step": 336860 }, { "epoch": 0.0035072, "grad_norm": 0.8613781929016113, "learning_rate": 9.224091235171506e-07, "loss": 2.6379, "step": 336870 }, { "epoch": 0.0035328, "grad_norm": 0.8518073558807373, "learning_rate": 9.220713451673246e-07, "loss": 2.76, "step": 336880 }, { "epoch": 0.0035584, "grad_norm": 0.9705018997192383, "learning_rate": 9.217336256854559e-07, "loss": 2.5598, "step": 336890 }, { "epoch": 0.003584, "grad_norm": 0.9649633169174194, "learning_rate": 9.213959650737292e-07, "loss": 3.0067, "step": 336900 }, { "epoch": 0.0036096, "grad_norm": 0.8802932500839233, "learning_rate": 9.210583633343373e-07, "loss": 2.7831, "step": 336910 }, { "epoch": 0.0036352, "grad_norm": 0.8400582075119019, "learning_rate": 9.207208204694684e-07, "loss": 2.806, "step": 336920 }, { "epoch": 0.0036608, "grad_norm": 5.087880611419678, "learning_rate": 9.20383336481313e-07, "loss": 2.9487, "step": 336930 }, { "epoch": 0.0036864, "grad_norm": 0.8280954360961914, "learning_rate": 9.200459113720572e-07, "loss": 2.829, "step": 336940 }, { "epoch": 0.003712, "grad_norm": 0.8248967528343201, "learning_rate": 9.197085451438915e-07, "loss": 2.6074, "step": 336950 }, { "epoch": 0.0037376, "grad_norm": 0.8001227378845215, "learning_rate": 9.193712377990027e-07, "loss": 2.8943, "step": 336960 }, { "epoch": 0.0037632, "grad_norm": 0.9971824884414673, "learning_rate": 9.190339893395783e-07, "loss": 2.5882, "step": 336970 }, { "epoch": 0.0037888, "grad_norm": 0.9619799256324768, "learning_rate": 9.186967997678043e-07, "loss": 2.9368, "step": 336980 }, { "epoch": 0.0038144, "grad_norm": 0.8591087460517883, "learning_rate": 9.183596690858676e-07, "loss": 2.7732, "step": 336990 }, { "epoch": 0.00384, "grad_norm": 0.8375579714775085, "learning_rate": 9.180225972959556e-07, "loss": 2.7934, "step": 337000 }, { "epoch": 0.0038656, "grad_norm": 0.8561888933181763, "learning_rate": 9.176855844002541e-07, "loss": 2.8685, "step": 337010 }, { "epoch": 0.0038912, "grad_norm": 0.7947531938552856, "learning_rate": 9.173486304009482e-07, "loss": 2.9071, "step": 337020 }, { "epoch": 0.0039168, "grad_norm": 0.8600497841835022, "learning_rate": 9.170117353002217e-07, "loss": 2.801, "step": 337030 }, { "epoch": 0.0039424, "grad_norm": 1.2241394519805908, "learning_rate": 9.166748991002594e-07, "loss": 2.7269, "step": 337040 }, { "epoch": 0.003968, "grad_norm": 0.8828281760215759, "learning_rate": 9.163381218032474e-07, "loss": 2.9541, "step": 337050 }, { "epoch": 0.0039936, "grad_norm": 0.7608388066291809, "learning_rate": 9.160014034113674e-07, "loss": 2.7116, "step": 337060 }, { "epoch": 0.0040192, "grad_norm": 0.8052069544792175, "learning_rate": 9.156647439268041e-07, "loss": 2.8239, "step": 337070 }, { "epoch": 0.0040448, "grad_norm": 0.8417986035346985, "learning_rate": 9.153281433517414e-07, "loss": 2.8377, "step": 337080 }, { "epoch": 0.0040704, "grad_norm": 0.8660770058631897, "learning_rate": 9.149916016883631e-07, "loss": 2.8467, "step": 337090 }, { "epoch": 0.004096, "grad_norm": 0.7910860180854797, "learning_rate": 9.146551189388475e-07, "loss": 2.8495, "step": 337100 }, { "epoch": 0.0041216, "grad_norm": 0.9785885810852051, "learning_rate": 9.143186951053784e-07, "loss": 3.0618, "step": 337110 }, { "epoch": 0.0041472, "grad_norm": 0.9623923301696777, "learning_rate": 9.139823301901363e-07, "loss": 2.8985, "step": 337120 }, { "epoch": 0.0041728, "grad_norm": 0.8439671993255615, "learning_rate": 9.13646024195306e-07, "loss": 2.6732, "step": 337130 }, { "epoch": 0.0041984, "grad_norm": 0.9221177101135254, "learning_rate": 9.13309777123067e-07, "loss": 2.8734, "step": 337140 }, { "epoch": 0.004224, "grad_norm": 0.8620396852493286, "learning_rate": 9.129735889755997e-07, "loss": 2.807, "step": 337150 }, { "epoch": 0.0042496, "grad_norm": 0.8081175088882446, "learning_rate": 9.126374597550824e-07, "loss": 2.9942, "step": 337160 }, { "epoch": 0.0042752, "grad_norm": 0.8168120980262756, "learning_rate": 9.123013894636967e-07, "loss": 2.8005, "step": 337170 }, { "epoch": 0.0043008, "grad_norm": 0.7771791815757751, "learning_rate": 9.119653781036208e-07, "loss": 2.708, "step": 337180 }, { "epoch": 0.0043264, "grad_norm": 0.8407449126243591, "learning_rate": 9.116294256770342e-07, "loss": 2.8169, "step": 337190 }, { "epoch": 0.004352, "grad_norm": 0.8368179798126221, "learning_rate": 9.11293532186115e-07, "loss": 2.5924, "step": 337200 }, { "epoch": 0.0043776, "grad_norm": 0.9032500982284546, "learning_rate": 9.109576976330414e-07, "loss": 2.9088, "step": 337210 }, { "epoch": 0.0044032, "grad_norm": 0.8505740165710449, "learning_rate": 9.106219220199941e-07, "loss": 2.7795, "step": 337220 }, { "epoch": 0.0044288, "grad_norm": 0.7885172367095947, "learning_rate": 9.102862053491468e-07, "loss": 2.573, "step": 337230 }, { "epoch": 0.0044544, "grad_norm": 0.8196792006492615, "learning_rate": 9.099505476226766e-07, "loss": 2.9333, "step": 337240 }, { "epoch": 0.00448, "grad_norm": 0.8921248912811279, "learning_rate": 9.096149488427608e-07, "loss": 2.8903, "step": 337250 }, { "epoch": 0.0045056, "grad_norm": 0.8269745111465454, "learning_rate": 9.092794090115775e-07, "loss": 2.7969, "step": 337260 }, { "epoch": 0.0045312, "grad_norm": 0.9089165329933167, "learning_rate": 9.089439281313006e-07, "loss": 2.7938, "step": 337270 }, { "epoch": 0.0045568, "grad_norm": 0.7557870745658875, "learning_rate": 9.086085062041062e-07, "loss": 2.8808, "step": 337280 }, { "epoch": 0.0045824, "grad_norm": 0.8299180269241333, "learning_rate": 9.082731432321712e-07, "loss": 2.6993, "step": 337290 }, { "epoch": 0.004608, "grad_norm": 0.8162498474121094, "learning_rate": 9.079378392176652e-07, "loss": 2.6164, "step": 337300 }, { "epoch": 0.0046336, "grad_norm": 0.8254241347312927, "learning_rate": 9.076025941627675e-07, "loss": 2.9587, "step": 337310 }, { "epoch": 0.0046592, "grad_norm": 0.8514885306358337, "learning_rate": 9.072674080696497e-07, "loss": 2.8306, "step": 337320 }, { "epoch": 0.0046848, "grad_norm": 0.8152075409889221, "learning_rate": 9.069322809404879e-07, "loss": 2.7373, "step": 337330 }, { "epoch": 0.0047104, "grad_norm": 0.8019256591796875, "learning_rate": 9.065972127774524e-07, "loss": 2.8526, "step": 337340 }, { "epoch": 0.004736, "grad_norm": 0.8406617641448975, "learning_rate": 9.062622035827195e-07, "loss": 3.189, "step": 337350 }, { "epoch": 0.0047616, "grad_norm": 0.7513964176177979, "learning_rate": 9.059272533584573e-07, "loss": 2.7415, "step": 337360 }, { "epoch": 0.0047872, "grad_norm": 0.858073890209198, "learning_rate": 9.055923621068396e-07, "loss": 2.7542, "step": 337370 }, { "epoch": 0.0048128, "grad_norm": 0.8911199569702148, "learning_rate": 9.052575298300393e-07, "loss": 2.8014, "step": 337380 }, { "epoch": 0.0048384, "grad_norm": 0.7797344923019409, "learning_rate": 9.049227565302265e-07, "loss": 2.7509, "step": 337390 }, { "epoch": 0.004864, "grad_norm": 0.8686263561248779, "learning_rate": 9.045880422095721e-07, "loss": 2.8899, "step": 337400 }, { "epoch": 0.0048896, "grad_norm": 0.8460330367088318, "learning_rate": 9.042533868702474e-07, "loss": 2.6107, "step": 337410 }, { "epoch": 0.0049152, "grad_norm": 0.7969167232513428, "learning_rate": 9.039187905144242e-07, "loss": 2.753, "step": 337420 }, { "epoch": 0.0049408, "grad_norm": 0.8018032312393188, "learning_rate": 9.035842531442684e-07, "loss": 2.7675, "step": 337430 }, { "epoch": 0.0049664, "grad_norm": 0.7723913192749023, "learning_rate": 9.032497747619506e-07, "loss": 2.7181, "step": 337440 }, { "epoch": 0.004992, "grad_norm": 0.822500467300415, "learning_rate": 9.029153553696402e-07, "loss": 2.7814, "step": 337450 }, { "epoch": 0.0050176, "grad_norm": 0.7801783084869385, "learning_rate": 9.025809949695063e-07, "loss": 2.7014, "step": 337460 }, { "epoch": 0.0050432, "grad_norm": 4.304584503173828, "learning_rate": 9.022466935637176e-07, "loss": 2.5699, "step": 337470 }, { "epoch": 0.0050688, "grad_norm": 0.876761257648468, "learning_rate": 9.019124511544397e-07, "loss": 2.6927, "step": 337480 }, { "epoch": 0.0050944, "grad_norm": 0.920875072479248, "learning_rate": 9.015782677438422e-07, "loss": 2.8845, "step": 337490 }, { "epoch": 0.00512, "grad_norm": 0.8621246814727783, "learning_rate": 9.012441433340923e-07, "loss": 3.0, "step": 337500 }, { "epoch": 0.0051456, "grad_norm": 0.8417222499847412, "learning_rate": 9.00910077927355e-07, "loss": 2.7521, "step": 337510 }, { "epoch": 0.0051712, "grad_norm": 0.8502156734466553, "learning_rate": 9.005760715257983e-07, "loss": 2.834, "step": 337520 }, { "epoch": 0.0051968, "grad_norm": 0.8629723191261292, "learning_rate": 9.002421241315873e-07, "loss": 2.5318, "step": 337530 }, { "epoch": 0.0052224, "grad_norm": 0.7726938724517822, "learning_rate": 8.999082357468892e-07, "loss": 2.8318, "step": 337540 }, { "epoch": 0.005248, "grad_norm": 0.9564240574836731, "learning_rate": 8.995744063738665e-07, "loss": 2.8742, "step": 337550 }, { "epoch": 0.0052736, "grad_norm": 0.7908117771148682, "learning_rate": 8.992406360146844e-07, "loss": 2.9293, "step": 337560 }, { "epoch": 0.0052992, "grad_norm": 0.8472579717636108, "learning_rate": 8.989069246715088e-07, "loss": 2.644, "step": 337570 }, { "epoch": 0.0053248, "grad_norm": 0.8332052826881409, "learning_rate": 8.985732723465024e-07, "loss": 2.7633, "step": 337580 }, { "epoch": 0.0053504, "grad_norm": 1.180313229560852, "learning_rate": 8.982396790418291e-07, "loss": 2.8241, "step": 337590 }, { "epoch": 0.005376, "grad_norm": 0.8084332346916199, "learning_rate": 8.979061447596537e-07, "loss": 2.8947, "step": 337600 }, { "epoch": 0.0054016, "grad_norm": 0.8823037147521973, "learning_rate": 8.97572669502137e-07, "loss": 2.6157, "step": 337610 }, { "epoch": 0.0054272, "grad_norm": 0.8448503017425537, "learning_rate": 8.972392532714447e-07, "loss": 2.7731, "step": 337620 }, { "epoch": 0.0054528, "grad_norm": 0.7736420035362244, "learning_rate": 8.969058960697341e-07, "loss": 2.8604, "step": 337630 }, { "epoch": 0.0054784, "grad_norm": 0.8950115442276001, "learning_rate": 8.965725978991702e-07, "loss": 2.8434, "step": 337640 }, { "epoch": 0.005504, "grad_norm": 0.8368809223175049, "learning_rate": 8.962393587619111e-07, "loss": 2.611, "step": 337650 }, { "epoch": 0.0055296, "grad_norm": 0.8178136348724365, "learning_rate": 8.959061786601231e-07, "loss": 2.8075, "step": 337660 }, { "epoch": 0.0055552, "grad_norm": 0.9804081320762634, "learning_rate": 8.955730575959653e-07, "loss": 2.6188, "step": 337670 }, { "epoch": 0.0055808, "grad_norm": 0.955563485622406, "learning_rate": 8.95239995571594e-07, "loss": 3.0297, "step": 337680 }, { "epoch": 0.0056064, "grad_norm": 0.9554051160812378, "learning_rate": 8.949069925891729e-07, "loss": 2.8693, "step": 337690 }, { "epoch": 0.005632, "grad_norm": 0.9666185975074768, "learning_rate": 8.945740486508591e-07, "loss": 2.7353, "step": 337700 }, { "epoch": 0.0056576, "grad_norm": 0.8014793395996094, "learning_rate": 8.942411637588133e-07, "loss": 2.7808, "step": 337710 }, { "epoch": 0.0056832, "grad_norm": 0.8482881188392639, "learning_rate": 8.939083379151925e-07, "loss": 2.7021, "step": 337720 }, { "epoch": 0.0057088, "grad_norm": 0.8811891674995422, "learning_rate": 8.935755711221572e-07, "loss": 2.8265, "step": 337730 }, { "epoch": 0.0057344, "grad_norm": 0.9473466873168945, "learning_rate": 8.932428633818657e-07, "loss": 3.0683, "step": 337740 }, { "epoch": 0.00576, "grad_norm": 1.0202326774597168, "learning_rate": 8.929102146964718e-07, "loss": 2.8492, "step": 337750 }, { "epoch": 0.0057856, "grad_norm": 0.8086302280426025, "learning_rate": 8.925776250681351e-07, "loss": 2.6907, "step": 337760 }, { "epoch": 0.0058112, "grad_norm": 0.8006066083908081, "learning_rate": 8.922450944990114e-07, "loss": 2.7495, "step": 337770 }, { "epoch": 0.0058368, "grad_norm": 0.8522638082504272, "learning_rate": 8.919126229912578e-07, "loss": 2.7221, "step": 337780 }, { "epoch": 0.0058624, "grad_norm": 0.8616777062416077, "learning_rate": 8.915802105470306e-07, "loss": 2.7753, "step": 337790 }, { "epoch": 0.005888, "grad_norm": 0.810090184211731, "learning_rate": 8.912478571684858e-07, "loss": 2.732, "step": 337800 }, { "epoch": 0.0059136, "grad_norm": 0.8084713816642761, "learning_rate": 8.909155628577771e-07, "loss": 2.7851, "step": 337810 }, { "epoch": 0.0059392, "grad_norm": 0.8380578756332397, "learning_rate": 8.905833276170594e-07, "loss": 2.8049, "step": 337820 }, { "epoch": 0.0059648, "grad_norm": 0.7989261746406555, "learning_rate": 8.90251151448488e-07, "loss": 2.6254, "step": 337830 }, { "epoch": 0.0059904, "grad_norm": 0.8455715179443359, "learning_rate": 8.899190343542175e-07, "loss": 2.6888, "step": 337840 }, { "epoch": 0.006016, "grad_norm": 0.8615542650222778, "learning_rate": 8.895869763364007e-07, "loss": 2.8458, "step": 337850 }, { "epoch": 0.0060416, "grad_norm": 0.7700750231742859, "learning_rate": 8.892549773971903e-07, "loss": 2.8311, "step": 337860 }, { "epoch": 0.0060672, "grad_norm": 0.9425415396690369, "learning_rate": 8.889230375387425e-07, "loss": 2.6111, "step": 337870 }, { "epoch": 0.0060928, "grad_norm": 0.9065897464752197, "learning_rate": 8.885911567632066e-07, "loss": 2.8743, "step": 337880 }, { "epoch": 0.0061184, "grad_norm": 0.860680341720581, "learning_rate": 8.882593350727342e-07, "loss": 2.7306, "step": 337890 }, { "epoch": 0.006144, "grad_norm": 0.8517003655433655, "learning_rate": 8.87927572469478e-07, "loss": 2.8743, "step": 337900 }, { "epoch": 0.0061696, "grad_norm": 0.8514701724052429, "learning_rate": 8.875958689555908e-07, "loss": 2.8148, "step": 337910 }, { "epoch": 0.0061952, "grad_norm": 0.8968101143836975, "learning_rate": 8.872642245332219e-07, "loss": 2.7905, "step": 337920 }, { "epoch": 0.0062208, "grad_norm": 0.8394189476966858, "learning_rate": 8.869326392045241e-07, "loss": 2.8268, "step": 337930 }, { "epoch": 0.0062464, "grad_norm": 1.22300386428833, "learning_rate": 8.866011129716468e-07, "loss": 2.4254, "step": 337940 }, { "epoch": 0.006272, "grad_norm": 0.7740250825881958, "learning_rate": 8.86269645836737e-07, "loss": 2.8311, "step": 337950 }, { "epoch": 0.0062976, "grad_norm": 0.8295902609825134, "learning_rate": 8.859382378019465e-07, "loss": 2.7448, "step": 337960 }, { "epoch": 0.0063232, "grad_norm": 0.886586606502533, "learning_rate": 8.856068888694258e-07, "loss": 2.8512, "step": 337970 }, { "epoch": 0.0063488, "grad_norm": 0.7858036756515503, "learning_rate": 8.852755990413208e-07, "loss": 2.7508, "step": 337980 }, { "epoch": 0.0063744, "grad_norm": 0.7942598462104797, "learning_rate": 8.84944368319779e-07, "loss": 2.7238, "step": 337990 }, { "epoch": 0.0064, "grad_norm": 0.9884633421897888, "learning_rate": 8.846131967069549e-07, "loss": 2.9844, "step": 338000 }, { "epoch": 0.0064256, "grad_norm": 0.9866311550140381, "learning_rate": 8.842820842049892e-07, "loss": 2.7786, "step": 338010 }, { "epoch": 0.0064512, "grad_norm": 0.9498933553695679, "learning_rate": 8.839510308160327e-07, "loss": 2.8329, "step": 338020 }, { "epoch": 0.0064768, "grad_norm": 0.7749385833740234, "learning_rate": 8.836200365422299e-07, "loss": 2.7272, "step": 338030 }, { "epoch": 0.0065024, "grad_norm": 1.066606044769287, "learning_rate": 8.832891013857281e-07, "loss": 2.8408, "step": 338040 }, { "epoch": 0.006528, "grad_norm": 0.8342379331588745, "learning_rate": 8.829582253486746e-07, "loss": 2.8507, "step": 338050 }, { "epoch": 0.0065536, "grad_norm": 0.8168785572052002, "learning_rate": 8.826274084332131e-07, "loss": 2.7273, "step": 338060 }, { "epoch": 0.0065792, "grad_norm": 0.858957827091217, "learning_rate": 8.822966506414921e-07, "loss": 2.6418, "step": 338070 }, { "epoch": 0.0066048, "grad_norm": 0.8211709856987, "learning_rate": 8.819659519756519e-07, "loss": 2.7307, "step": 338080 }, { "epoch": 0.0066304, "grad_norm": 0.8501774668693542, "learning_rate": 8.816353124378396e-07, "loss": 2.718, "step": 338090 }, { "epoch": 0.006656, "grad_norm": 1.050728440284729, "learning_rate": 8.813047320301993e-07, "loss": 2.8851, "step": 338100 }, { "epoch": 0.0066816, "grad_norm": 1.1049411296844482, "learning_rate": 8.809742107548736e-07, "loss": 3.0487, "step": 338110 }, { "epoch": 0.0067072, "grad_norm": 0.9645025730133057, "learning_rate": 8.806437486140074e-07, "loss": 2.8323, "step": 338120 }, { "epoch": 0.0067328, "grad_norm": 0.8153219819068909, "learning_rate": 8.803133456097412e-07, "loss": 2.6854, "step": 338130 }, { "epoch": 0.0067584, "grad_norm": 1.187598466873169, "learning_rate": 8.799830017442223e-07, "loss": 2.7773, "step": 338140 }, { "epoch": 0.006784, "grad_norm": 0.8611119389533997, "learning_rate": 8.796527170195879e-07, "loss": 2.7456, "step": 338150 }, { "epoch": 0.0068096, "grad_norm": 0.8019713163375854, "learning_rate": 8.793224914379816e-07, "loss": 2.8187, "step": 338160 }, { "epoch": 0.0068352, "grad_norm": 0.9954916834831238, "learning_rate": 8.789923250015442e-07, "loss": 2.9795, "step": 338170 }, { "epoch": 0.0068608, "grad_norm": 0.8236088156700134, "learning_rate": 8.786622177124182e-07, "loss": 2.6481, "step": 338180 }, { "epoch": 0.0068864, "grad_norm": 0.835422158241272, "learning_rate": 8.783321695727432e-07, "loss": 2.6981, "step": 338190 }, { "epoch": 0.006912, "grad_norm": 0.8303605914115906, "learning_rate": 8.780021805846628e-07, "loss": 2.6647, "step": 338200 }, { "epoch": 0.0069376, "grad_norm": 0.7907372117042542, "learning_rate": 8.77672250750311e-07, "loss": 2.8603, "step": 338210 }, { "epoch": 0.0069632, "grad_norm": 0.8356109857559204, "learning_rate": 8.773423800718306e-07, "loss": 2.564, "step": 338220 }, { "epoch": 0.0069888, "grad_norm": 0.8102943897247314, "learning_rate": 8.770125685513608e-07, "loss": 2.8634, "step": 338230 }, { "epoch": 0.0070144, "grad_norm": 0.9396551847457886, "learning_rate": 8.76682816191039e-07, "loss": 2.7443, "step": 338240 }, { "epoch": 0.00704, "grad_norm": 0.8321098685264587, "learning_rate": 8.763531229930056e-07, "loss": 2.9404, "step": 338250 }, { "epoch": 0.0070656, "grad_norm": 0.8685417175292969, "learning_rate": 8.760234889593966e-07, "loss": 2.8339, "step": 338260 }, { "epoch": 0.0070912, "grad_norm": 0.7899550795555115, "learning_rate": 8.756939140923526e-07, "loss": 2.5542, "step": 338270 }, { "epoch": 0.0071168, "grad_norm": 0.8311007022857666, "learning_rate": 8.753643983940063e-07, "loss": 2.6435, "step": 338280 }, { "epoch": 0.0071424, "grad_norm": 0.9708808660507202, "learning_rate": 8.750349418664971e-07, "loss": 2.7435, "step": 338290 }, { "epoch": 0.007168, "grad_norm": 0.8382037281990051, "learning_rate": 8.747055445119612e-07, "loss": 2.7643, "step": 338300 }, { "epoch": 0.0071936, "grad_norm": 0.8409537076950073, "learning_rate": 8.743762063325345e-07, "loss": 2.881, "step": 338310 }, { "epoch": 0.0072192, "grad_norm": 0.8253667950630188, "learning_rate": 8.740469273303531e-07, "loss": 2.8737, "step": 338320 }, { "epoch": 0.0072448, "grad_norm": 0.7736184000968933, "learning_rate": 8.737177075075521e-07, "loss": 2.5843, "step": 338330 }, { "epoch": 0.0072704, "grad_norm": 0.8288024663925171, "learning_rate": 8.733885468662661e-07, "loss": 2.9059, "step": 338340 }, { "epoch": 0.007296, "grad_norm": 0.8002236485481262, "learning_rate": 8.730594454086295e-07, "loss": 2.7739, "step": 338350 }, { "epoch": 0.0073216, "grad_norm": 0.8951451182365417, "learning_rate": 8.727304031367767e-07, "loss": 2.9004, "step": 338360 }, { "epoch": 0.0073472, "grad_norm": 0.9001874327659607, "learning_rate": 8.724014200528419e-07, "loss": 2.4705, "step": 338370 }, { "epoch": 0.0073728, "grad_norm": 0.7956358194351196, "learning_rate": 8.720724961589588e-07, "loss": 2.8921, "step": 338380 }, { "epoch": 0.0073984, "grad_norm": 0.8995971083641052, "learning_rate": 8.717436314572581e-07, "loss": 2.6492, "step": 338390 }, { "epoch": 0.007424, "grad_norm": 1.0566531419754028, "learning_rate": 8.714148259498767e-07, "loss": 2.8167, "step": 338400 }, { "epoch": 0.0074496, "grad_norm": 0.884952187538147, "learning_rate": 8.71086079638942e-07, "loss": 2.8222, "step": 338410 }, { "epoch": 0.0074752, "grad_norm": 0.8500131964683533, "learning_rate": 8.707573925265878e-07, "loss": 2.938, "step": 338420 }, { "epoch": 0.0075008, "grad_norm": 0.8149593472480774, "learning_rate": 8.704287646149456e-07, "loss": 2.8923, "step": 338430 }, { "epoch": 0.0075264, "grad_norm": 0.8233075141906738, "learning_rate": 8.701001959061473e-07, "loss": 2.743, "step": 338440 }, { "epoch": 0.007552, "grad_norm": 0.7576778531074524, "learning_rate": 8.69771686402322e-07, "loss": 2.8375, "step": 338450 }, { "epoch": 0.0075776, "grad_norm": 0.8633913397789001, "learning_rate": 8.694432361056015e-07, "loss": 2.8135, "step": 338460 }, { "epoch": 0.0076032, "grad_norm": 0.8523681163787842, "learning_rate": 8.691148450181164e-07, "loss": 2.8025, "step": 338470 }, { "epoch": 0.0076288, "grad_norm": 0.8743675947189331, "learning_rate": 8.687865131419926e-07, "loss": 2.9726, "step": 338480 }, { "epoch": 0.0076544, "grad_norm": 0.8787896633148193, "learning_rate": 8.684582404793617e-07, "loss": 2.7871, "step": 338490 }, { "epoch": 0.00768, "grad_norm": 0.917420506477356, "learning_rate": 8.681300270323523e-07, "loss": 2.9012, "step": 338500 }, { "epoch": 0.0077056, "grad_norm": 0.9350385665893555, "learning_rate": 8.678018728030924e-07, "loss": 2.8187, "step": 338510 }, { "epoch": 0.0077312, "grad_norm": 0.8696634769439697, "learning_rate": 8.674737777937093e-07, "loss": 3.0134, "step": 338520 }, { "epoch": 0.0077568, "grad_norm": 0.8260453343391418, "learning_rate": 8.671457420063345e-07, "loss": 2.8656, "step": 338530 }, { "epoch": 0.0077824, "grad_norm": 0.8748322129249573, "learning_rate": 8.66817765443091e-07, "loss": 2.9093, "step": 338540 }, { "epoch": 0.007808, "grad_norm": 0.9062970280647278, "learning_rate": 8.664898481061068e-07, "loss": 2.8397, "step": 338550 }, { "epoch": 0.0078336, "grad_norm": 1.0764141082763672, "learning_rate": 8.661619899975082e-07, "loss": 2.5605, "step": 338560 }, { "epoch": 0.0078592, "grad_norm": 0.826835036277771, "learning_rate": 8.658341911194223e-07, "loss": 2.7716, "step": 338570 }, { "epoch": 0.0078848, "grad_norm": 0.7876999974250793, "learning_rate": 8.655064514739741e-07, "loss": 2.7678, "step": 338580 }, { "epoch": 0.0079104, "grad_norm": 0.983116090297699, "learning_rate": 8.651787710632887e-07, "loss": 2.7278, "step": 338590 }, { "epoch": 0.007936, "grad_norm": 0.9478554129600525, "learning_rate": 8.648511498894929e-07, "loss": 2.789, "step": 338600 }, { "epoch": 0.0079616, "grad_norm": 0.8554540276527405, "learning_rate": 8.645235879547076e-07, "loss": 2.9612, "step": 338610 }, { "epoch": 0.0079872, "grad_norm": 0.9196307063102722, "learning_rate": 8.641960852610587e-07, "loss": 2.8896, "step": 338620 }, { "epoch": 0.0080128, "grad_norm": 0.832111120223999, "learning_rate": 8.638686418106701e-07, "loss": 2.806, "step": 338630 }, { "epoch": 0.0080384, "grad_norm": 0.8032025694847107, "learning_rate": 8.635412576056656e-07, "loss": 2.8642, "step": 338640 }, { "epoch": 0.008064, "grad_norm": 0.9239169955253601, "learning_rate": 8.632139326481681e-07, "loss": 3.0193, "step": 338650 }, { "epoch": 0.0080896, "grad_norm": 0.9406335949897766, "learning_rate": 8.628866669403002e-07, "loss": 2.9141, "step": 338660 }, { "epoch": 0.0081152, "grad_norm": 0.7616788744926453, "learning_rate": 8.625594604841859e-07, "loss": 2.9188, "step": 338670 }, { "epoch": 0.0081408, "grad_norm": 0.9935427308082581, "learning_rate": 8.622323132819421e-07, "loss": 2.648, "step": 338680 }, { "epoch": 0.0081664, "grad_norm": 0.7943527698516846, "learning_rate": 8.619052253356919e-07, "loss": 2.8693, "step": 338690 }, { "epoch": 0.008192, "grad_norm": 0.8763550519943237, "learning_rate": 8.6157819664756e-07, "loss": 2.8509, "step": 338700 }, { "epoch": 0.0082176, "grad_norm": 0.835738480091095, "learning_rate": 8.612512272196638e-07, "loss": 3.0578, "step": 338710 }, { "epoch": 0.0082432, "grad_norm": 0.8509367108345032, "learning_rate": 8.609243170541259e-07, "loss": 2.7424, "step": 338720 }, { "epoch": 0.0082688, "grad_norm": 0.8519923090934753, "learning_rate": 8.605974661530658e-07, "loss": 2.8341, "step": 338730 }, { "epoch": 0.0082944, "grad_norm": 0.8122439980506897, "learning_rate": 8.602706745186007e-07, "loss": 2.7345, "step": 338740 }, { "epoch": 0.00832, "grad_norm": 0.8125451803207397, "learning_rate": 8.599439421528511e-07, "loss": 2.8037, "step": 338750 }, { "epoch": 0.0083456, "grad_norm": 0.8307704925537109, "learning_rate": 8.596172690579363e-07, "loss": 2.7052, "step": 338760 }, { "epoch": 0.0083712, "grad_norm": 0.8734477162361145, "learning_rate": 8.592906552359737e-07, "loss": 2.8067, "step": 338770 }, { "epoch": 0.0083968, "grad_norm": 0.8451113104820251, "learning_rate": 8.589641006890825e-07, "loss": 2.856, "step": 338780 }, { "epoch": 0.0084224, "grad_norm": 0.7820470333099365, "learning_rate": 8.58637605419379e-07, "loss": 2.8779, "step": 338790 }, { "epoch": 0.008448, "grad_norm": 0.8112220764160156, "learning_rate": 8.583111694289836e-07, "loss": 2.9332, "step": 338800 }, { "epoch": 0.0084736, "grad_norm": 0.7712209224700928, "learning_rate": 8.57984792720008e-07, "loss": 2.9185, "step": 338810 }, { "epoch": 0.0084992, "grad_norm": 0.8067310452461243, "learning_rate": 8.576584752945726e-07, "loss": 2.9713, "step": 338820 }, { "epoch": 0.0085248, "grad_norm": 0.9654321670532227, "learning_rate": 8.573322171547916e-07, "loss": 2.863, "step": 338830 }, { "epoch": 0.0085504, "grad_norm": 0.8082612156867981, "learning_rate": 8.570060183027806e-07, "loss": 2.9522, "step": 338840 }, { "epoch": 0.008576, "grad_norm": 0.8895509839057922, "learning_rate": 8.566798787406572e-07, "loss": 2.7902, "step": 338850 }, { "epoch": 0.0086016, "grad_norm": 1.888681173324585, "learning_rate": 8.563537984705339e-07, "loss": 2.975, "step": 338860 }, { "epoch": 0.0086272, "grad_norm": 0.8110969662666321, "learning_rate": 8.560277774945258e-07, "loss": 2.8332, "step": 338870 }, { "epoch": 0.0086528, "grad_norm": 0.940078854560852, "learning_rate": 8.557018158147479e-07, "loss": 2.8783, "step": 338880 }, { "epoch": 0.0086784, "grad_norm": 0.8616249561309814, "learning_rate": 8.553759134333128e-07, "loss": 2.779, "step": 338890 }, { "epoch": 0.008704, "grad_norm": 0.8180977702140808, "learning_rate": 8.550500703523356e-07, "loss": 2.7211, "step": 338900 }, { "epoch": 0.0087296, "grad_norm": 0.9457148909568787, "learning_rate": 8.54724286573928e-07, "loss": 2.7817, "step": 338910 }, { "epoch": 0.0087552, "grad_norm": 0.9035813212394714, "learning_rate": 8.543985621002026e-07, "loss": 2.7323, "step": 338920 }, { "epoch": 0.0087808, "grad_norm": 0.7802016139030457, "learning_rate": 8.540728969332734e-07, "loss": 2.7941, "step": 338930 }, { "epoch": 0.0088064, "grad_norm": 1.1191251277923584, "learning_rate": 8.537472910752498e-07, "loss": 2.8063, "step": 338940 }, { "epoch": 0.008832, "grad_norm": 0.8874281048774719, "learning_rate": 8.534217445282434e-07, "loss": 2.8664, "step": 338950 }, { "epoch": 0.0088576, "grad_norm": 0.7802249789237976, "learning_rate": 8.530962572943668e-07, "loss": 2.6941, "step": 338960 }, { "epoch": 0.0088832, "grad_norm": 0.9367630481719971, "learning_rate": 8.527708293757297e-07, "loss": 2.6551, "step": 338970 }, { "epoch": 0.0089088, "grad_norm": 0.8253310918807983, "learning_rate": 8.524454607744426e-07, "loss": 2.837, "step": 338980 }, { "epoch": 0.0089344, "grad_norm": 0.7910746932029724, "learning_rate": 8.521201514926158e-07, "loss": 2.8499, "step": 338990 }, { "epoch": 0.00896, "grad_norm": 0.7717753052711487, "learning_rate": 8.517949015323601e-07, "loss": 2.7803, "step": 339000 }, { "epoch": 0.0089856, "grad_norm": 0.7739831805229187, "learning_rate": 8.514697108957814e-07, "loss": 2.833, "step": 339010 }, { "epoch": 0.0090112, "grad_norm": 0.7836863994598389, "learning_rate": 8.511445795849904e-07, "loss": 2.9406, "step": 339020 }, { "epoch": 0.0090368, "grad_norm": 0.8444833159446716, "learning_rate": 8.508195076020953e-07, "loss": 2.6422, "step": 339030 }, { "epoch": 0.0090624, "grad_norm": 0.8005121350288391, "learning_rate": 8.504944949492023e-07, "loss": 2.9311, "step": 339040 }, { "epoch": 0.009088, "grad_norm": 0.8330118656158447, "learning_rate": 8.50169541628425e-07, "loss": 2.7985, "step": 339050 }, { "epoch": 0.0091136, "grad_norm": 0.8657951951026917, "learning_rate": 8.498446476418643e-07, "loss": 2.9237, "step": 339060 }, { "epoch": 0.0091392, "grad_norm": 0.7871949672698975, "learning_rate": 8.495198129916282e-07, "loss": 2.7596, "step": 339070 }, { "epoch": 0.0091648, "grad_norm": 0.9475791454315186, "learning_rate": 8.491950376798252e-07, "loss": 2.8995, "step": 339080 }, { "epoch": 0.0091904, "grad_norm": 0.9679244160652161, "learning_rate": 8.488703217085603e-07, "loss": 3.1745, "step": 339090 }, { "epoch": 0.009216, "grad_norm": 0.7627512216567993, "learning_rate": 8.485456650799395e-07, "loss": 3.0041, "step": 339100 }, { "epoch": 0.0092416, "grad_norm": 0.8171462416648865, "learning_rate": 8.482210677960678e-07, "loss": 2.7403, "step": 339110 }, { "epoch": 0.0092672, "grad_norm": 0.8072196841239929, "learning_rate": 8.478965298590502e-07, "loss": 2.6999, "step": 339120 }, { "epoch": 0.0092928, "grad_norm": 1.2238339185714722, "learning_rate": 8.475720512709928e-07, "loss": 2.4023, "step": 339130 }, { "epoch": 0.0093184, "grad_norm": 0.8813668489456177, "learning_rate": 8.472476320339973e-07, "loss": 2.8797, "step": 339140 }, { "epoch": 0.009344, "grad_norm": 0.8918208479881287, "learning_rate": 8.469232721501674e-07, "loss": 2.9161, "step": 339150 }, { "epoch": 0.0093696, "grad_norm": 0.8717830777168274, "learning_rate": 8.465989716216072e-07, "loss": 2.8476, "step": 339160 }, { "epoch": 0.0093952, "grad_norm": 0.7919518947601318, "learning_rate": 8.462747304504204e-07, "loss": 2.8367, "step": 339170 }, { "epoch": 0.0094208, "grad_norm": 0.7533743977546692, "learning_rate": 8.459505486387099e-07, "loss": 2.784, "step": 339180 }, { "epoch": 0.0094464, "grad_norm": 0.7697253227233887, "learning_rate": 8.456264261885771e-07, "loss": 2.9131, "step": 339190 }, { "epoch": 0.009472, "grad_norm": 0.8125613927841187, "learning_rate": 8.453023631021251e-07, "loss": 2.7793, "step": 339200 }, { "epoch": 0.0094976, "grad_norm": 0.9084057211875916, "learning_rate": 8.449783593814509e-07, "loss": 2.8463, "step": 339210 }, { "epoch": 0.0095232, "grad_norm": 0.8096882700920105, "learning_rate": 8.446544150286606e-07, "loss": 2.7803, "step": 339220 }, { "epoch": 0.0095488, "grad_norm": 1.5742216110229492, "learning_rate": 8.443305300458537e-07, "loss": 2.7229, "step": 339230 }, { "epoch": 0.0095744, "grad_norm": 0.7977712154388428, "learning_rate": 8.440067044351297e-07, "loss": 2.8806, "step": 339240 }, { "epoch": 0.0096, "grad_norm": 0.7961212992668152, "learning_rate": 8.436829381985901e-07, "loss": 2.788, "step": 339250 }, { "epoch": 0.0096256, "grad_norm": 0.8131951093673706, "learning_rate": 8.433592313383321e-07, "loss": 2.832, "step": 339260 }, { "epoch": 0.0096512, "grad_norm": 0.9676151871681213, "learning_rate": 8.430355838564563e-07, "loss": 3.035, "step": 339270 }, { "epoch": 0.0096768, "grad_norm": 0.9819188714027405, "learning_rate": 8.4271199575506e-07, "loss": 2.6533, "step": 339280 }, { "epoch": 0.0097024, "grad_norm": 0.8171953558921814, "learning_rate": 8.423884670362437e-07, "loss": 2.5684, "step": 339290 }, { "epoch": 0.009728, "grad_norm": 0.8583455085754395, "learning_rate": 8.420649977021045e-07, "loss": 2.7611, "step": 339300 }, { "epoch": 0.0097536, "grad_norm": 0.8730859756469727, "learning_rate": 8.417415877547385e-07, "loss": 2.9108, "step": 339310 }, { "epoch": 0.0097792, "grad_norm": 0.8459438681602478, "learning_rate": 8.414182371962453e-07, "loss": 2.6965, "step": 339320 }, { "epoch": 0.0098048, "grad_norm": 0.8292255401611328, "learning_rate": 8.410949460287232e-07, "loss": 3.129, "step": 339330 }, { "epoch": 0.0098304, "grad_norm": 0.8464087247848511, "learning_rate": 8.407717142542638e-07, "loss": 2.8815, "step": 339340 }, { "epoch": 0.009856, "grad_norm": 0.978294849395752, "learning_rate": 8.404485418749653e-07, "loss": 2.7858, "step": 339350 }, { "epoch": 0.0098816, "grad_norm": 1.048721432685852, "learning_rate": 8.40125428892924e-07, "loss": 2.9168, "step": 339360 }, { "epoch": 0.0099072, "grad_norm": 0.947054386138916, "learning_rate": 8.398023753102346e-07, "loss": 2.6574, "step": 339370 }, { "epoch": 0.0099328, "grad_norm": 1.0881980657577515, "learning_rate": 8.394793811289926e-07, "loss": 3.0172, "step": 339380 }, { "epoch": 0.0099584, "grad_norm": 0.8744928240776062, "learning_rate": 8.391564463512924e-07, "loss": 2.9939, "step": 339390 }, { "epoch": 0.009984, "grad_norm": 0.7925512194633484, "learning_rate": 8.388335709792272e-07, "loss": 2.8588, "step": 339400 }, { "epoch": 0.0100096, "grad_norm": 0.8248863220214844, "learning_rate": 8.385107550148918e-07, "loss": 2.7589, "step": 339410 }, { "epoch": 0.0100352, "grad_norm": 0.7724155783653259, "learning_rate": 8.381879984603803e-07, "loss": 2.7864, "step": 339420 }, { "epoch": 0.0100608, "grad_norm": 0.7663249969482422, "learning_rate": 8.37865301317784e-07, "loss": 2.8863, "step": 339430 }, { "epoch": 0.0100864, "grad_norm": 0.8464488983154297, "learning_rate": 8.375426635891959e-07, "loss": 2.7609, "step": 339440 }, { "epoch": 0.010112, "grad_norm": 0.9782276749610901, "learning_rate": 8.37220085276711e-07, "loss": 2.6747, "step": 339450 }, { "epoch": 0.0101376, "grad_norm": 0.8630722165107727, "learning_rate": 8.368975663824164e-07, "loss": 3.1021, "step": 339460 }, { "epoch": 0.0101632, "grad_norm": 0.8134934306144714, "learning_rate": 8.365751069084061e-07, "loss": 2.8728, "step": 339470 }, { "epoch": 0.0101888, "grad_norm": 0.794482409954071, "learning_rate": 8.362527068567706e-07, "loss": 2.796, "step": 339480 }, { "epoch": 0.0102144, "grad_norm": 0.8038676381111145, "learning_rate": 8.359303662296015e-07, "loss": 2.7751, "step": 339490 }, { "epoch": 0.01024, "grad_norm": 0.7752072215080261, "learning_rate": 8.356080850289872e-07, "loss": 2.7081, "step": 339500 }, { "epoch": 0.0102656, "grad_norm": 0.8226297497749329, "learning_rate": 8.352858632570204e-07, "loss": 2.9793, "step": 339510 }, { "epoch": 0.0102912, "grad_norm": 0.8783748149871826, "learning_rate": 8.349637009157896e-07, "loss": 2.7956, "step": 339520 }, { "epoch": 0.0103168, "grad_norm": 0.9845902919769287, "learning_rate": 8.346415980073818e-07, "loss": 2.7252, "step": 339530 }, { "epoch": 0.0103424, "grad_norm": 1.3432676792144775, "learning_rate": 8.343195545338878e-07, "loss": 2.4434, "step": 339540 }, { "epoch": 0.010368, "grad_norm": 0.8754068613052368, "learning_rate": 8.339975704973957e-07, "loss": 2.832, "step": 339550 }, { "epoch": 0.0103936, "grad_norm": 0.8692219853401184, "learning_rate": 8.336756458999917e-07, "loss": 2.8474, "step": 339560 }, { "epoch": 0.0104192, "grad_norm": 0.7992146611213684, "learning_rate": 8.333537807437664e-07, "loss": 2.6874, "step": 339570 }, { "epoch": 0.0104448, "grad_norm": 0.867946445941925, "learning_rate": 8.330319750308069e-07, "loss": 3.0519, "step": 339580 }, { "epoch": 0.0104704, "grad_norm": 0.7680538296699524, "learning_rate": 8.327102287631983e-07, "loss": 2.8035, "step": 339590 }, { "epoch": 0.010496, "grad_norm": 0.8982051014900208, "learning_rate": 8.323885419430267e-07, "loss": 2.9019, "step": 339600 }, { "epoch": 0.0105216, "grad_norm": 0.9025477170944214, "learning_rate": 8.320669145723781e-07, "loss": 2.7516, "step": 339610 }, { "epoch": 0.0105472, "grad_norm": 0.9248078465461731, "learning_rate": 8.317453466533399e-07, "loss": 2.7839, "step": 339620 }, { "epoch": 0.0105728, "grad_norm": 0.9278677105903625, "learning_rate": 8.314238381879969e-07, "loss": 2.5901, "step": 339630 }, { "epoch": 0.0105984, "grad_norm": 0.8015637397766113, "learning_rate": 8.311023891784332e-07, "loss": 2.8159, "step": 339640 }, { "epoch": 0.010624, "grad_norm": 0.8615289330482483, "learning_rate": 8.307809996267357e-07, "loss": 2.7203, "step": 339650 }, { "epoch": 0.0106496, "grad_norm": 0.8043615221977234, "learning_rate": 8.304596695349843e-07, "loss": 2.886, "step": 339660 }, { "epoch": 0.0106752, "grad_norm": 0.9942379593849182, "learning_rate": 8.301383989052647e-07, "loss": 2.8909, "step": 339670 }, { "epoch": 0.0107008, "grad_norm": 0.8439637422561646, "learning_rate": 8.29817187739661e-07, "loss": 2.9123, "step": 339680 }, { "epoch": 0.0107264, "grad_norm": 1.1456009149551392, "learning_rate": 8.29496036040256e-07, "loss": 2.7754, "step": 339690 }, { "epoch": 0.010752, "grad_norm": 0.796509861946106, "learning_rate": 8.291749438091312e-07, "loss": 2.6636, "step": 339700 }, { "epoch": 0.0107776, "grad_norm": 0.8796952366828918, "learning_rate": 8.288539110483707e-07, "loss": 2.749, "step": 339710 }, { "epoch": 0.0108032, "grad_norm": 0.8249518871307373, "learning_rate": 8.285329377600559e-07, "loss": 2.8089, "step": 339720 }, { "epoch": 0.0108288, "grad_norm": 0.8829290270805359, "learning_rate": 8.282120239462643e-07, "loss": 2.7348, "step": 339730 }, { "epoch": 0.0108544, "grad_norm": 0.8544175028800964, "learning_rate": 8.278911696090819e-07, "loss": 2.7948, "step": 339740 }, { "epoch": 0.01088, "grad_norm": 0.7965789437294006, "learning_rate": 8.275703747505881e-07, "loss": 2.866, "step": 339750 }, { "epoch": 0.0109056, "grad_norm": 0.798014760017395, "learning_rate": 8.272496393728624e-07, "loss": 2.7539, "step": 339760 }, { "epoch": 0.0109312, "grad_norm": 0.8772576451301575, "learning_rate": 8.269289634779842e-07, "loss": 2.8435, "step": 339770 }, { "epoch": 0.0109568, "grad_norm": 2.236086368560791, "learning_rate": 8.266083470680364e-07, "loss": 2.7209, "step": 339780 }, { "epoch": 0.0109824, "grad_norm": 0.8604965209960938, "learning_rate": 8.262877901450939e-07, "loss": 2.9238, "step": 339790 }, { "epoch": 0.011008, "grad_norm": 0.9229706525802612, "learning_rate": 8.259672927112361e-07, "loss": 2.9297, "step": 339800 }, { "epoch": 0.0110336, "grad_norm": 0.905234694480896, "learning_rate": 8.256468547685425e-07, "loss": 2.7976, "step": 339810 }, { "epoch": 0.0110592, "grad_norm": 0.8554940819740295, "learning_rate": 8.253264763190916e-07, "loss": 2.827, "step": 339820 }, { "epoch": 0.0110848, "grad_norm": 0.915279746055603, "learning_rate": 8.250061573649592e-07, "loss": 2.6636, "step": 339830 }, { "epoch": 0.0111104, "grad_norm": 0.8617889881134033, "learning_rate": 8.24685897908224e-07, "loss": 2.8706, "step": 339840 }, { "epoch": 0.011136, "grad_norm": 0.8873745799064636, "learning_rate": 8.24365697950964e-07, "loss": 2.8502, "step": 339850 }, { "epoch": 0.0111616, "grad_norm": 0.9499377608299255, "learning_rate": 8.240455574952522e-07, "loss": 2.8704, "step": 339860 }, { "epoch": 0.0111872, "grad_norm": 0.908909022808075, "learning_rate": 8.237254765431657e-07, "loss": 2.7756, "step": 339870 }, { "epoch": 0.0112128, "grad_norm": 0.9433378577232361, "learning_rate": 8.234054550967807e-07, "loss": 2.7525, "step": 339880 }, { "epoch": 0.0112384, "grad_norm": 0.8859347105026245, "learning_rate": 8.230854931581733e-07, "loss": 3.0141, "step": 339890 }, { "epoch": 0.011264, "grad_norm": 0.8413042426109314, "learning_rate": 8.227655907294163e-07, "loss": 2.9885, "step": 339900 }, { "epoch": 0.0112896, "grad_norm": 0.9994478225708008, "learning_rate": 8.224457478125869e-07, "loss": 2.8933, "step": 339910 }, { "epoch": 0.0113152, "grad_norm": 0.8083884119987488, "learning_rate": 8.221259644097568e-07, "loss": 2.8442, "step": 339920 }, { "epoch": 0.0113408, "grad_norm": 0.8691335916519165, "learning_rate": 8.218062405230009e-07, "loss": 2.822, "step": 339930 }, { "epoch": 0.0113664, "grad_norm": 1.083876371383667, "learning_rate": 8.214865761543922e-07, "loss": 2.7549, "step": 339940 }, { "epoch": 0.011392, "grad_norm": 0.7543980479240417, "learning_rate": 8.211669713060033e-07, "loss": 2.6702, "step": 339950 }, { "epoch": 0.0114176, "grad_norm": 0.8087019324302673, "learning_rate": 8.208474259799082e-07, "loss": 2.8412, "step": 339960 }, { "epoch": 0.0114432, "grad_norm": 0.8293070793151855, "learning_rate": 8.205279401781773e-07, "loss": 2.7831, "step": 339970 }, { "epoch": 0.0114688, "grad_norm": 0.9004818797111511, "learning_rate": 8.202085139028848e-07, "loss": 2.8758, "step": 339980 }, { "epoch": 0.0114944, "grad_norm": 0.9336040616035461, "learning_rate": 8.198891471560988e-07, "loss": 2.8776, "step": 339990 }, { "epoch": 0.01152, "grad_norm": 0.9228086471557617, "learning_rate": 8.195698399398921e-07, "loss": 3.048, "step": 340000 }, { "epoch": 0.0115456, "grad_norm": 0.9189521670341492, "learning_rate": 8.192505922563343e-07, "loss": 2.7873, "step": 340010 }, { "epoch": 0.0115712, "grad_norm": 0.9675172567367554, "learning_rate": 8.18931404107497e-07, "loss": 3.3702, "step": 340020 }, { "epoch": 0.0115968, "grad_norm": 0.9770752787590027, "learning_rate": 8.186122754954495e-07, "loss": 2.896, "step": 340030 }, { "epoch": 0.0116224, "grad_norm": 0.8992058038711548, "learning_rate": 8.182932064222615e-07, "loss": 2.8986, "step": 340040 }, { "epoch": 0.011648, "grad_norm": 0.7734149694442749, "learning_rate": 8.179741968900035e-07, "loss": 2.845, "step": 340050 }, { "epoch": 0.0116736, "grad_norm": 0.8319277167320251, "learning_rate": 8.176552469007403e-07, "loss": 2.9305, "step": 340060 }, { "epoch": 0.0116992, "grad_norm": 0.8734126091003418, "learning_rate": 8.173363564565429e-07, "loss": 2.9882, "step": 340070 }, { "epoch": 0.0117248, "grad_norm": 0.8842250108718872, "learning_rate": 8.17017525559477e-07, "loss": 2.9292, "step": 340080 }, { "epoch": 0.0117504, "grad_norm": 0.8629974126815796, "learning_rate": 8.166987542116144e-07, "loss": 2.662, "step": 340090 }, { "epoch": 0.011776, "grad_norm": 0.8866902589797974, "learning_rate": 8.163800424150181e-07, "loss": 2.7835, "step": 340100 }, { "epoch": 0.0118016, "grad_norm": 0.9193760752677917, "learning_rate": 8.160613901717596e-07, "loss": 2.7677, "step": 340110 }, { "epoch": 0.0118272, "grad_norm": 0.9047386646270752, "learning_rate": 8.157427974839005e-07, "loss": 3.0115, "step": 340120 }, { "epoch": 0.0118528, "grad_norm": 0.8558060526847839, "learning_rate": 8.154242643535071e-07, "loss": 2.8965, "step": 340130 }, { "epoch": 0.0118784, "grad_norm": 0.8209421634674072, "learning_rate": 8.151057907826476e-07, "loss": 2.8115, "step": 340140 }, { "epoch": 0.011904, "grad_norm": 0.8882346153259277, "learning_rate": 8.14787376773386e-07, "loss": 2.8691, "step": 340150 }, { "epoch": 0.0119296, "grad_norm": 0.8308545351028442, "learning_rate": 8.144690223277863e-07, "loss": 2.9233, "step": 340160 }, { "epoch": 0.0119552, "grad_norm": 0.8024107813835144, "learning_rate": 8.141507274479144e-07, "loss": 2.6889, "step": 340170 }, { "epoch": 0.0119808, "grad_norm": 0.8965538740158081, "learning_rate": 8.138324921358354e-07, "loss": 2.7075, "step": 340180 }, { "epoch": 0.0120064, "grad_norm": 0.7749036550521851, "learning_rate": 8.135143163936088e-07, "loss": 2.8164, "step": 340190 }, { "epoch": 0.012032, "grad_norm": 0.8058093786239624, "learning_rate": 8.131962002233018e-07, "loss": 2.749, "step": 340200 }, { "epoch": 0.0120576, "grad_norm": 0.8364585638046265, "learning_rate": 8.128781436269751e-07, "loss": 2.9029, "step": 340210 }, { "epoch": 0.0120832, "grad_norm": 0.8023548126220703, "learning_rate": 8.125601466066913e-07, "loss": 2.7272, "step": 340220 }, { "epoch": 0.0121088, "grad_norm": 0.9606231451034546, "learning_rate": 8.122422091645144e-07, "loss": 2.8952, "step": 340230 }, { "epoch": 0.0121344, "grad_norm": 0.9450041055679321, "learning_rate": 8.119243313025049e-07, "loss": 2.9416, "step": 340240 }, { "epoch": 0.01216, "grad_norm": 0.894517719745636, "learning_rate": 8.116065130227246e-07, "loss": 3.1001, "step": 340250 }, { "epoch": 0.0121856, "grad_norm": 0.8149450421333313, "learning_rate": 8.112887543272318e-07, "loss": 2.7709, "step": 340260 }, { "epoch": 0.0122112, "grad_norm": 0.8728802800178528, "learning_rate": 8.109710552180905e-07, "loss": 2.8233, "step": 340270 }, { "epoch": 0.0122368, "grad_norm": 0.8483132719993591, "learning_rate": 8.106534156973611e-07, "loss": 2.9967, "step": 340280 }, { "epoch": 0.0122624, "grad_norm": 0.8386427164077759, "learning_rate": 8.10335835767101e-07, "loss": 2.8073, "step": 340290 }, { "epoch": 0.012288, "grad_norm": 0.88557368516922, "learning_rate": 8.100183154293706e-07, "loss": 3.1041, "step": 340300 }, { "epoch": 0.0123136, "grad_norm": 0.8092606067657471, "learning_rate": 8.097008546862306e-07, "loss": 2.936, "step": 340310 }, { "epoch": 0.0123392, "grad_norm": 0.8940982818603516, "learning_rate": 8.093834535397372e-07, "loss": 3.003, "step": 340320 }, { "epoch": 0.0123648, "grad_norm": 1.363693118095398, "learning_rate": 8.090661119919485e-07, "loss": 2.7322, "step": 340330 }, { "epoch": 0.0123904, "grad_norm": 0.7628681659698486, "learning_rate": 8.087488300449242e-07, "loss": 2.8148, "step": 340340 }, { "epoch": 0.012416, "grad_norm": 0.7973239421844482, "learning_rate": 8.084316077007203e-07, "loss": 2.9604, "step": 340350 }, { "epoch": 0.0124416, "grad_norm": 0.9327028393745422, "learning_rate": 8.081144449613953e-07, "loss": 3.1113, "step": 340360 }, { "epoch": 0.0124672, "grad_norm": 0.895892322063446, "learning_rate": 8.077973418290042e-07, "loss": 2.6854, "step": 340370 }, { "epoch": 0.0124928, "grad_norm": 1.219339370727539, "learning_rate": 8.074802983056074e-07, "loss": 2.7094, "step": 340380 }, { "epoch": 0.0125184, "grad_norm": 0.8863463997840881, "learning_rate": 8.071633143932556e-07, "loss": 2.7667, "step": 340390 }, { "epoch": 0.012544, "grad_norm": 0.8632345795631409, "learning_rate": 8.06846390094006e-07, "loss": 3.0354, "step": 340400 }, { "epoch": 0.0125696, "grad_norm": 0.8763354420661926, "learning_rate": 8.065295254099148e-07, "loss": 3.1334, "step": 340410 }, { "epoch": 0.0125952, "grad_norm": 1.0490797758102417, "learning_rate": 8.062127203430358e-07, "loss": 3.0215, "step": 340420 }, { "epoch": 0.0126208, "grad_norm": 0.818132221698761, "learning_rate": 8.05895974895422e-07, "loss": 2.4781, "step": 340430 }, { "epoch": 0.0126464, "grad_norm": 0.9277215600013733, "learning_rate": 8.055792890691339e-07, "loss": 2.9264, "step": 340440 }, { "epoch": 0.012672, "grad_norm": 0.8004551529884338, "learning_rate": 8.052626628662186e-07, "loss": 2.7265, "step": 340450 }, { "epoch": 0.0126976, "grad_norm": 0.8278459906578064, "learning_rate": 8.049460962887301e-07, "loss": 2.7234, "step": 340460 }, { "epoch": 0.0127232, "grad_norm": 0.858539342880249, "learning_rate": 8.046295893387235e-07, "loss": 2.8517, "step": 340470 }, { "epoch": 0.0127488, "grad_norm": 0.9192917943000793, "learning_rate": 8.043131420182493e-07, "loss": 2.8994, "step": 340480 }, { "epoch": 0.0127744, "grad_norm": 0.7896771430969238, "learning_rate": 8.039967543293614e-07, "loss": 2.7886, "step": 340490 }, { "epoch": 0.0128, "grad_norm": 0.8401105999946594, "learning_rate": 8.036804262741105e-07, "loss": 2.8323, "step": 340500 }, { "epoch": 0.0128256, "grad_norm": 0.8763288855552673, "learning_rate": 8.033641578545492e-07, "loss": 2.6282, "step": 340510 }, { "epoch": 0.0128512, "grad_norm": 1.1161274909973145, "learning_rate": 8.03047949072725e-07, "loss": 2.9055, "step": 340520 }, { "epoch": 0.0128768, "grad_norm": 1.0151184797286987, "learning_rate": 8.027317999306916e-07, "loss": 2.9113, "step": 340530 }, { "epoch": 0.0129024, "grad_norm": 0.9515738487243652, "learning_rate": 8.024157104304975e-07, "loss": 2.6285, "step": 340540 }, { "epoch": 0.012928, "grad_norm": 1.0362889766693115, "learning_rate": 8.020996805741932e-07, "loss": 2.9158, "step": 340550 }, { "epoch": 0.0129536, "grad_norm": 0.9091178178787231, "learning_rate": 8.017837103638282e-07, "loss": 2.8589, "step": 340560 }, { "epoch": 0.0129792, "grad_norm": 1.1028069257736206, "learning_rate": 8.014677998014509e-07, "loss": 2.8495, "step": 340570 }, { "epoch": 0.0130048, "grad_norm": 1.0626857280731201, "learning_rate": 8.011519488891128e-07, "loss": 3.0039, "step": 340580 }, { "epoch": 0.0130304, "grad_norm": 1.0332071781158447, "learning_rate": 8.008361576288571e-07, "loss": 2.7511, "step": 340590 }, { "epoch": 0.013056, "grad_norm": 0.893747866153717, "learning_rate": 8.005204260227317e-07, "loss": 2.9832, "step": 340600 }, { "epoch": 0.0130816, "grad_norm": 1.3266282081604004, "learning_rate": 8.002047540727887e-07, "loss": 2.6112, "step": 340610 }, { "epoch": 0.0131072, "grad_norm": 0.9550742506980896, "learning_rate": 7.998891417810739e-07, "loss": 2.8598, "step": 340620 }, { "epoch": 0.0131328, "grad_norm": 0.8563041687011719, "learning_rate": 7.995735891496337e-07, "loss": 2.6705, "step": 340630 }, { "epoch": 0.0131584, "grad_norm": 0.8615648746490479, "learning_rate": 7.992580961805119e-07, "loss": 2.7749, "step": 340640 }, { "epoch": 0.013184, "grad_norm": 0.80247563123703, "learning_rate": 7.989426628757557e-07, "loss": 2.7429, "step": 340650 }, { "epoch": 0.0132096, "grad_norm": 0.8292964696884155, "learning_rate": 7.986272892374114e-07, "loss": 2.8506, "step": 340660 }, { "epoch": 0.0132352, "grad_norm": 0.8931251168251038, "learning_rate": 7.983119752675239e-07, "loss": 2.7805, "step": 340670 }, { "epoch": 0.0132608, "grad_norm": 0.8398290872573853, "learning_rate": 7.979967209681372e-07, "loss": 2.7753, "step": 340680 }, { "epoch": 0.0132864, "grad_norm": 0.8611195087432861, "learning_rate": 7.976815263412963e-07, "loss": 2.7534, "step": 340690 }, { "epoch": 0.013312, "grad_norm": 0.8025956153869629, "learning_rate": 7.97366391389045e-07, "loss": 2.7696, "step": 340700 }, { "epoch": 2.56e-05, "grad_norm": 0.8593146204948425, "learning_rate": 7.970513161134296e-07, "loss": 2.2159, "step": 340710 }, { "epoch": 5.12e-05, "grad_norm": 0.8335390090942383, "learning_rate": 7.967363005164874e-07, "loss": 2.6158, "step": 340720 }, { "epoch": 7.68e-05, "grad_norm": 0.8009431958198547, "learning_rate": 7.964213446002655e-07, "loss": 2.7292, "step": 340730 }, { "epoch": 0.0001024, "grad_norm": 0.83150714635849, "learning_rate": 7.961064483668046e-07, "loss": 2.8275, "step": 340740 }, { "epoch": 0.000128, "grad_norm": 0.9439178705215454, "learning_rate": 7.957916118181475e-07, "loss": 2.5523, "step": 340750 }, { "epoch": 0.0001536, "grad_norm": 0.9715688228607178, "learning_rate": 7.954768349563347e-07, "loss": 2.6563, "step": 340760 }, { "epoch": 0.0001792, "grad_norm": 0.7734018564224243, "learning_rate": 7.951621177834091e-07, "loss": 2.8287, "step": 340770 }, { "epoch": 0.0002048, "grad_norm": 0.8291379809379578, "learning_rate": 7.948474603014112e-07, "loss": 2.8092, "step": 340780 }, { "epoch": 0.0002304, "grad_norm": 0.808708667755127, "learning_rate": 7.945328625123805e-07, "loss": 2.653, "step": 340790 }, { "epoch": 0.000256, "grad_norm": 0.8937944769859314, "learning_rate": 7.942183244183577e-07, "loss": 2.6547, "step": 340800 }, { "epoch": 0.0002816, "grad_norm": 0.861725926399231, "learning_rate": 7.939038460213833e-07, "loss": 2.5195, "step": 340810 }, { "epoch": 0.0003072, "grad_norm": 0.8379940390586853, "learning_rate": 7.935894273234946e-07, "loss": 2.6868, "step": 340820 }, { "epoch": 0.0003328, "grad_norm": 0.8123665452003479, "learning_rate": 7.932750683267342e-07, "loss": 2.9466, "step": 340830 }, { "epoch": 0.0003584, "grad_norm": 0.8402495980262756, "learning_rate": 7.929607690331364e-07, "loss": 2.5686, "step": 340840 }, { "epoch": 0.000384, "grad_norm": 0.8236294388771057, "learning_rate": 7.926465294447405e-07, "loss": 2.7389, "step": 340850 }, { "epoch": 0.0004096, "grad_norm": 0.7873246073722839, "learning_rate": 7.923323495635849e-07, "loss": 2.6625, "step": 340860 }, { "epoch": 0.0004352, "grad_norm": 0.7611145377159119, "learning_rate": 7.920182293917078e-07, "loss": 2.7667, "step": 340870 }, { "epoch": 0.0004608, "grad_norm": 1.1734074354171753, "learning_rate": 7.917041689311444e-07, "loss": 2.7428, "step": 340880 }, { "epoch": 0.0004864, "grad_norm": 0.8150089979171753, "learning_rate": 7.91390168183932e-07, "loss": 2.8732, "step": 340890 }, { "epoch": 0.000512, "grad_norm": 0.8107773065567017, "learning_rate": 7.910762271521077e-07, "loss": 2.5316, "step": 340900 }, { "epoch": 0.0005376, "grad_norm": 0.8410989046096802, "learning_rate": 7.907623458377079e-07, "loss": 2.6663, "step": 340910 }, { "epoch": 0.0005632, "grad_norm": 0.9659985303878784, "learning_rate": 7.904485242427651e-07, "loss": 2.332, "step": 340920 }, { "epoch": 0.0005888, "grad_norm": 0.8095685243606567, "learning_rate": 7.901347623693167e-07, "loss": 2.6594, "step": 340930 }, { "epoch": 0.0006144, "grad_norm": 0.8213629126548767, "learning_rate": 7.898210602193956e-07, "loss": 2.5442, "step": 340940 }, { "epoch": 0.00064, "grad_norm": 0.908380925655365, "learning_rate": 7.895074177950357e-07, "loss": 2.4534, "step": 340950 }, { "epoch": 0.0006656, "grad_norm": 0.9480249285697937, "learning_rate": 7.891938350982764e-07, "loss": 2.6903, "step": 340960 }, { "epoch": 0.0006912, "grad_norm": 0.8215486407279968, "learning_rate": 7.88880312131145e-07, "loss": 2.7289, "step": 340970 }, { "epoch": 0.0007168, "grad_norm": 1.0129750967025757, "learning_rate": 7.885668488956765e-07, "loss": 2.7928, "step": 340980 }, { "epoch": 0.0007424, "grad_norm": 0.9596628546714783, "learning_rate": 7.88253445393905e-07, "loss": 2.7022, "step": 340990 }, { "epoch": 0.000768, "grad_norm": 0.921308696269989, "learning_rate": 7.879401016278609e-07, "loss": 2.6845, "step": 341000 }, { "epoch": 0.0007936, "grad_norm": 0.9699082374572754, "learning_rate": 7.876268175995771e-07, "loss": 2.7435, "step": 341010 }, { "epoch": 0.0008192, "grad_norm": 0.8319149613380432, "learning_rate": 7.873135933110842e-07, "loss": 2.5767, "step": 341020 }, { "epoch": 0.0008448, "grad_norm": 0.8337294459342957, "learning_rate": 7.870004287644173e-07, "loss": 2.5146, "step": 341030 }, { "epoch": 0.0008704, "grad_norm": 0.9774466753005981, "learning_rate": 7.866873239616013e-07, "loss": 2.6668, "step": 341040 }, { "epoch": 0.000896, "grad_norm": 0.8856611251831055, "learning_rate": 7.863742789046702e-07, "loss": 2.4834, "step": 341050 }, { "epoch": 0.0009216, "grad_norm": 0.7293861508369446, "learning_rate": 7.860612935956524e-07, "loss": 2.4095, "step": 341060 }, { "epoch": 0.0009472, "grad_norm": 0.8636677861213684, "learning_rate": 7.857483680365785e-07, "loss": 2.4796, "step": 341070 }, { "epoch": 0.0009728, "grad_norm": 1.0305166244506836, "learning_rate": 7.854355022294779e-07, "loss": 2.4982, "step": 341080 }, { "epoch": 0.0009984, "grad_norm": 0.7906078100204468, "learning_rate": 7.851226961763791e-07, "loss": 2.4991, "step": 341090 }, { "epoch": 0.001024, "grad_norm": 0.9919029474258423, "learning_rate": 7.848099498793105e-07, "loss": 2.5072, "step": 341100 }, { "epoch": 0.0010496, "grad_norm": 0.8317157030105591, "learning_rate": 7.844972633403026e-07, "loss": 2.7114, "step": 341110 }, { "epoch": 0.0010752, "grad_norm": 0.8087929487228394, "learning_rate": 7.841846365613781e-07, "loss": 2.6072, "step": 341120 }, { "epoch": 0.0011008, "grad_norm": 0.8782738447189331, "learning_rate": 7.838720695445657e-07, "loss": 2.6188, "step": 341130 }, { "epoch": 0.0011264, "grad_norm": 0.7992514371871948, "learning_rate": 7.835595622918957e-07, "loss": 2.6323, "step": 341140 }, { "epoch": 0.001152, "grad_norm": 0.8104068636894226, "learning_rate": 7.832471148053922e-07, "loss": 2.7405, "step": 341150 }, { "epoch": 0.0011776, "grad_norm": 0.8110936284065247, "learning_rate": 7.829347270870824e-07, "loss": 2.6568, "step": 341160 }, { "epoch": 0.0012032, "grad_norm": 0.9680420756340027, "learning_rate": 7.826223991389903e-07, "loss": 2.8318, "step": 341170 }, { "epoch": 0.0012288, "grad_norm": 0.8802482485771179, "learning_rate": 7.823101309631431e-07, "loss": 2.6449, "step": 341180 }, { "epoch": 0.0012544, "grad_norm": 0.7842570543289185, "learning_rate": 7.819979225615638e-07, "loss": 2.5254, "step": 341190 }, { "epoch": 0.00128, "grad_norm": 0.8464350700378418, "learning_rate": 7.816857739362782e-07, "loss": 2.6172, "step": 341200 }, { "epoch": 0.0013056, "grad_norm": 0.9875450134277344, "learning_rate": 7.813736850893105e-07, "loss": 2.5374, "step": 341210 }, { "epoch": 0.0013312, "grad_norm": 0.9860211610794067, "learning_rate": 7.810616560226847e-07, "loss": 2.7998, "step": 341220 }, { "epoch": 0.0013568, "grad_norm": 0.8719937801361084, "learning_rate": 7.807496867384257e-07, "loss": 2.6653, "step": 341230 }, { "epoch": 0.0013824, "grad_norm": 1.49604070186615, "learning_rate": 7.80437777238553e-07, "loss": 2.7408, "step": 341240 }, { "epoch": 0.001408, "grad_norm": 0.8298701047897339, "learning_rate": 7.801259275250917e-07, "loss": 2.6471, "step": 341250 }, { "epoch": 0.0014336, "grad_norm": 0.7536318898200989, "learning_rate": 7.798141376000624e-07, "loss": 2.4584, "step": 341260 }, { "epoch": 0.0014592, "grad_norm": 0.9780122637748718, "learning_rate": 7.795024074654878e-07, "loss": 2.6022, "step": 341270 }, { "epoch": 0.0014848, "grad_norm": 0.853426456451416, "learning_rate": 7.791907371233908e-07, "loss": 2.2526, "step": 341280 }, { "epoch": 0.0015104, "grad_norm": 0.865493655204773, "learning_rate": 7.7887912657579e-07, "loss": 2.7194, "step": 341290 }, { "epoch": 0.001536, "grad_norm": 0.8201336860656738, "learning_rate": 7.78567575824708e-07, "loss": 2.5916, "step": 341300 }, { "epoch": 2.56e-05, "grad_norm": 0.9224228858947754, "learning_rate": 7.782560848721643e-07, "loss": 1.8822, "step": 341310 }, { "epoch": 5.12e-05, "grad_norm": 1.0800015926361084, "learning_rate": 7.779446537201796e-07, "loss": 2.8474, "step": 341320 }, { "epoch": 7.68e-05, "grad_norm": 0.9341338872909546, "learning_rate": 7.776332823707722e-07, "loss": 2.7988, "step": 341330 }, { "epoch": 0.0001024, "grad_norm": 0.8744654655456543, "learning_rate": 7.773219708259627e-07, "loss": 2.7885, "step": 341340 }, { "epoch": 0.000128, "grad_norm": 0.8703789114952087, "learning_rate": 7.770107190877685e-07, "loss": 2.7359, "step": 341350 }, { "epoch": 0.0001536, "grad_norm": 0.8523438572883606, "learning_rate": 7.766995271582111e-07, "loss": 2.7375, "step": 341360 }, { "epoch": 0.0001792, "grad_norm": 0.8593406081199646, "learning_rate": 7.763883950393036e-07, "loss": 2.5202, "step": 341370 }, { "epoch": 0.0002048, "grad_norm": 0.882646381855011, "learning_rate": 7.760773227330654e-07, "loss": 2.6129, "step": 341380 }, { "epoch": 0.0002304, "grad_norm": 0.8628478646278381, "learning_rate": 7.757663102415158e-07, "loss": 2.5914, "step": 341390 }, { "epoch": 0.000256, "grad_norm": 0.8426867723464966, "learning_rate": 7.75455357566669e-07, "loss": 2.824, "step": 341400 }, { "epoch": 0.0002816, "grad_norm": 1.0026170015335083, "learning_rate": 7.751444647105433e-07, "loss": 2.612, "step": 341410 }, { "epoch": 0.0003072, "grad_norm": 0.7776743769645691, "learning_rate": 7.748336316751537e-07, "loss": 2.5048, "step": 341420 }, { "epoch": 0.0003328, "grad_norm": 0.9739827513694763, "learning_rate": 7.745228584625175e-07, "loss": 2.7732, "step": 341430 }, { "epoch": 0.0003584, "grad_norm": 0.9065678119659424, "learning_rate": 7.742121450746476e-07, "loss": 2.6747, "step": 341440 }, { "epoch": 0.000384, "grad_norm": 0.9836235046386719, "learning_rate": 7.739014915135601e-07, "loss": 2.6074, "step": 341450 }, { "epoch": 0.0004096, "grad_norm": 0.7910675406455994, "learning_rate": 7.735908977812701e-07, "loss": 2.6661, "step": 341460 }, { "epoch": 0.0004352, "grad_norm": 0.8136487007141113, "learning_rate": 7.732803638797881e-07, "loss": 2.7759, "step": 341470 }, { "epoch": 0.0004608, "grad_norm": 0.8733058571815491, "learning_rate": 7.729698898111326e-07, "loss": 2.5652, "step": 341480 }, { "epoch": 0.0004864, "grad_norm": 0.9074473977088928, "learning_rate": 7.726594755773175e-07, "loss": 2.5832, "step": 341490 }, { "epoch": 0.000512, "grad_norm": 0.7846843004226685, "learning_rate": 7.723491211803513e-07, "loss": 2.501, "step": 341500 }, { "epoch": 0.0005376, "grad_norm": 0.971429169178009, "learning_rate": 7.720388266222489e-07, "loss": 2.6526, "step": 341510 }, { "epoch": 0.0005632, "grad_norm": 0.9304726123809814, "learning_rate": 7.717285919050221e-07, "loss": 2.4051, "step": 341520 }, { "epoch": 0.0005888, "grad_norm": 0.8591592311859131, "learning_rate": 7.714184170306826e-07, "loss": 2.5947, "step": 341530 }, { "epoch": 0.0006144, "grad_norm": 0.8312366008758545, "learning_rate": 7.711083020012422e-07, "loss": 2.5443, "step": 341540 }, { "epoch": 0.00064, "grad_norm": 0.888830304145813, "learning_rate": 7.707982468187114e-07, "loss": 2.8309, "step": 341550 }, { "epoch": 0.0006656, "grad_norm": 0.9233748912811279, "learning_rate": 7.704882514851031e-07, "loss": 3.0351, "step": 341560 }, { "epoch": 0.0006912, "grad_norm": 0.8351134657859802, "learning_rate": 7.701783160024234e-07, "loss": 2.8228, "step": 341570 }, { "epoch": 0.0007168, "grad_norm": 0.8410393595695496, "learning_rate": 7.698684403726853e-07, "loss": 2.5509, "step": 341580 }, { "epoch": 0.0007424, "grad_norm": 0.8301195502281189, "learning_rate": 7.69558624597897e-07, "loss": 2.6899, "step": 341590 }, { "epoch": 0.000768, "grad_norm": 0.8612253069877625, "learning_rate": 7.69248868680067e-07, "loss": 2.6165, "step": 341600 }, { "epoch": 2.56e-05, "grad_norm": 0.851277768611908, "learning_rate": 7.689391726212059e-07, "loss": 1.7467, "step": 341610 }, { "epoch": 5.12e-05, "grad_norm": 0.8490919470787048, "learning_rate": 7.686295364233198e-07, "loss": 2.5718, "step": 341620 }, { "epoch": 7.68e-05, "grad_norm": 0.8375365734100342, "learning_rate": 7.683199600884206e-07, "loss": 2.6943, "step": 341630 }, { "epoch": 0.0001024, "grad_norm": 0.7693363428115845, "learning_rate": 7.68010443618511e-07, "loss": 2.3707, "step": 341640 }, { "epoch": 0.000128, "grad_norm": 0.8586790561676025, "learning_rate": 7.677009870155983e-07, "loss": 2.7092, "step": 341650 }, { "epoch": 0.0001536, "grad_norm": 0.9867141246795654, "learning_rate": 7.67391590281692e-07, "loss": 2.6451, "step": 341660 }, { "epoch": 0.0001792, "grad_norm": 0.909222424030304, "learning_rate": 7.670822534187983e-07, "loss": 2.6156, "step": 341670 }, { "epoch": 0.0002048, "grad_norm": 0.8775516748428345, "learning_rate": 7.667729764289223e-07, "loss": 2.6796, "step": 341680 }, { "epoch": 0.0002304, "grad_norm": 0.8637421131134033, "learning_rate": 7.664637593140712e-07, "loss": 2.8429, "step": 341690 }, { "epoch": 0.000256, "grad_norm": 0.8584046363830566, "learning_rate": 7.661546020762467e-07, "loss": 2.801, "step": 341700 }, { "epoch": 0.0002816, "grad_norm": 0.8103947043418884, "learning_rate": 7.658455047174562e-07, "loss": 2.4977, "step": 341710 }, { "epoch": 0.0003072, "grad_norm": 0.8315973281860352, "learning_rate": 7.655364672397036e-07, "loss": 2.4746, "step": 341720 }, { "epoch": 0.0003328, "grad_norm": 0.8335447907447815, "learning_rate": 7.652274896449918e-07, "loss": 2.5478, "step": 341730 }, { "epoch": 0.0003584, "grad_norm": 0.8990256786346436, "learning_rate": 7.649185719353259e-07, "loss": 2.8372, "step": 341740 }, { "epoch": 0.000384, "grad_norm": 0.8749334216117859, "learning_rate": 7.646097141127085e-07, "loss": 2.6632, "step": 341750 }, { "epoch": 0.0004096, "grad_norm": 0.8882768154144287, "learning_rate": 7.643009161791448e-07, "loss": 2.8497, "step": 341760 }, { "epoch": 0.0004352, "grad_norm": 0.8392636179924011, "learning_rate": 7.639921781366333e-07, "loss": 3.0722, "step": 341770 }, { "epoch": 0.0004608, "grad_norm": 0.7901732921600342, "learning_rate": 7.636834999871778e-07, "loss": 2.6153, "step": 341780 }, { "epoch": 0.0004864, "grad_norm": 0.9490554928779602, "learning_rate": 7.633748817327802e-07, "loss": 2.6698, "step": 341790 }, { "epoch": 0.000512, "grad_norm": 0.7911054491996765, "learning_rate": 7.63066323375441e-07, "loss": 2.3318, "step": 341800 }, { "epoch": 0.0005376, "grad_norm": 0.8043571710586548, "learning_rate": 7.62757824917163e-07, "loss": 2.6385, "step": 341810 }, { "epoch": 0.0005632, "grad_norm": 0.943240761756897, "learning_rate": 7.624493863599447e-07, "loss": 2.4362, "step": 341820 }, { "epoch": 0.0005888, "grad_norm": 0.8315603137016296, "learning_rate": 7.621410077057867e-07, "loss": 2.5012, "step": 341830 }, { "epoch": 0.0006144, "grad_norm": 0.7616007328033447, "learning_rate": 7.618326889566896e-07, "loss": 2.406, "step": 341840 }, { "epoch": 0.00064, "grad_norm": 0.9276673197746277, "learning_rate": 7.615244301146529e-07, "loss": 2.6651, "step": 341850 }, { "epoch": 0.0006656, "grad_norm": 0.8605361580848694, "learning_rate": 7.61216231181674e-07, "loss": 2.7934, "step": 341860 }, { "epoch": 0.0006912, "grad_norm": 0.883929431438446, "learning_rate": 7.609080921597534e-07, "loss": 2.675, "step": 341870 }, { "epoch": 0.0007168, "grad_norm": 0.8686836361885071, "learning_rate": 7.606000130508872e-07, "loss": 2.6089, "step": 341880 }, { "epoch": 0.0007424, "grad_norm": 0.8245125412940979, "learning_rate": 7.602919938570774e-07, "loss": 2.5033, "step": 341890 }, { "epoch": 0.000768, "grad_norm": 0.935256838798523, "learning_rate": 7.599840345803155e-07, "loss": 2.7921, "step": 341900 }, { "epoch": 0.0007936, "grad_norm": 0.8552492260932922, "learning_rate": 7.596761352226023e-07, "loss": 2.7145, "step": 341910 }, { "epoch": 0.0008192, "grad_norm": 0.8905975222587585, "learning_rate": 7.593682957859327e-07, "loss": 2.689, "step": 341920 }, { "epoch": 0.0008448, "grad_norm": 0.8872944116592407, "learning_rate": 7.590605162723053e-07, "loss": 2.6576, "step": 341930 }, { "epoch": 0.0008704, "grad_norm": 0.9697219133377075, "learning_rate": 7.587527966837127e-07, "loss": 2.5246, "step": 341940 }, { "epoch": 0.000896, "grad_norm": 0.8227419853210449, "learning_rate": 7.584451370221535e-07, "loss": 2.588, "step": 341950 }, { "epoch": 0.0009216, "grad_norm": 0.8770915865898132, "learning_rate": 7.581375372896227e-07, "loss": 2.7601, "step": 341960 }, { "epoch": 0.0009472, "grad_norm": 0.8072284460067749, "learning_rate": 7.57829997488112e-07, "loss": 2.5225, "step": 341970 }, { "epoch": 0.0009728, "grad_norm": 0.8424193263053894, "learning_rate": 7.575225176196177e-07, "loss": 2.6857, "step": 341980 }, { "epoch": 0.0009984, "grad_norm": 0.8197401762008667, "learning_rate": 7.572150976861337e-07, "loss": 2.6589, "step": 341990 }, { "epoch": 0.001024, "grad_norm": 0.8581780791282654, "learning_rate": 7.569077376896516e-07, "loss": 2.7086, "step": 342000 }, { "epoch": 0.0010496, "grad_norm": 0.8816387057304382, "learning_rate": 7.566004376321679e-07, "loss": 2.8481, "step": 342010 }, { "epoch": 0.0010752, "grad_norm": 0.8315589427947998, "learning_rate": 7.562931975156751e-07, "loss": 2.6951, "step": 342020 }, { "epoch": 0.0011008, "grad_norm": 0.875503420829773, "learning_rate": 7.559860173421629e-07, "loss": 2.697, "step": 342030 }, { "epoch": 0.0011264, "grad_norm": 0.8467735052108765, "learning_rate": 7.556788971136242e-07, "loss": 2.4889, "step": 342040 }, { "epoch": 0.001152, "grad_norm": 0.8118081092834473, "learning_rate": 7.553718368320517e-07, "loss": 2.7439, "step": 342050 }, { "epoch": 0.0011776, "grad_norm": 0.8091480731964111, "learning_rate": 7.550648364994351e-07, "loss": 2.6328, "step": 342060 }, { "epoch": 0.0012032, "grad_norm": 0.8250882029533386, "learning_rate": 7.54757896117767e-07, "loss": 2.6917, "step": 342070 }, { "epoch": 0.0012288, "grad_norm": 0.811758279800415, "learning_rate": 7.544510156890361e-07, "loss": 2.7682, "step": 342080 }, { "epoch": 0.0012544, "grad_norm": 0.8536276817321777, "learning_rate": 7.541441952152351e-07, "loss": 2.7449, "step": 342090 }, { "epoch": 0.00128, "grad_norm": 0.8947057723999023, "learning_rate": 7.538374346983501e-07, "loss": 2.6642, "step": 342100 }, { "epoch": 0.0013056, "grad_norm": 0.8997709155082703, "learning_rate": 7.53530734140372e-07, "loss": 2.6321, "step": 342110 }, { "epoch": 0.0013312, "grad_norm": 0.9439429044723511, "learning_rate": 7.5322409354329e-07, "loss": 2.694, "step": 342120 }, { "epoch": 0.0013568, "grad_norm": 0.788945734500885, "learning_rate": 7.529175129090916e-07, "loss": 2.5168, "step": 342130 }, { "epoch": 0.0013824, "grad_norm": 0.8978392481803894, "learning_rate": 7.526109922397662e-07, "loss": 2.6683, "step": 342140 }, { "epoch": 0.001408, "grad_norm": 0.8132690787315369, "learning_rate": 7.523045315373001e-07, "loss": 2.5423, "step": 342150 }, { "epoch": 0.0014336, "grad_norm": 0.8059104084968567, "learning_rate": 7.519981308036828e-07, "loss": 2.8394, "step": 342160 }, { "epoch": 0.0014592, "grad_norm": 0.7393787503242493, "learning_rate": 7.516917900408971e-07, "loss": 2.6595, "step": 342170 }, { "epoch": 0.0014848, "grad_norm": 1.2300708293914795, "learning_rate": 7.513855092509347e-07, "loss": 2.2216, "step": 342180 }, { "epoch": 0.0015104, "grad_norm": 0.7957682013511658, "learning_rate": 7.510792884357776e-07, "loss": 2.0201, "step": 342190 }, { "epoch": 0.001536, "grad_norm": 0.7802578806877136, "learning_rate": 7.507731275974151e-07, "loss": 2.5663, "step": 342200 }, { "epoch": 2.56e-05, "grad_norm": 0.960769772529602, "learning_rate": 7.504670267378289e-07, "loss": 2.084, "step": 342210 }, { "epoch": 5.12e-05, "grad_norm": 0.8633583784103394, "learning_rate": 7.501609858590087e-07, "loss": 2.606, "step": 342220 }, { "epoch": 7.68e-05, "grad_norm": 0.8146755695343018, "learning_rate": 7.498550049629338e-07, "loss": 2.6999, "step": 342230 }, { "epoch": 0.0001024, "grad_norm": 0.9468700289726257, "learning_rate": 7.495490840515918e-07, "loss": 2.506, "step": 342240 }, { "epoch": 0.000128, "grad_norm": 0.8902190923690796, "learning_rate": 7.492432231269641e-07, "loss": 2.6488, "step": 342250 }, { "epoch": 0.0001536, "grad_norm": 0.8742616176605225, "learning_rate": 7.489374221910362e-07, "loss": 2.8716, "step": 342260 }, { "epoch": 0.0001792, "grad_norm": 0.8621306419372559, "learning_rate": 7.486316812457906e-07, "loss": 2.6674, "step": 342270 }, { "epoch": 0.0002048, "grad_norm": 0.7759837508201599, "learning_rate": 7.483260002932103e-07, "loss": 2.6224, "step": 342280 }, { "epoch": 0.0002304, "grad_norm": 0.7885217070579529, "learning_rate": 7.480203793352781e-07, "loss": 2.5722, "step": 342290 }, { "epoch": 0.000256, "grad_norm": 0.7893285155296326, "learning_rate": 7.477148183739735e-07, "loss": 2.6073, "step": 342300 }, { "epoch": 0.0002816, "grad_norm": 0.9548067450523376, "learning_rate": 7.474093174112784e-07, "loss": 2.621, "step": 342310 }, { "epoch": 0.0003072, "grad_norm": 0.9289996027946472, "learning_rate": 7.471038764491767e-07, "loss": 2.5789, "step": 342320 }, { "epoch": 0.0003328, "grad_norm": 1.0335609912872314, "learning_rate": 7.467984954896456e-07, "loss": 2.7526, "step": 342330 }, { "epoch": 0.0003584, "grad_norm": 0.8254112005233765, "learning_rate": 7.464931745346671e-07, "loss": 2.7283, "step": 342340 }, { "epoch": 0.000384, "grad_norm": 1.0781456232070923, "learning_rate": 7.461879135862216e-07, "loss": 2.5847, "step": 342350 }, { "epoch": 0.0004096, "grad_norm": 0.9325507283210754, "learning_rate": 7.458827126462887e-07, "loss": 2.7481, "step": 342360 }, { "epoch": 0.0004352, "grad_norm": 0.8981302380561829, "learning_rate": 7.455775717168467e-07, "loss": 2.7283, "step": 342370 }, { "epoch": 0.0004608, "grad_norm": 0.9122962951660156, "learning_rate": 7.452724907998743e-07, "loss": 2.6534, "step": 342380 }, { "epoch": 0.0004864, "grad_norm": 0.7869250178337097, "learning_rate": 7.449674698973497e-07, "loss": 2.7515, "step": 342390 }, { "epoch": 0.000512, "grad_norm": 0.7936570048332214, "learning_rate": 7.446625090112513e-07, "loss": 2.479, "step": 342400 }, { "epoch": 0.0005376, "grad_norm": 0.8742550611495972, "learning_rate": 7.443576081435578e-07, "loss": 2.6245, "step": 342410 }, { "epoch": 0.0005632, "grad_norm": 0.8635907769203186, "learning_rate": 7.44052767296246e-07, "loss": 2.51, "step": 342420 }, { "epoch": 0.0005888, "grad_norm": 0.8706178665161133, "learning_rate": 7.437479864712915e-07, "loss": 2.4912, "step": 342430 }, { "epoch": 0.0006144, "grad_norm": 0.9655483365058899, "learning_rate": 7.434432656706703e-07, "loss": 2.6195, "step": 342440 }, { "epoch": 0.00064, "grad_norm": 0.8885006904602051, "learning_rate": 7.431386048963596e-07, "loss": 2.5916, "step": 342450 }, { "epoch": 0.0006656, "grad_norm": 0.8958317637443542, "learning_rate": 7.428340041503357e-07, "loss": 2.9604, "step": 342460 }, { "epoch": 0.0006912, "grad_norm": 0.8743563294410706, "learning_rate": 7.425294634345725e-07, "loss": 2.6248, "step": 342470 }, { "epoch": 0.0007168, "grad_norm": 0.8502532243728638, "learning_rate": 7.422249827510452e-07, "loss": 2.4857, "step": 342480 }, { "epoch": 0.0007424, "grad_norm": 1.1050745248794556, "learning_rate": 7.4192056210173e-07, "loss": 2.7622, "step": 342490 }, { "epoch": 0.000768, "grad_norm": 0.8050664067268372, "learning_rate": 7.416162014885974e-07, "loss": 2.6599, "step": 342500 }, { "epoch": 0.0007936, "grad_norm": 0.8943543434143066, "learning_rate": 7.413119009136238e-07, "loss": 2.6459, "step": 342510 }, { "epoch": 0.0008192, "grad_norm": 0.951507568359375, "learning_rate": 7.410076603787808e-07, "loss": 2.5403, "step": 342520 }, { "epoch": 0.0008448, "grad_norm": 0.7824926972389221, "learning_rate": 7.407034798860436e-07, "loss": 2.7464, "step": 342530 }, { "epoch": 0.0008704, "grad_norm": 0.7935687899589539, "learning_rate": 7.403993594373849e-07, "loss": 2.8285, "step": 342540 }, { "epoch": 0.000896, "grad_norm": 0.8258506655693054, "learning_rate": 7.400952990347731e-07, "loss": 2.5897, "step": 342550 }, { "epoch": 0.0009216, "grad_norm": 0.7314659357070923, "learning_rate": 7.397912986801836e-07, "loss": 2.6335, "step": 342560 }, { "epoch": 0.0009472, "grad_norm": 0.8186291456222534, "learning_rate": 7.394873583755857e-07, "loss": 2.7278, "step": 342570 }, { "epoch": 0.0009728, "grad_norm": 0.9336840510368347, "learning_rate": 7.391834781229512e-07, "loss": 2.6936, "step": 342580 }, { "epoch": 0.0009984, "grad_norm": 0.9354785084724426, "learning_rate": 7.388796579242507e-07, "loss": 2.478, "step": 342590 }, { "epoch": 0.001024, "grad_norm": 0.8659514784812927, "learning_rate": 7.38575897781455e-07, "loss": 2.6388, "step": 342600 }, { "epoch": 0.0010496, "grad_norm": 0.8618926405906677, "learning_rate": 7.382721976965334e-07, "loss": 2.8072, "step": 342610 }, { "epoch": 0.0010752, "grad_norm": 1.0322256088256836, "learning_rate": 7.379685576714546e-07, "loss": 2.6977, "step": 342620 }, { "epoch": 0.0011008, "grad_norm": 0.7602007985115051, "learning_rate": 7.376649777081879e-07, "loss": 2.4244, "step": 342630 }, { "epoch": 0.0011264, "grad_norm": 0.8203611373901367, "learning_rate": 7.373614578087018e-07, "loss": 2.6553, "step": 342640 }, { "epoch": 0.001152, "grad_norm": 0.8398727774620056, "learning_rate": 7.370579979749659e-07, "loss": 2.9135, "step": 342650 }, { "epoch": 0.0011776, "grad_norm": 0.9624552726745605, "learning_rate": 7.367545982089463e-07, "loss": 2.3204, "step": 342660 }, { "epoch": 0.0012032, "grad_norm": 0.802445650100708, "learning_rate": 7.364512585126116e-07, "loss": 2.5997, "step": 342670 }, { "epoch": 0.0012288, "grad_norm": 0.8569592833518982, "learning_rate": 7.361479788879278e-07, "loss": 2.7815, "step": 342680 }, { "epoch": 0.0012544, "grad_norm": 0.9720642566680908, "learning_rate": 7.358447593368634e-07, "loss": 2.7816, "step": 342690 }, { "epoch": 0.00128, "grad_norm": 0.880774974822998, "learning_rate": 7.355415998613824e-07, "loss": 2.688, "step": 342700 }, { "epoch": 0.0013056, "grad_norm": 0.8754860758781433, "learning_rate": 7.352385004634532e-07, "loss": 2.5471, "step": 342710 }, { "epoch": 0.0013312, "grad_norm": 0.8682904243469238, "learning_rate": 7.349354611450388e-07, "loss": 2.6281, "step": 342720 }, { "epoch": 0.0013568, "grad_norm": 0.8780467510223389, "learning_rate": 7.346324819081064e-07, "loss": 2.6499, "step": 342730 }, { "epoch": 0.0013824, "grad_norm": 0.8128551244735718, "learning_rate": 7.34329562754621e-07, "loss": 2.65, "step": 342740 }, { "epoch": 0.001408, "grad_norm": 0.8415260314941406, "learning_rate": 7.340267036865445e-07, "loss": 2.6933, "step": 342750 }, { "epoch": 0.0014336, "grad_norm": 0.8513371348381042, "learning_rate": 7.337239047058431e-07, "loss": 2.8483, "step": 342760 }, { "epoch": 0.0014592, "grad_norm": 0.849898636341095, "learning_rate": 7.334211658144786e-07, "loss": 2.577, "step": 342770 }, { "epoch": 0.0014848, "grad_norm": 0.9013099074363708, "learning_rate": 7.331184870144148e-07, "loss": 2.3927, "step": 342780 }, { "epoch": 0.0015104, "grad_norm": 0.8252038359642029, "learning_rate": 7.328158683076159e-07, "loss": 2.4542, "step": 342790 }, { "epoch": 0.001536, "grad_norm": 0.8976205587387085, "learning_rate": 7.325133096960424e-07, "loss": 2.6365, "step": 342800 }, { "epoch": 0.0015616, "grad_norm": 0.892415463924408, "learning_rate": 7.322108111816584e-07, "loss": 2.6402, "step": 342810 }, { "epoch": 0.0015872, "grad_norm": 0.9456131458282471, "learning_rate": 7.319083727664234e-07, "loss": 2.8357, "step": 342820 }, { "epoch": 0.0016128, "grad_norm": 0.9038628935813904, "learning_rate": 7.31605994452298e-07, "loss": 2.5041, "step": 342830 }, { "epoch": 0.0016384, "grad_norm": 0.9736919403076172, "learning_rate": 7.313036762412462e-07, "loss": 2.7806, "step": 342840 }, { "epoch": 0.001664, "grad_norm": 0.889607846736908, "learning_rate": 7.310014181352255e-07, "loss": 2.7033, "step": 342850 }, { "epoch": 0.0016896, "grad_norm": 0.9388120770454407, "learning_rate": 7.306992201361974e-07, "loss": 2.8153, "step": 342860 }, { "epoch": 0.0017152, "grad_norm": 0.9308205842971802, "learning_rate": 7.303970822461226e-07, "loss": 2.8582, "step": 342870 }, { "epoch": 0.0017408, "grad_norm": 1.1321018934249878, "learning_rate": 7.300950044669575e-07, "loss": 2.4303, "step": 342880 }, { "epoch": 0.0017664, "grad_norm": 0.8967387080192566, "learning_rate": 7.297929868006637e-07, "loss": 2.8091, "step": 342890 }, { "epoch": 0.001792, "grad_norm": 0.7765092253684998, "learning_rate": 7.294910292491986e-07, "loss": 2.3596, "step": 342900 }, { "epoch": 0.0018176, "grad_norm": 1.1058441400527954, "learning_rate": 7.291891318145195e-07, "loss": 2.7445, "step": 342910 }, { "epoch": 0.0018432, "grad_norm": 1.3595207929611206, "learning_rate": 7.288872944985858e-07, "loss": 2.0203, "step": 342920 }, { "epoch": 0.0018688, "grad_norm": 0.7968209981918335, "learning_rate": 7.28585517303354e-07, "loss": 2.7863, "step": 342930 }, { "epoch": 0.0018944, "grad_norm": 0.9367736577987671, "learning_rate": 7.282838002307824e-07, "loss": 2.6828, "step": 342940 }, { "epoch": 0.00192, "grad_norm": 0.8233746290206909, "learning_rate": 7.279821432828249e-07, "loss": 2.595, "step": 342950 }, { "epoch": 0.0019456, "grad_norm": 1.5259320735931396, "learning_rate": 7.27680546461439e-07, "loss": 2.5711, "step": 342960 }, { "epoch": 0.0019712, "grad_norm": 0.7796148657798767, "learning_rate": 7.273790097685796e-07, "loss": 2.8426, "step": 342970 }, { "epoch": 0.0019968, "grad_norm": 0.9469735622406006, "learning_rate": 7.270775332062042e-07, "loss": 2.5654, "step": 342980 }, { "epoch": 0.0020224, "grad_norm": 0.8180961608886719, "learning_rate": 7.267761167762655e-07, "loss": 2.4186, "step": 342990 }, { "epoch": 0.002048, "grad_norm": 0.8859190940856934, "learning_rate": 7.264747604807199e-07, "loss": 2.7647, "step": 343000 }, { "epoch": 0.0020736, "grad_norm": 0.876622200012207, "learning_rate": 7.261734643215224e-07, "loss": 2.6404, "step": 343010 }, { "epoch": 0.0020992, "grad_norm": 0.8435524702072144, "learning_rate": 7.258722283006236e-07, "loss": 2.809, "step": 343020 }, { "epoch": 0.0021248, "grad_norm": 0.8334654569625854, "learning_rate": 7.255710524199777e-07, "loss": 2.7474, "step": 343030 }, { "epoch": 0.0021504, "grad_norm": 0.9551721215248108, "learning_rate": 7.252699366815386e-07, "loss": 2.8925, "step": 343040 }, { "epoch": 0.002176, "grad_norm": 0.8429842591285706, "learning_rate": 7.249688810872602e-07, "loss": 2.7645, "step": 343050 }, { "epoch": 0.0022016, "grad_norm": 0.8226860761642456, "learning_rate": 7.246678856390932e-07, "loss": 2.5757, "step": 343060 }, { "epoch": 0.0022272, "grad_norm": 0.9286614656448364, "learning_rate": 7.243669503389917e-07, "loss": 2.7643, "step": 343070 }, { "epoch": 0.0022528, "grad_norm": 0.8988156318664551, "learning_rate": 7.240660751889028e-07, "loss": 2.559, "step": 343080 }, { "epoch": 0.0022784, "grad_norm": 0.8978911638259888, "learning_rate": 7.237652601907818e-07, "loss": 2.8844, "step": 343090 }, { "epoch": 0.002304, "grad_norm": 1.0079573392868042, "learning_rate": 7.234645053465772e-07, "loss": 2.4821, "step": 343100 }, { "epoch": 0.0023296, "grad_norm": 1.0205479860305786, "learning_rate": 7.231638106582395e-07, "loss": 2.7198, "step": 343110 }, { "epoch": 0.0023552, "grad_norm": 1.1447583436965942, "learning_rate": 7.228631761277183e-07, "loss": 2.526, "step": 343120 }, { "epoch": 0.0023808, "grad_norm": 0.8688279390335083, "learning_rate": 7.225626017569653e-07, "loss": 2.8098, "step": 343130 }, { "epoch": 0.0024064, "grad_norm": 0.7968810796737671, "learning_rate": 7.222620875479292e-07, "loss": 2.5579, "step": 343140 }, { "epoch": 0.002432, "grad_norm": 0.8341289162635803, "learning_rate": 7.219616335025559e-07, "loss": 2.5952, "step": 343150 }, { "epoch": 0.0024576, "grad_norm": 0.8034313321113586, "learning_rate": 7.216612396227962e-07, "loss": 2.6838, "step": 343160 }, { "epoch": 0.0024832, "grad_norm": 0.7933242917060852, "learning_rate": 7.213609059105964e-07, "loss": 2.5705, "step": 343170 }, { "epoch": 0.0025088, "grad_norm": 0.9636204242706299, "learning_rate": 7.210606323679059e-07, "loss": 2.7744, "step": 343180 }, { "epoch": 0.0025344, "grad_norm": 0.8104730844497681, "learning_rate": 7.20760418996671e-07, "loss": 2.6596, "step": 343190 }, { "epoch": 0.00256, "grad_norm": 0.9159190058708191, "learning_rate": 7.204602657988391e-07, "loss": 2.7507, "step": 343200 }, { "epoch": 0.0025856, "grad_norm": 0.9362682700157166, "learning_rate": 7.201601727763563e-07, "loss": 2.3565, "step": 343210 }, { "epoch": 0.0026112, "grad_norm": 0.8749353289604187, "learning_rate": 7.198601399311678e-07, "loss": 2.6557, "step": 343220 }, { "epoch": 0.0026368, "grad_norm": 0.8286206126213074, "learning_rate": 7.195601672652197e-07, "loss": 2.773, "step": 343230 }, { "epoch": 0.0026624, "grad_norm": 1.1672457456588745, "learning_rate": 7.192602547804583e-07, "loss": 2.819, "step": 343240 }, { "epoch": 0.002688, "grad_norm": 0.8777638673782349, "learning_rate": 7.189604024788277e-07, "loss": 2.6939, "step": 343250 }, { "epoch": 0.0027136, "grad_norm": 0.8196470737457275, "learning_rate": 7.186606103622717e-07, "loss": 2.7266, "step": 343260 }, { "epoch": 0.0027392, "grad_norm": 0.9301405549049377, "learning_rate": 7.183608784327367e-07, "loss": 2.7397, "step": 343270 }, { "epoch": 0.0027648, "grad_norm": 0.994402289390564, "learning_rate": 7.180612066921621e-07, "loss": 2.9023, "step": 343280 }, { "epoch": 0.0027904, "grad_norm": 0.8726357221603394, "learning_rate": 7.177615951424955e-07, "loss": 2.7985, "step": 343290 }, { "epoch": 0.002816, "grad_norm": 0.8683280944824219, "learning_rate": 7.174620437856772e-07, "loss": 2.7251, "step": 343300 }, { "epoch": 0.0028416, "grad_norm": 1.0626107454299927, "learning_rate": 7.171625526236503e-07, "loss": 2.8701, "step": 343310 }, { "epoch": 0.0028672, "grad_norm": 0.8126087188720703, "learning_rate": 7.168631216583576e-07, "loss": 2.9013, "step": 343320 }, { "epoch": 0.0028928, "grad_norm": 0.9782446026802063, "learning_rate": 7.1656375089174e-07, "loss": 2.7489, "step": 343330 }, { "epoch": 0.0029184, "grad_norm": 0.8816097974777222, "learning_rate": 7.162644403257412e-07, "loss": 2.6913, "step": 343340 }, { "epoch": 0.002944, "grad_norm": 1.0381242036819458, "learning_rate": 7.159651899622977e-07, "loss": 2.9102, "step": 343350 }, { "epoch": 0.0029696, "grad_norm": 0.7921067476272583, "learning_rate": 7.156659998033533e-07, "loss": 2.6836, "step": 343360 }, { "epoch": 0.0029952, "grad_norm": 1.04803466796875, "learning_rate": 7.153668698508476e-07, "loss": 2.9954, "step": 343370 }, { "epoch": 0.0030208, "grad_norm": 0.8336398005485535, "learning_rate": 7.150678001067201e-07, "loss": 2.9138, "step": 343380 }, { "epoch": 0.0030464, "grad_norm": 0.8911908864974976, "learning_rate": 7.147687905729084e-07, "loss": 2.7479, "step": 343390 }, { "epoch": 0.003072, "grad_norm": 0.7938947677612305, "learning_rate": 7.144698412513573e-07, "loss": 2.5724, "step": 343400 }, { "epoch": 2.56e-05, "grad_norm": 0.8646548390388489, "learning_rate": 7.141709521439988e-07, "loss": 2.2173, "step": 343410 }, { "epoch": 5.12e-05, "grad_norm": 0.7605156302452087, "learning_rate": 7.138721232527746e-07, "loss": 2.6039, "step": 343420 }, { "epoch": 7.68e-05, "grad_norm": 0.7806981205940247, "learning_rate": 7.135733545796209e-07, "loss": 2.6066, "step": 343430 }, { "epoch": 0.0001024, "grad_norm": 0.8828171491622925, "learning_rate": 7.132746461264762e-07, "loss": 2.5475, "step": 343440 }, { "epoch": 0.000128, "grad_norm": 0.957892119884491, "learning_rate": 7.129759978952777e-07, "loss": 2.6395, "step": 343450 }, { "epoch": 0.0001536, "grad_norm": 0.8503103852272034, "learning_rate": 7.126774098879619e-07, "loss": 2.524, "step": 343460 }, { "epoch": 0.0001792, "grad_norm": 0.8658186197280884, "learning_rate": 7.123788821064659e-07, "loss": 2.6538, "step": 343470 }, { "epoch": 0.0002048, "grad_norm": 0.8674077987670898, "learning_rate": 7.120804145527227e-07, "loss": 2.9123, "step": 343480 }, { "epoch": 0.0002304, "grad_norm": 0.8532932996749878, "learning_rate": 7.117820072286708e-07, "loss": 2.5229, "step": 343490 }, { "epoch": 0.000256, "grad_norm": 0.7974575161933899, "learning_rate": 7.114836601362429e-07, "loss": 2.7074, "step": 343500 }, { "epoch": 0.0002816, "grad_norm": 0.8456234931945801, "learning_rate": 7.111853732773755e-07, "loss": 2.5908, "step": 343510 }, { "epoch": 0.0003072, "grad_norm": 0.8002519011497498, "learning_rate": 7.108871466540012e-07, "loss": 2.7171, "step": 343520 }, { "epoch": 0.0003328, "grad_norm": 1.099077582359314, "learning_rate": 7.105889802680554e-07, "loss": 2.3036, "step": 343530 }, { "epoch": 0.0003584, "grad_norm": 0.89884352684021, "learning_rate": 7.102908741214742e-07, "loss": 3.0073, "step": 343540 }, { "epoch": 0.000384, "grad_norm": 0.9182901382446289, "learning_rate": 7.099928282161849e-07, "loss": 2.6308, "step": 343550 }, { "epoch": 0.0004096, "grad_norm": 0.8961630463600159, "learning_rate": 7.096948425541217e-07, "loss": 2.4837, "step": 343560 }, { "epoch": 0.0004352, "grad_norm": 0.9138275980949402, "learning_rate": 7.093969171372195e-07, "loss": 2.5191, "step": 343570 }, { "epoch": 0.0004608, "grad_norm": 0.80874103307724, "learning_rate": 7.090990519674102e-07, "loss": 2.5833, "step": 343580 }, { "epoch": 0.0004864, "grad_norm": 0.9969192743301392, "learning_rate": 7.088012470466232e-07, "loss": 2.7532, "step": 343590 }, { "epoch": 0.000512, "grad_norm": 0.9752798080444336, "learning_rate": 7.085035023767939e-07, "loss": 2.6263, "step": 343600 }, { "epoch": 0.0005376, "grad_norm": 0.8835785984992981, "learning_rate": 7.082058179598472e-07, "loss": 2.4879, "step": 343610 }, { "epoch": 0.0005632, "grad_norm": 0.9059610366821289, "learning_rate": 7.079081937977172e-07, "loss": 2.6353, "step": 343620 }, { "epoch": 0.0005888, "grad_norm": 0.9689130783081055, "learning_rate": 7.076106298923324e-07, "loss": 2.4441, "step": 343630 }, { "epoch": 0.0006144, "grad_norm": 0.7440598011016846, "learning_rate": 7.073131262456224e-07, "loss": 2.4249, "step": 343640 }, { "epoch": 0.00064, "grad_norm": 0.8670905232429504, "learning_rate": 7.070156828595176e-07, "loss": 2.574, "step": 343650 }, { "epoch": 0.0006656, "grad_norm": 0.8645307421684265, "learning_rate": 7.067182997359467e-07, "loss": 2.8463, "step": 343660 }, { "epoch": 0.0006912, "grad_norm": 0.8226223587989807, "learning_rate": 7.064209768768393e-07, "loss": 2.7068, "step": 343670 }, { "epoch": 0.0007168, "grad_norm": 0.9927654266357422, "learning_rate": 7.061237142841204e-07, "loss": 2.6671, "step": 343680 }, { "epoch": 0.0007424, "grad_norm": 0.8319278955459595, "learning_rate": 7.058265119597185e-07, "loss": 2.8161, "step": 343690 }, { "epoch": 0.000768, "grad_norm": 0.9346502423286438, "learning_rate": 7.055293699055621e-07, "loss": 2.8286, "step": 343700 }, { "epoch": 0.0007936, "grad_norm": 0.8127217888832092, "learning_rate": 7.052322881235785e-07, "loss": 2.4823, "step": 343710 }, { "epoch": 0.0008192, "grad_norm": 0.8816305994987488, "learning_rate": 7.049352666156917e-07, "loss": 2.7593, "step": 343720 }, { "epoch": 0.0008448, "grad_norm": 0.8486818075180054, "learning_rate": 7.046383053838302e-07, "loss": 2.6315, "step": 343730 }, { "epoch": 0.0008704, "grad_norm": 0.8198041319847107, "learning_rate": 7.04341404429918e-07, "loss": 2.5858, "step": 343740 }, { "epoch": 0.000896, "grad_norm": 0.8390738368034363, "learning_rate": 7.040445637558824e-07, "loss": 2.6203, "step": 343750 }, { "epoch": 0.0009216, "grad_norm": 1.3605895042419434, "learning_rate": 7.037477833636463e-07, "loss": 2.6748, "step": 343760 }, { "epoch": 0.0009472, "grad_norm": 0.7577134966850281, "learning_rate": 7.034510632551361e-07, "loss": 2.4999, "step": 343770 }, { "epoch": 0.0009728, "grad_norm": 0.8347252011299133, "learning_rate": 7.031544034322746e-07, "loss": 2.678, "step": 343780 }, { "epoch": 0.0009984, "grad_norm": 0.8085824251174927, "learning_rate": 7.028578038969858e-07, "loss": 2.7393, "step": 343790 }, { "epoch": 0.001024, "grad_norm": 0.8771748542785645, "learning_rate": 7.025612646511947e-07, "loss": 2.6119, "step": 343800 }, { "epoch": 0.0010496, "grad_norm": 0.8321967720985413, "learning_rate": 7.022647856968223e-07, "loss": 2.5688, "step": 343810 }, { "epoch": 0.0010752, "grad_norm": 0.8613590002059937, "learning_rate": 7.019683670357901e-07, "loss": 2.7518, "step": 343820 }, { "epoch": 0.0011008, "grad_norm": 0.9075140357017517, "learning_rate": 7.016720086700235e-07, "loss": 2.6566, "step": 343830 }, { "epoch": 0.0011264, "grad_norm": 0.9060090780258179, "learning_rate": 7.013757106014418e-07, "loss": 2.4476, "step": 343840 }, { "epoch": 0.001152, "grad_norm": 0.8945751190185547, "learning_rate": 7.010794728319681e-07, "loss": 2.699, "step": 343850 }, { "epoch": 0.0011776, "grad_norm": 0.9871870875358582, "learning_rate": 7.007832953635219e-07, "loss": 2.8373, "step": 343860 }, { "epoch": 0.0012032, "grad_norm": 0.9579872488975525, "learning_rate": 7.00487178198026e-07, "loss": 2.7927, "step": 343870 }, { "epoch": 0.0012288, "grad_norm": 0.7698272466659546, "learning_rate": 7.00191121337398e-07, "loss": 2.5167, "step": 343880 }, { "epoch": 0.0012544, "grad_norm": 0.8240554332733154, "learning_rate": 6.998951247835594e-07, "loss": 2.8058, "step": 343890 }, { "epoch": 0.00128, "grad_norm": 0.9055245518684387, "learning_rate": 6.99599188538429e-07, "loss": 2.7628, "step": 343900 }, { "epoch": 0.0013056, "grad_norm": 0.922130823135376, "learning_rate": 6.99303312603924e-07, "loss": 2.7794, "step": 343910 }, { "epoch": 0.0013312, "grad_norm": 0.9423545002937317, "learning_rate": 6.990074969819693e-07, "loss": 2.7785, "step": 343920 }, { "epoch": 0.0013568, "grad_norm": 0.953861653804779, "learning_rate": 6.987117416744771e-07, "loss": 2.501, "step": 343930 }, { "epoch": 0.0013824, "grad_norm": 1.2404768466949463, "learning_rate": 6.984160466833667e-07, "loss": 2.906, "step": 343940 }, { "epoch": 0.001408, "grad_norm": 0.8273583054542542, "learning_rate": 6.981204120105566e-07, "loss": 2.5623, "step": 343950 }, { "epoch": 0.0014336, "grad_norm": 0.8285709023475647, "learning_rate": 6.978248376579632e-07, "loss": 2.617, "step": 343960 }, { "epoch": 0.0014592, "grad_norm": 0.7989079356193542, "learning_rate": 6.975293236275038e-07, "loss": 2.5259, "step": 343970 }, { "epoch": 0.0014848, "grad_norm": 0.8023689389228821, "learning_rate": 6.972338699210946e-07, "loss": 2.535, "step": 343980 }, { "epoch": 0.0015104, "grad_norm": 1.1308578252792358, "learning_rate": 6.969384765406506e-07, "loss": 2.1436, "step": 343990 }, { "epoch": 0.001536, "grad_norm": 0.8249325752258301, "learning_rate": 6.966431434880905e-07, "loss": 2.6582, "step": 344000 }, { "epoch": 0.0015616, "grad_norm": 0.8673208951950073, "learning_rate": 6.963478707653259e-07, "loss": 2.6647, "step": 344010 }, { "epoch": 0.0015872, "grad_norm": 0.9027307629585266, "learning_rate": 6.960526583742722e-07, "loss": 2.7601, "step": 344020 }, { "epoch": 0.0016128, "grad_norm": 0.8104478120803833, "learning_rate": 6.957575063168442e-07, "loss": 2.6368, "step": 344030 }, { "epoch": 0.0016384, "grad_norm": 0.8186855912208557, "learning_rate": 6.954624145949562e-07, "loss": 2.5707, "step": 344040 }, { "epoch": 0.001664, "grad_norm": 1.3489717245101929, "learning_rate": 6.95167383210521e-07, "loss": 2.7351, "step": 344050 }, { "epoch": 0.0016896, "grad_norm": 0.9761861562728882, "learning_rate": 6.948724121654526e-07, "loss": 2.6587, "step": 344060 }, { "epoch": 0.0017152, "grad_norm": 0.8220334053039551, "learning_rate": 6.945775014616662e-07, "loss": 2.3441, "step": 344070 }, { "epoch": 0.0017408, "grad_norm": 0.8546478748321533, "learning_rate": 6.942826511010669e-07, "loss": 2.7092, "step": 344080 }, { "epoch": 0.0017664, "grad_norm": 0.9035123586654663, "learning_rate": 6.939878610855743e-07, "loss": 2.3974, "step": 344090 }, { "epoch": 0.001792, "grad_norm": 0.9227421879768372, "learning_rate": 6.936931314170969e-07, "loss": 2.2633, "step": 344100 }, { "epoch": 0.0018176, "grad_norm": 0.9700284600257874, "learning_rate": 6.933984620975454e-07, "loss": 2.5714, "step": 344110 }, { "epoch": 0.0018432, "grad_norm": 0.7935689687728882, "learning_rate": 6.931038531288337e-07, "loss": 2.3821, "step": 344120 }, { "epoch": 0.0018688, "grad_norm": 0.8956437706947327, "learning_rate": 6.928093045128692e-07, "loss": 2.8074, "step": 344130 }, { "epoch": 0.0018944, "grad_norm": 0.827924907207489, "learning_rate": 6.925148162515616e-07, "loss": 2.5002, "step": 344140 }, { "epoch": 0.00192, "grad_norm": 0.8455248475074768, "learning_rate": 6.922203883468226e-07, "loss": 2.614, "step": 344150 }, { "epoch": 0.0019456, "grad_norm": 0.7756439447402954, "learning_rate": 6.919260208005596e-07, "loss": 2.8677, "step": 344160 }, { "epoch": 0.0019712, "grad_norm": 0.8365830183029175, "learning_rate": 6.916317136146833e-07, "loss": 3.0408, "step": 344170 }, { "epoch": 0.0019968, "grad_norm": 0.7934269309043884, "learning_rate": 6.91337466791101e-07, "loss": 2.5293, "step": 344180 }, { "epoch": 0.0020224, "grad_norm": 0.8622098565101624, "learning_rate": 6.910432803317224e-07, "loss": 2.5921, "step": 344190 }, { "epoch": 0.002048, "grad_norm": 1.6052638292312622, "learning_rate": 6.907491542384548e-07, "loss": 2.6172, "step": 344200 }, { "epoch": 0.0020736, "grad_norm": 0.7508506178855896, "learning_rate": 6.904550885132033e-07, "loss": 2.326, "step": 344210 }, { "epoch": 0.0020992, "grad_norm": 0.8261383771896362, "learning_rate": 6.901610831578764e-07, "loss": 2.7576, "step": 344220 }, { "epoch": 0.0021248, "grad_norm": 0.9754723310470581, "learning_rate": 6.898671381743804e-07, "loss": 2.6945, "step": 344230 }, { "epoch": 0.0021504, "grad_norm": 0.8952194452285767, "learning_rate": 6.895732535646226e-07, "loss": 2.9478, "step": 344240 }, { "epoch": 0.002176, "grad_norm": 0.8920625448226929, "learning_rate": 6.892794293305072e-07, "loss": 2.7926, "step": 344250 }, { "epoch": 0.0022016, "grad_norm": 0.8932071924209595, "learning_rate": 6.889856654739402e-07, "loss": 2.8096, "step": 344260 }, { "epoch": 0.0022272, "grad_norm": 0.8782886862754822, "learning_rate": 6.886919619968268e-07, "loss": 2.7179, "step": 344270 }, { "epoch": 0.0022528, "grad_norm": 0.7996385097503662, "learning_rate": 6.883983189010712e-07, "loss": 2.5615, "step": 344280 }, { "epoch": 0.0022784, "grad_norm": 0.7945809364318848, "learning_rate": 6.881047361885773e-07, "loss": 2.9585, "step": 344290 }, { "epoch": 0.002304, "grad_norm": 0.8815982341766357, "learning_rate": 6.878112138612503e-07, "loss": 2.7263, "step": 344300 }, { "epoch": 0.0023296, "grad_norm": 1.0541378259658813, "learning_rate": 6.875177519209919e-07, "loss": 2.5937, "step": 344310 }, { "epoch": 0.0023552, "grad_norm": 0.8091428279876709, "learning_rate": 6.872243503697085e-07, "loss": 2.6486, "step": 344320 }, { "epoch": 0.0023808, "grad_norm": 0.7856493592262268, "learning_rate": 6.869310092092974e-07, "loss": 2.3102, "step": 344330 }, { "epoch": 0.0024064, "grad_norm": 0.8782237768173218, "learning_rate": 6.866377284416636e-07, "loss": 2.6336, "step": 344340 }, { "epoch": 0.002432, "grad_norm": 0.8498802781105042, "learning_rate": 6.863445080687103e-07, "loss": 2.7188, "step": 344350 }, { "epoch": 0.0024576, "grad_norm": 0.8431428074836731, "learning_rate": 6.86051348092337e-07, "loss": 2.5445, "step": 344360 }, { "epoch": 0.0024832, "grad_norm": 0.8756846189498901, "learning_rate": 6.857582485144443e-07, "loss": 2.7902, "step": 344370 }, { "epoch": 0.0025088, "grad_norm": 0.892204999923706, "learning_rate": 6.854652093369351e-07, "loss": 2.8611, "step": 344380 }, { "epoch": 0.0025344, "grad_norm": 1.252637505531311, "learning_rate": 6.851722305617081e-07, "loss": 2.8342, "step": 344390 }, { "epoch": 0.00256, "grad_norm": 1.2621128559112549, "learning_rate": 6.848793121906649e-07, "loss": 3.0413, "step": 344400 }, { "epoch": 0.0025856, "grad_norm": 0.8570565581321716, "learning_rate": 6.84586454225703e-07, "loss": 2.7421, "step": 344410 }, { "epoch": 0.0026112, "grad_norm": 0.9202445149421692, "learning_rate": 6.842936566687208e-07, "loss": 2.8084, "step": 344420 }, { "epoch": 0.0026368, "grad_norm": 0.8737165927886963, "learning_rate": 6.840009195216179e-07, "loss": 2.763, "step": 344430 }, { "epoch": 0.0026624, "grad_norm": 0.8419034481048584, "learning_rate": 6.837082427862951e-07, "loss": 2.8828, "step": 344440 }, { "epoch": 0.002688, "grad_norm": 0.8132434487342834, "learning_rate": 6.834156264646485e-07, "loss": 2.8182, "step": 344450 }, { "epoch": 0.0027136, "grad_norm": 0.8042832016944885, "learning_rate": 6.831230705585757e-07, "loss": 2.8436, "step": 344460 }, { "epoch": 0.0027392, "grad_norm": 0.730469822883606, "learning_rate": 6.828305750699727e-07, "loss": 2.5749, "step": 344470 }, { "epoch": 0.0027648, "grad_norm": 0.9157702922821045, "learning_rate": 6.825381400007369e-07, "loss": 2.4854, "step": 344480 }, { "epoch": 0.0027904, "grad_norm": 0.98307865858078, "learning_rate": 6.822457653527648e-07, "loss": 2.6519, "step": 344490 }, { "epoch": 0.002816, "grad_norm": 0.8421657681465149, "learning_rate": 6.819534511279536e-07, "loss": 2.7041, "step": 344500 }, { "epoch": 0.0028416, "grad_norm": 0.8629916906356812, "learning_rate": 6.816611973281972e-07, "loss": 2.8925, "step": 344510 }, { "epoch": 0.0028672, "grad_norm": 0.891609787940979, "learning_rate": 6.813690039553933e-07, "loss": 2.7087, "step": 344520 }, { "epoch": 0.0028928, "grad_norm": 0.9410315155982971, "learning_rate": 6.810768710114324e-07, "loss": 2.8716, "step": 344530 }, { "epoch": 0.0029184, "grad_norm": 0.9794909358024597, "learning_rate": 6.807847984982119e-07, "loss": 2.8281, "step": 344540 }, { "epoch": 0.002944, "grad_norm": 0.8928069472312927, "learning_rate": 6.804927864176259e-07, "loss": 2.8316, "step": 344550 }, { "epoch": 0.0029696, "grad_norm": 0.8660265803337097, "learning_rate": 6.802008347715672e-07, "loss": 2.9028, "step": 344560 }, { "epoch": 0.0029952, "grad_norm": 0.8102326393127441, "learning_rate": 6.799089435619277e-07, "loss": 2.9864, "step": 344570 }, { "epoch": 0.0030208, "grad_norm": 0.8142077922821045, "learning_rate": 6.796171127906037e-07, "loss": 2.756, "step": 344580 }, { "epoch": 0.0030464, "grad_norm": 0.9206501245498657, "learning_rate": 6.793253424594859e-07, "loss": 2.8752, "step": 344590 }, { "epoch": 0.003072, "grad_norm": 0.9366231560707092, "learning_rate": 6.790336325704649e-07, "loss": 2.8184, "step": 344600 }, { "epoch": 0.0030976, "grad_norm": 0.9640365242958069, "learning_rate": 6.787419831254316e-07, "loss": 3.1336, "step": 344610 }, { "epoch": 0.0031232, "grad_norm": 0.8283284306526184, "learning_rate": 6.78450394126281e-07, "loss": 2.7784, "step": 344620 }, { "epoch": 0.0031488, "grad_norm": 0.8318033218383789, "learning_rate": 6.781588655749027e-07, "loss": 2.8825, "step": 344630 }, { "epoch": 0.0031744, "grad_norm": 0.9474977254867554, "learning_rate": 6.778673974731864e-07, "loss": 2.8673, "step": 344640 }, { "epoch": 0.0032, "grad_norm": 0.8971545696258545, "learning_rate": 6.77575989823025e-07, "loss": 2.6939, "step": 344650 }, { "epoch": 0.0032256, "grad_norm": 0.8157772421836853, "learning_rate": 6.772846426263025e-07, "loss": 2.8172, "step": 344660 }, { "epoch": 0.0032512, "grad_norm": 0.8992791175842285, "learning_rate": 6.769933558849129e-07, "loss": 2.6035, "step": 344670 }, { "epoch": 0.0032768, "grad_norm": 0.8970391750335693, "learning_rate": 6.767021296007426e-07, "loss": 2.9925, "step": 344680 }, { "epoch": 0.0033024, "grad_norm": 0.8185625076293945, "learning_rate": 6.764109637756811e-07, "loss": 2.4881, "step": 344690 }, { "epoch": 0.003328, "grad_norm": 0.8909828662872314, "learning_rate": 6.761198584116168e-07, "loss": 2.8747, "step": 344700 }, { "epoch": 0.0033536, "grad_norm": 0.8225047588348389, "learning_rate": 6.758288135104374e-07, "loss": 2.6745, "step": 344710 }, { "epoch": 0.0033792, "grad_norm": 0.9924147129058838, "learning_rate": 6.755378290740322e-07, "loss": 2.6862, "step": 344720 }, { "epoch": 0.0034048, "grad_norm": 0.8229420185089111, "learning_rate": 6.752469051042843e-07, "loss": 2.7992, "step": 344730 }, { "epoch": 0.0034304, "grad_norm": 0.9917294383049011, "learning_rate": 6.749560416030809e-07, "loss": 2.8338, "step": 344740 }, { "epoch": 0.003456, "grad_norm": 0.8964492082595825, "learning_rate": 6.746652385723095e-07, "loss": 2.7138, "step": 344750 }, { "epoch": 0.0034816, "grad_norm": 0.837884783744812, "learning_rate": 6.743744960138565e-07, "loss": 2.7436, "step": 344760 }, { "epoch": 0.0035072, "grad_norm": 0.8489679098129272, "learning_rate": 6.740838139296057e-07, "loss": 2.7431, "step": 344770 }, { "epoch": 0.0035328, "grad_norm": 0.8858928680419922, "learning_rate": 6.737931923214436e-07, "loss": 2.6396, "step": 344780 }, { "epoch": 0.0035584, "grad_norm": 0.8934288620948792, "learning_rate": 6.73502631191254e-07, "loss": 2.9529, "step": 344790 }, { "epoch": 0.003584, "grad_norm": 0.8598085045814514, "learning_rate": 6.7321213054092e-07, "loss": 2.8039, "step": 344800 }, { "epoch": 0.0036096, "grad_norm": 0.9121577143669128, "learning_rate": 6.729216903723279e-07, "loss": 2.8236, "step": 344810 }, { "epoch": 0.0036352, "grad_norm": 0.9370828866958618, "learning_rate": 6.726313106873594e-07, "loss": 2.8573, "step": 344820 }, { "epoch": 0.0036608, "grad_norm": 0.8436880707740784, "learning_rate": 6.723409914878976e-07, "loss": 2.8236, "step": 344830 }, { "epoch": 0.0036864, "grad_norm": 0.8501610159873962, "learning_rate": 6.720507327758252e-07, "loss": 2.8509, "step": 344840 }, { "epoch": 0.003712, "grad_norm": 0.896030843257904, "learning_rate": 6.717605345530276e-07, "loss": 2.9163, "step": 344850 }, { "epoch": 0.0037376, "grad_norm": 0.8655058145523071, "learning_rate": 6.714703968213809e-07, "loss": 3.0062, "step": 344860 }, { "epoch": 0.0037632, "grad_norm": 0.9569944739341736, "learning_rate": 6.711803195827704e-07, "loss": 2.9971, "step": 344870 }, { "epoch": 0.0037888, "grad_norm": 0.8115509152412415, "learning_rate": 6.708903028390757e-07, "loss": 2.8559, "step": 344880 }, { "epoch": 0.0038144, "grad_norm": 0.848127007484436, "learning_rate": 6.706003465921773e-07, "loss": 2.7233, "step": 344890 }, { "epoch": 0.00384, "grad_norm": 0.7816829085350037, "learning_rate": 6.703104508439573e-07, "loss": 2.6458, "step": 344900 }, { "epoch": 0.0038656, "grad_norm": 0.8924999833106995, "learning_rate": 6.70020615596294e-07, "loss": 2.6259, "step": 344910 }, { "epoch": 0.0038912, "grad_norm": 0.8700714707374573, "learning_rate": 6.697308408510695e-07, "loss": 2.9997, "step": 344920 }, { "epoch": 0.0039168, "grad_norm": 0.8466795086860657, "learning_rate": 6.694411266101586e-07, "loss": 2.9052, "step": 344930 }, { "epoch": 0.0039424, "grad_norm": 0.808589518070221, "learning_rate": 6.691514728754422e-07, "loss": 2.8536, "step": 344940 }, { "epoch": 0.003968, "grad_norm": 0.9173145294189453, "learning_rate": 6.688618796487978e-07, "loss": 2.7942, "step": 344950 }, { "epoch": 0.0039936, "grad_norm": 0.7702368497848511, "learning_rate": 6.685723469321048e-07, "loss": 2.7605, "step": 344960 }, { "epoch": 0.0040192, "grad_norm": 0.8220256567001343, "learning_rate": 6.682828747272407e-07, "loss": 2.8334, "step": 344970 }, { "epoch": 0.0040448, "grad_norm": 0.8887747526168823, "learning_rate": 6.67993463036084e-07, "loss": 2.796, "step": 344980 }, { "epoch": 0.0040704, "grad_norm": 0.8062780499458313, "learning_rate": 6.677041118605076e-07, "loss": 2.9562, "step": 344990 }, { "epoch": 0.004096, "grad_norm": 0.9473998546600342, "learning_rate": 6.67414821202389e-07, "loss": 2.8638, "step": 345000 }, { "epoch": 0.0041216, "grad_norm": 0.9223604202270508, "learning_rate": 6.671255910636066e-07, "loss": 2.628, "step": 345010 }, { "epoch": 0.0041472, "grad_norm": 0.8549197912216187, "learning_rate": 6.668364214460333e-07, "loss": 2.8629, "step": 345020 }, { "epoch": 0.0041728, "grad_norm": 0.9188092350959778, "learning_rate": 6.665473123515453e-07, "loss": 3.0615, "step": 345030 }, { "epoch": 0.0041984, "grad_norm": 0.9331684708595276, "learning_rate": 6.662582637820181e-07, "loss": 2.6849, "step": 345040 }, { "epoch": 0.004224, "grad_norm": 0.8656244277954102, "learning_rate": 6.659692757393266e-07, "loss": 2.9006, "step": 345050 }, { "epoch": 0.0042496, "grad_norm": 9.402009963989258, "learning_rate": 6.656803482253416e-07, "loss": 2.7173, "step": 345060 }, { "epoch": 0.0042752, "grad_norm": 0.8801894783973694, "learning_rate": 6.653914812419393e-07, "loss": 2.7209, "step": 345070 }, { "epoch": 0.0043008, "grad_norm": 0.7891750931739807, "learning_rate": 6.651026747909928e-07, "loss": 2.9005, "step": 345080 }, { "epoch": 0.0043264, "grad_norm": 0.7772735953330994, "learning_rate": 6.648139288743738e-07, "loss": 2.6426, "step": 345090 }, { "epoch": 0.004352, "grad_norm": 0.8566871285438538, "learning_rate": 6.645252434939564e-07, "loss": 2.9273, "step": 345100 }, { "epoch": 0.0043776, "grad_norm": 0.7659762501716614, "learning_rate": 6.642366186516103e-07, "loss": 2.8897, "step": 345110 }, { "epoch": 0.0044032, "grad_norm": 0.7875062823295593, "learning_rate": 6.639480543492116e-07, "loss": 2.7638, "step": 345120 }, { "epoch": 0.0044288, "grad_norm": 0.8075883388519287, "learning_rate": 6.636595505886245e-07, "loss": 2.8806, "step": 345130 }, { "epoch": 0.0044544, "grad_norm": 0.8871557116508484, "learning_rate": 6.633711073717264e-07, "loss": 2.7896, "step": 345140 }, { "epoch": 0.00448, "grad_norm": 1.1498273611068726, "learning_rate": 6.630827247003846e-07, "loss": 3.1256, "step": 345150 }, { "epoch": 0.0045056, "grad_norm": 1.0184458494186401, "learning_rate": 6.627944025764699e-07, "loss": 2.9753, "step": 345160 }, { "epoch": 0.0045312, "grad_norm": 0.7980347871780396, "learning_rate": 6.625061410018519e-07, "loss": 2.7611, "step": 345170 }, { "epoch": 0.0045568, "grad_norm": 0.8735318183898926, "learning_rate": 6.622179399784012e-07, "loss": 2.884, "step": 345180 }, { "epoch": 0.0045824, "grad_norm": 0.8316059112548828, "learning_rate": 6.619297995079844e-07, "loss": 2.8317, "step": 345190 }, { "epoch": 0.004608, "grad_norm": 1.231586217880249, "learning_rate": 6.616417195924707e-07, "loss": 3.0107, "step": 345200 }, { "epoch": 0.0046336, "grad_norm": 0.948025643825531, "learning_rate": 6.613537002337289e-07, "loss": 3.0197, "step": 345210 }, { "epoch": 0.0046592, "grad_norm": 0.9881672263145447, "learning_rate": 6.610657414336274e-07, "loss": 2.9412, "step": 345220 }, { "epoch": 0.0046848, "grad_norm": 0.8114980459213257, "learning_rate": 6.607778431940315e-07, "loss": 2.9259, "step": 345230 }, { "epoch": 0.0047104, "grad_norm": 0.8703979849815369, "learning_rate": 6.604900055168096e-07, "loss": 2.7056, "step": 345240 }, { "epoch": 0.004736, "grad_norm": 0.9484314322471619, "learning_rate": 6.60202228403829e-07, "loss": 2.9298, "step": 345250 }, { "epoch": 0.0047616, "grad_norm": 0.8410553336143494, "learning_rate": 6.599145118569528e-07, "loss": 2.7431, "step": 345260 }, { "epoch": 0.0047872, "grad_norm": 0.8489828109741211, "learning_rate": 6.596268558780494e-07, "loss": 2.7699, "step": 345270 }, { "epoch": 0.0048128, "grad_norm": 0.8489450812339783, "learning_rate": 6.59339260468983e-07, "loss": 2.88, "step": 345280 }, { "epoch": 0.0048384, "grad_norm": 0.9344878792762756, "learning_rate": 6.590517256316197e-07, "loss": 2.7725, "step": 345290 }, { "epoch": 0.004864, "grad_norm": 0.9783785939216614, "learning_rate": 6.587642513678216e-07, "loss": 2.798, "step": 345300 }, { "epoch": 0.0048896, "grad_norm": 0.7927162051200867, "learning_rate": 6.584768376794581e-07, "loss": 2.7566, "step": 345310 }, { "epoch": 0.0049152, "grad_norm": 0.8385183215141296, "learning_rate": 6.581894845683878e-07, "loss": 2.6513, "step": 345320 }, { "epoch": 0.0049408, "grad_norm": 0.7775233387947083, "learning_rate": 6.579021920364759e-07, "loss": 2.7107, "step": 345330 }, { "epoch": 0.0049664, "grad_norm": 0.7436272501945496, "learning_rate": 6.576149600855852e-07, "loss": 2.8891, "step": 345340 }, { "epoch": 0.004992, "grad_norm": 0.8203125596046448, "learning_rate": 6.573277887175788e-07, "loss": 2.777, "step": 345350 }, { "epoch": 0.0050176, "grad_norm": 0.8508734703063965, "learning_rate": 6.570406779343186e-07, "loss": 2.7481, "step": 345360 }, { "epoch": 0.0050432, "grad_norm": 0.8645015358924866, "learning_rate": 6.567536277376674e-07, "loss": 2.5489, "step": 345370 }, { "epoch": 0.0050688, "grad_norm": 0.808051347732544, "learning_rate": 6.56466638129486e-07, "loss": 2.7358, "step": 345380 }, { "epoch": 0.0050944, "grad_norm": 0.9077112078666687, "learning_rate": 6.56179709111634e-07, "loss": 2.7271, "step": 345390 }, { "epoch": 0.00512, "grad_norm": 0.9388511180877686, "learning_rate": 6.558928406859744e-07, "loss": 2.7925, "step": 345400 }, { "epoch": 0.0051456, "grad_norm": 0.8423381447792053, "learning_rate": 6.556060328543656e-07, "loss": 2.7999, "step": 345410 }, { "epoch": 0.0051712, "grad_norm": 0.8265862464904785, "learning_rate": 6.553192856186674e-07, "loss": 2.8089, "step": 345420 }, { "epoch": 0.0051968, "grad_norm": 0.9068069458007812, "learning_rate": 6.550325989807416e-07, "loss": 2.8695, "step": 345430 }, { "epoch": 0.0052224, "grad_norm": 0.8458771705627441, "learning_rate": 6.547459729424444e-07, "loss": 2.6136, "step": 345440 }, { "epoch": 0.005248, "grad_norm": 0.8298386335372925, "learning_rate": 6.544594075056387e-07, "loss": 2.803, "step": 345450 }, { "epoch": 0.0052736, "grad_norm": 0.7373030781745911, "learning_rate": 6.541729026721777e-07, "loss": 2.7769, "step": 345460 }, { "epoch": 0.0052992, "grad_norm": 0.8821479082107544, "learning_rate": 6.538864584439219e-07, "loss": 2.6803, "step": 345470 }, { "epoch": 0.0053248, "grad_norm": 0.8787548542022705, "learning_rate": 6.536000748227267e-07, "loss": 2.8367, "step": 345480 }, { "epoch": 0.0053504, "grad_norm": 0.8403658866882324, "learning_rate": 6.533137518104526e-07, "loss": 2.6309, "step": 345490 }, { "epoch": 0.005376, "grad_norm": 0.8424771428108215, "learning_rate": 6.53027489408955e-07, "loss": 2.8788, "step": 345500 }, { "epoch": 0.0054016, "grad_norm": 0.7829804420471191, "learning_rate": 6.527412876200922e-07, "loss": 2.6402, "step": 345510 }, { "epoch": 0.0054272, "grad_norm": 0.8982354402542114, "learning_rate": 6.524551464457151e-07, "loss": 2.6314, "step": 345520 }, { "epoch": 0.0054528, "grad_norm": 0.8610154390335083, "learning_rate": 6.521690658876823e-07, "loss": 2.7971, "step": 345530 }, { "epoch": 0.0054784, "grad_norm": 0.8534964919090271, "learning_rate": 6.518830459478497e-07, "loss": 2.5741, "step": 345540 }, { "epoch": 0.005504, "grad_norm": 0.8247687220573425, "learning_rate": 6.515970866280708e-07, "loss": 2.8513, "step": 345550 }, { "epoch": 0.0055296, "grad_norm": 0.8274833559989929, "learning_rate": 6.513111879301992e-07, "loss": 2.6926, "step": 345560 }, { "epoch": 0.0055552, "grad_norm": 0.8158808946609497, "learning_rate": 6.510253498560915e-07, "loss": 2.8635, "step": 345570 }, { "epoch": 0.0055808, "grad_norm": 0.778452455997467, "learning_rate": 6.507395724076005e-07, "loss": 2.7824, "step": 345580 }, { "epoch": 0.0056064, "grad_norm": 0.851382315158844, "learning_rate": 6.504538555865769e-07, "loss": 2.6977, "step": 345590 }, { "epoch": 0.005632, "grad_norm": 0.8826770186424255, "learning_rate": 6.50168199394876e-07, "loss": 2.6605, "step": 345600 }, { "epoch": 0.0056576, "grad_norm": 0.7587625980377197, "learning_rate": 6.498826038343486e-07, "loss": 2.7184, "step": 345610 }, { "epoch": 0.0056832, "grad_norm": 0.8580876588821411, "learning_rate": 6.495970689068476e-07, "loss": 2.8045, "step": 345620 }, { "epoch": 0.0057088, "grad_norm": 0.963060736656189, "learning_rate": 6.493115946142248e-07, "loss": 2.8665, "step": 345630 }, { "epoch": 0.0057344, "grad_norm": 0.9002411961555481, "learning_rate": 6.490261809583321e-07, "loss": 2.6096, "step": 345640 }, { "epoch": 0.00576, "grad_norm": 0.8758597373962402, "learning_rate": 6.487408279410179e-07, "loss": 2.755, "step": 345650 }, { "epoch": 0.0057856, "grad_norm": 0.8002524375915527, "learning_rate": 6.484555355641354e-07, "loss": 2.8181, "step": 345660 }, { "epoch": 0.0058112, "grad_norm": 0.813055157661438, "learning_rate": 6.481703038295329e-07, "loss": 2.8283, "step": 345670 }, { "epoch": 0.0058368, "grad_norm": 0.8194794058799744, "learning_rate": 6.478851327390601e-07, "loss": 2.7556, "step": 345680 }, { "epoch": 0.0058624, "grad_norm": 0.7964984774589539, "learning_rate": 6.476000222945678e-07, "loss": 2.6728, "step": 345690 }, { "epoch": 0.005888, "grad_norm": 1.4426591396331787, "learning_rate": 6.473149724979022e-07, "loss": 2.8066, "step": 345700 }, { "epoch": 0.0059136, "grad_norm": 0.8771694302558899, "learning_rate": 6.470299833509164e-07, "loss": 2.6281, "step": 345710 }, { "epoch": 0.0059392, "grad_norm": 0.8417352437973022, "learning_rate": 6.467450548554532e-07, "loss": 2.644, "step": 345720 }, { "epoch": 0.0059648, "grad_norm": 0.7943432927131653, "learning_rate": 6.464601870133624e-07, "loss": 2.6512, "step": 345730 }, { "epoch": 0.0059904, "grad_norm": 0.8128944039344788, "learning_rate": 6.461753798264913e-07, "loss": 2.8531, "step": 345740 }, { "epoch": 0.006016, "grad_norm": 0.9408989548683167, "learning_rate": 6.458906332966874e-07, "loss": 2.6738, "step": 345750 }, { "epoch": 0.0060416, "grad_norm": 0.9086571931838989, "learning_rate": 6.45605947425797e-07, "loss": 2.6592, "step": 345760 }, { "epoch": 0.0060672, "grad_norm": 0.9241592288017273, "learning_rate": 6.453213222156651e-07, "loss": 2.6477, "step": 345770 }, { "epoch": 0.0060928, "grad_norm": 0.8936825394630432, "learning_rate": 6.450367576681404e-07, "loss": 2.7869, "step": 345780 }, { "epoch": 0.0061184, "grad_norm": 0.8686597347259521, "learning_rate": 6.447522537850648e-07, "loss": 2.6556, "step": 345790 }, { "epoch": 0.006144, "grad_norm": 0.8986815810203552, "learning_rate": 6.444678105682833e-07, "loss": 2.801, "step": 345800 }, { "epoch": 0.0061696, "grad_norm": 0.8664180040359497, "learning_rate": 6.441834280196424e-07, "loss": 2.7507, "step": 345810 }, { "epoch": 0.0061952, "grad_norm": 0.897976815700531, "learning_rate": 6.43899106140985e-07, "loss": 2.8513, "step": 345820 }, { "epoch": 0.0062208, "grad_norm": 0.8189851641654968, "learning_rate": 6.436148449341573e-07, "loss": 2.8676, "step": 345830 }, { "epoch": 0.0062464, "grad_norm": 1.015104055404663, "learning_rate": 6.43330644400999e-07, "loss": 2.738, "step": 345840 }, { "epoch": 0.006272, "grad_norm": 0.8077702522277832, "learning_rate": 6.430465045433542e-07, "loss": 2.6506, "step": 345850 }, { "epoch": 0.0062976, "grad_norm": 0.8364429473876953, "learning_rate": 6.427624253630671e-07, "loss": 2.6827, "step": 345860 }, { "epoch": 0.0063232, "grad_norm": 0.8840748071670532, "learning_rate": 6.424784068619794e-07, "loss": 2.7205, "step": 345870 }, { "epoch": 0.0063488, "grad_norm": 0.9748128652572632, "learning_rate": 6.421944490419307e-07, "loss": 2.7165, "step": 345880 }, { "epoch": 0.0063744, "grad_norm": 0.807214617729187, "learning_rate": 6.419105519047652e-07, "loss": 2.7777, "step": 345890 }, { "epoch": 0.0064, "grad_norm": 1.0571563243865967, "learning_rate": 6.416267154523237e-07, "loss": 2.7697, "step": 345900 }, { "epoch": 0.0064256, "grad_norm": 0.8009405136108398, "learning_rate": 6.413429396864445e-07, "loss": 2.6859, "step": 345910 }, { "epoch": 0.0064512, "grad_norm": 0.8386836647987366, "learning_rate": 6.410592246089686e-07, "loss": 2.663, "step": 345920 }, { "epoch": 0.0064768, "grad_norm": 0.8775103688240051, "learning_rate": 6.407755702217367e-07, "loss": 2.8285, "step": 345930 }, { "epoch": 0.0065024, "grad_norm": 0.8232018351554871, "learning_rate": 6.404919765265883e-07, "loss": 2.5943, "step": 345940 }, { "epoch": 0.006528, "grad_norm": 0.8678941130638123, "learning_rate": 6.402084435253608e-07, "loss": 2.7356, "step": 345950 }, { "epoch": 0.0065536, "grad_norm": 0.7691233158111572, "learning_rate": 6.399249712198952e-07, "loss": 2.702, "step": 345960 }, { "epoch": 0.0065792, "grad_norm": 0.9522619247436523, "learning_rate": 6.396415596120275e-07, "loss": 2.7808, "step": 345970 }, { "epoch": 0.0066048, "grad_norm": 0.8401975631713867, "learning_rate": 6.393582087035988e-07, "loss": 2.818, "step": 345980 }, { "epoch": 0.0066304, "grad_norm": 0.9259312152862549, "learning_rate": 6.390749184964429e-07, "loss": 2.9462, "step": 345990 }, { "epoch": 0.006656, "grad_norm": 0.8922718167304993, "learning_rate": 6.387916889923961e-07, "loss": 2.7371, "step": 346000 }, { "epoch": 0.0066816, "grad_norm": 0.9259646534919739, "learning_rate": 6.385085201932995e-07, "loss": 2.9515, "step": 346010 }, { "epoch": 0.0067072, "grad_norm": 0.8559521436691284, "learning_rate": 6.382254121009868e-07, "loss": 2.7916, "step": 346020 }, { "epoch": 0.0067328, "grad_norm": 0.8177400231361389, "learning_rate": 6.379423647172966e-07, "loss": 2.8998, "step": 346030 }, { "epoch": 0.0067584, "grad_norm": 0.8476433753967285, "learning_rate": 6.376593780440599e-07, "loss": 2.7906, "step": 346040 }, { "epoch": 0.006784, "grad_norm": 0.8985262513160706, "learning_rate": 6.373764520831138e-07, "loss": 2.7714, "step": 346050 }, { "epoch": 0.0068096, "grad_norm": 0.8257694840431213, "learning_rate": 6.370935868362926e-07, "loss": 2.7484, "step": 346060 }, { "epoch": 0.0068352, "grad_norm": 0.8313912153244019, "learning_rate": 6.368107823054314e-07, "loss": 2.8414, "step": 346070 }, { "epoch": 0.0068608, "grad_norm": 0.8291869759559631, "learning_rate": 6.365280384923644e-07, "loss": 2.847, "step": 346080 }, { "epoch": 0.0068864, "grad_norm": 1.0868092775344849, "learning_rate": 6.362453553989245e-07, "loss": 2.7595, "step": 346090 }, { "epoch": 0.006912, "grad_norm": 0.8013933300971985, "learning_rate": 6.359627330269458e-07, "loss": 2.6105, "step": 346100 }, { "epoch": 0.0069376, "grad_norm": 0.8166119456291199, "learning_rate": 6.356801713782589e-07, "loss": 2.7618, "step": 346110 }, { "epoch": 0.0069632, "grad_norm": 0.8998928666114807, "learning_rate": 6.353976704546982e-07, "loss": 2.8479, "step": 346120 }, { "epoch": 0.0069888, "grad_norm": 0.8599623441696167, "learning_rate": 6.351152302580932e-07, "loss": 2.8921, "step": 346130 }, { "epoch": 0.0070144, "grad_norm": 0.8723490834236145, "learning_rate": 6.348328507902779e-07, "loss": 2.7726, "step": 346140 }, { "epoch": 0.00704, "grad_norm": 1.1684719324111938, "learning_rate": 6.345505320530832e-07, "loss": 2.8147, "step": 346150 }, { "epoch": 0.0070656, "grad_norm": 0.8056842088699341, "learning_rate": 6.342682740483386e-07, "loss": 2.7843, "step": 346160 }, { "epoch": 0.0070912, "grad_norm": 0.761486291885376, "learning_rate": 6.339860767778749e-07, "loss": 2.7908, "step": 346170 }, { "epoch": 0.0071168, "grad_norm": 0.8052988648414612, "learning_rate": 6.33703940243523e-07, "loss": 2.7553, "step": 346180 }, { "epoch": 0.0071424, "grad_norm": 0.9216548800468445, "learning_rate": 6.334218644471113e-07, "loss": 2.6794, "step": 346190 }, { "epoch": 0.007168, "grad_norm": 0.8258796334266663, "learning_rate": 6.331398493904694e-07, "loss": 3.0213, "step": 346200 }, { "epoch": 0.0071936, "grad_norm": 0.9880358576774597, "learning_rate": 6.32857895075426e-07, "loss": 2.9957, "step": 346210 }, { "epoch": 0.0072192, "grad_norm": 0.8577410578727722, "learning_rate": 6.325760015038095e-07, "loss": 2.7841, "step": 346220 }, { "epoch": 0.0072448, "grad_norm": 0.9627076387405396, "learning_rate": 6.322941686774509e-07, "loss": 2.793, "step": 346230 }, { "epoch": 0.0072704, "grad_norm": 1.0369902849197388, "learning_rate": 6.320123965981728e-07, "loss": 3.1477, "step": 346240 }, { "epoch": 0.007296, "grad_norm": 0.7918531894683838, "learning_rate": 6.31730685267804e-07, "loss": 2.8057, "step": 346250 }, { "epoch": 0.0073216, "grad_norm": 0.8977828621864319, "learning_rate": 6.314490346881719e-07, "loss": 2.905, "step": 346260 }, { "epoch": 0.0073472, "grad_norm": 0.7683430910110474, "learning_rate": 6.311674448611038e-07, "loss": 2.657, "step": 346270 }, { "epoch": 0.0073728, "grad_norm": 0.8722265958786011, "learning_rate": 6.30885915788424e-07, "loss": 2.682, "step": 346280 }, { "epoch": 0.0073984, "grad_norm": 0.9166572690010071, "learning_rate": 6.306044474719598e-07, "loss": 2.9001, "step": 346290 }, { "epoch": 0.007424, "grad_norm": 0.9768415093421936, "learning_rate": 6.303230399135363e-07, "loss": 2.7054, "step": 346300 }, { "epoch": 0.0074496, "grad_norm": 0.795612096786499, "learning_rate": 6.300416931149778e-07, "loss": 2.7975, "step": 346310 }, { "epoch": 0.0074752, "grad_norm": 0.8680872321128845, "learning_rate": 6.297604070781072e-07, "loss": 2.8331, "step": 346320 }, { "epoch": 0.0075008, "grad_norm": 0.8418217301368713, "learning_rate": 6.294791818047508e-07, "loss": 2.7813, "step": 346330 }, { "epoch": 0.0075264, "grad_norm": 0.8504241108894348, "learning_rate": 6.291980172967305e-07, "loss": 2.7336, "step": 346340 }, { "epoch": 0.007552, "grad_norm": 1.0434216260910034, "learning_rate": 6.289169135558703e-07, "loss": 2.885, "step": 346350 }, { "epoch": 0.0075776, "grad_norm": 0.8046556711196899, "learning_rate": 6.286358705839957e-07, "loss": 2.8258, "step": 346360 }, { "epoch": 0.0076032, "grad_norm": 0.8195915222167969, "learning_rate": 6.283548883829261e-07, "loss": 2.8468, "step": 346370 }, { "epoch": 0.0076288, "grad_norm": 0.8276851773262024, "learning_rate": 6.280739669544833e-07, "loss": 2.7331, "step": 346380 }, { "epoch": 0.0076544, "grad_norm": 0.7958106994628906, "learning_rate": 6.277931063004905e-07, "loss": 2.8276, "step": 346390 }, { "epoch": 0.00768, "grad_norm": 0.853212296962738, "learning_rate": 6.275123064227684e-07, "loss": 2.9698, "step": 346400 }, { "epoch": 0.0077056, "grad_norm": 0.8457121253013611, "learning_rate": 6.272315673231388e-07, "loss": 2.759, "step": 346410 }, { "epoch": 0.0077312, "grad_norm": 0.8705182671546936, "learning_rate": 6.269508890034204e-07, "loss": 2.7984, "step": 346420 }, { "epoch": 0.0077568, "grad_norm": 0.8063643574714661, "learning_rate": 6.266702714654371e-07, "loss": 2.6721, "step": 346430 }, { "epoch": 0.0077824, "grad_norm": 0.8447285890579224, "learning_rate": 6.263897147110043e-07, "loss": 2.7677, "step": 346440 }, { "epoch": 0.007808, "grad_norm": 0.9157115817070007, "learning_rate": 6.261092187419426e-07, "loss": 2.5465, "step": 346450 }, { "epoch": 0.0078336, "grad_norm": 0.9719769358634949, "learning_rate": 6.258287835600707e-07, "loss": 2.8994, "step": 346460 }, { "epoch": 0.0078592, "grad_norm": 0.8687463998794556, "learning_rate": 6.255484091672082e-07, "loss": 2.7579, "step": 346470 }, { "epoch": 0.0078848, "grad_norm": 0.7992487549781799, "learning_rate": 6.252680955651736e-07, "loss": 2.7829, "step": 346480 }, { "epoch": 0.0079104, "grad_norm": 0.8141602873802185, "learning_rate": 6.249878427557831e-07, "loss": 2.8477, "step": 346490 }, { "epoch": 0.007936, "grad_norm": 0.9432787299156189, "learning_rate": 6.247076507408567e-07, "loss": 2.7609, "step": 346500 }, { "epoch": 0.0079616, "grad_norm": 0.858770489692688, "learning_rate": 6.244275195222083e-07, "loss": 2.7275, "step": 346510 }, { "epoch": 0.0079872, "grad_norm": 0.8154556751251221, "learning_rate": 6.241474491016542e-07, "loss": 2.7169, "step": 346520 }, { "epoch": 0.0080128, "grad_norm": 0.8997745513916016, "learning_rate": 6.238674394810129e-07, "loss": 2.5913, "step": 346530 }, { "epoch": 0.0080384, "grad_norm": 0.8794372081756592, "learning_rate": 6.235874906620998e-07, "loss": 2.8345, "step": 346540 }, { "epoch": 0.008064, "grad_norm": 0.8184460997581482, "learning_rate": 6.2330760264673e-07, "loss": 2.6619, "step": 346550 }, { "epoch": 0.0080896, "grad_norm": 0.851107120513916, "learning_rate": 6.230277754367198e-07, "loss": 2.6212, "step": 346560 }, { "epoch": 0.0081152, "grad_norm": 0.7990986108779907, "learning_rate": 6.2274800903388e-07, "loss": 2.7331, "step": 346570 }, { "epoch": 0.0081408, "grad_norm": 0.9600942730903625, "learning_rate": 6.224683034400281e-07, "loss": 2.7925, "step": 346580 }, { "epoch": 0.0081664, "grad_norm": 0.828744113445282, "learning_rate": 6.22188658656977e-07, "loss": 2.76, "step": 346590 }, { "epoch": 0.008192, "grad_norm": 0.8602617979049683, "learning_rate": 6.219090746865397e-07, "loss": 2.7878, "step": 346600 }, { "epoch": 0.0082176, "grad_norm": 0.814953088760376, "learning_rate": 6.216295515305304e-07, "loss": 2.7959, "step": 346610 }, { "epoch": 0.0082432, "grad_norm": 0.8258646726608276, "learning_rate": 6.213500891907598e-07, "loss": 2.7779, "step": 346620 }, { "epoch": 0.0082688, "grad_norm": 0.7490041255950928, "learning_rate": 6.210706876690443e-07, "loss": 2.837, "step": 346630 }, { "epoch": 0.0082944, "grad_norm": 0.9530748724937439, "learning_rate": 6.2079134696719e-07, "loss": 2.6084, "step": 346640 }, { "epoch": 0.00832, "grad_norm": 0.8035206198692322, "learning_rate": 6.205120670870113e-07, "loss": 2.8304, "step": 346650 }, { "epoch": 0.0083456, "grad_norm": 0.8997234106063843, "learning_rate": 6.202328480303199e-07, "loss": 2.6476, "step": 346660 }, { "epoch": 0.0083712, "grad_norm": 0.8040429949760437, "learning_rate": 6.199536897989255e-07, "loss": 2.6903, "step": 346670 }, { "epoch": 0.0083968, "grad_norm": 0.8173391819000244, "learning_rate": 6.196745923946379e-07, "loss": 2.8621, "step": 346680 }, { "epoch": 0.0084224, "grad_norm": 0.8177932500839233, "learning_rate": 6.193955558192677e-07, "loss": 2.9652, "step": 346690 }, { "epoch": 0.008448, "grad_norm": 0.9432900547981262, "learning_rate": 6.191165800746246e-07, "loss": 2.8168, "step": 346700 }, { "epoch": 0.0084736, "grad_norm": 0.9084033966064453, "learning_rate": 6.188376651625172e-07, "loss": 2.8512, "step": 346710 }, { "epoch": 0.0084992, "grad_norm": 0.931660532951355, "learning_rate": 6.18558811084754e-07, "loss": 2.8617, "step": 346720 }, { "epoch": 0.0085248, "grad_norm": 1.1641120910644531, "learning_rate": 6.182800178431436e-07, "loss": 2.8355, "step": 346730 }, { "epoch": 0.0085504, "grad_norm": 0.8661662340164185, "learning_rate": 6.180012854394945e-07, "loss": 2.8882, "step": 346740 }, { "epoch": 0.008576, "grad_norm": 0.897352933883667, "learning_rate": 6.177226138756131e-07, "loss": 2.9133, "step": 346750 }, { "epoch": 0.0086016, "grad_norm": 0.8056995272636414, "learning_rate": 6.174440031533091e-07, "loss": 2.6685, "step": 346760 }, { "epoch": 0.0086272, "grad_norm": 0.8364745378494263, "learning_rate": 6.171654532743853e-07, "loss": 2.8261, "step": 346770 }, { "epoch": 0.0086528, "grad_norm": 0.7916631698608398, "learning_rate": 6.168869642406494e-07, "loss": 2.7553, "step": 346780 }, { "epoch": 0.0086784, "grad_norm": 0.7852660417556763, "learning_rate": 6.166085360539087e-07, "loss": 2.8221, "step": 346790 }, { "epoch": 0.008704, "grad_norm": 0.9132006168365479, "learning_rate": 6.163301687159673e-07, "loss": 2.7409, "step": 346800 }, { "epoch": 0.0087296, "grad_norm": 0.8636777400970459, "learning_rate": 6.160518622286305e-07, "loss": 2.7077, "step": 346810 }, { "epoch": 0.0087552, "grad_norm": 0.8147357702255249, "learning_rate": 6.157736165937045e-07, "loss": 2.8945, "step": 346820 }, { "epoch": 0.0087808, "grad_norm": 0.9062533974647522, "learning_rate": 6.154954318129936e-07, "loss": 2.7802, "step": 346830 }, { "epoch": 0.0088064, "grad_norm": 0.8185858726501465, "learning_rate": 6.152173078882983e-07, "loss": 2.783, "step": 346840 }, { "epoch": 0.008832, "grad_norm": 0.839468240737915, "learning_rate": 6.149392448214253e-07, "loss": 2.868, "step": 346850 }, { "epoch": 0.0088576, "grad_norm": 0.8446794748306274, "learning_rate": 6.146612426141774e-07, "loss": 2.7875, "step": 346860 }, { "epoch": 0.0088832, "grad_norm": 0.7650529146194458, "learning_rate": 6.143833012683554e-07, "loss": 2.8705, "step": 346870 }, { "epoch": 0.0089088, "grad_norm": 1.1741690635681152, "learning_rate": 6.141054207857645e-07, "loss": 2.8181, "step": 346880 }, { "epoch": 0.0089344, "grad_norm": 0.8499672412872314, "learning_rate": 6.138276011682076e-07, "loss": 2.8044, "step": 346890 }, { "epoch": 0.00896, "grad_norm": 0.7969455122947693, "learning_rate": 6.135498424174824e-07, "loss": 2.7259, "step": 346900 }, { "epoch": 0.0089856, "grad_norm": 0.7909001111984253, "learning_rate": 6.132721445353907e-07, "loss": 2.7396, "step": 346910 }, { "epoch": 0.0090112, "grad_norm": 0.8776979446411133, "learning_rate": 6.129945075237354e-07, "loss": 2.5481, "step": 346920 }, { "epoch": 0.0090368, "grad_norm": 0.8842483162879944, "learning_rate": 6.127169313843162e-07, "loss": 2.9073, "step": 346930 }, { "epoch": 0.0090624, "grad_norm": 0.829587459564209, "learning_rate": 6.124394161189328e-07, "loss": 2.7195, "step": 346940 }, { "epoch": 0.009088, "grad_norm": 0.8105077743530273, "learning_rate": 6.121619617293851e-07, "loss": 2.7517, "step": 346950 }, { "epoch": 0.0091136, "grad_norm": 0.9050664901733398, "learning_rate": 6.118845682174724e-07, "loss": 2.8708, "step": 346960 }, { "epoch": 0.0091392, "grad_norm": 0.8498857617378235, "learning_rate": 6.116072355849923e-07, "loss": 2.8908, "step": 346970 }, { "epoch": 0.0091648, "grad_norm": 0.8496431708335876, "learning_rate": 6.113299638337444e-07, "loss": 2.8111, "step": 346980 }, { "epoch": 0.0091904, "grad_norm": 0.8236566185951233, "learning_rate": 6.110527529655264e-07, "loss": 2.6333, "step": 346990 }, { "epoch": 0.009216, "grad_norm": 1.0231423377990723, "learning_rate": 6.107756029821366e-07, "loss": 2.9169, "step": 347000 }, { "epoch": 0.0092416, "grad_norm": 0.8433867692947388, "learning_rate": 6.104985138853703e-07, "loss": 2.9041, "step": 347010 }, { "epoch": 0.0092672, "grad_norm": 0.9368089437484741, "learning_rate": 6.102214856770272e-07, "loss": 2.7843, "step": 347020 }, { "epoch": 0.0092928, "grad_norm": 0.846821129322052, "learning_rate": 6.099445183589026e-07, "loss": 2.6233, "step": 347030 }, { "epoch": 0.0093184, "grad_norm": 0.8069620132446289, "learning_rate": 6.096676119327893e-07, "loss": 2.9057, "step": 347040 }, { "epoch": 0.009344, "grad_norm": 0.8091047406196594, "learning_rate": 6.093907664004883e-07, "loss": 2.6927, "step": 347050 }, { "epoch": 0.0093696, "grad_norm": 0.9263724684715271, "learning_rate": 6.091139817637914e-07, "loss": 2.9643, "step": 347060 }, { "epoch": 0.0093952, "grad_norm": 0.806711733341217, "learning_rate": 6.08837258024495e-07, "loss": 2.8313, "step": 347070 }, { "epoch": 0.0094208, "grad_norm": 0.8983010649681091, "learning_rate": 6.085605951843931e-07, "loss": 2.9508, "step": 347080 }, { "epoch": 0.0094464, "grad_norm": 0.7924553155899048, "learning_rate": 6.082839932452822e-07, "loss": 2.7175, "step": 347090 }, { "epoch": 0.009472, "grad_norm": 0.7673080563545227, "learning_rate": 6.080074522089507e-07, "loss": 2.6852, "step": 347100 }, { "epoch": 0.0094976, "grad_norm": 0.8773941993713379, "learning_rate": 6.077309720771951e-07, "loss": 2.7754, "step": 347110 }, { "epoch": 0.0095232, "grad_norm": 0.9790900349617004, "learning_rate": 6.074545528518083e-07, "loss": 2.9572, "step": 347120 }, { "epoch": 0.0095488, "grad_norm": 0.8460507988929749, "learning_rate": 6.071781945345823e-07, "loss": 2.9818, "step": 347130 }, { "epoch": 0.0095744, "grad_norm": 0.7806259989738464, "learning_rate": 6.0690189712731e-07, "loss": 2.8607, "step": 347140 }, { "epoch": 0.0096, "grad_norm": 0.8594427108764648, "learning_rate": 6.066256606317811e-07, "loss": 2.9375, "step": 347150 }, { "epoch": 0.0096256, "grad_norm": 0.8624967932701111, "learning_rate": 6.063494850497909e-07, "loss": 2.9003, "step": 347160 }, { "epoch": 0.0096512, "grad_norm": 0.7882050275802612, "learning_rate": 6.060733703831267e-07, "loss": 2.9545, "step": 347170 }, { "epoch": 0.0096768, "grad_norm": 0.9057795405387878, "learning_rate": 6.057973166335784e-07, "loss": 2.7769, "step": 347180 }, { "epoch": 0.0097024, "grad_norm": 0.8303849101066589, "learning_rate": 6.05521323802939e-07, "loss": 2.8153, "step": 347190 }, { "epoch": 0.009728, "grad_norm": 0.9586015939712524, "learning_rate": 6.052453918929968e-07, "loss": 2.6539, "step": 347200 }, { "epoch": 0.0097536, "grad_norm": 1.0590497255325317, "learning_rate": 6.049695209055418e-07, "loss": 2.7438, "step": 347210 }, { "epoch": 0.0097792, "grad_norm": 0.815778911113739, "learning_rate": 6.046937108423623e-07, "loss": 2.652, "step": 347220 }, { "epoch": 0.0098048, "grad_norm": 1.0162404775619507, "learning_rate": 6.044179617052471e-07, "loss": 2.79, "step": 347230 }, { "epoch": 0.0098304, "grad_norm": 0.9605357646942139, "learning_rate": 6.041422734959846e-07, "loss": 2.811, "step": 347240 }, { "epoch": 0.009856, "grad_norm": 0.8369733095169067, "learning_rate": 6.038666462163623e-07, "loss": 2.6295, "step": 347250 }, { "epoch": 0.0098816, "grad_norm": 0.7776085734367371, "learning_rate": 6.035910798681677e-07, "loss": 3.1499, "step": 347260 }, { "epoch": 0.0099072, "grad_norm": 0.8307235836982727, "learning_rate": 6.033155744531872e-07, "loss": 2.757, "step": 347270 }, { "epoch": 0.0099328, "grad_norm": 0.7901933193206787, "learning_rate": 6.03040129973208e-07, "loss": 2.7886, "step": 347280 }, { "epoch": 0.0099584, "grad_norm": 0.8120728731155396, "learning_rate": 6.027647464300168e-07, "loss": 2.6735, "step": 347290 }, { "epoch": 0.009984, "grad_norm": 0.901077389717102, "learning_rate": 6.024894238253987e-07, "loss": 2.7019, "step": 347300 }, { "epoch": 0.0100096, "grad_norm": 1.2270894050598145, "learning_rate": 6.022141621611389e-07, "loss": 3.1155, "step": 347310 }, { "epoch": 0.0100352, "grad_norm": 0.8226301670074463, "learning_rate": 6.019389614390214e-07, "loss": 2.6843, "step": 347320 }, { "epoch": 0.0100608, "grad_norm": 1.031994104385376, "learning_rate": 6.01663821660834e-07, "loss": 2.8591, "step": 347330 }, { "epoch": 0.0100864, "grad_norm": 0.8120096921920776, "learning_rate": 6.013887428283571e-07, "loss": 2.826, "step": 347340 }, { "epoch": 0.010112, "grad_norm": 0.774893581867218, "learning_rate": 6.011137249433774e-07, "loss": 2.8122, "step": 347350 }, { "epoch": 0.0101376, "grad_norm": 0.8695253133773804, "learning_rate": 6.008387680076788e-07, "loss": 2.9532, "step": 347360 }, { "epoch": 0.0101632, "grad_norm": 0.8683069348335266, "learning_rate": 6.0056387202304e-07, "loss": 2.826, "step": 347370 }, { "epoch": 0.0101888, "grad_norm": 0.9018341302871704, "learning_rate": 6.002890369912473e-07, "loss": 3.2407, "step": 347380 }, { "epoch": 0.0102144, "grad_norm": 0.9961867332458496, "learning_rate": 6.000142629140815e-07, "loss": 3.0, "step": 347390 }, { "epoch": 0.01024, "grad_norm": 1.1154805421829224, "learning_rate": 5.997395497933256e-07, "loss": 2.86, "step": 347400 }, { "epoch": 0.0102656, "grad_norm": 0.9275990724563599, "learning_rate": 5.994648976307616e-07, "loss": 2.8237, "step": 347410 }, { "epoch": 0.0102912, "grad_norm": 0.8704720735549927, "learning_rate": 5.991903064281679e-07, "loss": 2.9091, "step": 347420 }, { "epoch": 0.0103168, "grad_norm": 0.8782855868339539, "learning_rate": 5.989157761873276e-07, "loss": 2.9019, "step": 347430 }, { "epoch": 0.0103424, "grad_norm": 0.9692162275314331, "learning_rate": 5.986413069100195e-07, "loss": 2.7794, "step": 347440 }, { "epoch": 0.010368, "grad_norm": 0.7564689517021179, "learning_rate": 5.983668985980229e-07, "loss": 2.8428, "step": 347450 }, { "epoch": 0.0103936, "grad_norm": 0.903533399105072, "learning_rate": 5.980925512531199e-07, "loss": 2.9347, "step": 347460 }, { "epoch": 0.0104192, "grad_norm": 0.8670058846473694, "learning_rate": 5.978182648770869e-07, "loss": 2.6278, "step": 347470 }, { "epoch": 0.0104448, "grad_norm": 0.8158991932868958, "learning_rate": 5.975440394717036e-07, "loss": 2.9192, "step": 347480 }, { "epoch": 0.0104704, "grad_norm": 0.8164783120155334, "learning_rate": 5.972698750387507e-07, "loss": 2.7662, "step": 347490 }, { "epoch": 0.010496, "grad_norm": 0.8464527130126953, "learning_rate": 5.969957715800012e-07, "loss": 2.5708, "step": 347500 }, { "epoch": 0.0105216, "grad_norm": 0.9945509433746338, "learning_rate": 5.96721729097236e-07, "loss": 2.8861, "step": 347510 }, { "epoch": 0.0105472, "grad_norm": 0.9153493046760559, "learning_rate": 5.964477475922315e-07, "loss": 2.8734, "step": 347520 }, { "epoch": 0.0105728, "grad_norm": 0.8060303926467896, "learning_rate": 5.961738270667627e-07, "loss": 2.8706, "step": 347530 }, { "epoch": 0.0105984, "grad_norm": 0.8087325096130371, "learning_rate": 5.958999675226085e-07, "loss": 2.8261, "step": 347540 }, { "epoch": 0.010624, "grad_norm": 0.9534764289855957, "learning_rate": 5.956261689615439e-07, "loss": 2.8851, "step": 347550 }, { "epoch": 0.0106496, "grad_norm": 0.8338521718978882, "learning_rate": 5.953524313853443e-07, "loss": 2.8896, "step": 347560 }, { "epoch": 0.0106752, "grad_norm": 0.8798498511314392, "learning_rate": 5.950787547957837e-07, "loss": 2.8867, "step": 347570 }, { "epoch": 0.0107008, "grad_norm": 1.0486035346984863, "learning_rate": 5.948051391946385e-07, "loss": 2.8083, "step": 347580 }, { "epoch": 0.0107264, "grad_norm": 0.9151228070259094, "learning_rate": 5.945315845836819e-07, "loss": 2.6855, "step": 347590 }, { "epoch": 0.010752, "grad_norm": 0.8942798376083374, "learning_rate": 5.942580909646889e-07, "loss": 2.831, "step": 347600 }, { "epoch": 0.0107776, "grad_norm": 0.8113926649093628, "learning_rate": 5.939846583394337e-07, "loss": 2.8036, "step": 347610 }, { "epoch": 0.0108032, "grad_norm": 0.7889653444290161, "learning_rate": 5.937112867096873e-07, "loss": 2.8644, "step": 347620 }, { "epoch": 0.0108288, "grad_norm": 0.8745374083518982, "learning_rate": 5.934379760772224e-07, "loss": 2.8171, "step": 347630 }, { "epoch": 0.0108544, "grad_norm": 0.9553562998771667, "learning_rate": 5.931647264438123e-07, "loss": 2.8486, "step": 347640 }, { "epoch": 0.01088, "grad_norm": 1.1296861171722412, "learning_rate": 5.928915378112299e-07, "loss": 2.8211, "step": 347650 }, { "epoch": 0.0109056, "grad_norm": 0.8092600703239441, "learning_rate": 5.926184101812449e-07, "loss": 2.6891, "step": 347660 }, { "epoch": 0.0109312, "grad_norm": 566.645751953125, "learning_rate": 5.923453435556292e-07, "loss": 4.6283, "step": 347670 }, { "epoch": 0.0109568, "grad_norm": 0.9713683128356934, "learning_rate": 5.920723379361537e-07, "loss": 2.8679, "step": 347680 }, { "epoch": 0.0109824, "grad_norm": 0.7793570756912231, "learning_rate": 5.917993933245902e-07, "loss": 2.6457, "step": 347690 }, { "epoch": 0.011008, "grad_norm": 0.8153229355812073, "learning_rate": 5.915265097227063e-07, "loss": 2.7527, "step": 347700 }, { "epoch": 0.0110336, "grad_norm": 0.744869589805603, "learning_rate": 5.912536871322728e-07, "loss": 2.5672, "step": 347710 }, { "epoch": 0.0110592, "grad_norm": 0.8662900924682617, "learning_rate": 5.90980925555058e-07, "loss": 2.992, "step": 347720 }, { "epoch": 0.0110848, "grad_norm": 1.0441043376922607, "learning_rate": 5.907082249928309e-07, "loss": 2.873, "step": 347730 }, { "epoch": 0.0111104, "grad_norm": 1.5051418542861938, "learning_rate": 5.904355854473609e-07, "loss": 2.7652, "step": 347740 }, { "epoch": 0.011136, "grad_norm": 1.0750001668930054, "learning_rate": 5.901630069204156e-07, "loss": 3.1308, "step": 347750 }, { "epoch": 0.0111616, "grad_norm": 0.8209498524665833, "learning_rate": 5.898904894137614e-07, "loss": 2.8029, "step": 347760 }, { "epoch": 0.0111872, "grad_norm": 0.8587725758552551, "learning_rate": 5.896180329291668e-07, "loss": 3.0198, "step": 347770 }, { "epoch": 0.0112128, "grad_norm": 0.993083655834198, "learning_rate": 5.893456374683992e-07, "loss": 2.8753, "step": 347780 }, { "epoch": 0.0112384, "grad_norm": 0.8075871467590332, "learning_rate": 5.89073303033223e-07, "loss": 2.895, "step": 347790 }, { "epoch": 0.011264, "grad_norm": 0.8241326808929443, "learning_rate": 5.888010296254054e-07, "loss": 2.8445, "step": 347800 }, { "epoch": 0.0112896, "grad_norm": 0.8089697360992432, "learning_rate": 5.885288172467141e-07, "loss": 2.8186, "step": 347810 }, { "epoch": 0.0113152, "grad_norm": 1.3327035903930664, "learning_rate": 5.882566658989108e-07, "loss": 2.8683, "step": 347820 }, { "epoch": 0.0113408, "grad_norm": 0.8614577054977417, "learning_rate": 5.87984575583761e-07, "loss": 2.8653, "step": 347830 }, { "epoch": 0.0113664, "grad_norm": 0.8493186831474304, "learning_rate": 5.877125463030297e-07, "loss": 2.5441, "step": 347840 }, { "epoch": 0.011392, "grad_norm": 0.883352518081665, "learning_rate": 5.874405780584813e-07, "loss": 2.5911, "step": 347850 }, { "epoch": 0.0114176, "grad_norm": 1.6417077779769897, "learning_rate": 5.871686708518786e-07, "loss": 3.5159, "step": 347860 }, { "epoch": 0.0114432, "grad_norm": 0.8199136257171631, "learning_rate": 5.868968246849871e-07, "loss": 2.6836, "step": 347870 }, { "epoch": 0.0114688, "grad_norm": 0.9257999658584595, "learning_rate": 5.866250395595674e-07, "loss": 2.8863, "step": 347880 }, { "epoch": 0.0114944, "grad_norm": 0.8163939714431763, "learning_rate": 5.863533154773826e-07, "loss": 2.8177, "step": 347890 }, { "epoch": 0.01152, "grad_norm": 0.7890012264251709, "learning_rate": 5.860816524401947e-07, "loss": 2.9491, "step": 347900 }, { "epoch": 0.0115456, "grad_norm": 1.0943909883499146, "learning_rate": 5.858100504497633e-07, "loss": 2.8168, "step": 347910 }, { "epoch": 0.0115712, "grad_norm": 0.8643054962158203, "learning_rate": 5.855385095078526e-07, "loss": 2.7765, "step": 347920 }, { "epoch": 0.0115968, "grad_norm": 0.8202794194221497, "learning_rate": 5.852670296162233e-07, "loss": 2.876, "step": 347930 }, { "epoch": 0.0116224, "grad_norm": 0.7624835968017578, "learning_rate": 5.849956107766363e-07, "loss": 2.7931, "step": 347940 }, { "epoch": 0.011648, "grad_norm": 0.7729390859603882, "learning_rate": 5.847242529908503e-07, "loss": 2.8074, "step": 347950 }, { "epoch": 0.0116736, "grad_norm": 0.9209871292114258, "learning_rate": 5.844529562606238e-07, "loss": 2.7254, "step": 347960 }, { "epoch": 0.0116992, "grad_norm": 1.0807865858078003, "learning_rate": 5.841817205877187e-07, "loss": 2.9227, "step": 347970 }, { "epoch": 0.0117248, "grad_norm": 0.8240377306938171, "learning_rate": 5.839105459738914e-07, "loss": 2.7803, "step": 347980 }, { "epoch": 0.0117504, "grad_norm": 1.3808327913284302, "learning_rate": 5.836394324209027e-07, "loss": 2.9069, "step": 347990 }, { "epoch": 0.011776, "grad_norm": 0.9038357734680176, "learning_rate": 5.83368379930509e-07, "loss": 2.8296, "step": 348000 }, { "epoch": 0.0118016, "grad_norm": 0.9368056058883667, "learning_rate": 5.83097388504471e-07, "loss": 2.8801, "step": 348010 }, { "epoch": 0.0118272, "grad_norm": 0.8626670241355896, "learning_rate": 5.828264581445419e-07, "loss": 2.7912, "step": 348020 }, { "epoch": 0.0118528, "grad_norm": 0.9194375872612, "learning_rate": 5.825555888524803e-07, "loss": 2.7, "step": 348030 }, { "epoch": 0.0118784, "grad_norm": 0.8638468980789185, "learning_rate": 5.822847806300425e-07, "loss": 2.8091, "step": 348040 }, { "epoch": 0.011904, "grad_norm": 0.854871928691864, "learning_rate": 5.820140334789859e-07, "loss": 2.5587, "step": 348050 }, { "epoch": 0.0119296, "grad_norm": 0.7843626737594604, "learning_rate": 5.817433474010648e-07, "loss": 2.5103, "step": 348060 }, { "epoch": 0.0119552, "grad_norm": 0.8714041113853455, "learning_rate": 5.814727223980354e-07, "loss": 2.768, "step": 348070 }, { "epoch": 0.0119808, "grad_norm": 0.7886574864387512, "learning_rate": 5.812021584716532e-07, "loss": 2.5822, "step": 348080 }, { "epoch": 0.0120064, "grad_norm": 0.8913326859474182, "learning_rate": 5.809316556236688e-07, "loss": 2.7209, "step": 348090 }, { "epoch": 0.012032, "grad_norm": 0.8365811705589294, "learning_rate": 5.806612138558409e-07, "loss": 2.5842, "step": 348100 }, { "epoch": 0.0120576, "grad_norm": 0.8487169742584229, "learning_rate": 5.803908331699216e-07, "loss": 2.8361, "step": 348110 }, { "epoch": 0.0120832, "grad_norm": 0.7860698103904724, "learning_rate": 5.801205135676646e-07, "loss": 2.9693, "step": 348120 }, { "epoch": 0.0121088, "grad_norm": 0.8126827478408813, "learning_rate": 5.798502550508223e-07, "loss": 2.7216, "step": 348130 }, { "epoch": 0.0121344, "grad_norm": 0.8748254776000977, "learning_rate": 5.795800576211497e-07, "loss": 2.8786, "step": 348140 }, { "epoch": 0.01216, "grad_norm": 0.7955288887023926, "learning_rate": 5.793099212803943e-07, "loss": 2.6036, "step": 348150 }, { "epoch": 0.0121856, "grad_norm": 0.8284873962402344, "learning_rate": 5.790398460303103e-07, "loss": 2.8998, "step": 348160 }, { "epoch": 0.0122112, "grad_norm": 0.7981054782867432, "learning_rate": 5.787698318726498e-07, "loss": 2.95, "step": 348170 }, { "epoch": 0.0122368, "grad_norm": 0.8821656107902527, "learning_rate": 5.784998788091634e-07, "loss": 2.9206, "step": 348180 }, { "epoch": 0.0122624, "grad_norm": 0.8797772526741028, "learning_rate": 5.782299868416008e-07, "loss": 2.8762, "step": 348190 }, { "epoch": 0.012288, "grad_norm": 0.8067883253097534, "learning_rate": 5.779601559717129e-07, "loss": 3.039, "step": 348200 }, { "epoch": 2.56e-05, "grad_norm": 0.7966413497924805, "learning_rate": 5.776903862012506e-07, "loss": 1.804, "step": 348210 }, { "epoch": 5.12e-05, "grad_norm": 0.7910172343254089, "learning_rate": 5.774206775319602e-07, "loss": 2.8377, "step": 348220 }, { "epoch": 7.68e-05, "grad_norm": 1.5899447202682495, "learning_rate": 5.771510299655924e-07, "loss": 3.0935, "step": 348230 }, { "epoch": 0.0001024, "grad_norm": 0.8024471402168274, "learning_rate": 5.76881443503896e-07, "loss": 2.5809, "step": 348240 }, { "epoch": 0.000128, "grad_norm": 0.8049434423446655, "learning_rate": 5.766119181486196e-07, "loss": 2.5649, "step": 348250 }, { "epoch": 0.0001536, "grad_norm": 0.8198651075363159, "learning_rate": 5.763424539015095e-07, "loss": 2.5452, "step": 348260 }, { "epoch": 0.0001792, "grad_norm": 0.8678796291351318, "learning_rate": 5.760730507643164e-07, "loss": 2.5814, "step": 348270 }, { "epoch": 0.0002048, "grad_norm": 0.7983577847480774, "learning_rate": 5.758037087387824e-07, "loss": 2.5987, "step": 348280 }, { "epoch": 0.0002304, "grad_norm": 0.894096314907074, "learning_rate": 5.755344278266584e-07, "loss": 2.5342, "step": 348290 }, { "epoch": 0.000256, "grad_norm": 0.8159319758415222, "learning_rate": 5.752652080296883e-07, "loss": 2.3709, "step": 348300 }, { "epoch": 0.0002816, "grad_norm": 0.9286146759986877, "learning_rate": 5.749960493496187e-07, "loss": 2.5795, "step": 348310 }, { "epoch": 0.0003072, "grad_norm": 0.7855417132377625, "learning_rate": 5.747269517881937e-07, "loss": 2.3836, "step": 348320 }, { "epoch": 0.0003328, "grad_norm": 0.8261249661445618, "learning_rate": 5.744579153471608e-07, "loss": 2.6994, "step": 348330 }, { "epoch": 0.0003584, "grad_norm": 0.9310190677642822, "learning_rate": 5.741889400282641e-07, "loss": 2.749, "step": 348340 }, { "epoch": 0.000384, "grad_norm": 0.8097965717315674, "learning_rate": 5.739200258332467e-07, "loss": 2.5927, "step": 348350 }, { "epoch": 0.0004096, "grad_norm": 0.7812308669090271, "learning_rate": 5.736511727638516e-07, "loss": 2.6236, "step": 348360 }, { "epoch": 0.0004352, "grad_norm": 0.7901282906532288, "learning_rate": 5.73382380821823e-07, "loss": 2.6489, "step": 348370 }, { "epoch": 0.0004608, "grad_norm": 0.9604964852333069, "learning_rate": 5.731136500089052e-07, "loss": 2.6365, "step": 348380 }, { "epoch": 0.0004864, "grad_norm": 0.8200964331626892, "learning_rate": 5.728449803268399e-07, "loss": 2.5109, "step": 348390 }, { "epoch": 0.000512, "grad_norm": 0.7176139950752258, "learning_rate": 5.725763717773703e-07, "loss": 2.3291, "step": 348400 }, { "epoch": 0.0005376, "grad_norm": 0.8439581990242004, "learning_rate": 5.723078243622382e-07, "loss": 2.5403, "step": 348410 }, { "epoch": 0.0005632, "grad_norm": 0.9644824862480164, "learning_rate": 5.720393380831823e-07, "loss": 2.5243, "step": 348420 }, { "epoch": 0.0005888, "grad_norm": 0.9096746444702148, "learning_rate": 5.717709129419446e-07, "loss": 2.5344, "step": 348430 }, { "epoch": 0.0006144, "grad_norm": 0.9785164594650269, "learning_rate": 5.715025489402682e-07, "loss": 2.5614, "step": 348440 }, { "epoch": 0.00064, "grad_norm": 0.8153619170188904, "learning_rate": 5.712342460798915e-07, "loss": 2.7473, "step": 348450 }, { "epoch": 0.0006656, "grad_norm": 0.8370224237442017, "learning_rate": 5.709660043625553e-07, "loss": 2.8225, "step": 348460 }, { "epoch": 0.0006912, "grad_norm": 0.7295762300491333, "learning_rate": 5.706978237900007e-07, "loss": 2.6204, "step": 348470 }, { "epoch": 0.0007168, "grad_norm": 0.9189844727516174, "learning_rate": 5.704297043639628e-07, "loss": 2.7335, "step": 348480 }, { "epoch": 0.0007424, "grad_norm": 0.8352077603340149, "learning_rate": 5.701616460861814e-07, "loss": 2.6637, "step": 348490 }, { "epoch": 0.000768, "grad_norm": 1.0030227899551392, "learning_rate": 5.698936489583973e-07, "loss": 2.6864, "step": 348500 }, { "epoch": 0.0007936, "grad_norm": 0.8440917134284973, "learning_rate": 5.696257129823446e-07, "loss": 2.6158, "step": 348510 }, { "epoch": 0.0008192, "grad_norm": 0.8468858003616333, "learning_rate": 5.693578381597642e-07, "loss": 2.6252, "step": 348520 }, { "epoch": 0.0008448, "grad_norm": 0.8775398135185242, "learning_rate": 5.690900244923914e-07, "loss": 2.5723, "step": 348530 }, { "epoch": 0.0008704, "grad_norm": 0.8467774987220764, "learning_rate": 5.688222719819658e-07, "loss": 2.466, "step": 348540 }, { "epoch": 0.000896, "grad_norm": 0.8463860750198364, "learning_rate": 5.685545806302184e-07, "loss": 2.3765, "step": 348550 }, { "epoch": 0.0009216, "grad_norm": 0.8619681596755981, "learning_rate": 5.682869504388889e-07, "loss": 2.8063, "step": 348560 }, { "epoch": 0.0009472, "grad_norm": 0.9115432500839233, "learning_rate": 5.680193814097112e-07, "loss": 2.5611, "step": 348570 }, { "epoch": 0.0009728, "grad_norm": 0.9830630421638489, "learning_rate": 5.67751873544421e-07, "loss": 2.516, "step": 348580 }, { "epoch": 0.0009984, "grad_norm": 0.7348389625549316, "learning_rate": 5.674844268447543e-07, "loss": 2.562, "step": 348590 }, { "epoch": 0.001024, "grad_norm": 0.8375670909881592, "learning_rate": 5.672170413124434e-07, "loss": 2.7319, "step": 348600 }, { "epoch": 0.0010496, "grad_norm": 0.8612847924232483, "learning_rate": 5.669497169492221e-07, "loss": 2.6595, "step": 348610 }, { "epoch": 0.0010752, "grad_norm": 0.8531004190444946, "learning_rate": 5.66682453756826e-07, "loss": 2.625, "step": 348620 }, { "epoch": 0.0011008, "grad_norm": 1.0068200826644897, "learning_rate": 5.66415251736987e-07, "loss": 2.8454, "step": 348630 }, { "epoch": 0.0011264, "grad_norm": 0.8195807933807373, "learning_rate": 5.661481108914379e-07, "loss": 2.6207, "step": 348640 }, { "epoch": 0.001152, "grad_norm": 0.9866927266120911, "learning_rate": 5.65881031221911e-07, "loss": 2.6755, "step": 348650 }, { "epoch": 0.0011776, "grad_norm": 0.8486892580986023, "learning_rate": 5.656140127301391e-07, "loss": 2.7846, "step": 348660 }, { "epoch": 0.0012032, "grad_norm": 0.8801909685134888, "learning_rate": 5.65347055417853e-07, "loss": 2.6505, "step": 348670 }, { "epoch": 0.0012288, "grad_norm": 0.877906858921051, "learning_rate": 5.650801592867839e-07, "loss": 2.8953, "step": 348680 }, { "epoch": 0.0012544, "grad_norm": 0.8338509798049927, "learning_rate": 5.648133243386611e-07, "loss": 2.5139, "step": 348690 }, { "epoch": 0.00128, "grad_norm": 0.8673874139785767, "learning_rate": 5.645465505752168e-07, "loss": 2.8122, "step": 348700 }, { "epoch": 0.0013056, "grad_norm": 0.8210430145263672, "learning_rate": 5.642798379981806e-07, "loss": 2.6915, "step": 348710 }, { "epoch": 0.0013312, "grad_norm": 0.8607224225997925, "learning_rate": 5.640131866092824e-07, "loss": 2.4311, "step": 348720 }, { "epoch": 0.0013568, "grad_norm": 0.9116901755332947, "learning_rate": 5.637465964102506e-07, "loss": 2.8907, "step": 348730 }, { "epoch": 0.0013824, "grad_norm": 0.7977197766304016, "learning_rate": 5.634800674028151e-07, "loss": 2.5066, "step": 348740 }, { "epoch": 0.001408, "grad_norm": 0.849134087562561, "learning_rate": 5.632135995887022e-07, "loss": 2.678, "step": 348750 }, { "epoch": 0.0014336, "grad_norm": 0.8451412320137024, "learning_rate": 5.629471929696417e-07, "loss": 2.6103, "step": 348760 }, { "epoch": 0.0014592, "grad_norm": 0.8278781771659851, "learning_rate": 5.6268084754736e-07, "loss": 2.6521, "step": 348770 }, { "epoch": 0.0014848, "grad_norm": 1.0545003414154053, "learning_rate": 5.624145633235833e-07, "loss": 2.2508, "step": 348780 }, { "epoch": 0.0015104, "grad_norm": 0.8900772929191589, "learning_rate": 5.621483403000416e-07, "loss": 2.2942, "step": 348790 }, { "epoch": 0.001536, "grad_norm": 1.2450132369995117, "learning_rate": 5.618821784784622e-07, "loss": 2.9875, "step": 348800 }, { "epoch": 2.56e-05, "grad_norm": 0.8310501575469971, "learning_rate": 5.616160778605661e-07, "loss": 1.6561, "step": 348810 }, { "epoch": 5.12e-05, "grad_norm": 0.8279669880867004, "learning_rate": 5.613500384480808e-07, "loss": 2.7339, "step": 348820 }, { "epoch": 7.68e-05, "grad_norm": 0.9874172806739807, "learning_rate": 5.610840602427325e-07, "loss": 2.842, "step": 348830 }, { "epoch": 0.0001024, "grad_norm": 0.866029679775238, "learning_rate": 5.608181432462457e-07, "loss": 2.6451, "step": 348840 }, { "epoch": 0.000128, "grad_norm": 0.9175143241882324, "learning_rate": 5.605522874603442e-07, "loss": 3.1682, "step": 348850 }, { "epoch": 0.0001536, "grad_norm": 0.7698187828063965, "learning_rate": 5.602864928867524e-07, "loss": 2.4078, "step": 348860 }, { "epoch": 0.0001792, "grad_norm": 0.8853026032447815, "learning_rate": 5.600207595271956e-07, "loss": 2.7678, "step": 348870 }, { "epoch": 0.0002048, "grad_norm": 0.8928279280662537, "learning_rate": 5.597550873833946e-07, "loss": 2.8519, "step": 348880 }, { "epoch": 0.0002304, "grad_norm": 0.804252028465271, "learning_rate": 5.594894764570713e-07, "loss": 2.5922, "step": 348890 }, { "epoch": 0.000256, "grad_norm": 1.0045969486236572, "learning_rate": 5.592239267499511e-07, "loss": 2.5573, "step": 348900 }, { "epoch": 0.0002816, "grad_norm": 0.951540470123291, "learning_rate": 5.589584382637536e-07, "loss": 2.7117, "step": 348910 }, { "epoch": 0.0003072, "grad_norm": 0.8470761179924011, "learning_rate": 5.586930110002031e-07, "loss": 2.5952, "step": 348920 }, { "epoch": 0.0003328, "grad_norm": 0.8183044791221619, "learning_rate": 5.584276449610182e-07, "loss": 2.6518, "step": 348930 }, { "epoch": 0.0003584, "grad_norm": 0.8840672373771667, "learning_rate": 5.581623401479219e-07, "loss": 2.8326, "step": 348940 }, { "epoch": 0.000384, "grad_norm": 0.912912130355835, "learning_rate": 5.578970965626329e-07, "loss": 2.5308, "step": 348950 }, { "epoch": 0.0004096, "grad_norm": 1.1233752965927124, "learning_rate": 5.576319142068698e-07, "loss": 3.077, "step": 348960 }, { "epoch": 0.0004352, "grad_norm": 0.8033508062362671, "learning_rate": 5.573667930823556e-07, "loss": 2.8089, "step": 348970 }, { "epoch": 0.0004608, "grad_norm": 0.8702021837234497, "learning_rate": 5.57101733190808e-07, "loss": 2.6314, "step": 348980 }, { "epoch": 0.0004864, "grad_norm": 0.8009441494941711, "learning_rate": 5.568367345339476e-07, "loss": 2.5666, "step": 348990 }, { "epoch": 0.000512, "grad_norm": 0.8300380110740662, "learning_rate": 5.56571797113491e-07, "loss": 2.621, "step": 349000 }, { "epoch": 0.0005376, "grad_norm": 0.8493019938468933, "learning_rate": 5.563069209311556e-07, "loss": 2.6416, "step": 349010 }, { "epoch": 0.0005632, "grad_norm": 0.8538175821304321, "learning_rate": 5.5604210598866e-07, "loss": 2.5846, "step": 349020 }, { "epoch": 0.0005888, "grad_norm": 0.9898551106452942, "learning_rate": 5.557773522877208e-07, "loss": 2.4218, "step": 349030 }, { "epoch": 0.0006144, "grad_norm": 0.7973692417144775, "learning_rate": 5.555126598300564e-07, "loss": 2.5925, "step": 349040 }, { "epoch": 0.00064, "grad_norm": 0.883924663066864, "learning_rate": 5.552480286173822e-07, "loss": 2.6853, "step": 349050 }, { "epoch": 0.0006656, "grad_norm": 0.8843384385108948, "learning_rate": 5.549834586514147e-07, "loss": 3.0969, "step": 349060 }, { "epoch": 0.0006912, "grad_norm": 0.8727647066116333, "learning_rate": 5.547189499338701e-07, "loss": 2.6068, "step": 349070 }, { "epoch": 0.0007168, "grad_norm": 0.8479192852973938, "learning_rate": 5.544545024664616e-07, "loss": 2.7402, "step": 349080 }, { "epoch": 0.0007424, "grad_norm": 0.9392288327217102, "learning_rate": 5.541901162509056e-07, "loss": 2.5394, "step": 349090 }, { "epoch": 0.000768, "grad_norm": 0.9035298824310303, "learning_rate": 5.539257912889162e-07, "loss": 2.6137, "step": 349100 }, { "epoch": 0.0007936, "grad_norm": 0.9139775037765503, "learning_rate": 5.536615275822066e-07, "loss": 2.6199, "step": 349110 }, { "epoch": 0.0008192, "grad_norm": 0.8594834208488464, "learning_rate": 5.533973251324931e-07, "loss": 2.5775, "step": 349120 }, { "epoch": 0.0008448, "grad_norm": 0.7940225601196289, "learning_rate": 5.531331839414866e-07, "loss": 2.5087, "step": 349130 }, { "epoch": 0.0008704, "grad_norm": 0.9213999509811401, "learning_rate": 5.528691040109002e-07, "loss": 2.7482, "step": 349140 }, { "epoch": 0.000896, "grad_norm": 0.7722503542900085, "learning_rate": 5.52605085342447e-07, "loss": 2.4002, "step": 349150 }, { "epoch": 0.0009216, "grad_norm": 0.8555559515953064, "learning_rate": 5.523411279378398e-07, "loss": 2.5495, "step": 349160 }, { "epoch": 0.0009472, "grad_norm": 0.8447521328926086, "learning_rate": 5.520772317987888e-07, "loss": 2.6936, "step": 349170 }, { "epoch": 0.0009728, "grad_norm": 1.0124313831329346, "learning_rate": 5.518133969270068e-07, "loss": 2.608, "step": 349180 }, { "epoch": 0.0009984, "grad_norm": 0.8394002914428711, "learning_rate": 5.515496233242046e-07, "loss": 2.6112, "step": 349190 }, { "epoch": 0.001024, "grad_norm": 0.8758513927459717, "learning_rate": 5.51285910992091e-07, "loss": 2.4208, "step": 349200 }, { "epoch": 0.0010496, "grad_norm": 1.0253510475158691, "learning_rate": 5.510222599323767e-07, "loss": 2.5351, "step": 349210 }, { "epoch": 0.0010752, "grad_norm": 0.809556245803833, "learning_rate": 5.507586701467727e-07, "loss": 2.7589, "step": 349220 }, { "epoch": 0.0011008, "grad_norm": 0.847825825214386, "learning_rate": 5.504951416369864e-07, "loss": 2.6117, "step": 349230 }, { "epoch": 0.0011264, "grad_norm": 0.9104542136192322, "learning_rate": 5.502316744047275e-07, "loss": 2.7586, "step": 349240 }, { "epoch": 0.001152, "grad_norm": 0.8158723711967468, "learning_rate": 5.499682684517061e-07, "loss": 2.517, "step": 349250 }, { "epoch": 0.0011776, "grad_norm": 0.9773159027099609, "learning_rate": 5.497049237796282e-07, "loss": 2.7972, "step": 349260 }, { "epoch": 0.0012032, "grad_norm": 0.8725246787071228, "learning_rate": 5.494416403902037e-07, "loss": 2.757, "step": 349270 }, { "epoch": 0.0012288, "grad_norm": 0.788474977016449, "learning_rate": 5.491784182851367e-07, "loss": 2.5779, "step": 349280 }, { "epoch": 0.0012544, "grad_norm": 0.8340736627578735, "learning_rate": 5.48915257466136e-07, "loss": 2.5364, "step": 349290 }, { "epoch": 0.00128, "grad_norm": 0.9062339067459106, "learning_rate": 5.486521579349069e-07, "loss": 2.5468, "step": 349300 }, { "epoch": 0.0013056, "grad_norm": 0.8123624324798584, "learning_rate": 5.483891196931579e-07, "loss": 2.6028, "step": 349310 }, { "epoch": 0.0013312, "grad_norm": 0.8795206546783447, "learning_rate": 5.481261427425954e-07, "loss": 2.5293, "step": 349320 }, { "epoch": 0.0013568, "grad_norm": 0.7748145461082458, "learning_rate": 5.478632270849216e-07, "loss": 2.5367, "step": 349330 }, { "epoch": 0.0013824, "grad_norm": 0.9819616079330444, "learning_rate": 5.476003727218415e-07, "loss": 2.478, "step": 349340 }, { "epoch": 0.001408, "grad_norm": 0.7731900215148926, "learning_rate": 5.473375796550617e-07, "loss": 2.4524, "step": 349350 }, { "epoch": 0.0014336, "grad_norm": 0.8628987073898315, "learning_rate": 5.470748478862852e-07, "loss": 2.5302, "step": 349360 }, { "epoch": 0.0014592, "grad_norm": 0.7834569215774536, "learning_rate": 5.468121774172164e-07, "loss": 2.7404, "step": 349370 }, { "epoch": 0.0014848, "grad_norm": 0.9840549230575562, "learning_rate": 5.46549568249558e-07, "loss": 2.0242, "step": 349380 }, { "epoch": 0.0015104, "grad_norm": 0.843743085861206, "learning_rate": 5.462870203850156e-07, "loss": 2.5188, "step": 349390 }, { "epoch": 0.001536, "grad_norm": 0.8401195406913757, "learning_rate": 5.460245338252867e-07, "loss": 2.5565, "step": 349400 }, { "epoch": 2.56e-05, "grad_norm": 0.9317243099212646, "learning_rate": 5.457621085720765e-07, "loss": 1.8264, "step": 349410 }, { "epoch": 5.12e-05, "grad_norm": 0.8074687719345093, "learning_rate": 5.454997446270871e-07, "loss": 2.6704, "step": 349420 }, { "epoch": 7.68e-05, "grad_norm": 0.8066387176513672, "learning_rate": 5.452374419920182e-07, "loss": 2.6855, "step": 349430 }, { "epoch": 0.0001024, "grad_norm": 0.906856119632721, "learning_rate": 5.449752006685716e-07, "loss": 2.3995, "step": 349440 }, { "epoch": 0.000128, "grad_norm": 0.8625527024269104, "learning_rate": 5.447130206584483e-07, "loss": 2.5649, "step": 349450 }, { "epoch": 0.0001536, "grad_norm": 0.840597927570343, "learning_rate": 5.444509019633482e-07, "loss": 2.4993, "step": 349460 }, { "epoch": 0.0001792, "grad_norm": 0.7931626439094543, "learning_rate": 5.44188844584973e-07, "loss": 2.63, "step": 349470 }, { "epoch": 0.0002048, "grad_norm": 0.848363995552063, "learning_rate": 5.43926848525016e-07, "loss": 2.6146, "step": 349480 }, { "epoch": 0.0002304, "grad_norm": 0.9644367098808289, "learning_rate": 5.436649137851824e-07, "loss": 2.662, "step": 349490 }, { "epoch": 0.000256, "grad_norm": 0.9290728569030762, "learning_rate": 5.434030403671686e-07, "loss": 2.405, "step": 349500 }, { "epoch": 0.0002816, "grad_norm": 0.8768886923789978, "learning_rate": 5.431412282726733e-07, "loss": 2.5079, "step": 349510 }, { "epoch": 0.0003072, "grad_norm": 0.8164005279541016, "learning_rate": 5.428794775033952e-07, "loss": 2.3664, "step": 349520 }, { "epoch": 0.0003328, "grad_norm": 0.8518190383911133, "learning_rate": 5.426177880610283e-07, "loss": 2.6519, "step": 349530 }, { "epoch": 0.0003584, "grad_norm": 0.869239330291748, "learning_rate": 5.423561599472716e-07, "loss": 2.6499, "step": 349540 }, { "epoch": 0.000384, "grad_norm": 1.023133635520935, "learning_rate": 5.420945931638221e-07, "loss": 2.6662, "step": 349550 }, { "epoch": 0.0004096, "grad_norm": 0.9251331090927124, "learning_rate": 5.418330877123745e-07, "loss": 2.8303, "step": 349560 }, { "epoch": 0.0004352, "grad_norm": 0.8431347012519836, "learning_rate": 5.415716435946261e-07, "loss": 2.6071, "step": 349570 }, { "epoch": 0.0004608, "grad_norm": 0.8140577077865601, "learning_rate": 5.413102608122723e-07, "loss": 2.3741, "step": 349580 }, { "epoch": 0.0004864, "grad_norm": 0.9140141606330872, "learning_rate": 5.410489393670093e-07, "loss": 2.7285, "step": 349590 }, { "epoch": 0.000512, "grad_norm": 0.8038704991340637, "learning_rate": 5.407876792605282e-07, "loss": 2.3364, "step": 349600 }, { "epoch": 0.0005376, "grad_norm": 0.8485333323478699, "learning_rate": 5.405264804945253e-07, "loss": 2.6394, "step": 349610 }, { "epoch": 0.0005632, "grad_norm": 0.8881922364234924, "learning_rate": 5.402653430706939e-07, "loss": 2.4401, "step": 349620 }, { "epoch": 0.0005888, "grad_norm": 0.9497599601745605, "learning_rate": 5.400042669907268e-07, "loss": 2.6901, "step": 349630 }, { "epoch": 0.0006144, "grad_norm": 0.8107668161392212, "learning_rate": 5.397432522563196e-07, "loss": 2.5655, "step": 349640 }, { "epoch": 0.00064, "grad_norm": 0.8681325912475586, "learning_rate": 5.394822988691617e-07, "loss": 2.4917, "step": 349650 }, { "epoch": 0.0006656, "grad_norm": 0.9128350019454956, "learning_rate": 5.392214068309476e-07, "loss": 2.709, "step": 349660 }, { "epoch": 0.0006912, "grad_norm": 0.7890488505363464, "learning_rate": 5.38960576143368e-07, "loss": 2.6994, "step": 349670 }, { "epoch": 0.0007168, "grad_norm": 0.8233095407485962, "learning_rate": 5.38699806808115e-07, "loss": 2.4329, "step": 349680 }, { "epoch": 0.0007424, "grad_norm": 0.8307976126670837, "learning_rate": 5.384390988268784e-07, "loss": 2.5967, "step": 349690 }, { "epoch": 0.000768, "grad_norm": 0.9743698835372925, "learning_rate": 5.3817845220135e-07, "loss": 2.7727, "step": 349700 }, { "epoch": 0.0007936, "grad_norm": 0.8081462383270264, "learning_rate": 5.379178669332207e-07, "loss": 2.5974, "step": 349710 }, { "epoch": 0.0008192, "grad_norm": 0.8395048379898071, "learning_rate": 5.376573430241794e-07, "loss": 2.7278, "step": 349720 }, { "epoch": 0.0008448, "grad_norm": 0.9375646114349365, "learning_rate": 5.373968804759144e-07, "loss": 2.663, "step": 349730 }, { "epoch": 0.0008704, "grad_norm": 0.8547065258026123, "learning_rate": 5.371364792901168e-07, "loss": 2.6235, "step": 349740 }, { "epoch": 0.000896, "grad_norm": 0.8688641786575317, "learning_rate": 5.368761394684729e-07, "loss": 2.298, "step": 349750 }, { "epoch": 0.0009216, "grad_norm": 0.8138684034347534, "learning_rate": 5.366158610126726e-07, "loss": 2.5353, "step": 349760 }, { "epoch": 0.0009472, "grad_norm": 0.878545343875885, "learning_rate": 5.363556439244034e-07, "loss": 2.2216, "step": 349770 }, { "epoch": 0.0009728, "grad_norm": 0.8206475377082825, "learning_rate": 5.360954882053526e-07, "loss": 2.5549, "step": 349780 }, { "epoch": 0.0009984, "grad_norm": 0.8266573548316956, "learning_rate": 5.358353938572092e-07, "loss": 2.5362, "step": 349790 }, { "epoch": 0.001024, "grad_norm": 0.9711016416549683, "learning_rate": 5.35575360881656e-07, "loss": 2.7283, "step": 349800 }, { "epoch": 0.0010496, "grad_norm": 0.9049937129020691, "learning_rate": 5.35315389280382e-07, "loss": 2.7664, "step": 349810 }, { "epoch": 0.0010752, "grad_norm": 0.9489143490791321, "learning_rate": 5.35055479055071e-07, "loss": 2.7863, "step": 349820 }, { "epoch": 0.0011008, "grad_norm": 0.8973376750946045, "learning_rate": 5.347956302074098e-07, "loss": 2.6689, "step": 349830 }, { "epoch": 0.0011264, "grad_norm": 0.8352967500686646, "learning_rate": 5.345358427390846e-07, "loss": 2.7428, "step": 349840 }, { "epoch": 0.001152, "grad_norm": 0.8125854134559631, "learning_rate": 5.342761166517796e-07, "loss": 2.5581, "step": 349850 }, { "epoch": 0.0011776, "grad_norm": 0.8560827970504761, "learning_rate": 5.340164519471769e-07, "loss": 2.7685, "step": 349860 }, { "epoch": 0.0012032, "grad_norm": 0.8617955446243286, "learning_rate": 5.337568486269618e-07, "loss": 2.8621, "step": 349870 }, { "epoch": 0.0012288, "grad_norm": 0.8321749567985535, "learning_rate": 5.334973066928173e-07, "loss": 2.5868, "step": 349880 }, { "epoch": 0.0012544, "grad_norm": 0.8953008651733398, "learning_rate": 5.332378261464277e-07, "loss": 2.5199, "step": 349890 }, { "epoch": 0.00128, "grad_norm": 0.9160609841346741, "learning_rate": 5.32978406989475e-07, "loss": 2.6699, "step": 349900 }, { "epoch": 0.0013056, "grad_norm": 0.8689252734184265, "learning_rate": 5.327190492236411e-07, "loss": 2.7705, "step": 349910 }, { "epoch": 0.0013312, "grad_norm": 0.8972148299217224, "learning_rate": 5.324597528506092e-07, "loss": 2.5515, "step": 349920 }, { "epoch": 0.0013568, "grad_norm": 0.8986034989356995, "learning_rate": 5.322005178720579e-07, "loss": 2.5777, "step": 349930 }, { "epoch": 0.0013824, "grad_norm": 0.9467219710350037, "learning_rate": 5.319413442896703e-07, "loss": 2.9541, "step": 349940 }, { "epoch": 0.001408, "grad_norm": 0.8872143626213074, "learning_rate": 5.316822321051273e-07, "loss": 2.5987, "step": 349950 }, { "epoch": 0.0014336, "grad_norm": 0.8346425890922546, "learning_rate": 5.314231813201076e-07, "loss": 2.814, "step": 349960 }, { "epoch": 0.0014592, "grad_norm": 0.7997066974639893, "learning_rate": 5.311641919362931e-07, "loss": 2.7384, "step": 349970 }, { "epoch": 0.0014848, "grad_norm": 1.0139861106872559, "learning_rate": 5.309052639553614e-07, "loss": 2.3078, "step": 349980 }, { "epoch": 0.0015104, "grad_norm": 0.8641291260719299, "learning_rate": 5.306463973789944e-07, "loss": 2.3748, "step": 349990 }, { "epoch": 0.001536, "grad_norm": 0.7774025797843933, "learning_rate": 5.303875922088664e-07, "loss": 2.5765, "step": 350000 }, { "epoch": 2.56e-05, "grad_norm": 0.897608757019043, "learning_rate": 5.301288484466594e-07, "loss": 1.85, "step": 350010 }, { "epoch": 5.12e-05, "grad_norm": 0.9067579507827759, "learning_rate": 5.298701660940498e-07, "loss": 2.7155, "step": 350020 }, { "epoch": 7.68e-05, "grad_norm": 0.8156661987304688, "learning_rate": 5.296115451527151e-07, "loss": 2.6341, "step": 350030 }, { "epoch": 0.0001024, "grad_norm": 0.8929005265235901, "learning_rate": 5.293529856243329e-07, "loss": 2.5033, "step": 350040 }, { "epoch": 0.000128, "grad_norm": 0.9382767081260681, "learning_rate": 5.290944875105808e-07, "loss": 3.0794, "step": 350050 }, { "epoch": 0.0001536, "grad_norm": 0.8973445296287537, "learning_rate": 5.288360508131329e-07, "loss": 2.4484, "step": 350060 }, { "epoch": 0.0001792, "grad_norm": 0.9412304759025574, "learning_rate": 5.285776755336658e-07, "loss": 2.6002, "step": 350070 }, { "epoch": 0.0002048, "grad_norm": 0.809540331363678, "learning_rate": 5.283193616738546e-07, "loss": 2.6691, "step": 350080 }, { "epoch": 0.0002304, "grad_norm": 0.8259446024894714, "learning_rate": 5.28061109235376e-07, "loss": 2.4711, "step": 350090 }, { "epoch": 0.000256, "grad_norm": 0.8733989000320435, "learning_rate": 5.278029182199029e-07, "loss": 2.6696, "step": 350100 }, { "epoch": 0.0002816, "grad_norm": 0.8446752429008484, "learning_rate": 5.275447886291119e-07, "loss": 2.4667, "step": 350110 }, { "epoch": 0.0003072, "grad_norm": 0.7993299961090088, "learning_rate": 5.272867204646758e-07, "loss": 2.5742, "step": 350120 }, { "epoch": 0.0003328, "grad_norm": 0.8817965388298035, "learning_rate": 5.270287137282659e-07, "loss": 2.6305, "step": 350130 }, { "epoch": 0.0003584, "grad_norm": 0.911128044128418, "learning_rate": 5.267707684215584e-07, "loss": 2.7413, "step": 350140 }, { "epoch": 0.000384, "grad_norm": 0.9237406253814697, "learning_rate": 5.265128845462242e-07, "loss": 2.6035, "step": 350150 }, { "epoch": 0.0004096, "grad_norm": 0.8996067643165588, "learning_rate": 5.262550621039364e-07, "loss": 2.6427, "step": 350160 }, { "epoch": 0.0004352, "grad_norm": 1.0160210132598877, "learning_rate": 5.259973010963648e-07, "loss": 2.7573, "step": 350170 }, { "epoch": 0.0004608, "grad_norm": 0.7727740406990051, "learning_rate": 5.257396015251859e-07, "loss": 2.6847, "step": 350180 }, { "epoch": 0.0004864, "grad_norm": 0.9649778008460999, "learning_rate": 5.254819633920672e-07, "loss": 2.7649, "step": 350190 }, { "epoch": 0.000512, "grad_norm": 0.8628746271133423, "learning_rate": 5.252243866986783e-07, "loss": 2.4906, "step": 350200 }, { "epoch": 0.0005376, "grad_norm": 0.9283989667892456, "learning_rate": 5.249668714466927e-07, "loss": 2.4922, "step": 350210 }, { "epoch": 0.0005632, "grad_norm": 0.8201594352722168, "learning_rate": 5.247094176377776e-07, "loss": 2.4914, "step": 350220 }, { "epoch": 0.0005888, "grad_norm": 0.9402357935905457, "learning_rate": 5.244520252736041e-07, "loss": 2.4942, "step": 350230 }, { "epoch": 0.0006144, "grad_norm": 0.783036470413208, "learning_rate": 5.241946943558407e-07, "loss": 2.5466, "step": 350240 }, { "epoch": 0.00064, "grad_norm": 0.9078009128570557, "learning_rate": 5.239374248861584e-07, "loss": 2.6938, "step": 350250 }, { "epoch": 0.0006656, "grad_norm": 0.9297026991844177, "learning_rate": 5.236802168662214e-07, "loss": 2.9095, "step": 350260 }, { "epoch": 0.0006912, "grad_norm": 0.8281332850456238, "learning_rate": 5.234230702977006e-07, "loss": 2.6297, "step": 350270 }, { "epoch": 0.0007168, "grad_norm": 0.8561676144599915, "learning_rate": 5.231659851822613e-07, "loss": 2.5505, "step": 350280 }, { "epoch": 0.0007424, "grad_norm": 0.8835237622261047, "learning_rate": 5.229089615215732e-07, "loss": 2.695, "step": 350290 }, { "epoch": 0.000768, "grad_norm": 0.8593960404396057, "learning_rate": 5.226519993173007e-07, "loss": 2.5427, "step": 350300 }, { "epoch": 0.0007936, "grad_norm": 0.8099113702774048, "learning_rate": 5.223950985711124e-07, "loss": 2.527, "step": 350310 }, { "epoch": 0.0008192, "grad_norm": 0.7981533408164978, "learning_rate": 5.221382592846735e-07, "loss": 2.488, "step": 350320 }, { "epoch": 0.0008448, "grad_norm": 0.9161601066589355, "learning_rate": 5.218814814596484e-07, "loss": 2.5435, "step": 350330 }, { "epoch": 0.0008704, "grad_norm": 0.9700300097465515, "learning_rate": 5.216247650977025e-07, "loss": 2.5835, "step": 350340 }, { "epoch": 0.000896, "grad_norm": 1.0849604606628418, "learning_rate": 5.213681102004986e-07, "loss": 2.534, "step": 350350 }, { "epoch": 0.0009216, "grad_norm": 0.806177020072937, "learning_rate": 5.211115167697056e-07, "loss": 3.0544, "step": 350360 }, { "epoch": 0.0009472, "grad_norm": 0.9187182784080505, "learning_rate": 5.208549848069855e-07, "loss": 2.265, "step": 350370 }, { "epoch": 0.0009728, "grad_norm": 0.8498068451881409, "learning_rate": 5.205985143140035e-07, "loss": 2.4294, "step": 350380 }, { "epoch": 0.0009984, "grad_norm": 0.8838139772415161, "learning_rate": 5.203421052924185e-07, "loss": 2.4746, "step": 350390 }, { "epoch": 0.001024, "grad_norm": 0.9385965466499329, "learning_rate": 5.200857577438956e-07, "loss": 2.8848, "step": 350400 }, { "epoch": 0.0010496, "grad_norm": 0.9601166248321533, "learning_rate": 5.198294716700969e-07, "loss": 2.7889, "step": 350410 }, { "epoch": 0.0010752, "grad_norm": 0.7902770042419434, "learning_rate": 5.195732470726855e-07, "loss": 2.5307, "step": 350420 }, { "epoch": 0.0011008, "grad_norm": 0.7978842854499817, "learning_rate": 5.193170839533213e-07, "loss": 2.7158, "step": 350430 }, { "epoch": 0.0011264, "grad_norm": 0.9881918430328369, "learning_rate": 5.190609823136661e-07, "loss": 2.5152, "step": 350440 }, { "epoch": 0.001152, "grad_norm": 0.8528633713722229, "learning_rate": 5.18804942155382e-07, "loss": 2.8704, "step": 350450 }, { "epoch": 0.0011776, "grad_norm": 0.8733841776847839, "learning_rate": 5.185489634801266e-07, "loss": 2.561, "step": 350460 }, { "epoch": 0.0012032, "grad_norm": 0.8753854632377625, "learning_rate": 5.182930462895619e-07, "loss": 2.622, "step": 350470 }, { "epoch": 0.0012288, "grad_norm": 0.8266398906707764, "learning_rate": 5.180371905853465e-07, "loss": 2.6308, "step": 350480 }, { "epoch": 0.0012544, "grad_norm": 0.841059684753418, "learning_rate": 5.177813963691402e-07, "loss": 2.6818, "step": 350490 }, { "epoch": 0.00128, "grad_norm": 0.8908947706222534, "learning_rate": 5.175256636426007e-07, "loss": 2.8289, "step": 350500 }, { "epoch": 0.0013056, "grad_norm": 0.8424166440963745, "learning_rate": 5.172699924073876e-07, "loss": 2.3919, "step": 350510 }, { "epoch": 0.0013312, "grad_norm": 0.9069783687591553, "learning_rate": 5.170143826651586e-07, "loss": 2.681, "step": 350520 }, { "epoch": 0.0013568, "grad_norm": 0.8922181725502014, "learning_rate": 5.167588344175711e-07, "loss": 2.7093, "step": 350530 }, { "epoch": 0.0013824, "grad_norm": 0.8866962194442749, "learning_rate": 5.165033476662817e-07, "loss": 2.8398, "step": 350540 }, { "epoch": 0.001408, "grad_norm": 0.8752549290657043, "learning_rate": 5.16247922412948e-07, "loss": 2.738, "step": 350550 }, { "epoch": 0.0014336, "grad_norm": 0.8526394963264465, "learning_rate": 5.159925586592263e-07, "loss": 2.4704, "step": 350560 }, { "epoch": 0.0014592, "grad_norm": 0.8707787990570068, "learning_rate": 5.157372564067719e-07, "loss": 2.7332, "step": 350570 }, { "epoch": 0.0014848, "grad_norm": 1.0581104755401611, "learning_rate": 5.154820156572427e-07, "loss": 2.1967, "step": 350580 }, { "epoch": 0.0015104, "grad_norm": 0.8093494176864624, "learning_rate": 5.152268364122903e-07, "loss": 2.0503, "step": 350590 }, { "epoch": 0.001536, "grad_norm": 0.8907436728477478, "learning_rate": 5.149717186735714e-07, "loss": 2.7314, "step": 350600 }, { "epoch": 0.0015616, "grad_norm": 0.9283519387245178, "learning_rate": 5.147166624427392e-07, "loss": 2.7483, "step": 350610 }, { "epoch": 0.0015872, "grad_norm": 0.8818942308425903, "learning_rate": 5.144616677214486e-07, "loss": 2.6515, "step": 350620 }, { "epoch": 0.0016128, "grad_norm": 0.9342600107192993, "learning_rate": 5.142067345113544e-07, "loss": 2.6953, "step": 350630 }, { "epoch": 0.0016384, "grad_norm": 0.865084171295166, "learning_rate": 5.139518628141071e-07, "loss": 2.6619, "step": 350640 }, { "epoch": 0.001664, "grad_norm": 0.8479894399642944, "learning_rate": 5.136970526313634e-07, "loss": 2.7559, "step": 350650 }, { "epoch": 0.0016896, "grad_norm": 0.8489115834236145, "learning_rate": 5.134423039647707e-07, "loss": 2.6297, "step": 350660 }, { "epoch": 0.0017152, "grad_norm": 0.8837941884994507, "learning_rate": 5.131876168159844e-07, "loss": 2.6212, "step": 350670 }, { "epoch": 0.0017408, "grad_norm": 0.8017551898956299, "learning_rate": 5.129329911866543e-07, "loss": 2.6445, "step": 350680 }, { "epoch": 0.0017664, "grad_norm": 0.8094173073768616, "learning_rate": 5.126784270784324e-07, "loss": 2.7421, "step": 350690 }, { "epoch": 0.001792, "grad_norm": 0.9041216373443604, "learning_rate": 5.124239244929708e-07, "loss": 2.4745, "step": 350700 }, { "epoch": 0.0018176, "grad_norm": 0.8367669582366943, "learning_rate": 5.121694834319168e-07, "loss": 2.6796, "step": 350710 }, { "epoch": 0.0018432, "grad_norm": 1.1769628524780273, "learning_rate": 5.119151038969228e-07, "loss": 2.466, "step": 350720 }, { "epoch": 0.0018688, "grad_norm": 0.8056312799453735, "learning_rate": 5.116607858896383e-07, "loss": 2.6653, "step": 350730 }, { "epoch": 0.0018944, "grad_norm": 0.8647946715354919, "learning_rate": 5.11406529411711e-07, "loss": 2.7261, "step": 350740 }, { "epoch": 0.00192, "grad_norm": 0.8075494170188904, "learning_rate": 5.111523344647917e-07, "loss": 2.5452, "step": 350750 }, { "epoch": 0.0019456, "grad_norm": 0.9403879046440125, "learning_rate": 5.10898201050527e-07, "loss": 2.8577, "step": 350760 }, { "epoch": 0.0019712, "grad_norm": 2.1319568157196045, "learning_rate": 5.106441291705655e-07, "loss": 2.356, "step": 350770 }, { "epoch": 0.0019968, "grad_norm": 1.1527336835861206, "learning_rate": 5.103901188265559e-07, "loss": 2.7047, "step": 350780 }, { "epoch": 0.0020224, "grad_norm": 0.8884162306785583, "learning_rate": 5.101361700201435e-07, "loss": 3.0384, "step": 350790 }, { "epoch": 0.002048, "grad_norm": 0.7817161679267883, "learning_rate": 5.09882282752976e-07, "loss": 2.4121, "step": 350800 }, { "epoch": 0.0020736, "grad_norm": 0.9210757613182068, "learning_rate": 5.096284570266996e-07, "loss": 2.5811, "step": 350810 }, { "epoch": 0.0020992, "grad_norm": 0.7938932180404663, "learning_rate": 5.093746928429599e-07, "loss": 2.9458, "step": 350820 }, { "epoch": 0.0021248, "grad_norm": 0.8251776695251465, "learning_rate": 5.091209902034034e-07, "loss": 2.8051, "step": 350830 }, { "epoch": 0.0021504, "grad_norm": 0.8399580121040344, "learning_rate": 5.088673491096751e-07, "loss": 2.7464, "step": 350840 }, { "epoch": 0.002176, "grad_norm": 0.8986846804618835, "learning_rate": 5.086137695634208e-07, "loss": 2.8124, "step": 350850 }, { "epoch": 0.0022016, "grad_norm": 0.8594548106193542, "learning_rate": 5.083602515662822e-07, "loss": 2.7318, "step": 350860 }, { "epoch": 0.0022272, "grad_norm": 0.9105708003044128, "learning_rate": 5.081067951199037e-07, "loss": 2.386, "step": 350870 }, { "epoch": 0.0022528, "grad_norm": 0.8705661296844482, "learning_rate": 5.078534002259316e-07, "loss": 2.8284, "step": 350880 }, { "epoch": 0.0022784, "grad_norm": 1.2512924671173096, "learning_rate": 5.076000668860071e-07, "loss": 2.7681, "step": 350890 }, { "epoch": 0.002304, "grad_norm": 0.8221815228462219, "learning_rate": 5.073467951017752e-07, "loss": 2.5254, "step": 350900 }, { "epoch": 0.0023296, "grad_norm": 0.8278098106384277, "learning_rate": 5.070935848748748e-07, "loss": 2.9132, "step": 350910 }, { "epoch": 0.0023552, "grad_norm": 0.8222830891609192, "learning_rate": 5.06840436206949e-07, "loss": 2.5574, "step": 350920 }, { "epoch": 0.0023808, "grad_norm": 0.8741312026977539, "learning_rate": 5.065873490996409e-07, "loss": 2.7963, "step": 350930 }, { "epoch": 0.0024064, "grad_norm": 0.8758158087730408, "learning_rate": 5.063343235545903e-07, "loss": 2.784, "step": 350940 }, { "epoch": 0.002432, "grad_norm": 0.8438246846199036, "learning_rate": 5.060813595734382e-07, "loss": 2.5466, "step": 350950 }, { "epoch": 0.0024576, "grad_norm": 0.9122306704521179, "learning_rate": 5.058284571578254e-07, "loss": 3.0353, "step": 350960 }, { "epoch": 0.0024832, "grad_norm": 0.8773837089538574, "learning_rate": 5.055756163093939e-07, "loss": 2.9355, "step": 350970 }, { "epoch": 0.0025088, "grad_norm": 0.8847016096115112, "learning_rate": 5.053228370297791e-07, "loss": 2.9993, "step": 350980 }, { "epoch": 0.0025344, "grad_norm": 0.8126504421234131, "learning_rate": 5.050701193206231e-07, "loss": 2.7759, "step": 350990 }, { "epoch": 0.00256, "grad_norm": 0.9288262724876404, "learning_rate": 5.048174631835634e-07, "loss": 2.6453, "step": 351000 }, { "epoch": 0.0025856, "grad_norm": 0.8724657297134399, "learning_rate": 5.045648686202386e-07, "loss": 2.741, "step": 351010 }, { "epoch": 0.0026112, "grad_norm": 0.8010614514350891, "learning_rate": 5.043123356322877e-07, "loss": 2.7562, "step": 351020 }, { "epoch": 0.0026368, "grad_norm": 0.8060387969017029, "learning_rate": 5.040598642213479e-07, "loss": 2.961, "step": 351030 }, { "epoch": 0.0026624, "grad_norm": 0.8505656719207764, "learning_rate": 5.038074543890547e-07, "loss": 2.7353, "step": 351040 }, { "epoch": 0.002688, "grad_norm": 0.8677623271942139, "learning_rate": 5.03555106137048e-07, "loss": 2.8741, "step": 351050 }, { "epoch": 0.0027136, "grad_norm": 0.858403205871582, "learning_rate": 5.03302819466962e-07, "loss": 2.9192, "step": 351060 }, { "epoch": 0.0027392, "grad_norm": 0.7627670764923096, "learning_rate": 5.030505943804331e-07, "loss": 2.7706, "step": 351070 }, { "epoch": 0.0027648, "grad_norm": 0.8382155895233154, "learning_rate": 5.027984308790967e-07, "loss": 2.7561, "step": 351080 }, { "epoch": 0.0027904, "grad_norm": 0.7985320091247559, "learning_rate": 5.025463289645893e-07, "loss": 2.7343, "step": 351090 }, { "epoch": 0.002816, "grad_norm": 0.8357048034667969, "learning_rate": 5.022942886385462e-07, "loss": 2.8535, "step": 351100 }, { "epoch": 0.0028416, "grad_norm": 0.8663468956947327, "learning_rate": 5.020423099025984e-07, "loss": 2.8178, "step": 351110 }, { "epoch": 0.0028672, "grad_norm": 0.8739065527915955, "learning_rate": 5.017903927583823e-07, "loss": 2.8874, "step": 351120 }, { "epoch": 0.0028928, "grad_norm": 0.8137204647064209, "learning_rate": 5.01538537207531e-07, "loss": 2.8062, "step": 351130 }, { "epoch": 0.0029184, "grad_norm": 0.8274560570716858, "learning_rate": 5.012867432516777e-07, "loss": 2.6421, "step": 351140 }, { "epoch": 0.002944, "grad_norm": 0.861392080783844, "learning_rate": 5.010350108924567e-07, "loss": 2.9954, "step": 351150 }, { "epoch": 0.0029696, "grad_norm": 0.8243095278739929, "learning_rate": 5.007833401314977e-07, "loss": 2.7515, "step": 351160 }, { "epoch": 0.0029952, "grad_norm": 0.8319186568260193, "learning_rate": 5.00531730970436e-07, "loss": 2.7647, "step": 351170 }, { "epoch": 0.0030208, "grad_norm": 0.9695500731468201, "learning_rate": 5.002801834109006e-07, "loss": 2.916, "step": 351180 }, { "epoch": 0.0030464, "grad_norm": 0.9404884576797485, "learning_rate": 5.000286974545221e-07, "loss": 2.6702, "step": 351190 }, { "epoch": 0.003072, "grad_norm": 0.79917311668396, "learning_rate": 4.997772731029338e-07, "loss": 2.8431, "step": 351200 }, { "epoch": 0.0030976, "grad_norm": 0.8428704142570496, "learning_rate": 4.995259103577643e-07, "loss": 2.9258, "step": 351210 }, { "epoch": 0.0031232, "grad_norm": 0.9556376934051514, "learning_rate": 4.992746092206435e-07, "loss": 2.83, "step": 351220 }, { "epoch": 0.0031488, "grad_norm": 0.8636236190795898, "learning_rate": 4.990233696932045e-07, "loss": 3.0215, "step": 351230 }, { "epoch": 0.0031744, "grad_norm": 0.8336655497550964, "learning_rate": 4.987721917770727e-07, "loss": 2.9724, "step": 351240 }, { "epoch": 0.0032, "grad_norm": 0.9054128527641296, "learning_rate": 4.985210754738779e-07, "loss": 2.768, "step": 351250 }, { "epoch": 0.0032256, "grad_norm": 0.802214503288269, "learning_rate": 4.982700207852487e-07, "loss": 2.9928, "step": 351260 }, { "epoch": 0.0032512, "grad_norm": 0.8524966835975647, "learning_rate": 4.980190277128138e-07, "loss": 2.6839, "step": 351270 }, { "epoch": 0.0032768, "grad_norm": 0.8592745661735535, "learning_rate": 4.977680962581999e-07, "loss": 2.8129, "step": 351280 }, { "epoch": 0.0033024, "grad_norm": 0.8121400475502014, "learning_rate": 4.975172264230343e-07, "loss": 2.6412, "step": 351290 }, { "epoch": 0.003328, "grad_norm": 0.8426274061203003, "learning_rate": 4.972664182089449e-07, "loss": 2.8617, "step": 351300 }, { "epoch": 0.0033536, "grad_norm": 1.3109554052352905, "learning_rate": 4.970156716175567e-07, "loss": 2.5928, "step": 351310 }, { "epoch": 0.0033792, "grad_norm": 0.925957202911377, "learning_rate": 4.967649866504954e-07, "loss": 2.9538, "step": 351320 }, { "epoch": 0.0034048, "grad_norm": 0.8322237133979797, "learning_rate": 4.965143633093882e-07, "loss": 2.8975, "step": 351330 }, { "epoch": 0.0034304, "grad_norm": 1.0216294527053833, "learning_rate": 4.962638015958598e-07, "loss": 2.9832, "step": 351340 }, { "epoch": 0.003456, "grad_norm": 0.7983812093734741, "learning_rate": 4.960133015115343e-07, "loss": 2.8833, "step": 351350 }, { "epoch": 0.0034816, "grad_norm": 0.9379638433456421, "learning_rate": 4.957628630580368e-07, "loss": 2.7285, "step": 351360 }, { "epoch": 0.0035072, "grad_norm": 0.8049780130386353, "learning_rate": 4.955124862369931e-07, "loss": 2.6581, "step": 351370 }, { "epoch": 0.0035328, "grad_norm": 0.85185307264328, "learning_rate": 4.952621710500227e-07, "loss": 2.6961, "step": 351380 }, { "epoch": 0.0035584, "grad_norm": 0.9061744809150696, "learning_rate": 4.9501191749875e-07, "loss": 2.7253, "step": 351390 }, { "epoch": 0.003584, "grad_norm": 0.8581629395484924, "learning_rate": 4.947617255848003e-07, "loss": 2.8548, "step": 351400 }, { "epoch": 0.0036096, "grad_norm": 0.8680216073989868, "learning_rate": 4.945115953097946e-07, "loss": 3.0306, "step": 351410 }, { "epoch": 0.0036352, "grad_norm": 0.813416063785553, "learning_rate": 4.942615266753558e-07, "loss": 2.8316, "step": 351420 }, { "epoch": 0.0036608, "grad_norm": 0.9000096321105957, "learning_rate": 4.940115196831052e-07, "loss": 2.8959, "step": 351430 }, { "epoch": 0.0036864, "grad_norm": 1.0285495519638062, "learning_rate": 4.937615743346624e-07, "loss": 2.9677, "step": 351440 }, { "epoch": 0.003712, "grad_norm": 0.8575134873390198, "learning_rate": 4.935116906316484e-07, "loss": 2.7068, "step": 351450 }, { "epoch": 0.0037376, "grad_norm": 0.8008434772491455, "learning_rate": 4.932618685756852e-07, "loss": 2.5248, "step": 351460 }, { "epoch": 0.0037632, "grad_norm": 0.9993950724601746, "learning_rate": 4.930121081683925e-07, "loss": 2.8427, "step": 351470 }, { "epoch": 0.0037888, "grad_norm": 0.7710146307945251, "learning_rate": 4.927624094113892e-07, "loss": 2.8154, "step": 351480 }, { "epoch": 0.0038144, "grad_norm": 1.193275809288025, "learning_rate": 4.925127723062961e-07, "loss": 3.0182, "step": 351490 }, { "epoch": 0.00384, "grad_norm": 0.7977451086044312, "learning_rate": 4.922631968547309e-07, "loss": 2.7315, "step": 351500 }, { "epoch": 0.0038656, "grad_norm": 0.8742096424102783, "learning_rate": 4.920136830583111e-07, "loss": 2.5582, "step": 351510 }, { "epoch": 0.0038912, "grad_norm": 0.9640178084373474, "learning_rate": 4.917642309186555e-07, "loss": 2.8609, "step": 351520 }, { "epoch": 0.0039168, "grad_norm": 0.7860305905342102, "learning_rate": 4.915148404373826e-07, "loss": 2.8595, "step": 351530 }, { "epoch": 0.0039424, "grad_norm": 0.8190034627914429, "learning_rate": 4.91265511616108e-07, "loss": 2.6916, "step": 351540 }, { "epoch": 0.003968, "grad_norm": 0.822101354598999, "learning_rate": 4.910162444564504e-07, "loss": 2.6445, "step": 351550 }, { "epoch": 0.0039936, "grad_norm": 0.7635137438774109, "learning_rate": 4.907670389600239e-07, "loss": 2.7732, "step": 351560 }, { "epoch": 0.0040192, "grad_norm": 0.8849213123321533, "learning_rate": 4.905178951284473e-07, "loss": 2.8357, "step": 351570 }, { "epoch": 0.0040448, "grad_norm": 0.8508462309837341, "learning_rate": 4.902688129633337e-07, "loss": 2.9204, "step": 351580 }, { "epoch": 0.0040704, "grad_norm": 0.8189353942871094, "learning_rate": 4.900197924662998e-07, "loss": 2.8183, "step": 351590 }, { "epoch": 0.004096, "grad_norm": 0.7848419547080994, "learning_rate": 4.897708336389606e-07, "loss": 2.8355, "step": 351600 }, { "epoch": 0.0041216, "grad_norm": 0.8448570370674133, "learning_rate": 4.895219364829296e-07, "loss": 2.7888, "step": 351610 }, { "epoch": 0.0041472, "grad_norm": 0.8041236400604248, "learning_rate": 4.892731009998208e-07, "loss": 2.6066, "step": 351620 }, { "epoch": 0.0041728, "grad_norm": 0.8419344425201416, "learning_rate": 4.890243271912509e-07, "loss": 2.8421, "step": 351630 }, { "epoch": 0.0041984, "grad_norm": 0.916447103023529, "learning_rate": 4.887756150588285e-07, "loss": 2.8246, "step": 351640 }, { "epoch": 0.004224, "grad_norm": 1.0325754880905151, "learning_rate": 4.885269646041679e-07, "loss": 2.9412, "step": 351650 }, { "epoch": 0.0042496, "grad_norm": 0.8719403743743896, "learning_rate": 4.882783758288823e-07, "loss": 2.8698, "step": 351660 }, { "epoch": 0.0042752, "grad_norm": 0.8852275609970093, "learning_rate": 4.880298487345847e-07, "loss": 2.8299, "step": 351670 }, { "epoch": 0.0043008, "grad_norm": 0.8324098587036133, "learning_rate": 4.877813833228839e-07, "loss": 2.7559, "step": 351680 }, { "epoch": 0.0043264, "grad_norm": 0.8066483736038208, "learning_rate": 4.875329795953943e-07, "loss": 2.7552, "step": 351690 }, { "epoch": 0.004352, "grad_norm": 0.8549336194992065, "learning_rate": 4.872846375537255e-07, "loss": 2.6987, "step": 351700 }, { "epoch": 0.0043776, "grad_norm": 0.9014918804168701, "learning_rate": 4.870363571994862e-07, "loss": 2.6558, "step": 351710 }, { "epoch": 0.0044032, "grad_norm": 0.8004995584487915, "learning_rate": 4.867881385342888e-07, "loss": 2.7195, "step": 351720 }, { "epoch": 0.0044288, "grad_norm": 0.7979987263679504, "learning_rate": 4.865399815597416e-07, "loss": 2.6712, "step": 351730 }, { "epoch": 0.0044544, "grad_norm": 0.9827778935432434, "learning_rate": 4.862918862774524e-07, "loss": 2.8672, "step": 351740 }, { "epoch": 0.00448, "grad_norm": 0.9031880497932434, "learning_rate": 4.860438526890332e-07, "loss": 2.7451, "step": 351750 }, { "epoch": 0.0045056, "grad_norm": 0.9104956388473511, "learning_rate": 4.857958807960938e-07, "loss": 2.6432, "step": 351760 }, { "epoch": 0.0045312, "grad_norm": 0.7846724987030029, "learning_rate": 4.855479706002363e-07, "loss": 2.8689, "step": 351770 }, { "epoch": 0.0045568, "grad_norm": 0.8783848285675049, "learning_rate": 4.853001221030728e-07, "loss": 2.9795, "step": 351780 }, { "epoch": 0.0045824, "grad_norm": 0.9233590960502625, "learning_rate": 4.850523353062097e-07, "loss": 2.7911, "step": 351790 }, { "epoch": 0.004608, "grad_norm": 1.1384329795837402, "learning_rate": 4.848046102112524e-07, "loss": 2.8048, "step": 351800 }, { "epoch": 0.0046336, "grad_norm": 0.8906714916229248, "learning_rate": 4.845569468198085e-07, "loss": 2.8699, "step": 351810 }, { "epoch": 0.0046592, "grad_norm": 1.0106784105300903, "learning_rate": 4.843093451334846e-07, "loss": 2.8109, "step": 351820 }, { "epoch": 0.0046848, "grad_norm": 0.7727528214454651, "learning_rate": 4.840618051538859e-07, "loss": 2.9537, "step": 351830 }, { "epoch": 0.0047104, "grad_norm": 0.8736231923103333, "learning_rate": 4.838143268826167e-07, "loss": 2.7738, "step": 351840 }, { "epoch": 0.004736, "grad_norm": 0.806848406791687, "learning_rate": 4.835669103212825e-07, "loss": 2.6158, "step": 351850 }, { "epoch": 0.0047616, "grad_norm": 0.8449020385742188, "learning_rate": 4.833195554714875e-07, "loss": 2.5122, "step": 351860 }, { "epoch": 0.0047872, "grad_norm": 0.7927678227424622, "learning_rate": 4.830722623348361e-07, "loss": 2.9494, "step": 351870 }, { "epoch": 0.0048128, "grad_norm": 0.9515573978424072, "learning_rate": 4.828250309129323e-07, "loss": 2.8878, "step": 351880 }, { "epoch": 0.0048384, "grad_norm": 0.815557599067688, "learning_rate": 4.825778612073783e-07, "loss": 2.8923, "step": 351890 }, { "epoch": 0.004864, "grad_norm": 0.8837867975234985, "learning_rate": 4.823307532197786e-07, "loss": 2.7187, "step": 351900 }, { "epoch": 0.0048896, "grad_norm": 0.8063586950302124, "learning_rate": 4.820837069517326e-07, "loss": 2.7891, "step": 351910 }, { "epoch": 0.0049152, "grad_norm": 0.8729557394981384, "learning_rate": 4.818367224048448e-07, "loss": 2.7872, "step": 351920 }, { "epoch": 0.0049408, "grad_norm": 0.8326566815376282, "learning_rate": 4.815897995807162e-07, "loss": 2.8632, "step": 351930 }, { "epoch": 0.0049664, "grad_norm": 0.8340063095092773, "learning_rate": 4.813429384809476e-07, "loss": 2.7819, "step": 351940 }, { "epoch": 0.004992, "grad_norm": 1.7407233715057373, "learning_rate": 4.810961391071411e-07, "loss": 2.7547, "step": 351950 }, { "epoch": 0.0050176, "grad_norm": 0.9502248167991638, "learning_rate": 4.808494014608978e-07, "loss": 2.6366, "step": 351960 }, { "epoch": 0.0050432, "grad_norm": 0.8069524765014648, "learning_rate": 4.806027255438139e-07, "loss": 2.9325, "step": 351970 }, { "epoch": 0.0050688, "grad_norm": 0.8754297494888306, "learning_rate": 4.803561113574928e-07, "loss": 2.9716, "step": 351980 }, { "epoch": 0.0050944, "grad_norm": 0.8766047358512878, "learning_rate": 4.801095589035309e-07, "loss": 2.8244, "step": 351990 }, { "epoch": 0.00512, "grad_norm": 1.086303472518921, "learning_rate": 4.798630681835292e-07, "loss": 2.7147, "step": 352000 }, { "epoch": 0.0051456, "grad_norm": 0.8431174755096436, "learning_rate": 4.796166391990853e-07, "loss": 2.6948, "step": 352010 }, { "epoch": 0.0051712, "grad_norm": 0.8526707887649536, "learning_rate": 4.793702719517979e-07, "loss": 2.7577, "step": 352020 }, { "epoch": 0.0051968, "grad_norm": 0.8159906268119812, "learning_rate": 4.791239664432645e-07, "loss": 2.8446, "step": 352030 }, { "epoch": 0.0052224, "grad_norm": 0.8321895599365234, "learning_rate": 4.788777226750807e-07, "loss": 2.7491, "step": 352040 }, { "epoch": 0.005248, "grad_norm": 2.1457595825195312, "learning_rate": 4.78631540648844e-07, "loss": 3.335, "step": 352050 }, { "epoch": 0.0052736, "grad_norm": 0.8395318388938904, "learning_rate": 4.78385420366152e-07, "loss": 2.8733, "step": 352060 }, { "epoch": 0.0052992, "grad_norm": 0.8722394108772278, "learning_rate": 4.781393618285989e-07, "loss": 2.7853, "step": 352070 }, { "epoch": 0.0053248, "grad_norm": 0.7723860740661621, "learning_rate": 4.778933650377826e-07, "loss": 2.6505, "step": 352080 }, { "epoch": 0.0053504, "grad_norm": 0.8967999219894409, "learning_rate": 4.776474299952972e-07, "loss": 2.8325, "step": 352090 }, { "epoch": 0.005376, "grad_norm": 0.9504229426383972, "learning_rate": 4.774015567027368e-07, "loss": 2.7141, "step": 352100 }, { "epoch": 0.0054016, "grad_norm": 0.7526656985282898, "learning_rate": 4.77155745161696e-07, "loss": 2.5356, "step": 352110 }, { "epoch": 0.0054272, "grad_norm": 0.787237286567688, "learning_rate": 4.769099953737711e-07, "loss": 2.7921, "step": 352120 }, { "epoch": 0.0054528, "grad_norm": 0.830647349357605, "learning_rate": 4.766643073405519e-07, "loss": 2.942, "step": 352130 }, { "epoch": 0.0054784, "grad_norm": 1.7305999994277954, "learning_rate": 4.764186810636351e-07, "loss": 2.7008, "step": 352140 }, { "epoch": 0.005504, "grad_norm": 0.764342188835144, "learning_rate": 4.761731165446115e-07, "loss": 2.8047, "step": 352150 }, { "epoch": 0.0055296, "grad_norm": 0.9079641103744507, "learning_rate": 4.759276137850755e-07, "loss": 2.9123, "step": 352160 }, { "epoch": 0.0055552, "grad_norm": 1.2166138887405396, "learning_rate": 4.7568217278661676e-07, "loss": 2.7888, "step": 352170 }, { "epoch": 0.0055808, "grad_norm": 0.8668347001075745, "learning_rate": 4.754367935508275e-07, "loss": 2.8275, "step": 352180 }, { "epoch": 0.0056064, "grad_norm": 0.8030349016189575, "learning_rate": 4.751914760792997e-07, "loss": 2.5146, "step": 352190 }, { "epoch": 0.005632, "grad_norm": 0.7948160767555237, "learning_rate": 4.7494622037362327e-07, "loss": 2.7945, "step": 352200 }, { "epoch": 0.0056576, "grad_norm": 0.8393798470497131, "learning_rate": 4.7470102643538907e-07, "loss": 2.5715, "step": 352210 }, { "epoch": 0.0056832, "grad_norm": 0.7813737988471985, "learning_rate": 4.744558942661881e-07, "loss": 2.843, "step": 352220 }, { "epoch": 0.0057088, "grad_norm": 1.2686995267868042, "learning_rate": 4.742108238676102e-07, "loss": 2.9539, "step": 352230 }, { "epoch": 0.0057344, "grad_norm": 0.9307237863540649, "learning_rate": 4.7396581524124077e-07, "loss": 2.8259, "step": 352240 }, { "epoch": 0.00576, "grad_norm": 0.8024680018424988, "learning_rate": 4.7372086838867293e-07, "loss": 2.5709, "step": 352250 }, { "epoch": 0.0057856, "grad_norm": 0.801297128200531, "learning_rate": 4.7347598331149103e-07, "loss": 2.7294, "step": 352260 }, { "epoch": 0.0058112, "grad_norm": 0.7974653244018555, "learning_rate": 4.732311600112871e-07, "loss": 2.592, "step": 352270 }, { "epoch": 0.0058368, "grad_norm": 0.8130019903182983, "learning_rate": 4.729863984896488e-07, "loss": 2.789, "step": 352280 }, { "epoch": 0.0058624, "grad_norm": 0.8145472407341003, "learning_rate": 4.7274169874816033e-07, "loss": 2.6486, "step": 352290 }, { "epoch": 0.005888, "grad_norm": 0.8297227025032043, "learning_rate": 4.724970607884105e-07, "loss": 2.8585, "step": 352300 }, { "epoch": 0.0059136, "grad_norm": 0.8845117688179016, "learning_rate": 4.722524846119847e-07, "loss": 2.7505, "step": 352310 }, { "epoch": 0.0059392, "grad_norm": 0.7791675329208374, "learning_rate": 4.720079702204694e-07, "loss": 2.6878, "step": 352320 }, { "epoch": 0.0059648, "grad_norm": 0.8762128949165344, "learning_rate": 4.7176351761544893e-07, "loss": 2.7329, "step": 352330 }, { "epoch": 0.0059904, "grad_norm": 0.8515294790267944, "learning_rate": 4.7151912679851086e-07, "loss": 2.7545, "step": 352340 }, { "epoch": 0.006016, "grad_norm": 0.9061924815177917, "learning_rate": 4.7127479777123953e-07, "loss": 2.8809, "step": 352350 }, { "epoch": 0.0060416, "grad_norm": 0.8584676384925842, "learning_rate": 4.710305305352192e-07, "loss": 2.7577, "step": 352360 }, { "epoch": 0.0060672, "grad_norm": 0.821323812007904, "learning_rate": 4.7078632509203194e-07, "loss": 2.6859, "step": 352370 }, { "epoch": 0.0060928, "grad_norm": 1.057686448097229, "learning_rate": 4.7054218144326314e-07, "loss": 2.7132, "step": 352380 }, { "epoch": 0.0061184, "grad_norm": 0.8636132478713989, "learning_rate": 4.70298099590496e-07, "loss": 2.5539, "step": 352390 }, { "epoch": 0.006144, "grad_norm": 0.9394811391830444, "learning_rate": 4.7005407953531257e-07, "loss": 2.7878, "step": 352400 }, { "epoch": 0.0061696, "grad_norm": 0.8699447512626648, "learning_rate": 4.6981012127929494e-07, "loss": 2.8168, "step": 352410 }, { "epoch": 0.0061952, "grad_norm": 1.3908991813659668, "learning_rate": 4.695662248240274e-07, "loss": 2.9445, "step": 352420 }, { "epoch": 0.0062208, "grad_norm": 0.7673304677009583, "learning_rate": 4.693223901710897e-07, "loss": 2.7197, "step": 352430 }, { "epoch": 0.0062464, "grad_norm": 0.8306951522827148, "learning_rate": 4.6907861732206186e-07, "loss": 2.604, "step": 352440 }, { "epoch": 0.006272, "grad_norm": 0.8902711272239685, "learning_rate": 4.688349062785269e-07, "loss": 2.7721, "step": 352450 }, { "epoch": 0.0062976, "grad_norm": 0.8329707384109497, "learning_rate": 4.6859125704206476e-07, "loss": 2.7099, "step": 352460 }, { "epoch": 0.0063232, "grad_norm": 0.8685725927352905, "learning_rate": 4.6834766961425527e-07, "loss": 2.6297, "step": 352470 }, { "epoch": 0.0063488, "grad_norm": 0.8523768186569214, "learning_rate": 4.681041439966794e-07, "loss": 2.7763, "step": 352480 }, { "epoch": 0.0063744, "grad_norm": 0.9096477031707764, "learning_rate": 4.678606801909136e-07, "loss": 2.965, "step": 352490 }, { "epoch": 0.0064, "grad_norm": 0.8544920682907104, "learning_rate": 4.6761727819853666e-07, "loss": 2.8579, "step": 352500 }, { "epoch": 0.0064256, "grad_norm": 0.8734423518180847, "learning_rate": 4.6737393802112953e-07, "loss": 2.8087, "step": 352510 }, { "epoch": 0.0064512, "grad_norm": 0.7992380857467651, "learning_rate": 4.6713065966026873e-07, "loss": 2.5781, "step": 352520 }, { "epoch": 0.0064768, "grad_norm": 0.8270416259765625, "learning_rate": 4.6688744311753187e-07, "loss": 2.8398, "step": 352530 }, { "epoch": 0.0065024, "grad_norm": 0.8306363821029663, "learning_rate": 4.666442883944966e-07, "loss": 2.8491, "step": 352540 }, { "epoch": 0.006528, "grad_norm": 0.804168164730072, "learning_rate": 4.6640119549273944e-07, "loss": 2.8649, "step": 352550 }, { "epoch": 0.0065536, "grad_norm": 0.7867670059204102, "learning_rate": 4.66158164413838e-07, "loss": 2.5382, "step": 352560 }, { "epoch": 0.0065792, "grad_norm": 0.7891609072685242, "learning_rate": 4.6591519515936546e-07, "loss": 2.809, "step": 352570 }, { "epoch": 0.0066048, "grad_norm": 0.8166806697845459, "learning_rate": 4.6567228773089943e-07, "loss": 2.842, "step": 352580 }, { "epoch": 0.0066304, "grad_norm": 0.7945295572280884, "learning_rate": 4.654294421300143e-07, "loss": 2.7344, "step": 352590 }, { "epoch": 0.006656, "grad_norm": 0.8243559002876282, "learning_rate": 4.6518665835828646e-07, "loss": 2.9351, "step": 352600 }, { "epoch": 0.0066816, "grad_norm": 0.9565250277519226, "learning_rate": 4.649439364172881e-07, "loss": 2.8497, "step": 352610 }, { "epoch": 0.0067072, "grad_norm": 0.9570237398147583, "learning_rate": 4.647012763085934e-07, "loss": 2.8448, "step": 352620 }, { "epoch": 0.0067328, "grad_norm": 0.8350675106048584, "learning_rate": 4.6445867803377786e-07, "loss": 2.8853, "step": 352630 }, { "epoch": 0.0067584, "grad_norm": 0.7696576714515686, "learning_rate": 4.642161415944135e-07, "loss": 2.6276, "step": 352640 }, { "epoch": 0.006784, "grad_norm": 0.8406043648719788, "learning_rate": 4.639736669920736e-07, "loss": 2.7398, "step": 352650 }, { "epoch": 0.0068096, "grad_norm": 0.8517913222312927, "learning_rate": 4.63731254228329e-07, "loss": 2.7716, "step": 352660 }, { "epoch": 0.0068352, "grad_norm": 0.7981236577033997, "learning_rate": 4.6348890330475295e-07, "loss": 2.6339, "step": 352670 }, { "epoch": 0.0068608, "grad_norm": 1.0334773063659668, "learning_rate": 4.632466142229186e-07, "loss": 2.9114, "step": 352680 }, { "epoch": 0.0068864, "grad_norm": 0.8274282217025757, "learning_rate": 4.6300438698439367e-07, "loss": 2.7992, "step": 352690 }, { "epoch": 0.006912, "grad_norm": 0.8724181652069092, "learning_rate": 4.627622215907512e-07, "loss": 2.7323, "step": 352700 }, { "epoch": 0.0069376, "grad_norm": 0.8775588274002075, "learning_rate": 4.625201180435612e-07, "loss": 2.8968, "step": 352710 }, { "epoch": 0.0069632, "grad_norm": 0.8608707785606384, "learning_rate": 4.6227807634439236e-07, "loss": 2.6574, "step": 352720 }, { "epoch": 0.0069888, "grad_norm": 0.8064048290252686, "learning_rate": 4.6203609649481674e-07, "loss": 2.6705, "step": 352730 }, { "epoch": 0.0070144, "grad_norm": 1.198146104812622, "learning_rate": 4.617941784964008e-07, "loss": 3.0845, "step": 352740 }, { "epoch": 0.00704, "grad_norm": 0.9253256320953369, "learning_rate": 4.615523223507157e-07, "loss": 2.8021, "step": 352750 }, { "epoch": 0.0070656, "grad_norm": 0.969516932964325, "learning_rate": 4.6131052805932995e-07, "loss": 2.8083, "step": 352760 }, { "epoch": 0.0070912, "grad_norm": 0.8180283308029175, "learning_rate": 4.61068795623808e-07, "loss": 2.4214, "step": 352770 }, { "epoch": 0.0071168, "grad_norm": 0.8798481822013855, "learning_rate": 4.608271250457197e-07, "loss": 3.1361, "step": 352780 }, { "epoch": 0.0071424, "grad_norm": 0.8287530541419983, "learning_rate": 4.6058551632663376e-07, "loss": 2.8865, "step": 352790 }, { "epoch": 0.007168, "grad_norm": 0.8389014601707458, "learning_rate": 4.6034396946811446e-07, "loss": 2.8322, "step": 352800 }, { "epoch": 0.0071936, "grad_norm": 0.8314492702484131, "learning_rate": 4.6010248447173055e-07, "loss": 2.9313, "step": 352810 }, { "epoch": 0.0072192, "grad_norm": 0.8170209527015686, "learning_rate": 4.598610613390464e-07, "loss": 2.7391, "step": 352820 }, { "epoch": 0.0072448, "grad_norm": 0.9183225631713867, "learning_rate": 4.596197000716263e-07, "loss": 2.7572, "step": 352830 }, { "epoch": 0.0072704, "grad_norm": 0.8542091250419617, "learning_rate": 4.5937840067103776e-07, "loss": 3.0382, "step": 352840 }, { "epoch": 0.007296, "grad_norm": 0.853709876537323, "learning_rate": 4.5913716313884527e-07, "loss": 2.8607, "step": 352850 }, { "epoch": 0.0073216, "grad_norm": 0.7348785996437073, "learning_rate": 4.588959874766119e-07, "loss": 2.8252, "step": 352860 }, { "epoch": 0.0073472, "grad_norm": 0.8556313514709473, "learning_rate": 4.58654873685902e-07, "loss": 2.7536, "step": 352870 }, { "epoch": 0.0073728, "grad_norm": 0.8407496809959412, "learning_rate": 4.58413821768281e-07, "loss": 2.8152, "step": 352880 }, { "epoch": 0.0073984, "grad_norm": 0.8873947262763977, "learning_rate": 4.581728317253098e-07, "loss": 2.8478, "step": 352890 }, { "epoch": 0.007424, "grad_norm": 0.8440901041030884, "learning_rate": 4.5793190355855056e-07, "loss": 2.8266, "step": 352900 }, { "epoch": 0.0074496, "grad_norm": 0.8473105430603027, "learning_rate": 4.576910372695686e-07, "loss": 2.9686, "step": 352910 }, { "epoch": 0.0074752, "grad_norm": 0.9879277348518372, "learning_rate": 4.5745023285992283e-07, "loss": 2.8573, "step": 352920 }, { "epoch": 0.0075008, "grad_norm": 0.7807134985923767, "learning_rate": 4.5720949033117745e-07, "loss": 2.7416, "step": 352930 }, { "epoch": 0.0075264, "grad_norm": 0.9192603826522827, "learning_rate": 4.569688096848912e-07, "loss": 2.7459, "step": 352940 }, { "epoch": 0.007552, "grad_norm": 0.8448039293289185, "learning_rate": 4.5672819092262845e-07, "loss": 2.7892, "step": 352950 }, { "epoch": 0.0075776, "grad_norm": 0.8414396047592163, "learning_rate": 4.564876340459434e-07, "loss": 2.7824, "step": 352960 }, { "epoch": 0.0076032, "grad_norm": 0.8551763892173767, "learning_rate": 4.562471390564016e-07, "loss": 2.8864, "step": 352970 }, { "epoch": 0.0076288, "grad_norm": 0.8086345195770264, "learning_rate": 4.560067059555606e-07, "loss": 3.0587, "step": 352980 }, { "epoch": 0.0076544, "grad_norm": 0.8022440075874329, "learning_rate": 4.5576633474497924e-07, "loss": 2.9345, "step": 352990 }, { "epoch": 0.00768, "grad_norm": 0.7989379167556763, "learning_rate": 4.555260254262173e-07, "loss": 2.9713, "step": 353000 }, { "epoch": 0.0077056, "grad_norm": 0.9031038880348206, "learning_rate": 4.552857780008335e-07, "loss": 2.8154, "step": 353010 }, { "epoch": 0.0077312, "grad_norm": 0.8953186869621277, "learning_rate": 4.550455924703834e-07, "loss": 2.9853, "step": 353020 }, { "epoch": 0.0077568, "grad_norm": 0.8395503163337708, "learning_rate": 4.548054688364256e-07, "loss": 2.7861, "step": 353030 }, { "epoch": 0.0077824, "grad_norm": 0.8528426289558411, "learning_rate": 4.5456540710051786e-07, "loss": 2.7806, "step": 353040 }, { "epoch": 0.007808, "grad_norm": 0.803119957447052, "learning_rate": 4.543254072642156e-07, "loss": 2.9034, "step": 353050 }, { "epoch": 0.0078336, "grad_norm": 0.8249615430831909, "learning_rate": 4.540854693290775e-07, "loss": 2.8069, "step": 353060 }, { "epoch": 0.0078592, "grad_norm": 0.8462989330291748, "learning_rate": 4.538455932966579e-07, "loss": 2.6342, "step": 353070 }, { "epoch": 0.0078848, "grad_norm": 0.845885694026947, "learning_rate": 4.536057791685133e-07, "loss": 2.8944, "step": 353080 }, { "epoch": 0.0079104, "grad_norm": 0.9163857102394104, "learning_rate": 4.53366026946197e-07, "loss": 2.7564, "step": 353090 }, { "epoch": 0.007936, "grad_norm": 0.8195629715919495, "learning_rate": 4.5312633663126547e-07, "loss": 2.7991, "step": 353100 }, { "epoch": 0.0079616, "grad_norm": 0.8236821889877319, "learning_rate": 4.528867082252719e-07, "loss": 2.9093, "step": 353110 }, { "epoch": 0.0079872, "grad_norm": 0.8219122886657715, "learning_rate": 4.5264714172977063e-07, "loss": 2.743, "step": 353120 }, { "epoch": 0.0080128, "grad_norm": 0.8075352907180786, "learning_rate": 4.5240763714631374e-07, "loss": 2.7042, "step": 353130 }, { "epoch": 0.0080384, "grad_norm": 1.2365063428878784, "learning_rate": 4.5216819447645886e-07, "loss": 2.8455, "step": 353140 }, { "epoch": 0.008064, "grad_norm": 0.8100870847702026, "learning_rate": 4.519288137217537e-07, "loss": 2.9136, "step": 353150 }, { "epoch": 0.0080896, "grad_norm": 0.8609246015548706, "learning_rate": 4.516894948837536e-07, "loss": 2.7746, "step": 353160 }, { "epoch": 0.0081152, "grad_norm": 0.9199474453926086, "learning_rate": 4.5145023796400957e-07, "loss": 2.8722, "step": 353170 }, { "epoch": 0.0081408, "grad_norm": 0.7863293886184692, "learning_rate": 4.5121104296407146e-07, "loss": 2.6623, "step": 353180 }, { "epoch": 0.0081664, "grad_norm": 0.8494586944580078, "learning_rate": 4.5097190988549365e-07, "loss": 2.8111, "step": 353190 }, { "epoch": 0.008192, "grad_norm": 1.070473313331604, "learning_rate": 4.5073283872982485e-07, "loss": 2.8017, "step": 353200 }, { "epoch": 0.0082176, "grad_norm": 0.9119670391082764, "learning_rate": 4.504938294986161e-07, "loss": 2.8396, "step": 353210 }, { "epoch": 0.0082432, "grad_norm": 0.7715638875961304, "learning_rate": 4.502548821934172e-07, "loss": 2.6282, "step": 353220 }, { "epoch": 0.0082688, "grad_norm": 0.854053795337677, "learning_rate": 4.500159968157758e-07, "loss": 2.7854, "step": 353230 }, { "epoch": 0.0082944, "grad_norm": 0.8128269910812378, "learning_rate": 4.497771733672429e-07, "loss": 2.743, "step": 353240 }, { "epoch": 0.00832, "grad_norm": 0.824481725692749, "learning_rate": 4.495384118493673e-07, "loss": 2.661, "step": 353250 }, { "epoch": 0.0083456, "grad_norm": 0.9314226508140564, "learning_rate": 4.492997122636977e-07, "loss": 2.7174, "step": 353260 }, { "epoch": 0.0083712, "grad_norm": 1.5842686891555786, "learning_rate": 4.490610746117796e-07, "loss": 2.8181, "step": 353270 }, { "epoch": 0.0083968, "grad_norm": 0.8218831419944763, "learning_rate": 4.4882249889516393e-07, "loss": 2.8106, "step": 353280 }, { "epoch": 0.0084224, "grad_norm": 0.8019695281982422, "learning_rate": 4.48583985115395e-07, "loss": 2.4514, "step": 353290 }, { "epoch": 0.008448, "grad_norm": 0.9083213210105896, "learning_rate": 4.483455332740205e-07, "loss": 2.7452, "step": 353300 }, { "epoch": 0.0084736, "grad_norm": 0.8834375739097595, "learning_rate": 4.4810714337258477e-07, "loss": 2.8465, "step": 353310 }, { "epoch": 0.0084992, "grad_norm": 1.00663423538208, "learning_rate": 4.478688154126376e-07, "loss": 2.8453, "step": 353320 }, { "epoch": 0.0085248, "grad_norm": 0.7747719883918762, "learning_rate": 4.476305493957223e-07, "loss": 2.8196, "step": 353330 }, { "epoch": 0.0085504, "grad_norm": 0.9879822731018066, "learning_rate": 4.473923453233853e-07, "loss": 2.7085, "step": 353340 }, { "epoch": 0.008576, "grad_norm": 1.0255697965621948, "learning_rate": 4.4715420319716876e-07, "loss": 2.8182, "step": 353350 }, { "epoch": 0.0086016, "grad_norm": 0.9538187384605408, "learning_rate": 4.469161230186192e-07, "loss": 2.6512, "step": 353360 }, { "epoch": 0.0086272, "grad_norm": 0.8061856031417847, "learning_rate": 4.466781047892788e-07, "loss": 2.9176, "step": 353370 }, { "epoch": 0.0086528, "grad_norm": 0.8472033739089966, "learning_rate": 4.4644014851069175e-07, "loss": 2.716, "step": 353380 }, { "epoch": 0.0086784, "grad_norm": 0.805401623249054, "learning_rate": 4.4620225418440244e-07, "loss": 2.7209, "step": 353390 }, { "epoch": 0.008704, "grad_norm": 0.9551897048950195, "learning_rate": 4.4596442181195186e-07, "loss": 2.5301, "step": 353400 }, { "epoch": 0.0087296, "grad_norm": 0.941936731338501, "learning_rate": 4.4572665139488325e-07, "loss": 2.6944, "step": 353410 }, { "epoch": 0.0087552, "grad_norm": 0.9026955366134644, "learning_rate": 4.454889429347375e-07, "loss": 2.7984, "step": 353420 }, { "epoch": 0.0087808, "grad_norm": 0.8464054465293884, "learning_rate": 4.4525129643305685e-07, "loss": 2.8596, "step": 353430 }, { "epoch": 0.0088064, "grad_norm": 1.4677021503448486, "learning_rate": 4.450137118913822e-07, "loss": 2.8922, "step": 353440 }, { "epoch": 0.008832, "grad_norm": 0.8592096567153931, "learning_rate": 4.447761893112545e-07, "loss": 2.7662, "step": 353450 }, { "epoch": 0.0088576, "grad_norm": 0.915418267250061, "learning_rate": 4.4453872869421267e-07, "loss": 3.1322, "step": 353460 }, { "epoch": 0.0088832, "grad_norm": 0.8889870643615723, "learning_rate": 4.443013300417987e-07, "loss": 2.7796, "step": 353470 }, { "epoch": 0.0089088, "grad_norm": 0.9154715538024902, "learning_rate": 4.440639933555502e-07, "loss": 2.6723, "step": 353480 }, { "epoch": 0.0089344, "grad_norm": 0.8957034349441528, "learning_rate": 4.438267186370071e-07, "loss": 3.0695, "step": 353490 }, { "epoch": 0.00896, "grad_norm": 0.8123064637184143, "learning_rate": 4.4358950588770934e-07, "loss": 2.4532, "step": 353500 }, { "epoch": 0.0089856, "grad_norm": 0.869875431060791, "learning_rate": 4.4335235510919226e-07, "loss": 2.5598, "step": 353510 }, { "epoch": 0.0090112, "grad_norm": 0.8697648644447327, "learning_rate": 4.4311526630299585e-07, "loss": 2.6615, "step": 353520 }, { "epoch": 0.0090368, "grad_norm": 0.8179499506950378, "learning_rate": 4.428782394706577e-07, "loss": 2.9104, "step": 353530 }, { "epoch": 0.0090624, "grad_norm": 1.2111862897872925, "learning_rate": 4.426412746137143e-07, "loss": 2.7416, "step": 353540 }, { "epoch": 0.009088, "grad_norm": 0.8135528564453125, "learning_rate": 4.4240437173370123e-07, "loss": 2.8526, "step": 353550 }, { "epoch": 0.0091136, "grad_norm": 0.8024439811706543, "learning_rate": 4.42167530832156e-07, "loss": 2.8289, "step": 353560 }, { "epoch": 0.0091392, "grad_norm": 0.8743990063667297, "learning_rate": 4.4193075191061416e-07, "loss": 3.0484, "step": 353570 }, { "epoch": 0.0091648, "grad_norm": 0.8659006357192993, "learning_rate": 4.4169403497061114e-07, "loss": 2.8425, "step": 353580 }, { "epoch": 0.0091904, "grad_norm": 0.9963567852973938, "learning_rate": 4.4145738001368234e-07, "loss": 3.1186, "step": 353590 }, { "epoch": 0.009216, "grad_norm": 0.8859434127807617, "learning_rate": 4.4122078704136096e-07, "loss": 2.8802, "step": 353600 }, { "epoch": 0.0092416, "grad_norm": 0.9048101902008057, "learning_rate": 4.409842560551847e-07, "loss": 2.7883, "step": 353610 }, { "epoch": 0.0092672, "grad_norm": 0.9211331009864807, "learning_rate": 4.4074778705668344e-07, "loss": 2.7499, "step": 353620 }, { "epoch": 0.0092928, "grad_norm": 0.8411283493041992, "learning_rate": 4.405113800473926e-07, "loss": 2.7287, "step": 353630 }, { "epoch": 0.0093184, "grad_norm": 0.8099237680435181, "learning_rate": 4.402750350288443e-07, "loss": 2.8323, "step": 353640 }, { "epoch": 0.009344, "grad_norm": 0.7809491157531738, "learning_rate": 4.4003875200257063e-07, "loss": 2.8405, "step": 353650 }, { "epoch": 0.0093696, "grad_norm": 0.9532079100608826, "learning_rate": 4.3980253097010596e-07, "loss": 2.9189, "step": 353660 }, { "epoch": 0.0093952, "grad_norm": 0.7892253398895264, "learning_rate": 4.3956637193298236e-07, "loss": 2.666, "step": 353670 }, { "epoch": 0.0094208, "grad_norm": 0.9975671172142029, "learning_rate": 4.393302748927286e-07, "loss": 2.9973, "step": 353680 }, { "epoch": 0.0094464, "grad_norm": 0.8946375250816345, "learning_rate": 4.3909423985087797e-07, "loss": 2.821, "step": 353690 }, { "epoch": 0.009472, "grad_norm": 0.8976513743400574, "learning_rate": 4.3885826680895914e-07, "loss": 3.041, "step": 353700 }, { "epoch": 0.0094976, "grad_norm": 0.9302221536636353, "learning_rate": 4.386223557685043e-07, "loss": 2.7339, "step": 353710 }, { "epoch": 0.0095232, "grad_norm": 0.8728201389312744, "learning_rate": 4.3838650673104224e-07, "loss": 2.8665, "step": 353720 }, { "epoch": 0.0095488, "grad_norm": 0.8450786471366882, "learning_rate": 4.3815071969810164e-07, "loss": 2.9806, "step": 353730 }, { "epoch": 0.0095744, "grad_norm": 0.808499813079834, "learning_rate": 4.379149946712147e-07, "loss": 2.9907, "step": 353740 }, { "epoch": 0.0096, "grad_norm": 0.864996075630188, "learning_rate": 4.3767933165190567e-07, "loss": 2.9109, "step": 353750 }, { "epoch": 0.0096256, "grad_norm": 0.8834238648414612, "learning_rate": 4.3744373064170566e-07, "loss": 2.7142, "step": 353760 }, { "epoch": 0.0096512, "grad_norm": 0.8304572701454163, "learning_rate": 4.372081916421411e-07, "loss": 2.7633, "step": 353770 }, { "epoch": 0.0096768, "grad_norm": 3.8461835384368896, "learning_rate": 4.3697271465473977e-07, "loss": 2.8389, "step": 353780 }, { "epoch": 0.0097024, "grad_norm": 1.5028659105300903, "learning_rate": 4.367372996810293e-07, "loss": 2.7922, "step": 353790 }, { "epoch": 0.009728, "grad_norm": 0.8019057512283325, "learning_rate": 4.365019467225351e-07, "loss": 2.8645, "step": 353800 }, { "epoch": 0.0097536, "grad_norm": 0.8983151912689209, "learning_rate": 4.36266655780786e-07, "loss": 2.6935, "step": 353810 }, { "epoch": 0.0097792, "grad_norm": 0.8318006992340088, "learning_rate": 4.3603142685730513e-07, "loss": 2.716, "step": 353820 }, { "epoch": 0.0098048, "grad_norm": 0.9201924800872803, "learning_rate": 4.3579625995361696e-07, "loss": 2.7067, "step": 353830 }, { "epoch": 0.0098304, "grad_norm": 0.8168751001358032, "learning_rate": 4.355611550712491e-07, "loss": 2.8625, "step": 353840 }, { "epoch": 0.009856, "grad_norm": 0.8967071175575256, "learning_rate": 4.353261122117247e-07, "loss": 2.9394, "step": 353850 }, { "epoch": 0.0098816, "grad_norm": 0.7765905857086182, "learning_rate": 4.3509113137656934e-07, "loss": 2.9907, "step": 353860 }, { "epoch": 0.0099072, "grad_norm": 0.9274066090583801, "learning_rate": 4.3485621256730724e-07, "loss": 2.7733, "step": 353870 }, { "epoch": 0.0099328, "grad_norm": 0.820851743221283, "learning_rate": 4.346213557854595e-07, "loss": 2.9957, "step": 353880 }, { "epoch": 0.0099584, "grad_norm": 0.8289868235588074, "learning_rate": 4.343865610325493e-07, "loss": 3.0683, "step": 353890 }, { "epoch": 0.009984, "grad_norm": 0.8444427847862244, "learning_rate": 4.3415182831009984e-07, "loss": 2.8696, "step": 353900 }, { "epoch": 0.0100096, "grad_norm": 0.7696180939674377, "learning_rate": 4.339171576196344e-07, "loss": 2.8568, "step": 353910 }, { "epoch": 0.0100352, "grad_norm": 1.214911699295044, "learning_rate": 4.3368254896267283e-07, "loss": 2.8205, "step": 353920 }, { "epoch": 0.0100608, "grad_norm": 0.9487895965576172, "learning_rate": 4.3344800234073834e-07, "loss": 2.8441, "step": 353930 }, { "epoch": 0.0100864, "grad_norm": 0.8488512635231018, "learning_rate": 4.33213517755352e-07, "loss": 2.9418, "step": 353940 }, { "epoch": 0.010112, "grad_norm": 0.8952998518943787, "learning_rate": 4.329790952080326e-07, "loss": 2.7412, "step": 353950 }, { "epoch": 0.0101376, "grad_norm": 0.8203372955322266, "learning_rate": 4.327447347003011e-07, "loss": 2.7875, "step": 353960 }, { "epoch": 0.0101632, "grad_norm": 0.8081491589546204, "learning_rate": 4.3251043623367627e-07, "loss": 2.7803, "step": 353970 }, { "epoch": 0.0101888, "grad_norm": 0.7807128429412842, "learning_rate": 4.3227619980967914e-07, "loss": 2.6756, "step": 353980 }, { "epoch": 0.0102144, "grad_norm": 0.8699051141738892, "learning_rate": 4.3204202542982855e-07, "loss": 2.825, "step": 353990 }, { "epoch": 0.01024, "grad_norm": 0.8232222199440002, "learning_rate": 4.318079130956421e-07, "loss": 3.1038, "step": 354000 }, { "epoch": 0.0102656, "grad_norm": 0.8952164053916931, "learning_rate": 4.3157386280863856e-07, "loss": 2.9112, "step": 354010 }, { "epoch": 0.0102912, "grad_norm": 0.8206319212913513, "learning_rate": 4.313398745703357e-07, "loss": 2.7233, "step": 354020 }, { "epoch": 0.0103168, "grad_norm": 0.8980787992477417, "learning_rate": 4.3110594838224995e-07, "loss": 2.7746, "step": 354030 }, { "epoch": 0.0103424, "grad_norm": 0.9256093502044678, "learning_rate": 4.308720842459002e-07, "loss": 2.9113, "step": 354040 }, { "epoch": 0.010368, "grad_norm": 0.8268648386001587, "learning_rate": 4.306382821628008e-07, "loss": 2.9221, "step": 354050 }, { "epoch": 0.0103936, "grad_norm": 0.8502691984176636, "learning_rate": 4.3040454213446934e-07, "loss": 2.7124, "step": 354060 }, { "epoch": 0.0104192, "grad_norm": 1.3075745105743408, "learning_rate": 4.301708641624225e-07, "loss": 2.7579, "step": 354070 }, { "epoch": 0.0104448, "grad_norm": 0.8475635051727295, "learning_rate": 4.299372482481734e-07, "loss": 2.8542, "step": 354080 }, { "epoch": 0.0104704, "grad_norm": 0.827053427696228, "learning_rate": 4.2970369439323647e-07, "loss": 2.7903, "step": 354090 }, { "epoch": 0.010496, "grad_norm": 0.7894940972328186, "learning_rate": 4.294702025991282e-07, "loss": 2.5617, "step": 354100 }, { "epoch": 0.0105216, "grad_norm": 0.7802181839942932, "learning_rate": 4.29236772867363e-07, "loss": 2.8413, "step": 354110 }, { "epoch": 0.0105472, "grad_norm": 0.8542219996452332, "learning_rate": 4.290034051994529e-07, "loss": 2.6398, "step": 354120 }, { "epoch": 0.0105728, "grad_norm": 0.8651036620140076, "learning_rate": 4.287700995969124e-07, "loss": 2.642, "step": 354130 }, { "epoch": 0.0105984, "grad_norm": 1.0459421873092651, "learning_rate": 4.285368560612546e-07, "loss": 2.714, "step": 354140 }, { "epoch": 0.010624, "grad_norm": 0.8340352177619934, "learning_rate": 4.283036745939917e-07, "loss": 2.7492, "step": 354150 }, { "epoch": 0.0106496, "grad_norm": 0.8431504368782043, "learning_rate": 4.2807055519663467e-07, "loss": 3.0292, "step": 354160 }, { "epoch": 0.0106752, "grad_norm": 1.5458333492279053, "learning_rate": 4.2783749787069563e-07, "loss": 2.7765, "step": 354170 }, { "epoch": 0.0107008, "grad_norm": 0.8480717539787292, "learning_rate": 4.276045026176856e-07, "loss": 2.7372, "step": 354180 }, { "epoch": 0.0107264, "grad_norm": 0.8667972087860107, "learning_rate": 4.2737156943911895e-07, "loss": 2.6707, "step": 354190 }, { "epoch": 0.010752, "grad_norm": 0.8947992920875549, "learning_rate": 4.2713869833650225e-07, "loss": 2.8469, "step": 354200 }, { "epoch": 0.0107776, "grad_norm": 0.8359084129333496, "learning_rate": 4.2690588931134757e-07, "loss": 2.8141, "step": 354210 }, { "epoch": 0.0108032, "grad_norm": 0.8469275236129761, "learning_rate": 4.2667314236516375e-07, "loss": 2.9983, "step": 354220 }, { "epoch": 0.0108288, "grad_norm": 1.0026065111160278, "learning_rate": 4.264404574994596e-07, "loss": 2.7349, "step": 354230 }, { "epoch": 0.0108544, "grad_norm": 0.7863929867744446, "learning_rate": 4.26207834715745e-07, "loss": 2.9736, "step": 354240 }, { "epoch": 0.01088, "grad_norm": 0.7955739498138428, "learning_rate": 4.259752740155287e-07, "loss": 2.8874, "step": 354250 }, { "epoch": 0.0109056, "grad_norm": 0.9256283640861511, "learning_rate": 4.2574277540031847e-07, "loss": 2.8761, "step": 354260 }, { "epoch": 0.0109312, "grad_norm": 0.9565203785896301, "learning_rate": 4.2551033887162195e-07, "loss": 2.9074, "step": 354270 }, { "epoch": 0.0109568, "grad_norm": 1.0598236322402954, "learning_rate": 4.2527796443094573e-07, "loss": 2.7103, "step": 354280 }, { "epoch": 0.0109824, "grad_norm": 0.8972530961036682, "learning_rate": 4.250456520797963e-07, "loss": 3.0039, "step": 354290 }, { "epoch": 0.011008, "grad_norm": 0.8365715146064758, "learning_rate": 4.248134018196826e-07, "loss": 2.7648, "step": 354300 }, { "epoch": 0.0110336, "grad_norm": 1.1858737468719482, "learning_rate": 4.245812136521088e-07, "loss": 2.7936, "step": 354310 }, { "epoch": 0.0110592, "grad_norm": 0.8865828514099121, "learning_rate": 4.243490875785805e-07, "loss": 2.8723, "step": 354320 }, { "epoch": 0.0110848, "grad_norm": 0.9772853851318359, "learning_rate": 4.2411702360060313e-07, "loss": 2.88, "step": 354330 }, { "epoch": 0.0111104, "grad_norm": 0.869860053062439, "learning_rate": 4.2388502171968436e-07, "loss": 2.7203, "step": 354340 }, { "epoch": 0.011136, "grad_norm": 0.859539270401001, "learning_rate": 4.23653081937323e-07, "loss": 2.9074, "step": 354350 }, { "epoch": 0.0111616, "grad_norm": 0.863962709903717, "learning_rate": 4.2342120425502673e-07, "loss": 2.7775, "step": 354360 }, { "epoch": 0.0111872, "grad_norm": 0.9487706422805786, "learning_rate": 4.231893886742999e-07, "loss": 3.0032, "step": 354370 }, { "epoch": 0.0112128, "grad_norm": 0.8652559518814087, "learning_rate": 4.229576351966447e-07, "loss": 2.8912, "step": 354380 }, { "epoch": 0.0112384, "grad_norm": 0.9212033748626709, "learning_rate": 4.2272594382356534e-07, "loss": 3.0447, "step": 354390 }, { "epoch": 0.011264, "grad_norm": 0.8509740829467773, "learning_rate": 4.224943145565608e-07, "loss": 3.0093, "step": 354400 }, { "epoch": 0.0112896, "grad_norm": 1.4777424335479736, "learning_rate": 4.2226274739713524e-07, "loss": 2.8034, "step": 354410 }, { "epoch": 0.0113152, "grad_norm": 0.9829877018928528, "learning_rate": 4.2203124234679094e-07, "loss": 3.0674, "step": 354420 }, { "epoch": 0.0113408, "grad_norm": 0.769919216632843, "learning_rate": 4.2179979940702886e-07, "loss": 2.9136, "step": 354430 }, { "epoch": 0.0113664, "grad_norm": 0.8235995173454285, "learning_rate": 4.2156841857934895e-07, "loss": 2.7928, "step": 354440 }, { "epoch": 0.011392, "grad_norm": 0.7845258116722107, "learning_rate": 4.213370998652522e-07, "loss": 2.6795, "step": 354450 }, { "epoch": 0.0114176, "grad_norm": 0.8125396370887756, "learning_rate": 4.211058432662396e-07, "loss": 2.6749, "step": 354460 }, { "epoch": 0.0114432, "grad_norm": 0.8756415247917175, "learning_rate": 4.208746487838089e-07, "loss": 3.0103, "step": 354470 }, { "epoch": 0.0114688, "grad_norm": 1.1888266801834106, "learning_rate": 4.2064351641946e-07, "loss": 3.0625, "step": 354480 }, { "epoch": 0.0114944, "grad_norm": 0.8316223621368408, "learning_rate": 4.204124461746928e-07, "loss": 2.8534, "step": 354490 }, { "epoch": 0.01152, "grad_norm": 0.7820842266082764, "learning_rate": 4.2018143805100496e-07, "loss": 2.9317, "step": 354500 }, { "epoch": 0.0115456, "grad_norm": 0.8955219984054565, "learning_rate": 4.1995049204989314e-07, "loss": 2.6261, "step": 354510 }, { "epoch": 0.0115712, "grad_norm": 0.8258287310600281, "learning_rate": 4.197196081728572e-07, "loss": 2.7512, "step": 354520 }, { "epoch": 0.0115968, "grad_norm": 0.8553439378738403, "learning_rate": 4.194887864213937e-07, "loss": 2.7396, "step": 354530 }, { "epoch": 0.0116224, "grad_norm": 0.8644465804100037, "learning_rate": 4.1925802679699924e-07, "loss": 2.8366, "step": 354540 }, { "epoch": 0.011648, "grad_norm": 0.8589099049568176, "learning_rate": 4.190273293011693e-07, "loss": 3.1019, "step": 354550 }, { "epoch": 0.0116736, "grad_norm": 0.8268159627914429, "learning_rate": 4.1879669393540155e-07, "loss": 2.8185, "step": 354560 }, { "epoch": 0.0116992, "grad_norm": 0.8546352386474609, "learning_rate": 4.185661207011904e-07, "loss": 2.7378, "step": 354570 }, { "epoch": 0.0117248, "grad_norm": 0.7821188569068909, "learning_rate": 4.183356096000324e-07, "loss": 2.8454, "step": 354580 }, { "epoch": 0.0117504, "grad_norm": 0.8923285603523254, "learning_rate": 4.181051606334219e-07, "loss": 2.7237, "step": 354590 }, { "epoch": 0.011776, "grad_norm": 1.0264917612075806, "learning_rate": 4.178747738028521e-07, "loss": 2.8422, "step": 354600 }, { "epoch": 0.0118016, "grad_norm": 0.8847548365592957, "learning_rate": 4.176444491098175e-07, "loss": 2.8519, "step": 354610 }, { "epoch": 0.0118272, "grad_norm": 0.8604389429092407, "learning_rate": 4.1741418655581235e-07, "loss": 2.7179, "step": 354620 }, { "epoch": 0.0118528, "grad_norm": 0.795234739780426, "learning_rate": 4.171839861423288e-07, "loss": 2.789, "step": 354630 }, { "epoch": 0.0118784, "grad_norm": 0.8273658752441406, "learning_rate": 4.169538478708601e-07, "loss": 2.8397, "step": 354640 }, { "epoch": 0.011904, "grad_norm": 0.8877252340316772, "learning_rate": 4.1672377174289957e-07, "loss": 2.9143, "step": 354650 }, { "epoch": 0.0119296, "grad_norm": 0.8982588648796082, "learning_rate": 4.164937577599404e-07, "loss": 2.9195, "step": 354660 }, { "epoch": 0.0119552, "grad_norm": 0.8596769571304321, "learning_rate": 4.1626380592346913e-07, "loss": 2.8323, "step": 354670 }, { "epoch": 0.0119808, "grad_norm": 1.3275582790374756, "learning_rate": 4.1603391623498136e-07, "loss": 2.8544, "step": 354680 }, { "epoch": 0.0120064, "grad_norm": 0.8935136198997498, "learning_rate": 4.158040886959658e-07, "loss": 2.3379, "step": 354690 }, { "epoch": 0.012032, "grad_norm": 0.9098778963088989, "learning_rate": 4.155743233079124e-07, "loss": 2.5614, "step": 354700 }, { "epoch": 0.0120576, "grad_norm": 0.8435202240943909, "learning_rate": 4.1534462007231437e-07, "loss": 2.822, "step": 354710 }, { "epoch": 0.0120832, "grad_norm": 1.0033535957336426, "learning_rate": 4.151149789906594e-07, "loss": 2.8807, "step": 354720 }, { "epoch": 0.0121088, "grad_norm": 0.7972124218940735, "learning_rate": 4.1488540006443535e-07, "loss": 2.6668, "step": 354730 }, { "epoch": 0.0121344, "grad_norm": 0.8811987042427063, "learning_rate": 4.1465588329513304e-07, "loss": 2.8924, "step": 354740 }, { "epoch": 0.01216, "grad_norm": 0.770729660987854, "learning_rate": 4.144264286842392e-07, "loss": 2.7074, "step": 354750 }, { "epoch": 0.0121856, "grad_norm": 0.7653173208236694, "learning_rate": 4.1419703623324147e-07, "loss": 2.934, "step": 354760 }, { "epoch": 0.0122112, "grad_norm": 0.9728206396102905, "learning_rate": 4.139677059436298e-07, "loss": 2.8409, "step": 354770 }, { "epoch": 0.0122368, "grad_norm": 0.8581702709197998, "learning_rate": 4.137384378168896e-07, "loss": 2.9158, "step": 354780 }, { "epoch": 0.0122624, "grad_norm": 0.7798588275909424, "learning_rate": 4.1350923185450867e-07, "loss": 2.8106, "step": 354790 }, { "epoch": 0.012288, "grad_norm": 1.2570921182632446, "learning_rate": 4.1328008805797126e-07, "loss": 3.0068, "step": 354800 }, { "epoch": 0.0123136, "grad_norm": 0.8473051190376282, "learning_rate": 4.1305100642876517e-07, "loss": 2.843, "step": 354810 }, { "epoch": 0.0123392, "grad_norm": 0.815899133682251, "learning_rate": 4.128219869683747e-07, "loss": 2.7885, "step": 354820 }, { "epoch": 0.0123648, "grad_norm": 0.8098056316375732, "learning_rate": 4.125930296782865e-07, "loss": 2.8746, "step": 354830 }, { "epoch": 0.0123904, "grad_norm": 0.8745241761207581, "learning_rate": 4.123641345599838e-07, "loss": 2.9308, "step": 354840 }, { "epoch": 0.012416, "grad_norm": 0.9944790601730347, "learning_rate": 4.121353016149521e-07, "loss": 2.8844, "step": 354850 }, { "epoch": 0.0124416, "grad_norm": 0.8004209995269775, "learning_rate": 4.119065308446757e-07, "loss": 2.8885, "step": 354860 }, { "epoch": 0.0124672, "grad_norm": 0.8323415517807007, "learning_rate": 4.116778222506357e-07, "loss": 2.8906, "step": 354870 }, { "epoch": 0.0124928, "grad_norm": 0.8400901556015015, "learning_rate": 4.114491758343164e-07, "loss": 2.7997, "step": 354880 }, { "epoch": 0.0125184, "grad_norm": 0.9868452548980713, "learning_rate": 4.112205915972023e-07, "loss": 2.9309, "step": 354890 }, { "epoch": 0.012544, "grad_norm": 0.9712984561920166, "learning_rate": 4.109920695407732e-07, "loss": 3.0332, "step": 354900 }, { "epoch": 0.0125696, "grad_norm": 0.8308760523796082, "learning_rate": 4.107636096665124e-07, "loss": 2.733, "step": 354910 }, { "epoch": 0.0125952, "grad_norm": 0.8844612836837769, "learning_rate": 4.105352119759032e-07, "loss": 2.9294, "step": 354920 }, { "epoch": 0.0126208, "grad_norm": 0.9895167350769043, "learning_rate": 4.103068764704221e-07, "loss": 3.2197, "step": 354930 }, { "epoch": 0.0126464, "grad_norm": 0.7868670225143433, "learning_rate": 4.1007860315155245e-07, "loss": 2.6052, "step": 354940 }, { "epoch": 0.012672, "grad_norm": 0.9258832335472107, "learning_rate": 4.0985039202077526e-07, "loss": 2.8638, "step": 354950 }, { "epoch": 0.0126976, "grad_norm": 0.8918982148170471, "learning_rate": 4.0962224307956823e-07, "loss": 2.8173, "step": 354960 }, { "epoch": 0.0127232, "grad_norm": 0.8718123435974121, "learning_rate": 4.0939415632941347e-07, "loss": 2.8894, "step": 354970 }, { "epoch": 0.0127488, "grad_norm": 0.9461256265640259, "learning_rate": 4.0916613177178767e-07, "loss": 3.0879, "step": 354980 }, { "epoch": 0.0127744, "grad_norm": 0.8698354363441467, "learning_rate": 4.089381694081729e-07, "loss": 2.9626, "step": 354990 }, { "epoch": 0.0128, "grad_norm": 0.8551095724105835, "learning_rate": 4.0871026924004245e-07, "loss": 3.1695, "step": 355000 }, { "epoch": 0.0128256, "grad_norm": 0.945955753326416, "learning_rate": 4.084824312688773e-07, "loss": 2.9047, "step": 355010 }, { "epoch": 0.0128512, "grad_norm": 0.9368022084236145, "learning_rate": 4.082546554961553e-07, "loss": 2.8669, "step": 355020 }, { "epoch": 0.0128768, "grad_norm": 0.9034727215766907, "learning_rate": 4.080269419233518e-07, "loss": 2.8921, "step": 355030 }, { "epoch": 0.0129024, "grad_norm": 0.8780671954154968, "learning_rate": 4.0779929055194457e-07, "loss": 2.8371, "step": 355040 }, { "epoch": 0.012928, "grad_norm": 0.7149859666824341, "learning_rate": 4.0757170138341017e-07, "loss": 2.6176, "step": 355050 }, { "epoch": 0.0129536, "grad_norm": 0.8367487192153931, "learning_rate": 4.07344174419223e-07, "loss": 3.0463, "step": 355060 }, { "epoch": 0.0129792, "grad_norm": 0.9778497219085693, "learning_rate": 4.071167096608608e-07, "loss": 2.8209, "step": 355070 }, { "epoch": 0.0130048, "grad_norm": 0.848031759262085, "learning_rate": 4.068893071097957e-07, "loss": 2.8692, "step": 355080 }, { "epoch": 0.0130304, "grad_norm": 0.9913409948348999, "learning_rate": 4.0666196676750534e-07, "loss": 2.9801, "step": 355090 }, { "epoch": 0.013056, "grad_norm": 0.7992827892303467, "learning_rate": 4.064346886354609e-07, "loss": 2.9323, "step": 355100 }, { "epoch": 0.0130816, "grad_norm": 0.9819557666778564, "learning_rate": 4.062074727151388e-07, "loss": 2.8788, "step": 355110 }, { "epoch": 0.0131072, "grad_norm": 0.8426122665405273, "learning_rate": 4.059803190080125e-07, "loss": 2.6768, "step": 355120 }, { "epoch": 0.0131328, "grad_norm": 0.9032624363899231, "learning_rate": 4.057532275155529e-07, "loss": 2.9713, "step": 355130 }, { "epoch": 0.0131584, "grad_norm": 1.0125577449798584, "learning_rate": 4.0552619823923443e-07, "loss": 2.8225, "step": 355140 }, { "epoch": 0.013184, "grad_norm": 0.917885959148407, "learning_rate": 4.0529923118052816e-07, "loss": 2.8918, "step": 355150 }, { "epoch": 0.0132096, "grad_norm": 1.6768174171447754, "learning_rate": 4.0507232634090623e-07, "loss": 3.0012, "step": 355160 }, { "epoch": 0.0132352, "grad_norm": 0.9562260508537292, "learning_rate": 4.0484548372184074e-07, "loss": 3.0284, "step": 355170 }, { "epoch": 0.0132608, "grad_norm": 0.8507057428359985, "learning_rate": 4.046187033248017e-07, "loss": 2.4052, "step": 355180 }, { "epoch": 0.0132864, "grad_norm": 0.7802550792694092, "learning_rate": 4.043919851512612e-07, "loss": 2.797, "step": 355190 }, { "epoch": 0.013312, "grad_norm": 1.1194435358047485, "learning_rate": 4.0416532920268704e-07, "loss": 2.8462, "step": 355200 }, { "epoch": 0.0133376, "grad_norm": 0.7710137367248535, "learning_rate": 4.039387354805513e-07, "loss": 2.8887, "step": 355210 }, { "epoch": 0.0133632, "grad_norm": 0.933664858341217, "learning_rate": 4.037122039863217e-07, "loss": 2.9691, "step": 355220 }, { "epoch": 0.0133888, "grad_norm": 1.0360345840454102, "learning_rate": 4.0348573472146826e-07, "loss": 2.7857, "step": 355230 }, { "epoch": 0.0134144, "grad_norm": 1.0022871494293213, "learning_rate": 4.032593276874597e-07, "loss": 2.855, "step": 355240 }, { "epoch": 0.01344, "grad_norm": 0.7585753798484802, "learning_rate": 4.0303298288576485e-07, "loss": 2.6578, "step": 355250 }, { "epoch": 0.0134656, "grad_norm": 0.8301810026168823, "learning_rate": 4.028067003178493e-07, "loss": 2.5848, "step": 355260 }, { "epoch": 0.0134912, "grad_norm": 0.8470381498336792, "learning_rate": 4.025804799851829e-07, "loss": 3.0315, "step": 355270 }, { "epoch": 0.0135168, "grad_norm": 0.7849418520927429, "learning_rate": 4.0235432188923007e-07, "loss": 2.696, "step": 355280 }, { "epoch": 0.0135424, "grad_norm": 1.4868898391723633, "learning_rate": 4.021282260314585e-07, "loss": 2.8679, "step": 355290 }, { "epoch": 0.013568, "grad_norm": 0.8211506009101868, "learning_rate": 4.0190219241333596e-07, "loss": 2.7959, "step": 355300 }, { "epoch": 0.0135936, "grad_norm": 0.7827345728874207, "learning_rate": 4.0167622103632563e-07, "loss": 2.879, "step": 355310 }, { "epoch": 0.0136192, "grad_norm": 1.203957200050354, "learning_rate": 4.0145031190189644e-07, "loss": 2.7757, "step": 355320 }, { "epoch": 0.0136448, "grad_norm": 0.8328819274902344, "learning_rate": 4.0122446501150825e-07, "loss": 2.7441, "step": 355330 }, { "epoch": 0.0136704, "grad_norm": 0.9186167120933533, "learning_rate": 4.0099868036662993e-07, "loss": 2.8542, "step": 355340 }, { "epoch": 0.013696, "grad_norm": 0.8133915066719055, "learning_rate": 4.007729579687225e-07, "loss": 3.0537, "step": 355350 }, { "epoch": 0.0137216, "grad_norm": 0.9016165733337402, "learning_rate": 4.005472978192526e-07, "loss": 2.6882, "step": 355360 }, { "epoch": 0.0137472, "grad_norm": 1.4540292024612427, "learning_rate": 4.0032169991968127e-07, "loss": 2.9394, "step": 355370 }, { "epoch": 0.0137728, "grad_norm": 0.822656512260437, "learning_rate": 4.000961642714729e-07, "loss": 2.9371, "step": 355380 }, { "epoch": 0.0137984, "grad_norm": 0.8606875538825989, "learning_rate": 3.9987069087608963e-07, "loss": 2.8421, "step": 355390 }, { "epoch": 0.013824, "grad_norm": 0.9418385028839111, "learning_rate": 3.9964527973499366e-07, "loss": 2.624, "step": 355400 }, { "epoch": 0.0138496, "grad_norm": 0.8650961518287659, "learning_rate": 3.99419930849646e-07, "loss": 2.7749, "step": 355410 }, { "epoch": 0.0138752, "grad_norm": 0.8255681991577148, "learning_rate": 3.991946442215089e-07, "loss": 2.8011, "step": 355420 }, { "epoch": 0.0139008, "grad_norm": 0.8302972316741943, "learning_rate": 3.989694198520433e-07, "loss": 2.881, "step": 355430 }, { "epoch": 0.0139264, "grad_norm": 0.7726385593414307, "learning_rate": 3.987442577427092e-07, "loss": 2.6392, "step": 355440 }, { "epoch": 0.013952, "grad_norm": 0.7724590301513672, "learning_rate": 3.9851915789496766e-07, "loss": 2.7997, "step": 355450 }, { "epoch": 0.0139776, "grad_norm": 0.8988723754882812, "learning_rate": 3.982941203102764e-07, "loss": 2.8778, "step": 355460 }, { "epoch": 0.0140032, "grad_norm": 0.8965915441513062, "learning_rate": 3.980691449900964e-07, "loss": 2.7372, "step": 355470 }, { "epoch": 0.0140288, "grad_norm": 0.859435498714447, "learning_rate": 3.978442319358855e-07, "loss": 2.8483, "step": 355480 }, { "epoch": 0.0140544, "grad_norm": 0.8992827534675598, "learning_rate": 3.976193811491036e-07, "loss": 2.612, "step": 355490 }, { "epoch": 0.01408, "grad_norm": 0.7844536304473877, "learning_rate": 3.973945926312073e-07, "loss": 2.6378, "step": 355500 }, { "epoch": 0.0141056, "grad_norm": 1.372352957725525, "learning_rate": 3.971698663836554e-07, "loss": 2.7349, "step": 355510 }, { "epoch": 0.0141312, "grad_norm": 0.8323403000831604, "learning_rate": 3.969452024079057e-07, "loss": 2.8528, "step": 355520 }, { "epoch": 0.0141568, "grad_norm": 0.8139328360557556, "learning_rate": 3.967206007054136e-07, "loss": 2.8219, "step": 355530 }, { "epoch": 0.0141824, "grad_norm": 0.8528648614883423, "learning_rate": 3.9649606127763584e-07, "loss": 2.8302, "step": 355540 }, { "epoch": 0.014208, "grad_norm": 0.851969838142395, "learning_rate": 3.962715841260278e-07, "loss": 2.9575, "step": 355550 }, { "epoch": 0.0142336, "grad_norm": 0.8973116278648376, "learning_rate": 3.9604716925204736e-07, "loss": 2.8282, "step": 355560 }, { "epoch": 0.0142592, "grad_norm": 0.8616177439689636, "learning_rate": 3.958228166571487e-07, "loss": 3.0018, "step": 355570 }, { "epoch": 0.0142848, "grad_norm": 0.8054623603820801, "learning_rate": 3.9559852634278637e-07, "loss": 2.9113, "step": 355580 }, { "epoch": 0.0143104, "grad_norm": 0.8280615210533142, "learning_rate": 3.9537429831041476e-07, "loss": 2.9095, "step": 355590 }, { "epoch": 0.014336, "grad_norm": 0.9444705247879028, "learning_rate": 3.951501325614882e-07, "loss": 2.8136, "step": 355600 }, { "epoch": 0.0143616, "grad_norm": 0.8375621438026428, "learning_rate": 3.9492602909745993e-07, "loss": 2.9777, "step": 355610 }, { "epoch": 0.0143872, "grad_norm": 1.365175724029541, "learning_rate": 3.947019879197844e-07, "loss": 2.9086, "step": 355620 }, { "epoch": 0.0144128, "grad_norm": 0.8313985466957092, "learning_rate": 3.9447800902991387e-07, "loss": 2.8702, "step": 355630 }, { "epoch": 0.0144384, "grad_norm": 0.8190113306045532, "learning_rate": 3.942540924293003e-07, "loss": 2.7972, "step": 355640 }, { "epoch": 0.014464, "grad_norm": 0.8103138208389282, "learning_rate": 3.9403023811939724e-07, "loss": 2.8207, "step": 355650 }, { "epoch": 0.0144896, "grad_norm": 0.7820383906364441, "learning_rate": 3.938064461016533e-07, "loss": 2.6776, "step": 355660 }, { "epoch": 0.0145152, "grad_norm": 0.822343111038208, "learning_rate": 3.9358271637752297e-07, "loss": 2.8872, "step": 355670 }, { "epoch": 0.0145408, "grad_norm": 0.9883444905281067, "learning_rate": 3.93359048948454e-07, "loss": 2.8254, "step": 355680 }, { "epoch": 0.0145664, "grad_norm": 0.9315394759178162, "learning_rate": 3.9313544381589964e-07, "loss": 2.9648, "step": 355690 }, { "epoch": 0.014592, "grad_norm": 0.7809758186340332, "learning_rate": 3.9291190098130874e-07, "loss": 2.7123, "step": 355700 }, { "epoch": 0.0146176, "grad_norm": 0.8787478804588318, "learning_rate": 3.9268842044613007e-07, "loss": 2.9452, "step": 355710 }, { "epoch": 0.0146432, "grad_norm": 0.8890570402145386, "learning_rate": 3.9246500221181594e-07, "loss": 2.7885, "step": 355720 }, { "epoch": 0.0146688, "grad_norm": 1.3968117237091064, "learning_rate": 3.922416462798107e-07, "loss": 2.7659, "step": 355730 }, { "epoch": 0.0146944, "grad_norm": 0.9540555477142334, "learning_rate": 3.920183526515642e-07, "loss": 2.7845, "step": 355740 }, { "epoch": 0.01472, "grad_norm": 0.8559561371803284, "learning_rate": 3.917951213285265e-07, "loss": 2.7031, "step": 355750 }, { "epoch": 0.0147456, "grad_norm": 0.8642202615737915, "learning_rate": 3.9157195231214307e-07, "loss": 2.9067, "step": 355760 }, { "epoch": 0.0147712, "grad_norm": 1.125803828239441, "learning_rate": 3.913488456038639e-07, "loss": 2.9722, "step": 355770 }, { "epoch": 0.0147968, "grad_norm": 0.8624262809753418, "learning_rate": 3.911258012051322e-07, "loss": 2.9586, "step": 355780 }, { "epoch": 0.0148224, "grad_norm": 0.8337966203689575, "learning_rate": 3.909028191173969e-07, "loss": 2.7765, "step": 355790 }, { "epoch": 0.014848, "grad_norm": 0.8465895652770996, "learning_rate": 3.9067989934210126e-07, "loss": 2.889, "step": 355800 }, { "epoch": 0.0148736, "grad_norm": 0.8398146033287048, "learning_rate": 3.904570418806941e-07, "loss": 2.8276, "step": 355810 }, { "epoch": 0.0148992, "grad_norm": 0.9140965938568115, "learning_rate": 3.902342467346187e-07, "loss": 3.1026, "step": 355820 }, { "epoch": 0.0149248, "grad_norm": 0.9359618425369263, "learning_rate": 3.9001151390532064e-07, "loss": 2.8824, "step": 355830 }, { "epoch": 0.0149504, "grad_norm": 0.9133939146995544, "learning_rate": 3.8978884339424316e-07, "loss": 2.6915, "step": 355840 }, { "epoch": 0.014976, "grad_norm": 0.8592029809951782, "learning_rate": 3.8956623520283285e-07, "loss": 2.8427, "step": 355850 }, { "epoch": 0.0150016, "grad_norm": 1.0976872444152832, "learning_rate": 3.8934368933252974e-07, "loss": 2.783, "step": 355860 }, { "epoch": 0.0150272, "grad_norm": 0.7826147079467773, "learning_rate": 3.891212057847782e-07, "loss": 2.8303, "step": 355870 }, { "epoch": 0.0150528, "grad_norm": 0.81141597032547, "learning_rate": 3.888987845610215e-07, "loss": 2.8684, "step": 355880 }, { "epoch": 0.0150784, "grad_norm": 0.9004056453704834, "learning_rate": 3.886764256627018e-07, "loss": 2.9634, "step": 355890 }, { "epoch": 0.015104, "grad_norm": 0.8221061825752258, "learning_rate": 3.8845412909126246e-07, "loss": 2.9512, "step": 355900 }, { "epoch": 0.0151296, "grad_norm": 0.7497913241386414, "learning_rate": 3.8823189484814226e-07, "loss": 2.8585, "step": 355910 }, { "epoch": 0.0151552, "grad_norm": 0.7946900129318237, "learning_rate": 3.8800972293478343e-07, "loss": 2.9742, "step": 355920 }, { "epoch": 0.0151808, "grad_norm": 0.9426855444908142, "learning_rate": 3.8778761335262816e-07, "loss": 2.9404, "step": 355930 }, { "epoch": 0.0152064, "grad_norm": 0.7761751413345337, "learning_rate": 3.8756556610311523e-07, "loss": 2.9174, "step": 355940 }, { "epoch": 0.015232, "grad_norm": 0.8384215831756592, "learning_rate": 3.8734358118768465e-07, "loss": 3.0485, "step": 355950 }, { "epoch": 0.0152576, "grad_norm": 0.8951608538627625, "learning_rate": 3.871216586077764e-07, "loss": 2.8469, "step": 355960 }, { "epoch": 0.0152832, "grad_norm": 0.8164446949958801, "learning_rate": 3.868997983648304e-07, "loss": 2.8259, "step": 355970 }, { "epoch": 0.0153088, "grad_norm": 0.8755747675895691, "learning_rate": 3.866780004602833e-07, "loss": 2.9074, "step": 355980 }, { "epoch": 0.0153344, "grad_norm": 0.823090672492981, "learning_rate": 3.8645626489557385e-07, "loss": 2.726, "step": 355990 }, { "epoch": 0.01536, "grad_norm": 0.8582994341850281, "learning_rate": 3.8623459167214215e-07, "loss": 2.8927, "step": 356000 }, { "epoch": 0.0153856, "grad_norm": 0.8494754433631897, "learning_rate": 3.860129807914226e-07, "loss": 2.7898, "step": 356010 }, { "epoch": 0.0154112, "grad_norm": 0.9448773860931396, "learning_rate": 3.857914322548539e-07, "loss": 2.7818, "step": 356020 }, { "epoch": 0.0154368, "grad_norm": 0.8728298544883728, "learning_rate": 3.855699460638729e-07, "loss": 2.6851, "step": 356030 }, { "epoch": 0.0154624, "grad_norm": 0.8482863903045654, "learning_rate": 3.8534852221991604e-07, "loss": 3.066, "step": 356040 }, { "epoch": 0.015488, "grad_norm": 0.7925405502319336, "learning_rate": 3.8512716072442003e-07, "loss": 2.8352, "step": 356050 }, { "epoch": 0.0155136, "grad_norm": 0.8138425350189209, "learning_rate": 3.8490586157881704e-07, "loss": 2.7366, "step": 356060 }, { "epoch": 0.0155392, "grad_norm": 0.8363919258117676, "learning_rate": 3.846846247845448e-07, "loss": 2.6338, "step": 356070 }, { "epoch": 0.0155648, "grad_norm": 0.8264715075492859, "learning_rate": 3.8446345034303665e-07, "loss": 2.8201, "step": 356080 }, { "epoch": 0.0155904, "grad_norm": 0.8662041425704956, "learning_rate": 3.84242338255727e-07, "loss": 2.8723, "step": 356090 }, { "epoch": 0.015616, "grad_norm": 0.792937159538269, "learning_rate": 3.840212885240524e-07, "loss": 2.8425, "step": 356100 }, { "epoch": 0.0156416, "grad_norm": 0.7543958425521851, "learning_rate": 3.838003011494429e-07, "loss": 2.62, "step": 356110 }, { "epoch": 0.0156672, "grad_norm": 0.7939644455909729, "learning_rate": 3.8357937613333173e-07, "loss": 2.6917, "step": 356120 }, { "epoch": 0.0156928, "grad_norm": 0.8724422454833984, "learning_rate": 3.833585134771545e-07, "loss": 2.7854, "step": 356130 }, { "epoch": 0.0157184, "grad_norm": 0.8210118412971497, "learning_rate": 3.831377131823399e-07, "loss": 2.8548, "step": 356140 }, { "epoch": 0.015744, "grad_norm": 0.8246181011199951, "learning_rate": 3.8291697525032256e-07, "loss": 2.7215, "step": 356150 }, { "epoch": 0.0157696, "grad_norm": 1.2123020887374878, "learning_rate": 3.826962996825323e-07, "loss": 3.0254, "step": 356160 }, { "epoch": 0.0157952, "grad_norm": 0.82292640209198, "learning_rate": 3.8247568648040134e-07, "loss": 2.8188, "step": 356170 }, { "epoch": 0.0158208, "grad_norm": 1.0416985750198364, "learning_rate": 3.8225513564535855e-07, "loss": 2.7242, "step": 356180 }, { "epoch": 0.0158464, "grad_norm": 0.8492293953895569, "learning_rate": 3.82034647178835e-07, "loss": 2.8338, "step": 356190 }, { "epoch": 0.015872, "grad_norm": 1.0221173763275146, "learning_rate": 3.818142210822606e-07, "loss": 2.7265, "step": 356200 }, { "epoch": 0.0158976, "grad_norm": 0.9078836441040039, "learning_rate": 3.8159385735706545e-07, "loss": 2.9641, "step": 356210 }, { "epoch": 0.0159232, "grad_norm": 0.9264284372329712, "learning_rate": 3.8137355600467717e-07, "loss": 2.9932, "step": 356220 }, { "epoch": 0.0159488, "grad_norm": 0.9095993041992188, "learning_rate": 3.811533170265258e-07, "loss": 2.8046, "step": 356230 }, { "epoch": 0.0159744, "grad_norm": 0.984297513961792, "learning_rate": 3.8093314042403906e-07, "loss": 2.9646, "step": 356240 }, { "epoch": 0.016, "grad_norm": 0.9025633335113525, "learning_rate": 3.8071302619864356e-07, "loss": 2.818, "step": 356250 }, { "epoch": 0.0160256, "grad_norm": 0.8339889049530029, "learning_rate": 3.8049297435176604e-07, "loss": 2.8323, "step": 356260 }, { "epoch": 0.0160512, "grad_norm": 0.8688246011734009, "learning_rate": 3.8027298488483743e-07, "loss": 2.8424, "step": 356270 }, { "epoch": 0.0160768, "grad_norm": 0.8717886805534363, "learning_rate": 3.8005305779928114e-07, "loss": 2.7519, "step": 356280 }, { "epoch": 0.0161024, "grad_norm": 0.8104559779167175, "learning_rate": 3.7983319309652377e-07, "loss": 2.6543, "step": 356290 }, { "epoch": 0.016128, "grad_norm": 0.8583871126174927, "learning_rate": 3.796133907779931e-07, "loss": 2.7677, "step": 356300 }, { "epoch": 0.0161536, "grad_norm": 0.8546134233474731, "learning_rate": 3.7939365084511126e-07, "loss": 2.8429, "step": 356310 }, { "epoch": 0.0161792, "grad_norm": 1.309777021408081, "learning_rate": 3.7917397329930495e-07, "loss": 3.0424, "step": 356320 }, { "epoch": 0.0162048, "grad_norm": 0.857042670249939, "learning_rate": 3.7895435814199853e-07, "loss": 2.8647, "step": 356330 }, { "epoch": 0.0162304, "grad_norm": 1.4109753370285034, "learning_rate": 3.787348053746165e-07, "loss": 2.834, "step": 356340 }, { "epoch": 0.016256, "grad_norm": 0.7837526798248291, "learning_rate": 3.78515314998581e-07, "loss": 2.8641, "step": 356350 }, { "epoch": 0.0162816, "grad_norm": 0.7904261350631714, "learning_rate": 3.782958870153175e-07, "loss": 2.7749, "step": 356360 }, { "epoch": 0.0163072, "grad_norm": 0.793662428855896, "learning_rate": 3.780765214262494e-07, "loss": 2.726, "step": 356370 }, { "epoch": 0.0163328, "grad_norm": 0.8595956563949585, "learning_rate": 3.778572182327966e-07, "loss": 2.9113, "step": 356380 }, { "epoch": 0.0163584, "grad_norm": 0.8333421349525452, "learning_rate": 3.7763797743638254e-07, "loss": 2.7414, "step": 356390 }, { "epoch": 0.016384, "grad_norm": 1.0580447912216187, "learning_rate": 3.774187990384282e-07, "loss": 2.8542, "step": 356400 }, { "epoch": 0.0164096, "grad_norm": 0.8798726201057434, "learning_rate": 3.771996830403557e-07, "loss": 3.0166, "step": 356410 }, { "epoch": 0.0164352, "grad_norm": 0.9818991422653198, "learning_rate": 3.7698062944358627e-07, "loss": 2.762, "step": 356420 }, { "epoch": 0.0164608, "grad_norm": 1.1108835935592651, "learning_rate": 3.7676163824953983e-07, "loss": 2.939, "step": 356430 }, { "epoch": 0.0164864, "grad_norm": 2.278407335281372, "learning_rate": 3.7654270945963633e-07, "loss": 2.6009, "step": 356440 }, { "epoch": 0.016512, "grad_norm": 0.8141232132911682, "learning_rate": 3.763238430752969e-07, "loss": 2.7213, "step": 356450 }, { "epoch": 0.0165376, "grad_norm": 0.9170156121253967, "learning_rate": 3.761050390979382e-07, "loss": 3.1215, "step": 356460 }, { "epoch": 0.0165632, "grad_norm": 0.8886229991912842, "learning_rate": 3.758862975289812e-07, "loss": 3.0398, "step": 356470 }, { "epoch": 0.0165888, "grad_norm": 0.7874343395233154, "learning_rate": 3.7566761836984377e-07, "loss": 2.8801, "step": 356480 }, { "epoch": 0.0166144, "grad_norm": 0.850438117980957, "learning_rate": 3.7544900162194365e-07, "loss": 2.8484, "step": 356490 }, { "epoch": 0.01664, "grad_norm": 0.8482429385185242, "learning_rate": 3.752304472867008e-07, "loss": 2.8544, "step": 356500 }, { "epoch": 0.0166656, "grad_norm": 0.7691676020622253, "learning_rate": 3.7501195536552846e-07, "loss": 2.9109, "step": 356510 }, { "epoch": 0.0166912, "grad_norm": 0.7860456705093384, "learning_rate": 3.747935258598467e-07, "loss": 2.9945, "step": 356520 }, { "epoch": 0.0167168, "grad_norm": 0.8919698596000671, "learning_rate": 3.7457515877107e-07, "loss": 2.9993, "step": 356530 }, { "epoch": 0.0167424, "grad_norm": 0.8039852976799011, "learning_rate": 3.7435685410061597e-07, "loss": 2.9518, "step": 356540 }, { "epoch": 0.016768, "grad_norm": 1.1980818510055542, "learning_rate": 3.7413861184989907e-07, "loss": 2.8814, "step": 356550 }, { "epoch": 0.0167936, "grad_norm": 1.0211799144744873, "learning_rate": 3.739204320203349e-07, "loss": 2.7592, "step": 356560 }, { "epoch": 0.0168192, "grad_norm": 0.8389096856117249, "learning_rate": 3.7370231461333893e-07, "loss": 3.0545, "step": 356570 }, { "epoch": 0.0168448, "grad_norm": 0.8373697400093079, "learning_rate": 3.7348425963032563e-07, "loss": 2.5973, "step": 356580 }, { "epoch": 0.0168704, "grad_norm": 0.8300129771232605, "learning_rate": 3.732662670727072e-07, "loss": 3.0249, "step": 356590 }, { "epoch": 0.016896, "grad_norm": 1.3193161487579346, "learning_rate": 3.730483369418991e-07, "loss": 2.7609, "step": 356600 }, { "epoch": 0.0169216, "grad_norm": 0.8767089247703552, "learning_rate": 3.7283046923931253e-07, "loss": 2.7978, "step": 356610 }, { "epoch": 0.0169472, "grad_norm": 1.2913000583648682, "learning_rate": 3.7261266396636297e-07, "loss": 2.9437, "step": 356620 }, { "epoch": 0.0169728, "grad_norm": 0.9307937622070312, "learning_rate": 3.723949211244626e-07, "loss": 2.9632, "step": 356630 }, { "epoch": 0.0169984, "grad_norm": 0.9263268709182739, "learning_rate": 3.7217724071502147e-07, "loss": 2.7701, "step": 356640 }, { "epoch": 0.017024, "grad_norm": 0.8866972923278809, "learning_rate": 3.7195962273945284e-07, "loss": 2.7214, "step": 356650 }, { "epoch": 0.0170496, "grad_norm": 0.9763185977935791, "learning_rate": 3.717420671991667e-07, "loss": 2.842, "step": 356660 }, { "epoch": 0.0170752, "grad_norm": 0.8631784319877625, "learning_rate": 3.7152457409557417e-07, "loss": 2.7171, "step": 356670 }, { "epoch": 0.0171008, "grad_norm": 0.8188663125038147, "learning_rate": 3.7130714343008523e-07, "loss": 2.7814, "step": 356680 }, { "epoch": 0.0171264, "grad_norm": 0.9207231998443604, "learning_rate": 3.7108977520411207e-07, "loss": 2.9625, "step": 356690 }, { "epoch": 0.017152, "grad_norm": 1.0609456300735474, "learning_rate": 3.708724694190624e-07, "loss": 2.824, "step": 356700 }, { "epoch": 0.0171776, "grad_norm": 1.5441702604293823, "learning_rate": 3.706552260763452e-07, "loss": 2.6722, "step": 356710 }, { "epoch": 0.0172032, "grad_norm": 0.8378818035125732, "learning_rate": 3.7043804517736926e-07, "loss": 2.4161, "step": 356720 }, { "epoch": 0.0172288, "grad_norm": 0.9179863333702087, "learning_rate": 3.702209267235435e-07, "loss": 2.8868, "step": 356730 }, { "epoch": 0.0172544, "grad_norm": 0.7668533325195312, "learning_rate": 3.7000387071627675e-07, "loss": 2.9255, "step": 356740 }, { "epoch": 0.01728, "grad_norm": 0.9855888485908508, "learning_rate": 3.697868771569746e-07, "loss": 2.9237, "step": 356750 }, { "epoch": 0.0173056, "grad_norm": 0.8790473341941833, "learning_rate": 3.6956994604704475e-07, "loss": 2.7546, "step": 356760 }, { "epoch": 0.0173312, "grad_norm": 0.8389224410057068, "learning_rate": 3.6935307738789614e-07, "loss": 2.895, "step": 356770 }, { "epoch": 0.0173568, "grad_norm": 0.9228066802024841, "learning_rate": 3.6913627118093207e-07, "loss": 2.8612, "step": 356780 }, { "epoch": 0.0173824, "grad_norm": 0.7983108162879944, "learning_rate": 3.689195274275581e-07, "loss": 2.8312, "step": 356790 }, { "epoch": 0.017408, "grad_norm": 0.8759590983390808, "learning_rate": 3.6870284612918307e-07, "loss": 2.8324, "step": 356800 }, { "epoch": 0.0174336, "grad_norm": 0.8766544461250305, "learning_rate": 3.684862272872103e-07, "loss": 2.8789, "step": 356810 }, { "epoch": 0.0174592, "grad_norm": 0.8682771921157837, "learning_rate": 3.6826967090304424e-07, "loss": 2.8456, "step": 356820 }, { "epoch": 0.0174848, "grad_norm": 0.8472569584846497, "learning_rate": 3.680531769780915e-07, "loss": 2.862, "step": 356830 }, { "epoch": 0.0175104, "grad_norm": 0.8703341484069824, "learning_rate": 3.6783674551375215e-07, "loss": 2.8946, "step": 356840 }, { "epoch": 0.017536, "grad_norm": 0.806994616985321, "learning_rate": 3.6762037651143166e-07, "loss": 2.9864, "step": 356850 }, { "epoch": 0.0175616, "grad_norm": 0.8299776911735535, "learning_rate": 3.6740406997253344e-07, "loss": 2.6782, "step": 356860 }, { "epoch": 0.0175872, "grad_norm": 0.8227347135543823, "learning_rate": 3.671878258984596e-07, "loss": 2.7531, "step": 356870 }, { "epoch": 0.0176128, "grad_norm": 0.7792096138000488, "learning_rate": 3.669716442906124e-07, "loss": 3.0548, "step": 356880 }, { "epoch": 0.0176384, "grad_norm": 0.9493094086647034, "learning_rate": 3.6675552515039404e-07, "loss": 2.7482, "step": 356890 }, { "epoch": 0.017664, "grad_norm": 0.8358091711997986, "learning_rate": 3.6653946847920675e-07, "loss": 2.7964, "step": 356900 }, { "epoch": 0.0176896, "grad_norm": 0.795693576335907, "learning_rate": 3.663234742784505e-07, "loss": 2.7852, "step": 356910 }, { "epoch": 0.0177152, "grad_norm": 0.8741291761398315, "learning_rate": 3.661075425495253e-07, "loss": 2.8879, "step": 356920 }, { "epoch": 0.0177408, "grad_norm": 0.8038560748100281, "learning_rate": 3.658916732938322e-07, "loss": 2.8262, "step": 356930 }, { "epoch": 0.0177664, "grad_norm": 0.8979613184928894, "learning_rate": 3.656758665127713e-07, "loss": 2.7062, "step": 356940 }, { "epoch": 0.017792, "grad_norm": 0.9336890578269958, "learning_rate": 3.654601222077425e-07, "loss": 2.8389, "step": 356950 }, { "epoch": 0.0178176, "grad_norm": 0.8193153738975525, "learning_rate": 3.6524444038014364e-07, "loss": 2.9731, "step": 356960 }, { "epoch": 0.0178432, "grad_norm": 0.8553789258003235, "learning_rate": 3.650288210313746e-07, "loss": 2.76, "step": 356970 }, { "epoch": 0.0178688, "grad_norm": 0.8077735304832458, "learning_rate": 3.6481326416283327e-07, "loss": 2.9176, "step": 356980 }, { "epoch": 0.0178944, "grad_norm": 0.8997255563735962, "learning_rate": 3.645977697759162e-07, "loss": 2.7739, "step": 356990 }, { "epoch": 0.01792, "grad_norm": 0.9271358251571655, "learning_rate": 3.6438233787202237e-07, "loss": 2.933, "step": 357000 }, { "epoch": 0.0179456, "grad_norm": 0.7882974743843079, "learning_rate": 3.641669684525473e-07, "loss": 2.5665, "step": 357010 }, { "epoch": 0.0179712, "grad_norm": 0.8780937790870667, "learning_rate": 3.639516615188898e-07, "loss": 2.7962, "step": 357020 }, { "epoch": 0.0179968, "grad_norm": 0.8721827268600464, "learning_rate": 3.6373641707244554e-07, "loss": 2.8515, "step": 357030 }, { "epoch": 0.0180224, "grad_norm": 0.8270830512046814, "learning_rate": 3.635212351146089e-07, "loss": 2.7801, "step": 357040 }, { "epoch": 0.018048, "grad_norm": 0.855841338634491, "learning_rate": 3.6330611564677544e-07, "loss": 2.93, "step": 357050 }, { "epoch": 0.0180736, "grad_norm": 0.8515569567680359, "learning_rate": 3.630910586703407e-07, "loss": 2.7131, "step": 357060 }, { "epoch": 0.0180992, "grad_norm": 0.9722879528999329, "learning_rate": 3.628760641866991e-07, "loss": 2.6271, "step": 357070 }, { "epoch": 0.0181248, "grad_norm": 0.8578132390975952, "learning_rate": 3.626611321972451e-07, "loss": 2.9875, "step": 357080 }, { "epoch": 0.0181504, "grad_norm": 0.9205694198608398, "learning_rate": 3.6244626270337204e-07, "loss": 2.9096, "step": 357090 }, { "epoch": 0.018176, "grad_norm": 0.8434097170829773, "learning_rate": 3.622314557064754e-07, "loss": 3.1181, "step": 357100 }, { "epoch": 0.0182016, "grad_norm": 0.8103192448616028, "learning_rate": 3.6201671120794536e-07, "loss": 2.8047, "step": 357110 }, { "epoch": 0.0182272, "grad_norm": 0.8405083417892456, "learning_rate": 3.618020292091751e-07, "loss": 2.7594, "step": 357120 }, { "epoch": 0.0182528, "grad_norm": 0.7940076589584351, "learning_rate": 3.615874097115557e-07, "loss": 2.9456, "step": 357130 }, { "epoch": 0.0182784, "grad_norm": 1.0510600805282593, "learning_rate": 3.613728527164817e-07, "loss": 2.9398, "step": 357140 }, { "epoch": 0.018304, "grad_norm": 1.024303913116455, "learning_rate": 3.6115835822534307e-07, "loss": 2.7637, "step": 357150 }, { "epoch": 0.0183296, "grad_norm": 0.9900856018066406, "learning_rate": 3.60943926239532e-07, "loss": 2.9697, "step": 357160 }, { "epoch": 0.0183552, "grad_norm": 0.8317590951919556, "learning_rate": 3.6072955676043633e-07, "loss": 2.913, "step": 357170 }, { "epoch": 0.0183808, "grad_norm": 0.8194674253463745, "learning_rate": 3.605152497894482e-07, "loss": 2.7407, "step": 357180 }, { "epoch": 0.0184064, "grad_norm": 0.9554177522659302, "learning_rate": 3.6030100532795656e-07, "loss": 2.6759, "step": 357190 }, { "epoch": 0.018432, "grad_norm": 0.8603959083557129, "learning_rate": 3.6008682337735133e-07, "loss": 2.8323, "step": 357200 }, { "epoch": 0.0184576, "grad_norm": 0.8083444237709045, "learning_rate": 3.5987270393902154e-07, "loss": 2.7091, "step": 357210 }, { "epoch": 0.0184832, "grad_norm": 0.9552183747291565, "learning_rate": 3.596586470143537e-07, "loss": 2.7631, "step": 357220 }, { "epoch": 0.0185088, "grad_norm": 0.8296881318092346, "learning_rate": 3.5944465260474016e-07, "loss": 2.6077, "step": 357230 }, { "epoch": 0.0185344, "grad_norm": 1.0458465814590454, "learning_rate": 3.5923072071156415e-07, "loss": 2.7851, "step": 357240 }, { "epoch": 0.01856, "grad_norm": 0.8032313585281372, "learning_rate": 3.5901685133621354e-07, "loss": 2.7241, "step": 357250 }, { "epoch": 0.0185856, "grad_norm": 1.0201917886734009, "learning_rate": 3.588030444800783e-07, "loss": 2.7282, "step": 357260 }, { "epoch": 0.0186112, "grad_norm": 0.8737740516662598, "learning_rate": 3.5858930014454176e-07, "loss": 2.8049, "step": 357270 }, { "epoch": 0.0186368, "grad_norm": 0.8045428395271301, "learning_rate": 3.5837561833099167e-07, "loss": 2.8502, "step": 357280 }, { "epoch": 0.0186624, "grad_norm": 0.9172534346580505, "learning_rate": 3.581619990408125e-07, "loss": 2.9636, "step": 357290 }, { "epoch": 0.018688, "grad_norm": 0.777588427066803, "learning_rate": 3.5794844227539313e-07, "loss": 3.0661, "step": 357300 }, { "epoch": 0.0187136, "grad_norm": 0.7584623098373413, "learning_rate": 3.577349480361114e-07, "loss": 2.7442, "step": 357310 }, { "epoch": 0.0187392, "grad_norm": 0.8711616396903992, "learning_rate": 3.575215163243584e-07, "loss": 2.8137, "step": 357320 }, { "epoch": 0.0187648, "grad_norm": 0.82660973072052, "learning_rate": 3.573081471415152e-07, "loss": 2.9127, "step": 357330 }, { "epoch": 0.0187904, "grad_norm": 0.796889066696167, "learning_rate": 3.5709484048896513e-07, "loss": 2.7764, "step": 357340 }, { "epoch": 0.018816, "grad_norm": 0.7684995532035828, "learning_rate": 3.568815963680949e-07, "loss": 2.7656, "step": 357350 }, { "epoch": 0.0188416, "grad_norm": 0.8620963096618652, "learning_rate": 3.5666841478028235e-07, "loss": 2.8105, "step": 357360 }, { "epoch": 0.0188672, "grad_norm": 1.147603988647461, "learning_rate": 3.5645529572691296e-07, "loss": 2.963, "step": 357370 }, { "epoch": 0.0188928, "grad_norm": 0.9579992294311523, "learning_rate": 3.5624223920936785e-07, "loss": 2.7668, "step": 357380 }, { "epoch": 0.0189184, "grad_norm": 0.809008002281189, "learning_rate": 3.5602924522902925e-07, "loss": 2.7534, "step": 357390 }, { "epoch": 0.018944, "grad_norm": 0.9789981245994568, "learning_rate": 3.558163137872772e-07, "loss": 3.1288, "step": 357400 }, { "epoch": 0.0189696, "grad_norm": 0.827223002910614, "learning_rate": 3.5560344488549283e-07, "loss": 2.7827, "step": 357410 }, { "epoch": 0.0189952, "grad_norm": 1.576269268989563, "learning_rate": 3.553906385250583e-07, "loss": 2.7825, "step": 357420 }, { "epoch": 0.0190208, "grad_norm": 0.8635405898094177, "learning_rate": 3.551778947073525e-07, "loss": 2.8658, "step": 357430 }, { "epoch": 0.0190464, "grad_norm": 0.8482396602630615, "learning_rate": 3.5496521343375335e-07, "loss": 2.6912, "step": 357440 }, { "epoch": 0.019072, "grad_norm": 0.9180181622505188, "learning_rate": 3.547525947056418e-07, "loss": 2.9517, "step": 357450 }, { "epoch": 0.0190976, "grad_norm": 1.155950665473938, "learning_rate": 3.545400385243969e-07, "loss": 2.8614, "step": 357460 }, { "epoch": 0.0191232, "grad_norm": 0.9351534843444824, "learning_rate": 3.543275448913963e-07, "loss": 2.9304, "step": 357470 }, { "epoch": 0.0191488, "grad_norm": 0.8626660108566284, "learning_rate": 3.5411511380801676e-07, "loss": 2.8257, "step": 357480 }, { "epoch": 0.0191744, "grad_norm": 0.810751736164093, "learning_rate": 3.539027452756383e-07, "loss": 2.8712, "step": 357490 }, { "epoch": 0.0192, "grad_norm": 0.9126148223876953, "learning_rate": 3.536904392956364e-07, "loss": 2.7974, "step": 357500 }, { "epoch": 0.0192256, "grad_norm": 0.9139586687088013, "learning_rate": 3.5347819586938893e-07, "loss": 3.0594, "step": 357510 }, { "epoch": 0.0192512, "grad_norm": 0.8975963592529297, "learning_rate": 3.532660149982714e-07, "loss": 2.8342, "step": 357520 }, { "epoch": 0.0192768, "grad_norm": 1.0606130361557007, "learning_rate": 3.530538966836594e-07, "loss": 2.8298, "step": 357530 }, { "epoch": 0.0193024, "grad_norm": 0.8282232284545898, "learning_rate": 3.5284184092692963e-07, "loss": 2.6477, "step": 357540 }, { "epoch": 0.019328, "grad_norm": 0.852294385433197, "learning_rate": 3.5262984772945766e-07, "loss": 2.6623, "step": 357550 }, { "epoch": 0.0193536, "grad_norm": 0.8324778079986572, "learning_rate": 3.5241791709261563e-07, "loss": 2.5356, "step": 357560 }, { "epoch": 0.0193792, "grad_norm": 0.9213930368423462, "learning_rate": 3.5220604901778035e-07, "loss": 2.7298, "step": 357570 }, { "epoch": 0.0194048, "grad_norm": 0.9170143008232117, "learning_rate": 3.5199424350632395e-07, "loss": 2.5974, "step": 357580 }, { "epoch": 0.0194304, "grad_norm": 0.918937623500824, "learning_rate": 3.517825005596209e-07, "loss": 2.6155, "step": 357590 }, { "epoch": 0.019456, "grad_norm": 0.8403602838516235, "learning_rate": 3.515708201790435e-07, "loss": 2.7309, "step": 357600 }, { "epoch": 0.0194816, "grad_norm": 1.0749739408493042, "learning_rate": 3.513592023659662e-07, "loss": 2.6663, "step": 357610 }, { "epoch": 0.0195072, "grad_norm": 0.9244095087051392, "learning_rate": 3.51147647121759e-07, "loss": 2.7121, "step": 357620 }, { "epoch": 0.0195328, "grad_norm": 0.8519610166549683, "learning_rate": 3.509361544477974e-07, "loss": 2.6818, "step": 357630 }, { "epoch": 0.0195584, "grad_norm": 0.8254101872444153, "learning_rate": 3.5072472434544814e-07, "loss": 2.6561, "step": 357640 }, { "epoch": 0.019584, "grad_norm": 0.8439866304397583, "learning_rate": 3.505133568160857e-07, "loss": 2.4911, "step": 357650 }, { "epoch": 0.0196096, "grad_norm": 0.8207749724388123, "learning_rate": 3.503020518610778e-07, "loss": 2.5537, "step": 357660 }, { "epoch": 0.0196352, "grad_norm": 0.867655336856842, "learning_rate": 3.5009080948179785e-07, "loss": 2.6487, "step": 357670 }, { "epoch": 0.0196608, "grad_norm": 0.7446701526641846, "learning_rate": 3.4987962967961475e-07, "loss": 2.6455, "step": 357680 }, { "epoch": 0.0196864, "grad_norm": 0.7523216009140015, "learning_rate": 3.496685124558974e-07, "loss": 2.6172, "step": 357690 }, { "epoch": 0.019712, "grad_norm": 0.8027570843696594, "learning_rate": 3.494574578120147e-07, "loss": 2.3096, "step": 357700 }, { "epoch": 0.0197376, "grad_norm": 0.7991381883621216, "learning_rate": 3.492464657493355e-07, "loss": 2.4197, "step": 357710 }, { "epoch": 0.0197632, "grad_norm": 0.9936993718147278, "learning_rate": 3.490355362692277e-07, "loss": 2.6336, "step": 357720 }, { "epoch": 0.0197888, "grad_norm": 0.7676852345466614, "learning_rate": 3.4882466937306017e-07, "loss": 2.5347, "step": 357730 }, { "epoch": 0.0198144, "grad_norm": 0.7524250745773315, "learning_rate": 3.486138650621995e-07, "loss": 2.4667, "step": 357740 }, { "epoch": 0.01984, "grad_norm": 0.8219428658485413, "learning_rate": 3.484031233380136e-07, "loss": 2.7184, "step": 357750 }, { "epoch": 0.0198656, "grad_norm": 0.8093791604042053, "learning_rate": 3.4819244420186804e-07, "loss": 2.5916, "step": 357760 }, { "epoch": 0.0198912, "grad_norm": 0.8165967464447021, "learning_rate": 3.4798182765512946e-07, "loss": 2.4908, "step": 357770 }, { "epoch": 0.0199168, "grad_norm": 0.8049386143684387, "learning_rate": 3.477712736991623e-07, "loss": 2.6223, "step": 357780 }, { "epoch": 0.0199424, "grad_norm": 0.8944780826568604, "learning_rate": 3.475607823353344e-07, "loss": 2.4673, "step": 357790 }, { "epoch": 0.019968, "grad_norm": 0.8028469085693359, "learning_rate": 3.4735035356500804e-07, "loss": 2.9904, "step": 357800 }, { "epoch": 0.0199936, "grad_norm": 0.8506789803504944, "learning_rate": 3.471399873895509e-07, "loss": 2.511, "step": 357810 }, { "epoch": 0.0200192, "grad_norm": 0.8768380880355835, "learning_rate": 3.469296838103242e-07, "loss": 2.6507, "step": 357820 }, { "epoch": 0.0200448, "grad_norm": 0.7562704086303711, "learning_rate": 3.4671944282869353e-07, "loss": 2.5766, "step": 357830 }, { "epoch": 0.0200704, "grad_norm": 0.8355880379676819, "learning_rate": 3.4650926444602217e-07, "loss": 2.5352, "step": 357840 }, { "epoch": 0.020096, "grad_norm": 0.8093586564064026, "learning_rate": 3.4629914866367245e-07, "loss": 2.6134, "step": 357850 }, { "epoch": 0.0201216, "grad_norm": 0.7863557934761047, "learning_rate": 3.460890954830065e-07, "loss": 2.4669, "step": 357860 }, { "epoch": 0.0201472, "grad_norm": 0.9318779110908508, "learning_rate": 3.458791049053878e-07, "loss": 2.4795, "step": 357870 }, { "epoch": 0.0201728, "grad_norm": 0.7538451552391052, "learning_rate": 3.4566917693217847e-07, "loss": 2.3964, "step": 357880 }, { "epoch": 0.0201984, "grad_norm": 0.7625318765640259, "learning_rate": 3.454593115647376e-07, "loss": 2.5309, "step": 357890 }, { "epoch": 0.020224, "grad_norm": 1.0094777345657349, "learning_rate": 3.452495088044272e-07, "loss": 2.5423, "step": 357900 }, { "epoch": 0.0202496, "grad_norm": 0.9575576186180115, "learning_rate": 3.4503976865260747e-07, "loss": 2.6599, "step": 357910 }, { "epoch": 0.0202752, "grad_norm": 0.7875673174858093, "learning_rate": 3.4483009111063834e-07, "loss": 2.5729, "step": 357920 }, { "epoch": 0.0203008, "grad_norm": 0.8322022557258606, "learning_rate": 3.4462047617988106e-07, "loss": 2.5566, "step": 357930 }, { "epoch": 0.0203264, "grad_norm": 0.8892752528190613, "learning_rate": 3.444109238616933e-07, "loss": 2.889, "step": 357940 }, { "epoch": 0.020352, "grad_norm": 0.8270470499992371, "learning_rate": 3.4420143415743625e-07, "loss": 2.5839, "step": 357950 }, { "epoch": 0.0203776, "grad_norm": 0.8682546019554138, "learning_rate": 3.439920070684644e-07, "loss": 2.8892, "step": 357960 }, { "epoch": 0.0204032, "grad_norm": 0.8530411124229431, "learning_rate": 3.437826425961388e-07, "loss": 2.776, "step": 357970 }, { "epoch": 0.0204288, "grad_norm": 0.7646711468696594, "learning_rate": 3.4357334074181736e-07, "loss": 2.4847, "step": 357980 }, { "epoch": 0.0204544, "grad_norm": 0.865808367729187, "learning_rate": 3.433641015068556e-07, "loss": 2.6319, "step": 357990 }, { "epoch": 0.02048, "grad_norm": 0.8764150738716125, "learning_rate": 3.431549248926103e-07, "loss": 2.6459, "step": 358000 }, { "epoch": 0.0205056, "grad_norm": 0.918997585773468, "learning_rate": 3.429458109004402e-07, "loss": 2.8447, "step": 358010 }, { "epoch": 0.0205312, "grad_norm": 0.7970659136772156, "learning_rate": 3.4273675953169995e-07, "loss": 2.7185, "step": 358020 }, { "epoch": 0.0205568, "grad_norm": 0.834740400314331, "learning_rate": 3.4252777078774503e-07, "loss": 2.5746, "step": 358030 }, { "epoch": 0.0205824, "grad_norm": 0.8934484124183655, "learning_rate": 3.4231884466992995e-07, "loss": 2.5694, "step": 358040 }, { "epoch": 0.020608, "grad_norm": 0.9071245193481445, "learning_rate": 3.4210998117961135e-07, "loss": 2.8831, "step": 358050 }, { "epoch": 0.0206336, "grad_norm": 1.0180152654647827, "learning_rate": 3.4190118031814267e-07, "loss": 2.2752, "step": 358060 }, { "epoch": 0.0206592, "grad_norm": 0.8994042277336121, "learning_rate": 3.4169244208687837e-07, "loss": 2.6633, "step": 358070 }, { "epoch": 0.0206848, "grad_norm": 1.006837010383606, "learning_rate": 3.4148376648717175e-07, "loss": 2.4156, "step": 358080 }, { "epoch": 0.0207104, "grad_norm": 0.8276848793029785, "learning_rate": 3.412751535203762e-07, "loss": 2.4775, "step": 358090 }, { "epoch": 0.020736, "grad_norm": 0.8110013604164124, "learning_rate": 3.410666031878429e-07, "loss": 2.6771, "step": 358100 }, { "epoch": 0.0207616, "grad_norm": 0.8284925222396851, "learning_rate": 3.408581154909263e-07, "loss": 2.413, "step": 358110 }, { "epoch": 0.0207872, "grad_norm": 0.8160762190818787, "learning_rate": 3.406496904309775e-07, "loss": 2.3885, "step": 358120 }, { "epoch": 0.0208128, "grad_norm": 0.9403979182243347, "learning_rate": 3.404413280093488e-07, "loss": 2.055, "step": 358130 }, { "epoch": 0.0208384, "grad_norm": 0.9254236817359924, "learning_rate": 3.4023302822739134e-07, "loss": 2.6111, "step": 358140 }, { "epoch": 0.020864, "grad_norm": 0.883939802646637, "learning_rate": 3.4002479108645513e-07, "loss": 2.8945, "step": 358150 }, { "epoch": 0.0208896, "grad_norm": 0.8858767747879028, "learning_rate": 3.3981661658789136e-07, "loss": 2.6043, "step": 358160 }, { "epoch": 0.0209152, "grad_norm": 0.9154184460639954, "learning_rate": 3.3960850473304776e-07, "loss": 2.4432, "step": 358170 }, { "epoch": 0.0209408, "grad_norm": 0.9165037274360657, "learning_rate": 3.394004555232755e-07, "loss": 2.6897, "step": 358180 }, { "epoch": 0.0209664, "grad_norm": 0.7719820141792297, "learning_rate": 3.391924689599246e-07, "loss": 2.7534, "step": 358190 }, { "epoch": 0.020992, "grad_norm": 0.9069293737411499, "learning_rate": 3.38984545044343e-07, "loss": 2.6199, "step": 358200 }, { "epoch": 2.56e-05, "grad_norm": 0.8493655323982239, "learning_rate": 3.3877668377788055e-07, "loss": 2.1202, "step": 358210 }, { "epoch": 5.12e-05, "grad_norm": 0.8336993455886841, "learning_rate": 3.385688851618818e-07, "loss": 2.528, "step": 358220 }, { "epoch": 7.68e-05, "grad_norm": 0.8377329111099243, "learning_rate": 3.383611491976968e-07, "loss": 2.6588, "step": 358230 }, { "epoch": 0.0001024, "grad_norm": 0.893092155456543, "learning_rate": 3.3815347588667224e-07, "loss": 2.7087, "step": 358240 }, { "epoch": 0.000128, "grad_norm": 0.8239572644233704, "learning_rate": 3.379458652301537e-07, "loss": 2.6685, "step": 358250 }, { "epoch": 0.0001536, "grad_norm": 0.8444437980651855, "learning_rate": 3.37738317229489e-07, "loss": 2.8966, "step": 358260 }, { "epoch": 0.0001792, "grad_norm": 0.921577513217926, "learning_rate": 3.3753083188602264e-07, "loss": 2.6925, "step": 358270 }, { "epoch": 0.0002048, "grad_norm": 0.8279072046279907, "learning_rate": 3.373234092011024e-07, "loss": 2.7573, "step": 358280 }, { "epoch": 0.0002304, "grad_norm": 0.9503729939460754, "learning_rate": 3.3711604917607163e-07, "loss": 2.6698, "step": 358290 }, { "epoch": 0.000256, "grad_norm": 0.9888485670089722, "learning_rate": 3.3690875181227377e-07, "loss": 2.7001, "step": 358300 }, { "epoch": 0.0002816, "grad_norm": 0.9078336358070374, "learning_rate": 3.367015171110555e-07, "loss": 2.4357, "step": 358310 }, { "epoch": 0.0003072, "grad_norm": 0.8117787837982178, "learning_rate": 3.36494345073759e-07, "loss": 2.5551, "step": 358320 }, { "epoch": 0.0003328, "grad_norm": 0.8579913973808289, "learning_rate": 3.362872357017288e-07, "loss": 2.6895, "step": 358330 }, { "epoch": 0.0003584, "grad_norm": 0.8086233139038086, "learning_rate": 3.3608018899630723e-07, "loss": 2.7002, "step": 358340 }, { "epoch": 0.000384, "grad_norm": 0.8790589570999146, "learning_rate": 3.3587320495883757e-07, "loss": 2.6291, "step": 358350 }, { "epoch": 0.0004096, "grad_norm": 0.8411518931388855, "learning_rate": 3.356662835906621e-07, "loss": 2.612, "step": 358360 }, { "epoch": 0.0004352, "grad_norm": 0.8399873971939087, "learning_rate": 3.35459424893122e-07, "loss": 2.5905, "step": 358370 }, { "epoch": 0.0004608, "grad_norm": 0.8374341130256653, "learning_rate": 3.352526288675584e-07, "loss": 2.5514, "step": 358380 }, { "epoch": 0.0004864, "grad_norm": 0.9775924682617188, "learning_rate": 3.350458955153135e-07, "loss": 2.6876, "step": 358390 }, { "epoch": 0.000512, "grad_norm": 0.83266681432724, "learning_rate": 3.348392248377275e-07, "loss": 2.6492, "step": 358400 }, { "epoch": 0.0005376, "grad_norm": 0.7870023846626282, "learning_rate": 3.3463261683614136e-07, "loss": 2.4774, "step": 358410 }, { "epoch": 0.0005632, "grad_norm": 0.8509982228279114, "learning_rate": 3.34426071511893e-07, "loss": 2.6133, "step": 358420 }, { "epoch": 0.0005888, "grad_norm": 0.9192668795585632, "learning_rate": 3.3421958886632245e-07, "loss": 2.4255, "step": 358430 }, { "epoch": 0.0006144, "grad_norm": 0.8058155179023743, "learning_rate": 3.3401316890076863e-07, "loss": 2.591, "step": 358440 }, { "epoch": 0.00064, "grad_norm": 0.8633183240890503, "learning_rate": 3.3380681161657156e-07, "loss": 2.5878, "step": 358450 }, { "epoch": 0.0006656, "grad_norm": 0.9371931552886963, "learning_rate": 3.33600517015068e-07, "loss": 2.7232, "step": 358460 }, { "epoch": 0.0006912, "grad_norm": 0.8160141706466675, "learning_rate": 3.3339428509759575e-07, "loss": 2.6443, "step": 358470 }, { "epoch": 0.0007168, "grad_norm": 0.8701672554016113, "learning_rate": 3.331881158654937e-07, "loss": 2.5917, "step": 358480 }, { "epoch": 0.0007424, "grad_norm": 0.8737747669219971, "learning_rate": 3.329820093200964e-07, "loss": 2.7606, "step": 358490 }, { "epoch": 0.000768, "grad_norm": 1.288076400756836, "learning_rate": 3.327759654627416e-07, "loss": 2.6801, "step": 358500 }, { "epoch": 0.0007936, "grad_norm": 0.8546943664550781, "learning_rate": 3.3256998429476494e-07, "loss": 2.4505, "step": 358510 }, { "epoch": 0.0008192, "grad_norm": 0.7786909937858582, "learning_rate": 3.323640658175031e-07, "loss": 2.3003, "step": 358520 }, { "epoch": 0.0008448, "grad_norm": 0.9283780455589294, "learning_rate": 3.3215821003229064e-07, "loss": 2.8724, "step": 358530 }, { "epoch": 0.0008704, "grad_norm": 0.8614253401756287, "learning_rate": 3.319524169404642e-07, "loss": 2.6647, "step": 358540 }, { "epoch": 0.000896, "grad_norm": 0.9510716199874878, "learning_rate": 3.3174668654335606e-07, "loss": 2.5113, "step": 358550 }, { "epoch": 0.0009216, "grad_norm": 0.8467705249786377, "learning_rate": 3.3154101884230074e-07, "loss": 2.459, "step": 358560 }, { "epoch": 0.0009472, "grad_norm": 0.9667323231697083, "learning_rate": 3.3133541383863264e-07, "loss": 2.5842, "step": 358570 }, { "epoch": 0.0009728, "grad_norm": 0.9804608225822449, "learning_rate": 3.3112987153368416e-07, "loss": 2.689, "step": 358580 }, { "epoch": 0.0009984, "grad_norm": 0.7913295030593872, "learning_rate": 3.309243919287897e-07, "loss": 2.5056, "step": 358590 }, { "epoch": 0.001024, "grad_norm": 0.8682906627655029, "learning_rate": 3.3071897502528036e-07, "loss": 2.8303, "step": 358600 }, { "epoch": 0.0010496, "grad_norm": 0.8692767024040222, "learning_rate": 3.3051362082448966e-07, "loss": 2.5271, "step": 358610 }, { "epoch": 0.0010752, "grad_norm": 0.8794887661933899, "learning_rate": 3.3030832932774756e-07, "loss": 2.6493, "step": 358620 }, { "epoch": 0.0011008, "grad_norm": 1.262496829032898, "learning_rate": 3.301031005363853e-07, "loss": 2.774, "step": 358630 }, { "epoch": 0.0011264, "grad_norm": 0.8805108666419983, "learning_rate": 3.2989793445173504e-07, "loss": 2.5052, "step": 358640 }, { "epoch": 0.001152, "grad_norm": 0.8204418420791626, "learning_rate": 3.2969283107512683e-07, "loss": 2.6836, "step": 358650 }, { "epoch": 0.0011776, "grad_norm": 0.9122073650360107, "learning_rate": 3.2948779040789083e-07, "loss": 2.646, "step": 358660 }, { "epoch": 0.0012032, "grad_norm": 0.8597409129142761, "learning_rate": 3.2928281245135583e-07, "loss": 2.5259, "step": 358670 }, { "epoch": 0.0012288, "grad_norm": 0.8396261930465698, "learning_rate": 3.2907789720685204e-07, "loss": 3.5243, "step": 358680 }, { "epoch": 0.0012544, "grad_norm": 1.0079951286315918, "learning_rate": 3.2887304467570824e-07, "loss": 2.9347, "step": 358690 }, { "epoch": 0.00128, "grad_norm": 0.8364352583885193, "learning_rate": 3.2866825485925014e-07, "loss": 2.7624, "step": 358700 }, { "epoch": 0.0013056, "grad_norm": 0.8422907590866089, "learning_rate": 3.2846352775880997e-07, "loss": 2.5699, "step": 358710 }, { "epoch": 0.0013312, "grad_norm": 0.8549548387527466, "learning_rate": 3.2825886337571333e-07, "loss": 2.5511, "step": 358720 }, { "epoch": 0.0013568, "grad_norm": 0.8354431986808777, "learning_rate": 3.2805426171128695e-07, "loss": 2.3183, "step": 358730 }, { "epoch": 0.0013824, "grad_norm": 0.8062424063682556, "learning_rate": 3.2784972276685976e-07, "loss": 2.718, "step": 358740 }, { "epoch": 0.001408, "grad_norm": 0.8195026516914368, "learning_rate": 3.276452465437552e-07, "loss": 2.8011, "step": 358750 }, { "epoch": 0.0014336, "grad_norm": 0.8760941624641418, "learning_rate": 3.27440833043301e-07, "loss": 2.5784, "step": 358760 }, { "epoch": 0.0014592, "grad_norm": 0.8201715350151062, "learning_rate": 3.272364822668228e-07, "loss": 2.573, "step": 358770 }, { "epoch": 0.0014848, "grad_norm": 0.8026418685913086, "learning_rate": 3.27032194215644e-07, "loss": 2.4114, "step": 358780 }, { "epoch": 0.0015104, "grad_norm": 1.0675989389419556, "learning_rate": 3.268279688910914e-07, "loss": 2.4178, "step": 358790 }, { "epoch": 0.001536, "grad_norm": 0.848239541053772, "learning_rate": 3.266238062944893e-07, "loss": 2.6308, "step": 358800 }, { "epoch": 2.56e-05, "grad_norm": 0.8417665362358093, "learning_rate": 3.2641970642716123e-07, "loss": 1.7056, "step": 358810 }, { "epoch": 5.12e-05, "grad_norm": 0.8012857437133789, "learning_rate": 3.2621566929042947e-07, "loss": 2.6713, "step": 358820 }, { "epoch": 7.68e-05, "grad_norm": 0.7945088148117065, "learning_rate": 3.2601169488561736e-07, "loss": 2.5496, "step": 358830 }, { "epoch": 0.0001024, "grad_norm": 0.8053775429725647, "learning_rate": 3.258077832140494e-07, "loss": 2.5292, "step": 358840 }, { "epoch": 0.000128, "grad_norm": 0.8516979217529297, "learning_rate": 3.256039342770467e-07, "loss": 2.585, "step": 358850 }, { "epoch": 0.0001536, "grad_norm": 0.9224601984024048, "learning_rate": 3.2540014807593057e-07, "loss": 2.6731, "step": 358860 }, { "epoch": 0.0001792, "grad_norm": 0.9824492931365967, "learning_rate": 3.2519642461202426e-07, "loss": 2.8836, "step": 358870 }, { "epoch": 0.0002048, "grad_norm": 1.225730061531067, "learning_rate": 3.249927638866468e-07, "loss": 2.4497, "step": 358880 }, { "epoch": 0.0002304, "grad_norm": 0.8841897249221802, "learning_rate": 3.247891659011204e-07, "loss": 2.5908, "step": 358890 }, { "epoch": 0.000256, "grad_norm": 0.9436725974082947, "learning_rate": 3.245856306567652e-07, "loss": 2.7134, "step": 358900 }, { "epoch": 0.0002816, "grad_norm": 0.8488925695419312, "learning_rate": 3.2438215815490114e-07, "loss": 2.6068, "step": 358910 }, { "epoch": 0.0003072, "grad_norm": 0.7930445671081543, "learning_rate": 3.2417874839684726e-07, "loss": 2.519, "step": 358920 }, { "epoch": 0.0003328, "grad_norm": 0.8329598903656006, "learning_rate": 3.2397540138392135e-07, "loss": 2.7438, "step": 358930 }, { "epoch": 0.0003584, "grad_norm": 1.0491772890090942, "learning_rate": 3.237721171174457e-07, "loss": 2.776, "step": 358940 }, { "epoch": 0.000384, "grad_norm": 0.8820163607597351, "learning_rate": 3.235688955987348e-07, "loss": 2.4745, "step": 358950 }, { "epoch": 0.0004096, "grad_norm": 0.8850710391998291, "learning_rate": 3.233657368291088e-07, "loss": 2.7509, "step": 358960 }, { "epoch": 0.0004352, "grad_norm": 0.9364029765129089, "learning_rate": 3.231626408098831e-07, "loss": 2.4835, "step": 358970 }, { "epoch": 0.0004608, "grad_norm": 0.851759672164917, "learning_rate": 3.2295960754237577e-07, "loss": 2.7464, "step": 358980 }, { "epoch": 0.0004864, "grad_norm": 0.9426130056381226, "learning_rate": 3.227566370279045e-07, "loss": 2.5004, "step": 358990 }, { "epoch": 0.000512, "grad_norm": 0.774100661277771, "learning_rate": 3.225537292677838e-07, "loss": 2.6752, "step": 359000 }, { "epoch": 0.0005376, "grad_norm": 0.8584672808647156, "learning_rate": 3.223508842633316e-07, "loss": 2.4963, "step": 359010 }, { "epoch": 0.0005632, "grad_norm": 0.8364156484603882, "learning_rate": 3.221481020158601e-07, "loss": 2.5469, "step": 359020 }, { "epoch": 0.0005888, "grad_norm": 0.8853554725646973, "learning_rate": 3.219453825266872e-07, "loss": 2.4289, "step": 359030 }, { "epoch": 0.0006144, "grad_norm": 0.7574223279953003, "learning_rate": 3.217427257971262e-07, "loss": 2.6292, "step": 359040 }, { "epoch": 0.00064, "grad_norm": 0.9947985410690308, "learning_rate": 3.2154013182848944e-07, "loss": 2.5855, "step": 359050 }, { "epoch": 0.0006656, "grad_norm": 0.878995954990387, "learning_rate": 3.2133760062209585e-07, "loss": 2.8596, "step": 359060 }, { "epoch": 0.0006912, "grad_norm": 0.8880777955055237, "learning_rate": 3.211351321792544e-07, "loss": 2.7024, "step": 359070 }, { "epoch": 0.0007168, "grad_norm": 0.8565109372138977, "learning_rate": 3.209327265012785e-07, "loss": 2.6759, "step": 359080 }, { "epoch": 0.0007424, "grad_norm": 0.9282662272453308, "learning_rate": 3.2073038358948263e-07, "loss": 2.8248, "step": 359090 }, { "epoch": 0.000768, "grad_norm": 0.7692101001739502, "learning_rate": 3.205281034451768e-07, "loss": 2.6209, "step": 359100 }, { "epoch": 0.0007936, "grad_norm": 0.8263222575187683, "learning_rate": 3.203258860696745e-07, "loss": 2.675, "step": 359110 }, { "epoch": 0.0008192, "grad_norm": 1.0831618309020996, "learning_rate": 3.201237314642858e-07, "loss": 2.6915, "step": 359120 }, { "epoch": 0.0008448, "grad_norm": 0.915712833404541, "learning_rate": 3.199216396303217e-07, "loss": 2.5606, "step": 359130 }, { "epoch": 0.0008704, "grad_norm": 0.8047309517860413, "learning_rate": 3.197196105690947e-07, "loss": 2.6169, "step": 359140 }, { "epoch": 0.000896, "grad_norm": 0.9269860982894897, "learning_rate": 3.1951764428191254e-07, "loss": 2.4507, "step": 359150 }, { "epoch": 0.0009216, "grad_norm": 0.8366005420684814, "learning_rate": 3.1931574077008533e-07, "loss": 2.5826, "step": 359160 }, { "epoch": 0.0009472, "grad_norm": 0.8931027054786682, "learning_rate": 3.19113900034923e-07, "loss": 2.5608, "step": 359170 }, { "epoch": 0.0009728, "grad_norm": 0.8224082589149475, "learning_rate": 3.189121220777336e-07, "loss": 2.4295, "step": 359180 }, { "epoch": 0.0009984, "grad_norm": 0.9380356669425964, "learning_rate": 3.18710406899827e-07, "loss": 2.8652, "step": 359190 }, { "epoch": 0.001024, "grad_norm": 1.01740562915802, "learning_rate": 3.1850875450250896e-07, "loss": 2.8171, "step": 359200 }, { "epoch": 0.0010496, "grad_norm": 0.9391812682151794, "learning_rate": 3.1830716488709166e-07, "loss": 2.6939, "step": 359210 }, { "epoch": 0.0010752, "grad_norm": 0.7981637716293335, "learning_rate": 3.181056380548753e-07, "loss": 2.6285, "step": 359220 }, { "epoch": 0.0011008, "grad_norm": 0.8326682448387146, "learning_rate": 3.179041740071731e-07, "loss": 2.6579, "step": 359230 }, { "epoch": 0.0011264, "grad_norm": 0.7707617878913879, "learning_rate": 3.177027727452886e-07, "loss": 2.5212, "step": 359240 }, { "epoch": 0.001152, "grad_norm": 0.8726404905319214, "learning_rate": 3.175014342705285e-07, "loss": 2.6388, "step": 359250 }, { "epoch": 0.0011776, "grad_norm": 0.8525384068489075, "learning_rate": 3.173001585841995e-07, "loss": 2.7794, "step": 359260 }, { "epoch": 0.0012032, "grad_norm": 0.8886842131614685, "learning_rate": 3.170989456876039e-07, "loss": 2.5702, "step": 359270 }, { "epoch": 0.0012288, "grad_norm": 0.8866816163063049, "learning_rate": 3.168977955820496e-07, "loss": 2.9141, "step": 359280 }, { "epoch": 0.0012544, "grad_norm": 0.8739024996757507, "learning_rate": 3.1669670826883883e-07, "loss": 2.5496, "step": 359290 }, { "epoch": 0.00128, "grad_norm": 0.8854517340660095, "learning_rate": 3.164956837492761e-07, "loss": 2.5754, "step": 359300 }, { "epoch": 0.0013056, "grad_norm": 0.9413154721260071, "learning_rate": 3.162947220246648e-07, "loss": 2.6395, "step": 359310 }, { "epoch": 0.0013312, "grad_norm": 0.9869586229324341, "learning_rate": 3.160938230963095e-07, "loss": 2.7294, "step": 359320 }, { "epoch": 0.0013568, "grad_norm": 0.8207823038101196, "learning_rate": 3.158929869655136e-07, "loss": 2.8831, "step": 359330 }, { "epoch": 0.0013824, "grad_norm": 0.8128343820571899, "learning_rate": 3.1569221363357606e-07, "loss": 2.8023, "step": 359340 }, { "epoch": 0.001408, "grad_norm": 0.8828743696212769, "learning_rate": 3.154915031018013e-07, "loss": 2.6654, "step": 359350 }, { "epoch": 0.0014336, "grad_norm": 0.86972576379776, "learning_rate": 3.1529085537148953e-07, "loss": 2.5992, "step": 359360 }, { "epoch": 0.0014592, "grad_norm": 0.8907178044319153, "learning_rate": 3.150902704439429e-07, "loss": 2.6034, "step": 359370 }, { "epoch": 0.0014848, "grad_norm": 1.1150931119918823, "learning_rate": 3.148897483204627e-07, "loss": 2.1484, "step": 359380 }, { "epoch": 0.0015104, "grad_norm": 0.7729915976524353, "learning_rate": 3.1468928900234785e-07, "loss": 2.1316, "step": 359390 }, { "epoch": 0.001536, "grad_norm": 0.8930898904800415, "learning_rate": 3.1448889249089843e-07, "loss": 2.7732, "step": 359400 }, { "epoch": 0.0015616, "grad_norm": 1.0040806531906128, "learning_rate": 3.142885587874156e-07, "loss": 2.6981, "step": 359410 }, { "epoch": 0.0015872, "grad_norm": 0.8532537817955017, "learning_rate": 3.1408828789319615e-07, "loss": 2.6212, "step": 359420 }, { "epoch": 0.0016128, "grad_norm": 0.9577513337135315, "learning_rate": 3.1388807980954007e-07, "loss": 2.413, "step": 359430 }, { "epoch": 0.0016384, "grad_norm": 0.9471181631088257, "learning_rate": 3.1368793453774635e-07, "loss": 2.693, "step": 359440 }, { "epoch": 0.001664, "grad_norm": 0.9530557990074158, "learning_rate": 3.134878520791118e-07, "loss": 2.829, "step": 359450 }, { "epoch": 0.0016896, "grad_norm": 0.8464002013206482, "learning_rate": 3.132878324349353e-07, "loss": 2.8318, "step": 359460 }, { "epoch": 0.0017152, "grad_norm": 0.897718608379364, "learning_rate": 3.1308787560651146e-07, "loss": 2.6024, "step": 359470 }, { "epoch": 0.0017408, "grad_norm": 0.9985032081604004, "learning_rate": 3.1288798159513803e-07, "loss": 2.6443, "step": 359480 }, { "epoch": 0.0017664, "grad_norm": 0.7625822424888611, "learning_rate": 3.126881504021118e-07, "loss": 2.6507, "step": 359490 }, { "epoch": 0.001792, "grad_norm": 0.8393094539642334, "learning_rate": 3.124883820287283e-07, "loss": 2.4978, "step": 359500 }, { "epoch": 0.0018176, "grad_norm": 0.8131115436553955, "learning_rate": 3.1228867647628336e-07, "loss": 2.7358, "step": 359510 }, { "epoch": 0.0018432, "grad_norm": 0.9028803706169128, "learning_rate": 3.1208903374607135e-07, "loss": 2.6684, "step": 359520 }, { "epoch": 0.0018688, "grad_norm": 0.8536331653594971, "learning_rate": 3.1188945383938794e-07, "loss": 2.6212, "step": 359530 }, { "epoch": 0.0018944, "grad_norm": 1.7030664682388306, "learning_rate": 3.116899367575266e-07, "loss": 2.6061, "step": 359540 }, { "epoch": 0.00192, "grad_norm": 0.8239572644233704, "learning_rate": 3.114904825017806e-07, "loss": 2.635, "step": 359550 }, { "epoch": 0.0019456, "grad_norm": 0.7948781847953796, "learning_rate": 3.1129109107344344e-07, "loss": 2.65, "step": 359560 }, { "epoch": 0.0019712, "grad_norm": 1.2644366025924683, "learning_rate": 3.1109176247380747e-07, "loss": 2.474, "step": 359570 }, { "epoch": 0.0019968, "grad_norm": 0.831080436706543, "learning_rate": 3.108924967041682e-07, "loss": 2.7243, "step": 359580 }, { "epoch": 0.0020224, "grad_norm": 0.8609279990196228, "learning_rate": 3.106932937658169e-07, "loss": 2.6133, "step": 359590 }, { "epoch": 0.002048, "grad_norm": 0.7533696293830872, "learning_rate": 3.1049415366004365e-07, "loss": 2.4093, "step": 359600 }, { "epoch": 0.0020736, "grad_norm": 0.8374148607254028, "learning_rate": 3.102950763881396e-07, "loss": 2.5559, "step": 359610 }, { "epoch": 0.0020992, "grad_norm": 0.8053135275840759, "learning_rate": 3.1009606195139816e-07, "loss": 2.6207, "step": 359620 }, { "epoch": 0.0021248, "grad_norm": 0.9636070728302002, "learning_rate": 3.0989711035110727e-07, "loss": 2.6404, "step": 359630 }, { "epoch": 0.0021504, "grad_norm": 0.9036421775817871, "learning_rate": 3.0969822158855913e-07, "loss": 2.7735, "step": 359640 }, { "epoch": 0.002176, "grad_norm": 0.9097692370414734, "learning_rate": 3.094993956650427e-07, "loss": 2.8133, "step": 359650 }, { "epoch": 0.0022016, "grad_norm": 0.8667019605636597, "learning_rate": 3.0930063258184817e-07, "loss": 2.6767, "step": 359660 }, { "epoch": 0.0022272, "grad_norm": 1.1094435453414917, "learning_rate": 3.091019323402622e-07, "loss": 2.4593, "step": 359670 }, { "epoch": 0.0022528, "grad_norm": 0.8580755591392517, "learning_rate": 3.089032949415749e-07, "loss": 2.7005, "step": 359680 }, { "epoch": 0.0022784, "grad_norm": 0.8627166152000427, "learning_rate": 3.087047203870741e-07, "loss": 2.2982, "step": 359690 }, { "epoch": 0.002304, "grad_norm": 1.068751573562622, "learning_rate": 3.085062086780477e-07, "loss": 1.9628, "step": 359700 }, { "epoch": 0.0023296, "grad_norm": 0.888016402721405, "learning_rate": 3.083077598157824e-07, "loss": 2.5795, "step": 359710 }, { "epoch": 0.0023552, "grad_norm": 0.8710764646530151, "learning_rate": 3.0810937380156615e-07, "loss": 2.4785, "step": 359720 }, { "epoch": 0.0023808, "grad_norm": 0.9347205758094788, "learning_rate": 3.079110506366845e-07, "loss": 2.7425, "step": 359730 }, { "epoch": 0.0024064, "grad_norm": 0.8963475823402405, "learning_rate": 3.077127903224231e-07, "loss": 2.6065, "step": 359740 }, { "epoch": 0.002432, "grad_norm": 0.8825911283493042, "learning_rate": 3.075145928600687e-07, "loss": 2.8102, "step": 359750 }, { "epoch": 0.0024576, "grad_norm": 0.8820439577102661, "learning_rate": 3.073164582509069e-07, "loss": 2.7365, "step": 359760 }, { "epoch": 0.0024832, "grad_norm": 0.853210985660553, "learning_rate": 3.0711838649622126e-07, "loss": 2.6032, "step": 359770 }, { "epoch": 0.0025088, "grad_norm": 0.8183149099349976, "learning_rate": 3.069203775972973e-07, "loss": 2.8035, "step": 359780 }, { "epoch": 0.0025344, "grad_norm": 0.8187352418899536, "learning_rate": 3.067224315554196e-07, "loss": 2.6023, "step": 359790 }, { "epoch": 0.00256, "grad_norm": 0.8659395575523376, "learning_rate": 3.065245483718693e-07, "loss": 2.8595, "step": 359800 }, { "epoch": 0.0025856, "grad_norm": 1.00346839427948, "learning_rate": 3.0632672804793207e-07, "loss": 2.6752, "step": 359810 }, { "epoch": 0.0026112, "grad_norm": 0.8123335838317871, "learning_rate": 3.06128970584888e-07, "loss": 2.7672, "step": 359820 }, { "epoch": 0.0026368, "grad_norm": 0.8458513617515564, "learning_rate": 3.0593127598402275e-07, "loss": 2.9252, "step": 359830 }, { "epoch": 0.0026624, "grad_norm": 0.9297922253608704, "learning_rate": 3.0573364424661635e-07, "loss": 2.8447, "step": 359840 }, { "epoch": 0.002688, "grad_norm": 0.8177646994590759, "learning_rate": 3.0553607537395115e-07, "loss": 2.9062, "step": 359850 }, { "epoch": 0.0027136, "grad_norm": 0.9421570897102356, "learning_rate": 3.053385693673083e-07, "loss": 2.4338, "step": 359860 }, { "epoch": 0.0027392, "grad_norm": 0.8475180864334106, "learning_rate": 3.05141126227968e-07, "loss": 2.961, "step": 359870 }, { "epoch": 0.0027648, "grad_norm": 0.8967019319534302, "learning_rate": 3.0494374595721024e-07, "loss": 2.7242, "step": 359880 }, { "epoch": 0.0027904, "grad_norm": 0.9397884011268616, "learning_rate": 3.047464285563162e-07, "loss": 2.9355, "step": 359890 }, { "epoch": 0.002816, "grad_norm": 1.024614930152893, "learning_rate": 3.0454917402656383e-07, "loss": 2.9184, "step": 359900 }, { "epoch": 0.0028416, "grad_norm": 0.8296196460723877, "learning_rate": 3.0435198236923425e-07, "loss": 2.7439, "step": 359910 }, { "epoch": 0.0028672, "grad_norm": 0.855284571647644, "learning_rate": 3.041548535856054e-07, "loss": 2.7969, "step": 359920 }, { "epoch": 0.0028928, "grad_norm": 0.8704053163528442, "learning_rate": 3.0395778767695503e-07, "loss": 2.8627, "step": 359930 }, { "epoch": 0.0029184, "grad_norm": 0.8008702993392944, "learning_rate": 3.0376078464456227e-07, "loss": 2.8898, "step": 359940 }, { "epoch": 0.002944, "grad_norm": 0.8148054480552673, "learning_rate": 3.0356384448970264e-07, "loss": 2.7564, "step": 359950 }, { "epoch": 0.0029696, "grad_norm": 0.8231488466262817, "learning_rate": 3.033669672136552e-07, "loss": 2.7552, "step": 359960 }, { "epoch": 0.0029952, "grad_norm": 0.9206733107566833, "learning_rate": 3.031701528176956e-07, "loss": 2.4348, "step": 359970 }, { "epoch": 0.0030208, "grad_norm": 0.816474974155426, "learning_rate": 3.029734013031016e-07, "loss": 2.8977, "step": 359980 }, { "epoch": 0.0030464, "grad_norm": 0.886938214302063, "learning_rate": 3.027767126711478e-07, "loss": 2.8724, "step": 359990 }, { "epoch": 0.003072, "grad_norm": 0.8220286965370178, "learning_rate": 3.025800869231088e-07, "loss": 2.8241, "step": 360000 }, { "epoch": 0.0030976, "grad_norm": 0.9053717851638794, "learning_rate": 3.0238352406026127e-07, "loss": 2.8077, "step": 360010 }, { "epoch": 0.0031232, "grad_norm": 0.9178534746170044, "learning_rate": 3.021870240838787e-07, "loss": 2.7514, "step": 360020 }, { "epoch": 0.0031488, "grad_norm": 0.7121727466583252, "learning_rate": 3.0199058699523664e-07, "loss": 2.6403, "step": 360030 }, { "epoch": 0.0031744, "grad_norm": 0.9093044400215149, "learning_rate": 3.017942127956075e-07, "loss": 2.7333, "step": 360040 }, { "epoch": 0.0032, "grad_norm": 1.0111970901489258, "learning_rate": 3.015979014862658e-07, "loss": 2.8222, "step": 360050 }, { "epoch": 0.0032256, "grad_norm": 0.8983656764030457, "learning_rate": 3.014016530684849e-07, "loss": 2.8261, "step": 360060 }, { "epoch": 0.0032512, "grad_norm": 0.8156994581222534, "learning_rate": 3.0120546754353496e-07, "loss": 2.7494, "step": 360070 }, { "epoch": 0.0032768, "grad_norm": 0.9350175261497498, "learning_rate": 3.0100934491269163e-07, "loss": 2.5101, "step": 360080 }, { "epoch": 0.0033024, "grad_norm": 0.8872793912887573, "learning_rate": 3.0081328517722275e-07, "loss": 2.9394, "step": 360090 }, { "epoch": 0.003328, "grad_norm": 0.9085145592689514, "learning_rate": 3.0061728833840285e-07, "loss": 2.8196, "step": 360100 }, { "epoch": 0.0033536, "grad_norm": 0.8659526705741882, "learning_rate": 3.004213543975032e-07, "loss": 2.7438, "step": 360110 }, { "epoch": 0.0033792, "grad_norm": 0.8568825125694275, "learning_rate": 3.002254833557938e-07, "loss": 2.5294, "step": 360120 }, { "epoch": 0.0034048, "grad_norm": 0.9514431953430176, "learning_rate": 3.0002967521454265e-07, "loss": 2.8026, "step": 360130 }, { "epoch": 0.0034304, "grad_norm": 0.8318968415260315, "learning_rate": 2.998339299750219e-07, "loss": 2.7769, "step": 360140 }, { "epoch": 0.003456, "grad_norm": 0.872972309589386, "learning_rate": 2.9963824763849957e-07, "loss": 2.694, "step": 360150 }, { "epoch": 0.0034816, "grad_norm": 0.8290106058120728, "learning_rate": 2.9944262820624457e-07, "loss": 2.6466, "step": 360160 }, { "epoch": 0.0035072, "grad_norm": 0.8807954788208008, "learning_rate": 2.9924707167952705e-07, "loss": 2.7524, "step": 360170 }, { "epoch": 0.0035328, "grad_norm": 0.8957972526550293, "learning_rate": 2.9905157805961373e-07, "loss": 2.72, "step": 360180 }, { "epoch": 0.0035584, "grad_norm": 0.8713822364807129, "learning_rate": 2.988561473477747e-07, "loss": 2.8788, "step": 360190 }, { "epoch": 0.003584, "grad_norm": 0.8836238980293274, "learning_rate": 2.986607795452734e-07, "loss": 2.734, "step": 360200 }, { "epoch": 0.0036096, "grad_norm": 0.8741977214813232, "learning_rate": 2.984654746533788e-07, "loss": 2.9995, "step": 360210 }, { "epoch": 0.0036352, "grad_norm": 0.9122756719589233, "learning_rate": 2.9827023267335665e-07, "loss": 2.8194, "step": 360220 }, { "epoch": 0.0036608, "grad_norm": 0.8993634581565857, "learning_rate": 2.980750536064736e-07, "loss": 2.9386, "step": 360230 }, { "epoch": 0.0036864, "grad_norm": 0.9867888689041138, "learning_rate": 2.978799374539965e-07, "loss": 3.0274, "step": 360240 }, { "epoch": 0.003712, "grad_norm": 0.8385710716247559, "learning_rate": 2.9768488421718866e-07, "loss": 2.851, "step": 360250 }, { "epoch": 0.0037376, "grad_norm": 0.9463236331939697, "learning_rate": 2.97489893897317e-07, "loss": 2.8273, "step": 360260 }, { "epoch": 0.0037632, "grad_norm": 0.8528454303741455, "learning_rate": 2.972949664956426e-07, "loss": 2.8701, "step": 360270 }, { "epoch": 0.0037888, "grad_norm": 0.8469123840332031, "learning_rate": 2.9710010201343345e-07, "loss": 2.7694, "step": 360280 }, { "epoch": 0.0038144, "grad_norm": 0.8040942549705505, "learning_rate": 2.9690530045195066e-07, "loss": 2.6514, "step": 360290 }, { "epoch": 0.00384, "grad_norm": 0.85391765832901, "learning_rate": 2.9671056181245774e-07, "loss": 2.8881, "step": 360300 }, { "epoch": 0.0038656, "grad_norm": 0.8638335466384888, "learning_rate": 2.965158860962192e-07, "loss": 2.5943, "step": 360310 }, { "epoch": 0.0038912, "grad_norm": 0.879356861114502, "learning_rate": 2.9632127330449623e-07, "loss": 2.7253, "step": 360320 }, { "epoch": 0.0039168, "grad_norm": 0.8682277202606201, "learning_rate": 2.961267234385501e-07, "loss": 2.7617, "step": 360330 }, { "epoch": 0.0039424, "grad_norm": 0.8731677532196045, "learning_rate": 2.9593223649964307e-07, "loss": 2.8137, "step": 360340 }, { "epoch": 0.003968, "grad_norm": 1.1599479913711548, "learning_rate": 2.957378124890375e-07, "loss": 2.6824, "step": 360350 }, { "epoch": 0.0039936, "grad_norm": 0.8377076983451843, "learning_rate": 2.9554345140799246e-07, "loss": 2.7836, "step": 360360 }, { "epoch": 0.0040192, "grad_norm": 0.7812326550483704, "learning_rate": 2.95349153257769e-07, "loss": 2.7891, "step": 360370 }, { "epoch": 0.0040448, "grad_norm": 0.8942041397094727, "learning_rate": 2.9515491803962626e-07, "loss": 2.7372, "step": 360380 }, { "epoch": 0.0040704, "grad_norm": 0.838847279548645, "learning_rate": 2.9496074575482647e-07, "loss": 2.7455, "step": 360390 }, { "epoch": 0.004096, "grad_norm": 0.880936324596405, "learning_rate": 2.9476663640462643e-07, "loss": 3.0047, "step": 360400 }, { "epoch": 0.0041216, "grad_norm": 1.4484531879425049, "learning_rate": 2.9457258999028516e-07, "loss": 2.7113, "step": 360410 }, { "epoch": 0.0041472, "grad_norm": 0.9129427075386047, "learning_rate": 2.943786065130605e-07, "loss": 2.7713, "step": 360420 }, { "epoch": 0.0041728, "grad_norm": 0.7459953427314758, "learning_rate": 2.9418468597421146e-07, "loss": 2.7091, "step": 360430 }, { "epoch": 0.0041984, "grad_norm": 0.8639402389526367, "learning_rate": 2.939908283749948e-07, "loss": 2.7577, "step": 360440 }, { "epoch": 0.004224, "grad_norm": 0.7746111154556274, "learning_rate": 2.937970337166696e-07, "loss": 2.7685, "step": 360450 }, { "epoch": 0.0042496, "grad_norm": 0.861749529838562, "learning_rate": 2.9360330200049027e-07, "loss": 2.9866, "step": 360460 }, { "epoch": 0.0042752, "grad_norm": 0.8284888863563538, "learning_rate": 2.9340963322771366e-07, "loss": 2.6324, "step": 360470 }, { "epoch": 0.0043008, "grad_norm": 0.9376768469810486, "learning_rate": 2.9321602739959543e-07, "loss": 2.8753, "step": 360480 }, { "epoch": 0.0043264, "grad_norm": 0.8358775973320007, "learning_rate": 2.9302248451739236e-07, "loss": 2.7172, "step": 360490 }, { "epoch": 0.004352, "grad_norm": 0.9004737734794617, "learning_rate": 2.9282900458235786e-07, "loss": 2.8399, "step": 360500 }, { "epoch": 0.0043776, "grad_norm": 0.8491239547729492, "learning_rate": 2.9263558759574764e-07, "loss": 2.6549, "step": 360510 }, { "epoch": 0.0044032, "grad_norm": 0.9446026682853699, "learning_rate": 2.9244223355881727e-07, "loss": 2.9635, "step": 360520 }, { "epoch": 0.0044288, "grad_norm": 0.847048819065094, "learning_rate": 2.922489424728181e-07, "loss": 2.827, "step": 360530 }, { "epoch": 0.0044544, "grad_norm": 0.8145902752876282, "learning_rate": 2.920557143390035e-07, "loss": 2.6853, "step": 360540 }, { "epoch": 0.00448, "grad_norm": 0.8119754791259766, "learning_rate": 2.918625491586291e-07, "loss": 2.7938, "step": 360550 }, { "epoch": 0.0045056, "grad_norm": 0.851121723651886, "learning_rate": 2.9166944693294397e-07, "loss": 2.9339, "step": 360560 }, { "epoch": 0.0045312, "grad_norm": 0.80475252866745, "learning_rate": 2.914764076632037e-07, "loss": 2.6134, "step": 360570 }, { "epoch": 0.0045568, "grad_norm": 0.8117938041687012, "learning_rate": 2.912834313506585e-07, "loss": 2.7946, "step": 360580 }, { "epoch": 0.0045824, "grad_norm": 1.166170597076416, "learning_rate": 2.9109051799656064e-07, "loss": 2.7888, "step": 360590 }, { "epoch": 0.004608, "grad_norm": 0.8343316316604614, "learning_rate": 2.908976676021602e-07, "loss": 2.8051, "step": 360600 }, { "epoch": 0.0046336, "grad_norm": 1.7602877616882324, "learning_rate": 2.9070488016870625e-07, "loss": 2.3261, "step": 360610 }, { "epoch": 0.0046592, "grad_norm": 1.2897717952728271, "learning_rate": 2.905121556974522e-07, "loss": 2.8522, "step": 360620 }, { "epoch": 0.0046848, "grad_norm": 1.0949134826660156, "learning_rate": 2.903194941896459e-07, "loss": 2.9978, "step": 360630 }, { "epoch": 0.0047104, "grad_norm": 0.8052533864974976, "learning_rate": 2.9012689564653864e-07, "loss": 2.8333, "step": 360640 }, { "epoch": 0.004736, "grad_norm": 0.825786828994751, "learning_rate": 2.89934360069376e-07, "loss": 2.7313, "step": 360650 }, { "epoch": 0.0047616, "grad_norm": 0.9032337069511414, "learning_rate": 2.8974188745940935e-07, "loss": 3.1138, "step": 360660 }, { "epoch": 0.0047872, "grad_norm": 0.8325718641281128, "learning_rate": 2.8954947781788533e-07, "loss": 2.8151, "step": 360670 }, { "epoch": 0.0048128, "grad_norm": 0.8606631755828857, "learning_rate": 2.89357131146053e-07, "loss": 2.6786, "step": 360680 }, { "epoch": 0.0048384, "grad_norm": 0.8360366225242615, "learning_rate": 2.8916484744515804e-07, "loss": 2.7977, "step": 360690 }, { "epoch": 0.004864, "grad_norm": 0.8512260913848877, "learning_rate": 2.8897262671644834e-07, "loss": 2.8019, "step": 360700 }, { "epoch": 0.0048896, "grad_norm": 0.8534073233604431, "learning_rate": 2.887804689611706e-07, "loss": 2.8451, "step": 360710 }, { "epoch": 0.0049152, "grad_norm": 1.0104398727416992, "learning_rate": 2.8858837418057175e-07, "loss": 2.5432, "step": 360720 }, { "epoch": 0.0049408, "grad_norm": 0.7914049625396729, "learning_rate": 2.883963423758951e-07, "loss": 2.5727, "step": 360730 }, { "epoch": 0.0049664, "grad_norm": 0.8588154911994934, "learning_rate": 2.8820437354838637e-07, "loss": 2.7626, "step": 360740 }, { "epoch": 0.004992, "grad_norm": 0.9201818704605103, "learning_rate": 2.8801246769929235e-07, "loss": 2.7481, "step": 360750 }, { "epoch": 0.0050176, "grad_norm": 0.8109952211380005, "learning_rate": 2.878206248298554e-07, "loss": 2.8056, "step": 360760 }, { "epoch": 0.0050432, "grad_norm": 0.8274464011192322, "learning_rate": 2.8762884494132114e-07, "loss": 2.7616, "step": 360770 }, { "epoch": 0.0050688, "grad_norm": 0.8719793558120728, "learning_rate": 2.8743712803493306e-07, "loss": 2.7403, "step": 360780 }, { "epoch": 0.0050944, "grad_norm": 0.7928882241249084, "learning_rate": 2.8724547411193236e-07, "loss": 2.7708, "step": 360790 }, { "epoch": 0.00512, "grad_norm": 0.801089882850647, "learning_rate": 2.8705388317356475e-07, "loss": 2.9084, "step": 360800 }, { "epoch": 0.0051456, "grad_norm": 0.8250362873077393, "learning_rate": 2.868623552210703e-07, "loss": 2.8177, "step": 360810 }, { "epoch": 0.0051712, "grad_norm": 0.829119086265564, "learning_rate": 2.8667089025569364e-07, "loss": 2.9107, "step": 360820 }, { "epoch": 0.0051968, "grad_norm": 0.8788798451423645, "learning_rate": 2.864794882786737e-07, "loss": 2.8439, "step": 360830 }, { "epoch": 0.0052224, "grad_norm": 0.908197820186615, "learning_rate": 2.862881492912528e-07, "loss": 2.9291, "step": 360840 }, { "epoch": 0.005248, "grad_norm": 0.8674789071083069, "learning_rate": 2.8609687329467227e-07, "loss": 2.8689, "step": 360850 }, { "epoch": 0.0052736, "grad_norm": 1.143129587173462, "learning_rate": 2.8590566029017107e-07, "loss": 2.7044, "step": 360860 }, { "epoch": 0.0052992, "grad_norm": 0.9600434303283691, "learning_rate": 2.8571451027899046e-07, "loss": 2.8309, "step": 360870 }, { "epoch": 0.0053248, "grad_norm": 0.7707140445709229, "learning_rate": 2.8552342326236825e-07, "loss": 2.897, "step": 360880 }, { "epoch": 0.0053504, "grad_norm": 0.8126978874206543, "learning_rate": 2.853323992415458e-07, "loss": 2.6777, "step": 360890 }, { "epoch": 0.005376, "grad_norm": 0.7974080443382263, "learning_rate": 2.8514143821776084e-07, "loss": 2.482, "step": 360900 }, { "epoch": 0.0054016, "grad_norm": 0.9264965653419495, "learning_rate": 2.8495054019225257e-07, "loss": 2.7365, "step": 360910 }, { "epoch": 0.0054272, "grad_norm": 0.7926034927368164, "learning_rate": 2.847597051662576e-07, "loss": 2.5801, "step": 360920 }, { "epoch": 0.0054528, "grad_norm": 0.8386869430541992, "learning_rate": 2.8456893314101395e-07, "loss": 2.6843, "step": 360930 }, { "epoch": 0.0054784, "grad_norm": 0.831028938293457, "learning_rate": 2.843782241177584e-07, "loss": 2.8274, "step": 360940 }, { "epoch": 0.005504, "grad_norm": 0.8366524577140808, "learning_rate": 2.841875780977277e-07, "loss": 2.8362, "step": 360950 }, { "epoch": 0.0055296, "grad_norm": 0.7840197682380676, "learning_rate": 2.839969950821575e-07, "loss": 2.7, "step": 360960 }, { "epoch": 0.0055552, "grad_norm": 0.8200573921203613, "learning_rate": 2.8380647507228687e-07, "loss": 2.7851, "step": 360970 }, { "epoch": 0.0055808, "grad_norm": 0.834202766418457, "learning_rate": 2.836160180693481e-07, "loss": 2.8588, "step": 360980 }, { "epoch": 0.0056064, "grad_norm": 0.8536860942840576, "learning_rate": 2.8342562407457696e-07, "loss": 2.8422, "step": 360990 }, { "epoch": 0.005632, "grad_norm": 0.7783385515213013, "learning_rate": 2.8323529308920793e-07, "loss": 2.7603, "step": 361000 }, { "epoch": 0.0056576, "grad_norm": 0.8732569813728333, "learning_rate": 2.8304502511447676e-07, "loss": 2.8298, "step": 361010 }, { "epoch": 0.0056832, "grad_norm": 0.8388901948928833, "learning_rate": 2.828548201516146e-07, "loss": 2.8058, "step": 361020 }, { "epoch": 0.0057088, "grad_norm": 0.869281530380249, "learning_rate": 2.8266467820185716e-07, "loss": 2.9644, "step": 361030 }, { "epoch": 0.0057344, "grad_norm": 1.0273082256317139, "learning_rate": 2.8247459926643794e-07, "loss": 2.9011, "step": 361040 }, { "epoch": 0.00576, "grad_norm": 0.8907639980316162, "learning_rate": 2.8228458334658705e-07, "loss": 2.626, "step": 361050 }, { "epoch": 0.0057856, "grad_norm": 0.9746415019035339, "learning_rate": 2.820946304435379e-07, "loss": 2.7733, "step": 361060 }, { "epoch": 0.0058112, "grad_norm": 0.8511093258857727, "learning_rate": 2.8190474055852293e-07, "loss": 2.7212, "step": 361070 }, { "epoch": 0.0058368, "grad_norm": 0.876373827457428, "learning_rate": 2.8171491369277213e-07, "loss": 2.8117, "step": 361080 }, { "epoch": 0.0058624, "grad_norm": 0.8146148920059204, "learning_rate": 2.81525149847518e-07, "loss": 2.7903, "step": 361090 }, { "epoch": 0.005888, "grad_norm": 0.8214396238327026, "learning_rate": 2.8133544902398947e-07, "loss": 2.7095, "step": 361100 }, { "epoch": 0.0059136, "grad_norm": 0.8023264408111572, "learning_rate": 2.8114581122341886e-07, "loss": 2.6858, "step": 361110 }, { "epoch": 0.0059392, "grad_norm": 0.8637736439704895, "learning_rate": 2.809562364470353e-07, "loss": 2.8731, "step": 361120 }, { "epoch": 0.0059648, "grad_norm": 0.9704412817955017, "learning_rate": 2.807667246960666e-07, "loss": 2.9219, "step": 361130 }, { "epoch": 0.0059904, "grad_norm": 0.7807960510253906, "learning_rate": 2.8057727597174066e-07, "loss": 2.8797, "step": 361140 }, { "epoch": 0.006016, "grad_norm": 0.8001233339309692, "learning_rate": 2.803878902752899e-07, "loss": 2.8744, "step": 361150 }, { "epoch": 0.0060416, "grad_norm": 0.8039575815200806, "learning_rate": 2.8019856760794e-07, "loss": 2.8287, "step": 361160 }, { "epoch": 0.0060672, "grad_norm": 0.7775998711585999, "learning_rate": 2.8000930797092094e-07, "loss": 2.6211, "step": 361170 }, { "epoch": 0.0060928, "grad_norm": 0.7976440191268921, "learning_rate": 2.798201113654564e-07, "loss": 2.6822, "step": 361180 }, { "epoch": 0.0061184, "grad_norm": 0.8650289177894592, "learning_rate": 2.7963097779277527e-07, "loss": 2.857, "step": 361190 }, { "epoch": 0.006144, "grad_norm": 0.8899646401405334, "learning_rate": 2.794419072541044e-07, "loss": 2.7231, "step": 361200 }, { "epoch": 0.0061696, "grad_norm": 0.8810059428215027, "learning_rate": 2.792528997506694e-07, "loss": 2.828, "step": 361210 }, { "epoch": 0.0061952, "grad_norm": 0.8364047408103943, "learning_rate": 2.790639552836949e-07, "loss": 2.8349, "step": 361220 }, { "epoch": 0.0062208, "grad_norm": 0.8786189556121826, "learning_rate": 2.788750738544088e-07, "loss": 2.6854, "step": 361230 }, { "epoch": 0.0062464, "grad_norm": 0.7565369009971619, "learning_rate": 2.786862554640335e-07, "loss": 2.6058, "step": 361240 }, { "epoch": 0.006272, "grad_norm": 0.8947702050209045, "learning_rate": 2.7849750011379464e-07, "loss": 2.8554, "step": 361250 }, { "epoch": 0.0062976, "grad_norm": 0.9016766548156738, "learning_rate": 2.783088078049145e-07, "loss": 2.6442, "step": 361260 }, { "epoch": 0.0063232, "grad_norm": 0.7550061941146851, "learning_rate": 2.7812017853861895e-07, "loss": 2.8815, "step": 361270 }, { "epoch": 0.0063488, "grad_norm": 0.8467217683792114, "learning_rate": 2.779316123161302e-07, "loss": 2.866, "step": 361280 }, { "epoch": 0.0063744, "grad_norm": 0.8938449621200562, "learning_rate": 2.777431091386706e-07, "loss": 2.6223, "step": 361290 }, { "epoch": 0.0064, "grad_norm": 0.8937451243400574, "learning_rate": 2.7755466900746373e-07, "loss": 2.628, "step": 361300 }, { "epoch": 0.0064256, "grad_norm": 0.848920464515686, "learning_rate": 2.7736629192373077e-07, "loss": 2.7832, "step": 361310 }, { "epoch": 0.0064512, "grad_norm": 0.8852716684341431, "learning_rate": 2.7717797788869296e-07, "loss": 2.7712, "step": 361320 }, { "epoch": 0.0064768, "grad_norm": 0.8263595700263977, "learning_rate": 2.769897269035726e-07, "loss": 2.7705, "step": 361330 }, { "epoch": 0.0065024, "grad_norm": 0.8363216519355774, "learning_rate": 2.7680153896958883e-07, "loss": 2.9335, "step": 361340 }, { "epoch": 0.006528, "grad_norm": 0.8405662775039673, "learning_rate": 2.7661341408796393e-07, "loss": 2.6781, "step": 361350 }, { "epoch": 0.0065536, "grad_norm": 0.831912100315094, "learning_rate": 2.7642535225991695e-07, "loss": 2.7103, "step": 361360 }, { "epoch": 0.0065792, "grad_norm": 0.9253130555152893, "learning_rate": 2.7623735348666804e-07, "loss": 2.761, "step": 361370 }, { "epoch": 0.0066048, "grad_norm": 0.8283244967460632, "learning_rate": 2.7604941776943397e-07, "loss": 2.7708, "step": 361380 }, { "epoch": 0.0066304, "grad_norm": 0.8126504421234131, "learning_rate": 2.7586154510943594e-07, "loss": 2.5488, "step": 361390 }, { "epoch": 0.006656, "grad_norm": 0.8432349562644958, "learning_rate": 2.756737355078909e-07, "loss": 2.7974, "step": 361400 }, { "epoch": 0.0066816, "grad_norm": 0.8900271058082581, "learning_rate": 2.754859889660177e-07, "loss": 2.8148, "step": 361410 }, { "epoch": 0.0067072, "grad_norm": 0.8590095639228821, "learning_rate": 2.7529830548503333e-07, "loss": 2.8757, "step": 361420 }, { "epoch": 0.0067328, "grad_norm": 0.8895487189292908, "learning_rate": 2.7511068506615447e-07, "loss": 2.7973, "step": 361430 }, { "epoch": 0.0067584, "grad_norm": 0.810936450958252, "learning_rate": 2.749231277106001e-07, "loss": 2.8469, "step": 361440 }, { "epoch": 0.006784, "grad_norm": 0.8260511755943298, "learning_rate": 2.7473563341958277e-07, "loss": 2.6505, "step": 361450 }, { "epoch": 0.0068096, "grad_norm": 0.7953823804855347, "learning_rate": 2.7454820219432023e-07, "loss": 2.6844, "step": 361460 }, { "epoch": 0.0068352, "grad_norm": 0.890000581741333, "learning_rate": 2.743608340360282e-07, "loss": 2.7401, "step": 361470 }, { "epoch": 0.0068608, "grad_norm": 0.8004732131958008, "learning_rate": 2.741735289459202e-07, "loss": 2.707, "step": 361480 }, { "epoch": 0.0068864, "grad_norm": 1.0043948888778687, "learning_rate": 2.739862869252141e-07, "loss": 2.7231, "step": 361490 }, { "epoch": 0.006912, "grad_norm": 0.8983734846115112, "learning_rate": 2.7379910797512235e-07, "loss": 2.9624, "step": 361500 }, { "epoch": 0.0069376, "grad_norm": 0.7982003688812256, "learning_rate": 2.7361199209685716e-07, "loss": 2.8388, "step": 361510 }, { "epoch": 0.0069632, "grad_norm": 2.0876998901367188, "learning_rate": 2.734249392916333e-07, "loss": 3.0251, "step": 361520 }, { "epoch": 0.0069888, "grad_norm": 0.947895884513855, "learning_rate": 2.7323794956066297e-07, "loss": 2.6907, "step": 361530 }, { "epoch": 0.0070144, "grad_norm": 0.8373022079467773, "learning_rate": 2.730510229051608e-07, "loss": 2.8162, "step": 361540 }, { "epoch": 0.00704, "grad_norm": 0.7915387153625488, "learning_rate": 2.7286415932633704e-07, "loss": 2.8121, "step": 361550 }, { "epoch": 0.0070656, "grad_norm": 0.8502511382102966, "learning_rate": 2.7267735882540394e-07, "loss": 2.5735, "step": 361560 }, { "epoch": 0.0070912, "grad_norm": 0.8031792640686035, "learning_rate": 2.7249062140357494e-07, "loss": 2.6658, "step": 361570 }, { "epoch": 0.0071168, "grad_norm": 0.8288913369178772, "learning_rate": 2.7230394706205696e-07, "loss": 2.964, "step": 361580 }, { "epoch": 0.0071424, "grad_norm": 0.8076215982437134, "learning_rate": 2.721173358020623e-07, "loss": 2.617, "step": 361590 }, { "epoch": 0.007168, "grad_norm": 0.904761016368866, "learning_rate": 2.719307876248023e-07, "loss": 2.7387, "step": 361600 }, { "epoch": 0.0071936, "grad_norm": 0.7872766256332397, "learning_rate": 2.7174430253148476e-07, "loss": 2.8427, "step": 361610 }, { "epoch": 0.0072192, "grad_norm": 1.0887690782546997, "learning_rate": 2.71557880523321e-07, "loss": 2.854, "step": 361620 }, { "epoch": 0.0072448, "grad_norm": 1.1227147579193115, "learning_rate": 2.713715216015189e-07, "loss": 2.6792, "step": 361630 }, { "epoch": 0.0072704, "grad_norm": 0.9183961153030396, "learning_rate": 2.7118522576728646e-07, "loss": 2.9292, "step": 361640 }, { "epoch": 0.007296, "grad_norm": 0.9021956920623779, "learning_rate": 2.709989930218326e-07, "loss": 2.7286, "step": 361650 }, { "epoch": 0.0073216, "grad_norm": 0.8988887071609497, "learning_rate": 2.708128233663632e-07, "loss": 2.8697, "step": 361660 }, { "epoch": 0.0073472, "grad_norm": 0.7714747190475464, "learning_rate": 2.706267168020882e-07, "loss": 2.8906, "step": 361670 }, { "epoch": 0.0073728, "grad_norm": 0.7901046276092529, "learning_rate": 2.704406733302123e-07, "loss": 2.6478, "step": 361680 }, { "epoch": 0.0073984, "grad_norm": 0.9056600332260132, "learning_rate": 2.702546929519434e-07, "loss": 2.7757, "step": 361690 }, { "epoch": 0.007424, "grad_norm": 0.80672687292099, "learning_rate": 2.7006877566848834e-07, "loss": 2.9033, "step": 361700 }, { "epoch": 0.0074496, "grad_norm": 0.7870028018951416, "learning_rate": 2.698829214810494e-07, "loss": 2.833, "step": 361710 }, { "epoch": 0.0074752, "grad_norm": 0.9823905825614929, "learning_rate": 2.6969713039083466e-07, "loss": 2.8713, "step": 361720 }, { "epoch": 0.0075008, "grad_norm": 0.8597726821899414, "learning_rate": 2.695114023990475e-07, "loss": 2.8698, "step": 361730 }, { "epoch": 0.0075264, "grad_norm": 0.8954282999038696, "learning_rate": 2.693257375068936e-07, "loss": 2.8232, "step": 361740 }, { "epoch": 0.007552, "grad_norm": 0.8287953734397888, "learning_rate": 2.691401357155754e-07, "loss": 2.8857, "step": 361750 }, { "epoch": 0.0075776, "grad_norm": 0.8481448888778687, "learning_rate": 2.6895459702629744e-07, "loss": 3.0298, "step": 361760 }, { "epoch": 0.0076032, "grad_norm": 0.8500460982322693, "learning_rate": 2.6876912144026436e-07, "loss": 2.8136, "step": 361770 }, { "epoch": 0.0076288, "grad_norm": 0.9724722504615784, "learning_rate": 2.6858370895867516e-07, "loss": 2.7843, "step": 361780 }, { "epoch": 0.0076544, "grad_norm": 0.8835186958312988, "learning_rate": 2.683983595827355e-07, "loss": 2.6754, "step": 361790 }, { "epoch": 0.00768, "grad_norm": 0.823977530002594, "learning_rate": 2.682130733136457e-07, "loss": 2.6945, "step": 361800 }, { "epoch": 0.0077056, "grad_norm": 0.8794766664505005, "learning_rate": 2.680278501526068e-07, "loss": 2.9988, "step": 361810 }, { "epoch": 0.0077312, "grad_norm": 0.8639912009239197, "learning_rate": 2.6784269010082134e-07, "loss": 2.6758, "step": 361820 }, { "epoch": 0.0077568, "grad_norm": 0.8591173887252808, "learning_rate": 2.676575931594905e-07, "loss": 2.7748, "step": 361830 }, { "epoch": 0.0077824, "grad_norm": 0.8541174530982971, "learning_rate": 2.674725593298122e-07, "loss": 2.8066, "step": 361840 }, { "epoch": 0.007808, "grad_norm": 0.9669308662414551, "learning_rate": 2.672875886129889e-07, "loss": 2.9591, "step": 361850 }, { "epoch": 0.0078336, "grad_norm": 0.8302295207977295, "learning_rate": 2.6710268101021843e-07, "loss": 2.8265, "step": 361860 }, { "epoch": 0.0078592, "grad_norm": 0.822834849357605, "learning_rate": 2.6691783652269985e-07, "loss": 2.7329, "step": 361870 }, { "epoch": 0.0078848, "grad_norm": 0.8822996616363525, "learning_rate": 2.667330551516334e-07, "loss": 2.7692, "step": 361880 }, { "epoch": 0.0079104, "grad_norm": 0.8429097533226013, "learning_rate": 2.665483368982158e-07, "loss": 2.513, "step": 361890 }, { "epoch": 0.007936, "grad_norm": 0.8662504553794861, "learning_rate": 2.663636817636461e-07, "loss": 2.8244, "step": 361900 }, { "epoch": 0.0079616, "grad_norm": 0.8404454588890076, "learning_rate": 2.6617908974912123e-07, "loss": 2.8272, "step": 361910 }, { "epoch": 0.0079872, "grad_norm": 1.28057861328125, "learning_rate": 2.6599456085583673e-07, "loss": 2.8016, "step": 361920 }, { "epoch": 0.0080128, "grad_norm": 0.8591449856758118, "learning_rate": 2.6581009508499065e-07, "loss": 2.5998, "step": 361930 }, { "epoch": 0.0080384, "grad_norm": 0.8946437835693359, "learning_rate": 2.656256924377798e-07, "loss": 2.6319, "step": 361940 }, { "epoch": 0.008064, "grad_norm": 0.9166824221611023, "learning_rate": 2.6544135291539873e-07, "loss": 2.7007, "step": 361950 }, { "epoch": 0.0080896, "grad_norm": 1.0802665948867798, "learning_rate": 2.652570765190432e-07, "loss": 3.0149, "step": 361960 }, { "epoch": 0.0081152, "grad_norm": 0.8951040506362915, "learning_rate": 2.6507286324991e-07, "loss": 2.97, "step": 361970 }, { "epoch": 0.0081408, "grad_norm": 0.8166884779930115, "learning_rate": 2.648887131091904e-07, "loss": 2.6295, "step": 361980 }, { "epoch": 0.0081664, "grad_norm": 0.839026689529419, "learning_rate": 2.647046260980801e-07, "loss": 2.9127, "step": 361990 }, { "epoch": 0.008192, "grad_norm": 0.8779134154319763, "learning_rate": 2.6452060221777376e-07, "loss": 2.904, "step": 362000 }, { "epoch": 0.0082176, "grad_norm": 0.8063019514083862, "learning_rate": 2.643366414694626e-07, "loss": 2.874, "step": 362010 }, { "epoch": 0.0082432, "grad_norm": 0.7867788076400757, "learning_rate": 2.641527438543412e-07, "loss": 2.9573, "step": 362020 }, { "epoch": 0.0082688, "grad_norm": 0.8338903784751892, "learning_rate": 2.639689093736031e-07, "loss": 2.7776, "step": 362030 }, { "epoch": 0.0082944, "grad_norm": 0.8623434901237488, "learning_rate": 2.637851380284384e-07, "loss": 2.7316, "step": 362040 }, { "epoch": 0.00832, "grad_norm": 0.8339487314224243, "learning_rate": 2.6360142982003954e-07, "loss": 2.7672, "step": 362050 }, { "epoch": 0.0083456, "grad_norm": 1.1330277919769287, "learning_rate": 2.6341778474959776e-07, "loss": 2.8233, "step": 362060 }, { "epoch": 0.0083712, "grad_norm": 0.8945806622505188, "learning_rate": 2.6323420281830435e-07, "loss": 2.624, "step": 362070 }, { "epoch": 0.0083968, "grad_norm": 0.9271154403686523, "learning_rate": 2.630506840273483e-07, "loss": 2.8222, "step": 362080 }, { "epoch": 0.0084224, "grad_norm": 0.8037452697753906, "learning_rate": 2.628672283779221e-07, "loss": 3.0182, "step": 362090 }, { "epoch": 0.008448, "grad_norm": 0.8184521794319153, "learning_rate": 2.626838358712136e-07, "loss": 2.7436, "step": 362100 }, { "epoch": 0.0084736, "grad_norm": 0.7872598171234131, "learning_rate": 2.6250050650841296e-07, "loss": 2.7447, "step": 362110 }, { "epoch": 0.0084992, "grad_norm": 0.8608465194702148, "learning_rate": 2.6231724029070814e-07, "loss": 2.8828, "step": 362120 }, { "epoch": 0.0085248, "grad_norm": 0.9648033380508423, "learning_rate": 2.621340372192882e-07, "loss": 2.8157, "step": 362130 }, { "epoch": 0.0085504, "grad_norm": 0.8320255279541016, "learning_rate": 2.6195089729533997e-07, "loss": 2.8045, "step": 362140 }, { "epoch": 0.008576, "grad_norm": 0.9203642010688782, "learning_rate": 2.617678205200536e-07, "loss": 2.7867, "step": 362150 }, { "epoch": 0.0086016, "grad_norm": 0.8748550415039062, "learning_rate": 2.6158480689461365e-07, "loss": 2.8775, "step": 362160 }, { "epoch": 0.0086272, "grad_norm": 0.8821027874946594, "learning_rate": 2.614018564202103e-07, "loss": 2.8417, "step": 362170 }, { "epoch": 0.0086528, "grad_norm": 0.8099480867385864, "learning_rate": 2.612189690980249e-07, "loss": 2.8345, "step": 362180 }, { "epoch": 0.0086784, "grad_norm": 0.8529967665672302, "learning_rate": 2.6103614492924644e-07, "loss": 2.6742, "step": 362190 }, { "epoch": 0.008704, "grad_norm": 0.9198806881904602, "learning_rate": 2.6085338391506176e-07, "loss": 2.5792, "step": 362200 }, { "epoch": 0.0087296, "grad_norm": 0.9087340235710144, "learning_rate": 2.6067068605665323e-07, "loss": 2.6556, "step": 362210 }, { "epoch": 0.0087552, "grad_norm": 0.8827683329582214, "learning_rate": 2.604880513552077e-07, "loss": 2.8468, "step": 362220 }, { "epoch": 0.0087808, "grad_norm": 0.8746462464332581, "learning_rate": 2.6030547981190976e-07, "loss": 2.7451, "step": 362230 }, { "epoch": 0.0088064, "grad_norm": 1.0121861696243286, "learning_rate": 2.6012297142794075e-07, "loss": 2.7836, "step": 362240 }, { "epoch": 0.008832, "grad_norm": 0.869986355304718, "learning_rate": 2.599405262044863e-07, "loss": 2.7577, "step": 362250 }, { "epoch": 0.0088576, "grad_norm": 0.9473204016685486, "learning_rate": 2.597581441427288e-07, "loss": 2.6497, "step": 362260 }, { "epoch": 0.0088832, "grad_norm": 0.8920488953590393, "learning_rate": 2.595758252438507e-07, "loss": 2.7908, "step": 362270 }, { "epoch": 0.0089088, "grad_norm": 0.7946478724479675, "learning_rate": 2.593935695090355e-07, "loss": 2.6695, "step": 362280 }, { "epoch": 0.0089344, "grad_norm": 0.8685744404792786, "learning_rate": 2.592113769394633e-07, "loss": 2.7406, "step": 362290 }, { "epoch": 0.00896, "grad_norm": 0.8751125931739807, "learning_rate": 2.5902924753631763e-07, "loss": 2.7733, "step": 362300 }, { "epoch": 0.0089856, "grad_norm": 3.425705909729004, "learning_rate": 2.5884718130077866e-07, "loss": 2.5415, "step": 362310 }, { "epoch": 0.0090112, "grad_norm": 1.0880930423736572, "learning_rate": 2.5866517823402547e-07, "loss": 2.809, "step": 362320 }, { "epoch": 0.0090368, "grad_norm": 0.8920176029205322, "learning_rate": 2.584832383372404e-07, "loss": 2.9098, "step": 362330 }, { "epoch": 0.0090624, "grad_norm": 0.8380365967750549, "learning_rate": 2.5830136161160257e-07, "loss": 2.7867, "step": 362340 }, { "epoch": 0.009088, "grad_norm": 0.8030878901481628, "learning_rate": 2.58119548058291e-07, "loss": 2.9416, "step": 362350 }, { "epoch": 0.0091136, "grad_norm": 0.8466911315917969, "learning_rate": 2.579377976784847e-07, "loss": 2.8225, "step": 362360 }, { "epoch": 0.0091392, "grad_norm": 0.8890205025672913, "learning_rate": 2.577561104733628e-07, "loss": 2.7271, "step": 362370 }, { "epoch": 0.0091648, "grad_norm": 0.872208833694458, "learning_rate": 2.575744864441043e-07, "loss": 2.7331, "step": 362380 }, { "epoch": 0.0091904, "grad_norm": 0.8944977521896362, "learning_rate": 2.57392925591885e-07, "loss": 2.9862, "step": 362390 }, { "epoch": 0.009216, "grad_norm": 0.8519682884216309, "learning_rate": 2.572114279178839e-07, "loss": 2.7576, "step": 362400 }, { "epoch": 0.0092416, "grad_norm": 0.9994596838951111, "learning_rate": 2.570299934232767e-07, "loss": 2.7986, "step": 362410 }, { "epoch": 0.0092672, "grad_norm": 0.9582151770591736, "learning_rate": 2.568486221092414e-07, "loss": 2.8173, "step": 362420 }, { "epoch": 0.0092928, "grad_norm": 0.8965982794761658, "learning_rate": 2.5666731397695375e-07, "loss": 2.7602, "step": 362430 }, { "epoch": 0.0093184, "grad_norm": 1.7618452310562134, "learning_rate": 2.564860690275883e-07, "loss": 3.0511, "step": 362440 }, { "epoch": 0.009344, "grad_norm": 0.8068059682846069, "learning_rate": 2.5630488726232195e-07, "loss": 2.8733, "step": 362450 }, { "epoch": 0.0093696, "grad_norm": 0.8235632181167603, "learning_rate": 2.5612376868232814e-07, "loss": 2.8047, "step": 362460 }, { "epoch": 0.0093952, "grad_norm": 0.8991013169288635, "learning_rate": 2.5594271328878153e-07, "loss": 2.9075, "step": 362470 }, { "epoch": 0.0094208, "grad_norm": 0.9357622861862183, "learning_rate": 2.5576172108285777e-07, "loss": 2.7855, "step": 362480 }, { "epoch": 0.0094464, "grad_norm": 0.8528813719749451, "learning_rate": 2.5558079206572936e-07, "loss": 2.8459, "step": 362490 }, { "epoch": 0.009472, "grad_norm": 0.8295310139656067, "learning_rate": 2.553999262385698e-07, "loss": 2.5918, "step": 362500 }, { "epoch": 0.0094976, "grad_norm": 0.8585569858551025, "learning_rate": 2.552191236025514e-07, "loss": 2.7762, "step": 362510 }, { "epoch": 0.0095232, "grad_norm": 0.8571256995201111, "learning_rate": 2.5503838415884774e-07, "loss": 2.7259, "step": 362520 }, { "epoch": 0.0095488, "grad_norm": 0.8075106143951416, "learning_rate": 2.5485770790862896e-07, "loss": 2.906, "step": 362530 }, { "epoch": 0.0095744, "grad_norm": 0.8971423506736755, "learning_rate": 2.546770948530686e-07, "loss": 2.9322, "step": 362540 }, { "epoch": 0.0096, "grad_norm": 0.8908414244651794, "learning_rate": 2.5449654499333896e-07, "loss": 2.7885, "step": 362550 }, { "epoch": 0.0096256, "grad_norm": 0.823203980922699, "learning_rate": 2.5431605833060814e-07, "loss": 2.8742, "step": 362560 }, { "epoch": 0.0096512, "grad_norm": 0.8183453679084778, "learning_rate": 2.541356348660473e-07, "loss": 2.7926, "step": 362570 }, { "epoch": 0.0096768, "grad_norm": 0.8354263305664062, "learning_rate": 2.539552746008267e-07, "loss": 2.8532, "step": 362580 }, { "epoch": 0.0097024, "grad_norm": 0.9660682678222656, "learning_rate": 2.5377497753611535e-07, "loss": 2.8922, "step": 362590 }, { "epoch": 0.009728, "grad_norm": 0.8678390979766846, "learning_rate": 2.5359474367308344e-07, "loss": 2.7477, "step": 362600 }, { "epoch": 0.0097536, "grad_norm": 0.83238285779953, "learning_rate": 2.5341457301290006e-07, "loss": 2.6514, "step": 362610 }, { "epoch": 0.0097792, "grad_norm": 0.8711931705474854, "learning_rate": 2.5323446555673315e-07, "loss": 3.0593, "step": 362620 }, { "epoch": 0.0098048, "grad_norm": 0.8202279806137085, "learning_rate": 2.530544213057495e-07, "loss": 2.8269, "step": 362630 }, { "epoch": 0.0098304, "grad_norm": 0.9656457304954529, "learning_rate": 2.528744402611172e-07, "loss": 2.936, "step": 362640 }, { "epoch": 0.009856, "grad_norm": 0.7543172836303711, "learning_rate": 2.5269452242400406e-07, "loss": 2.7522, "step": 362650 }, { "epoch": 0.0098816, "grad_norm": 0.8184288740158081, "learning_rate": 2.5251466779557587e-07, "loss": 2.851, "step": 362660 }, { "epoch": 0.0099072, "grad_norm": 0.7944364547729492, "learning_rate": 2.523348763770006e-07, "loss": 2.9195, "step": 362670 }, { "epoch": 0.0099328, "grad_norm": 0.8719912767410278, "learning_rate": 2.5215514816944177e-07, "loss": 3.155, "step": 362680 }, { "epoch": 0.0099584, "grad_norm": 1.1755194664001465, "learning_rate": 2.5197548317406727e-07, "loss": 2.8008, "step": 362690 }, { "epoch": 0.009984, "grad_norm": 0.9938275218009949, "learning_rate": 2.517958813920407e-07, "loss": 2.5738, "step": 362700 }, { "epoch": 0.0100096, "grad_norm": 0.8894641995429993, "learning_rate": 2.5161634282452664e-07, "loss": 2.6674, "step": 362710 }, { "epoch": 0.0100352, "grad_norm": 0.8793523907661438, "learning_rate": 2.514368674726897e-07, "loss": 3.1309, "step": 362720 }, { "epoch": 0.0100608, "grad_norm": 0.8348132371902466, "learning_rate": 2.5125745533769453e-07, "loss": 2.8918, "step": 362730 }, { "epoch": 0.0100864, "grad_norm": 0.7983936071395874, "learning_rate": 2.5107810642070354e-07, "loss": 2.6819, "step": 362740 }, { "epoch": 0.010112, "grad_norm": 0.8787005543708801, "learning_rate": 2.508988207228813e-07, "loss": 2.9611, "step": 362750 }, { "epoch": 0.0101376, "grad_norm": 0.8992324471473694, "learning_rate": 2.507195982453881e-07, "loss": 2.9574, "step": 362760 }, { "epoch": 0.0101632, "grad_norm": 0.8584340810775757, "learning_rate": 2.505404389893873e-07, "loss": 2.7401, "step": 362770 }, { "epoch": 0.0101888, "grad_norm": 1.7014415264129639, "learning_rate": 2.503613429560403e-07, "loss": 2.9278, "step": 362780 }, { "epoch": 0.0102144, "grad_norm": 0.8016423583030701, "learning_rate": 2.501823101465084e-07, "loss": 2.7076, "step": 362790 }, { "epoch": 0.01024, "grad_norm": 0.8389585614204407, "learning_rate": 2.50003340561954e-07, "loss": 2.6528, "step": 362800 }, { "epoch": 0.0102656, "grad_norm": 1.0307292938232422, "learning_rate": 2.498244342035361e-07, "loss": 2.8837, "step": 362810 }, { "epoch": 0.0102912, "grad_norm": 0.9813902378082275, "learning_rate": 2.4964559107241714e-07, "loss": 3.1694, "step": 362820 }, { "epoch": 0.0103168, "grad_norm": 0.9034523367881775, "learning_rate": 2.494668111697529e-07, "loss": 3.1232, "step": 362830 }, { "epoch": 0.0103424, "grad_norm": 0.8158898949623108, "learning_rate": 2.492880944967069e-07, "loss": 2.8641, "step": 362840 }, { "epoch": 0.010368, "grad_norm": 0.8974273204803467, "learning_rate": 2.491094410544348e-07, "loss": 3.4606, "step": 362850 }, { "epoch": 0.0103936, "grad_norm": 0.8350685834884644, "learning_rate": 2.4893085084409687e-07, "loss": 2.8579, "step": 362860 }, { "epoch": 0.0104192, "grad_norm": 0.954091489315033, "learning_rate": 2.48752323866851e-07, "loss": 2.7016, "step": 362870 }, { "epoch": 0.0104448, "grad_norm": 0.7501057982444763, "learning_rate": 2.485738601238541e-07, "loss": 2.8359, "step": 362880 }, { "epoch": 0.0104704, "grad_norm": 0.7891282439231873, "learning_rate": 2.483954596162641e-07, "loss": 2.8241, "step": 362890 }, { "epoch": 0.010496, "grad_norm": 0.8084729909896851, "learning_rate": 2.48217122345239e-07, "loss": 2.7501, "step": 362900 }, { "epoch": 0.0105216, "grad_norm": 0.9872139692306519, "learning_rate": 2.480388483119334e-07, "loss": 2.669, "step": 362910 }, { "epoch": 0.0105472, "grad_norm": 1.1778160333633423, "learning_rate": 2.4786063751750413e-07, "loss": 2.8864, "step": 362920 }, { "epoch": 0.0105728, "grad_norm": 0.8755242824554443, "learning_rate": 2.4768248996310694e-07, "loss": 2.8545, "step": 362930 }, { "epoch": 0.0105984, "grad_norm": 0.7596769332885742, "learning_rate": 2.475044056498965e-07, "loss": 2.804, "step": 362940 }, { "epoch": 0.010624, "grad_norm": 0.8537918925285339, "learning_rate": 2.473263845790297e-07, "loss": 2.8526, "step": 362950 }, { "epoch": 0.0106496, "grad_norm": 0.861839771270752, "learning_rate": 2.471484267516577e-07, "loss": 2.9281, "step": 362960 }, { "epoch": 0.0106752, "grad_norm": 0.8257150650024414, "learning_rate": 2.469705321689364e-07, "loss": 2.8331, "step": 362970 }, { "epoch": 0.0107008, "grad_norm": 1.0815080404281616, "learning_rate": 2.4679270083202035e-07, "loss": 2.7306, "step": 362980 }, { "epoch": 0.0107264, "grad_norm": 0.8411905169487, "learning_rate": 2.4661493274205974e-07, "loss": 2.8681, "step": 362990 }, { "epoch": 0.010752, "grad_norm": 0.8647293448448181, "learning_rate": 2.464372279002103e-07, "loss": 3.0333, "step": 363000 }, { "epoch": 0.0107776, "grad_norm": 0.8672186136245728, "learning_rate": 2.4625958630762227e-07, "loss": 2.9466, "step": 363010 }, { "epoch": 0.0108032, "grad_norm": 0.9106113910675049, "learning_rate": 2.4608200796545024e-07, "loss": 2.8593, "step": 363020 }, { "epoch": 0.0108288, "grad_norm": 0.9212074875831604, "learning_rate": 2.459044928748433e-07, "loss": 2.8253, "step": 363030 }, { "epoch": 0.0108544, "grad_norm": 0.7899981737136841, "learning_rate": 2.457270410369528e-07, "loss": 2.7943, "step": 363040 }, { "epoch": 0.01088, "grad_norm": 0.7640841603279114, "learning_rate": 2.4554965245292994e-07, "loss": 2.7615, "step": 363050 }, { "epoch": 0.0109056, "grad_norm": 0.7848767638206482, "learning_rate": 2.453723271239261e-07, "loss": 2.7371, "step": 363060 }, { "epoch": 0.0109312, "grad_norm": 0.7973921895027161, "learning_rate": 2.4519506505108926e-07, "loss": 2.7083, "step": 363070 }, { "epoch": 0.0109568, "grad_norm": 0.8729894161224365, "learning_rate": 2.4501786623557177e-07, "loss": 2.7632, "step": 363080 }, { "epoch": 0.0109824, "grad_norm": 0.8293752074241638, "learning_rate": 2.448407306785194e-07, "loss": 2.829, "step": 363090 }, { "epoch": 0.011008, "grad_norm": 0.8749448657035828, "learning_rate": 2.446636583810824e-07, "loss": 2.6259, "step": 363100 }, { "epoch": 0.0110336, "grad_norm": 0.9585132598876953, "learning_rate": 2.444866493444087e-07, "loss": 2.6114, "step": 363110 }, { "epoch": 0.0110592, "grad_norm": 0.8459774851799011, "learning_rate": 2.443097035696462e-07, "loss": 2.7605, "step": 363120 }, { "epoch": 0.0110848, "grad_norm": 1.0576075315475464, "learning_rate": 2.44132821057943e-07, "loss": 2.9789, "step": 363130 }, { "epoch": 0.0111104, "grad_norm": 0.9172202944755554, "learning_rate": 2.4395600181044586e-07, "loss": 2.7305, "step": 363140 }, { "epoch": 0.011136, "grad_norm": 0.7738659381866455, "learning_rate": 2.437792458283017e-07, "loss": 2.7286, "step": 363150 }, { "epoch": 0.0111616, "grad_norm": 0.9562162756919861, "learning_rate": 2.4360255311265626e-07, "loss": 2.8799, "step": 363160 }, { "epoch": 0.0111872, "grad_norm": 0.9820334911346436, "learning_rate": 2.4342592366465414e-07, "loss": 2.7098, "step": 363170 }, { "epoch": 0.0112128, "grad_norm": 0.8567832112312317, "learning_rate": 2.432493574854422e-07, "loss": 2.7481, "step": 363180 }, { "epoch": 0.0112384, "grad_norm": 1.6968753337860107, "learning_rate": 2.4307285457616515e-07, "loss": 2.7811, "step": 363190 }, { "epoch": 0.011264, "grad_norm": 0.8447450995445251, "learning_rate": 2.4289641493796755e-07, "loss": 2.9335, "step": 363200 }, { "epoch": 0.0112896, "grad_norm": 0.8235515356063843, "learning_rate": 2.427200385719941e-07, "loss": 2.8489, "step": 363210 }, { "epoch": 0.0113152, "grad_norm": 0.8232543468475342, "learning_rate": 2.425437254793872e-07, "loss": 3.1263, "step": 363220 }, { "epoch": 0.0113408, "grad_norm": 0.8176999092102051, "learning_rate": 2.423674756612915e-07, "loss": 2.8854, "step": 363230 }, { "epoch": 0.0113664, "grad_norm": 0.8835060000419617, "learning_rate": 2.4219128911884935e-07, "loss": 2.7488, "step": 363240 }, { "epoch": 0.011392, "grad_norm": 0.7995017170906067, "learning_rate": 2.420151658532033e-07, "loss": 2.8384, "step": 363250 }, { "epoch": 0.0114176, "grad_norm": 0.9403156042098999, "learning_rate": 2.4183910586549565e-07, "loss": 2.699, "step": 363260 }, { "epoch": 0.0114432, "grad_norm": 0.9306178092956543, "learning_rate": 2.4166310915686776e-07, "loss": 2.8091, "step": 363270 }, { "epoch": 0.0114688, "grad_norm": 0.9882218837738037, "learning_rate": 2.4148717572846204e-07, "loss": 2.7691, "step": 363280 }, { "epoch": 0.0114944, "grad_norm": 0.7943774461746216, "learning_rate": 2.4131130558141867e-07, "loss": 2.95, "step": 363290 }, { "epoch": 0.01152, "grad_norm": 0.8513442277908325, "learning_rate": 2.411354987168768e-07, "loss": 2.8005, "step": 363300 }, { "epoch": 0.0115456, "grad_norm": 0.783576488494873, "learning_rate": 2.409597551359777e-07, "loss": 2.6856, "step": 363310 }, { "epoch": 0.0115712, "grad_norm": 0.7931017279624939, "learning_rate": 2.4078407483986157e-07, "loss": 2.7176, "step": 363320 }, { "epoch": 0.0115968, "grad_norm": 0.9277819991111755, "learning_rate": 2.4060845782966633e-07, "loss": 2.5124, "step": 363330 }, { "epoch": 0.0116224, "grad_norm": 0.9595214128494263, "learning_rate": 2.404329041065312e-07, "loss": 2.7556, "step": 363340 }, { "epoch": 0.011648, "grad_norm": 0.8142890334129333, "learning_rate": 2.402574136715974e-07, "loss": 2.7326, "step": 363350 }, { "epoch": 0.0116736, "grad_norm": 0.8060120940208435, "learning_rate": 2.400819865259985e-07, "loss": 2.7815, "step": 363360 }, { "epoch": 0.0116992, "grad_norm": 0.761506199836731, "learning_rate": 2.399066226708735e-07, "loss": 2.7791, "step": 363370 }, { "epoch": 0.0117248, "grad_norm": 0.8914819955825806, "learning_rate": 2.397313221073605e-07, "loss": 2.7918, "step": 363380 }, { "epoch": 0.0117504, "grad_norm": 0.9282498359680176, "learning_rate": 2.3955608483659633e-07, "loss": 2.7589, "step": 363390 }, { "epoch": 0.011776, "grad_norm": 0.8451080918312073, "learning_rate": 2.393809108597156e-07, "loss": 2.7889, "step": 363400 }, { "epoch": 0.0118016, "grad_norm": 0.8194382190704346, "learning_rate": 2.3920580017785743e-07, "loss": 2.8554, "step": 363410 }, { "epoch": 0.0118272, "grad_norm": 0.9387438893318176, "learning_rate": 2.390307527921554e-07, "loss": 2.9296, "step": 363420 }, { "epoch": 0.0118528, "grad_norm": 0.8132844567298889, "learning_rate": 2.3885576870374405e-07, "loss": 2.8585, "step": 363430 }, { "epoch": 0.0118784, "grad_norm": 0.9407564401626587, "learning_rate": 2.386808479137592e-07, "loss": 2.7524, "step": 363440 }, { "epoch": 0.011904, "grad_norm": 0.8412395715713501, "learning_rate": 2.3850599042333446e-07, "loss": 2.8395, "step": 363450 }, { "epoch": 0.0119296, "grad_norm": 0.848646879196167, "learning_rate": 2.3833119623360434e-07, "loss": 2.6206, "step": 363460 }, { "epoch": 0.0119552, "grad_norm": 1.0475550889968872, "learning_rate": 2.3815646534570135e-07, "loss": 2.6535, "step": 363470 }, { "epoch": 0.0119808, "grad_norm": 0.7202942371368408, "learning_rate": 2.3798179776076124e-07, "loss": 2.6692, "step": 363480 }, { "epoch": 0.0120064, "grad_norm": 0.8786847591400146, "learning_rate": 2.378071934799131e-07, "loss": 2.9718, "step": 363490 }, { "epoch": 0.012032, "grad_norm": 0.7819257378578186, "learning_rate": 2.3763265250429158e-07, "loss": 2.8392, "step": 363500 }, { "epoch": 0.0120576, "grad_norm": 0.8693711161613464, "learning_rate": 2.3745817483502687e-07, "loss": 2.8618, "step": 363510 }, { "epoch": 0.0120832, "grad_norm": 2.256286382675171, "learning_rate": 2.3728376047325141e-07, "loss": 2.9161, "step": 363520 }, { "epoch": 0.0121088, "grad_norm": 1.017187476158142, "learning_rate": 2.3710940942009654e-07, "loss": 3.0269, "step": 363530 }, { "epoch": 0.0121344, "grad_norm": 0.7654657363891602, "learning_rate": 2.369351216766924e-07, "loss": 2.8596, "step": 363540 }, { "epoch": 0.01216, "grad_norm": 0.8776144981384277, "learning_rate": 2.3676089724417038e-07, "loss": 2.8307, "step": 363550 }, { "epoch": 0.0121856, "grad_norm": 0.8248263597488403, "learning_rate": 2.3658673612365847e-07, "loss": 2.9251, "step": 363560 }, { "epoch": 0.0122112, "grad_norm": 0.8562418818473816, "learning_rate": 2.364126383162857e-07, "loss": 2.8037, "step": 363570 }, { "epoch": 0.0122368, "grad_norm": 0.8585139513015747, "learning_rate": 2.3623860382318232e-07, "loss": 2.9941, "step": 363580 }, { "epoch": 0.0122624, "grad_norm": 0.8507795929908752, "learning_rate": 2.360646326454774e-07, "loss": 2.7385, "step": 363590 }, { "epoch": 0.012288, "grad_norm": 0.827155590057373, "learning_rate": 2.3589072478429897e-07, "loss": 2.9055, "step": 363600 }, { "epoch": 0.0123136, "grad_norm": 1.0178537368774414, "learning_rate": 2.357168802407739e-07, "loss": 2.7383, "step": 363610 }, { "epoch": 0.0123392, "grad_norm": 0.8059407472610474, "learning_rate": 2.35543099016029e-07, "loss": 2.8625, "step": 363620 }, { "epoch": 0.0123648, "grad_norm": 0.8358075618743896, "learning_rate": 2.3536938111119234e-07, "loss": 2.9279, "step": 363630 }, { "epoch": 0.0123904, "grad_norm": 0.8899069428443909, "learning_rate": 2.3519572652739077e-07, "loss": 3.108, "step": 363640 }, { "epoch": 0.012416, "grad_norm": 1.051124095916748, "learning_rate": 2.350221352657489e-07, "loss": 3.188, "step": 363650 }, { "epoch": 0.0124416, "grad_norm": 0.8468402624130249, "learning_rate": 2.3484860732739368e-07, "loss": 2.7784, "step": 363660 }, { "epoch": 0.0124672, "grad_norm": 0.8106138706207275, "learning_rate": 2.346751427134497e-07, "loss": 2.9609, "step": 363670 }, { "epoch": 0.0124928, "grad_norm": 0.7750692963600159, "learning_rate": 2.3450174142504278e-07, "loss": 2.743, "step": 363680 }, { "epoch": 0.0125184, "grad_norm": 0.9705590009689331, "learning_rate": 2.3432840346329643e-07, "loss": 2.879, "step": 363690 }, { "epoch": 0.012544, "grad_norm": 0.7717176675796509, "learning_rate": 2.3415512882933534e-07, "loss": 2.7608, "step": 363700 }, { "epoch": 0.0125696, "grad_norm": 0.9338473677635193, "learning_rate": 2.339819175242819e-07, "loss": 2.9626, "step": 363710 }, { "epoch": 0.0125952, "grad_norm": 0.8374669551849365, "learning_rate": 2.338087695492597e-07, "loss": 2.9248, "step": 363720 }, { "epoch": 0.0126208, "grad_norm": 0.9143600463867188, "learning_rate": 2.3363568490539334e-07, "loss": 2.7919, "step": 363730 }, { "epoch": 0.0126464, "grad_norm": 1.1249091625213623, "learning_rate": 2.334626635938031e-07, "loss": 2.7641, "step": 363740 }, { "epoch": 0.012672, "grad_norm": 0.8228736519813538, "learning_rate": 2.332897056156125e-07, "loss": 2.864, "step": 363750 }, { "epoch": 0.0126976, "grad_norm": 0.9433528780937195, "learning_rate": 2.3311681097194284e-07, "loss": 2.8957, "step": 363760 }, { "epoch": 0.0127232, "grad_norm": 0.9455218315124512, "learning_rate": 2.3294397966391435e-07, "loss": 2.8643, "step": 363770 }, { "epoch": 0.0127488, "grad_norm": 1.1546010971069336, "learning_rate": 2.327712116926484e-07, "loss": 3.0313, "step": 363780 }, { "epoch": 0.0127744, "grad_norm": 1.0144051313400269, "learning_rate": 2.325985070592651e-07, "loss": 2.8331, "step": 363790 }, { "epoch": 0.0128, "grad_norm": 0.9955174326896667, "learning_rate": 2.3242586576488479e-07, "loss": 2.786, "step": 363800 }, { "epoch": 0.0128256, "grad_norm": 1.0775038003921509, "learning_rate": 2.3225328781062762e-07, "loss": 2.8855, "step": 363810 }, { "epoch": 0.0128512, "grad_norm": 0.9588645696640015, "learning_rate": 2.3208077319761045e-07, "loss": 2.8155, "step": 363820 }, { "epoch": 0.0128768, "grad_norm": 1.0632541179656982, "learning_rate": 2.3190832192695468e-07, "loss": 2.7135, "step": 363830 }, { "epoch": 0.0129024, "grad_norm": 0.9899585247039795, "learning_rate": 2.3173593399977713e-07, "loss": 3.0717, "step": 363840 }, { "epoch": 0.012928, "grad_norm": 0.9041382670402527, "learning_rate": 2.315636094171958e-07, "loss": 2.8177, "step": 363850 }, { "epoch": 0.0129536, "grad_norm": 0.8971526622772217, "learning_rate": 2.313913481803276e-07, "loss": 2.7895, "step": 363860 }, { "epoch": 0.0129792, "grad_norm": 0.823885977268219, "learning_rate": 2.3121915029029051e-07, "loss": 2.765, "step": 363870 }, { "epoch": 0.0130048, "grad_norm": 1.1384611129760742, "learning_rate": 2.3104701574820254e-07, "loss": 3.0743, "step": 363880 }, { "epoch": 0.0130304, "grad_norm": 1.7950763702392578, "learning_rate": 2.3087494455517723e-07, "loss": 3.1361, "step": 363890 }, { "epoch": 0.013056, "grad_norm": 0.8469346165657043, "learning_rate": 2.3070293671233256e-07, "loss": 2.7881, "step": 363900 }, { "epoch": 0.0130816, "grad_norm": 0.871024489402771, "learning_rate": 2.30530992220781e-07, "loss": 2.8194, "step": 363910 }, { "epoch": 0.0131072, "grad_norm": 0.8337764739990234, "learning_rate": 2.3035911108164054e-07, "loss": 2.8593, "step": 363920 }, { "epoch": 0.0131328, "grad_norm": 0.8101897835731506, "learning_rate": 2.3018729329602584e-07, "loss": 2.7017, "step": 363930 }, { "epoch": 0.0131584, "grad_norm": 0.8690321445465088, "learning_rate": 2.3001553886504823e-07, "loss": 2.7705, "step": 363940 }, { "epoch": 0.013184, "grad_norm": 1.0941015481948853, "learning_rate": 2.298438477898246e-07, "loss": 3.0587, "step": 363950 }, { "epoch": 0.0132096, "grad_norm": 1.1330844163894653, "learning_rate": 2.2967222007146628e-07, "loss": 2.8169, "step": 363960 }, { "epoch": 0.0132352, "grad_norm": 0.930022656917572, "learning_rate": 2.295006557110868e-07, "loss": 2.7055, "step": 363970 }, { "epoch": 0.0132608, "grad_norm": 0.8890236020088196, "learning_rate": 2.2932915470979977e-07, "loss": 2.6833, "step": 363980 }, { "epoch": 0.0132864, "grad_norm": 0.9133943915367126, "learning_rate": 2.2915771706871647e-07, "loss": 2.8657, "step": 363990 }, { "epoch": 0.013312, "grad_norm": 0.8499343991279602, "learning_rate": 2.2898634278894717e-07, "loss": 2.7064, "step": 364000 }, { "epoch": 0.0133376, "grad_norm": 0.8660539388656616, "learning_rate": 2.2881503187160647e-07, "loss": 2.9978, "step": 364010 }, { "epoch": 0.0133632, "grad_norm": 0.8323609232902527, "learning_rate": 2.2864378431780242e-07, "loss": 2.703, "step": 364020 }, { "epoch": 0.0133888, "grad_norm": 0.8309236168861389, "learning_rate": 2.2847260012864746e-07, "loss": 3.104, "step": 364030 }, { "epoch": 0.0134144, "grad_norm": 0.7797548770904541, "learning_rate": 2.283014793052496e-07, "loss": 2.752, "step": 364040 }, { "epoch": 0.01344, "grad_norm": 1.2114039659500122, "learning_rate": 2.2813042184872014e-07, "loss": 2.8438, "step": 364050 }, { "epoch": 0.0134656, "grad_norm": 0.8172939419746399, "learning_rate": 2.279594277601671e-07, "loss": 2.7462, "step": 364060 }, { "epoch": 0.0134912, "grad_norm": 1.0541375875473022, "learning_rate": 2.277884970407007e-07, "loss": 2.8684, "step": 364070 }, { "epoch": 0.0135168, "grad_norm": 0.7940062880516052, "learning_rate": 2.2761762969142897e-07, "loss": 2.855, "step": 364080 }, { "epoch": 0.0135424, "grad_norm": 0.8678818941116333, "learning_rate": 2.2744682571345878e-07, "loss": 2.8598, "step": 364090 }, { "epoch": 0.013568, "grad_norm": 0.8491756319999695, "learning_rate": 2.2727608510789923e-07, "loss": 2.8065, "step": 364100 }, { "epoch": 0.0135936, "grad_norm": 0.8726961016654968, "learning_rate": 2.2710540787585723e-07, "loss": 2.9695, "step": 364110 }, { "epoch": 0.0136192, "grad_norm": 0.8650698661804199, "learning_rate": 2.2693479401843965e-07, "loss": 2.8625, "step": 364120 }, { "epoch": 0.0136448, "grad_norm": 0.9103763699531555, "learning_rate": 2.267642435367523e-07, "loss": 2.8149, "step": 364130 }, { "epoch": 0.0136704, "grad_norm": 0.8009986281394958, "learning_rate": 2.2659375643190095e-07, "loss": 2.6864, "step": 364140 }, { "epoch": 0.013696, "grad_norm": 0.7940796613693237, "learning_rate": 2.264233327049925e-07, "loss": 2.7796, "step": 364150 }, { "epoch": 0.0137216, "grad_norm": 0.8525314331054688, "learning_rate": 2.2625297235713052e-07, "loss": 2.9819, "step": 364160 }, { "epoch": 0.0137472, "grad_norm": 0.8143289685249329, "learning_rate": 2.2608267538942074e-07, "loss": 2.9036, "step": 364170 }, { "epoch": 0.0137728, "grad_norm": 0.8274737000465393, "learning_rate": 2.2591244180296678e-07, "loss": 2.6329, "step": 364180 }, { "epoch": 0.0137984, "grad_norm": 0.8797023892402649, "learning_rate": 2.257422715988733e-07, "loss": 2.9098, "step": 364190 }, { "epoch": 0.013824, "grad_norm": 0.8655573725700378, "learning_rate": 2.2557216477824272e-07, "loss": 2.7708, "step": 364200 }, { "epoch": 0.0138496, "grad_norm": 0.8374969959259033, "learning_rate": 2.2540212134218086e-07, "loss": 2.4468, "step": 364210 }, { "epoch": 0.0138752, "grad_norm": 0.7346143126487732, "learning_rate": 2.2523214129178683e-07, "loss": 2.7135, "step": 364220 }, { "epoch": 0.0139008, "grad_norm": 0.9693794250488281, "learning_rate": 2.250622246281653e-07, "loss": 2.9814, "step": 364230 }, { "epoch": 0.0139264, "grad_norm": 0.8929356336593628, "learning_rate": 2.2489237135241758e-07, "loss": 2.966, "step": 364240 }, { "epoch": 0.013952, "grad_norm": 1.066356897354126, "learning_rate": 2.2472258146564396e-07, "loss": 2.8291, "step": 364250 }, { "epoch": 0.0139776, "grad_norm": 0.8036434650421143, "learning_rate": 2.2455285496894686e-07, "loss": 2.84, "step": 364260 }, { "epoch": 0.0140032, "grad_norm": 0.8461859226226807, "learning_rate": 2.2438319186342649e-07, "loss": 2.7344, "step": 364270 }, { "epoch": 0.0140288, "grad_norm": 0.7894585728645325, "learning_rate": 2.2421359215018312e-07, "loss": 2.9029, "step": 364280 }, { "epoch": 0.0140544, "grad_norm": 0.8190479278564453, "learning_rate": 2.2404405583031586e-07, "loss": 2.7001, "step": 364290 }, { "epoch": 0.01408, "grad_norm": 0.8494136929512024, "learning_rate": 2.2387458290492604e-07, "loss": 2.7374, "step": 364300 }, { "epoch": 0.0141056, "grad_norm": 0.8458041548728943, "learning_rate": 2.2370517337511055e-07, "loss": 2.8237, "step": 364310 }, { "epoch": 0.0141312, "grad_norm": 0.9095882177352905, "learning_rate": 2.2353582724196855e-07, "loss": 2.7266, "step": 364320 }, { "epoch": 0.0141568, "grad_norm": 0.9133421778678894, "learning_rate": 2.2336654450660022e-07, "loss": 2.9825, "step": 364330 }, { "epoch": 0.0141824, "grad_norm": 0.8765522837638855, "learning_rate": 2.2319732517009917e-07, "loss": 2.8975, "step": 364340 }, { "epoch": 0.014208, "grad_norm": 0.7983289957046509, "learning_rate": 2.2302816923356675e-07, "loss": 2.9107, "step": 364350 }, { "epoch": 0.0142336, "grad_norm": 1.5929791927337646, "learning_rate": 2.228590766980976e-07, "loss": 3.2351, "step": 364360 }, { "epoch": 0.0142592, "grad_norm": 0.8599867224693298, "learning_rate": 2.2269004756478863e-07, "loss": 2.784, "step": 364370 }, { "epoch": 0.0142848, "grad_norm": 0.821770429611206, "learning_rate": 2.2252108183473674e-07, "loss": 2.8355, "step": 364380 }, { "epoch": 0.0143104, "grad_norm": 0.82927006483078, "learning_rate": 2.2235217950903664e-07, "loss": 2.7288, "step": 364390 }, { "epoch": 0.014336, "grad_norm": 0.9053561091423035, "learning_rate": 2.221833405887841e-07, "loss": 2.9295, "step": 364400 }, { "epoch": 0.0143616, "grad_norm": 0.8528144955635071, "learning_rate": 2.2201456507507492e-07, "loss": 2.8264, "step": 364410 }, { "epoch": 0.0143872, "grad_norm": 0.7537058591842651, "learning_rate": 2.2184585296900152e-07, "loss": 2.7776, "step": 364420 }, { "epoch": 0.0144128, "grad_norm": 0.8778796792030334, "learning_rate": 2.216772042716586e-07, "loss": 2.7643, "step": 364430 }, { "epoch": 0.0144384, "grad_norm": 0.9209412336349487, "learning_rate": 2.215086189841409e-07, "loss": 2.6441, "step": 364440 }, { "epoch": 0.014464, "grad_norm": 0.7917913198471069, "learning_rate": 2.213400971075408e-07, "loss": 2.8922, "step": 364450 }, { "epoch": 0.0144896, "grad_norm": 1.0134024620056152, "learning_rate": 2.2117163864295187e-07, "loss": 2.8198, "step": 364460 }, { "epoch": 0.0145152, "grad_norm": 0.743910551071167, "learning_rate": 2.2100324359146552e-07, "loss": 2.6959, "step": 364470 }, { "epoch": 0.0145408, "grad_norm": 1.3877052068710327, "learning_rate": 2.2083491195417417e-07, "loss": 2.8, "step": 364480 }, { "epoch": 0.0145664, "grad_norm": 0.8628432750701904, "learning_rate": 2.2066664373217028e-07, "loss": 3.0622, "step": 364490 }, { "epoch": 0.014592, "grad_norm": 0.8131266832351685, "learning_rate": 2.20498438926543e-07, "loss": 2.87, "step": 364500 }, { "epoch": 0.0146176, "grad_norm": 1.2102738618850708, "learning_rate": 2.2033029753838476e-07, "loss": 2.8927, "step": 364510 }, { "epoch": 0.0146432, "grad_norm": 0.7797832489013672, "learning_rate": 2.201622195687847e-07, "loss": 2.4664, "step": 364520 }, { "epoch": 0.0146688, "grad_norm": 0.7963441610336304, "learning_rate": 2.199942050188353e-07, "loss": 2.8448, "step": 364530 }, { "epoch": 0.0146944, "grad_norm": 1.0228084325790405, "learning_rate": 2.1982625388962342e-07, "loss": 2.8833, "step": 364540 }, { "epoch": 0.01472, "grad_norm": 0.8844044208526611, "learning_rate": 2.1965836618223824e-07, "loss": 2.9585, "step": 364550 }, { "epoch": 0.0147456, "grad_norm": 1.6124541759490967, "learning_rate": 2.1949054189776998e-07, "loss": 2.821, "step": 364560 }, { "epoch": 0.0147712, "grad_norm": 0.7768877148628235, "learning_rate": 2.1932278103730554e-07, "loss": 2.7499, "step": 364570 }, { "epoch": 0.0147968, "grad_norm": 0.791305422782898, "learning_rate": 2.1915508360193405e-07, "loss": 2.7568, "step": 364580 }, { "epoch": 0.0148224, "grad_norm": 0.8151006698608398, "learning_rate": 2.1898744959274242e-07, "loss": 2.9122, "step": 364590 }, { "epoch": 0.014848, "grad_norm": 0.7766486406326294, "learning_rate": 2.1881987901081868e-07, "loss": 2.8516, "step": 364600 }, { "epoch": 0.0148736, "grad_norm": 0.9669168591499329, "learning_rate": 2.186523718572464e-07, "loss": 2.8743, "step": 364610 }, { "epoch": 0.0148992, "grad_norm": 0.9367361664772034, "learning_rate": 2.1848492813311474e-07, "loss": 2.7574, "step": 364620 }, { "epoch": 0.0149248, "grad_norm": 1.3336068391799927, "learning_rate": 2.183175478395083e-07, "loss": 2.7405, "step": 364630 }, { "epoch": 0.0149504, "grad_norm": 0.9816034436225891, "learning_rate": 2.1815023097751296e-07, "loss": 2.6864, "step": 364640 }, { "epoch": 0.014976, "grad_norm": 1.2614644765853882, "learning_rate": 2.1798297754821452e-07, "loss": 2.9853, "step": 364650 }, { "epoch": 0.0150016, "grad_norm": 0.8285378813743591, "learning_rate": 2.1781578755269762e-07, "loss": 2.6647, "step": 364660 }, { "epoch": 0.0150272, "grad_norm": 0.7568603754043579, "learning_rate": 2.1764866099204361e-07, "loss": 2.7468, "step": 364670 }, { "epoch": 0.0150528, "grad_norm": 1.1177541017532349, "learning_rate": 2.1748159786733946e-07, "loss": 2.7302, "step": 364680 }, { "epoch": 0.0150784, "grad_norm": 1.4452455043792725, "learning_rate": 2.173145981796665e-07, "loss": 2.9462, "step": 364690 }, { "epoch": 0.015104, "grad_norm": 0.9964115619659424, "learning_rate": 2.1714766193010827e-07, "loss": 2.831, "step": 364700 }, { "epoch": 0.0151296, "grad_norm": 0.9249062538146973, "learning_rate": 2.169807891197473e-07, "loss": 2.8451, "step": 364710 }, { "epoch": 0.0151552, "grad_norm": 0.9198371171951294, "learning_rate": 2.1681397974966712e-07, "loss": 2.8086, "step": 364720 }, { "epoch": 0.0151808, "grad_norm": 0.7892135977745056, "learning_rate": 2.1664723382094798e-07, "loss": 2.8998, "step": 364730 }, { "epoch": 0.0152064, "grad_norm": 1.0196292400360107, "learning_rate": 2.1648055133467127e-07, "loss": 2.794, "step": 364740 }, { "epoch": 0.015232, "grad_norm": 0.8209384679794312, "learning_rate": 2.163139322919172e-07, "loss": 2.7963, "step": 364750 }, { "epoch": 0.0152576, "grad_norm": 0.9072578549385071, "learning_rate": 2.1614737669376718e-07, "loss": 2.8279, "step": 364760 }, { "epoch": 0.0152832, "grad_norm": 1.0545576810836792, "learning_rate": 2.159808845413025e-07, "loss": 2.8264, "step": 364770 }, { "epoch": 0.0153088, "grad_norm": 0.824021577835083, "learning_rate": 2.1581445583560013e-07, "loss": 2.7695, "step": 364780 }, { "epoch": 0.0153344, "grad_norm": 0.8994940519332886, "learning_rate": 2.1564809057774138e-07, "loss": 2.8097, "step": 364790 }, { "epoch": 0.01536, "grad_norm": 0.9929676651954651, "learning_rate": 2.1548178876880431e-07, "loss": 2.7636, "step": 364800 }, { "epoch": 0.0153856, "grad_norm": 0.7800456285476685, "learning_rate": 2.1531555040986696e-07, "loss": 2.5823, "step": 364810 }, { "epoch": 0.0154112, "grad_norm": 0.8496848344802856, "learning_rate": 2.1514937550200842e-07, "loss": 2.8643, "step": 364820 }, { "epoch": 0.0154368, "grad_norm": 0.9071352481842041, "learning_rate": 2.1498326404630453e-07, "loss": 3.0035, "step": 364830 }, { "epoch": 0.0154624, "grad_norm": 1.0140485763549805, "learning_rate": 2.1481721604383443e-07, "loss": 2.8322, "step": 364840 }, { "epoch": 0.015488, "grad_norm": 0.8480036854743958, "learning_rate": 2.1465123149567502e-07, "loss": 2.8131, "step": 364850 }, { "epoch": 0.0155136, "grad_norm": 0.9173872470855713, "learning_rate": 2.1448531040290098e-07, "loss": 2.913, "step": 364860 }, { "epoch": 0.0155392, "grad_norm": 0.9272898435592651, "learning_rate": 2.1431945276658928e-07, "loss": 2.6497, "step": 364870 }, { "epoch": 0.0155648, "grad_norm": 0.9477879405021667, "learning_rate": 2.1415365858781456e-07, "loss": 3.0066, "step": 364880 }, { "epoch": 0.0155904, "grad_norm": 0.8844931125640869, "learning_rate": 2.1398792786765266e-07, "loss": 2.6184, "step": 364890 }, { "epoch": 0.015616, "grad_norm": 0.9722638130187988, "learning_rate": 2.1382226060717827e-07, "loss": 2.8148, "step": 364900 }, { "epoch": 0.0156416, "grad_norm": 0.7899674773216248, "learning_rate": 2.1365665680746607e-07, "loss": 2.7274, "step": 364910 }, { "epoch": 0.0156672, "grad_norm": 0.8354282379150391, "learning_rate": 2.1349111646958965e-07, "loss": 2.7491, "step": 364920 }, { "epoch": 0.0156928, "grad_norm": 0.7859142422676086, "learning_rate": 2.133256395946226e-07, "loss": 2.9364, "step": 364930 }, { "epoch": 0.0157184, "grad_norm": 0.9279155135154724, "learning_rate": 2.1316022618363741e-07, "loss": 2.7572, "step": 364940 }, { "epoch": 0.015744, "grad_norm": 0.7898391485214233, "learning_rate": 2.1299487623770654e-07, "loss": 2.9344, "step": 364950 }, { "epoch": 0.0157696, "grad_norm": 1.2197078466415405, "learning_rate": 2.1282958975790247e-07, "loss": 2.7111, "step": 364960 }, { "epoch": 0.0157952, "grad_norm": 0.8807592391967773, "learning_rate": 2.1266436674529877e-07, "loss": 2.6639, "step": 364970 }, { "epoch": 0.0158208, "grad_norm": 1.4496190547943115, "learning_rate": 2.1249920720096461e-07, "loss": 2.9028, "step": 364980 }, { "epoch": 0.0158464, "grad_norm": 0.8272917866706848, "learning_rate": 2.1233411112597358e-07, "loss": 2.7182, "step": 364990 }, { "epoch": 0.015872, "grad_norm": 0.9983782768249512, "learning_rate": 2.1216907852139258e-07, "loss": 3.168, "step": 365000 }, { "epoch": 0.0158976, "grad_norm": 0.7812590599060059, "learning_rate": 2.1200410938829409e-07, "loss": 2.6144, "step": 365010 }, { "epoch": 0.0159232, "grad_norm": 0.7777432799339294, "learning_rate": 2.1183920372774836e-07, "loss": 2.9172, "step": 365020 }, { "epoch": 0.0159488, "grad_norm": 0.8283504247665405, "learning_rate": 2.1167436154082345e-07, "loss": 2.74, "step": 365030 }, { "epoch": 0.0159744, "grad_norm": 1.4291616678237915, "learning_rate": 2.1150958282858958e-07, "loss": 2.8732, "step": 365040 }, { "epoch": 0.016, "grad_norm": 1.3425005674362183, "learning_rate": 2.113448675921137e-07, "loss": 2.7372, "step": 365050 }, { "epoch": 0.0160256, "grad_norm": 0.8813576698303223, "learning_rate": 2.1118021583246607e-07, "loss": 2.7886, "step": 365060 }, { "epoch": 0.0160512, "grad_norm": 0.8838376998901367, "learning_rate": 2.1101562755071249e-07, "loss": 2.9569, "step": 365070 }, { "epoch": 0.0160768, "grad_norm": 0.9188330769538879, "learning_rate": 2.1085110274791986e-07, "loss": 2.8426, "step": 365080 }, { "epoch": 0.0161024, "grad_norm": 0.8864340782165527, "learning_rate": 2.1068664142515737e-07, "loss": 2.7927, "step": 365090 }, { "epoch": 0.016128, "grad_norm": 1.280700445175171, "learning_rate": 2.1052224358348973e-07, "loss": 2.7696, "step": 365100 }, { "epoch": 0.0161536, "grad_norm": 0.8014338612556458, "learning_rate": 2.1035790922398492e-07, "loss": 2.5634, "step": 365110 }, { "epoch": 0.0161792, "grad_norm": 1.1490633487701416, "learning_rate": 2.1019363834770546e-07, "loss": 2.9049, "step": 365120 }, { "epoch": 0.0162048, "grad_norm": 1.029894471168518, "learning_rate": 2.100294309557205e-07, "loss": 2.602, "step": 365130 }, { "epoch": 0.0162304, "grad_norm": 0.9326921105384827, "learning_rate": 2.0986528704909026e-07, "loss": 2.8917, "step": 365140 }, { "epoch": 0.016256, "grad_norm": 0.8829818367958069, "learning_rate": 2.0970120662888283e-07, "loss": 2.6984, "step": 365150 }, { "epoch": 0.0162816, "grad_norm": 0.868075966835022, "learning_rate": 2.0953718969616177e-07, "loss": 3.0289, "step": 365160 }, { "epoch": 0.0163072, "grad_norm": 0.8487340807914734, "learning_rate": 2.0937323625198956e-07, "loss": 2.898, "step": 365170 }, { "epoch": 0.0163328, "grad_norm": 0.7874289155006409, "learning_rate": 2.0920934629742984e-07, "loss": 2.7421, "step": 365180 }, { "epoch": 0.0163584, "grad_norm": 0.9080002307891846, "learning_rate": 2.0904551983354615e-07, "loss": 2.6746, "step": 365190 }, { "epoch": 0.016384, "grad_norm": 0.8078106045722961, "learning_rate": 2.088817568613999e-07, "loss": 2.7947, "step": 365200 }, { "epoch": 0.0164096, "grad_norm": 1.019513487815857, "learning_rate": 2.0871805738205352e-07, "loss": 2.861, "step": 365210 }, { "epoch": 0.0164352, "grad_norm": 0.8974212408065796, "learning_rate": 2.0855442139656735e-07, "loss": 2.8936, "step": 365220 }, { "epoch": 0.0164608, "grad_norm": 0.783980131149292, "learning_rate": 2.083908489060038e-07, "loss": 3.0031, "step": 365230 }, { "epoch": 0.0164864, "grad_norm": 1.1238141059875488, "learning_rate": 2.082273399114232e-07, "loss": 3.0095, "step": 365240 }, { "epoch": 0.016512, "grad_norm": 0.8781567811965942, "learning_rate": 2.0806389441388575e-07, "loss": 2.7946, "step": 365250 }, { "epoch": 0.0165376, "grad_norm": 0.82115238904953, "learning_rate": 2.0790051241445286e-07, "loss": 2.6641, "step": 365260 }, { "epoch": 0.0165632, "grad_norm": 0.8324518799781799, "learning_rate": 2.0773719391418256e-07, "loss": 2.718, "step": 365270 }, { "epoch": 0.0165888, "grad_norm": 0.9230080842971802, "learning_rate": 2.075739389141329e-07, "loss": 2.7713, "step": 365280 }, { "epoch": 0.0166144, "grad_norm": 0.8021242022514343, "learning_rate": 2.0741074741536416e-07, "loss": 2.9133, "step": 365290 }, { "epoch": 0.01664, "grad_norm": 0.903079628944397, "learning_rate": 2.0724761941893434e-07, "loss": 2.8319, "step": 365300 }, { "epoch": 0.0166656, "grad_norm": 0.943805456161499, "learning_rate": 2.0708455492589929e-07, "loss": 2.9983, "step": 365310 }, { "epoch": 0.0166912, "grad_norm": 0.8991912007331848, "learning_rate": 2.069215539373215e-07, "loss": 2.7992, "step": 365320 }, { "epoch": 0.0167168, "grad_norm": 0.9129119515419006, "learning_rate": 2.0675861645425232e-07, "loss": 2.6872, "step": 365330 }, { "epoch": 0.0167424, "grad_norm": 0.8154653310775757, "learning_rate": 2.0659574247775204e-07, "loss": 3.0246, "step": 365340 }, { "epoch": 0.016768, "grad_norm": 0.8331882357597351, "learning_rate": 2.0643293200887537e-07, "loss": 3.038, "step": 365350 }, { "epoch": 0.0167936, "grad_norm": 1.2984619140625, "learning_rate": 2.0627018504867813e-07, "loss": 2.8012, "step": 365360 }, { "epoch": 0.0168192, "grad_norm": 0.844196081161499, "learning_rate": 2.0610750159821614e-07, "loss": 2.7716, "step": 365370 }, { "epoch": 0.0168448, "grad_norm": 0.9470561742782593, "learning_rate": 2.0594488165854297e-07, "loss": 2.7986, "step": 365380 }, { "epoch": 0.0168704, "grad_norm": 0.8393251895904541, "learning_rate": 2.0578232523071674e-07, "loss": 2.9211, "step": 365390 }, { "epoch": 0.016896, "grad_norm": 0.8623736500740051, "learning_rate": 2.0561983231578652e-07, "loss": 2.7299, "step": 365400 }, { "epoch": 0.0169216, "grad_norm": 0.7854230999946594, "learning_rate": 2.0545740291481042e-07, "loss": 2.776, "step": 365410 }, { "epoch": 0.0169472, "grad_norm": 0.8832940459251404, "learning_rate": 2.0529503702883868e-07, "loss": 2.8471, "step": 365420 }, { "epoch": 0.0169728, "grad_norm": 0.8060401678085327, "learning_rate": 2.0513273465892602e-07, "loss": 2.7924, "step": 365430 }, { "epoch": 0.0169984, "grad_norm": 0.8778260350227356, "learning_rate": 2.0497049580612495e-07, "loss": 3.1053, "step": 365440 }, { "epoch": 0.017024, "grad_norm": 1.6378084421157837, "learning_rate": 2.0480832047148568e-07, "loss": 3.0494, "step": 365450 }, { "epoch": 0.0170496, "grad_norm": 0.9349460005760193, "learning_rate": 2.04646208656063e-07, "loss": 2.8568, "step": 365460 }, { "epoch": 0.0170752, "grad_norm": 0.8254210948944092, "learning_rate": 2.04484160360906e-07, "loss": 2.6955, "step": 365470 }, { "epoch": 0.0171008, "grad_norm": 0.9312901496887207, "learning_rate": 2.0432217558706503e-07, "loss": 2.6612, "step": 365480 }, { "epoch": 0.0171264, "grad_norm": 0.7940033674240112, "learning_rate": 2.041602543355903e-07, "loss": 2.8566, "step": 365490 }, { "epoch": 0.017152, "grad_norm": 1.1248323917388916, "learning_rate": 2.0399839660753428e-07, "loss": 2.806, "step": 365500 }, { "epoch": 0.0171776, "grad_norm": 0.8499471545219421, "learning_rate": 2.038366024039451e-07, "loss": 2.9753, "step": 365510 }, { "epoch": 0.0172032, "grad_norm": 1.0066701173782349, "learning_rate": 2.0367487172587297e-07, "loss": 2.7587, "step": 365520 }, { "epoch": 0.0172288, "grad_norm": 1.9979292154312134, "learning_rate": 2.0351320457436487e-07, "loss": 2.8229, "step": 365530 }, { "epoch": 0.0172544, "grad_norm": 0.9328619241714478, "learning_rate": 2.033516009504699e-07, "loss": 2.842, "step": 365540 }, { "epoch": 0.01728, "grad_norm": 0.9016994833946228, "learning_rate": 2.0319006085523618e-07, "loss": 2.7534, "step": 365550 }, { "epoch": 0.0173056, "grad_norm": 0.8106333017349243, "learning_rate": 2.0302858428971172e-07, "loss": 2.8431, "step": 365560 }, { "epoch": 0.0173312, "grad_norm": 0.9247840046882629, "learning_rate": 2.0286717125494238e-07, "loss": 2.8339, "step": 365570 }, { "epoch": 0.0173568, "grad_norm": 1.0051556825637817, "learning_rate": 2.0270582175197618e-07, "loss": 2.859, "step": 365580 }, { "epoch": 0.0173824, "grad_norm": 1.1338260173797607, "learning_rate": 2.0254453578186006e-07, "loss": 2.6669, "step": 365590 }, { "epoch": 0.017408, "grad_norm": 0.9764415621757507, "learning_rate": 2.0238331334563766e-07, "loss": 2.8603, "step": 365600 }, { "epoch": 0.0174336, "grad_norm": 0.8343079090118408, "learning_rate": 2.0222215444435478e-07, "loss": 2.8928, "step": 365610 }, { "epoch": 0.0174592, "grad_norm": 0.9968980550765991, "learning_rate": 2.0206105907905838e-07, "loss": 2.7635, "step": 365620 }, { "epoch": 0.0174848, "grad_norm": 0.9870953559875488, "learning_rate": 2.0190002725079205e-07, "loss": 2.8206, "step": 365630 }, { "epoch": 0.0175104, "grad_norm": 0.9363947510719299, "learning_rate": 2.0173905896059942e-07, "loss": 2.9386, "step": 365640 }, { "epoch": 0.017536, "grad_norm": 0.8454164862632751, "learning_rate": 2.015781542095252e-07, "loss": 2.8742, "step": 365650 }, { "epoch": 0.0175616, "grad_norm": 1.23631751537323, "learning_rate": 2.0141731299861188e-07, "loss": 3.0468, "step": 365660 }, { "epoch": 0.0175872, "grad_norm": 0.802462637424469, "learning_rate": 2.012565353289042e-07, "loss": 2.7742, "step": 365670 }, { "epoch": 0.0176128, "grad_norm": 0.802431583404541, "learning_rate": 2.0109582120144244e-07, "loss": 2.9231, "step": 365680 }, { "epoch": 0.0176384, "grad_norm": 0.9023292660713196, "learning_rate": 2.009351706172713e-07, "loss": 2.8605, "step": 365690 }, { "epoch": 0.017664, "grad_norm": 0.7909576296806335, "learning_rate": 2.0077458357742995e-07, "loss": 2.831, "step": 365700 }, { "epoch": 0.0176896, "grad_norm": 1.0202064514160156, "learning_rate": 2.0061406008296312e-07, "loss": 2.9463, "step": 365710 }, { "epoch": 0.0177152, "grad_norm": 0.8913434743881226, "learning_rate": 2.0045360013490888e-07, "loss": 2.7355, "step": 365720 }, { "epoch": 0.0177408, "grad_norm": 0.926188051700592, "learning_rate": 2.002932037343075e-07, "loss": 2.9664, "step": 365730 }, { "epoch": 0.0177664, "grad_norm": 0.7625964283943176, "learning_rate": 2.0013287088220145e-07, "loss": 2.907, "step": 365740 }, { "epoch": 0.017792, "grad_norm": 0.806807816028595, "learning_rate": 1.9997260157962882e-07, "loss": 2.8947, "step": 365750 }, { "epoch": 0.0178176, "grad_norm": 0.9175333380699158, "learning_rate": 1.998123958276299e-07, "loss": 2.8161, "step": 365760 }, { "epoch": 0.0178432, "grad_norm": 0.8932063579559326, "learning_rate": 1.9965225362724162e-07, "loss": 3.0502, "step": 365770 }, { "epoch": 0.0178688, "grad_norm": 0.950161874294281, "learning_rate": 1.9949217497950535e-07, "loss": 2.7941, "step": 365780 }, { "epoch": 0.0178944, "grad_norm": 0.8255488276481628, "learning_rate": 1.9933215988545806e-07, "loss": 2.597, "step": 365790 }, { "epoch": 0.01792, "grad_norm": 0.8164666891098022, "learning_rate": 1.9917220834613559e-07, "loss": 2.8006, "step": 365800 }, { "epoch": 0.0179456, "grad_norm": 0.8846801519393921, "learning_rate": 1.9901232036257712e-07, "loss": 2.8202, "step": 365810 }, { "epoch": 0.0179712, "grad_norm": 0.815159022808075, "learning_rate": 1.9885249593581845e-07, "loss": 2.9838, "step": 365820 }, { "epoch": 0.0179968, "grad_norm": 0.844235360622406, "learning_rate": 1.9869273506689658e-07, "loss": 2.8682, "step": 365830 }, { "epoch": 0.0180224, "grad_norm": 0.843027651309967, "learning_rate": 1.985330377568484e-07, "loss": 2.8176, "step": 365840 }, { "epoch": 0.018048, "grad_norm": 0.9094981551170349, "learning_rate": 1.9837340400670758e-07, "loss": 2.899, "step": 365850 }, { "epoch": 0.0180736, "grad_norm": 0.8569741249084473, "learning_rate": 1.9821383381751102e-07, "loss": 2.7178, "step": 365860 }, { "epoch": 0.0180992, "grad_norm": 0.8434032797813416, "learning_rate": 1.9805432719029239e-07, "loss": 2.6496, "step": 365870 }, { "epoch": 0.0181248, "grad_norm": 0.8312823176383972, "learning_rate": 1.9789488412608526e-07, "loss": 2.9281, "step": 365880 }, { "epoch": 0.0181504, "grad_norm": 0.8225303292274475, "learning_rate": 1.977355046259266e-07, "loss": 2.8973, "step": 365890 }, { "epoch": 0.018176, "grad_norm": 0.8901997804641724, "learning_rate": 1.975761886908467e-07, "loss": 2.8617, "step": 365900 }, { "epoch": 0.0182016, "grad_norm": 0.8492472767829895, "learning_rate": 1.9741693632188142e-07, "loss": 2.9957, "step": 365910 }, { "epoch": 0.0182272, "grad_norm": 0.930731475353241, "learning_rate": 1.9725774752006099e-07, "loss": 2.8446, "step": 365920 }, { "epoch": 0.0182528, "grad_norm": 0.8300180435180664, "learning_rate": 1.9709862228641907e-07, "loss": 2.7951, "step": 365930 }, { "epoch": 0.0182784, "grad_norm": 1.4666961431503296, "learning_rate": 1.9693956062198816e-07, "loss": 2.7819, "step": 365940 }, { "epoch": 0.018304, "grad_norm": 0.9326438307762146, "learning_rate": 1.9678056252779742e-07, "loss": 2.9246, "step": 365950 }, { "epoch": 0.0183296, "grad_norm": 0.8196426630020142, "learning_rate": 1.966216280048805e-07, "loss": 2.8859, "step": 365960 }, { "epoch": 0.0183552, "grad_norm": 0.8108111023902893, "learning_rate": 1.9646275705426543e-07, "loss": 2.8002, "step": 365970 }, { "epoch": 0.0183808, "grad_norm": 1.3403139114379883, "learning_rate": 1.9630394967698585e-07, "loss": 2.9466, "step": 365980 }, { "epoch": 0.0184064, "grad_norm": 1.0328372716903687, "learning_rate": 1.9614520587406872e-07, "loss": 2.7334, "step": 365990 }, { "epoch": 0.018432, "grad_norm": 0.9406548738479614, "learning_rate": 1.959865256465443e-07, "loss": 2.8054, "step": 366000 }, { "epoch": 0.0184576, "grad_norm": 0.9939990639686584, "learning_rate": 1.9582790899544067e-07, "loss": 2.8206, "step": 366010 }, { "epoch": 0.0184832, "grad_norm": 0.8747028112411499, "learning_rate": 1.9566935592178815e-07, "loss": 3.0853, "step": 366020 }, { "epoch": 0.0185088, "grad_norm": 1.1468650102615356, "learning_rate": 1.9551086642661476e-07, "loss": 2.8964, "step": 366030 }, { "epoch": 0.0185344, "grad_norm": 0.9197209477424622, "learning_rate": 1.9535244051094859e-07, "loss": 2.8035, "step": 366040 }, { "epoch": 0.01856, "grad_norm": 0.8690014481544495, "learning_rate": 1.9519407817581438e-07, "loss": 2.8409, "step": 366050 }, { "epoch": 0.0185856, "grad_norm": 0.7846123576164246, "learning_rate": 1.9503577942224128e-07, "loss": 2.7457, "step": 366060 }, { "epoch": 0.0186112, "grad_norm": 0.8296294808387756, "learning_rate": 1.9487754425125516e-07, "loss": 2.9438, "step": 366070 }, { "epoch": 0.0186368, "grad_norm": 0.7707160115242004, "learning_rate": 1.9471937266388186e-07, "loss": 2.8523, "step": 366080 }, { "epoch": 0.0186624, "grad_norm": 0.8367302417755127, "learning_rate": 1.9456126466114833e-07, "loss": 2.8795, "step": 366090 }, { "epoch": 0.018688, "grad_norm": 0.8715137839317322, "learning_rate": 1.9440322024407932e-07, "loss": 2.9755, "step": 366100 }, { "epoch": 0.0187136, "grad_norm": 0.822658360004425, "learning_rate": 1.9424523941369954e-07, "loss": 2.8266, "step": 366110 }, { "epoch": 0.0187392, "grad_norm": 0.8286828994750977, "learning_rate": 1.9408732217103156e-07, "loss": 2.9689, "step": 366120 }, { "epoch": 0.0187648, "grad_norm": 0.9860554337501526, "learning_rate": 1.9392946851710225e-07, "loss": 2.9835, "step": 366130 }, { "epoch": 0.0187904, "grad_norm": 0.7762678265571594, "learning_rate": 1.937716784529342e-07, "loss": 2.9923, "step": 366140 }, { "epoch": 0.018816, "grad_norm": 0.7828819155693054, "learning_rate": 1.9361395197955102e-07, "loss": 2.9228, "step": 366150 }, { "epoch": 0.0188416, "grad_norm": 0.9372053742408752, "learning_rate": 1.934562890979741e-07, "loss": 2.7678, "step": 366160 }, { "epoch": 0.0188672, "grad_norm": 0.8474844694137573, "learning_rate": 1.9329868980922817e-07, "loss": 2.9065, "step": 366170 }, { "epoch": 0.0188928, "grad_norm": 0.801654040813446, "learning_rate": 1.9314115411433243e-07, "loss": 2.7176, "step": 366180 }, { "epoch": 0.0189184, "grad_norm": 0.9094372987747192, "learning_rate": 1.9298368201431161e-07, "loss": 2.8284, "step": 366190 }, { "epoch": 0.018944, "grad_norm": 0.8912447690963745, "learning_rate": 1.928262735101838e-07, "loss": 2.8358, "step": 366200 }, { "epoch": 0.0189696, "grad_norm": 0.8270766735076904, "learning_rate": 1.926689286029715e-07, "loss": 2.8335, "step": 366210 }, { "epoch": 0.0189952, "grad_norm": 0.8587064743041992, "learning_rate": 1.9251164729369498e-07, "loss": 2.853, "step": 366220 }, { "epoch": 0.0190208, "grad_norm": 0.7674379944801331, "learning_rate": 1.923544295833746e-07, "loss": 2.7336, "step": 366230 }, { "epoch": 0.0190464, "grad_norm": 0.8909704089164734, "learning_rate": 1.9219727547302836e-07, "loss": 2.7045, "step": 366240 }, { "epoch": 0.019072, "grad_norm": 0.8331953883171082, "learning_rate": 1.920401849636766e-07, "loss": 2.9132, "step": 366250 }, { "epoch": 0.0190976, "grad_norm": 0.8532371520996094, "learning_rate": 1.918831580563374e-07, "loss": 3.0836, "step": 366260 }, { "epoch": 0.0191232, "grad_norm": 0.8068373203277588, "learning_rate": 1.917261947520288e-07, "loss": 2.8809, "step": 366270 }, { "epoch": 0.0191488, "grad_norm": 1.2998781204223633, "learning_rate": 1.9156929505177002e-07, "loss": 2.8815, "step": 366280 }, { "epoch": 0.0191744, "grad_norm": 0.859412431716919, "learning_rate": 1.914124589565769e-07, "loss": 2.8595, "step": 366290 }, { "epoch": 0.0192, "grad_norm": 0.830269455909729, "learning_rate": 1.912556864674675e-07, "loss": 2.8694, "step": 366300 }, { "epoch": 0.0192256, "grad_norm": 0.8429325819015503, "learning_rate": 1.910989775854588e-07, "loss": 2.9581, "step": 366310 }, { "epoch": 0.0192512, "grad_norm": 0.8224455714225769, "learning_rate": 1.9094233231156445e-07, "loss": 2.9938, "step": 366320 }, { "epoch": 0.0192768, "grad_norm": 0.8672990798950195, "learning_rate": 1.9078575064680361e-07, "loss": 2.7576, "step": 366330 }, { "epoch": 0.0193024, "grad_norm": 1.2938759326934814, "learning_rate": 1.906292325921899e-07, "loss": 2.6064, "step": 366340 }, { "epoch": 0.019328, "grad_norm": 0.9164291620254517, "learning_rate": 1.9047277814873812e-07, "loss": 2.6166, "step": 366350 }, { "epoch": 0.0193536, "grad_norm": 0.7870656251907349, "learning_rate": 1.9031638731746294e-07, "loss": 2.6861, "step": 366360 }, { "epoch": 0.0193792, "grad_norm": 0.7826436161994934, "learning_rate": 1.9016006009938027e-07, "loss": 2.5462, "step": 366370 }, { "epoch": 0.0194048, "grad_norm": 0.7731128334999084, "learning_rate": 1.900037964955015e-07, "loss": 2.5046, "step": 366380 }, { "epoch": 0.0194304, "grad_norm": 0.8500122427940369, "learning_rate": 1.8984759650684137e-07, "loss": 2.6494, "step": 366390 }, { "epoch": 0.019456, "grad_norm": 0.8290590643882751, "learning_rate": 1.8969146013441243e-07, "loss": 2.5718, "step": 366400 }, { "epoch": 0.0194816, "grad_norm": 0.7700377106666565, "learning_rate": 1.8953538737922606e-07, "loss": 2.7003, "step": 366410 }, { "epoch": 0.0195072, "grad_norm": 0.7831859588623047, "learning_rate": 1.893793782422959e-07, "loss": 2.6697, "step": 366420 }, { "epoch": 0.0195328, "grad_norm": 0.8233271837234497, "learning_rate": 1.8922343272463338e-07, "loss": 2.5925, "step": 366430 }, { "epoch": 0.0195584, "grad_norm": 0.9717197418212891, "learning_rate": 1.8906755082725104e-07, "loss": 2.6541, "step": 366440 }, { "epoch": 0.019584, "grad_norm": 1.4775009155273438, "learning_rate": 1.889117325511569e-07, "loss": 2.1023, "step": 366450 }, { "epoch": 0.0196096, "grad_norm": 0.8071603775024414, "learning_rate": 1.8875597789736245e-07, "loss": 2.687, "step": 366460 }, { "epoch": 0.0196352, "grad_norm": 1.0166943073272705, "learning_rate": 1.8860028686687794e-07, "loss": 3.0383, "step": 366470 }, { "epoch": 0.0196608, "grad_norm": 0.7672187685966492, "learning_rate": 1.8844465946071255e-07, "loss": 2.4995, "step": 366480 }, { "epoch": 0.0196864, "grad_norm": 0.993826687335968, "learning_rate": 1.8828909567987663e-07, "loss": 2.7395, "step": 366490 }, { "epoch": 0.019712, "grad_norm": 1.0452905893325806, "learning_rate": 1.8813359552537824e-07, "loss": 2.5004, "step": 366500 }, { "epoch": 0.0197376, "grad_norm": 1.098505973815918, "learning_rate": 1.879781589982266e-07, "loss": 2.7021, "step": 366510 }, { "epoch": 0.0197632, "grad_norm": 0.8087701201438904, "learning_rate": 1.878227860994275e-07, "loss": 2.631, "step": 366520 }, { "epoch": 0.0197888, "grad_norm": 0.9594295620918274, "learning_rate": 1.876674768299891e-07, "loss": 2.4544, "step": 366530 }, { "epoch": 0.0198144, "grad_norm": 0.7545586228370667, "learning_rate": 1.8751223119092055e-07, "loss": 2.5989, "step": 366540 }, { "epoch": 0.01984, "grad_norm": 0.8546837568283081, "learning_rate": 1.8735704918322773e-07, "loss": 2.5814, "step": 366550 }, { "epoch": 0.0198656, "grad_norm": 0.8201121687889099, "learning_rate": 1.8720193080791648e-07, "loss": 2.5616, "step": 366560 }, { "epoch": 0.0198912, "grad_norm": 1.0781073570251465, "learning_rate": 1.870468760659927e-07, "loss": 2.8365, "step": 366570 }, { "epoch": 0.0199168, "grad_norm": 0.9245366454124451, "learning_rate": 1.8689188495846223e-07, "loss": 2.6375, "step": 366580 }, { "epoch": 0.0199424, "grad_norm": 0.8878142237663269, "learning_rate": 1.867369574863287e-07, "loss": 3.0344, "step": 366590 }, { "epoch": 0.019968, "grad_norm": 0.8136813044548035, "learning_rate": 1.865820936505991e-07, "loss": 2.8677, "step": 366600 }, { "epoch": 0.0199936, "grad_norm": 0.8828948140144348, "learning_rate": 1.8642729345227595e-07, "loss": 2.6505, "step": 366610 }, { "epoch": 0.0200192, "grad_norm": 0.8048132658004761, "learning_rate": 1.862725568923629e-07, "loss": 2.6702, "step": 366620 }, { "epoch": 0.0200448, "grad_norm": 0.7888104915618896, "learning_rate": 1.861178839718658e-07, "loss": 2.5663, "step": 366630 }, { "epoch": 0.0200704, "grad_norm": 0.7573742866516113, "learning_rate": 1.85963274691785e-07, "loss": 2.3406, "step": 366640 }, { "epoch": 0.020096, "grad_norm": 0.7979806661605835, "learning_rate": 1.8580872905312518e-07, "loss": 2.6931, "step": 366650 }, { "epoch": 0.0201216, "grad_norm": 0.8531313538551331, "learning_rate": 1.8565424705688562e-07, "loss": 2.5232, "step": 366660 }, { "epoch": 0.0201472, "grad_norm": 0.8999614119529724, "learning_rate": 1.8549982870407102e-07, "loss": 2.7056, "step": 366670 }, { "epoch": 0.0201728, "grad_norm": 0.835681676864624, "learning_rate": 1.853454739956817e-07, "loss": 2.4647, "step": 366680 }, { "epoch": 0.0201984, "grad_norm": 1.0641309022903442, "learning_rate": 1.8519118293271799e-07, "loss": 2.9199, "step": 366690 }, { "epoch": 0.020224, "grad_norm": 0.8267982006072998, "learning_rate": 1.8503695551618128e-07, "loss": 2.7544, "step": 366700 }, { "epoch": 0.0202496, "grad_norm": 0.8920598030090332, "learning_rate": 1.848827917470719e-07, "loss": 2.504, "step": 366710 }, { "epoch": 0.0202752, "grad_norm": 0.8508153557777405, "learning_rate": 1.8472869162638907e-07, "loss": 2.6653, "step": 366720 }, { "epoch": 0.0203008, "grad_norm": 0.9852798581123352, "learning_rate": 1.8457465515513084e-07, "loss": 2.8512, "step": 366730 }, { "epoch": 0.0203264, "grad_norm": 0.9133526682853699, "learning_rate": 1.8442068233429867e-07, "loss": 2.7472, "step": 366740 }, { "epoch": 0.020352, "grad_norm": 0.8049320578575134, "learning_rate": 1.842667731648895e-07, "loss": 2.5588, "step": 366750 }, { "epoch": 0.0203776, "grad_norm": 0.855380654335022, "learning_rate": 1.8411292764790035e-07, "loss": 2.5776, "step": 366760 }, { "epoch": 0.0204032, "grad_norm": 0.8118993043899536, "learning_rate": 1.8395914578433261e-07, "loss": 2.7015, "step": 366770 }, { "epoch": 0.0204288, "grad_norm": 0.8909431099891663, "learning_rate": 1.8380542757517882e-07, "loss": 2.7486, "step": 366780 }, { "epoch": 0.0204544, "grad_norm": 1.0658231973648071, "learning_rate": 1.8365177302143934e-07, "loss": 2.8642, "step": 366790 }, { "epoch": 0.02048, "grad_norm": 0.8136914968490601, "learning_rate": 1.8349818212410776e-07, "loss": 2.7422, "step": 366800 }, { "epoch": 0.0205056, "grad_norm": 0.7851590514183044, "learning_rate": 1.833446548841833e-07, "loss": 2.7109, "step": 366810 }, { "epoch": 0.0205312, "grad_norm": 0.8367952108383179, "learning_rate": 1.831911913026585e-07, "loss": 2.7628, "step": 366820 }, { "epoch": 0.0205568, "grad_norm": 1.0085437297821045, "learning_rate": 1.8303779138053036e-07, "loss": 2.6628, "step": 366830 }, { "epoch": 0.0205824, "grad_norm": 0.8612490296363831, "learning_rate": 1.828844551187936e-07, "loss": 2.6611, "step": 366840 }, { "epoch": 0.020608, "grad_norm": 0.9060025811195374, "learning_rate": 1.827311825184408e-07, "loss": 2.4517, "step": 366850 }, { "epoch": 0.0206336, "grad_norm": 0.8236687183380127, "learning_rate": 1.825779735804678e-07, "loss": 2.5192, "step": 366860 }, { "epoch": 0.0206592, "grad_norm": 0.809404194355011, "learning_rate": 1.8242482830586826e-07, "loss": 2.7471, "step": 366870 }, { "epoch": 0.0206848, "grad_norm": 1.0102369785308838, "learning_rate": 1.8227174669563252e-07, "loss": 2.597, "step": 366880 }, { "epoch": 0.0207104, "grad_norm": 0.7770842909812927, "learning_rate": 1.8211872875075641e-07, "loss": 2.5674, "step": 366890 }, { "epoch": 0.020736, "grad_norm": 0.766329288482666, "learning_rate": 1.819657744722314e-07, "loss": 2.4913, "step": 366900 }, { "epoch": 0.0207616, "grad_norm": 1.3866709470748901, "learning_rate": 1.818128838610489e-07, "loss": 2.0568, "step": 366910 }, { "epoch": 0.0207872, "grad_norm": 0.8812383413314819, "learning_rate": 1.8166005691820033e-07, "loss": 2.6163, "step": 366920 }, { "epoch": 0.0208128, "grad_norm": 0.8697801232337952, "learning_rate": 1.815072936446771e-07, "loss": 2.6602, "step": 366930 }, { "epoch": 0.0208384, "grad_norm": 0.7930081486701965, "learning_rate": 1.813545940414696e-07, "loss": 2.8607, "step": 366940 }, { "epoch": 0.020864, "grad_norm": 0.834923267364502, "learning_rate": 1.8120195810956698e-07, "loss": 2.7814, "step": 366950 }, { "epoch": 0.0208896, "grad_norm": 0.8951787352561951, "learning_rate": 1.810493858499618e-07, "loss": 2.6811, "step": 366960 }, { "epoch": 0.0209152, "grad_norm": 0.864902913570404, "learning_rate": 1.8089687726364103e-07, "loss": 2.798, "step": 366970 }, { "epoch": 0.0209408, "grad_norm": 0.801296591758728, "learning_rate": 1.80744432351595e-07, "loss": 2.5881, "step": 366980 }, { "epoch": 0.0209664, "grad_norm": 0.9032878875732422, "learning_rate": 1.8059205111481072e-07, "loss": 2.5305, "step": 366990 }, { "epoch": 0.020992, "grad_norm": 0.966477632522583, "learning_rate": 1.8043973355427735e-07, "loss": 2.9438, "step": 367000 }, { "epoch": 2.56e-05, "grad_norm": 0.8267191052436829, "learning_rate": 1.8028747967098305e-07, "loss": 1.9203, "step": 367010 }, { "epoch": 5.12e-05, "grad_norm": 0.8337881565093994, "learning_rate": 1.8013528946591474e-07, "loss": 2.5118, "step": 367020 }, { "epoch": 7.68e-05, "grad_norm": 0.8038837909698486, "learning_rate": 1.7998316294005946e-07, "loss": 2.5788, "step": 367030 }, { "epoch": 0.0001024, "grad_norm": 0.7397436499595642, "learning_rate": 1.7983110009440308e-07, "loss": 2.7175, "step": 367040 }, { "epoch": 0.000128, "grad_norm": 0.8880089521408081, "learning_rate": 1.7967910092993145e-07, "loss": 2.4957, "step": 367050 }, { "epoch": 0.0001536, "grad_norm": 0.9095824360847473, "learning_rate": 1.7952716544763161e-07, "loss": 2.4841, "step": 367060 }, { "epoch": 0.0001792, "grad_norm": 0.8175983428955078, "learning_rate": 1.7937529364848827e-07, "loss": 2.5844, "step": 367070 }, { "epoch": 0.0002048, "grad_norm": 0.8145293593406677, "learning_rate": 1.7922348553348623e-07, "loss": 2.6662, "step": 367080 }, { "epoch": 0.0002304, "grad_norm": 0.7906306385993958, "learning_rate": 1.7907174110360914e-07, "loss": 2.5774, "step": 367090 }, { "epoch": 0.000256, "grad_norm": 0.9518892765045166, "learning_rate": 1.7892006035984289e-07, "loss": 2.7299, "step": 367100 }, { "epoch": 0.0002816, "grad_norm": 0.7968716025352478, "learning_rate": 1.7876844330317e-07, "loss": 2.5992, "step": 367110 }, { "epoch": 0.0003072, "grad_norm": 0.8054817318916321, "learning_rate": 1.7861688993457193e-07, "loss": 2.5681, "step": 367120 }, { "epoch": 0.0003328, "grad_norm": 0.8697441816329956, "learning_rate": 1.7846540025503455e-07, "loss": 2.6314, "step": 367130 }, { "epoch": 0.0003584, "grad_norm": 0.8364700675010681, "learning_rate": 1.7831397426553821e-07, "loss": 2.8386, "step": 367140 }, { "epoch": 0.000384, "grad_norm": 0.8542020320892334, "learning_rate": 1.781626119670654e-07, "loss": 2.5822, "step": 367150 }, { "epoch": 0.0004096, "grad_norm": 0.7774695158004761, "learning_rate": 1.7801131336059764e-07, "loss": 2.5287, "step": 367160 }, { "epoch": 0.0004352, "grad_norm": 0.7836558222770691, "learning_rate": 1.778600784471174e-07, "loss": 2.6785, "step": 367170 }, { "epoch": 0.0004608, "grad_norm": 0.9077991247177124, "learning_rate": 1.777089072276028e-07, "loss": 2.8024, "step": 367180 }, { "epoch": 0.0004864, "grad_norm": 0.8727476000785828, "learning_rate": 1.7755779970303533e-07, "loss": 2.5169, "step": 367190 }, { "epoch": 0.000512, "grad_norm": 0.8141863346099854, "learning_rate": 1.7740675587439638e-07, "loss": 2.6438, "step": 367200 }, { "epoch": 0.0005376, "grad_norm": 0.774531364440918, "learning_rate": 1.7725577574266294e-07, "loss": 2.4574, "step": 367210 }, { "epoch": 0.0005632, "grad_norm": 0.8264868855476379, "learning_rate": 1.771048593088154e-07, "loss": 2.5991, "step": 367220 }, { "epoch": 0.0005888, "grad_norm": 0.7917738556861877, "learning_rate": 1.7695400657383177e-07, "loss": 2.5646, "step": 367230 }, { "epoch": 0.0006144, "grad_norm": 0.8414123058319092, "learning_rate": 1.7680321753869135e-07, "loss": 2.5907, "step": 367240 }, { "epoch": 0.00064, "grad_norm": 0.7617685198783875, "learning_rate": 1.7665249220437107e-07, "loss": 2.453, "step": 367250 }, { "epoch": 0.0006656, "grad_norm": 0.8986615538597107, "learning_rate": 1.7650183057184911e-07, "loss": 2.5539, "step": 367260 }, { "epoch": 0.0006912, "grad_norm": 0.8148935437202454, "learning_rate": 1.763512326421013e-07, "loss": 2.5956, "step": 367270 }, { "epoch": 0.0007168, "grad_norm": 0.8198615312576294, "learning_rate": 1.7620069841610578e-07, "loss": 2.544, "step": 367280 }, { "epoch": 0.0007424, "grad_norm": 0.8241762518882751, "learning_rate": 1.7605022789483728e-07, "loss": 2.6031, "step": 367290 }, { "epoch": 0.000768, "grad_norm": 0.8747535943984985, "learning_rate": 1.7589982107927283e-07, "loss": 2.7366, "step": 367300 }, { "epoch": 0.0007936, "grad_norm": 0.9502895474433899, "learning_rate": 1.757494779703872e-07, "loss": 2.7969, "step": 367310 }, { "epoch": 0.0008192, "grad_norm": 0.8323652148246765, "learning_rate": 1.755991985691541e-07, "loss": 2.2951, "step": 367320 }, { "epoch": 0.0008448, "grad_norm": 0.8752267360687256, "learning_rate": 1.7544898287655044e-07, "loss": 2.6873, "step": 367330 }, { "epoch": 0.0008704, "grad_norm": 1.0454217195510864, "learning_rate": 1.7529883089354771e-07, "loss": 2.596, "step": 367340 }, { "epoch": 0.000896, "grad_norm": 0.9827035069465637, "learning_rate": 1.7514874262112181e-07, "loss": 2.6567, "step": 367350 }, { "epoch": 0.0009216, "grad_norm": 0.8451061844825745, "learning_rate": 1.7499871806024527e-07, "loss": 2.5281, "step": 367360 }, { "epoch": 0.0009472, "grad_norm": 1.0052504539489746, "learning_rate": 1.7484875721189178e-07, "loss": 2.626, "step": 367370 }, { "epoch": 0.0009728, "grad_norm": 0.8360684514045715, "learning_rate": 1.7469886007703163e-07, "loss": 2.6532, "step": 367380 }, { "epoch": 0.0009984, "grad_norm": 0.7870548963546753, "learning_rate": 1.745490266566374e-07, "loss": 2.5319, "step": 367390 }, { "epoch": 0.001024, "grad_norm": 0.7822665572166443, "learning_rate": 1.743992569516817e-07, "loss": 2.4357, "step": 367400 }, { "epoch": 0.0010496, "grad_norm": 0.8200180530548096, "learning_rate": 1.7424955096313588e-07, "loss": 2.6866, "step": 367410 }, { "epoch": 0.0010752, "grad_norm": 0.8240225911140442, "learning_rate": 1.7409990869197146e-07, "loss": 2.5891, "step": 367420 }, { "epoch": 0.0011008, "grad_norm": 0.8366459012031555, "learning_rate": 1.7395033013915653e-07, "loss": 2.522, "step": 367430 }, { "epoch": 0.0011264, "grad_norm": 0.8602209687232971, "learning_rate": 1.738008153056625e-07, "loss": 2.607, "step": 367440 }, { "epoch": 0.001152, "grad_norm": 0.8162164092063904, "learning_rate": 1.7365136419245864e-07, "loss": 2.4757, "step": 367450 }, { "epoch": 0.0011776, "grad_norm": 0.9314992427825928, "learning_rate": 1.7350197680051417e-07, "loss": 2.6729, "step": 367460 }, { "epoch": 0.0012032, "grad_norm": 0.8355234861373901, "learning_rate": 1.733526531307983e-07, "loss": 2.8411, "step": 367470 }, { "epoch": 0.0012288, "grad_norm": 0.8509983420372009, "learning_rate": 1.7320339318427804e-07, "loss": 2.8699, "step": 367480 }, { "epoch": 0.0012544, "grad_norm": 0.8132219314575195, "learning_rate": 1.730541969619226e-07, "loss": 2.8816, "step": 367490 }, { "epoch": 0.00128, "grad_norm": 0.8836382031440735, "learning_rate": 1.7290506446470013e-07, "loss": 2.5201, "step": 367500 }, { "epoch": 0.0013056, "grad_norm": 0.960821270942688, "learning_rate": 1.7275599569357538e-07, "loss": 2.567, "step": 367510 }, { "epoch": 0.0013312, "grad_norm": 0.8890186548233032, "learning_rate": 1.7260699064951537e-07, "loss": 2.5247, "step": 367520 }, { "epoch": 0.0013568, "grad_norm": 0.9241647720336914, "learning_rate": 1.724580493334882e-07, "loss": 2.5347, "step": 367530 }, { "epoch": 0.0013824, "grad_norm": 0.9225901961326599, "learning_rate": 1.7230917174645866e-07, "loss": 2.6373, "step": 367540 }, { "epoch": 0.001408, "grad_norm": 0.816670298576355, "learning_rate": 1.7216035788939267e-07, "loss": 2.798, "step": 367550 }, { "epoch": 0.0014336, "grad_norm": 0.8368635773658752, "learning_rate": 1.7201160776325387e-07, "loss": 2.5289, "step": 367560 }, { "epoch": 0.0014592, "grad_norm": 0.8199846148490906, "learning_rate": 1.7186292136900817e-07, "loss": 2.4993, "step": 367570 }, { "epoch": 0.0014848, "grad_norm": 0.8727653622627258, "learning_rate": 1.7171429870761923e-07, "loss": 2.4608, "step": 367580 }, { "epoch": 0.0015104, "grad_norm": 0.8498931527137756, "learning_rate": 1.7156573978005185e-07, "loss": 2.4014, "step": 367590 }, { "epoch": 0.001536, "grad_norm": 0.8153261542320251, "learning_rate": 1.714172445872675e-07, "loss": 2.1365, "step": 367600 }, { "epoch": 0.0015616, "grad_norm": 0.9299904108047485, "learning_rate": 1.7126881313023093e-07, "loss": 2.631, "step": 367610 }, { "epoch": 0.0015872, "grad_norm": 0.9496352076530457, "learning_rate": 1.711204454099047e-07, "loss": 2.519, "step": 367620 }, { "epoch": 0.0016128, "grad_norm": 1.0269241333007812, "learning_rate": 1.7097214142724915e-07, "loss": 2.6552, "step": 367630 }, { "epoch": 0.0016384, "grad_norm": 0.8360219597816467, "learning_rate": 1.708239011832269e-07, "loss": 2.7501, "step": 367640 }, { "epoch": 0.001664, "grad_norm": 0.8975163102149963, "learning_rate": 1.7067572467879933e-07, "loss": 2.6969, "step": 367650 }, { "epoch": 0.0016896, "grad_norm": 0.9335447549819946, "learning_rate": 1.7052761191492795e-07, "loss": 2.9345, "step": 367660 }, { "epoch": 0.0017152, "grad_norm": 0.9258158802986145, "learning_rate": 1.7037956289257195e-07, "loss": 2.5086, "step": 367670 }, { "epoch": 0.0017408, "grad_norm": 0.8751955628395081, "learning_rate": 1.7023157761269282e-07, "loss": 2.726, "step": 367680 }, { "epoch": 0.0017664, "grad_norm": 0.7875151634216309, "learning_rate": 1.7008365607624977e-07, "loss": 2.6073, "step": 367690 }, { "epoch": 0.001792, "grad_norm": 0.87748122215271, "learning_rate": 1.6993579828420092e-07, "loss": 2.8742, "step": 367700 }, { "epoch": 0.0018176, "grad_norm": 0.8367792963981628, "learning_rate": 1.6978800423750663e-07, "loss": 2.5061, "step": 367710 }, { "epoch": 0.0018432, "grad_norm": 0.8704999089241028, "learning_rate": 1.696402739371239e-07, "loss": 2.5925, "step": 367720 }, { "epoch": 0.0018688, "grad_norm": 0.8710216283798218, "learning_rate": 1.6949260738401086e-07, "loss": 2.4746, "step": 367730 }, { "epoch": 0.0018944, "grad_norm": 1.0371180772781372, "learning_rate": 1.6934500457912672e-07, "loss": 2.9416, "step": 367740 }, { "epoch": 0.00192, "grad_norm": 0.8935965895652771, "learning_rate": 1.6919746552342742e-07, "loss": 2.6962, "step": 367750 }, { "epoch": 0.0019456, "grad_norm": 0.7748476266860962, "learning_rate": 1.6904999021786882e-07, "loss": 2.8599, "step": 367760 }, { "epoch": 0.0019712, "grad_norm": 0.726135790348053, "learning_rate": 1.6890257866340909e-07, "loss": 2.447, "step": 367770 }, { "epoch": 0.0019968, "grad_norm": 0.8415725231170654, "learning_rate": 1.6875523086100297e-07, "loss": 2.6564, "step": 367780 }, { "epoch": 0.0020224, "grad_norm": 0.9486402273178101, "learning_rate": 1.6860794681160642e-07, "loss": 2.9465, "step": 367790 }, { "epoch": 0.002048, "grad_norm": 0.8548578023910522, "learning_rate": 1.684607265161753e-07, "loss": 2.5099, "step": 367800 }, { "epoch": 0.0020736, "grad_norm": 0.8275979161262512, "learning_rate": 1.6831356997566219e-07, "loss": 2.679, "step": 367810 }, { "epoch": 0.0020992, "grad_norm": 0.8779551982879639, "learning_rate": 1.6816647719102407e-07, "loss": 2.8666, "step": 367820 }, { "epoch": 0.0021248, "grad_norm": 0.8998724818229675, "learning_rate": 1.6801944816321248e-07, "loss": 2.8225, "step": 367830 }, { "epoch": 0.0021504, "grad_norm": 0.9004455804824829, "learning_rate": 1.6787248289318215e-07, "loss": 2.7555, "step": 367840 }, { "epoch": 0.002176, "grad_norm": 0.9785685539245605, "learning_rate": 1.6772558138188456e-07, "loss": 2.8577, "step": 367850 }, { "epoch": 0.0022016, "grad_norm": 0.935064435005188, "learning_rate": 1.6757874363027448e-07, "loss": 2.8131, "step": 367860 }, { "epoch": 0.0022272, "grad_norm": 0.9030582308769226, "learning_rate": 1.6743196963930342e-07, "loss": 2.6719, "step": 367870 }, { "epoch": 0.0022528, "grad_norm": 0.8616436719894409, "learning_rate": 1.672852594099217e-07, "loss": 2.7653, "step": 367880 }, { "epoch": 0.0022784, "grad_norm": 0.8210144639015198, "learning_rate": 1.6713861294308298e-07, "loss": 2.4029, "step": 367890 }, { "epoch": 0.002304, "grad_norm": 1.1459712982177734, "learning_rate": 1.6699203023973652e-07, "loss": 2.2888, "step": 367900 }, { "epoch": 0.0023296, "grad_norm": 0.9826784729957581, "learning_rate": 1.6684551130083382e-07, "loss": 2.5288, "step": 367910 }, { "epoch": 0.0023552, "grad_norm": 0.8795647621154785, "learning_rate": 1.6669905612732405e-07, "loss": 2.5513, "step": 367920 }, { "epoch": 0.0023808, "grad_norm": 1.0239601135253906, "learning_rate": 1.6655266472015762e-07, "loss": 2.7217, "step": 367930 }, { "epoch": 0.0024064, "grad_norm": 0.8992058038711548, "learning_rate": 1.6640633708028376e-07, "loss": 2.6401, "step": 367940 }, { "epoch": 0.002432, "grad_norm": 0.888455331325531, "learning_rate": 1.6626007320865167e-07, "loss": 2.7465, "step": 367950 }, { "epoch": 0.0024576, "grad_norm": 0.7969071865081787, "learning_rate": 1.6611387310620953e-07, "loss": 2.8358, "step": 367960 }, { "epoch": 0.0024832, "grad_norm": 0.7880277633666992, "learning_rate": 1.6596773677390543e-07, "loss": 2.5631, "step": 367970 }, { "epoch": 0.0025088, "grad_norm": 0.8497190475463867, "learning_rate": 1.658216642126864e-07, "loss": 2.7044, "step": 367980 }, { "epoch": 0.0025344, "grad_norm": 0.8087220191955566, "learning_rate": 1.6567565542350062e-07, "loss": 2.5924, "step": 367990 }, { "epoch": 0.00256, "grad_norm": 0.8320755958557129, "learning_rate": 1.6552971040729503e-07, "loss": 2.6027, "step": 368000 }, { "epoch": 0.0025856, "grad_norm": 0.857306182384491, "learning_rate": 1.653838291650145e-07, "loss": 2.7819, "step": 368010 }, { "epoch": 0.0026112, "grad_norm": 0.7732520699501038, "learning_rate": 1.6523801169760712e-07, "loss": 2.7848, "step": 368020 }, { "epoch": 0.0026368, "grad_norm": 0.83184415102005, "learning_rate": 1.6509225800601768e-07, "loss": 2.7925, "step": 368030 }, { "epoch": 0.0026624, "grad_norm": 0.8883686065673828, "learning_rate": 1.6494656809118992e-07, "loss": 2.9184, "step": 368040 }, { "epoch": 0.002688, "grad_norm": 0.9090946316719055, "learning_rate": 1.6480094195407083e-07, "loss": 2.3823, "step": 368050 }, { "epoch": 0.0027136, "grad_norm": 1.5706933736801147, "learning_rate": 1.6465537959560296e-07, "loss": 2.8461, "step": 368060 }, { "epoch": 0.0027392, "grad_norm": 0.8433667421340942, "learning_rate": 1.6450988101673226e-07, "loss": 2.7598, "step": 368070 }, { "epoch": 0.0027648, "grad_norm": 0.8105230331420898, "learning_rate": 1.6436444621839909e-07, "loss": 2.7052, "step": 368080 }, { "epoch": 0.0027904, "grad_norm": 0.8356795310974121, "learning_rate": 1.6421907520155046e-07, "loss": 2.8313, "step": 368090 }, { "epoch": 0.002816, "grad_norm": 0.8650602102279663, "learning_rate": 1.640737679671256e-07, "loss": 2.6807, "step": 368100 }, { "epoch": 0.0028416, "grad_norm": 0.8339412212371826, "learning_rate": 1.6392852451606933e-07, "loss": 2.8444, "step": 368110 }, { "epoch": 0.0028672, "grad_norm": 0.8375227451324463, "learning_rate": 1.63783344849322e-07, "loss": 2.8545, "step": 368120 }, { "epoch": 0.0028928, "grad_norm": 0.8679588437080383, "learning_rate": 1.6363822896782512e-07, "loss": 2.8379, "step": 368130 }, { "epoch": 0.0029184, "grad_norm": 0.8910360336303711, "learning_rate": 1.6349317687252009e-07, "loss": 2.7728, "step": 368140 }, { "epoch": 0.002944, "grad_norm": 0.8842982649803162, "learning_rate": 1.6334818856434953e-07, "loss": 2.8075, "step": 368150 }, { "epoch": 0.0029696, "grad_norm": 0.8654004335403442, "learning_rate": 1.6320326404425047e-07, "loss": 2.813, "step": 368160 }, { "epoch": 0.0029952, "grad_norm": 0.9053752422332764, "learning_rate": 1.6305840331316326e-07, "loss": 2.9038, "step": 368170 }, { "epoch": 0.0030208, "grad_norm": 0.9095154404640198, "learning_rate": 1.6291360637202825e-07, "loss": 2.9011, "step": 368180 }, { "epoch": 0.0030464, "grad_norm": 1.244239091873169, "learning_rate": 1.627688732217847e-07, "loss": 2.8916, "step": 368190 }, { "epoch": 0.003072, "grad_norm": 0.7913639545440674, "learning_rate": 1.6262420386336963e-07, "loss": 2.7624, "step": 368200 }, { "epoch": 0.0030976, "grad_norm": 0.8298439979553223, "learning_rate": 1.624795982977223e-07, "loss": 2.7865, "step": 368210 }, { "epoch": 0.0031232, "grad_norm": 0.9113932847976685, "learning_rate": 1.6233505652578196e-07, "loss": 2.7691, "step": 368220 }, { "epoch": 0.0031488, "grad_norm": 0.8423729538917542, "learning_rate": 1.621905785484823e-07, "loss": 2.877, "step": 368230 }, { "epoch": 0.0031744, "grad_norm": 0.957572340965271, "learning_rate": 1.6204616436676257e-07, "loss": 2.5671, "step": 368240 }, { "epoch": 0.0032, "grad_norm": 0.8612474203109741, "learning_rate": 1.619018139815598e-07, "loss": 2.6734, "step": 368250 }, { "epoch": 0.0032256, "grad_norm": 0.797757625579834, "learning_rate": 1.6175752739380768e-07, "loss": 3.0635, "step": 368260 }, { "epoch": 0.0032512, "grad_norm": 0.8366543054580688, "learning_rate": 1.6161330460444324e-07, "loss": 2.8704, "step": 368270 }, { "epoch": 0.0032768, "grad_norm": 0.8423455953598022, "learning_rate": 1.6146914561440353e-07, "loss": 2.8799, "step": 368280 }, { "epoch": 0.0033024, "grad_norm": 1.5560188293457031, "learning_rate": 1.6132505042462e-07, "loss": 2.6492, "step": 368290 }, { "epoch": 0.003328, "grad_norm": 0.8201000690460205, "learning_rate": 1.6118101903602967e-07, "loss": 2.7214, "step": 368300 }, { "epoch": 0.0033536, "grad_norm": 0.8755051493644714, "learning_rate": 1.6103705144956516e-07, "loss": 2.8238, "step": 368310 }, { "epoch": 0.0033792, "grad_norm": 0.8736170530319214, "learning_rate": 1.6089314766616125e-07, "loss": 2.9069, "step": 368320 }, { "epoch": 0.0034048, "grad_norm": 1.150964379310608, "learning_rate": 1.6074930768674947e-07, "loss": 2.7038, "step": 368330 }, { "epoch": 0.0034304, "grad_norm": 0.8296375274658203, "learning_rate": 1.6060553151226345e-07, "loss": 2.9648, "step": 368340 }, { "epoch": 0.003456, "grad_norm": 1.245091199874878, "learning_rate": 1.604618191436369e-07, "loss": 3.0319, "step": 368350 }, { "epoch": 0.0034816, "grad_norm": 0.8646541833877563, "learning_rate": 1.603181705817991e-07, "loss": 2.7603, "step": 368360 }, { "epoch": 0.0035072, "grad_norm": 0.7911182641983032, "learning_rate": 1.6017458582768263e-07, "loss": 2.676, "step": 368370 }, { "epoch": 0.0035328, "grad_norm": 0.785224199295044, "learning_rate": 1.6003106488222008e-07, "loss": 2.5508, "step": 368380 }, { "epoch": 0.0035584, "grad_norm": 1.0579078197479248, "learning_rate": 1.598876077463396e-07, "loss": 2.8901, "step": 368390 }, { "epoch": 0.003584, "grad_norm": 0.8509277105331421, "learning_rate": 1.5974421442097377e-07, "loss": 2.8731, "step": 368400 }, { "epoch": 0.0036096, "grad_norm": 0.8652120232582092, "learning_rate": 1.5960088490705071e-07, "loss": 2.949, "step": 368410 }, { "epoch": 0.0036352, "grad_norm": 0.9587907195091248, "learning_rate": 1.5945761920550196e-07, "loss": 2.7927, "step": 368420 }, { "epoch": 0.0036608, "grad_norm": 0.8363537192344666, "learning_rate": 1.593144173172545e-07, "loss": 2.9182, "step": 368430 }, { "epoch": 0.0036864, "grad_norm": 0.8545597195625305, "learning_rate": 1.591712792432365e-07, "loss": 2.8419, "step": 368440 }, { "epoch": 0.003712, "grad_norm": 0.8244367241859436, "learning_rate": 1.5902820498437832e-07, "loss": 2.835, "step": 368450 }, { "epoch": 0.0037376, "grad_norm": 0.8367874026298523, "learning_rate": 1.58885194541607e-07, "loss": 2.901, "step": 368460 }, { "epoch": 0.0037632, "grad_norm": 0.8284346461296082, "learning_rate": 1.5874224791584959e-07, "loss": 2.6379, "step": 368470 }, { "epoch": 0.0037888, "grad_norm": 0.8336584568023682, "learning_rate": 1.5859936510803419e-07, "loss": 2.7776, "step": 368480 }, { "epoch": 0.0038144, "grad_norm": 0.9353331327438354, "learning_rate": 1.5845654611908457e-07, "loss": 2.7787, "step": 368490 }, { "epoch": 0.00384, "grad_norm": 0.8965495228767395, "learning_rate": 1.5831379094992993e-07, "loss": 2.6706, "step": 368500 }, { "epoch": 0.0038656, "grad_norm": 0.9851146340370178, "learning_rate": 1.5817109960149401e-07, "loss": 2.7837, "step": 368510 }, { "epoch": 0.0038912, "grad_norm": 1.0037875175476074, "learning_rate": 1.5802847207470384e-07, "loss": 2.8102, "step": 368520 }, { "epoch": 0.0039168, "grad_norm": 0.9823287725448608, "learning_rate": 1.5788590837048202e-07, "loss": 2.6295, "step": 368530 }, { "epoch": 0.0039424, "grad_norm": 0.9104170799255371, "learning_rate": 1.5774340848975554e-07, "loss": 2.5635, "step": 368540 }, { "epoch": 0.003968, "grad_norm": 0.7972959876060486, "learning_rate": 1.5760097243344708e-07, "loss": 2.7, "step": 368550 }, { "epoch": 0.0039936, "grad_norm": 0.851955771446228, "learning_rate": 1.5745860020248027e-07, "loss": 2.7679, "step": 368560 }, { "epoch": 0.0040192, "grad_norm": 0.8522566556930542, "learning_rate": 1.5731629179777886e-07, "loss": 2.6837, "step": 368570 }, { "epoch": 0.0040448, "grad_norm": 1.2363641262054443, "learning_rate": 1.5717404722026543e-07, "loss": 2.864, "step": 368580 }, { "epoch": 0.0040704, "grad_norm": 0.8924828171730042, "learning_rate": 1.5703186647086145e-07, "loss": 2.7962, "step": 368590 }, { "epoch": 0.004096, "grad_norm": 0.8435060977935791, "learning_rate": 1.5688974955049064e-07, "loss": 2.7267, "step": 368600 }, { "epoch": 0.0041216, "grad_norm": 0.8508713245391846, "learning_rate": 1.5674769646007336e-07, "loss": 2.7773, "step": 368610 }, { "epoch": 0.0041472, "grad_norm": 0.864500105381012, "learning_rate": 1.5660570720053226e-07, "loss": 2.8937, "step": 368620 }, { "epoch": 0.0041728, "grad_norm": 0.8318835496902466, "learning_rate": 1.5646378177278653e-07, "loss": 2.8955, "step": 368630 }, { "epoch": 0.0041984, "grad_norm": 0.8566522598266602, "learning_rate": 1.563219201777566e-07, "loss": 2.7781, "step": 368640 }, { "epoch": 0.004224, "grad_norm": 0.9450586438179016, "learning_rate": 1.5618012241636284e-07, "loss": 2.7765, "step": 368650 }, { "epoch": 0.0042496, "grad_norm": 0.9859218001365662, "learning_rate": 1.5603838848952558e-07, "loss": 2.8685, "step": 368660 }, { "epoch": 0.0042752, "grad_norm": 0.7881785035133362, "learning_rate": 1.55896718398163e-07, "loss": 2.7539, "step": 368670 }, { "epoch": 0.0043008, "grad_norm": 1.085877537727356, "learning_rate": 1.5575511214319438e-07, "loss": 2.8276, "step": 368680 }, { "epoch": 0.0043264, "grad_norm": 0.7976226210594177, "learning_rate": 1.5561356972553677e-07, "loss": 2.9089, "step": 368690 }, { "epoch": 0.004352, "grad_norm": 0.9146000742912292, "learning_rate": 1.554720911461094e-07, "loss": 2.8699, "step": 368700 }, { "epoch": 0.0043776, "grad_norm": 0.8716070652008057, "learning_rate": 1.5533067640582822e-07, "loss": 2.6796, "step": 368710 }, { "epoch": 0.0044032, "grad_norm": 0.8540029525756836, "learning_rate": 1.551893255056114e-07, "loss": 2.6477, "step": 368720 }, { "epoch": 0.0044288, "grad_norm": 0.8196583986282349, "learning_rate": 1.5504803844637594e-07, "loss": 2.9947, "step": 368730 }, { "epoch": 0.0044544, "grad_norm": 0.8253549337387085, "learning_rate": 1.5490681522903783e-07, "loss": 2.8835, "step": 368740 }, { "epoch": 0.00448, "grad_norm": 0.9855602979660034, "learning_rate": 1.5476565585451187e-07, "loss": 2.8416, "step": 368750 }, { "epoch": 0.0045056, "grad_norm": 0.7760425209999084, "learning_rate": 1.546245603237151e-07, "loss": 2.8532, "step": 368760 }, { "epoch": 0.0045312, "grad_norm": 0.9084932208061218, "learning_rate": 1.5448352863756012e-07, "loss": 2.8461, "step": 368770 }, { "epoch": 0.0045568, "grad_norm": 0.8523931503295898, "learning_rate": 1.5434256079696286e-07, "loss": 2.5416, "step": 368780 }, { "epoch": 0.0045824, "grad_norm": 0.8202101588249207, "learning_rate": 1.5420165680283706e-07, "loss": 2.8658, "step": 368790 }, { "epoch": 0.004608, "grad_norm": 0.8308282494544983, "learning_rate": 1.5406081665609752e-07, "loss": 2.7337, "step": 368800 }, { "epoch": 0.0046336, "grad_norm": 1.1675742864608765, "learning_rate": 1.5392004035765795e-07, "loss": 2.8305, "step": 368810 }, { "epoch": 0.0046592, "grad_norm": 0.8006896376609802, "learning_rate": 1.5377932790842877e-07, "loss": 2.7869, "step": 368820 }, { "epoch": 0.0046848, "grad_norm": 1.1101845502853394, "learning_rate": 1.5363867930932475e-07, "loss": 2.9753, "step": 368830 }, { "epoch": 0.0047104, "grad_norm": 0.8300601840019226, "learning_rate": 1.534980945612563e-07, "loss": 2.8366, "step": 368840 }, { "epoch": 0.004736, "grad_norm": 0.9397443532943726, "learning_rate": 1.5335757366513604e-07, "loss": 2.9678, "step": 368850 }, { "epoch": 0.0047616, "grad_norm": 0.8179602026939392, "learning_rate": 1.5321711662187432e-07, "loss": 2.6753, "step": 368860 }, { "epoch": 0.0047872, "grad_norm": 0.7741715908050537, "learning_rate": 1.5307672343238379e-07, "loss": 2.9804, "step": 368870 }, { "epoch": 0.0048128, "grad_norm": 0.8237026929855347, "learning_rate": 1.5293639409757366e-07, "loss": 2.8479, "step": 368880 }, { "epoch": 0.0048384, "grad_norm": 0.8084895014762878, "learning_rate": 1.5279612861835324e-07, "loss": 2.7327, "step": 368890 }, { "epoch": 0.004864, "grad_norm": 1.3350805044174194, "learning_rate": 1.5265592699563292e-07, "loss": 2.7716, "step": 368900 }, { "epoch": 0.0048896, "grad_norm": 0.8315574526786804, "learning_rate": 1.5251578923032196e-07, "loss": 2.8702, "step": 368910 }, { "epoch": 0.0049152, "grad_norm": 0.7888122797012329, "learning_rate": 1.5237571532332961e-07, "loss": 2.769, "step": 368920 }, { "epoch": 0.0049408, "grad_norm": 0.7898059487342834, "learning_rate": 1.5223570527556298e-07, "loss": 2.8193, "step": 368930 }, { "epoch": 0.0049664, "grad_norm": 0.7867558002471924, "learning_rate": 1.5209575908793016e-07, "loss": 2.7108, "step": 368940 }, { "epoch": 0.004992, "grad_norm": 0.9776653051376343, "learning_rate": 1.519558767613405e-07, "loss": 2.778, "step": 368950 }, { "epoch": 0.0050176, "grad_norm": 0.8142401576042175, "learning_rate": 1.5181605829669765e-07, "loss": 2.9126, "step": 368960 }, { "epoch": 0.0050432, "grad_norm": 0.9272149205207825, "learning_rate": 1.5167630369491094e-07, "loss": 2.9686, "step": 368970 }, { "epoch": 0.0050688, "grad_norm": 0.9218355417251587, "learning_rate": 1.5153661295688628e-07, "loss": 2.7673, "step": 368980 }, { "epoch": 0.0050944, "grad_norm": 0.8033239245414734, "learning_rate": 1.513969860835296e-07, "loss": 2.6726, "step": 368990 }, { "epoch": 0.00512, "grad_norm": 0.8869098424911499, "learning_rate": 1.5125742307574686e-07, "loss": 2.8849, "step": 369000 }, { "epoch": 0.0051456, "grad_norm": 0.8004458546638489, "learning_rate": 1.511179239344407e-07, "loss": 2.6125, "step": 369010 }, { "epoch": 0.0051712, "grad_norm": 0.8800248503684998, "learning_rate": 1.509784886605181e-07, "loss": 2.5535, "step": 369020 }, { "epoch": 0.0051968, "grad_norm": 1.1796363592147827, "learning_rate": 1.5083911725488176e-07, "loss": 2.8677, "step": 369030 }, { "epoch": 0.0052224, "grad_norm": 0.8753762245178223, "learning_rate": 1.5069980971843645e-07, "loss": 2.6804, "step": 369040 }, { "epoch": 0.005248, "grad_norm": 0.7940032482147217, "learning_rate": 1.5056056605208592e-07, "loss": 2.6726, "step": 369050 }, { "epoch": 0.0052736, "grad_norm": 1.0395071506500244, "learning_rate": 1.5042138625673163e-07, "loss": 2.6555, "step": 369060 }, { "epoch": 0.0052992, "grad_norm": 0.8595229983329773, "learning_rate": 1.5028227033327625e-07, "loss": 2.7785, "step": 369070 }, { "epoch": 0.0053248, "grad_norm": 0.9064670205116272, "learning_rate": 1.5014321828262458e-07, "loss": 2.802, "step": 369080 }, { "epoch": 0.0053504, "grad_norm": 0.8648979067802429, "learning_rate": 1.500042301056759e-07, "loss": 2.8851, "step": 369090 }, { "epoch": 0.005376, "grad_norm": 0.8659330606460571, "learning_rate": 1.498653058033306e-07, "loss": 2.8154, "step": 369100 }, { "epoch": 2.56e-05, "grad_norm": 1.123917818069458, "learning_rate": 1.4972644537649238e-07, "loss": 1.1352, "step": 369110 }, { "epoch": 5.12e-05, "grad_norm": 0.7908453941345215, "learning_rate": 1.4958764882605948e-07, "loss": 2.5765, "step": 369120 }, { "epoch": 7.68e-05, "grad_norm": 0.9795511364936829, "learning_rate": 1.494489161529322e-07, "loss": 2.6666, "step": 369130 }, { "epoch": 0.0001024, "grad_norm": 0.6925373673439026, "learning_rate": 1.4931024735801213e-07, "loss": 2.6101, "step": 369140 }, { "epoch": 0.000128, "grad_norm": 0.8292679190635681, "learning_rate": 1.4917164244219628e-07, "loss": 2.4119, "step": 369150 }, { "epoch": 0.0001536, "grad_norm": 0.7766066789627075, "learning_rate": 1.4903310140638394e-07, "loss": 2.9392, "step": 369160 }, { "epoch": 0.0001792, "grad_norm": 0.9008472561836243, "learning_rate": 1.488946242514744e-07, "loss": 2.6366, "step": 369170 }, { "epoch": 0.0002048, "grad_norm": 0.7961693406105042, "learning_rate": 1.4875621097836467e-07, "loss": 2.5247, "step": 369180 }, { "epoch": 0.0002304, "grad_norm": 0.8163701295852661, "learning_rate": 1.48617861587953e-07, "loss": 2.6622, "step": 369190 }, { "epoch": 0.000256, "grad_norm": 0.9118313789367676, "learning_rate": 1.484795760811375e-07, "loss": 2.5976, "step": 369200 }, { "epoch": 0.0002816, "grad_norm": 0.9120933413505554, "learning_rate": 1.4834135445881303e-07, "loss": 2.5309, "step": 369210 }, { "epoch": 0.0003072, "grad_norm": 0.7947728633880615, "learning_rate": 1.4820319672187556e-07, "loss": 2.3129, "step": 369220 }, { "epoch": 0.0003328, "grad_norm": 0.9782671928405762, "learning_rate": 1.4806510287122322e-07, "loss": 2.5437, "step": 369230 }, { "epoch": 0.0003584, "grad_norm": 0.849075973033905, "learning_rate": 1.4792707290774977e-07, "loss": 2.6353, "step": 369240 }, { "epoch": 0.000384, "grad_norm": 0.7970107197761536, "learning_rate": 1.4778910683235115e-07, "loss": 2.6437, "step": 369250 }, { "epoch": 0.0004096, "grad_norm": 0.7624837756156921, "learning_rate": 1.476512046459211e-07, "loss": 2.6072, "step": 369260 }, { "epoch": 0.0004352, "grad_norm": 0.8644902110099792, "learning_rate": 1.4751336634935555e-07, "loss": 2.7945, "step": 369270 }, { "epoch": 0.0004608, "grad_norm": 0.8148291110992432, "learning_rate": 1.4737559194354712e-07, "loss": 2.6706, "step": 369280 }, { "epoch": 0.0004864, "grad_norm": 0.7987117767333984, "learning_rate": 1.4723788142938954e-07, "loss": 2.4147, "step": 369290 }, { "epoch": 0.000512, "grad_norm": 0.8457469344139099, "learning_rate": 1.4710023480777547e-07, "loss": 2.654, "step": 369300 }, { "epoch": 0.0005376, "grad_norm": 0.8750073313713074, "learning_rate": 1.4696265207959748e-07, "loss": 2.6462, "step": 369310 }, { "epoch": 0.0005632, "grad_norm": 0.8022705316543579, "learning_rate": 1.468251332457482e-07, "loss": 2.6066, "step": 369320 }, { "epoch": 0.0005888, "grad_norm": 0.9534989595413208, "learning_rate": 1.4668767830712027e-07, "loss": 2.3383, "step": 369330 }, { "epoch": 0.0006144, "grad_norm": 0.8175421953201294, "learning_rate": 1.4655028726460296e-07, "loss": 2.6411, "step": 369340 }, { "epoch": 0.00064, "grad_norm": 0.9770694971084595, "learning_rate": 1.464129601190878e-07, "loss": 2.8318, "step": 369350 }, { "epoch": 0.0006656, "grad_norm": 0.8422291874885559, "learning_rate": 1.4627569687146737e-07, "loss": 2.5842, "step": 369360 }, { "epoch": 0.0006912, "grad_norm": 0.9513553977012634, "learning_rate": 1.4613849752262876e-07, "loss": 2.7885, "step": 369370 }, { "epoch": 0.0007168, "grad_norm": 0.7307326197624207, "learning_rate": 1.4600136207346349e-07, "loss": 2.5339, "step": 369380 }, { "epoch": 0.0007424, "grad_norm": 0.8673896193504333, "learning_rate": 1.4586429052486085e-07, "loss": 2.8241, "step": 369390 }, { "epoch": 0.000768, "grad_norm": 0.9469876885414124, "learning_rate": 1.457272828777101e-07, "loss": 2.6342, "step": 369400 }, { "epoch": 0.0007936, "grad_norm": 0.7802558541297913, "learning_rate": 1.4559033913289834e-07, "loss": 2.6273, "step": 369410 }, { "epoch": 0.0008192, "grad_norm": 0.7797764539718628, "learning_rate": 1.4545345929131482e-07, "loss": 2.0623, "step": 369420 }, { "epoch": 0.0008448, "grad_norm": 0.7996724247932434, "learning_rate": 1.4531664335384553e-07, "loss": 2.5439, "step": 369430 }, { "epoch": 0.0008704, "grad_norm": 0.8326260447502136, "learning_rate": 1.4517989132137866e-07, "loss": 2.4312, "step": 369440 }, { "epoch": 0.000896, "grad_norm": 0.7406184077262878, "learning_rate": 1.4504320319480236e-07, "loss": 2.4307, "step": 369450 }, { "epoch": 0.0009216, "grad_norm": 0.862287700176239, "learning_rate": 1.4490657897500038e-07, "loss": 2.6109, "step": 369460 }, { "epoch": 0.0009472, "grad_norm": 0.8724889755249023, "learning_rate": 1.4477001866286088e-07, "loss": 2.6821, "step": 369470 }, { "epoch": 0.0009728, "grad_norm": 0.8411180377006531, "learning_rate": 1.4463352225926875e-07, "loss": 2.513, "step": 369480 }, { "epoch": 0.0009984, "grad_norm": 0.8360802531242371, "learning_rate": 1.444970897651077e-07, "loss": 2.6419, "step": 369490 }, { "epoch": 0.001024, "grad_norm": 0.7833195328712463, "learning_rate": 1.4436072118126476e-07, "loss": 2.579, "step": 369500 }, { "epoch": 0.0010496, "grad_norm": 0.7613917589187622, "learning_rate": 1.4422441650862373e-07, "loss": 2.7602, "step": 369510 }, { "epoch": 0.0010752, "grad_norm": 0.9523198008537292, "learning_rate": 1.4408817574806722e-07, "loss": 2.566, "step": 369520 }, { "epoch": 0.0011008, "grad_norm": 0.7979658842086792, "learning_rate": 1.4395199890048005e-07, "loss": 2.5107, "step": 369530 }, { "epoch": 0.0011264, "grad_norm": 0.8846632242202759, "learning_rate": 1.4381588596674377e-07, "loss": 2.7438, "step": 369540 }, { "epoch": 0.001152, "grad_norm": 0.8654708862304688, "learning_rate": 1.4367983694774325e-07, "loss": 2.6545, "step": 369550 }, { "epoch": 0.0011776, "grad_norm": 0.8606047630310059, "learning_rate": 1.435438518443588e-07, "loss": 2.6823, "step": 369560 }, { "epoch": 0.0012032, "grad_norm": 0.845087468624115, "learning_rate": 1.4340793065747316e-07, "loss": 2.6662, "step": 369570 }, { "epoch": 0.0012288, "grad_norm": 0.8654444813728333, "learning_rate": 1.4327207338796667e-07, "loss": 2.8592, "step": 369580 }, { "epoch": 0.0012544, "grad_norm": 0.8490439653396606, "learning_rate": 1.4313628003672198e-07, "loss": 2.7649, "step": 369590 }, { "epoch": 0.00128, "grad_norm": 0.800477147102356, "learning_rate": 1.4300055060461948e-07, "loss": 2.4489, "step": 369600 }, { "epoch": 0.0013056, "grad_norm": 0.8385822176933289, "learning_rate": 1.428648850925385e-07, "loss": 2.589, "step": 369610 }, { "epoch": 0.0013312, "grad_norm": 0.8877631425857544, "learning_rate": 1.4272928350135827e-07, "loss": 2.468, "step": 369620 }, { "epoch": 0.0013568, "grad_norm": 0.8727290630340576, "learning_rate": 1.425937458319593e-07, "loss": 2.399, "step": 369630 }, { "epoch": 0.0013824, "grad_norm": 0.8242512941360474, "learning_rate": 1.4245827208521967e-07, "loss": 2.5618, "step": 369640 }, { "epoch": 0.001408, "grad_norm": 0.8755210638046265, "learning_rate": 1.4232286226201874e-07, "loss": 2.7688, "step": 369650 }, { "epoch": 0.0014336, "grad_norm": 1.0516074895858765, "learning_rate": 1.4218751636323358e-07, "loss": 2.7931, "step": 369660 }, { "epoch": 0.0014592, "grad_norm": 0.8093065619468689, "learning_rate": 1.4205223438974348e-07, "loss": 2.4207, "step": 369670 }, { "epoch": 0.0014848, "grad_norm": 0.8545674085617065, "learning_rate": 1.4191701634242328e-07, "loss": 2.4171, "step": 369680 }, { "epoch": 0.0015104, "grad_norm": 0.7635222673416138, "learning_rate": 1.4178186222215228e-07, "loss": 2.403, "step": 369690 }, { "epoch": 0.001536, "grad_norm": 0.7993581891059875, "learning_rate": 1.4164677202980647e-07, "loss": 2.2684, "step": 369700 }, { "epoch": 0.0015616, "grad_norm": 0.8436033129692078, "learning_rate": 1.415117457662607e-07, "loss": 2.6133, "step": 369710 }, { "epoch": 0.0015872, "grad_norm": 0.8314237594604492, "learning_rate": 1.413767834323909e-07, "loss": 2.6545, "step": 369720 }, { "epoch": 0.0016128, "grad_norm": 0.7953745722770691, "learning_rate": 1.412418850290731e-07, "loss": 2.8424, "step": 369730 }, { "epoch": 0.0016384, "grad_norm": 0.8130884766578674, "learning_rate": 1.41107050557181e-07, "loss": 2.5486, "step": 369740 }, { "epoch": 0.001664, "grad_norm": 0.8565011024475098, "learning_rate": 1.4097228001758944e-07, "loss": 2.5947, "step": 369750 }, { "epoch": 0.0016896, "grad_norm": 0.8579436540603638, "learning_rate": 1.4083757341117332e-07, "loss": 2.7111, "step": 369760 }, { "epoch": 0.0017152, "grad_norm": 0.9537191987037659, "learning_rate": 1.4070293073880416e-07, "loss": 2.7116, "step": 369770 }, { "epoch": 0.0017408, "grad_norm": 0.9693049788475037, "learning_rate": 1.4056835200135676e-07, "loss": 2.5146, "step": 369780 }, { "epoch": 0.0017664, "grad_norm": 0.975088894367218, "learning_rate": 1.4043383719970384e-07, "loss": 2.8717, "step": 369790 }, { "epoch": 0.001792, "grad_norm": 0.8830755352973938, "learning_rate": 1.4029938633471685e-07, "loss": 2.717, "step": 369800 }, { "epoch": 2.56e-05, "grad_norm": 0.8395363688468933, "learning_rate": 1.4016499940726735e-07, "loss": 1.7301, "step": 369810 }, { "epoch": 5.12e-05, "grad_norm": 0.8199413418769836, "learning_rate": 1.4003067641822798e-07, "loss": 2.6221, "step": 369820 }, { "epoch": 7.68e-05, "grad_norm": 1.151381254196167, "learning_rate": 1.3989641736846915e-07, "loss": 2.7409, "step": 369830 }, { "epoch": 0.0001024, "grad_norm": 0.802920937538147, "learning_rate": 1.3976222225886017e-07, "loss": 2.3567, "step": 369840 }, { "epoch": 0.000128, "grad_norm": 0.9273270964622498, "learning_rate": 1.3962809109027365e-07, "loss": 2.6106, "step": 369850 }, { "epoch": 0.0001536, "grad_norm": 0.8019103407859802, "learning_rate": 1.394940238635789e-07, "loss": 2.7459, "step": 369860 }, { "epoch": 0.0001792, "grad_norm": 0.9270559549331665, "learning_rate": 1.3936002057964415e-07, "loss": 2.6561, "step": 369870 }, { "epoch": 0.0002048, "grad_norm": 1.186734676361084, "learning_rate": 1.3922608123933978e-07, "loss": 2.6515, "step": 369880 }, { "epoch": 0.0002304, "grad_norm": 0.9193629622459412, "learning_rate": 1.390922058435329e-07, "loss": 2.5455, "step": 369890 }, { "epoch": 0.000256, "grad_norm": 0.9352750182151794, "learning_rate": 1.389583943930928e-07, "loss": 2.7683, "step": 369900 }, { "epoch": 0.0002816, "grad_norm": 0.8266250491142273, "learning_rate": 1.3882464688888653e-07, "loss": 2.8249, "step": 369910 }, { "epoch": 0.0003072, "grad_norm": 1.1182293891906738, "learning_rate": 1.3869096333178122e-07, "loss": 2.1555, "step": 369920 }, { "epoch": 0.0003328, "grad_norm": 0.8474999070167542, "learning_rate": 1.3855734372264506e-07, "loss": 2.6543, "step": 369930 }, { "epoch": 0.0003584, "grad_norm": 0.8230413198471069, "learning_rate": 1.3842378806234292e-07, "loss": 2.8437, "step": 369940 }, { "epoch": 0.000384, "grad_norm": 0.8655993938446045, "learning_rate": 1.3829029635174186e-07, "loss": 2.6614, "step": 369950 }, { "epoch": 0.0004096, "grad_norm": 0.811069905757904, "learning_rate": 1.3815686859170784e-07, "loss": 2.4076, "step": 369960 }, { "epoch": 0.0004352, "grad_norm": 0.7934468388557434, "learning_rate": 1.3802350478310466e-07, "loss": 2.6838, "step": 369970 }, { "epoch": 0.0004608, "grad_norm": 0.8838524222373962, "learning_rate": 1.3789020492679828e-07, "loss": 2.7818, "step": 369980 }, { "epoch": 0.0004864, "grad_norm": 0.8982244729995728, "learning_rate": 1.3775696902365355e-07, "loss": 2.6534, "step": 369990 }, { "epoch": 0.000512, "grad_norm": 0.8618359565734863, "learning_rate": 1.376237970745331e-07, "loss": 2.475, "step": 370000 }, { "epoch": 0.0005376, "grad_norm": 0.804407000541687, "learning_rate": 1.3749068908030072e-07, "loss": 2.4634, "step": 370010 }, { "epoch": 0.0005632, "grad_norm": 0.9091906547546387, "learning_rate": 1.3735764504182126e-07, "loss": 2.6356, "step": 370020 }, { "epoch": 0.0005888, "grad_norm": 0.831807553768158, "learning_rate": 1.3722466495995624e-07, "loss": 2.5052, "step": 370030 }, { "epoch": 0.0006144, "grad_norm": 0.9602652788162231, "learning_rate": 1.370917488355672e-07, "loss": 2.7462, "step": 370040 }, { "epoch": 0.00064, "grad_norm": 0.805331289768219, "learning_rate": 1.3695889666951678e-07, "loss": 2.6447, "step": 370050 }, { "epoch": 0.0006656, "grad_norm": 0.8516277074813843, "learning_rate": 1.3682610846266875e-07, "loss": 2.8562, "step": 370060 }, { "epoch": 0.0006912, "grad_norm": 0.8314051032066345, "learning_rate": 1.366933842158802e-07, "loss": 2.7587, "step": 370070 }, { "epoch": 0.0007168, "grad_norm": 0.8810084462165833, "learning_rate": 1.3656072393001375e-07, "loss": 2.6023, "step": 370080 }, { "epoch": 0.0007424, "grad_norm": 1.2665776014328003, "learning_rate": 1.3642812760592983e-07, "loss": 2.9085, "step": 370090 }, { "epoch": 0.000768, "grad_norm": 0.8189488053321838, "learning_rate": 1.3629559524448775e-07, "loss": 2.6488, "step": 370100 }, { "epoch": 0.0007936, "grad_norm": 0.7729530930519104, "learning_rate": 1.3616312684654797e-07, "loss": 2.6464, "step": 370110 }, { "epoch": 0.0008192, "grad_norm": 0.7723572254180908, "learning_rate": 1.3603072241296867e-07, "loss": 2.6343, "step": 370120 }, { "epoch": 0.0008448, "grad_norm": 0.8336641192436218, "learning_rate": 1.3589838194460913e-07, "loss": 2.7137, "step": 370130 }, { "epoch": 0.0008704, "grad_norm": 0.9231253862380981, "learning_rate": 1.357661054423265e-07, "loss": 2.4889, "step": 370140 }, { "epoch": 0.000896, "grad_norm": 0.8444979786872864, "learning_rate": 1.3563389290697894e-07, "loss": 2.4995, "step": 370150 }, { "epoch": 0.0009216, "grad_norm": 1.0384914875030518, "learning_rate": 1.3550174433942353e-07, "loss": 2.785, "step": 370160 }, { "epoch": 0.0009472, "grad_norm": 0.8491869568824768, "learning_rate": 1.353696597405185e-07, "loss": 2.6505, "step": 370170 }, { "epoch": 0.0009728, "grad_norm": 0.8510572910308838, "learning_rate": 1.3523763911111876e-07, "loss": 2.6945, "step": 370180 }, { "epoch": 0.0009984, "grad_norm": 0.8612778186798096, "learning_rate": 1.3510568245208134e-07, "loss": 2.7539, "step": 370190 }, { "epoch": 0.001024, "grad_norm": 0.8142139911651611, "learning_rate": 1.3497378976426224e-07, "loss": 2.5259, "step": 370200 }, { "epoch": 0.0010496, "grad_norm": 0.7645981311798096, "learning_rate": 1.3484196104851633e-07, "loss": 2.5585, "step": 370210 }, { "epoch": 0.0010752, "grad_norm": 0.8697900772094727, "learning_rate": 1.3471019630569738e-07, "loss": 2.5342, "step": 370220 }, { "epoch": 0.0011008, "grad_norm": 0.8827338814735413, "learning_rate": 1.3457849553666136e-07, "loss": 2.7095, "step": 370230 }, { "epoch": 0.0011264, "grad_norm": 0.894176721572876, "learning_rate": 1.3444685874226315e-07, "loss": 2.5145, "step": 370240 }, { "epoch": 0.001152, "grad_norm": 0.9216595888137817, "learning_rate": 1.3431528592335318e-07, "loss": 2.7842, "step": 370250 }, { "epoch": 0.0011776, "grad_norm": 0.7816994190216064, "learning_rate": 1.3418377708078856e-07, "loss": 2.7214, "step": 370260 }, { "epoch": 0.0012032, "grad_norm": 0.7519239187240601, "learning_rate": 1.340523322154197e-07, "loss": 2.6128, "step": 370270 }, { "epoch": 0.0012288, "grad_norm": 0.7888473272323608, "learning_rate": 1.339209513280981e-07, "loss": 2.8216, "step": 370280 }, { "epoch": 0.0012544, "grad_norm": 0.881348729133606, "learning_rate": 1.3378963441967873e-07, "loss": 2.6493, "step": 370290 }, { "epoch": 0.00128, "grad_norm": 0.8190761804580688, "learning_rate": 1.3365838149100972e-07, "loss": 2.6191, "step": 370300 }, { "epoch": 0.0013056, "grad_norm": 0.9668306708335876, "learning_rate": 1.3352719254294487e-07, "loss": 2.523, "step": 370310 }, { "epoch": 0.0013312, "grad_norm": 0.8864583969116211, "learning_rate": 1.333960675763335e-07, "loss": 2.4937, "step": 370320 }, { "epoch": 0.0013568, "grad_norm": 0.8188528418540955, "learning_rate": 1.3326500659202824e-07, "loss": 2.8545, "step": 370330 }, { "epoch": 0.0013824, "grad_norm": 0.8055851459503174, "learning_rate": 1.3313400959087508e-07, "loss": 2.7661, "step": 370340 }, { "epoch": 0.001408, "grad_norm": 0.9467320442199707, "learning_rate": 1.3300307657372668e-07, "loss": 2.6839, "step": 370350 }, { "epoch": 0.0014336, "grad_norm": 0.9097937345504761, "learning_rate": 1.3287220754142905e-07, "loss": 2.6224, "step": 370360 }, { "epoch": 0.0014592, "grad_norm": 0.7585725784301758, "learning_rate": 1.327414024948348e-07, "loss": 2.3646, "step": 370370 }, { "epoch": 0.0014848, "grad_norm": 1.0469545125961304, "learning_rate": 1.3261066143478885e-07, "loss": 2.2675, "step": 370380 }, { "epoch": 0.0015104, "grad_norm": 1.0598009824752808, "learning_rate": 1.3247998436214272e-07, "loss": 2.7618, "step": 370390 }, { "epoch": 0.001536, "grad_norm": 0.8466328978538513, "learning_rate": 1.3234937127773907e-07, "loss": 2.6345, "step": 370400 }, { "epoch": 2.56e-05, "grad_norm": 0.891940176486969, "learning_rate": 1.322188221824283e-07, "loss": 1.9167, "step": 370410 }, { "epoch": 5.12e-05, "grad_norm": 0.8382031917572021, "learning_rate": 1.3208833707705537e-07, "loss": 2.65, "step": 370420 }, { "epoch": 7.68e-05, "grad_norm": 0.8497466444969177, "learning_rate": 1.3195791596246732e-07, "loss": 2.5227, "step": 370430 }, { "epoch": 0.0001024, "grad_norm": 0.7949230074882507, "learning_rate": 1.3182755883950904e-07, "loss": 2.6019, "step": 370440 }, { "epoch": 0.000128, "grad_norm": 0.9379404783248901, "learning_rate": 1.3169726570902653e-07, "loss": 3.0077, "step": 370450 }, { "epoch": 0.0001536, "grad_norm": 0.8889119625091553, "learning_rate": 1.3156703657186575e-07, "loss": 2.6637, "step": 370460 }, { "epoch": 0.0001792, "grad_norm": 0.960600733757019, "learning_rate": 1.314368714288694e-07, "loss": 2.7455, "step": 370470 }, { "epoch": 0.0002048, "grad_norm": 0.8291125893592834, "learning_rate": 1.3130677028088125e-07, "loss": 2.6119, "step": 370480 }, { "epoch": 0.0002304, "grad_norm": 0.787128746509552, "learning_rate": 1.3117673312874723e-07, "loss": 2.5982, "step": 370490 }, { "epoch": 0.000256, "grad_norm": 1.064042091369629, "learning_rate": 1.3104675997330785e-07, "loss": 2.5583, "step": 370500 }, { "epoch": 0.0002816, "grad_norm": 0.8987919688224792, "learning_rate": 1.3091685081540796e-07, "loss": 2.6977, "step": 370510 }, { "epoch": 0.0003072, "grad_norm": 0.8406310677528381, "learning_rate": 1.307870056558902e-07, "loss": 2.4619, "step": 370520 }, { "epoch": 0.0003328, "grad_norm": 0.8341906070709229, "learning_rate": 1.3065722449559503e-07, "loss": 2.2033, "step": 370530 }, { "epoch": 0.0003584, "grad_norm": 0.7927395105361938, "learning_rate": 1.3052750733536512e-07, "loss": 2.8797, "step": 370540 }, { "epoch": 0.000384, "grad_norm": 0.8258081078529358, "learning_rate": 1.303978541760409e-07, "loss": 2.5708, "step": 370550 }, { "epoch": 0.0004096, "grad_norm": 0.8031421303749084, "learning_rate": 1.302682650184639e-07, "loss": 2.5025, "step": 370560 }, { "epoch": 0.0004352, "grad_norm": 0.8841541409492493, "learning_rate": 1.3013873986347346e-07, "loss": 2.8578, "step": 370570 }, { "epoch": 0.0004608, "grad_norm": 0.8473473787307739, "learning_rate": 1.3000927871191117e-07, "loss": 2.5667, "step": 370580 }, { "epoch": 0.0004864, "grad_norm": 0.9273120760917664, "learning_rate": 1.298798815646163e-07, "loss": 2.6902, "step": 370590 }, { "epoch": 0.000512, "grad_norm": 0.8396347165107727, "learning_rate": 1.2975054842242595e-07, "loss": 2.5427, "step": 370600 }, { "epoch": 0.0005376, "grad_norm": 0.8960116505622864, "learning_rate": 1.296212792861795e-07, "loss": 2.6593, "step": 370610 }, { "epoch": 0.0005632, "grad_norm": 0.9106935262680054, "learning_rate": 1.2949207415671738e-07, "loss": 2.3268, "step": 370620 }, { "epoch": 0.0005888, "grad_norm": 0.8276695609092712, "learning_rate": 1.2936293303487445e-07, "loss": 2.6378, "step": 370630 }, { "epoch": 0.0006144, "grad_norm": 0.9951728582382202, "learning_rate": 1.2923385592149006e-07, "loss": 2.5824, "step": 370640 }, { "epoch": 0.00064, "grad_norm": 0.9171199202537537, "learning_rate": 1.2910484281740133e-07, "loss": 2.5721, "step": 370650 }, { "epoch": 0.0006656, "grad_norm": 0.9165037870407104, "learning_rate": 1.289758937234442e-07, "loss": 2.8767, "step": 370660 }, { "epoch": 0.0006912, "grad_norm": 0.8539184331893921, "learning_rate": 1.2884700864045473e-07, "loss": 2.6747, "step": 370670 }, { "epoch": 0.0007168, "grad_norm": 0.8719145059585571, "learning_rate": 1.287181875692689e-07, "loss": 2.7129, "step": 370680 }, { "epoch": 0.0007424, "grad_norm": 0.9511085748672485, "learning_rate": 1.285894305107227e-07, "loss": 2.7624, "step": 370690 }, { "epoch": 0.000768, "grad_norm": 0.8632257580757141, "learning_rate": 1.284607374656499e-07, "loss": 2.5932, "step": 370700 }, { "epoch": 0.0007936, "grad_norm": 0.8867873549461365, "learning_rate": 1.2833210843488652e-07, "loss": 2.6452, "step": 370710 }, { "epoch": 0.0008192, "grad_norm": 0.9528893828392029, "learning_rate": 1.2820354341926411e-07, "loss": 2.4522, "step": 370720 }, { "epoch": 0.0008448, "grad_norm": 0.9133307337760925, "learning_rate": 1.2807504241961976e-07, "loss": 2.6028, "step": 370730 }, { "epoch": 0.0008704, "grad_norm": 0.7679728865623474, "learning_rate": 1.2794660543678395e-07, "loss": 2.5472, "step": 370740 }, { "epoch": 0.000896, "grad_norm": 0.8314725160598755, "learning_rate": 1.2781823247159153e-07, "loss": 2.4654, "step": 370750 }, { "epoch": 0.0009216, "grad_norm": 0.8121504783630371, "learning_rate": 1.276899235248741e-07, "loss": 2.6198, "step": 370760 }, { "epoch": 0.0009472, "grad_norm": 0.9178550839424133, "learning_rate": 1.2756167859746314e-07, "loss": 2.5938, "step": 370770 }, { "epoch": 0.0009728, "grad_norm": 0.858540415763855, "learning_rate": 1.2743349769019143e-07, "loss": 2.5074, "step": 370780 }, { "epoch": 0.0009984, "grad_norm": 0.818856418132782, "learning_rate": 1.2730538080389043e-07, "loss": 2.5197, "step": 370790 }, { "epoch": 0.001024, "grad_norm": 0.8615411520004272, "learning_rate": 1.2717732793938953e-07, "loss": 2.7631, "step": 370800 }, { "epoch": 0.0010496, "grad_norm": 0.8843905329704285, "learning_rate": 1.2704933909752025e-07, "loss": 2.6182, "step": 370810 }, { "epoch": 0.0010752, "grad_norm": 0.8123325705528259, "learning_rate": 1.2692141427911088e-07, "loss": 2.5615, "step": 370820 }, { "epoch": 0.0011008, "grad_norm": 0.9724084734916687, "learning_rate": 1.267935534849929e-07, "loss": 2.6236, "step": 370830 }, { "epoch": 0.0011264, "grad_norm": 0.8491687774658203, "learning_rate": 1.2666575671599456e-07, "loss": 2.6311, "step": 370840 }, { "epoch": 0.001152, "grad_norm": 0.8509774208068848, "learning_rate": 1.2653802397294524e-07, "loss": 2.8288, "step": 370850 }, { "epoch": 0.0011776, "grad_norm": 0.8731145262718201, "learning_rate": 1.264103552566731e-07, "loss": 2.9453, "step": 370860 }, { "epoch": 0.0012032, "grad_norm": 0.8529740571975708, "learning_rate": 1.2628275056800421e-07, "loss": 2.7569, "step": 370870 }, { "epoch": 0.0012288, "grad_norm": 0.802510678768158, "learning_rate": 1.2615520990776785e-07, "loss": 2.7374, "step": 370880 }, { "epoch": 0.0012544, "grad_norm": 0.8584945797920227, "learning_rate": 1.2602773327679118e-07, "loss": 2.6538, "step": 370890 }, { "epoch": 0.00128, "grad_norm": 0.9381782412528992, "learning_rate": 1.259003206759013e-07, "loss": 2.6764, "step": 370900 }, { "epoch": 2.56e-05, "grad_norm": 0.8426613807678223, "learning_rate": 1.2577297210592309e-07, "loss": 1.7407, "step": 370910 }, { "epoch": 5.12e-05, "grad_norm": 0.855082094669342, "learning_rate": 1.2564568756768258e-07, "loss": 2.7887, "step": 370920 }, { "epoch": 7.68e-05, "grad_norm": 0.9256843328475952, "learning_rate": 1.2551846706200578e-07, "loss": 2.662, "step": 370930 }, { "epoch": 0.0001024, "grad_norm": 0.9019607901573181, "learning_rate": 1.2539131058971755e-07, "loss": 2.4699, "step": 370940 }, { "epoch": 0.000128, "grad_norm": 0.9295216798782349, "learning_rate": 1.252642181516417e-07, "loss": 2.5909, "step": 370950 }, { "epoch": 0.0001536, "grad_norm": 0.8161939978599548, "learning_rate": 1.2513718974860313e-07, "loss": 2.6066, "step": 370960 }, { "epoch": 0.0001792, "grad_norm": 0.9686944484710693, "learning_rate": 1.250102253814256e-07, "loss": 2.6821, "step": 370970 }, { "epoch": 0.0002048, "grad_norm": 0.8171600699424744, "learning_rate": 1.2488332505093292e-07, "loss": 2.6542, "step": 370980 }, { "epoch": 0.0002304, "grad_norm": 0.8150277137756348, "learning_rate": 1.2475648875794777e-07, "loss": 2.7243, "step": 370990 }, { "epoch": 0.000256, "grad_norm": 0.7969090938568115, "learning_rate": 1.246297165032906e-07, "loss": 2.6015, "step": 371000 }, { "epoch": 0.0002816, "grad_norm": 0.8317875862121582, "learning_rate": 1.245030082877863e-07, "loss": 2.5209, "step": 371010 }, { "epoch": 0.0003072, "grad_norm": 0.8925228118896484, "learning_rate": 1.2437636411225418e-07, "loss": 2.6664, "step": 371020 }, { "epoch": 0.0003328, "grad_norm": 0.8700436353683472, "learning_rate": 1.2424978397751808e-07, "loss": 2.7366, "step": 371030 }, { "epoch": 0.0003584, "grad_norm": 0.9262571334838867, "learning_rate": 1.241232678843973e-07, "loss": 2.8038, "step": 371040 }, { "epoch": 0.000384, "grad_norm": 1.0406924486160278, "learning_rate": 1.2399681583371125e-07, "loss": 2.8384, "step": 371050 }, { "epoch": 0.0004096, "grad_norm": 0.7817019820213318, "learning_rate": 1.2387042782628257e-07, "loss": 2.4115, "step": 371060 }, { "epoch": 0.0004352, "grad_norm": 0.9771279692649841, "learning_rate": 1.2374410386292834e-07, "loss": 2.7842, "step": 371070 }, { "epoch": 0.0004608, "grad_norm": 0.9052402377128601, "learning_rate": 1.2361784394446908e-07, "loss": 2.5748, "step": 371080 }, { "epoch": 0.0004864, "grad_norm": 0.9582107663154602, "learning_rate": 1.234916480717241e-07, "loss": 2.7082, "step": 371090 }, { "epoch": 0.000512, "grad_norm": 0.7834042310714722, "learning_rate": 1.2336551624550941e-07, "loss": 2.5827, "step": 371100 }, { "epoch": 0.0005376, "grad_norm": 0.8246884346008301, "learning_rate": 1.232394484666466e-07, "loss": 2.5852, "step": 371110 }, { "epoch": 0.0005632, "grad_norm": 0.8782312273979187, "learning_rate": 1.2311344473594944e-07, "loss": 2.5834, "step": 371120 }, { "epoch": 0.0005888, "grad_norm": 0.8492186069488525, "learning_rate": 1.2298750505423617e-07, "loss": 2.5568, "step": 371130 }, { "epoch": 0.0006144, "grad_norm": 0.8792788982391357, "learning_rate": 1.2286162942232505e-07, "loss": 2.5913, "step": 371140 }, { "epoch": 0.00064, "grad_norm": 1.0363456010818481, "learning_rate": 1.2273581784103094e-07, "loss": 2.7602, "step": 371150 }, { "epoch": 0.0006656, "grad_norm": 0.835556149482727, "learning_rate": 1.226100703111699e-07, "loss": 2.812, "step": 371160 }, { "epoch": 0.0006912, "grad_norm": 0.8479049801826477, "learning_rate": 1.224843868335579e-07, "loss": 2.881, "step": 371170 }, { "epoch": 0.0007168, "grad_norm": 0.9331523776054382, "learning_rate": 1.2235876740900875e-07, "loss": 2.5419, "step": 371180 }, { "epoch": 0.0007424, "grad_norm": 0.9345728754997253, "learning_rate": 1.222332120383385e-07, "loss": 2.6762, "step": 371190 }, { "epoch": 0.000768, "grad_norm": 0.7727011442184448, "learning_rate": 1.2210772072236088e-07, "loss": 2.8, "step": 371200 }, { "epoch": 0.0007936, "grad_norm": 0.8050734400749207, "learning_rate": 1.2198229346188862e-07, "loss": 2.4575, "step": 371210 }, { "epoch": 0.0008192, "grad_norm": 0.9702664017677307, "learning_rate": 1.218569302577366e-07, "loss": 2.7501, "step": 371220 }, { "epoch": 0.0008448, "grad_norm": 1.0155134201049805, "learning_rate": 1.2173163111071529e-07, "loss": 2.5447, "step": 371230 }, { "epoch": 0.0008704, "grad_norm": 0.8752357959747314, "learning_rate": 1.2160639602164182e-07, "loss": 2.6097, "step": 371240 }, { "epoch": 0.000896, "grad_norm": 0.9251165390014648, "learning_rate": 1.214812249913233e-07, "loss": 2.6812, "step": 371250 }, { "epoch": 0.0009216, "grad_norm": 0.772919774055481, "learning_rate": 1.2135611802057466e-07, "loss": 2.4977, "step": 371260 }, { "epoch": 0.0009472, "grad_norm": 0.7281426191329956, "learning_rate": 1.212310751102064e-07, "loss": 2.5452, "step": 371270 }, { "epoch": 0.0009728, "grad_norm": 1.007517695426941, "learning_rate": 1.2110609626102888e-07, "loss": 2.651, "step": 371280 }, { "epoch": 0.0009984, "grad_norm": 0.8151404857635498, "learning_rate": 1.2098118147385264e-07, "loss": 2.5225, "step": 371290 }, { "epoch": 0.001024, "grad_norm": 0.7878261804580688, "learning_rate": 1.20856330749487e-07, "loss": 2.6617, "step": 371300 }, { "epoch": 0.0010496, "grad_norm": 0.9989016056060791, "learning_rate": 1.2073154408874466e-07, "loss": 2.6484, "step": 371310 }, { "epoch": 0.0010752, "grad_norm": 0.8712669610977173, "learning_rate": 1.2060682149243163e-07, "loss": 2.7746, "step": 371320 }, { "epoch": 0.0011008, "grad_norm": 0.816335141658783, "learning_rate": 1.204821629613573e-07, "loss": 2.8016, "step": 371330 }, { "epoch": 0.0011264, "grad_norm": 0.8023927211761475, "learning_rate": 1.2035756849633097e-07, "loss": 2.6197, "step": 371340 }, { "epoch": 0.001152, "grad_norm": 0.8197097182273865, "learning_rate": 1.2023303809815978e-07, "loss": 2.4931, "step": 371350 }, { "epoch": 0.0011776, "grad_norm": 0.8795168399810791, "learning_rate": 1.2010857176765088e-07, "loss": 2.7619, "step": 371360 }, { "epoch": 0.0012032, "grad_norm": 0.8371731638908386, "learning_rate": 1.1998416950561364e-07, "loss": 2.8807, "step": 371370 }, { "epoch": 0.0012288, "grad_norm": 0.8651583194732666, "learning_rate": 1.1985983131285295e-07, "loss": 2.6891, "step": 371380 }, { "epoch": 0.0012544, "grad_norm": 0.842668354511261, "learning_rate": 1.1973555719017482e-07, "loss": 3.0165, "step": 371390 }, { "epoch": 0.00128, "grad_norm": 0.8458667397499084, "learning_rate": 1.196113471383853e-07, "loss": 2.5435, "step": 371400 }, { "epoch": 0.0013056, "grad_norm": 0.8763586282730103, "learning_rate": 1.1948720115829148e-07, "loss": 2.8209, "step": 371410 }, { "epoch": 0.0013312, "grad_norm": 0.8582722544670105, "learning_rate": 1.193631192506961e-07, "loss": 2.6703, "step": 371420 }, { "epoch": 0.0013568, "grad_norm": 0.8562361001968384, "learning_rate": 1.1923910141640626e-07, "loss": 2.4167, "step": 371430 }, { "epoch": 0.0013824, "grad_norm": 0.8104995489120483, "learning_rate": 1.1911514765622468e-07, "loss": 2.6307, "step": 371440 }, { "epoch": 0.001408, "grad_norm": 0.8832956552505493, "learning_rate": 1.1899125797095513e-07, "loss": 2.8057, "step": 371450 }, { "epoch": 0.0014336, "grad_norm": 0.9454182386398315, "learning_rate": 1.188674323614003e-07, "loss": 2.7158, "step": 371460 }, { "epoch": 0.0014592, "grad_norm": 0.8288347721099854, "learning_rate": 1.1874367082836514e-07, "loss": 2.5722, "step": 371470 }, { "epoch": 0.0014848, "grad_norm": 0.8083382844924927, "learning_rate": 1.1861997337265007e-07, "loss": 2.7578, "step": 371480 }, { "epoch": 0.0015104, "grad_norm": 1.019010066986084, "learning_rate": 1.1849633999505895e-07, "loss": 2.0848, "step": 371490 }, { "epoch": 0.001536, "grad_norm": 0.8419094085693359, "learning_rate": 1.183727706963933e-07, "loss": 2.5496, "step": 371500 }, { "epoch": 0.0015616, "grad_norm": 0.7915528416633606, "learning_rate": 1.1824926547745364e-07, "loss": 2.5587, "step": 371510 }, { "epoch": 0.0015872, "grad_norm": 0.8979319334030151, "learning_rate": 1.1812582433904151e-07, "loss": 2.7202, "step": 371520 }, { "epoch": 0.0016128, "grad_norm": 0.8062240481376648, "learning_rate": 1.1800244728195742e-07, "loss": 2.7222, "step": 371530 }, { "epoch": 0.0016384, "grad_norm": 0.8329649567604065, "learning_rate": 1.178791343069996e-07, "loss": 2.5873, "step": 371540 }, { "epoch": 0.001664, "grad_norm": 1.015021562576294, "learning_rate": 1.1775588541497074e-07, "loss": 2.6637, "step": 371550 }, { "epoch": 0.0016896, "grad_norm": 0.835780143737793, "learning_rate": 1.1763270060666799e-07, "loss": 2.5606, "step": 371560 }, { "epoch": 0.0017152, "grad_norm": 0.8145661950111389, "learning_rate": 1.1750957988289069e-07, "loss": 2.5717, "step": 371570 }, { "epoch": 0.0017408, "grad_norm": 0.8693565726280212, "learning_rate": 1.173865232444371e-07, "loss": 2.7488, "step": 371580 }, { "epoch": 0.0017664, "grad_norm": 0.9374135136604309, "learning_rate": 1.172635306921055e-07, "loss": 2.8579, "step": 371590 }, { "epoch": 0.001792, "grad_norm": 1.0331906080245972, "learning_rate": 1.171406022266941e-07, "loss": 2.4481, "step": 371600 }, { "epoch": 0.0018176, "grad_norm": 0.8913015723228455, "learning_rate": 1.1701773784899895e-07, "loss": 2.7943, "step": 371610 }, { "epoch": 0.0018432, "grad_norm": 0.8166002631187439, "learning_rate": 1.1689493755981718e-07, "loss": 2.4345, "step": 371620 }, { "epoch": 0.0018688, "grad_norm": 0.931892454624176, "learning_rate": 1.1677220135994482e-07, "loss": 2.6148, "step": 371630 }, { "epoch": 0.0018944, "grad_norm": 0.8384263515472412, "learning_rate": 1.1664952925017903e-07, "loss": 2.6132, "step": 371640 }, { "epoch": 0.00192, "grad_norm": 0.8920643925666809, "learning_rate": 1.165269212313136e-07, "loss": 2.6603, "step": 371650 }, { "epoch": 0.0019456, "grad_norm": 0.9346044659614563, "learning_rate": 1.1640437730414455e-07, "loss": 2.7123, "step": 371660 }, { "epoch": 0.0019712, "grad_norm": 0.7360352873802185, "learning_rate": 1.1628189746946573e-07, "loss": 2.594, "step": 371670 }, { "epoch": 0.0019968, "grad_norm": 0.8570763468742371, "learning_rate": 1.1615948172807312e-07, "loss": 2.5579, "step": 371680 }, { "epoch": 0.0020224, "grad_norm": 0.8723891973495483, "learning_rate": 1.1603713008075835e-07, "loss": 2.6532, "step": 371690 }, { "epoch": 0.002048, "grad_norm": 0.9625536799430847, "learning_rate": 1.1591484252831631e-07, "loss": 2.5891, "step": 371700 }, { "epoch": 0.0020736, "grad_norm": 0.8714451193809509, "learning_rate": 1.1579261907154082e-07, "loss": 2.4661, "step": 371710 }, { "epoch": 0.0020992, "grad_norm": 0.7755446434020996, "learning_rate": 1.1567045971122237e-07, "loss": 2.6676, "step": 371720 }, { "epoch": 0.0021248, "grad_norm": 0.840926468372345, "learning_rate": 1.1554836444815365e-07, "loss": 2.6775, "step": 371730 }, { "epoch": 0.0021504, "grad_norm": 1.2686853408813477, "learning_rate": 1.1542633328312736e-07, "loss": 2.9406, "step": 371740 }, { "epoch": 0.002176, "grad_norm": 0.9050579071044922, "learning_rate": 1.1530436621693286e-07, "loss": 2.7256, "step": 371750 }, { "epoch": 0.0022016, "grad_norm": 0.91446453332901, "learning_rate": 1.1518246325036398e-07, "loss": 2.8823, "step": 371760 }, { "epoch": 0.0022272, "grad_norm": 0.8364784121513367, "learning_rate": 1.1506062438421007e-07, "loss": 2.844, "step": 371770 }, { "epoch": 0.0022528, "grad_norm": 0.9167898297309875, "learning_rate": 1.149388496192605e-07, "loss": 2.5665, "step": 371780 }, { "epoch": 0.0022784, "grad_norm": 1.0014398097991943, "learning_rate": 1.1481713895630464e-07, "loss": 2.8162, "step": 371790 }, { "epoch": 0.002304, "grad_norm": 0.8306073546409607, "learning_rate": 1.1469549239613298e-07, "loss": 2.3453, "step": 371800 }, { "epoch": 0.0023296, "grad_norm": 0.8322003483772278, "learning_rate": 1.1457390993953377e-07, "loss": 2.8278, "step": 371810 }, { "epoch": 0.0023552, "grad_norm": 0.8508733510971069, "learning_rate": 1.1445239158729527e-07, "loss": 2.7212, "step": 371820 }, { "epoch": 0.0023808, "grad_norm": 0.8982186317443848, "learning_rate": 1.1433093734020684e-07, "loss": 2.4062, "step": 371830 }, { "epoch": 0.0024064, "grad_norm": 0.8192375898361206, "learning_rate": 1.1420954719905453e-07, "loss": 2.7528, "step": 371840 }, { "epoch": 0.002432, "grad_norm": 0.9461553692817688, "learning_rate": 1.1408822116462548e-07, "loss": 2.8472, "step": 371850 }, { "epoch": 0.0024576, "grad_norm": 0.8891788721084595, "learning_rate": 1.1396695923770684e-07, "loss": 2.7998, "step": 371860 }, { "epoch": 0.0024832, "grad_norm": 1.0876436233520508, "learning_rate": 1.1384576141908577e-07, "loss": 2.7367, "step": 371870 }, { "epoch": 0.0025088, "grad_norm": 1.034458041191101, "learning_rate": 1.1372462770954717e-07, "loss": 2.4723, "step": 371880 }, { "epoch": 0.0025344, "grad_norm": 0.8116711974143982, "learning_rate": 1.1360355810987711e-07, "loss": 2.8135, "step": 371890 }, { "epoch": 0.00256, "grad_norm": 0.8764032125473022, "learning_rate": 1.134825526208605e-07, "loss": 2.6311, "step": 371900 }, { "epoch": 0.0025856, "grad_norm": 0.9132581949234009, "learning_rate": 1.1336161124328226e-07, "loss": 2.5791, "step": 371910 }, { "epoch": 0.0026112, "grad_norm": 0.9591127634048462, "learning_rate": 1.1324073397792512e-07, "loss": 2.6701, "step": 371920 }, { "epoch": 0.0026368, "grad_norm": 1.0041968822479248, "learning_rate": 1.131199208255751e-07, "loss": 2.81, "step": 371930 }, { "epoch": 0.0026624, "grad_norm": 0.9962717890739441, "learning_rate": 1.1299917178701492e-07, "loss": 2.7323, "step": 371940 }, { "epoch": 0.002688, "grad_norm": 0.8484994173049927, "learning_rate": 1.1287848686302616e-07, "loss": 2.8938, "step": 371950 }, { "epoch": 0.0027136, "grad_norm": 0.8240726590156555, "learning_rate": 1.1275786605439376e-07, "loss": 2.7981, "step": 371960 }, { "epoch": 0.0027392, "grad_norm": 0.9675368666648865, "learning_rate": 1.1263730936190043e-07, "loss": 2.7554, "step": 371970 }, { "epoch": 0.0027648, "grad_norm": 0.8943650126457214, "learning_rate": 1.1251681678632442e-07, "loss": 2.9015, "step": 371980 }, { "epoch": 0.0027904, "grad_norm": 0.841118335723877, "learning_rate": 1.1239638832844957e-07, "loss": 2.891, "step": 371990 }, { "epoch": 0.002816, "grad_norm": 1.7329000234603882, "learning_rate": 1.1227602398905635e-07, "loss": 2.8611, "step": 372000 }, { "epoch": 0.0028416, "grad_norm": 0.9152148365974426, "learning_rate": 1.1215572376892525e-07, "loss": 2.6923, "step": 372010 }, { "epoch": 0.0028672, "grad_norm": 0.8629988431930542, "learning_rate": 1.1203548766883677e-07, "loss": 2.8141, "step": 372020 }, { "epoch": 0.0028928, "grad_norm": 0.9607837796211243, "learning_rate": 1.1191531568956915e-07, "loss": 2.7115, "step": 372030 }, { "epoch": 0.0029184, "grad_norm": 0.8504763841629028, "learning_rate": 1.1179520783190512e-07, "loss": 2.7181, "step": 372040 }, { "epoch": 0.002944, "grad_norm": 0.9358628392219543, "learning_rate": 1.1167516409661849e-07, "loss": 2.7976, "step": 372050 }, { "epoch": 0.0029696, "grad_norm": 0.8494913578033447, "learning_rate": 1.1155518448449198e-07, "loss": 2.5956, "step": 372060 }, { "epoch": 0.0029952, "grad_norm": 0.867313802242279, "learning_rate": 1.1143526899630164e-07, "loss": 2.9755, "step": 372070 }, { "epoch": 0.0030208, "grad_norm": 0.85187828540802, "learning_rate": 1.1131541763282572e-07, "loss": 2.836, "step": 372080 }, { "epoch": 0.0030464, "grad_norm": 0.8867212533950806, "learning_rate": 1.1119563039484027e-07, "loss": 2.8975, "step": 372090 }, { "epoch": 0.003072, "grad_norm": 0.8209984302520752, "learning_rate": 1.1107590728312356e-07, "loss": 2.8274, "step": 372100 }, { "epoch": 0.0030976, "grad_norm": 0.8220681548118591, "learning_rate": 1.1095624829845164e-07, "loss": 2.8297, "step": 372110 }, { "epoch": 0.0031232, "grad_norm": 0.8536720275878906, "learning_rate": 1.1083665344160055e-07, "loss": 2.9265, "step": 372120 }, { "epoch": 0.0031488, "grad_norm": 0.9064351320266724, "learning_rate": 1.1071712271334522e-07, "loss": 2.8833, "step": 372130 }, { "epoch": 0.0031744, "grad_norm": 0.9299423694610596, "learning_rate": 1.1059765611446171e-07, "loss": 3.0545, "step": 372140 }, { "epoch": 0.0032, "grad_norm": 0.8781831860542297, "learning_rate": 1.1047825364572274e-07, "loss": 2.865, "step": 372150 }, { "epoch": 0.0032256, "grad_norm": 1.0503802299499512, "learning_rate": 1.1035891530790432e-07, "loss": 3.0677, "step": 372160 }, { "epoch": 0.0032512, "grad_norm": 0.9064871668815613, "learning_rate": 1.1023964110178142e-07, "loss": 2.982, "step": 372170 }, { "epoch": 0.0032768, "grad_norm": 0.8745357394218445, "learning_rate": 1.1012043102812563e-07, "loss": 2.9471, "step": 372180 }, { "epoch": 0.0033024, "grad_norm": 0.9281135201454163, "learning_rate": 1.1000128508770969e-07, "loss": 2.5266, "step": 372190 }, { "epoch": 0.003328, "grad_norm": 0.8503718376159668, "learning_rate": 1.0988220328130627e-07, "loss": 2.7807, "step": 372200 }, { "epoch": 0.0033536, "grad_norm": 0.9573344588279724, "learning_rate": 1.0976318560968924e-07, "loss": 2.934, "step": 372210 }, { "epoch": 0.0033792, "grad_norm": 0.9057306051254272, "learning_rate": 1.0964423207362906e-07, "loss": 2.757, "step": 372220 }, { "epoch": 0.0034048, "grad_norm": 0.8243812918663025, "learning_rate": 1.0952534267389847e-07, "loss": 2.7219, "step": 372230 }, { "epoch": 0.0034304, "grad_norm": 0.841198742389679, "learning_rate": 1.0940651741126685e-07, "loss": 2.9107, "step": 372240 }, { "epoch": 0.003456, "grad_norm": 0.8202880024909973, "learning_rate": 1.0928775628650579e-07, "loss": 2.9525, "step": 372250 }, { "epoch": 0.0034816, "grad_norm": 1.0166354179382324, "learning_rate": 1.0916905930038469e-07, "loss": 2.6275, "step": 372260 }, { "epoch": 0.0035072, "grad_norm": 0.8764519691467285, "learning_rate": 1.0905042645367181e-07, "loss": 2.5423, "step": 372270 }, { "epoch": 0.0035328, "grad_norm": 1.0789222717285156, "learning_rate": 1.0893185774713987e-07, "loss": 2.7597, "step": 372280 }, { "epoch": 0.0035584, "grad_norm": 0.9443726539611816, "learning_rate": 1.0881335318155716e-07, "loss": 2.6845, "step": 372290 }, { "epoch": 0.003584, "grad_norm": 0.8054100275039673, "learning_rate": 1.0869491275768973e-07, "loss": 2.9217, "step": 372300 }, { "epoch": 0.0036096, "grad_norm": 0.9452774524688721, "learning_rate": 1.0857653647630695e-07, "loss": 2.8564, "step": 372310 }, { "epoch": 0.0036352, "grad_norm": 0.8751360774040222, "learning_rate": 1.0845822433817709e-07, "loss": 2.9963, "step": 372320 }, { "epoch": 0.0036608, "grad_norm": 0.8620344996452332, "learning_rate": 1.0833997634406623e-07, "loss": 2.9447, "step": 372330 }, { "epoch": 0.0036864, "grad_norm": 0.9088513255119324, "learning_rate": 1.0822179249474152e-07, "loss": 2.9231, "step": 372340 }, { "epoch": 0.003712, "grad_norm": 0.9400078654289246, "learning_rate": 1.0810367279097011e-07, "loss": 2.688, "step": 372350 }, { "epoch": 0.0037376, "grad_norm": 0.8355779051780701, "learning_rate": 1.0798561723351698e-07, "loss": 2.6867, "step": 372360 }, { "epoch": 0.0037632, "grad_norm": 0.8538486361503601, "learning_rate": 1.0786762582314925e-07, "loss": 2.762, "step": 372370 }, { "epoch": 0.0037888, "grad_norm": 0.8835572004318237, "learning_rate": 1.0774969856063078e-07, "loss": 2.6825, "step": 372380 }, { "epoch": 0.0038144, "grad_norm": 0.8742626309394836, "learning_rate": 1.0763183544672539e-07, "loss": 2.6989, "step": 372390 }, { "epoch": 0.00384, "grad_norm": 0.9171218872070312, "learning_rate": 1.0751403648219916e-07, "loss": 3.1064, "step": 372400 }, { "epoch": 0.0038656, "grad_norm": 0.8917632102966309, "learning_rate": 1.0739630166781479e-07, "loss": 2.8078, "step": 372410 }, { "epoch": 0.0038912, "grad_norm": 1.0397812128067017, "learning_rate": 1.0727863100433611e-07, "loss": 2.7997, "step": 372420 }, { "epoch": 0.0039168, "grad_norm": 0.9387028813362122, "learning_rate": 1.0716102449252697e-07, "loss": 2.9687, "step": 372430 }, { "epoch": 0.0039424, "grad_norm": 0.7924585938453674, "learning_rate": 1.0704348213315007e-07, "loss": 2.7371, "step": 372440 }, { "epoch": 0.003968, "grad_norm": 0.92161625623703, "learning_rate": 1.0692600392696484e-07, "loss": 2.8811, "step": 372450 }, { "epoch": 0.0039936, "grad_norm": 0.9197085499763489, "learning_rate": 1.068085898747373e-07, "loss": 2.9207, "step": 372460 }, { "epoch": 0.0040192, "grad_norm": 0.8738544583320618, "learning_rate": 1.0669123997722574e-07, "loss": 2.7947, "step": 372470 }, { "epoch": 0.0040448, "grad_norm": 0.7925921082496643, "learning_rate": 1.0657395423519179e-07, "loss": 2.7808, "step": 372480 }, { "epoch": 0.0040704, "grad_norm": 1.0210120677947998, "learning_rate": 1.0645673264939816e-07, "loss": 2.6147, "step": 372490 }, { "epoch": 0.004096, "grad_norm": 0.8904767036437988, "learning_rate": 1.0633957522060201e-07, "loss": 2.7227, "step": 372500 }, { "epoch": 0.0041216, "grad_norm": 0.8934992551803589, "learning_rate": 1.0622248194956386e-07, "loss": 2.9166, "step": 372510 }, { "epoch": 0.0041472, "grad_norm": 0.9178336262702942, "learning_rate": 1.0610545283704421e-07, "loss": 2.8319, "step": 372520 }, { "epoch": 0.0041728, "grad_norm": 0.83414626121521, "learning_rate": 1.0598848788380023e-07, "loss": 2.8059, "step": 372530 }, { "epoch": 0.0041984, "grad_norm": 0.920409619808197, "learning_rate": 1.0587158709059242e-07, "loss": 2.7452, "step": 372540 }, { "epoch": 0.004224, "grad_norm": 0.8108920454978943, "learning_rate": 1.0575475045817685e-07, "loss": 2.8582, "step": 372550 }, { "epoch": 0.0042496, "grad_norm": 0.8295011520385742, "learning_rate": 1.056379779873129e-07, "loss": 2.9516, "step": 372560 }, { "epoch": 0.0042752, "grad_norm": 0.8087656497955322, "learning_rate": 1.0552126967875775e-07, "loss": 2.8699, "step": 372570 }, { "epoch": 0.0043008, "grad_norm": 0.9829626083374023, "learning_rate": 1.0540462553326636e-07, "loss": 2.8478, "step": 372580 }, { "epoch": 0.0043264, "grad_norm": 0.8090102672576904, "learning_rate": 1.05288045551597e-07, "loss": 2.7753, "step": 372590 }, { "epoch": 0.004352, "grad_norm": 1.0695648193359375, "learning_rate": 1.0517152973450462e-07, "loss": 2.6592, "step": 372600 }, { "epoch": 0.0043776, "grad_norm": 0.7957929968833923, "learning_rate": 1.0505507808274529e-07, "loss": 2.5845, "step": 372610 }, { "epoch": 0.0044032, "grad_norm": 0.8620430827140808, "learning_rate": 1.0493869059707396e-07, "loss": 2.7168, "step": 372620 }, { "epoch": 0.0044288, "grad_norm": 0.9523202776908875, "learning_rate": 1.0482236727824447e-07, "loss": 2.8608, "step": 372630 }, { "epoch": 0.0044544, "grad_norm": 0.8035600185394287, "learning_rate": 1.0470610812701287e-07, "loss": 2.8394, "step": 372640 }, { "epoch": 0.00448, "grad_norm": 0.961983859539032, "learning_rate": 1.0458991314413302e-07, "loss": 2.8443, "step": 372650 }, { "epoch": 0.0045056, "grad_norm": 0.8656925559043884, "learning_rate": 1.0447378233035654e-07, "loss": 2.8973, "step": 372660 }, { "epoch": 0.0045312, "grad_norm": 0.8356913328170776, "learning_rate": 1.0435771568643837e-07, "loss": 2.7355, "step": 372670 }, { "epoch": 0.0045568, "grad_norm": 0.9374784231185913, "learning_rate": 1.0424171321313014e-07, "loss": 2.7869, "step": 372680 }, { "epoch": 0.0045824, "grad_norm": 1.0042062997817993, "learning_rate": 1.0412577491118569e-07, "loss": 2.817, "step": 372690 }, { "epoch": 0.004608, "grad_norm": 1.2980101108551025, "learning_rate": 1.0400990078135442e-07, "loss": 2.7672, "step": 372700 }, { "epoch": 0.0046336, "grad_norm": 0.9139313697814941, "learning_rate": 1.0389409082438906e-07, "loss": 2.6677, "step": 372710 }, { "epoch": 0.0046592, "grad_norm": 0.8346738815307617, "learning_rate": 1.0377834504104012e-07, "loss": 2.8373, "step": 372720 }, { "epoch": 0.0046848, "grad_norm": 0.7934699654579163, "learning_rate": 1.0366266343205811e-07, "loss": 2.7187, "step": 372730 }, { "epoch": 0.0047104, "grad_norm": 0.8800737857818604, "learning_rate": 1.0354704599819466e-07, "loss": 2.7455, "step": 372740 }, { "epoch": 0.004736, "grad_norm": 0.8251137137413025, "learning_rate": 1.0343149274019804e-07, "loss": 2.6836, "step": 372750 }, { "epoch": 0.0047616, "grad_norm": 0.8192336559295654, "learning_rate": 1.0331600365881767e-07, "loss": 2.7906, "step": 372760 }, { "epoch": 0.0047872, "grad_norm": 0.8528293371200562, "learning_rate": 1.0320057875480405e-07, "loss": 2.8115, "step": 372770 }, { "epoch": 0.0048128, "grad_norm": 0.865421712398529, "learning_rate": 1.0308521802890326e-07, "loss": 2.7929, "step": 372780 }, { "epoch": 0.0048384, "grad_norm": 0.8352661728858948, "learning_rate": 1.0296992148186469e-07, "loss": 2.952, "step": 372790 }, { "epoch": 0.004864, "grad_norm": 0.9384347200393677, "learning_rate": 1.0285468911443553e-07, "loss": 2.7328, "step": 372800 }, { "epoch": 0.0048896, "grad_norm": 0.8051010370254517, "learning_rate": 1.0273952092736405e-07, "loss": 2.7678, "step": 372810 }, { "epoch": 0.0049152, "grad_norm": 1.1489622592926025, "learning_rate": 1.0262441692139636e-07, "loss": 2.7792, "step": 372820 }, { "epoch": 0.0049408, "grad_norm": 0.7822783589363098, "learning_rate": 1.0250937709727848e-07, "loss": 2.8719, "step": 372830 }, { "epoch": 0.0049664, "grad_norm": 0.7813701629638672, "learning_rate": 1.0239440145575652e-07, "loss": 2.6249, "step": 372840 }, { "epoch": 0.004992, "grad_norm": 0.8954648971557617, "learning_rate": 1.0227948999757765e-07, "loss": 2.7861, "step": 372850 }, { "epoch": 0.0050176, "grad_norm": 0.8721900582313538, "learning_rate": 1.0216464272348459e-07, "loss": 2.9722, "step": 372860 }, { "epoch": 0.0050432, "grad_norm": 0.8210049271583557, "learning_rate": 1.0204985963422342e-07, "loss": 2.8135, "step": 372870 }, { "epoch": 0.0050688, "grad_norm": 0.9449863433837891, "learning_rate": 1.0193514073053911e-07, "loss": 2.8676, "step": 372880 }, { "epoch": 0.0050944, "grad_norm": 0.8308498859405518, "learning_rate": 1.0182048601317552e-07, "loss": 2.6968, "step": 372890 }, { "epoch": 0.00512, "grad_norm": 0.9499295353889465, "learning_rate": 1.0170589548287424e-07, "loss": 2.8863, "step": 372900 }, { "epoch": 0.0051456, "grad_norm": 0.8153526186943054, "learning_rate": 1.0159136914037914e-07, "loss": 2.7922, "step": 372910 }, { "epoch": 0.0051712, "grad_norm": 0.90428227186203, "learning_rate": 1.0147690698643409e-07, "loss": 2.7298, "step": 372920 }, { "epoch": 0.0051968, "grad_norm": 0.8048972487449646, "learning_rate": 1.0136250902178068e-07, "loss": 2.8237, "step": 372930 }, { "epoch": 0.0052224, "grad_norm": 0.8577671647071838, "learning_rate": 1.0124817524715947e-07, "loss": 2.8069, "step": 372940 }, { "epoch": 0.005248, "grad_norm": 0.809965968132019, "learning_rate": 1.0113390566331426e-07, "loss": 2.5951, "step": 372950 }, { "epoch": 0.0052736, "grad_norm": 0.9855086803436279, "learning_rate": 1.0101970027098562e-07, "loss": 2.8376, "step": 372960 }, { "epoch": 0.0052992, "grad_norm": 1.0311427116394043, "learning_rate": 1.009055590709107e-07, "loss": 3.0806, "step": 372970 }, { "epoch": 0.0053248, "grad_norm": 0.7800122499465942, "learning_rate": 1.0079148206383449e-07, "loss": 2.8129, "step": 372980 }, { "epoch": 0.0053504, "grad_norm": 0.8815405964851379, "learning_rate": 1.0067746925049415e-07, "loss": 2.9555, "step": 372990 }, { "epoch": 0.005376, "grad_norm": 0.8542325496673584, "learning_rate": 1.0056352063162911e-07, "loss": 2.8685, "step": 373000 }, { "epoch": 0.0054016, "grad_norm": 0.9444904923439026, "learning_rate": 1.0044963620797877e-07, "loss": 2.8672, "step": 373010 }, { "epoch": 0.0054272, "grad_norm": 0.7818400859832764, "learning_rate": 1.0033581598028253e-07, "loss": 2.6657, "step": 373020 }, { "epoch": 0.0054528, "grad_norm": 0.9036775827407837, "learning_rate": 1.0022205994927758e-07, "loss": 2.636, "step": 373030 }, { "epoch": 0.0054784, "grad_norm": 1.0869359970092773, "learning_rate": 1.0010836811570001e-07, "loss": 2.6915, "step": 373040 }, { "epoch": 0.005504, "grad_norm": 0.8278729915618896, "learning_rate": 9.999474048028923e-08, "loss": 2.7759, "step": 373050 }, { "epoch": 0.0055296, "grad_norm": 0.8022192120552063, "learning_rate": 9.988117704378242e-08, "loss": 2.9266, "step": 373060 }, { "epoch": 0.0055552, "grad_norm": 1.4997166395187378, "learning_rate": 9.976767780691343e-08, "loss": 2.7287, "step": 373070 }, { "epoch": 0.0055808, "grad_norm": 0.7783251404762268, "learning_rate": 9.965424277042057e-08, "loss": 2.6401, "step": 373080 }, { "epoch": 0.0056064, "grad_norm": 0.9998246431350708, "learning_rate": 9.954087193503992e-08, "loss": 2.8012, "step": 373090 }, { "epoch": 0.005632, "grad_norm": 0.8131154179573059, "learning_rate": 9.942756530150422e-08, "loss": 3.0107, "step": 373100 }, { "epoch": 0.0056576, "grad_norm": 0.7733730673789978, "learning_rate": 9.931432287054954e-08, "loss": 2.6906, "step": 373110 }, { "epoch": 0.0056832, "grad_norm": 0.7740695476531982, "learning_rate": 9.920114464290976e-08, "loss": 2.739, "step": 373120 }, { "epoch": 0.0057088, "grad_norm": 0.9477931261062622, "learning_rate": 9.908803061931982e-08, "loss": 2.6388, "step": 373130 }, { "epoch": 0.0057344, "grad_norm": 0.835657000541687, "learning_rate": 9.897498080051249e-08, "loss": 2.7898, "step": 373140 }, { "epoch": 0.00576, "grad_norm": 0.8303887844085693, "learning_rate": 9.886199518722162e-08, "loss": 2.5954, "step": 373150 }, { "epoch": 0.0057856, "grad_norm": 0.876237154006958, "learning_rate": 9.874907378017884e-08, "loss": 2.7885, "step": 373160 }, { "epoch": 0.0058112, "grad_norm": 0.9448190927505493, "learning_rate": 9.86362165801169e-08, "loss": 2.8214, "step": 373170 }, { "epoch": 0.0058368, "grad_norm": 0.8339215517044067, "learning_rate": 9.852342358776745e-08, "loss": 2.8103, "step": 373180 }, { "epoch": 0.0058624, "grad_norm": 0.8653324842453003, "learning_rate": 9.841069480386212e-08, "loss": 2.6816, "step": 373190 }, { "epoch": 0.005888, "grad_norm": 0.9338118433952332, "learning_rate": 9.829803022913254e-08, "loss": 2.635, "step": 373200 }, { "epoch": 0.0059136, "grad_norm": 0.8149963617324829, "learning_rate": 9.818542986430813e-08, "loss": 2.7159, "step": 373210 }, { "epoch": 0.0059392, "grad_norm": 0.8869471549987793, "learning_rate": 9.807289371012052e-08, "loss": 2.7191, "step": 373220 }, { "epoch": 0.0059648, "grad_norm": 0.8890383839607239, "learning_rate": 9.796042176729692e-08, "loss": 2.6762, "step": 373230 }, { "epoch": 0.0059904, "grad_norm": 0.8017902970314026, "learning_rate": 9.784801403656896e-08, "loss": 2.5711, "step": 373240 }, { "epoch": 0.006016, "grad_norm": 0.8960568308830261, "learning_rate": 9.773567051866606e-08, "loss": 2.921, "step": 373250 }, { "epoch": 0.0060416, "grad_norm": 0.8666635751724243, "learning_rate": 9.76233912143143e-08, "loss": 2.9592, "step": 373260 }, { "epoch": 0.0060672, "grad_norm": 1.0951660871505737, "learning_rate": 9.75111761242431e-08, "loss": 2.5893, "step": 373270 }, { "epoch": 0.0060928, "grad_norm": 0.7762454152107239, "learning_rate": 9.739902524918077e-08, "loss": 2.7246, "step": 373280 }, { "epoch": 0.0061184, "grad_norm": 0.8339923620223999, "learning_rate": 9.728693858985449e-08, "loss": 2.7995, "step": 373290 }, { "epoch": 0.006144, "grad_norm": 0.8479441404342651, "learning_rate": 9.717491614699037e-08, "loss": 2.808, "step": 373300 }, { "epoch": 0.0061696, "grad_norm": 0.8011808395385742, "learning_rate": 9.706295792131448e-08, "loss": 2.739, "step": 373310 }, { "epoch": 0.0061952, "grad_norm": 0.8742256760597229, "learning_rate": 9.695106391355292e-08, "loss": 2.7979, "step": 373320 }, { "epoch": 0.0062208, "grad_norm": 0.8740690350532532, "learning_rate": 9.683923412443175e-08, "loss": 2.823, "step": 373330 }, { "epoch": 0.0062464, "grad_norm": 0.7612698674201965, "learning_rate": 9.672746855467707e-08, "loss": 2.6592, "step": 373340 }, { "epoch": 0.006272, "grad_norm": 0.8150994777679443, "learning_rate": 9.661576720501276e-08, "loss": 2.7388, "step": 373350 }, { "epoch": 0.0062976, "grad_norm": 0.9066623449325562, "learning_rate": 9.650413007616377e-08, "loss": 2.7444, "step": 373360 }, { "epoch": 0.0063232, "grad_norm": 0.8160685300827026, "learning_rate": 9.639255716885176e-08, "loss": 2.8107, "step": 373370 }, { "epoch": 0.0063488, "grad_norm": 0.9439440965652466, "learning_rate": 9.628104848380282e-08, "loss": 2.8037, "step": 373380 }, { "epoch": 0.0063744, "grad_norm": 1.1461155414581299, "learning_rate": 9.61696040217397e-08, "loss": 2.7236, "step": 373390 }, { "epoch": 0.0064, "grad_norm": 0.7835798859596252, "learning_rate": 9.605822378338403e-08, "loss": 2.9072, "step": 373400 }, { "epoch": 0.0064256, "grad_norm": 0.8301421403884888, "learning_rate": 9.594690776945859e-08, "loss": 2.8763, "step": 373410 }, { "epoch": 0.0064512, "grad_norm": 0.8457674980163574, "learning_rate": 9.583565598068612e-08, "loss": 2.8108, "step": 373420 }, { "epoch": 0.0064768, "grad_norm": 0.8361811637878418, "learning_rate": 9.572446841778604e-08, "loss": 2.8119, "step": 373430 }, { "epoch": 0.0065024, "grad_norm": 0.8969249129295349, "learning_rate": 9.561334508148112e-08, "loss": 2.7323, "step": 373440 }, { "epoch": 0.006528, "grad_norm": 0.7835311889648438, "learning_rate": 9.55022859724919e-08, "loss": 2.8293, "step": 373450 }, { "epoch": 0.0065536, "grad_norm": 0.8526574969291687, "learning_rate": 9.539129109153666e-08, "loss": 2.7083, "step": 373460 }, { "epoch": 0.0065792, "grad_norm": 0.9857044219970703, "learning_rate": 9.528036043933709e-08, "loss": 2.7999, "step": 373470 }, { "epoch": 0.0066048, "grad_norm": 0.8424628376960754, "learning_rate": 9.516949401661257e-08, "loss": 2.702, "step": 373480 }, { "epoch": 0.0066304, "grad_norm": 1.3711737394332886, "learning_rate": 9.505869182408145e-08, "loss": 2.8868, "step": 373490 }, { "epoch": 0.006656, "grad_norm": 0.931077241897583, "learning_rate": 9.494795386246202e-08, "loss": 2.7228, "step": 373500 }, { "epoch": 0.0066816, "grad_norm": 0.8430227041244507, "learning_rate": 9.483728013247262e-08, "loss": 3.1712, "step": 373510 }, { "epoch": 0.0067072, "grad_norm": 0.8833321332931519, "learning_rate": 9.472667063483043e-08, "loss": 2.8485, "step": 373520 }, { "epoch": 0.0067328, "grad_norm": 0.9763581156730652, "learning_rate": 9.461612537025378e-08, "loss": 2.8123, "step": 373530 }, { "epoch": 0.0067584, "grad_norm": 1.0544663667678833, "learning_rate": 9.450564433945985e-08, "loss": 2.9491, "step": 373540 }, { "epoch": 0.006784, "grad_norm": 0.8750224113464355, "learning_rate": 9.439522754316366e-08, "loss": 2.8148, "step": 373550 }, { "epoch": 0.0068096, "grad_norm": 0.9044339060783386, "learning_rate": 9.428487498208128e-08, "loss": 2.6978, "step": 373560 }, { "epoch": 0.0068352, "grad_norm": 0.8438351154327393, "learning_rate": 9.417458665692992e-08, "loss": 2.8575, "step": 373570 }, { "epoch": 0.0068608, "grad_norm": 0.7757424712181091, "learning_rate": 9.406436256842233e-08, "loss": 2.5641, "step": 373580 }, { "epoch": 0.0068864, "grad_norm": 1.05429208278656, "learning_rate": 9.395420271727573e-08, "loss": 2.8275, "step": 373590 }, { "epoch": 0.006912, "grad_norm": 0.790964663028717, "learning_rate": 9.384410710420289e-08, "loss": 2.6816, "step": 373600 }, { "epoch": 0.0069376, "grad_norm": 0.8148099184036255, "learning_rate": 9.373407572991878e-08, "loss": 2.7808, "step": 373610 }, { "epoch": 0.0069632, "grad_norm": 0.8423959612846375, "learning_rate": 9.362410859513615e-08, "loss": 2.7311, "step": 373620 }, { "epoch": 0.0069888, "grad_norm": 0.8798120617866516, "learning_rate": 9.351420570056891e-08, "loss": 2.6033, "step": 373630 }, { "epoch": 0.0070144, "grad_norm": 0.8015105724334717, "learning_rate": 9.340436704692868e-08, "loss": 2.8174, "step": 373640 }, { "epoch": 0.00704, "grad_norm": 0.8926591277122498, "learning_rate": 9.329459263492824e-08, "loss": 2.7961, "step": 373650 }, { "epoch": 0.0070656, "grad_norm": 0.7990656495094299, "learning_rate": 9.318488246527924e-08, "loss": 2.8499, "step": 373660 }, { "epoch": 0.0070912, "grad_norm": 0.8952022790908813, "learning_rate": 9.307523653869333e-08, "loss": 2.6529, "step": 373670 }, { "epoch": 0.0071168, "grad_norm": 1.017734408378601, "learning_rate": 9.296565485588215e-08, "loss": 2.6491, "step": 373680 }, { "epoch": 0.0071424, "grad_norm": 0.9708926677703857, "learning_rate": 9.285613741755516e-08, "loss": 2.7425, "step": 373690 }, { "epoch": 0.007168, "grad_norm": 0.9684304594993591, "learning_rate": 9.27466842244229e-08, "loss": 2.7124, "step": 373700 }, { "epoch": 0.0071936, "grad_norm": 0.9642883539199829, "learning_rate": 9.26372952771959e-08, "loss": 2.744, "step": 373710 }, { "epoch": 0.0072192, "grad_norm": 1.0741467475891113, "learning_rate": 9.252797057658358e-08, "loss": 2.9073, "step": 373720 }, { "epoch": 0.0072448, "grad_norm": 0.8026836514472961, "learning_rate": 9.241871012329317e-08, "loss": 2.6461, "step": 373730 }, { "epoch": 0.0072704, "grad_norm": 0.8286633491516113, "learning_rate": 9.230951391803522e-08, "loss": 3.0289, "step": 373740 }, { "epoch": 0.007296, "grad_norm": 0.914479672908783, "learning_rate": 9.220038196151693e-08, "loss": 2.7876, "step": 373750 }, { "epoch": 0.0073216, "grad_norm": 0.8949214220046997, "learning_rate": 9.209131425444551e-08, "loss": 2.9768, "step": 373760 }, { "epoch": 0.0073472, "grad_norm": 0.7652578353881836, "learning_rate": 9.198231079752817e-08, "loss": 2.9543, "step": 373770 }, { "epoch": 0.0073728, "grad_norm": 0.8214464783668518, "learning_rate": 9.187337159147325e-08, "loss": 2.8222, "step": 373780 }, { "epoch": 0.0073984, "grad_norm": 0.8190248608589172, "learning_rate": 9.176449663698572e-08, "loss": 2.8625, "step": 373790 }, { "epoch": 0.007424, "grad_norm": 0.8001357913017273, "learning_rate": 9.16556859347717e-08, "loss": 2.6445, "step": 373800 }, { "epoch": 0.0074496, "grad_norm": 0.969976007938385, "learning_rate": 9.154693948553839e-08, "loss": 2.86, "step": 373810 }, { "epoch": 0.0074752, "grad_norm": 0.8525617718696594, "learning_rate": 9.143825728998968e-08, "loss": 2.816, "step": 373820 }, { "epoch": 0.0075008, "grad_norm": 0.8097096085548401, "learning_rate": 9.132963934882943e-08, "loss": 2.8366, "step": 373830 }, { "epoch": 0.0075264, "grad_norm": 0.8288354277610779, "learning_rate": 9.122108566276377e-08, "loss": 2.9886, "step": 373840 }, { "epoch": 0.007552, "grad_norm": 0.8376903533935547, "learning_rate": 9.111259623249546e-08, "loss": 2.7704, "step": 373850 }, { "epoch": 0.0075776, "grad_norm": 0.8579772710800171, "learning_rate": 9.100417105872839e-08, "loss": 2.8346, "step": 373860 }, { "epoch": 0.0076032, "grad_norm": 0.8940499424934387, "learning_rate": 9.089581014216641e-08, "loss": 2.8024, "step": 373870 }, { "epoch": 0.0076288, "grad_norm": 0.7954014539718628, "learning_rate": 9.078751348351233e-08, "loss": 2.886, "step": 373880 }, { "epoch": 0.0076544, "grad_norm": 0.9357423186302185, "learning_rate": 9.067928108346668e-08, "loss": 2.8738, "step": 373890 }, { "epoch": 0.00768, "grad_norm": 0.7921644449234009, "learning_rate": 9.057111294273225e-08, "loss": 2.9914, "step": 373900 }, { "epoch": 0.0077056, "grad_norm": 0.8777740597724915, "learning_rate": 9.046300906201066e-08, "loss": 2.734, "step": 373910 }, { "epoch": 0.0077312, "grad_norm": 0.7627873420715332, "learning_rate": 9.03549694420025e-08, "loss": 2.8185, "step": 373920 }, { "epoch": 0.0077568, "grad_norm": 0.8163464069366455, "learning_rate": 9.024699408340942e-08, "loss": 2.7548, "step": 373930 }, { "epoch": 0.0077824, "grad_norm": 0.8747544288635254, "learning_rate": 9.013908298692975e-08, "loss": 2.9861, "step": 373940 }, { "epoch": 0.007808, "grad_norm": 0.7901886105537415, "learning_rate": 9.003123615326626e-08, "loss": 2.8373, "step": 373950 }, { "epoch": 0.0078336, "grad_norm": 0.9646729230880737, "learning_rate": 8.992345358311505e-08, "loss": 2.8692, "step": 373960 }, { "epoch": 0.0078592, "grad_norm": 0.8510227203369141, "learning_rate": 8.981573527717669e-08, "loss": 2.5777, "step": 373970 }, { "epoch": 0.0078848, "grad_norm": 0.927617073059082, "learning_rate": 8.970808123614949e-08, "loss": 2.9468, "step": 373980 }, { "epoch": 0.0079104, "grad_norm": 0.9775382876396179, "learning_rate": 8.96004914607318e-08, "loss": 2.7662, "step": 373990 }, { "epoch": 0.007936, "grad_norm": 0.8107665181159973, "learning_rate": 8.949296595162083e-08, "loss": 2.8087, "step": 374000 }, { "epoch": 0.0079616, "grad_norm": 0.808343231678009, "learning_rate": 8.938550470951379e-08, "loss": 2.9101, "step": 374010 }, { "epoch": 0.0079872, "grad_norm": 0.9786348342895508, "learning_rate": 8.927810773510792e-08, "loss": 2.7904, "step": 374020 }, { "epoch": 0.0080128, "grad_norm": 0.8404473662376404, "learning_rate": 8.917077502910043e-08, "loss": 2.7046, "step": 374030 }, { "epoch": 0.0080384, "grad_norm": 0.8584869503974915, "learning_rate": 8.906350659218521e-08, "loss": 2.7286, "step": 374040 }, { "epoch": 0.008064, "grad_norm": 0.8681621551513672, "learning_rate": 8.895630242505948e-08, "loss": 2.7756, "step": 374050 }, { "epoch": 0.0080896, "grad_norm": 0.843827486038208, "learning_rate": 8.884916252841824e-08, "loss": 2.8456, "step": 374060 }, { "epoch": 0.0081152, "grad_norm": 0.7844198942184448, "learning_rate": 8.87420869029576e-08, "loss": 2.9931, "step": 374070 }, { "epoch": 0.0081408, "grad_norm": 1.1528836488723755, "learning_rate": 8.863507554936812e-08, "loss": 2.7214, "step": 374080 }, { "epoch": 0.0081664, "grad_norm": 0.8012651205062866, "learning_rate": 8.852812846834701e-08, "loss": 2.8978, "step": 374090 }, { "epoch": 0.008192, "grad_norm": 0.8940058946609497, "learning_rate": 8.842124566058707e-08, "loss": 2.8034, "step": 374100 }, { "epoch": 0.0082176, "grad_norm": 0.8291801810264587, "learning_rate": 8.831442712678107e-08, "loss": 2.8662, "step": 374110 }, { "epoch": 0.0082432, "grad_norm": 0.8156648874282837, "learning_rate": 8.820767286762177e-08, "loss": 2.7875, "step": 374120 }, { "epoch": 0.0082688, "grad_norm": 0.851632297039032, "learning_rate": 8.810098288380086e-08, "loss": 2.831, "step": 374130 }, { "epoch": 0.0082944, "grad_norm": 0.8937585353851318, "learning_rate": 8.799435717601113e-08, "loss": 2.835, "step": 374140 }, { "epoch": 0.00832, "grad_norm": 0.872497022151947, "learning_rate": 8.788779574494533e-08, "loss": 2.9468, "step": 374150 }, { "epoch": 0.0083456, "grad_norm": 0.808867871761322, "learning_rate": 8.77812985912907e-08, "loss": 2.8636, "step": 374160 }, { "epoch": 0.0083712, "grad_norm": 0.8522369265556335, "learning_rate": 8.767486571574113e-08, "loss": 2.8167, "step": 374170 }, { "epoch": 0.0083968, "grad_norm": 0.8895010948181152, "learning_rate": 8.756849711898496e-08, "loss": 2.6634, "step": 374180 }, { "epoch": 0.0084224, "grad_norm": 0.7123852968215942, "learning_rate": 8.746219280171275e-08, "loss": 2.4192, "step": 374190 }, { "epoch": 0.008448, "grad_norm": 0.8275769948959351, "learning_rate": 8.735595276461394e-08, "loss": 2.79, "step": 374200 }, { "epoch": 0.0084736, "grad_norm": 0.8873562216758728, "learning_rate": 8.724977700837689e-08, "loss": 2.8057, "step": 374210 }, { "epoch": 0.0084992, "grad_norm": 0.933141827583313, "learning_rate": 8.714366553369102e-08, "loss": 2.9558, "step": 374220 }, { "epoch": 0.0085248, "grad_norm": 0.859515905380249, "learning_rate": 8.703761834124469e-08, "loss": 2.828, "step": 374230 }, { "epoch": 0.0085504, "grad_norm": 0.8221818804740906, "learning_rate": 8.6931635431724e-08, "loss": 2.7953, "step": 374240 }, { "epoch": 0.008576, "grad_norm": 0.8459571003913879, "learning_rate": 8.68257168058173e-08, "loss": 2.6954, "step": 374250 }, { "epoch": 0.0086016, "grad_norm": 0.9671432971954346, "learning_rate": 8.671986246421072e-08, "loss": 2.8058, "step": 374260 }, { "epoch": 0.0086272, "grad_norm": 0.9614924788475037, "learning_rate": 8.661407240759257e-08, "loss": 2.7626, "step": 374270 }, { "epoch": 0.0086528, "grad_norm": 0.8210406303405762, "learning_rate": 8.650834663664564e-08, "loss": 2.5745, "step": 374280 }, { "epoch": 0.0086784, "grad_norm": 0.8169235587120056, "learning_rate": 8.64026851520583e-08, "loss": 2.7371, "step": 374290 }, { "epoch": 0.008704, "grad_norm": 0.9452579021453857, "learning_rate": 8.629708795451552e-08, "loss": 2.7723, "step": 374300 }, { "epoch": 0.0087296, "grad_norm": 0.8629665970802307, "learning_rate": 8.619155504470012e-08, "loss": 2.757, "step": 374310 }, { "epoch": 0.0087552, "grad_norm": 0.8234354257583618, "learning_rate": 8.608608642329819e-08, "loss": 2.7713, "step": 374320 }, { "epoch": 0.0087808, "grad_norm": 1.004435420036316, "learning_rate": 8.598068209099364e-08, "loss": 2.6989, "step": 374330 }, { "epoch": 0.0088064, "grad_norm": 0.8491721153259277, "learning_rate": 8.587534204846814e-08, "loss": 2.8044, "step": 374340 }, { "epoch": 0.008832, "grad_norm": 0.8918392062187195, "learning_rate": 8.577006629640783e-08, "loss": 2.6938, "step": 374350 }, { "epoch": 0.0088576, "grad_norm": 0.8821796774864197, "learning_rate": 8.566485483549325e-08, "loss": 2.5656, "step": 374360 }, { "epoch": 0.0088832, "grad_norm": 0.8259584307670593, "learning_rate": 8.55597076664072e-08, "loss": 2.8628, "step": 374370 }, { "epoch": 0.0089088, "grad_norm": 0.8500401377677917, "learning_rate": 8.545462478983024e-08, "loss": 2.8596, "step": 374380 }, { "epoch": 0.0089344, "grad_norm": 0.8176271319389343, "learning_rate": 8.534960620644627e-08, "loss": 2.7506, "step": 374390 }, { "epoch": 0.00896, "grad_norm": 0.8763490915298462, "learning_rate": 8.524465191693587e-08, "loss": 2.8091, "step": 374400 }, { "epoch": 0.0089856, "grad_norm": 0.8933868408203125, "learning_rate": 8.513976192197848e-08, "loss": 2.891, "step": 374410 }, { "epoch": 0.0090112, "grad_norm": 0.8270788788795471, "learning_rate": 8.503493622225356e-08, "loss": 2.7626, "step": 374420 }, { "epoch": 0.0090368, "grad_norm": 0.8393080830574036, "learning_rate": 8.49301748184428e-08, "loss": 3.0816, "step": 374430 }, { "epoch": 0.0090624, "grad_norm": 0.9029313921928406, "learning_rate": 8.482547771122452e-08, "loss": 3.1611, "step": 374440 }, { "epoch": 0.009088, "grad_norm": 1.0474034547805786, "learning_rate": 8.472084490127819e-08, "loss": 2.8625, "step": 374450 }, { "epoch": 0.0091136, "grad_norm": 0.8357837796211243, "learning_rate": 8.461627638928105e-08, "loss": 2.8542, "step": 374460 }, { "epoch": 0.0091392, "grad_norm": 1.0722718238830566, "learning_rate": 8.451177217591366e-08, "loss": 3.2908, "step": 374470 }, { "epoch": 0.0091648, "grad_norm": 0.8008103370666504, "learning_rate": 8.440733226185105e-08, "loss": 2.8699, "step": 374480 }, { "epoch": 0.0091904, "grad_norm": 0.9055253863334656, "learning_rate": 8.430295664777266e-08, "loss": 2.9185, "step": 374490 }, { "epoch": 0.009216, "grad_norm": 0.8358142375946045, "learning_rate": 8.419864533435352e-08, "loss": 2.8421, "step": 374500 }, { "epoch": 0.0092416, "grad_norm": 0.8710264563560486, "learning_rate": 8.409439832227084e-08, "loss": 2.6697, "step": 374510 }, { "epoch": 0.0092672, "grad_norm": 1.1063326597213745, "learning_rate": 8.399021561220078e-08, "loss": 2.9101, "step": 374520 }, { "epoch": 0.0092928, "grad_norm": 0.8346542119979858, "learning_rate": 8.388609720481944e-08, "loss": 2.6861, "step": 374530 }, { "epoch": 0.0093184, "grad_norm": 0.7943387627601624, "learning_rate": 8.378204310080073e-08, "loss": 2.7104, "step": 374540 }, { "epoch": 0.009344, "grad_norm": 0.9005441069602966, "learning_rate": 8.367805330081969e-08, "loss": 2.8954, "step": 374550 }, { "epoch": 0.0093696, "grad_norm": 0.8770171403884888, "learning_rate": 8.35741278055513e-08, "loss": 2.8403, "step": 374560 }, { "epoch": 0.0093952, "grad_norm": 0.8042111396789551, "learning_rate": 8.347026661566948e-08, "loss": 2.9167, "step": 374570 }, { "epoch": 0.0094208, "grad_norm": 0.8865512609481812, "learning_rate": 8.336646973184703e-08, "loss": 2.8673, "step": 374580 }, { "epoch": 0.0094464, "grad_norm": 1.1291429996490479, "learning_rate": 8.326273715475785e-08, "loss": 2.8141, "step": 374590 }, { "epoch": 0.009472, "grad_norm": 0.8672139644622803, "learning_rate": 8.315906888507475e-08, "loss": 2.7155, "step": 374600 }, { "epoch": 0.0094976, "grad_norm": 0.8780958652496338, "learning_rate": 8.305546492346939e-08, "loss": 3.0215, "step": 374610 }, { "epoch": 0.0095232, "grad_norm": 0.8717651963233948, "learning_rate": 8.295192527061347e-08, "loss": 2.8646, "step": 374620 }, { "epoch": 0.0095488, "grad_norm": 0.8629282116889954, "learning_rate": 8.284844992717866e-08, "loss": 2.8439, "step": 374630 }, { "epoch": 0.0095744, "grad_norm": 0.8338721990585327, "learning_rate": 8.274503889383556e-08, "loss": 2.8023, "step": 374640 }, { "epoch": 0.0096, "grad_norm": 0.838717520236969, "learning_rate": 8.264169217125584e-08, "loss": 2.6411, "step": 374650 }, { "epoch": 0.0096256, "grad_norm": 0.7992100715637207, "learning_rate": 8.253840976010785e-08, "loss": 2.8232, "step": 374660 }, { "epoch": 0.0096512, "grad_norm": 0.8094617128372192, "learning_rate": 8.243519166106439e-08, "loss": 2.8296, "step": 374670 }, { "epoch": 0.0096768, "grad_norm": 0.8265411853790283, "learning_rate": 8.233203787479161e-08, "loss": 2.8948, "step": 374680 }, { "epoch": 0.0097024, "grad_norm": 0.8740963935852051, "learning_rate": 8.222894840196005e-08, "loss": 2.6964, "step": 374690 }, { "epoch": 0.009728, "grad_norm": 0.9082867503166199, "learning_rate": 8.212592324323809e-08, "loss": 2.7526, "step": 374700 }, { "epoch": 0.0097536, "grad_norm": 0.8073769807815552, "learning_rate": 8.202296239929297e-08, "loss": 2.9383, "step": 374710 }, { "epoch": 0.0097792, "grad_norm": 1.032896876335144, "learning_rate": 8.192006587079415e-08, "loss": 2.9346, "step": 374720 }, { "epoch": 0.0098048, "grad_norm": 0.9357693195343018, "learning_rate": 8.181723365840777e-08, "loss": 2.9984, "step": 374730 }, { "epoch": 0.0098304, "grad_norm": 0.9372437000274658, "learning_rate": 8.171446576279996e-08, "loss": 2.8835, "step": 374740 }, { "epoch": 0.009856, "grad_norm": 0.8946948051452637, "learning_rate": 8.161176218463795e-08, "loss": 2.9743, "step": 374750 }, { "epoch": 0.0098816, "grad_norm": 0.9282420873641968, "learning_rate": 8.15091229245879e-08, "loss": 2.842, "step": 374760 }, { "epoch": 0.0099072, "grad_norm": 0.8844736218452454, "learning_rate": 8.140654798331482e-08, "loss": 2.8807, "step": 374770 }, { "epoch": 0.0099328, "grad_norm": 0.8636658191680908, "learning_rate": 8.130403736148485e-08, "loss": 2.9, "step": 374780 }, { "epoch": 0.0099584, "grad_norm": 0.877210795879364, "learning_rate": 8.120159105976188e-08, "loss": 3.0452, "step": 374790 }, { "epoch": 0.009984, "grad_norm": 0.7826092839241028, "learning_rate": 8.109920907881097e-08, "loss": 2.6099, "step": 374800 }, { "epoch": 0.0100096, "grad_norm": 0.8244284987449646, "learning_rate": 8.09968914192949e-08, "loss": 2.6697, "step": 374810 }, { "epoch": 0.0100352, "grad_norm": 0.806308388710022, "learning_rate": 8.089463808187759e-08, "loss": 2.936, "step": 374820 }, { "epoch": 0.0100608, "grad_norm": 0.9108853936195374, "learning_rate": 8.079244906722294e-08, "loss": 2.8518, "step": 374830 }, { "epoch": 0.0100864, "grad_norm": 1.0768725872039795, "learning_rate": 8.069032437599267e-08, "loss": 2.6927, "step": 374840 }, { "epoch": 0.010112, "grad_norm": 0.9000257253646851, "learning_rate": 8.058826400884956e-08, "loss": 2.8104, "step": 374850 }, { "epoch": 0.0101376, "grad_norm": 0.9953621625900269, "learning_rate": 8.048626796645531e-08, "loss": 2.699, "step": 374860 }, { "epoch": 0.0101632, "grad_norm": 0.8663427829742432, "learning_rate": 8.038433624947161e-08, "loss": 2.869, "step": 374870 }, { "epoch": 0.0101888, "grad_norm": 0.9262087345123291, "learning_rate": 8.028246885855906e-08, "loss": 2.9539, "step": 374880 }, { "epoch": 0.0102144, "grad_norm": 0.8196138143539429, "learning_rate": 8.018066579437823e-08, "loss": 2.7335, "step": 374890 }, { "epoch": 0.01024, "grad_norm": 0.7789534330368042, "learning_rate": 8.007892705758858e-08, "loss": 2.898, "step": 374900 }, { "epoch": 0.0102656, "grad_norm": 0.8513184785842896, "learning_rate": 7.99772526488518e-08, "loss": 2.8631, "step": 374910 }, { "epoch": 0.0102912, "grad_norm": 0.8693890571594238, "learning_rate": 7.987564256882518e-08, "loss": 2.7256, "step": 374920 }, { "epoch": 0.0103168, "grad_norm": 0.8658442497253418, "learning_rate": 7.977409681817039e-08, "loss": 2.9395, "step": 374930 }, { "epoch": 0.0103424, "grad_norm": 0.793820858001709, "learning_rate": 7.967261539754245e-08, "loss": 3.0414, "step": 374940 }, { "epoch": 0.010368, "grad_norm": 0.8147531151771545, "learning_rate": 7.957119830760196e-08, "loss": 2.8024, "step": 374950 }, { "epoch": 0.0103936, "grad_norm": 1.2160568237304688, "learning_rate": 7.946984554900506e-08, "loss": 2.8848, "step": 374960 }, { "epoch": 0.0104192, "grad_norm": 0.7971901893615723, "learning_rate": 7.93685571224101e-08, "loss": 3.0552, "step": 374970 }, { "epoch": 0.0104448, "grad_norm": 0.9174196124076843, "learning_rate": 7.926733302847323e-08, "loss": 2.6835, "step": 374980 }, { "epoch": 0.0104704, "grad_norm": 1.0035241842269897, "learning_rate": 7.916617326785059e-08, "loss": 2.7768, "step": 374990 }, { "epoch": 0.010496, "grad_norm": 0.8357875347137451, "learning_rate": 7.906507784119943e-08, "loss": 2.7792, "step": 375000 }, { "epoch": 0.0105216, "grad_norm": 1.181652307510376, "learning_rate": 7.89640467491748e-08, "loss": 2.7519, "step": 375010 }, { "epoch": 0.0105472, "grad_norm": 0.7663217186927795, "learning_rate": 7.886307999243059e-08, "loss": 2.6254, "step": 375020 }, { "epoch": 0.0105728, "grad_norm": 0.9850071668624878, "learning_rate": 7.876217757162297e-08, "loss": 2.8674, "step": 375030 }, { "epoch": 0.0105984, "grad_norm": 0.7974961996078491, "learning_rate": 7.866133948740584e-08, "loss": 2.763, "step": 375040 }, { "epoch": 0.010624, "grad_norm": 0.9339834451675415, "learning_rate": 7.856056574043424e-08, "loss": 2.6429, "step": 375050 }, { "epoch": 0.0106496, "grad_norm": 0.7990968227386475, "learning_rate": 7.845985633135877e-08, "loss": 2.9253, "step": 375060 }, { "epoch": 0.0106752, "grad_norm": 0.8166605234146118, "learning_rate": 7.835921126083557e-08, "loss": 2.7037, "step": 375070 }, { "epoch": 0.0107008, "grad_norm": 0.8417556881904602, "learning_rate": 7.825863052951522e-08, "loss": 2.8308, "step": 375080 }, { "epoch": 0.0107264, "grad_norm": 0.7985029816627502, "learning_rate": 7.815811413805163e-08, "loss": 2.8464, "step": 375090 }, { "epoch": 0.010752, "grad_norm": 0.8267159461975098, "learning_rate": 7.805766208709542e-08, "loss": 2.994, "step": 375100 }, { "epoch": 0.0107776, "grad_norm": 0.8786731958389282, "learning_rate": 7.795727437729827e-08, "loss": 2.8526, "step": 375110 }, { "epoch": 0.0108032, "grad_norm": 0.9401420950889587, "learning_rate": 7.785695100931079e-08, "loss": 2.7822, "step": 375120 }, { "epoch": 0.0108288, "grad_norm": 0.9626733660697937, "learning_rate": 7.775669198378466e-08, "loss": 2.7341, "step": 375130 }, { "epoch": 0.0108544, "grad_norm": 0.8601813912391663, "learning_rate": 7.765649730136937e-08, "loss": 2.8179, "step": 375140 }, { "epoch": 0.01088, "grad_norm": 0.8625580668449402, "learning_rate": 7.755636696271441e-08, "loss": 2.75, "step": 375150 }, { "epoch": 0.0109056, "grad_norm": 0.7952471375465393, "learning_rate": 7.745630096846923e-08, "loss": 2.7898, "step": 375160 }, { "epoch": 0.0109312, "grad_norm": 0.8319797515869141, "learning_rate": 7.735629931928335e-08, "loss": 2.9077, "step": 375170 }, { "epoch": 0.0109568, "grad_norm": 0.7846043705940247, "learning_rate": 7.7256362015804e-08, "loss": 2.7668, "step": 375180 }, { "epoch": 0.0109824, "grad_norm": 0.7816277742385864, "learning_rate": 7.715648905868068e-08, "loss": 2.7941, "step": 375190 }, { "epoch": 0.011008, "grad_norm": 3.1515629291534424, "learning_rate": 7.705668044856063e-08, "loss": 2.7117, "step": 375200 }, { "epoch": 0.0110336, "grad_norm": 0.86112380027771, "learning_rate": 7.695693618609001e-08, "loss": 2.7208, "step": 375210 }, { "epoch": 0.0110592, "grad_norm": 0.8498995304107666, "learning_rate": 7.685725627191609e-08, "loss": 2.7794, "step": 375220 }, { "epoch": 0.0110848, "grad_norm": 0.8370541930198669, "learning_rate": 7.675764070668612e-08, "loss": 2.6125, "step": 375230 }, { "epoch": 0.0111104, "grad_norm": 0.843177080154419, "learning_rate": 7.665808949104515e-08, "loss": 3.3387, "step": 375240 }, { "epoch": 0.011136, "grad_norm": 0.9543368816375732, "learning_rate": 7.65586026256393e-08, "loss": 2.9697, "step": 375250 }, { "epoch": 0.0111616, "grad_norm": 0.8632746934890747, "learning_rate": 7.645918011111475e-08, "loss": 2.9719, "step": 375260 }, { "epoch": 0.0111872, "grad_norm": 0.9213758707046509, "learning_rate": 7.635982194811431e-08, "loss": 2.6097, "step": 375270 }, { "epoch": 0.0112128, "grad_norm": 0.7959764003753662, "learning_rate": 7.626052813728302e-08, "loss": 2.6848, "step": 375280 }, { "epoch": 0.0112384, "grad_norm": 0.8198168277740479, "learning_rate": 7.616129867926481e-08, "loss": 2.7043, "step": 375290 }, { "epoch": 0.011264, "grad_norm": 0.9495964050292969, "learning_rate": 7.606213357470249e-08, "loss": 2.8691, "step": 375300 }, { "epoch": 0.0112896, "grad_norm": 1.0375478267669678, "learning_rate": 7.596303282424e-08, "loss": 2.952, "step": 375310 }, { "epoch": 0.0113152, "grad_norm": 0.8985689878463745, "learning_rate": 7.586399642852016e-08, "loss": 2.8622, "step": 375320 }, { "epoch": 0.0113408, "grad_norm": 0.9387410879135132, "learning_rate": 7.576502438818579e-08, "loss": 2.7247, "step": 375330 }, { "epoch": 0.0113664, "grad_norm": 0.8510407209396362, "learning_rate": 7.566611670387635e-08, "loss": 2.9446, "step": 375340 }, { "epoch": 0.011392, "grad_norm": 0.823901891708374, "learning_rate": 7.556727337623471e-08, "loss": 2.7413, "step": 375350 }, { "epoch": 0.0114176, "grad_norm": 0.8912700414657593, "learning_rate": 7.546849440590142e-08, "loss": 2.837, "step": 375360 }, { "epoch": 0.0114432, "grad_norm": 0.8202436566352844, "learning_rate": 7.536977979351822e-08, "loss": 2.8177, "step": 375370 }, { "epoch": 0.0114688, "grad_norm": 0.860365629196167, "learning_rate": 7.527112953972348e-08, "loss": 2.6416, "step": 375380 }, { "epoch": 0.0114944, "grad_norm": 0.8360256552696228, "learning_rate": 7.51725436451578e-08, "loss": 2.6965, "step": 375390 }, { "epoch": 0.01152, "grad_norm": 0.7737678289413452, "learning_rate": 7.507402211046178e-08, "loss": 3.1394, "step": 375400 }, { "epoch": 2.56e-05, "grad_norm": 0.9831908941268921, "learning_rate": 7.497556493627045e-08, "loss": 1.8055, "step": 375410 }, { "epoch": 5.12e-05, "grad_norm": 0.7606781721115112, "learning_rate": 7.487717212322665e-08, "loss": 2.6596, "step": 375420 }, { "epoch": 7.68e-05, "grad_norm": 0.7369675636291504, "learning_rate": 7.477884367196653e-08, "loss": 2.6875, "step": 375430 }, { "epoch": 0.0001024, "grad_norm": 0.8791579604148865, "learning_rate": 7.468057958312735e-08, "loss": 2.6599, "step": 375440 }, { "epoch": 0.000128, "grad_norm": 0.8189435601234436, "learning_rate": 7.45823798573475e-08, "loss": 2.6614, "step": 375450 }, { "epoch": 0.0001536, "grad_norm": 0.7861924767494202, "learning_rate": 7.448424449526315e-08, "loss": 2.6544, "step": 375460 }, { "epoch": 0.0001792, "grad_norm": 0.8108267188072205, "learning_rate": 7.438617349750932e-08, "loss": 2.6441, "step": 375470 }, { "epoch": 0.0002048, "grad_norm": 0.8448500037193298, "learning_rate": 7.42881668647244e-08, "loss": 2.7975, "step": 375480 }, { "epoch": 0.0002304, "grad_norm": 0.7655691504478455, "learning_rate": 7.419022459754233e-08, "loss": 2.4591, "step": 375490 }, { "epoch": 0.000256, "grad_norm": 0.7961292862892151, "learning_rate": 7.409234669659815e-08, "loss": 2.561, "step": 375500 }, { "epoch": 0.0002816, "grad_norm": 0.7359746694564819, "learning_rate": 7.399453316252803e-08, "loss": 2.5905, "step": 375510 }, { "epoch": 0.0003072, "grad_norm": 0.821397602558136, "learning_rate": 7.389678399596479e-08, "loss": 2.416, "step": 375520 }, { "epoch": 0.0003328, "grad_norm": 0.8466334939002991, "learning_rate": 7.379909919754346e-08, "loss": 2.6726, "step": 375530 }, { "epoch": 0.0003584, "grad_norm": 0.7267674207687378, "learning_rate": 7.37014787678969e-08, "loss": 2.8778, "step": 375540 }, { "epoch": 0.000384, "grad_norm": 0.8272937536239624, "learning_rate": 7.360392270765682e-08, "loss": 2.6452, "step": 375550 }, { "epoch": 0.0004096, "grad_norm": 0.8149482607841492, "learning_rate": 7.350643101745825e-08, "loss": 2.6238, "step": 375560 }, { "epoch": 0.0004352, "grad_norm": 0.8865132927894592, "learning_rate": 7.340900369793292e-08, "loss": 2.7329, "step": 375570 }, { "epoch": 0.0004608, "grad_norm": 0.8736317753791809, "learning_rate": 7.331164074971031e-08, "loss": 2.6593, "step": 375580 }, { "epoch": 0.0004864, "grad_norm": 0.9659318923950195, "learning_rate": 7.321434217342438e-08, "loss": 2.7222, "step": 375590 }, { "epoch": 0.000512, "grad_norm": 0.7865561246871948, "learning_rate": 7.311710796970572e-08, "loss": 2.3738, "step": 375600 }, { "epoch": 0.0005376, "grad_norm": 0.8018705248832703, "learning_rate": 7.301993813918385e-08, "loss": 2.6292, "step": 375610 }, { "epoch": 0.0005632, "grad_norm": 0.807510495185852, "learning_rate": 7.292283268248934e-08, "loss": 2.6537, "step": 375620 }, { "epoch": 0.0005888, "grad_norm": 0.8169693350791931, "learning_rate": 7.282579160025283e-08, "loss": 2.5063, "step": 375630 }, { "epoch": 0.0006144, "grad_norm": 0.7502769231796265, "learning_rate": 7.272881489310158e-08, "loss": 2.6056, "step": 375640 }, { "epoch": 0.00064, "grad_norm": 0.8989804983139038, "learning_rate": 7.263190256166618e-08, "loss": 2.6212, "step": 375650 }, { "epoch": 0.0006656, "grad_norm": 0.8270279765129089, "learning_rate": 7.253505460657506e-08, "loss": 2.7595, "step": 375660 }, { "epoch": 0.0006912, "grad_norm": 0.8039965033531189, "learning_rate": 7.243827102845436e-08, "loss": 2.876, "step": 375670 }, { "epoch": 0.0007168, "grad_norm": 0.9254726767539978, "learning_rate": 7.234155182793467e-08, "loss": 2.6346, "step": 375680 }, { "epoch": 0.0007424, "grad_norm": 0.8597695827484131, "learning_rate": 7.224489700563998e-08, "loss": 2.8572, "step": 375690 }, { "epoch": 0.000768, "grad_norm": 0.7620880603790283, "learning_rate": 7.214830656219974e-08, "loss": 2.4718, "step": 375700 }, { "epoch": 0.0007936, "grad_norm": 0.860954761505127, "learning_rate": 7.205178049823791e-08, "loss": 2.6971, "step": 375710 }, { "epoch": 0.0008192, "grad_norm": 0.8911092877388, "learning_rate": 7.195531881438288e-08, "loss": 2.619, "step": 375720 }, { "epoch": 0.0008448, "grad_norm": 0.8520122170448303, "learning_rate": 7.185892151125973e-08, "loss": 2.6417, "step": 375730 }, { "epoch": 0.0008704, "grad_norm": 0.7915055751800537, "learning_rate": 7.176258858949125e-08, "loss": 2.5353, "step": 375740 }, { "epoch": 0.000896, "grad_norm": 0.9081261157989502, "learning_rate": 7.166632004970475e-08, "loss": 2.4526, "step": 375750 }, { "epoch": 0.0009216, "grad_norm": 0.7843286991119385, "learning_rate": 7.157011589252306e-08, "loss": 2.5342, "step": 375760 }, { "epoch": 0.0009472, "grad_norm": 0.7660489678382874, "learning_rate": 7.14739761185701e-08, "loss": 2.5367, "step": 375770 }, { "epoch": 0.0009728, "grad_norm": 0.9060394167900085, "learning_rate": 7.137790072847095e-08, "loss": 2.6744, "step": 375780 }, { "epoch": 0.0009984, "grad_norm": 0.8771412968635559, "learning_rate": 7.128188972284733e-08, "loss": 2.4101, "step": 375790 }, { "epoch": 0.001024, "grad_norm": 1.0652424097061157, "learning_rate": 7.118594310232096e-08, "loss": 2.7992, "step": 375800 }, { "epoch": 0.0010496, "grad_norm": 0.8407060503959656, "learning_rate": 7.10900608675158e-08, "loss": 2.6137, "step": 375810 }, { "epoch": 0.0010752, "grad_norm": 0.7923121452331543, "learning_rate": 7.099424301905245e-08, "loss": 2.6782, "step": 375820 }, { "epoch": 0.0011008, "grad_norm": 0.8396674990653992, "learning_rate": 7.089848955755263e-08, "loss": 2.7116, "step": 375830 }, { "epoch": 0.0011264, "grad_norm": 0.8536584973335266, "learning_rate": 7.080280048363697e-08, "loss": 2.775, "step": 375840 }, { "epoch": 0.001152, "grad_norm": 0.7349064946174622, "learning_rate": 7.070717579792608e-08, "loss": 2.464, "step": 375850 }, { "epoch": 0.0011776, "grad_norm": 0.8064792156219482, "learning_rate": 7.061161550104167e-08, "loss": 2.6683, "step": 375860 }, { "epoch": 0.0012032, "grad_norm": 0.8958728313446045, "learning_rate": 7.051611959360105e-08, "loss": 2.8325, "step": 375870 }, { "epoch": 0.0012288, "grad_norm": 0.7553789019584656, "learning_rate": 7.042068807622371e-08, "loss": 2.5738, "step": 375880 }, { "epoch": 0.0012544, "grad_norm": 0.8588447570800781, "learning_rate": 7.032532094953026e-08, "loss": 2.5384, "step": 375890 }, { "epoch": 0.00128, "grad_norm": 0.7730627655982971, "learning_rate": 7.0230018214138e-08, "loss": 2.5392, "step": 375900 }, { "epoch": 0.0013056, "grad_norm": 0.8606160283088684, "learning_rate": 7.01347798706653e-08, "loss": 2.5405, "step": 375910 }, { "epoch": 0.0013312, "grad_norm": 0.846876859664917, "learning_rate": 7.003960591972837e-08, "loss": 2.6417, "step": 375920 }, { "epoch": 0.0013568, "grad_norm": 0.7664819955825806, "learning_rate": 6.994449636194666e-08, "loss": 2.5507, "step": 375930 }, { "epoch": 0.0013824, "grad_norm": 0.7742927670478821, "learning_rate": 6.98494511979353e-08, "loss": 2.7059, "step": 375940 }, { "epoch": 0.001408, "grad_norm": 0.8871484994888306, "learning_rate": 6.975447042831152e-08, "loss": 2.6408, "step": 375950 }, { "epoch": 0.0014336, "grad_norm": 0.8858447670936584, "learning_rate": 6.96595540536904e-08, "loss": 2.6784, "step": 375960 }, { "epoch": 0.0014592, "grad_norm": 0.8571767807006836, "learning_rate": 6.956470207468813e-08, "loss": 2.5681, "step": 375970 }, { "epoch": 0.0014848, "grad_norm": 0.8389209508895874, "learning_rate": 6.946991449191975e-08, "loss": 2.3796, "step": 375980 }, { "epoch": 0.0015104, "grad_norm": 1.4089065790176392, "learning_rate": 6.937519130599923e-08, "loss": 1.6484, "step": 375990 }, { "epoch": 0.001536, "grad_norm": 0.8230963945388794, "learning_rate": 6.928053251754164e-08, "loss": 2.6717, "step": 376000 }, { "epoch": 0.0015616, "grad_norm": 0.8177809119224548, "learning_rate": 6.91859381271609e-08, "loss": 2.741, "step": 376010 }, { "epoch": 0.0015872, "grad_norm": 1.0469555854797363, "learning_rate": 6.909140813546989e-08, "loss": 2.9475, "step": 376020 }, { "epoch": 0.0016128, "grad_norm": 0.7847746014595032, "learning_rate": 6.899694254308143e-08, "loss": 2.5222, "step": 376030 }, { "epoch": 0.0016384, "grad_norm": 0.808807909488678, "learning_rate": 6.890254135060837e-08, "loss": 2.5673, "step": 376040 }, { "epoch": 0.001664, "grad_norm": 1.318618655204773, "learning_rate": 6.880820455866355e-08, "loss": 2.8098, "step": 376050 }, { "epoch": 0.0016896, "grad_norm": 0.8421937823295593, "learning_rate": 6.87139321678576e-08, "loss": 2.4779, "step": 376060 }, { "epoch": 0.0017152, "grad_norm": 0.7752150893211365, "learning_rate": 6.861972417880226e-08, "loss": 2.609, "step": 376070 }, { "epoch": 0.0017408, "grad_norm": 0.7487202882766724, "learning_rate": 6.852558059210924e-08, "loss": 2.6468, "step": 376080 }, { "epoch": 0.0017664, "grad_norm": 0.8684262633323669, "learning_rate": 6.843150140838806e-08, "loss": 2.9553, "step": 376090 }, { "epoch": 0.001792, "grad_norm": 1.1509894132614136, "learning_rate": 6.833748662824824e-08, "loss": 2.439, "step": 376100 }, { "epoch": 0.0018176, "grad_norm": 0.8062402606010437, "learning_rate": 6.824353625230152e-08, "loss": 2.6072, "step": 376110 }, { "epoch": 0.0018432, "grad_norm": 0.8338265419006348, "learning_rate": 6.814965028115517e-08, "loss": 2.2983, "step": 376120 }, { "epoch": 0.0018688, "grad_norm": 0.9353280067443848, "learning_rate": 6.805582871541983e-08, "loss": 2.8253, "step": 376130 }, { "epoch": 0.0018944, "grad_norm": 0.8789011240005493, "learning_rate": 6.796207155570167e-08, "loss": 2.654, "step": 376140 }, { "epoch": 0.00192, "grad_norm": 0.8369931578636169, "learning_rate": 6.786837880261022e-08, "loss": 2.6789, "step": 376150 }, { "epoch": 0.0019456, "grad_norm": 0.8131788969039917, "learning_rate": 6.777475045675275e-08, "loss": 2.9563, "step": 376160 }, { "epoch": 0.0019712, "grad_norm": 0.9390164017677307, "learning_rate": 6.768118651873656e-08, "loss": 2.5736, "step": 376170 }, { "epoch": 0.0019968, "grad_norm": 0.8388123512268066, "learning_rate": 6.758768698916785e-08, "loss": 2.827, "step": 376180 }, { "epoch": 0.0020224, "grad_norm": 0.8926137089729309, "learning_rate": 6.749425186865388e-08, "loss": 2.4558, "step": 376190 }, { "epoch": 0.002048, "grad_norm": 0.9404935836791992, "learning_rate": 6.740088115779975e-08, "loss": 2.6318, "step": 376200 }, { "epoch": 0.0020736, "grad_norm": 0.8683228492736816, "learning_rate": 6.730757485721051e-08, "loss": 2.325, "step": 376210 }, { "epoch": 0.0020992, "grad_norm": 0.9210571646690369, "learning_rate": 6.721433296749236e-08, "loss": 2.7686, "step": 376220 }, { "epoch": 0.0021248, "grad_norm": 0.8283714056015015, "learning_rate": 6.712115548924925e-08, "loss": 2.5716, "step": 376230 }, { "epoch": 0.0021504, "grad_norm": 0.8941661715507507, "learning_rate": 6.702804242308514e-08, "loss": 2.7043, "step": 376240 }, { "epoch": 0.002176, "grad_norm": 0.9276767373085022, "learning_rate": 6.693499376960511e-08, "loss": 2.5336, "step": 376250 }, { "epoch": 0.0022016, "grad_norm": 0.8030285239219666, "learning_rate": 6.6842009529412e-08, "loss": 2.7118, "step": 376260 }, { "epoch": 0.0022272, "grad_norm": 0.9004417657852173, "learning_rate": 6.674908970310756e-08, "loss": 2.8717, "step": 376270 }, { "epoch": 0.0022528, "grad_norm": 0.8461447954177856, "learning_rate": 6.665623429129576e-08, "loss": 2.5585, "step": 376280 }, { "epoch": 0.0022784, "grad_norm": 0.8931547403335571, "learning_rate": 6.656344329457832e-08, "loss": 3.0846, "step": 376290 }, { "epoch": 0.002304, "grad_norm": 0.8971548080444336, "learning_rate": 6.647071671355698e-08, "loss": 2.3069, "step": 376300 }, { "epoch": 0.0023296, "grad_norm": 0.9181554913520813, "learning_rate": 6.63780545488335e-08, "loss": 2.3806, "step": 376310 }, { "epoch": 0.0023552, "grad_norm": 0.7799447178840637, "learning_rate": 6.62854568010074e-08, "loss": 2.749, "step": 376320 }, { "epoch": 0.0023808, "grad_norm": 0.9586271047592163, "learning_rate": 6.619292347068041e-08, "loss": 2.7114, "step": 376330 }, { "epoch": 0.0024064, "grad_norm": 0.9212734699249268, "learning_rate": 6.610045455845205e-08, "loss": 2.568, "step": 376340 }, { "epoch": 0.002432, "grad_norm": 1.153509259223938, "learning_rate": 6.600805006492183e-08, "loss": 2.5824, "step": 376350 }, { "epoch": 0.0024576, "grad_norm": 0.8477442264556885, "learning_rate": 6.591570999069042e-08, "loss": 2.7216, "step": 376360 }, { "epoch": 0.0024832, "grad_norm": 0.8366966843605042, "learning_rate": 6.582343433635507e-08, "loss": 2.7074, "step": 376370 }, { "epoch": 0.0025088, "grad_norm": 0.9916548132896423, "learning_rate": 6.573122310251422e-08, "loss": 2.9274, "step": 376380 }, { "epoch": 0.0025344, "grad_norm": 0.905734658241272, "learning_rate": 6.563907628976629e-08, "loss": 2.6678, "step": 376390 }, { "epoch": 0.00256, "grad_norm": 0.8645778894424438, "learning_rate": 6.554699389870856e-08, "loss": 2.7118, "step": 376400 }, { "epoch": 0.0025856, "grad_norm": 0.9701701402664185, "learning_rate": 6.545497592993832e-08, "loss": 2.8492, "step": 376410 }, { "epoch": 0.0026112, "grad_norm": 1.1628906726837158, "learning_rate": 6.536302238405289e-08, "loss": 2.7391, "step": 376420 }, { "epoch": 0.0026368, "grad_norm": 0.8694751262664795, "learning_rate": 6.527113326164736e-08, "loss": 2.8076, "step": 376430 }, { "epoch": 0.0026624, "grad_norm": 0.882166862487793, "learning_rate": 6.51793085633179e-08, "loss": 2.8232, "step": 376440 }, { "epoch": 0.002688, "grad_norm": 0.8467676043510437, "learning_rate": 6.508754828966069e-08, "loss": 2.7691, "step": 376450 }, { "epoch": 0.0027136, "grad_norm": 0.8581069111824036, "learning_rate": 6.499585244127083e-08, "loss": 2.9396, "step": 376460 }, { "epoch": 0.0027392, "grad_norm": 0.9041261672973633, "learning_rate": 6.490422101874116e-08, "loss": 2.8592, "step": 376470 }, { "epoch": 0.0027648, "grad_norm": 0.8569172620773315, "learning_rate": 6.48126540226679e-08, "loss": 2.6717, "step": 376480 }, { "epoch": 0.0027904, "grad_norm": 0.9514691829681396, "learning_rate": 6.472115145364499e-08, "loss": 2.8713, "step": 376490 }, { "epoch": 0.002816, "grad_norm": 0.775825023651123, "learning_rate": 6.462971331226308e-08, "loss": 2.7849, "step": 376500 }, { "epoch": 0.0028416, "grad_norm": 0.8997583985328674, "learning_rate": 6.453833959911948e-08, "loss": 2.9439, "step": 376510 }, { "epoch": 0.0028672, "grad_norm": 1.0990915298461914, "learning_rate": 6.44470303148026e-08, "loss": 2.9439, "step": 376520 }, { "epoch": 0.0028928, "grad_norm": 0.806094765663147, "learning_rate": 6.43557854599064e-08, "loss": 2.8383, "step": 376530 }, { "epoch": 0.0029184, "grad_norm": 0.8995686173439026, "learning_rate": 6.426460503502153e-08, "loss": 2.976, "step": 376540 }, { "epoch": 0.002944, "grad_norm": 0.8560368418693542, "learning_rate": 6.417348904074083e-08, "loss": 2.8163, "step": 376550 }, { "epoch": 0.0029696, "grad_norm": 0.905224621295929, "learning_rate": 6.408243747765497e-08, "loss": 2.6138, "step": 376560 }, { "epoch": 0.0029952, "grad_norm": 0.9774209260940552, "learning_rate": 6.399145034635345e-08, "loss": 2.7058, "step": 376570 }, { "epoch": 0.0030208, "grad_norm": 0.8169979453086853, "learning_rate": 6.390052764742694e-08, "loss": 2.7458, "step": 376580 }, { "epoch": 0.0030464, "grad_norm": 0.9172064065933228, "learning_rate": 6.380966938146382e-08, "loss": 2.8778, "step": 376590 }, { "epoch": 0.003072, "grad_norm": 0.8700689673423767, "learning_rate": 6.371887554905476e-08, "loss": 2.7278, "step": 376600 }, { "epoch": 0.0030976, "grad_norm": 0.8950104117393494, "learning_rate": 6.362814615078816e-08, "loss": 2.9048, "step": 376610 }, { "epoch": 0.0031232, "grad_norm": 0.8368260860443115, "learning_rate": 6.353748118725244e-08, "loss": 2.9138, "step": 376620 }, { "epoch": 0.0031488, "grad_norm": 0.8092358112335205, "learning_rate": 6.344688065903492e-08, "loss": 3.0126, "step": 376630 }, { "epoch": 0.0031744, "grad_norm": 0.8437597751617432, "learning_rate": 6.335634456672401e-08, "loss": 2.8823, "step": 376640 }, { "epoch": 0.0032, "grad_norm": 0.9270833730697632, "learning_rate": 6.32658729109059e-08, "loss": 2.976, "step": 376650 }, { "epoch": 0.0032256, "grad_norm": 0.9998124837875366, "learning_rate": 6.317546569216904e-08, "loss": 2.9886, "step": 376660 }, { "epoch": 0.0032512, "grad_norm": 0.8707380890846252, "learning_rate": 6.308512291109736e-08, "loss": 2.9456, "step": 376670 }, { "epoch": 0.0032768, "grad_norm": 0.9094239473342896, "learning_rate": 6.29948445682771e-08, "loss": 2.7941, "step": 376680 }, { "epoch": 0.0033024, "grad_norm": 0.910287082195282, "learning_rate": 6.290463066429553e-08, "loss": 2.9301, "step": 376690 }, { "epoch": 0.003328, "grad_norm": 0.8852811455726624, "learning_rate": 6.281448119973555e-08, "loss": 2.6541, "step": 376700 }, { "epoch": 0.0033536, "grad_norm": 0.8406276702880859, "learning_rate": 6.272439617518333e-08, "loss": 2.8355, "step": 376710 }, { "epoch": 0.0033792, "grad_norm": 0.9212905764579773, "learning_rate": 6.263437559122176e-08, "loss": 2.5758, "step": 376720 }, { "epoch": 0.0034048, "grad_norm": 2.1263158321380615, "learning_rate": 6.254441944843592e-08, "loss": 2.5027, "step": 376730 }, { "epoch": 0.0034304, "grad_norm": 0.87363600730896, "learning_rate": 6.245452774740868e-08, "loss": 2.9304, "step": 376740 }, { "epoch": 0.003456, "grad_norm": 0.8127173185348511, "learning_rate": 6.236470048872178e-08, "loss": 2.766, "step": 376750 }, { "epoch": 0.0034816, "grad_norm": 0.9417999386787415, "learning_rate": 6.227493767295922e-08, "loss": 2.8669, "step": 376760 }, { "epoch": 0.0035072, "grad_norm": 0.8672205805778503, "learning_rate": 6.218523930070275e-08, "loss": 2.7415, "step": 376770 }, { "epoch": 0.0035328, "grad_norm": 0.8626033067703247, "learning_rate": 6.2095605372533e-08, "loss": 2.8577, "step": 376780 }, { "epoch": 0.0035584, "grad_norm": 0.8898715376853943, "learning_rate": 6.200603588903286e-08, "loss": 2.6654, "step": 376790 }, { "epoch": 0.003584, "grad_norm": 0.8005974292755127, "learning_rate": 6.191653085078186e-08, "loss": 2.7314, "step": 376800 }, { "epoch": 0.0036096, "grad_norm": 0.8502303957939148, "learning_rate": 6.182709025836065e-08, "loss": 2.7483, "step": 376810 }, { "epoch": 0.0036352, "grad_norm": 0.8359679579734802, "learning_rate": 6.173771411235097e-08, "loss": 2.7818, "step": 376820 }, { "epoch": 0.0036608, "grad_norm": 1.0761702060699463, "learning_rate": 6.164840241332904e-08, "loss": 3.0239, "step": 376830 }, { "epoch": 0.0036864, "grad_norm": 0.7707650065422058, "learning_rate": 6.155915516187772e-08, "loss": 2.917, "step": 376840 }, { "epoch": 0.003712, "grad_norm": 0.8313727974891663, "learning_rate": 6.146997235857321e-08, "loss": 2.8026, "step": 376850 }, { "epoch": 0.0037376, "grad_norm": 0.8513354063034058, "learning_rate": 6.138085400399507e-08, "loss": 2.7489, "step": 376860 }, { "epoch": 0.0037632, "grad_norm": 0.8451933264732361, "learning_rate": 6.129180009871949e-08, "loss": 2.8511, "step": 376870 }, { "epoch": 0.0037888, "grad_norm": 0.8854979276657104, "learning_rate": 6.1202810643326e-08, "loss": 2.7641, "step": 376880 }, { "epoch": 0.0038144, "grad_norm": 0.9739828109741211, "learning_rate": 6.111388563839082e-08, "loss": 2.9242, "step": 376890 }, { "epoch": 0.00384, "grad_norm": 0.8480282425880432, "learning_rate": 6.102502508449015e-08, "loss": 2.8421, "step": 376900 }, { "epoch": 0.0038656, "grad_norm": 0.8833736181259155, "learning_rate": 6.093622898220242e-08, "loss": 2.8996, "step": 376910 }, { "epoch": 0.0038912, "grad_norm": 0.882260262966156, "learning_rate": 6.084749733210049e-08, "loss": 2.8106, "step": 376920 }, { "epoch": 0.0039168, "grad_norm": 0.8930439949035645, "learning_rate": 6.075883013476059e-08, "loss": 2.8264, "step": 376930 }, { "epoch": 0.0039424, "grad_norm": 0.8843601942062378, "learning_rate": 6.067022739075889e-08, "loss": 2.7982, "step": 376940 }, { "epoch": 0.003968, "grad_norm": 0.8305221199989319, "learning_rate": 6.058168910066831e-08, "loss": 2.7601, "step": 376950 }, { "epoch": 0.0039936, "grad_norm": 0.8435476422309875, "learning_rate": 6.049321526506502e-08, "loss": 2.7198, "step": 376960 }, { "epoch": 0.0040192, "grad_norm": 0.8065429925918579, "learning_rate": 6.04048058845208e-08, "loss": 2.8081, "step": 376970 }, { "epoch": 0.0040448, "grad_norm": 0.853865921497345, "learning_rate": 6.031646095960963e-08, "loss": 2.8847, "step": 376980 }, { "epoch": 0.0040704, "grad_norm": 0.8903115391731262, "learning_rate": 6.022818049090551e-08, "loss": 3.0328, "step": 376990 }, { "epoch": 0.004096, "grad_norm": 0.8125960230827332, "learning_rate": 6.013996447897908e-08, "loss": 2.8925, "step": 377000 }, { "epoch": 0.0041216, "grad_norm": 0.8500085473060608, "learning_rate": 6.005181292440432e-08, "loss": 2.8284, "step": 377010 }, { "epoch": 0.0041472, "grad_norm": 0.8546403050422668, "learning_rate": 5.996372582775079e-08, "loss": 3.0121, "step": 377020 }, { "epoch": 0.0041728, "grad_norm": 1.2494100332260132, "learning_rate": 5.987570318959135e-08, "loss": 2.8941, "step": 377030 }, { "epoch": 0.0041984, "grad_norm": 0.8267151117324829, "learning_rate": 5.978774501049666e-08, "loss": 2.7463, "step": 377040 }, { "epoch": 0.004224, "grad_norm": 0.8503267765045166, "learning_rate": 5.969985129103517e-08, "loss": 2.8643, "step": 377050 }, { "epoch": 0.0042496, "grad_norm": 0.8511139154434204, "learning_rate": 5.961202203177974e-08, "loss": 2.9307, "step": 377060 }, { "epoch": 0.0042752, "grad_norm": 0.9601196646690369, "learning_rate": 5.9524257233297686e-08, "loss": 2.9632, "step": 377070 }, { "epoch": 0.0043008, "grad_norm": 0.8739250302314758, "learning_rate": 5.943655689615857e-08, "loss": 2.8103, "step": 377080 }, { "epoch": 0.0043264, "grad_norm": 0.8126674890518188, "learning_rate": 5.934892102093304e-08, "loss": 2.8375, "step": 377090 }, { "epoch": 0.004352, "grad_norm": 0.8398178815841675, "learning_rate": 5.926134960818619e-08, "loss": 2.9277, "step": 377100 }, { "epoch": 0.0043776, "grad_norm": 0.9573043584823608, "learning_rate": 5.917384265848869e-08, "loss": 2.8563, "step": 377110 }, { "epoch": 0.0044032, "grad_norm": 0.8364206552505493, "learning_rate": 5.9086400172406746e-08, "loss": 2.9986, "step": 377120 }, { "epoch": 0.0044288, "grad_norm": 1.0754692554473877, "learning_rate": 5.899902215050768e-08, "loss": 2.9221, "step": 377130 }, { "epoch": 0.0044544, "grad_norm": 1.1444947719573975, "learning_rate": 5.8911708593356596e-08, "loss": 2.9637, "step": 377140 }, { "epoch": 0.00448, "grad_norm": 0.8524772524833679, "learning_rate": 5.8824459501521937e-08, "loss": 2.5863, "step": 377150 }, { "epoch": 0.0045056, "grad_norm": 0.9730765223503113, "learning_rate": 5.873727487556879e-08, "loss": 2.8145, "step": 377160 }, { "epoch": 0.0045312, "grad_norm": 0.9811540842056274, "learning_rate": 5.865015471606228e-08, "loss": 2.0341, "step": 377170 }, { "epoch": 0.0045568, "grad_norm": 0.9210930466651917, "learning_rate": 5.8563099023567495e-08, "loss": 2.8007, "step": 377180 }, { "epoch": 0.0045824, "grad_norm": 1.4366099834442139, "learning_rate": 5.847610779864843e-08, "loss": 2.955, "step": 377190 }, { "epoch": 0.004608, "grad_norm": 0.8630909323692322, "learning_rate": 5.838918104187019e-08, "loss": 2.8774, "step": 377200 }, { "epoch": 0.0046336, "grad_norm": 0.854383111000061, "learning_rate": 5.830231875379566e-08, "loss": 2.7704, "step": 377210 }, { "epoch": 0.0046592, "grad_norm": 0.9510955214500427, "learning_rate": 5.821552093498883e-08, "loss": 2.9494, "step": 377220 }, { "epoch": 0.0046848, "grad_norm": 1.2603744268417358, "learning_rate": 5.812878758601148e-08, "loss": 2.8559, "step": 377230 }, { "epoch": 0.0047104, "grad_norm": 0.9130731821060181, "learning_rate": 5.80421187074276e-08, "loss": 2.9072, "step": 377240 }, { "epoch": 0.004736, "grad_norm": 0.8712113499641418, "learning_rate": 5.7955514299797846e-08, "loss": 2.7721, "step": 377250 }, { "epoch": 0.0047616, "grad_norm": 0.8359620571136475, "learning_rate": 5.7868974363684e-08, "loss": 2.9708, "step": 377260 }, { "epoch": 0.0047872, "grad_norm": 1.0717272758483887, "learning_rate": 5.778249889964782e-08, "loss": 2.9267, "step": 377270 }, { "epoch": 0.0048128, "grad_norm": 0.8322235941886902, "learning_rate": 5.7696087908249984e-08, "loss": 2.767, "step": 377280 }, { "epoch": 0.0048384, "grad_norm": 1.0502145290374756, "learning_rate": 5.760974139005005e-08, "loss": 2.7945, "step": 377290 }, { "epoch": 0.004864, "grad_norm": 0.9065102934837341, "learning_rate": 5.7523459345608655e-08, "loss": 2.741, "step": 377300 }, { "epoch": 0.0048896, "grad_norm": 0.8414976596832275, "learning_rate": 5.7437241775486484e-08, "loss": 2.8611, "step": 377310 }, { "epoch": 0.0049152, "grad_norm": 0.9009999632835388, "learning_rate": 5.7351088680239756e-08, "loss": 2.8525, "step": 377320 }, { "epoch": 0.0049408, "grad_norm": 1.0081734657287598, "learning_rate": 5.726500006042912e-08, "loss": 2.6015, "step": 377330 }, { "epoch": 0.0049664, "grad_norm": 0.9078567028045654, "learning_rate": 5.717897591661303e-08, "loss": 2.6229, "step": 377340 }, { "epoch": 0.004992, "grad_norm": 0.844016432762146, "learning_rate": 5.7093016249348823e-08, "loss": 2.7703, "step": 377350 }, { "epoch": 0.0050176, "grad_norm": 1.0823613405227661, "learning_rate": 5.7007121059193817e-08, "loss": 2.6621, "step": 377360 }, { "epoch": 0.0050432, "grad_norm": 0.9026087522506714, "learning_rate": 5.6921290346705346e-08, "loss": 2.8148, "step": 377370 }, { "epoch": 0.0050688, "grad_norm": 1.012222409248352, "learning_rate": 5.683552411243964e-08, "loss": 2.644, "step": 377380 }, { "epoch": 0.0050944, "grad_norm": 0.8580283522605896, "learning_rate": 5.674982235695181e-08, "loss": 2.8176, "step": 377390 }, { "epoch": 0.00512, "grad_norm": 0.8836396932601929, "learning_rate": 5.666418508080029e-08, "loss": 3.0149, "step": 377400 }, { "epoch": 0.0051456, "grad_norm": 0.9985923767089844, "learning_rate": 5.657861228453798e-08, "loss": 2.8185, "step": 377410 }, { "epoch": 0.0051712, "grad_norm": 0.8159706592559814, "learning_rate": 5.64931039687211e-08, "loss": 2.7442, "step": 377420 }, { "epoch": 0.0051968, "grad_norm": 0.9497767686843872, "learning_rate": 5.640766013390364e-08, "loss": 2.815, "step": 377430 }, { "epoch": 0.0052224, "grad_norm": 0.8507105708122253, "learning_rate": 5.632228078064073e-08, "loss": 2.714, "step": 377440 }, { "epoch": 0.005248, "grad_norm": 0.9292636513710022, "learning_rate": 5.623696590948413e-08, "loss": 2.9367, "step": 377450 }, { "epoch": 0.0052736, "grad_norm": 0.8808985352516174, "learning_rate": 5.615171552098786e-08, "loss": 2.8727, "step": 377460 }, { "epoch": 0.0052992, "grad_norm": 0.8421841263771057, "learning_rate": 5.6066529615705914e-08, "loss": 2.6821, "step": 377470 }, { "epoch": 0.0053248, "grad_norm": 0.7438357472419739, "learning_rate": 5.5981408194188955e-08, "loss": 2.827, "step": 377480 }, { "epoch": 0.0053504, "grad_norm": 0.9046623706817627, "learning_rate": 5.589635125698989e-08, "loss": 2.668, "step": 377490 }, { "epoch": 0.005376, "grad_norm": 0.8822310566902161, "learning_rate": 5.5811358804660486e-08, "loss": 2.7875, "step": 377500 }, { "epoch": 0.0054016, "grad_norm": 0.7835496664047241, "learning_rate": 5.572643083775031e-08, "loss": 2.5154, "step": 377510 }, { "epoch": 0.0054272, "grad_norm": 0.9241707921028137, "learning_rate": 5.5641567356813364e-08, "loss": 2.817, "step": 377520 }, { "epoch": 0.0054528, "grad_norm": 0.8614783883094788, "learning_rate": 5.555676836239698e-08, "loss": 2.8212, "step": 377530 }, { "epoch": 0.0054784, "grad_norm": 0.9000792503356934, "learning_rate": 5.5472033855051844e-08, "loss": 2.793, "step": 377540 }, { "epoch": 0.005504, "grad_norm": 0.9806340336799622, "learning_rate": 5.53873638353275e-08, "loss": 2.5717, "step": 377550 }, { "epoch": 0.0055296, "grad_norm": 0.7846813201904297, "learning_rate": 5.530275830377352e-08, "loss": 2.6146, "step": 377560 }, { "epoch": 0.0055552, "grad_norm": 0.8352458477020264, "learning_rate": 5.5218217260938344e-08, "loss": 2.6118, "step": 377570 }, { "epoch": 0.0055808, "grad_norm": 0.7862673997879028, "learning_rate": 5.513374070736932e-08, "loss": 2.7219, "step": 377580 }, { "epoch": 0.0056064, "grad_norm": 0.9213528633117676, "learning_rate": 5.5049328643616004e-08, "loss": 2.8209, "step": 377590 }, { "epoch": 0.005632, "grad_norm": 0.8240547180175781, "learning_rate": 5.4964981070223525e-08, "loss": 2.8301, "step": 377600 }, { "epoch": 2.56e-05, "grad_norm": 0.87729412317276, "learning_rate": 5.4880697987741425e-08, "loss": 1.9017, "step": 377610 }, { "epoch": 5.12e-05, "grad_norm": 0.8136734962463379, "learning_rate": 5.479647939671373e-08, "loss": 2.8279, "step": 377620 }, { "epoch": 7.68e-05, "grad_norm": 0.8482144474983215, "learning_rate": 5.471232529768777e-08, "loss": 2.6868, "step": 377630 }, { "epoch": 0.0001024, "grad_norm": 0.8152199983596802, "learning_rate": 5.46282356912109e-08, "loss": 2.8081, "step": 377640 }, { "epoch": 0.000128, "grad_norm": 0.9251822233200073, "learning_rate": 5.454421057782488e-08, "loss": 2.6043, "step": 377650 }, { "epoch": 0.0001536, "grad_norm": 0.8193190693855286, "learning_rate": 5.446024995807708e-08, "loss": 2.8126, "step": 377660 }, { "epoch": 0.0001792, "grad_norm": 0.8150245547294617, "learning_rate": 5.4376353832511494e-08, "loss": 2.7202, "step": 377670 }, { "epoch": 0.0002048, "grad_norm": 0.9465740323066711, "learning_rate": 5.429252220167103e-08, "loss": 2.6945, "step": 377680 }, { "epoch": 0.0002304, "grad_norm": 0.8484441637992859, "learning_rate": 5.4208755066101906e-08, "loss": 2.6646, "step": 377690 }, { "epoch": 0.000256, "grad_norm": 0.7851911783218384, "learning_rate": 5.41250524263448e-08, "loss": 2.5343, "step": 377700 }, { "epoch": 0.0002816, "grad_norm": 0.9357039332389832, "learning_rate": 5.404141428294374e-08, "loss": 2.5815, "step": 377710 }, { "epoch": 0.0003072, "grad_norm": 2.1005899906158447, "learning_rate": 5.3957840636440494e-08, "loss": 2.5538, "step": 377720 }, { "epoch": 0.0003328, "grad_norm": 0.8391183018684387, "learning_rate": 5.387433148737797e-08, "loss": 2.7905, "step": 377730 }, { "epoch": 0.0003584, "grad_norm": 0.8389845490455627, "learning_rate": 5.379088683629685e-08, "loss": 2.4989, "step": 377740 }, { "epoch": 0.000384, "grad_norm": 0.8713830709457397, "learning_rate": 5.370750668373781e-08, "loss": 2.4796, "step": 377750 }, { "epoch": 0.0004096, "grad_norm": 0.7707020044326782, "learning_rate": 5.362419103024263e-08, "loss": 2.479, "step": 377760 }, { "epoch": 0.0004352, "grad_norm": 0.7758143544197083, "learning_rate": 5.3540939876350895e-08, "loss": 2.782, "step": 377770 }, { "epoch": 0.0004608, "grad_norm": 0.9074452519416809, "learning_rate": 5.345775322260327e-08, "loss": 2.589, "step": 377780 }, { "epoch": 0.0004864, "grad_norm": 0.8023529052734375, "learning_rate": 5.337463106953822e-08, "loss": 2.6543, "step": 377790 }, { "epoch": 0.000512, "grad_norm": 0.8650569915771484, "learning_rate": 5.329157341769531e-08, "loss": 2.5319, "step": 377800 }, { "epoch": 0.0005376, "grad_norm": 0.7839244604110718, "learning_rate": 5.3208580267612994e-08, "loss": 2.5108, "step": 377810 }, { "epoch": 0.0005632, "grad_norm": 0.7636429667472839, "learning_rate": 5.3125651619828634e-08, "loss": 2.5386, "step": 377820 }, { "epoch": 0.0005888, "grad_norm": 0.8003979921340942, "learning_rate": 5.3042787474881785e-08, "loss": 2.7198, "step": 377830 }, { "epoch": 0.0006144, "grad_norm": 0.942969024181366, "learning_rate": 5.295998783330869e-08, "loss": 2.6642, "step": 377840 }, { "epoch": 0.00064, "grad_norm": 0.8995028138160706, "learning_rate": 5.28772526956467e-08, "loss": 2.441, "step": 377850 }, { "epoch": 0.0006656, "grad_norm": 0.8212065696716309, "learning_rate": 5.279458206243204e-08, "loss": 2.5553, "step": 377860 }, { "epoch": 0.0006912, "grad_norm": 0.8829169869422913, "learning_rate": 5.2711975934200967e-08, "loss": 2.5011, "step": 377870 }, { "epoch": 0.0007168, "grad_norm": 0.8088458776473999, "learning_rate": 5.262943431148859e-08, "loss": 2.6674, "step": 377880 }, { "epoch": 0.0007424, "grad_norm": 0.7763382196426392, "learning_rate": 5.254695719483116e-08, "loss": 2.6474, "step": 377890 }, { "epoch": 0.000768, "grad_norm": 1.006635308265686, "learning_rate": 5.246454458476269e-08, "loss": 2.7219, "step": 377900 }, { "epoch": 0.0007936, "grad_norm": 1.5201293230056763, "learning_rate": 5.23821964818183e-08, "loss": 2.6844, "step": 377910 }, { "epoch": 0.0008192, "grad_norm": 0.8494547605514526, "learning_rate": 5.229991288653091e-08, "loss": 2.6066, "step": 377920 }, { "epoch": 0.0008448, "grad_norm": 1.0839639902114868, "learning_rate": 5.221769379943564e-08, "loss": 2.7864, "step": 377930 }, { "epoch": 0.0008704, "grad_norm": 0.8872191905975342, "learning_rate": 5.213553922106429e-08, "loss": 2.5109, "step": 377940 }, { "epoch": 0.000896, "grad_norm": 0.869896650314331, "learning_rate": 5.205344915194977e-08, "loss": 2.274, "step": 377950 }, { "epoch": 0.0009216, "grad_norm": 0.8534374237060547, "learning_rate": 5.197142359262608e-08, "loss": 2.6009, "step": 377960 }, { "epoch": 0.0009472, "grad_norm": 0.7682996392250061, "learning_rate": 5.188946254362282e-08, "loss": 2.6168, "step": 377970 }, { "epoch": 0.0009728, "grad_norm": 0.8441111445426941, "learning_rate": 5.1807566005471764e-08, "loss": 2.6064, "step": 377980 }, { "epoch": 0.0009984, "grad_norm": 0.9768467545509338, "learning_rate": 5.172573397870584e-08, "loss": 2.8349, "step": 377990 }, { "epoch": 0.001024, "grad_norm": 0.9121897220611572, "learning_rate": 5.164396646385461e-08, "loss": 2.7257, "step": 378000 }, { "epoch": 0.0010496, "grad_norm": 0.8803273439407349, "learning_rate": 5.156226346144766e-08, "loss": 2.5459, "step": 378010 }, { "epoch": 0.0010752, "grad_norm": 0.8632107377052307, "learning_rate": 5.148062497201567e-08, "loss": 2.866, "step": 378020 }, { "epoch": 0.0011008, "grad_norm": 0.819578230381012, "learning_rate": 5.139905099608822e-08, "loss": 2.6368, "step": 378030 }, { "epoch": 0.0011264, "grad_norm": 0.8171787261962891, "learning_rate": 5.131754153419377e-08, "loss": 2.8273, "step": 378040 }, { "epoch": 0.001152, "grad_norm": 0.818553626537323, "learning_rate": 5.1236096586860796e-08, "loss": 2.6109, "step": 378050 }, { "epoch": 0.0011776, "grad_norm": 0.9955747723579407, "learning_rate": 5.115471615461776e-08, "loss": 2.6322, "step": 378060 }, { "epoch": 0.0012032, "grad_norm": 0.7951399087905884, "learning_rate": 5.107340023799201e-08, "loss": 2.8502, "step": 378070 }, { "epoch": 0.0012288, "grad_norm": 0.8225335478782654, "learning_rate": 5.0992148837512024e-08, "loss": 2.6321, "step": 378080 }, { "epoch": 0.0012544, "grad_norm": 0.8703746199607849, "learning_rate": 5.0910961953704044e-08, "loss": 2.8637, "step": 378090 }, { "epoch": 0.00128, "grad_norm": 0.7816188931465149, "learning_rate": 5.08298395870932e-08, "loss": 2.5312, "step": 378100 }, { "epoch": 0.0013056, "grad_norm": 0.7889348864555359, "learning_rate": 5.0748781738206855e-08, "loss": 2.8661, "step": 378110 }, { "epoch": 0.0013312, "grad_norm": 0.8412631750106812, "learning_rate": 5.066778840757125e-08, "loss": 2.6902, "step": 378120 }, { "epoch": 0.0013568, "grad_norm": 1.1634224653244019, "learning_rate": 5.0586859595709305e-08, "loss": 2.5789, "step": 378130 }, { "epoch": 0.0013824, "grad_norm": 0.8597590327262878, "learning_rate": 5.050599530314837e-08, "loss": 2.7022, "step": 378140 }, { "epoch": 0.001408, "grad_norm": 0.7458530068397522, "learning_rate": 5.0425195530411366e-08, "loss": 2.6272, "step": 378150 }, { "epoch": 0.0014336, "grad_norm": 0.8867030739784241, "learning_rate": 5.034446027802342e-08, "loss": 2.7822, "step": 378160 }, { "epoch": 0.0014592, "grad_norm": 0.8984004855155945, "learning_rate": 5.026378954650746e-08, "loss": 2.6854, "step": 378170 }, { "epoch": 0.0014848, "grad_norm": 0.8021774888038635, "learning_rate": 5.0183183336386385e-08, "loss": 2.5062, "step": 378180 }, { "epoch": 0.0015104, "grad_norm": 1.5870760679244995, "learning_rate": 5.0102641648182014e-08, "loss": 1.7085, "step": 378190 }, { "epoch": 0.001536, "grad_norm": 0.8749939799308777, "learning_rate": 5.0022164482418366e-08, "loss": 2.6466, "step": 378200 }, { "epoch": 2.56e-05, "grad_norm": 0.8401899933815002, "learning_rate": 4.9941751839616136e-08, "loss": 1.9421, "step": 378210 }, { "epoch": 5.12e-05, "grad_norm": 0.9041339159011841, "learning_rate": 4.986140372029824e-08, "loss": 2.6538, "step": 378220 }, { "epoch": 7.68e-05, "grad_norm": 0.8801757097244263, "learning_rate": 4.978112012498315e-08, "loss": 2.6187, "step": 378230 }, { "epoch": 0.0001024, "grad_norm": 0.7768009305000305, "learning_rate": 4.9700901054194896e-08, "loss": 2.4557, "step": 378240 }, { "epoch": 0.000128, "grad_norm": 0.8452238440513611, "learning_rate": 4.962074650845083e-08, "loss": 2.5764, "step": 378250 }, { "epoch": 0.0001536, "grad_norm": 0.7923588752746582, "learning_rate": 4.954065648827167e-08, "loss": 2.6402, "step": 378260 }, { "epoch": 0.0001792, "grad_norm": 0.7792333960533142, "learning_rate": 4.946063099417697e-08, "loss": 2.7514, "step": 378270 }, { "epoch": 0.0002048, "grad_norm": 0.8077211976051331, "learning_rate": 4.938067002668523e-08, "loss": 2.7095, "step": 378280 }, { "epoch": 0.0002304, "grad_norm": 0.8267549872398376, "learning_rate": 4.930077358631602e-08, "loss": 2.5391, "step": 378290 }, { "epoch": 0.000256, "grad_norm": 0.8299200534820557, "learning_rate": 4.92209416735856e-08, "loss": 2.6115, "step": 378300 }, { "epoch": 0.0002816, "grad_norm": 0.7754981517791748, "learning_rate": 4.914117428901355e-08, "loss": 2.3762, "step": 378310 }, { "epoch": 0.0003072, "grad_norm": 0.8116067051887512, "learning_rate": 4.906147143311612e-08, "loss": 2.5345, "step": 378320 }, { "epoch": 0.0003328, "grad_norm": 0.9240599274635315, "learning_rate": 4.898183310641069e-08, "loss": 2.3986, "step": 378330 }, { "epoch": 0.0003584, "grad_norm": 1.0194562673568726, "learning_rate": 4.89022593094135e-08, "loss": 2.8477, "step": 378340 }, { "epoch": 0.000384, "grad_norm": 0.8069765567779541, "learning_rate": 4.88227500426397e-08, "loss": 2.708, "step": 378350 }, { "epoch": 0.0004096, "grad_norm": 0.8287976384162903, "learning_rate": 4.874330530660665e-08, "loss": 2.6253, "step": 378360 }, { "epoch": 0.0004352, "grad_norm": 0.8056721091270447, "learning_rate": 4.866392510182838e-08, "loss": 2.7248, "step": 378370 }, { "epoch": 0.0004608, "grad_norm": 0.825307309627533, "learning_rate": 4.858460942881893e-08, "loss": 2.6991, "step": 378380 }, { "epoch": 0.0004864, "grad_norm": 0.973093569278717, "learning_rate": 4.850535828809455e-08, "loss": 2.4633, "step": 378390 }, { "epoch": 0.000512, "grad_norm": 0.8766486048698425, "learning_rate": 4.842617168016706e-08, "loss": 2.6546, "step": 378400 }, { "epoch": 0.0005376, "grad_norm": 0.8810821771621704, "learning_rate": 4.8347049605551585e-08, "loss": 2.684, "step": 378410 }, { "epoch": 0.0005632, "grad_norm": 0.8669642806053162, "learning_rate": 4.826799206476107e-08, "loss": 2.6552, "step": 378420 }, { "epoch": 0.0005888, "grad_norm": 0.810005784034729, "learning_rate": 4.818899905830732e-08, "loss": 2.3513, "step": 378430 }, { "epoch": 0.0006144, "grad_norm": 0.8455251455307007, "learning_rate": 4.811007058670325e-08, "loss": 2.5452, "step": 378440 }, { "epoch": 0.00064, "grad_norm": 0.8657501935958862, "learning_rate": 4.803120665046068e-08, "loss": 2.5891, "step": 378450 }, { "epoch": 0.0006656, "grad_norm": 0.8599207997322083, "learning_rate": 4.7952407250090315e-08, "loss": 2.6772, "step": 378460 }, { "epoch": 0.0006912, "grad_norm": 0.8607594966888428, "learning_rate": 4.7873672386103965e-08, "loss": 2.9403, "step": 378470 }, { "epoch": 0.0007168, "grad_norm": 0.7258589267730713, "learning_rate": 4.7795002059011217e-08, "loss": 2.6186, "step": 378480 }, { "epoch": 0.0007424, "grad_norm": 0.9152268767356873, "learning_rate": 4.7716396269323895e-08, "loss": 2.6213, "step": 378490 }, { "epoch": 0.000768, "grad_norm": 0.8949560523033142, "learning_rate": 4.763785501755047e-08, "loss": 2.6158, "step": 378500 }, { "epoch": 0.0007936, "grad_norm": 0.7588551044464111, "learning_rate": 4.7559378304200545e-08, "loss": 2.7828, "step": 378510 }, { "epoch": 0.0008192, "grad_norm": 0.8444768786430359, "learning_rate": 4.748096612978259e-08, "loss": 2.7591, "step": 378520 }, { "epoch": 0.0008448, "grad_norm": 0.860038161277771, "learning_rate": 4.740261849480621e-08, "loss": 2.5383, "step": 378530 }, { "epoch": 0.0008704, "grad_norm": 0.8754494190216064, "learning_rate": 4.732433539977765e-08, "loss": 2.5444, "step": 378540 }, { "epoch": 0.000896, "grad_norm": 0.9302913546562195, "learning_rate": 4.724611684520652e-08, "loss": 2.6151, "step": 378550 }, { "epoch": 0.0009216, "grad_norm": 1.520451307296753, "learning_rate": 4.716796283159908e-08, "loss": 2.3756, "step": 378560 }, { "epoch": 0.0009472, "grad_norm": 1.0440362691879272, "learning_rate": 4.708987335946269e-08, "loss": 2.7543, "step": 378570 }, { "epoch": 0.0009728, "grad_norm": 0.8055311441421509, "learning_rate": 4.701184842930251e-08, "loss": 2.6684, "step": 378580 }, { "epoch": 0.0009984, "grad_norm": 0.803052544593811, "learning_rate": 4.6933888041625905e-08, "loss": 2.5473, "step": 378590 }, { "epoch": 0.001024, "grad_norm": 0.7516932487487793, "learning_rate": 4.685599219693693e-08, "loss": 2.5441, "step": 378600 }, { "epoch": 0.0010496, "grad_norm": 0.8211978077888489, "learning_rate": 4.677816089574183e-08, "loss": 2.6726, "step": 378610 }, { "epoch": 0.0010752, "grad_norm": 0.8637049794197083, "learning_rate": 4.670039413854688e-08, "loss": 2.9238, "step": 378620 }, { "epoch": 0.0011008, "grad_norm": 0.7870285511016846, "learning_rate": 4.662269192585278e-08, "loss": 2.5419, "step": 378630 }, { "epoch": 0.0011264, "grad_norm": 0.752808690071106, "learning_rate": 4.6545054258165804e-08, "loss": 2.4908, "step": 378640 }, { "epoch": 0.001152, "grad_norm": 0.7904641032218933, "learning_rate": 4.646748113598887e-08, "loss": 2.6307, "step": 378650 }, { "epoch": 0.0011776, "grad_norm": 0.8958570957183838, "learning_rate": 4.6389972559823806e-08, "loss": 2.7694, "step": 378660 }, { "epoch": 0.0012032, "grad_norm": 0.8553005456924438, "learning_rate": 4.6312528530175757e-08, "loss": 2.652, "step": 378670 }, { "epoch": 0.0012288, "grad_norm": 0.854394793510437, "learning_rate": 4.623514904754545e-08, "loss": 2.7149, "step": 378680 }, { "epoch": 0.0012544, "grad_norm": 0.8993708491325378, "learning_rate": 4.6157834112434686e-08, "loss": 2.7289, "step": 378690 }, { "epoch": 0.00128, "grad_norm": 0.8081266283988953, "learning_rate": 4.60805837253453e-08, "loss": 2.4467, "step": 378700 }, { "epoch": 0.0013056, "grad_norm": 0.8880365490913391, "learning_rate": 4.600339788677688e-08, "loss": 2.6546, "step": 378710 }, { "epoch": 0.0013312, "grad_norm": 0.8236908912658691, "learning_rate": 4.592627659723126e-08, "loss": 2.6668, "step": 378720 }, { "epoch": 0.0013568, "grad_norm": 1.1436469554901123, "learning_rate": 4.584921985720914e-08, "loss": 2.8693, "step": 378730 }, { "epoch": 0.0013824, "grad_norm": 0.9034834504127502, "learning_rate": 4.57722276672079e-08, "loss": 2.5546, "step": 378740 }, { "epoch": 0.001408, "grad_norm": 0.7847341895103455, "learning_rate": 4.5695300027729373e-08, "loss": 2.4489, "step": 378750 }, { "epoch": 0.0014336, "grad_norm": 0.9345132112503052, "learning_rate": 4.5618436939272035e-08, "loss": 2.7293, "step": 378760 }, { "epoch": 0.0014592, "grad_norm": 0.7953652143478394, "learning_rate": 4.554163840233217e-08, "loss": 2.577, "step": 378770 }, { "epoch": 0.0014848, "grad_norm": 0.7743233442306519, "learning_rate": 4.546490441740936e-08, "loss": 2.6065, "step": 378780 }, { "epoch": 0.0015104, "grad_norm": 1.1464532613754272, "learning_rate": 4.5388234985001e-08, "loss": 2.0663, "step": 378790 }, { "epoch": 0.001536, "grad_norm": 0.8077074289321899, "learning_rate": 4.5311630105604465e-08, "loss": 2.5959, "step": 378800 }, { "epoch": 2.56e-05, "grad_norm": 1.7904771566390991, "learning_rate": 4.5235089779717134e-08, "loss": 1.2207, "step": 378810 }, { "epoch": 5.12e-05, "grad_norm": 0.8205316066741943, "learning_rate": 4.5158614007834165e-08, "loss": 2.5195, "step": 378820 }, { "epoch": 7.68e-05, "grad_norm": 0.83881014585495, "learning_rate": 4.508220279045183e-08, "loss": 2.7757, "step": 378830 }, { "epoch": 0.0001024, "grad_norm": 1.0201424360275269, "learning_rate": 4.5005856128065293e-08, "loss": 2.4351, "step": 378840 }, { "epoch": 0.000128, "grad_norm": 0.9009063839912415, "learning_rate": 4.492957402117082e-08, "loss": 2.585, "step": 378850 }, { "epoch": 0.0001536, "grad_norm": 0.8198914527893066, "learning_rate": 4.485335647026134e-08, "loss": 2.6765, "step": 378860 }, { "epoch": 0.0001792, "grad_norm": 0.8144790530204773, "learning_rate": 4.4777203475832034e-08, "loss": 2.6044, "step": 378870 }, { "epoch": 0.0002048, "grad_norm": 0.8881475329399109, "learning_rate": 4.470111503837804e-08, "loss": 2.6717, "step": 378880 }, { "epoch": 0.0002304, "grad_norm": 0.8515272736549377, "learning_rate": 4.462509115839009e-08, "loss": 2.6181, "step": 378890 }, { "epoch": 0.000256, "grad_norm": 0.7956646084785461, "learning_rate": 4.454913183636333e-08, "loss": 2.6738, "step": 378900 }, { "epoch": 0.0002816, "grad_norm": 0.7963961362838745, "learning_rate": 4.447323707278961e-08, "loss": 2.4941, "step": 378910 }, { "epoch": 0.0003072, "grad_norm": 0.7999107837677002, "learning_rate": 4.4397406868161854e-08, "loss": 2.4937, "step": 378920 }, { "epoch": 0.0003328, "grad_norm": 1.143550157546997, "learning_rate": 4.432164122296967e-08, "loss": 2.064, "step": 378930 }, { "epoch": 0.0003584, "grad_norm": 0.980606198310852, "learning_rate": 4.4245940137706e-08, "loss": 2.8252, "step": 378940 }, { "epoch": 0.000384, "grad_norm": 0.8351773619651794, "learning_rate": 4.4170303612862674e-08, "loss": 2.8389, "step": 378950 }, { "epoch": 0.0004096, "grad_norm": 1.0122052431106567, "learning_rate": 4.409473164892819e-08, "loss": 2.5286, "step": 378960 }, { "epoch": 0.0004352, "grad_norm": 0.8039798140525818, "learning_rate": 4.4019224246393266e-08, "loss": 2.8551, "step": 378970 }, { "epoch": 0.0004608, "grad_norm": 0.9394060969352722, "learning_rate": 4.394378140574751e-08, "loss": 2.4486, "step": 378980 }, { "epoch": 0.0004864, "grad_norm": 0.7750133872032166, "learning_rate": 4.386840312748053e-08, "loss": 2.4844, "step": 378990 }, { "epoch": 0.000512, "grad_norm": 0.8625848889350891, "learning_rate": 4.3793089412080826e-08, "loss": 2.5916, "step": 379000 }, { "epoch": 0.0005376, "grad_norm": 0.8578479886054993, "learning_rate": 4.3717840260036895e-08, "loss": 2.5973, "step": 379010 }, { "epoch": 0.0005632, "grad_norm": 0.8669312596321106, "learning_rate": 4.3642655671837233e-08, "loss": 2.5959, "step": 379020 }, { "epoch": 0.0005888, "grad_norm": 0.9268216490745544, "learning_rate": 4.356753564796923e-08, "loss": 2.2541, "step": 379030 }, { "epoch": 0.0006144, "grad_norm": 0.8992246985435486, "learning_rate": 4.349248018891805e-08, "loss": 2.4945, "step": 379040 }, { "epoch": 0.00064, "grad_norm": 0.7869786024093628, "learning_rate": 4.341748929517331e-08, "loss": 2.7054, "step": 379050 }, { "epoch": 0.0006656, "grad_norm": 0.876020610332489, "learning_rate": 4.334256296721906e-08, "loss": 2.7361, "step": 379060 }, { "epoch": 0.0006912, "grad_norm": 0.8853282928466797, "learning_rate": 4.326770120554269e-08, "loss": 2.9952, "step": 379070 }, { "epoch": 0.0007168, "grad_norm": 0.7947571873664856, "learning_rate": 4.319290401062937e-08, "loss": 2.6752, "step": 379080 }, { "epoch": 0.0007424, "grad_norm": 0.8923047780990601, "learning_rate": 4.3118171382963144e-08, "loss": 2.6986, "step": 379090 }, { "epoch": 0.000768, "grad_norm": 0.7680230140686035, "learning_rate": 4.304350332303031e-08, "loss": 2.5029, "step": 379100 }, { "epoch": 0.0007936, "grad_norm": 0.9048081040382385, "learning_rate": 4.2968899831313806e-08, "loss": 2.545, "step": 379110 }, { "epoch": 0.0008192, "grad_norm": 0.8290961980819702, "learning_rate": 4.289436090829657e-08, "loss": 2.726, "step": 379120 }, { "epoch": 0.0008448, "grad_norm": 0.8025592565536499, "learning_rate": 4.2819886554464894e-08, "loss": 2.5546, "step": 379130 }, { "epoch": 0.0008704, "grad_norm": 0.8141307830810547, "learning_rate": 4.274547677029839e-08, "loss": 2.5392, "step": 379140 }, { "epoch": 0.000896, "grad_norm": 0.8836378455162048, "learning_rate": 4.267113155628222e-08, "loss": 2.5638, "step": 379150 }, { "epoch": 0.0009216, "grad_norm": 0.9557872414588928, "learning_rate": 4.259685091289712e-08, "loss": 2.4764, "step": 379160 }, { "epoch": 0.0009472, "grad_norm": 0.812919557094574, "learning_rate": 4.252263484062491e-08, "loss": 2.6935, "step": 379170 }, { "epoch": 0.0009728, "grad_norm": 0.7870222330093384, "learning_rate": 4.2448483339947444e-08, "loss": 2.5861, "step": 379180 }, { "epoch": 0.0009984, "grad_norm": 0.8117906451225281, "learning_rate": 4.237439641134433e-08, "loss": 2.6257, "step": 379190 }, { "epoch": 0.001024, "grad_norm": 0.8980966210365295, "learning_rate": 4.230037405529741e-08, "loss": 2.5701, "step": 379200 }, { "epoch": 0.0010496, "grad_norm": 0.9533869028091431, "learning_rate": 4.2226416272286295e-08, "loss": 2.9396, "step": 379210 }, { "epoch": 0.0010752, "grad_norm": 0.9337859153747559, "learning_rate": 4.215252306279061e-08, "loss": 2.5487, "step": 379220 }, { "epoch": 0.0011008, "grad_norm": 0.8434877991676331, "learning_rate": 4.2078694427288845e-08, "loss": 2.5139, "step": 379230 }, { "epoch": 0.0011264, "grad_norm": 0.836421549320221, "learning_rate": 4.2004930366259523e-08, "loss": 2.6101, "step": 379240 }, { "epoch": 0.001152, "grad_norm": 0.7959022521972656, "learning_rate": 4.193123088018336e-08, "loss": 2.7405, "step": 379250 }, { "epoch": 0.0011776, "grad_norm": 0.8696139454841614, "learning_rate": 4.1857595969535534e-08, "loss": 2.4612, "step": 379260 }, { "epoch": 0.0012032, "grad_norm": 1.026585578918457, "learning_rate": 4.178402563479456e-08, "loss": 2.6359, "step": 379270 }, { "epoch": 0.0012288, "grad_norm": 0.8106064200401306, "learning_rate": 4.171051987643782e-08, "loss": 2.8473, "step": 379280 }, { "epoch": 0.0012544, "grad_norm": 0.8297988772392273, "learning_rate": 4.163707869494271e-08, "loss": 3.1142, "step": 379290 }, { "epoch": 0.00128, "grad_norm": 0.9572311043739319, "learning_rate": 4.156370209078331e-08, "loss": 2.6665, "step": 379300 }, { "epoch": 0.0013056, "grad_norm": 0.8657077550888062, "learning_rate": 4.1490390064437e-08, "loss": 2.6646, "step": 379310 }, { "epoch": 0.0013312, "grad_norm": 1.0119245052337646, "learning_rate": 4.141714261637897e-08, "loss": 2.5329, "step": 379320 }, { "epoch": 0.0013568, "grad_norm": 1.1488736867904663, "learning_rate": 4.134395974708327e-08, "loss": 2.5631, "step": 379330 }, { "epoch": 0.0013824, "grad_norm": 0.8388939499855042, "learning_rate": 4.127084145702731e-08, "loss": 2.8297, "step": 379340 }, { "epoch": 0.001408, "grad_norm": 0.9427763223648071, "learning_rate": 4.1197787746681815e-08, "loss": 2.4675, "step": 379350 }, { "epoch": 0.0014336, "grad_norm": 0.8064989447593689, "learning_rate": 4.1124798616521965e-08, "loss": 2.7527, "step": 379360 }, { "epoch": 0.0014592, "grad_norm": 0.9080030918121338, "learning_rate": 4.1051874067020716e-08, "loss": 2.636, "step": 379370 }, { "epoch": 0.0014848, "grad_norm": 0.7623753547668457, "learning_rate": 4.0979014098652124e-08, "loss": 2.3649, "step": 379380 }, { "epoch": 0.0015104, "grad_norm": 1.2154123783111572, "learning_rate": 4.0906218711886934e-08, "loss": 1.6466, "step": 379390 }, { "epoch": 0.001536, "grad_norm": 0.7656936049461365, "learning_rate": 4.083348790719921e-08, "loss": 2.5698, "step": 379400 }, { "epoch": 2.56e-05, "grad_norm": 0.8480501174926758, "learning_rate": 4.076082168505857e-08, "loss": 1.7143, "step": 379410 }, { "epoch": 5.12e-05, "grad_norm": 0.8249223232269287, "learning_rate": 4.0688220045937976e-08, "loss": 2.5568, "step": 379420 }, { "epoch": 7.68e-05, "grad_norm": 0.8507925868034363, "learning_rate": 4.061568299030594e-08, "loss": 2.8205, "step": 379430 }, { "epoch": 0.0001024, "grad_norm": 0.8068243861198425, "learning_rate": 4.054321051863541e-08, "loss": 2.5414, "step": 379440 }, { "epoch": 0.000128, "grad_norm": 0.8409696221351624, "learning_rate": 4.047080263139602e-08, "loss": 2.3492, "step": 379450 }, { "epoch": 0.0001536, "grad_norm": 0.8366379141807556, "learning_rate": 4.0398459329055174e-08, "loss": 2.5493, "step": 379460 }, { "epoch": 0.0001792, "grad_norm": 0.832241952419281, "learning_rate": 4.0326180612084706e-08, "loss": 2.5428, "step": 379470 }, { "epoch": 0.0002048, "grad_norm": 0.764065146446228, "learning_rate": 4.0253966480950926e-08, "loss": 2.5141, "step": 379480 }, { "epoch": 0.0002304, "grad_norm": 0.836781919002533, "learning_rate": 4.018181693612344e-08, "loss": 2.6566, "step": 379490 }, { "epoch": 0.000256, "grad_norm": 0.9401951432228088, "learning_rate": 4.010973197806967e-08, "loss": 2.3727, "step": 379500 }, { "epoch": 0.0002816, "grad_norm": 0.8159881830215454, "learning_rate": 4.0037711607257e-08, "loss": 2.6033, "step": 379510 }, { "epoch": 0.0003072, "grad_norm": 0.8289011716842651, "learning_rate": 3.9965755824152854e-08, "loss": 2.5248, "step": 379520 }, { "epoch": 0.0003328, "grad_norm": 0.9401506185531616, "learning_rate": 3.989386462922462e-08, "loss": 2.6568, "step": 379530 }, { "epoch": 0.0003584, "grad_norm": 0.8110364675521851, "learning_rate": 3.982203802293749e-08, "loss": 2.4839, "step": 379540 }, { "epoch": 0.000384, "grad_norm": 0.8132232427597046, "learning_rate": 3.975027600575665e-08, "loss": 2.577, "step": 379550 }, { "epoch": 0.0004096, "grad_norm": 0.80683434009552, "learning_rate": 3.9678578578148385e-08, "loss": 2.7278, "step": 379560 }, { "epoch": 0.0004352, "grad_norm": 0.8659632802009583, "learning_rate": 3.960694574057677e-08, "loss": 2.7041, "step": 379570 }, { "epoch": 0.0004608, "grad_norm": 0.8278274536132812, "learning_rate": 3.953537749350811e-08, "loss": 2.6979, "step": 379580 }, { "epoch": 0.0004864, "grad_norm": 0.8012290000915527, "learning_rate": 3.9463873837404245e-08, "loss": 2.701, "step": 379590 }, { "epoch": 0.000512, "grad_norm": 0.8305657505989075, "learning_rate": 3.939243477273147e-08, "loss": 2.5633, "step": 379600 }, { "epoch": 0.0005376, "grad_norm": 0.8523005247116089, "learning_rate": 3.932106029995053e-08, "loss": 2.57, "step": 379610 }, { "epoch": 0.0005632, "grad_norm": 0.771841287612915, "learning_rate": 3.924975041952439e-08, "loss": 2.4692, "step": 379620 }, { "epoch": 0.0005888, "grad_norm": 0.9099401235580444, "learning_rate": 3.917850513191823e-08, "loss": 2.5671, "step": 379630 }, { "epoch": 0.0006144, "grad_norm": 0.872648298740387, "learning_rate": 3.910732443759058e-08, "loss": 2.5409, "step": 379640 }, { "epoch": 0.00064, "grad_norm": 0.8714896440505981, "learning_rate": 3.90362083370055e-08, "loss": 2.7651, "step": 379650 }, { "epoch": 0.0006656, "grad_norm": 0.8887680768966675, "learning_rate": 3.896515683062374e-08, "loss": 2.9633, "step": 379660 }, { "epoch": 0.0006912, "grad_norm": 0.8484546542167664, "learning_rate": 3.889416991890493e-08, "loss": 2.6907, "step": 379670 }, { "epoch": 0.0007168, "grad_norm": 0.8356562852859497, "learning_rate": 3.882324760230982e-08, "loss": 2.7308, "step": 379680 }, { "epoch": 0.0007424, "grad_norm": 0.887783408164978, "learning_rate": 3.8752389881299146e-08, "loss": 2.9202, "step": 379690 }, { "epoch": 0.000768, "grad_norm": 0.857287585735321, "learning_rate": 3.868159675633143e-08, "loss": 2.5993, "step": 379700 }, { "epoch": 0.0007936, "grad_norm": 0.805932879447937, "learning_rate": 3.861086822786631e-08, "loss": 2.5619, "step": 379710 }, { "epoch": 0.0008192, "grad_norm": 0.8404780030250549, "learning_rate": 3.8540204296362295e-08, "loss": 2.5717, "step": 379720 }, { "epoch": 0.0008448, "grad_norm": 0.8240410685539246, "learning_rate": 3.846960496227792e-08, "loss": 2.7078, "step": 379730 }, { "epoch": 0.0008704, "grad_norm": 0.7934319376945496, "learning_rate": 3.8399070226070587e-08, "loss": 2.5383, "step": 379740 }, { "epoch": 0.000896, "grad_norm": 0.8753677606582642, "learning_rate": 3.832860008819772e-08, "loss": 2.5018, "step": 379750 }, { "epoch": 0.0009216, "grad_norm": 1.6079270839691162, "learning_rate": 3.825819454911672e-08, "loss": 2.618, "step": 379760 }, { "epoch": 0.0009472, "grad_norm": 0.8662344217300415, "learning_rate": 3.818785360928279e-08, "loss": 2.4888, "step": 379770 }, { "epoch": 0.0009728, "grad_norm": 0.7903842329978943, "learning_rate": 3.811757726915444e-08, "loss": 2.6654, "step": 379780 }, { "epoch": 0.0009984, "grad_norm": 0.8410694003105164, "learning_rate": 3.8047365529184645e-08, "loss": 2.534, "step": 379790 }, { "epoch": 0.001024, "grad_norm": 0.8426778316497803, "learning_rate": 3.7977218389831925e-08, "loss": 2.7552, "step": 379800 }, { "epoch": 0.0010496, "grad_norm": 1.0019893646240234, "learning_rate": 3.790713585154815e-08, "loss": 2.5894, "step": 379810 }, { "epoch": 0.0010752, "grad_norm": 0.8101213574409485, "learning_rate": 3.78371179147885e-08, "loss": 2.5207, "step": 379820 }, { "epoch": 0.0011008, "grad_norm": 0.8327924013137817, "learning_rate": 3.7767164580008176e-08, "loss": 2.6124, "step": 379830 }, { "epoch": 0.0011264, "grad_norm": 0.8155297040939331, "learning_rate": 3.7697275847660144e-08, "loss": 2.5477, "step": 379840 }, { "epoch": 0.001152, "grad_norm": 0.90217524766922, "learning_rate": 3.7627451718197375e-08, "loss": 2.6568, "step": 379850 }, { "epoch": 0.0011776, "grad_norm": 0.878417432308197, "learning_rate": 3.755769219207284e-08, "loss": 2.7412, "step": 379860 }, { "epoch": 0.0012032, "grad_norm": 0.8581405282020569, "learning_rate": 3.748799726973951e-08, "loss": 2.8695, "step": 379870 }, { "epoch": 0.0012288, "grad_norm": 0.9902030825614929, "learning_rate": 3.741836695164924e-08, "loss": 3.0783, "step": 379880 }, { "epoch": 0.0012544, "grad_norm": 0.8383988738059998, "learning_rate": 3.7348801238251686e-08, "loss": 2.67, "step": 379890 }, { "epoch": 0.00128, "grad_norm": 0.8317307233810425, "learning_rate": 3.727930013000092e-08, "loss": 2.8021, "step": 379900 }, { "epoch": 0.0013056, "grad_norm": 0.8677374720573425, "learning_rate": 3.720986362734436e-08, "loss": 2.811, "step": 379910 }, { "epoch": 0.0013312, "grad_norm": 1.0412846803665161, "learning_rate": 3.7140491730734974e-08, "loss": 2.3249, "step": 379920 }, { "epoch": 0.0013568, "grad_norm": 0.9099971055984497, "learning_rate": 3.707118444062241e-08, "loss": 2.8012, "step": 379930 }, { "epoch": 0.0013824, "grad_norm": 0.8380234837532043, "learning_rate": 3.7001941757454085e-08, "loss": 2.8487, "step": 379940 }, { "epoch": 0.001408, "grad_norm": 0.8182653188705444, "learning_rate": 3.693276368168075e-08, "loss": 2.6865, "step": 379950 }, { "epoch": 0.0014336, "grad_norm": 0.8906424045562744, "learning_rate": 3.6863650213750934e-08, "loss": 2.4447, "step": 379960 }, { "epoch": 0.0014592, "grad_norm": 0.8540152311325073, "learning_rate": 3.679460135411206e-08, "loss": 2.566, "step": 379970 }, { "epoch": 0.0014848, "grad_norm": 0.8005587458610535, "learning_rate": 3.672561710321265e-08, "loss": 2.413, "step": 379980 }, { "epoch": 0.0015104, "grad_norm": 0.7906073927879333, "learning_rate": 3.665669746150013e-08, "loss": 2.3318, "step": 379990 }, { "epoch": 0.001536, "grad_norm": 0.8813594579696655, "learning_rate": 3.6587842429420815e-08, "loss": 2.5106, "step": 380000 }, { "epoch": 2.56e-05, "grad_norm": 0.8466230034828186, "learning_rate": 3.651905200742101e-08, "loss": 1.6914, "step": 380010 }, { "epoch": 5.12e-05, "grad_norm": 1.296527624130249, "learning_rate": 3.6450326195947014e-08, "loss": 2.5597, "step": 380020 }, { "epoch": 7.68e-05, "grad_norm": 0.7606250643730164, "learning_rate": 3.6381664995444045e-08, "loss": 2.5585, "step": 380030 }, { "epoch": 0.0001024, "grad_norm": 0.8329922556877136, "learning_rate": 3.631306840635951e-08, "loss": 2.5987, "step": 380040 }, { "epoch": 0.000128, "grad_norm": 0.8567012548446655, "learning_rate": 3.624453642913528e-08, "loss": 2.6415, "step": 380050 }, { "epoch": 0.0001536, "grad_norm": 0.8590782284736633, "learning_rate": 3.617606906421767e-08, "loss": 2.5637, "step": 380060 }, { "epoch": 0.0001792, "grad_norm": 0.7433835864067078, "learning_rate": 3.610766631205076e-08, "loss": 2.5124, "step": 380070 }, { "epoch": 0.0002048, "grad_norm": 0.799005389213562, "learning_rate": 3.603932817307754e-08, "loss": 2.687, "step": 380080 }, { "epoch": 0.0002304, "grad_norm": 0.7817230820655823, "learning_rate": 3.597105464773987e-08, "loss": 2.6627, "step": 380090 }, { "epoch": 0.000256, "grad_norm": 0.8102098107337952, "learning_rate": 3.590284573648295e-08, "loss": 2.6246, "step": 380100 }, { "epoch": 2.56e-05, "grad_norm": 0.8899606466293335, "learning_rate": 3.583470143974754e-08, "loss": 2.109, "step": 380110 }, { "epoch": 5.12e-05, "grad_norm": 0.9064435958862305, "learning_rate": 3.576662175797552e-08, "loss": 2.4169, "step": 380120 }, { "epoch": 7.68e-05, "grad_norm": 0.8978599905967712, "learning_rate": 3.569860669160985e-08, "loss": 2.5274, "step": 380130 }, { "epoch": 0.0001024, "grad_norm": 0.8177512288093567, "learning_rate": 3.5630656241090186e-08, "loss": 2.7008, "step": 380140 }, { "epoch": 0.000128, "grad_norm": 0.8536467552185059, "learning_rate": 3.556277040685729e-08, "loss": 2.6491, "step": 380150 }, { "epoch": 0.0001536, "grad_norm": 0.7518869638442993, "learning_rate": 3.5494949189350815e-08, "loss": 2.4744, "step": 380160 }, { "epoch": 0.0001792, "grad_norm": 0.8660011887550354, "learning_rate": 3.5427192589012615e-08, "loss": 2.4731, "step": 380170 }, { "epoch": 0.0002048, "grad_norm": 0.840986430644989, "learning_rate": 3.535950060628013e-08, "loss": 2.6381, "step": 380180 }, { "epoch": 0.0002304, "grad_norm": 0.8754302263259888, "learning_rate": 3.5291873241593e-08, "loss": 2.621, "step": 380190 }, { "epoch": 0.000256, "grad_norm": 0.8452579379081726, "learning_rate": 3.5224310495390877e-08, "loss": 2.6628, "step": 380200 }, { "epoch": 0.0002816, "grad_norm": 0.8467193245887756, "learning_rate": 3.515681236810897e-08, "loss": 2.6095, "step": 380210 }, { "epoch": 0.0003072, "grad_norm": 0.784852921962738, "learning_rate": 3.5089378860188036e-08, "loss": 2.446, "step": 380220 }, { "epoch": 0.0003328, "grad_norm": 0.9285172820091248, "learning_rate": 3.5022009972064394e-08, "loss": 2.5964, "step": 380230 }, { "epoch": 0.0003584, "grad_norm": 0.8025938272476196, "learning_rate": 3.495470570417325e-08, "loss": 2.7931, "step": 380240 }, { "epoch": 0.000384, "grad_norm": 0.89206862449646, "learning_rate": 3.488746605695315e-08, "loss": 2.6188, "step": 380250 }, { "epoch": 0.0004096, "grad_norm": 0.8302233219146729, "learning_rate": 3.482029103083928e-08, "loss": 2.569, "step": 380260 }, { "epoch": 0.0004352, "grad_norm": 0.9083172678947449, "learning_rate": 3.475318062626798e-08, "loss": 2.7972, "step": 380270 }, { "epoch": 0.0004608, "grad_norm": 0.8548811674118042, "learning_rate": 3.468613484367334e-08, "loss": 2.66, "step": 380280 }, { "epoch": 0.0004864, "grad_norm": 0.9335700273513794, "learning_rate": 3.461915368349056e-08, "loss": 2.8194, "step": 380290 }, { "epoch": 0.000512, "grad_norm": 0.7893691062927246, "learning_rate": 3.455223714615486e-08, "loss": 2.4434, "step": 380300 }, { "epoch": 0.0005376, "grad_norm": 0.8596598505973816, "learning_rate": 3.44853852320981e-08, "loss": 2.6948, "step": 380310 }, { "epoch": 0.0005632, "grad_norm": 0.7947694063186646, "learning_rate": 3.4418597941756615e-08, "loss": 2.4072, "step": 380320 }, { "epoch": 0.0005888, "grad_norm": 0.8541346788406372, "learning_rate": 3.4351875275561164e-08, "loss": 2.3771, "step": 380330 }, { "epoch": 0.0006144, "grad_norm": 0.7866935133934021, "learning_rate": 3.4285217233945847e-08, "loss": 2.472, "step": 380340 }, { "epoch": 0.00064, "grad_norm": 0.8522375226020813, "learning_rate": 3.4218623817341424e-08, "loss": 2.4914, "step": 380350 }, { "epoch": 0.0006656, "grad_norm": 1.0008128881454468, "learning_rate": 3.415209502618089e-08, "loss": 3.0164, "step": 380360 }, { "epoch": 0.0006912, "grad_norm": 0.8164426684379578, "learning_rate": 3.408563086089611e-08, "loss": 2.6204, "step": 380370 }, { "epoch": 0.0007168, "grad_norm": 0.963075578212738, "learning_rate": 3.401923132191676e-08, "loss": 2.6683, "step": 380380 }, { "epoch": 0.0007424, "grad_norm": 0.8183284401893616, "learning_rate": 3.3952896409673586e-08, "loss": 2.5395, "step": 380390 }, { "epoch": 0.000768, "grad_norm": 0.9527423977851868, "learning_rate": 3.388662612459848e-08, "loss": 2.8285, "step": 380400 }, { "epoch": 0.0007936, "grad_norm": 0.7868545055389404, "learning_rate": 3.382042046711887e-08, "loss": 2.6218, "step": 380410 }, { "epoch": 0.0008192, "grad_norm": 0.8836758732795715, "learning_rate": 3.3754279437665516e-08, "loss": 2.5967, "step": 380420 }, { "epoch": 0.0008448, "grad_norm": 0.8037191033363342, "learning_rate": 3.3688203036666976e-08, "loss": 2.4531, "step": 380430 }, { "epoch": 0.0008704, "grad_norm": 0.8035788536071777, "learning_rate": 3.3622191264551795e-08, "loss": 2.6892, "step": 380440 }, { "epoch": 0.000896, "grad_norm": 0.8700395226478577, "learning_rate": 3.3556244121747405e-08, "loss": 2.7593, "step": 380450 }, { "epoch": 0.0009216, "grad_norm": 0.9080121517181396, "learning_rate": 3.349036160868235e-08, "loss": 2.5239, "step": 380460 }, { "epoch": 0.0009472, "grad_norm": 0.732963502407074, "learning_rate": 3.3424543725782967e-08, "loss": 2.3849, "step": 380470 }, { "epoch": 0.0009728, "grad_norm": 0.9181424975395203, "learning_rate": 3.3358790473477786e-08, "loss": 2.5478, "step": 380480 }, { "epoch": 0.0009984, "grad_norm": 0.8896231651306152, "learning_rate": 3.329310185219092e-08, "loss": 2.5571, "step": 380490 }, { "epoch": 0.001024, "grad_norm": 0.9327064156532288, "learning_rate": 3.322747786235092e-08, "loss": 2.8803, "step": 380500 }, { "epoch": 0.0010496, "grad_norm": 0.8353829383850098, "learning_rate": 3.316191850438077e-08, "loss": 2.6597, "step": 380510 }, { "epoch": 0.0010752, "grad_norm": 1.1453585624694824, "learning_rate": 3.309642377870681e-08, "loss": 2.883, "step": 380520 }, { "epoch": 0.0011008, "grad_norm": 0.8370274305343628, "learning_rate": 3.303099368575424e-08, "loss": 2.6721, "step": 380530 }, { "epoch": 0.0011264, "grad_norm": 0.8134083151817322, "learning_rate": 3.2965628225946065e-08, "loss": 2.4394, "step": 380540 }, { "epoch": 0.001152, "grad_norm": 0.8779875040054321, "learning_rate": 3.290032739970639e-08, "loss": 2.7946, "step": 380550 }, { "epoch": 0.0011776, "grad_norm": 0.9968319535255432, "learning_rate": 3.2835091207460426e-08, "loss": 2.8331, "step": 380560 }, { "epoch": 0.0012032, "grad_norm": 0.890073299407959, "learning_rate": 3.2769919649628947e-08, "loss": 2.9242, "step": 380570 }, { "epoch": 0.0012288, "grad_norm": 0.8282092213630676, "learning_rate": 3.270481272663606e-08, "loss": 2.674, "step": 380580 }, { "epoch": 0.0012544, "grad_norm": 0.8906823992729187, "learning_rate": 3.2639770438903654e-08, "loss": 2.7301, "step": 380590 }, { "epoch": 0.00128, "grad_norm": 1.2585906982421875, "learning_rate": 3.2574792786853604e-08, "loss": 2.6566, "step": 380600 }, { "epoch": 0.0013056, "grad_norm": 0.9723369479179382, "learning_rate": 3.250987977090669e-08, "loss": 2.6525, "step": 380610 }, { "epoch": 0.0013312, "grad_norm": 0.8984964489936829, "learning_rate": 3.2445031391483696e-08, "loss": 2.5212, "step": 380620 }, { "epoch": 0.0013568, "grad_norm": 0.9404512643814087, "learning_rate": 3.238024764900538e-08, "loss": 2.7007, "step": 380630 }, { "epoch": 0.0013824, "grad_norm": 0.8429102301597595, "learning_rate": 3.2315528543892526e-08, "loss": 2.36, "step": 380640 }, { "epoch": 0.001408, "grad_norm": 1.1987154483795166, "learning_rate": 3.22508740765648e-08, "loss": 2.7437, "step": 380650 }, { "epoch": 0.0014336, "grad_norm": 0.8780139088630676, "learning_rate": 3.218628424744075e-08, "loss": 2.6437, "step": 380660 }, { "epoch": 0.0014592, "grad_norm": 0.7963619232177734, "learning_rate": 3.2121759056938925e-08, "loss": 2.4126, "step": 380670 }, { "epoch": 0.0014848, "grad_norm": 1.1926405429840088, "learning_rate": 3.2057298505479004e-08, "loss": 1.6646, "step": 380680 }, { "epoch": 0.0015104, "grad_norm": 0.8412919640541077, "learning_rate": 3.199290259347843e-08, "loss": 2.6339, "step": 380690 }, { "epoch": 0.001536, "grad_norm": 0.85013347864151, "learning_rate": 3.192857132135463e-08, "loss": 2.6314, "step": 380700 }, { "epoch": 2.56e-05, "grad_norm": 0.9206287264823914, "learning_rate": 3.186430468952395e-08, "loss": 1.8643, "step": 380710 }, { "epoch": 5.12e-05, "grad_norm": 0.8778917193412781, "learning_rate": 3.180010269840494e-08, "loss": 2.7379, "step": 380720 }, { "epoch": 7.68e-05, "grad_norm": 0.8951184153556824, "learning_rate": 3.1735965348412834e-08, "loss": 2.5923, "step": 380730 }, { "epoch": 0.0001024, "grad_norm": 0.7543774843215942, "learning_rate": 3.167189263996395e-08, "loss": 2.5733, "step": 380740 }, { "epoch": 0.000128, "grad_norm": 0.8253009915351868, "learning_rate": 3.160788457347352e-08, "loss": 2.6136, "step": 380750 }, { "epoch": 0.0001536, "grad_norm": 0.8504757285118103, "learning_rate": 3.154394114935677e-08, "loss": 2.4061, "step": 380760 }, { "epoch": 0.0001792, "grad_norm": 0.9165390133857727, "learning_rate": 3.148006236802781e-08, "loss": 2.6911, "step": 380770 }, { "epoch": 0.0002048, "grad_norm": 0.8357550501823425, "learning_rate": 3.1416248229901856e-08, "loss": 2.7522, "step": 380780 }, { "epoch": 0.0002304, "grad_norm": 0.835722804069519, "learning_rate": 3.135249873539192e-08, "loss": 2.56, "step": 380790 }, { "epoch": 0.000256, "grad_norm": 0.953279972076416, "learning_rate": 3.128881388491101e-08, "loss": 2.3953, "step": 380800 }, { "epoch": 0.0002816, "grad_norm": 0.8692259192466736, "learning_rate": 3.122519367887433e-08, "loss": 2.5143, "step": 380810 }, { "epoch": 0.0003072, "grad_norm": 0.8520172834396362, "learning_rate": 3.116163811769157e-08, "loss": 2.6622, "step": 380820 }, { "epoch": 0.0003328, "grad_norm": 0.8231519460678101, "learning_rate": 3.109814720177684e-08, "loss": 2.5016, "step": 380830 }, { "epoch": 0.0003584, "grad_norm": 0.9293779134750366, "learning_rate": 3.103472093154092e-08, "loss": 2.7655, "step": 380840 }, { "epoch": 0.000384, "grad_norm": 0.8530084490776062, "learning_rate": 3.09713593073957e-08, "loss": 2.5687, "step": 380850 }, { "epoch": 0.0004096, "grad_norm": 0.8245231509208679, "learning_rate": 3.0908062329750854e-08, "loss": 2.4754, "step": 380860 }, { "epoch": 0.0004352, "grad_norm": 0.8237999081611633, "learning_rate": 3.084482999901939e-08, "loss": 2.6571, "step": 380870 }, { "epoch": 0.0004608, "grad_norm": 0.900961697101593, "learning_rate": 3.078166231560875e-08, "loss": 2.6207, "step": 380880 }, { "epoch": 0.0004864, "grad_norm": 0.9866151213645935, "learning_rate": 3.071855927992862e-08, "loss": 2.7366, "step": 380890 }, { "epoch": 0.000512, "grad_norm": 0.8376080989837646, "learning_rate": 3.065552089239088e-08, "loss": 2.5411, "step": 380900 }, { "epoch": 0.0005376, "grad_norm": 0.8479009866714478, "learning_rate": 3.059254715340076e-08, "loss": 2.5751, "step": 380910 }, { "epoch": 0.0005632, "grad_norm": 0.7869959473609924, "learning_rate": 3.052963806337017e-08, "loss": 2.62, "step": 380920 }, { "epoch": 0.0005888, "grad_norm": 0.8078933954238892, "learning_rate": 3.0466793622704326e-08, "loss": 2.5105, "step": 380930 }, { "epoch": 0.0006144, "grad_norm": 0.8568593859672546, "learning_rate": 3.04040138318118e-08, "loss": 2.6837, "step": 380940 }, { "epoch": 0.00064, "grad_norm": 0.8928492069244385, "learning_rate": 3.034129869110003e-08, "loss": 2.6683, "step": 380950 }, { "epoch": 0.0006656, "grad_norm": 0.8733196258544922, "learning_rate": 3.027864820097537e-08, "loss": 2.9242, "step": 380960 }, { "epoch": 0.0006912, "grad_norm": 0.9299532771110535, "learning_rate": 3.021606236184305e-08, "loss": 2.5618, "step": 380970 }, { "epoch": 0.0007168, "grad_norm": 0.8688300251960754, "learning_rate": 3.015354117411051e-08, "loss": 2.5897, "step": 380980 }, { "epoch": 0.0007424, "grad_norm": 0.9956091642379761, "learning_rate": 3.0091084638183e-08, "loss": 2.8085, "step": 380990 }, { "epoch": 0.000768, "grad_norm": 0.8876728415489197, "learning_rate": 3.0028692754464626e-08, "loss": 2.6021, "step": 381000 }, { "epoch": 0.0007936, "grad_norm": 0.8799805641174316, "learning_rate": 2.996636552336063e-08, "loss": 2.7246, "step": 381010 }, { "epoch": 0.0008192, "grad_norm": 1.060590147972107, "learning_rate": 2.9904102945275124e-08, "loss": 2.6291, "step": 381020 }, { "epoch": 0.0008448, "grad_norm": 0.9541494250297546, "learning_rate": 2.984190502061224e-08, "loss": 2.6313, "step": 381030 }, { "epoch": 0.0008704, "grad_norm": 0.8602679967880249, "learning_rate": 2.9779771749773867e-08, "loss": 2.5151, "step": 381040 }, { "epoch": 0.000896, "grad_norm": 0.7519595623016357, "learning_rate": 2.9717703133164134e-08, "loss": 2.3165, "step": 381050 }, { "epoch": 0.0009216, "grad_norm": 0.8874619603157043, "learning_rate": 2.9655699171186048e-08, "loss": 2.6018, "step": 381060 }, { "epoch": 0.0009472, "grad_norm": 0.8374640345573425, "learning_rate": 2.95937598642404e-08, "loss": 2.6882, "step": 381070 }, { "epoch": 0.0009728, "grad_norm": 0.9771841764450073, "learning_rate": 2.9531885212729094e-08, "loss": 2.6894, "step": 381080 }, { "epoch": 0.0009984, "grad_norm": 0.9607027173042297, "learning_rate": 2.947007521705403e-08, "loss": 2.7417, "step": 381090 }, { "epoch": 0.001024, "grad_norm": 0.9255398511886597, "learning_rate": 2.940832987761599e-08, "loss": 2.6755, "step": 381100 }, { "epoch": 0.0010496, "grad_norm": 0.8526195883750916, "learning_rate": 2.9346649194814668e-08, "loss": 2.5594, "step": 381110 }, { "epoch": 0.0010752, "grad_norm": 0.8627004027366638, "learning_rate": 2.9285033169050847e-08, "loss": 2.7452, "step": 381120 }, { "epoch": 0.0011008, "grad_norm": 0.9650018811225891, "learning_rate": 2.9223481800723096e-08, "loss": 2.7103, "step": 381130 }, { "epoch": 0.0011264, "grad_norm": 0.8583819270133972, "learning_rate": 2.9161995090232208e-08, "loss": 2.5675, "step": 381140 }, { "epoch": 0.001152, "grad_norm": 0.8488948345184326, "learning_rate": 2.9100573037974532e-08, "loss": 2.6307, "step": 381150 }, { "epoch": 0.0011776, "grad_norm": 0.8285694122314453, "learning_rate": 2.903921564434975e-08, "loss": 2.7272, "step": 381160 }, { "epoch": 0.0012032, "grad_norm": 1.1035571098327637, "learning_rate": 2.897792290975643e-08, "loss": 3.2711, "step": 381170 }, { "epoch": 0.0012288, "grad_norm": 0.745105504989624, "learning_rate": 2.891669483459092e-08, "loss": 2.8365, "step": 381180 }, { "epoch": 0.0012544, "grad_norm": 0.9032918810844421, "learning_rate": 2.8855531419250683e-08, "loss": 2.8112, "step": 381190 }, { "epoch": 0.00128, "grad_norm": 0.8920385837554932, "learning_rate": 2.8794432664133178e-08, "loss": 2.6963, "step": 381200 }, { "epoch": 0.0013056, "grad_norm": 0.9161121249198914, "learning_rate": 2.873339856963253e-08, "loss": 2.8207, "step": 381210 }, { "epoch": 0.0013312, "grad_norm": 0.9052241444587708, "learning_rate": 2.8672429136146208e-08, "loss": 2.6484, "step": 381220 }, { "epoch": 0.0013568, "grad_norm": 0.8485632538795471, "learning_rate": 2.8611524364069444e-08, "loss": 2.3024, "step": 381230 }, { "epoch": 0.0013824, "grad_norm": 0.8014005422592163, "learning_rate": 2.8550684253796367e-08, "loss": 2.462, "step": 381240 }, { "epoch": 0.001408, "grad_norm": 0.8827414512634277, "learning_rate": 2.848990880572222e-08, "loss": 2.7555, "step": 381250 }, { "epoch": 0.0014336, "grad_norm": 1.2331191301345825, "learning_rate": 2.8429198020241132e-08, "loss": 2.686, "step": 381260 }, { "epoch": 0.0014592, "grad_norm": 0.858913004398346, "learning_rate": 2.836855189774612e-08, "loss": 2.6788, "step": 381270 }, { "epoch": 0.0014848, "grad_norm": 0.8260610699653625, "learning_rate": 2.8307970438631315e-08, "loss": 2.2857, "step": 381280 }, { "epoch": 0.0015104, "grad_norm": 1.0005364418029785, "learning_rate": 2.8247453643288625e-08, "loss": 2.1152, "step": 381290 }, { "epoch": 0.001536, "grad_norm": 0.8941596746444702, "learning_rate": 2.8187001512111067e-08, "loss": 2.2499, "step": 381300 }, { "epoch": 0.0015616, "grad_norm": 0.8453956246376038, "learning_rate": 2.8126614045490553e-08, "loss": 2.6837, "step": 381310 }, { "epoch": 0.0015872, "grad_norm": 0.9267572164535522, "learning_rate": 2.8066291243818988e-08, "loss": 2.6462, "step": 381320 }, { "epoch": 0.0016128, "grad_norm": 0.8220072388648987, "learning_rate": 2.8006033107488283e-08, "loss": 2.8333, "step": 381330 }, { "epoch": 0.0016384, "grad_norm": 0.8492071032524109, "learning_rate": 2.7945839636887017e-08, "loss": 2.6404, "step": 381340 }, { "epoch": 0.001664, "grad_norm": 0.901608407497406, "learning_rate": 2.78857108324071e-08, "loss": 2.6531, "step": 381350 }, { "epoch": 0.0016896, "grad_norm": 0.7973741888999939, "learning_rate": 2.782564669443821e-08, "loss": 2.5362, "step": 381360 }, { "epoch": 0.0017152, "grad_norm": 0.907521665096283, "learning_rate": 2.776564722336894e-08, "loss": 2.7289, "step": 381370 }, { "epoch": 0.0017408, "grad_norm": 0.8768634796142578, "learning_rate": 2.770571241959008e-08, "loss": 2.6619, "step": 381380 }, { "epoch": 0.0017664, "grad_norm": 1.0049203634262085, "learning_rate": 2.76458422834891e-08, "loss": 2.9784, "step": 381390 }, { "epoch": 0.001792, "grad_norm": 0.8792562484741211, "learning_rate": 2.7586036815454574e-08, "loss": 2.816, "step": 381400 }, { "epoch": 2.56e-05, "grad_norm": 0.9294179677963257, "learning_rate": 2.7526296015873976e-08, "loss": 1.709, "step": 381410 }, { "epoch": 5.12e-05, "grad_norm": 0.9850875735282898, "learning_rate": 2.746661988513477e-08, "loss": 2.4922, "step": 381420 }, { "epoch": 7.68e-05, "grad_norm": 0.7994523644447327, "learning_rate": 2.7407008423624425e-08, "loss": 2.6709, "step": 381430 }, { "epoch": 0.0001024, "grad_norm": 0.8556673526763916, "learning_rate": 2.73474616317293e-08, "loss": 2.6271, "step": 381440 }, { "epoch": 0.000128, "grad_norm": 0.8717460632324219, "learning_rate": 2.7287979509835748e-08, "loss": 2.7849, "step": 381450 }, { "epoch": 0.0001536, "grad_norm": 0.8358680009841919, "learning_rate": 2.7228562058329023e-08, "loss": 2.5494, "step": 381460 }, { "epoch": 0.0001792, "grad_norm": 0.9909674525260925, "learning_rate": 2.7169209277594366e-08, "loss": 2.6957, "step": 381470 }, { "epoch": 0.0002048, "grad_norm": 0.7633565068244934, "learning_rate": 2.710992116801703e-08, "loss": 2.3917, "step": 381480 }, { "epoch": 0.0002304, "grad_norm": 0.956653356552124, "learning_rate": 2.7050697729981145e-08, "loss": 2.5901, "step": 381490 }, { "epoch": 0.000256, "grad_norm": 0.8468479514122009, "learning_rate": 2.6991538963870855e-08, "loss": 2.5889, "step": 381500 }, { "epoch": 0.0002816, "grad_norm": 0.8229809403419495, "learning_rate": 2.6932444870070296e-08, "loss": 2.5847, "step": 381510 }, { "epoch": 0.0003072, "grad_norm": 0.8893232941627502, "learning_rate": 2.6873415448962493e-08, "loss": 2.4756, "step": 381520 }, { "epoch": 0.0003328, "grad_norm": 0.8672518134117126, "learning_rate": 2.6814450700930473e-08, "loss": 2.2368, "step": 381530 }, { "epoch": 0.0003584, "grad_norm": 0.8444836735725403, "learning_rate": 2.6755550626355044e-08, "loss": 2.5457, "step": 381540 }, { "epoch": 0.000384, "grad_norm": 0.8021304607391357, "learning_rate": 2.669671522562034e-08, "loss": 2.4665, "step": 381550 }, { "epoch": 0.0004096, "grad_norm": 0.9174531102180481, "learning_rate": 2.663794449910606e-08, "loss": 2.4296, "step": 381560 }, { "epoch": 0.0004352, "grad_norm": 0.8483964204788208, "learning_rate": 2.657923844719412e-08, "loss": 2.5364, "step": 381570 }, { "epoch": 0.0004608, "grad_norm": 0.9421857595443726, "learning_rate": 2.652059707026644e-08, "loss": 2.6255, "step": 381580 }, { "epoch": 0.0004864, "grad_norm": 0.8191220164299011, "learning_rate": 2.646202036870049e-08, "loss": 2.5691, "step": 381590 }, { "epoch": 0.000512, "grad_norm": 0.8011715412139893, "learning_rate": 2.6403508342878194e-08, "loss": 2.6205, "step": 381600 }, { "epoch": 2.56e-05, "grad_norm": 0.8523068428039551, "learning_rate": 2.634506099317924e-08, "loss": 1.7389, "step": 381610 }, { "epoch": 5.12e-05, "grad_norm": 0.8391212821006775, "learning_rate": 2.6286678319981108e-08, "loss": 2.6071, "step": 381620 }, { "epoch": 7.68e-05, "grad_norm": 0.8536185622215271, "learning_rate": 2.6228360323663492e-08, "loss": 2.5091, "step": 381630 }, { "epoch": 0.0001024, "grad_norm": 0.847683310508728, "learning_rate": 2.6170107004603872e-08, "loss": 2.8094, "step": 381640 }, { "epoch": 0.000128, "grad_norm": 0.8384186029434204, "learning_rate": 2.6111918363180834e-08, "loss": 2.4639, "step": 381650 }, { "epoch": 0.0001536, "grad_norm": 0.9816453456878662, "learning_rate": 2.605379439977074e-08, "loss": 2.6939, "step": 381660 }, { "epoch": 0.0001792, "grad_norm": 0.8446692228317261, "learning_rate": 2.5995735114751065e-08, "loss": 2.5585, "step": 381670 }, { "epoch": 0.0002048, "grad_norm": 0.8720324635505676, "learning_rate": 2.5937740508498178e-08, "loss": 2.3695, "step": 381680 }, { "epoch": 0.0002304, "grad_norm": 0.8197225332260132, "learning_rate": 2.5879810581388443e-08, "loss": 2.4771, "step": 381690 }, { "epoch": 0.000256, "grad_norm": 0.899039089679718, "learning_rate": 2.5821945333798225e-08, "loss": 2.5717, "step": 381700 }, { "epoch": 0.0002816, "grad_norm": 0.8672345280647278, "learning_rate": 2.5764144766100562e-08, "loss": 2.5607, "step": 381710 }, { "epoch": 0.0003072, "grad_norm": 0.8794943690299988, "learning_rate": 2.570640887867293e-08, "loss": 2.5295, "step": 381720 }, { "epoch": 0.0003328, "grad_norm": 0.8272092938423157, "learning_rate": 2.564873767188725e-08, "loss": 2.2584, "step": 381730 }, { "epoch": 0.0003584, "grad_norm": 0.9451506733894348, "learning_rate": 2.5591131146118775e-08, "loss": 2.6933, "step": 381740 }, { "epoch": 0.000384, "grad_norm": 0.9770883321762085, "learning_rate": 2.5533589301741657e-08, "loss": 2.6548, "step": 381750 }, { "epoch": 0.0004096, "grad_norm": 1.011718511581421, "learning_rate": 2.5476112139127817e-08, "loss": 2.7558, "step": 381760 }, { "epoch": 0.0004352, "grad_norm": 0.7992907762527466, "learning_rate": 2.5418699658650293e-08, "loss": 2.6196, "step": 381770 }, { "epoch": 0.0004608, "grad_norm": 0.8848914504051208, "learning_rate": 2.5361351860682116e-08, "loss": 2.681, "step": 381780 }, { "epoch": 0.0004864, "grad_norm": 0.9335715174674988, "learning_rate": 2.5304068745594103e-08, "loss": 2.5482, "step": 381790 }, { "epoch": 0.000512, "grad_norm": 0.8409301042556763, "learning_rate": 2.5246850313758175e-08, "loss": 2.436, "step": 381800 }, { "epoch": 2.56e-05, "grad_norm": 0.8381220698356628, "learning_rate": 2.5189696565546262e-08, "loss": 1.7176, "step": 381810 }, { "epoch": 5.12e-05, "grad_norm": 0.8270421624183655, "learning_rate": 2.5132607501326956e-08, "loss": 2.5808, "step": 381820 }, { "epoch": 7.68e-05, "grad_norm": 0.8196269869804382, "learning_rate": 2.507558312147218e-08, "loss": 2.5301, "step": 381830 }, { "epoch": 0.0001024, "grad_norm": 0.8103161454200745, "learning_rate": 2.501862342635164e-08, "loss": 2.3915, "step": 381840 }, { "epoch": 0.000128, "grad_norm": 0.8913973569869995, "learning_rate": 2.496172841633393e-08, "loss": 2.7017, "step": 381850 }, { "epoch": 0.0001536, "grad_norm": 0.9125951528549194, "learning_rate": 2.4904898091788753e-08, "loss": 2.6469, "step": 381860 }, { "epoch": 0.0001792, "grad_norm": 0.9045976400375366, "learning_rate": 2.4848132453083596e-08, "loss": 2.5397, "step": 381870 }, { "epoch": 0.0002048, "grad_norm": 0.8301952481269836, "learning_rate": 2.479143150058816e-08, "loss": 2.7738, "step": 381880 }, { "epoch": 0.0002304, "grad_norm": 0.8412489295005798, "learning_rate": 2.4734795234668817e-08, "loss": 2.5697, "step": 381890 }, { "epoch": 0.000256, "grad_norm": 0.8512207269668579, "learning_rate": 2.4678223655693058e-08, "loss": 2.6989, "step": 381900 }, { "epoch": 0.0002816, "grad_norm": 0.8774315118789673, "learning_rate": 2.4621716764027248e-08, "loss": 2.5832, "step": 381910 }, { "epoch": 0.0003072, "grad_norm": 0.8309082388877869, "learning_rate": 2.4565274560039987e-08, "loss": 2.4385, "step": 381920 }, { "epoch": 0.0003328, "grad_norm": 0.9012773633003235, "learning_rate": 2.4508897044094317e-08, "loss": 2.4377, "step": 381930 }, { "epoch": 0.0003584, "grad_norm": 0.9766936898231506, "learning_rate": 2.4452584216557718e-08, "loss": 2.934, "step": 381940 }, { "epoch": 0.000384, "grad_norm": 0.8668614029884338, "learning_rate": 2.4396336077795457e-08, "loss": 2.6612, "step": 381950 }, { "epoch": 0.0004096, "grad_norm": 1.0230820178985596, "learning_rate": 2.4340152628171688e-08, "loss": 2.5239, "step": 381960 }, { "epoch": 0.0004352, "grad_norm": 0.8244448304176331, "learning_rate": 2.428403386804945e-08, "loss": 2.5707, "step": 381970 }, { "epoch": 0.0004608, "grad_norm": 0.8645327091217041, "learning_rate": 2.422797979779623e-08, "loss": 2.7369, "step": 381980 }, { "epoch": 0.0004864, "grad_norm": 0.960697591304779, "learning_rate": 2.417199041777174e-08, "loss": 2.5428, "step": 381990 }, { "epoch": 0.000512, "grad_norm": 0.8449596166610718, "learning_rate": 2.411606572834124e-08, "loss": 2.6193, "step": 382000 }, { "epoch": 0.0005376, "grad_norm": 0.8635565042495728, "learning_rate": 2.406020572986667e-08, "loss": 2.4572, "step": 382010 }, { "epoch": 0.0005632, "grad_norm": 0.8459627032279968, "learning_rate": 2.4004410422709955e-08, "loss": 2.75, "step": 382020 }, { "epoch": 0.0005888, "grad_norm": 0.8874329924583435, "learning_rate": 2.394867980723303e-08, "loss": 2.412, "step": 382030 }, { "epoch": 0.0006144, "grad_norm": 0.8696382641792297, "learning_rate": 2.3893013883798943e-08, "loss": 2.6191, "step": 382040 }, { "epoch": 0.00064, "grad_norm": 0.8157691955566406, "learning_rate": 2.383741265276629e-08, "loss": 2.519, "step": 382050 }, { "epoch": 0.0006656, "grad_norm": 1.0787639617919922, "learning_rate": 2.3781876114497005e-08, "loss": 2.8452, "step": 382060 }, { "epoch": 0.0006912, "grad_norm": 0.9172830581665039, "learning_rate": 2.37264042693508e-08, "loss": 2.8967, "step": 382070 }, { "epoch": 0.0007168, "grad_norm": 0.9242863655090332, "learning_rate": 2.367099711768739e-08, "loss": 2.5363, "step": 382080 }, { "epoch": 0.0007424, "grad_norm": 0.8366209268569946, "learning_rate": 2.3615654659866482e-08, "loss": 2.5761, "step": 382090 }, { "epoch": 0.000768, "grad_norm": 0.9487918615341187, "learning_rate": 2.3560376896246684e-08, "loss": 2.5562, "step": 382100 }, { "epoch": 2.56e-05, "grad_norm": 0.8782978653907776, "learning_rate": 2.3505163827186595e-08, "loss": 1.8245, "step": 382110 }, { "epoch": 5.12e-05, "grad_norm": 0.9565247297286987, "learning_rate": 2.3450015453043708e-08, "loss": 2.6891, "step": 382120 }, { "epoch": 7.68e-05, "grad_norm": 1.0009469985961914, "learning_rate": 2.3394931774176622e-08, "loss": 2.6896, "step": 382130 }, { "epoch": 0.0001024, "grad_norm": 0.968223512172699, "learning_rate": 2.333991279094061e-08, "loss": 2.597, "step": 382140 }, { "epoch": 0.000128, "grad_norm": 0.8581017851829529, "learning_rate": 2.3284958503695388e-08, "loss": 2.7634, "step": 382150 }, { "epoch": 0.0001536, "grad_norm": 0.8749873042106628, "learning_rate": 2.3230068912794e-08, "loss": 2.4922, "step": 382160 }, { "epoch": 0.0001792, "grad_norm": 0.8806818723678589, "learning_rate": 2.3175244018595056e-08, "loss": 2.6299, "step": 382170 }, { "epoch": 0.0002048, "grad_norm": 1.0798083543777466, "learning_rate": 2.312048382145382e-08, "loss": 2.8056, "step": 382180 }, { "epoch": 0.0002304, "grad_norm": 0.851417601108551, "learning_rate": 2.3065788321723348e-08, "loss": 2.7297, "step": 382190 }, { "epoch": 0.000256, "grad_norm": 0.8800392746925354, "learning_rate": 2.301115751976113e-08, "loss": 2.5318, "step": 382200 }, { "epoch": 0.0002816, "grad_norm": 0.9732917547225952, "learning_rate": 2.2956591415919105e-08, "loss": 2.6322, "step": 382210 }, { "epoch": 0.0003072, "grad_norm": 0.8441980481147766, "learning_rate": 2.2902090010551438e-08, "loss": 2.7476, "step": 382220 }, { "epoch": 0.0003328, "grad_norm": 0.8537170886993408, "learning_rate": 2.2847653304013395e-08, "loss": 2.5468, "step": 382230 }, { "epoch": 0.0003584, "grad_norm": 0.8963409066200256, "learning_rate": 2.279328129665581e-08, "loss": 2.5669, "step": 382240 }, { "epoch": 0.000384, "grad_norm": 1.0787354707717896, "learning_rate": 2.2738973988832848e-08, "loss": 2.5221, "step": 382250 }, { "epoch": 0.0004096, "grad_norm": 0.7961556315422058, "learning_rate": 2.2684731380895332e-08, "loss": 2.4776, "step": 382260 }, { "epoch": 0.0004352, "grad_norm": 0.8898716568946838, "learning_rate": 2.2630553473196315e-08, "loss": 2.7652, "step": 382270 }, { "epoch": 0.0004608, "grad_norm": 0.8329148888587952, "learning_rate": 2.2576440266085518e-08, "loss": 2.6373, "step": 382280 }, { "epoch": 0.0004864, "grad_norm": 0.9396579265594482, "learning_rate": 2.2522391759915996e-08, "loss": 2.7675, "step": 382290 }, { "epoch": 0.000512, "grad_norm": 0.960446298122406, "learning_rate": 2.2468407955036352e-08, "loss": 2.5707, "step": 382300 }, { "epoch": 2.56e-05, "grad_norm": 0.993421196937561, "learning_rate": 2.2414488851797423e-08, "loss": 1.7152, "step": 382310 }, { "epoch": 5.12e-05, "grad_norm": 0.7866402864456177, "learning_rate": 2.2360634450548925e-08, "loss": 2.6236, "step": 382320 }, { "epoch": 7.68e-05, "grad_norm": 0.9511616826057434, "learning_rate": 2.2306844751639467e-08, "loss": 2.5954, "step": 382330 }, { "epoch": 0.0001024, "grad_norm": 0.9464519619941711, "learning_rate": 2.2253119755418773e-08, "loss": 2.8221, "step": 382340 }, { "epoch": 0.000128, "grad_norm": 0.8350209593772888, "learning_rate": 2.219945946223545e-08, "loss": 2.5619, "step": 382350 }, { "epoch": 0.0001536, "grad_norm": 0.9995344877243042, "learning_rate": 2.214586387243589e-08, "loss": 2.6788, "step": 382360 }, { "epoch": 0.0001792, "grad_norm": 0.9576597213745117, "learning_rate": 2.2092332986369814e-08, "loss": 2.8471, "step": 382370 }, { "epoch": 0.0002048, "grad_norm": 0.8422287106513977, "learning_rate": 2.2038866804381387e-08, "loss": 2.8201, "step": 382380 }, { "epoch": 0.0002304, "grad_norm": 0.9674120545387268, "learning_rate": 2.198546532682033e-08, "loss": 2.6449, "step": 382390 }, { "epoch": 0.000256, "grad_norm": 1.1308050155639648, "learning_rate": 2.1932128554031928e-08, "loss": 2.7125, "step": 382400 }, { "epoch": 0.0002816, "grad_norm": 0.7999570369720459, "learning_rate": 2.1878856486361455e-08, "loss": 2.4021, "step": 382410 }, { "epoch": 0.0003072, "grad_norm": 1.0106679201126099, "learning_rate": 2.1825649124154192e-08, "loss": 2.7573, "step": 382420 }, { "epoch": 0.0003328, "grad_norm": 0.888488233089447, "learning_rate": 2.1772506467756525e-08, "loss": 2.6909, "step": 382430 }, { "epoch": 0.0003584, "grad_norm": 0.9114934206008911, "learning_rate": 2.171942851751152e-08, "loss": 2.8949, "step": 382440 }, { "epoch": 0.000384, "grad_norm": 0.8378502726554871, "learning_rate": 2.166641527376445e-08, "loss": 2.5897, "step": 382450 }, { "epoch": 0.0004096, "grad_norm": 0.8205728530883789, "learning_rate": 2.161346673685838e-08, "loss": 2.4185, "step": 382460 }, { "epoch": 0.0004352, "grad_norm": 0.8009781837463379, "learning_rate": 2.156058290713747e-08, "loss": 2.6049, "step": 382470 }, { "epoch": 0.0004608, "grad_norm": 0.8303716778755188, "learning_rate": 2.150776378494479e-08, "loss": 2.606, "step": 382480 }, { "epoch": 0.0004864, "grad_norm": 0.9235763549804688, "learning_rate": 2.145500937062117e-08, "loss": 2.7094, "step": 382490 }, { "epoch": 0.000512, "grad_norm": 0.8890935778617859, "learning_rate": 2.1402319664510785e-08, "loss": 2.6569, "step": 382500 }, { "epoch": 0.0005376, "grad_norm": 0.7781819105148315, "learning_rate": 2.1349694666953357e-08, "loss": 2.6301, "step": 382510 }, { "epoch": 0.0005632, "grad_norm": 0.8169724345207214, "learning_rate": 2.129713437829195e-08, "loss": 2.605, "step": 382520 }, { "epoch": 0.0005888, "grad_norm": 0.8437312841415405, "learning_rate": 2.124463879886629e-08, "loss": 2.5334, "step": 382530 }, { "epoch": 0.0006144, "grad_norm": 0.9652125835418701, "learning_rate": 2.1192207929017215e-08, "loss": 2.5456, "step": 382540 }, { "epoch": 0.00064, "grad_norm": 0.8011510968208313, "learning_rate": 2.1139841769084457e-08, "loss": 2.7217, "step": 382550 }, { "epoch": 0.0006656, "grad_norm": 0.8472099900245667, "learning_rate": 2.108754031940774e-08, "loss": 2.7877, "step": 382560 }, { "epoch": 0.0006912, "grad_norm": 0.9047561883926392, "learning_rate": 2.1035303580326793e-08, "loss": 2.5933, "step": 382570 }, { "epoch": 0.0007168, "grad_norm": 0.7765597701072693, "learning_rate": 2.0983131552179126e-08, "loss": 2.5925, "step": 382580 }, { "epoch": 0.0007424, "grad_norm": 0.912834107875824, "learning_rate": 2.0931024235304464e-08, "loss": 2.6313, "step": 382590 }, { "epoch": 0.000768, "grad_norm": 0.8264864683151245, "learning_rate": 2.0878981630039208e-08, "loss": 2.8832, "step": 382600 }, { "epoch": 0.0007936, "grad_norm": 1.1858758926391602, "learning_rate": 2.0827003736721973e-08, "loss": 2.2008, "step": 382610 }, { "epoch": 0.0008192, "grad_norm": 0.8401368260383606, "learning_rate": 2.077509055568916e-08, "loss": 2.5188, "step": 382620 }, { "epoch": 0.0008448, "grad_norm": 0.7965715527534485, "learning_rate": 2.0723242087278272e-08, "loss": 2.6508, "step": 382630 }, { "epoch": 0.0008704, "grad_norm": 0.7863269448280334, "learning_rate": 2.067145833182571e-08, "loss": 2.447, "step": 382640 }, { "epoch": 0.000896, "grad_norm": 0.8180289268493652, "learning_rate": 2.0619739289664544e-08, "loss": 2.2487, "step": 382650 }, { "epoch": 0.0009216, "grad_norm": 1.0464913845062256, "learning_rate": 2.0568084961133384e-08, "loss": 2.8698, "step": 382660 }, { "epoch": 0.0009472, "grad_norm": 0.8380740880966187, "learning_rate": 2.051649534656641e-08, "loss": 2.74, "step": 382670 }, { "epoch": 0.0009728, "grad_norm": 0.9428915977478027, "learning_rate": 2.0464970446296695e-08, "loss": 2.5106, "step": 382680 }, { "epoch": 0.0009984, "grad_norm": 0.8550780415534973, "learning_rate": 2.041351026065952e-08, "loss": 2.6655, "step": 382690 }, { "epoch": 0.001024, "grad_norm": 0.8016858100891113, "learning_rate": 2.0362114789989064e-08, "loss": 2.6223, "step": 382700 }, { "epoch": 0.0010496, "grad_norm": 1.0007890462875366, "learning_rate": 2.0310784034618392e-08, "loss": 2.5211, "step": 382710 }, { "epoch": 0.0010752, "grad_norm": 0.7963597178459167, "learning_rate": 2.0259517994878352e-08, "loss": 2.6506, "step": 382720 }, { "epoch": 0.0011008, "grad_norm": 0.7878838181495667, "learning_rate": 2.0208316671104233e-08, "loss": 2.6894, "step": 382730 }, { "epoch": 0.0011264, "grad_norm": 0.8223549723625183, "learning_rate": 2.015718006362688e-08, "loss": 2.4932, "step": 382740 }, { "epoch": 0.001152, "grad_norm": 0.8727031350135803, "learning_rate": 2.010610817277825e-08, "loss": 2.7052, "step": 382750 }, { "epoch": 0.0011776, "grad_norm": 0.8557990789413452, "learning_rate": 2.0055100998888076e-08, "loss": 2.7883, "step": 382760 }, { "epoch": 0.0012032, "grad_norm": 0.8036856651306152, "learning_rate": 2.000415854228943e-08, "loss": 2.8321, "step": 382770 }, { "epoch": 0.0012288, "grad_norm": 0.8569948673248291, "learning_rate": 1.9953280803310936e-08, "loss": 2.6398, "step": 382780 }, { "epoch": 0.0012544, "grad_norm": 0.8383651375770569, "learning_rate": 1.990246778228233e-08, "loss": 2.7371, "step": 382790 }, { "epoch": 0.00128, "grad_norm": 0.8296560645103455, "learning_rate": 1.985171947953446e-08, "loss": 2.7531, "step": 382800 }, { "epoch": 0.0013056, "grad_norm": 0.8753052949905396, "learning_rate": 1.980103589539595e-08, "loss": 2.7079, "step": 382810 }, { "epoch": 0.0013312, "grad_norm": 0.8914296627044678, "learning_rate": 1.9750417030194315e-08, "loss": 2.8522, "step": 382820 }, { "epoch": 0.0013568, "grad_norm": 0.862080991268158, "learning_rate": 1.9699862884260403e-08, "loss": 2.6958, "step": 382830 }, { "epoch": 0.0013824, "grad_norm": 0.8926692605018616, "learning_rate": 1.9649373457919507e-08, "loss": 2.8791, "step": 382840 }, { "epoch": 0.001408, "grad_norm": 0.8357216715812683, "learning_rate": 1.9598948751500256e-08, "loss": 2.5419, "step": 382850 }, { "epoch": 0.0014336, "grad_norm": 0.8770257234573364, "learning_rate": 1.954858876532906e-08, "loss": 2.6871, "step": 382860 }, { "epoch": 0.0014592, "grad_norm": 0.896578311920166, "learning_rate": 1.9498293499732314e-08, "loss": 2.4377, "step": 382870 }, { "epoch": 0.0014848, "grad_norm": 0.8753366470336914, "learning_rate": 1.944806295503754e-08, "loss": 2.3813, "step": 382880 }, { "epoch": 0.0015104, "grad_norm": 0.860349178314209, "learning_rate": 1.9397897131568923e-08, "loss": 2.0617, "step": 382890 }, { "epoch": 0.001536, "grad_norm": 1.0481315851211548, "learning_rate": 1.9347796029652867e-08, "loss": 2.5645, "step": 382900 }, { "epoch": 2.56e-05, "grad_norm": 0.8764859437942505, "learning_rate": 1.9297759649613557e-08, "loss": 1.8, "step": 382910 }, { "epoch": 5.12e-05, "grad_norm": 0.7807216048240662, "learning_rate": 1.924778799177629e-08, "loss": 2.5186, "step": 382920 }, { "epoch": 7.68e-05, "grad_norm": 0.9347567558288574, "learning_rate": 1.9197881056464142e-08, "loss": 2.8849, "step": 382930 }, { "epoch": 0.0001024, "grad_norm": 0.7892730832099915, "learning_rate": 1.914803884400018e-08, "loss": 2.5574, "step": 382940 }, { "epoch": 0.000128, "grad_norm": 0.8898608684539795, "learning_rate": 1.909826135470971e-08, "loss": 2.6155, "step": 382950 }, { "epoch": 0.0001536, "grad_norm": 0.916036069393158, "learning_rate": 1.9048548588914694e-08, "loss": 2.5831, "step": 382960 }, { "epoch": 0.0001792, "grad_norm": 0.8689417243003845, "learning_rate": 1.8998900546937093e-08, "loss": 2.6962, "step": 382970 }, { "epoch": 0.0002048, "grad_norm": 0.8327406644821167, "learning_rate": 1.8949317229097764e-08, "loss": 2.7066, "step": 382980 }, { "epoch": 0.0002304, "grad_norm": 0.8144673109054565, "learning_rate": 1.889979863572089e-08, "loss": 2.4688, "step": 382990 }, { "epoch": 0.000256, "grad_norm": 1.0810059309005737, "learning_rate": 1.8850344767125107e-08, "loss": 2.6456, "step": 383000 }, { "epoch": 0.0002816, "grad_norm": 0.8366479277610779, "learning_rate": 1.8800955623632378e-08, "loss": 2.4365, "step": 383010 }, { "epoch": 0.0003072, "grad_norm": 1.0366489887237549, "learning_rate": 1.8751631205563557e-08, "loss": 2.6859, "step": 383020 }, { "epoch": 0.0003328, "grad_norm": 1.0260868072509766, "learning_rate": 1.8702371513237284e-08, "loss": 2.7627, "step": 383030 }, { "epoch": 0.0003584, "grad_norm": 0.8249218463897705, "learning_rate": 1.8653176546973294e-08, "loss": 2.7613, "step": 383040 }, { "epoch": 0.000384, "grad_norm": 0.8631759285926819, "learning_rate": 1.860404630709134e-08, "loss": 2.6023, "step": 383050 }, { "epoch": 0.0004096, "grad_norm": 0.845249593257904, "learning_rate": 1.8554980793907828e-08, "loss": 2.464, "step": 383060 }, { "epoch": 0.0004352, "grad_norm": 0.8090591430664062, "learning_rate": 1.850598000774362e-08, "loss": 2.7215, "step": 383070 }, { "epoch": 0.0004608, "grad_norm": 0.7714436650276184, "learning_rate": 1.845704394891512e-08, "loss": 2.3812, "step": 383080 }, { "epoch": 0.0004864, "grad_norm": 0.8643559217453003, "learning_rate": 1.840817261774097e-08, "loss": 2.5794, "step": 383090 }, { "epoch": 0.000512, "grad_norm": 0.7812498211860657, "learning_rate": 1.8359366014535363e-08, "loss": 2.5232, "step": 383100 }, { "epoch": 0.0005376, "grad_norm": 0.8783533573150635, "learning_rate": 1.831062413961804e-08, "loss": 2.4481, "step": 383110 }, { "epoch": 0.0005632, "grad_norm": 0.928658127784729, "learning_rate": 1.8261946993302082e-08, "loss": 2.3121, "step": 383120 }, { "epoch": 0.0005888, "grad_norm": 0.8672024011611938, "learning_rate": 1.8213334575905016e-08, "loss": 2.7081, "step": 383130 }, { "epoch": 0.0006144, "grad_norm": 0.7757343649864197, "learning_rate": 1.8164786887742147e-08, "loss": 2.4966, "step": 383140 }, { "epoch": 0.00064, "grad_norm": 0.8206928968429565, "learning_rate": 1.8116303929127664e-08, "loss": 2.6324, "step": 383150 }, { "epoch": 0.0006656, "grad_norm": 0.9435679316520691, "learning_rate": 1.8067885700375763e-08, "loss": 2.6336, "step": 383160 }, { "epoch": 0.0006912, "grad_norm": 0.8691269755363464, "learning_rate": 1.8019532201800638e-08, "loss": 2.5415, "step": 383170 }, { "epoch": 0.0007168, "grad_norm": 0.9408698081970215, "learning_rate": 1.7971243433716478e-08, "loss": 2.6507, "step": 383180 }, { "epoch": 0.0007424, "grad_norm": 0.8862859010696411, "learning_rate": 1.792301939643526e-08, "loss": 2.5957, "step": 383190 }, { "epoch": 0.000768, "grad_norm": 0.8685112595558167, "learning_rate": 1.7874860090271174e-08, "loss": 2.6001, "step": 383200 }, { "epoch": 0.0007936, "grad_norm": 0.8198140263557434, "learning_rate": 1.7826765515535082e-08, "loss": 2.3706, "step": 383210 }, { "epoch": 0.0008192, "grad_norm": 0.7742181420326233, "learning_rate": 1.777873567253896e-08, "loss": 2.684, "step": 383220 }, { "epoch": 0.0008448, "grad_norm": 0.9470668435096741, "learning_rate": 1.7730770561595888e-08, "loss": 2.5172, "step": 383230 }, { "epoch": 0.0008704, "grad_norm": 0.8090770244598389, "learning_rate": 1.768287018301562e-08, "loss": 2.5347, "step": 383240 }, { "epoch": 0.000896, "grad_norm": 0.841945469379425, "learning_rate": 1.763503453710791e-08, "loss": 2.2857, "step": 383250 }, { "epoch": 0.0009216, "grad_norm": 0.8236480355262756, "learning_rate": 1.7587263624184725e-08, "loss": 2.811, "step": 383260 }, { "epoch": 0.0009472, "grad_norm": 0.9623865485191345, "learning_rate": 1.753955744455582e-08, "loss": 2.7014, "step": 383270 }, { "epoch": 0.0009728, "grad_norm": 0.8849484324455261, "learning_rate": 1.749191599852873e-08, "loss": 2.746, "step": 383280 }, { "epoch": 0.0009984, "grad_norm": 0.9559711813926697, "learning_rate": 1.7444339286414314e-08, "loss": 2.5593, "step": 383290 }, { "epoch": 0.001024, "grad_norm": 0.9707961082458496, "learning_rate": 1.7396827308520102e-08, "loss": 2.8816, "step": 383300 }, { "epoch": 0.0010496, "grad_norm": 0.9351794123649597, "learning_rate": 1.7349380065153633e-08, "loss": 2.5625, "step": 383310 }, { "epoch": 0.0010752, "grad_norm": 0.8268162608146667, "learning_rate": 1.730199755662465e-08, "loss": 2.4573, "step": 383320 }, { "epoch": 0.0011008, "grad_norm": 0.8014514446258545, "learning_rate": 1.725467978323847e-08, "loss": 2.6217, "step": 383330 }, { "epoch": 0.0011264, "grad_norm": 0.8467985391616821, "learning_rate": 1.7207426745302626e-08, "loss": 2.5152, "step": 383340 }, { "epoch": 0.001152, "grad_norm": 0.9149224162101746, "learning_rate": 1.7160238443123532e-08, "loss": 2.5627, "step": 383350 }, { "epoch": 0.0011776, "grad_norm": 0.8272507786750793, "learning_rate": 1.711311487700651e-08, "loss": 2.7555, "step": 383360 }, { "epoch": 0.0012032, "grad_norm": 0.936579704284668, "learning_rate": 1.7066056047259082e-08, "loss": 2.7266, "step": 383370 }, { "epoch": 0.0012288, "grad_norm": 0.8480087518692017, "learning_rate": 1.7019061954184347e-08, "loss": 2.7816, "step": 383380 }, { "epoch": 0.0012544, "grad_norm": 0.860532820224762, "learning_rate": 1.697213259808761e-08, "loss": 2.5881, "step": 383390 }, { "epoch": 0.00128, "grad_norm": 0.8095604777336121, "learning_rate": 1.6925267979274186e-08, "loss": 2.7785, "step": 383400 }, { "epoch": 0.0013056, "grad_norm": 0.9536285400390625, "learning_rate": 1.6878468098046053e-08, "loss": 2.8455, "step": 383410 }, { "epoch": 0.0013312, "grad_norm": 1.1743218898773193, "learning_rate": 1.6831732954709634e-08, "loss": 2.3623, "step": 383420 }, { "epoch": 0.0013568, "grad_norm": 1.2035651206970215, "learning_rate": 1.678506254956469e-08, "loss": 2.8633, "step": 383430 }, { "epoch": 0.0013824, "grad_norm": 0.8177244067192078, "learning_rate": 1.673845688291542e-08, "loss": 2.6335, "step": 383440 }, { "epoch": 0.001408, "grad_norm": 0.7662172317504883, "learning_rate": 1.669191595506492e-08, "loss": 2.5338, "step": 383450 }, { "epoch": 0.0014336, "grad_norm": 0.8296858072280884, "learning_rate": 1.6645439766314055e-08, "loss": 2.3732, "step": 383460 }, { "epoch": 0.0014592, "grad_norm": 0.9562384486198425, "learning_rate": 1.6599028316963694e-08, "loss": 2.1997, "step": 383470 }, { "epoch": 0.0014848, "grad_norm": 0.8560168743133545, "learning_rate": 1.6552681607315824e-08, "loss": 2.5273, "step": 383480 }, { "epoch": 0.0015104, "grad_norm": 0.9117460250854492, "learning_rate": 1.65063996376702e-08, "loss": 2.5194, "step": 383490 }, { "epoch": 0.001536, "grad_norm": 0.8694722056388855, "learning_rate": 1.6460182408327695e-08, "loss": 2.9087, "step": 383500 }, { "epoch": 0.0015616, "grad_norm": 0.8986956477165222, "learning_rate": 1.641402991958696e-08, "loss": 2.7954, "step": 383510 }, { "epoch": 0.0015872, "grad_norm": 0.8782846331596375, "learning_rate": 1.636794217174775e-08, "loss": 2.6541, "step": 383520 }, { "epoch": 0.0016128, "grad_norm": 0.9074561595916748, "learning_rate": 1.6321919165109836e-08, "loss": 2.7289, "step": 383530 }, { "epoch": 0.0016384, "grad_norm": 0.8098050355911255, "learning_rate": 1.6275960899970743e-08, "loss": 2.6665, "step": 383540 }, { "epoch": 0.001664, "grad_norm": 0.8219649195671082, "learning_rate": 1.6230067376629132e-08, "loss": 2.5864, "step": 383550 }, { "epoch": 0.0016896, "grad_norm": 1.0214011669158936, "learning_rate": 1.6184238595381428e-08, "loss": 2.7367, "step": 383560 }, { "epoch": 0.0017152, "grad_norm": 0.8919674158096313, "learning_rate": 1.6138474556525175e-08, "loss": 2.6981, "step": 383570 }, { "epoch": 0.0017408, "grad_norm": 0.8080316781997681, "learning_rate": 1.6092775260357905e-08, "loss": 2.6883, "step": 383580 }, { "epoch": 0.0017664, "grad_norm": 0.898712694644928, "learning_rate": 1.6047140707176058e-08, "loss": 2.4065, "step": 383590 }, { "epoch": 0.001792, "grad_norm": 0.8776135444641113, "learning_rate": 1.6001570897274943e-08, "loss": 2.766, "step": 383600 }, { "epoch": 0.0018176, "grad_norm": 0.8110093474388123, "learning_rate": 1.5956065830949884e-08, "loss": 2.4087, "step": 383610 }, { "epoch": 0.0018432, "grad_norm": 0.8993057608604431, "learning_rate": 1.5910625508497313e-08, "loss": 2.5767, "step": 383620 }, { "epoch": 0.0018688, "grad_norm": 0.9184089303016663, "learning_rate": 1.5865249930210325e-08, "loss": 2.7729, "step": 383630 }, { "epoch": 0.0018944, "grad_norm": 0.8784351348876953, "learning_rate": 1.581993909638313e-08, "loss": 2.8769, "step": 383640 }, { "epoch": 0.00192, "grad_norm": 0.9084694385528564, "learning_rate": 1.577469300730994e-08, "loss": 2.711, "step": 383650 }, { "epoch": 0.0019456, "grad_norm": 0.8830004930496216, "learning_rate": 1.5729511663284957e-08, "loss": 2.8665, "step": 383660 }, { "epoch": 0.0019712, "grad_norm": 1.0619165897369385, "learning_rate": 1.5684395064601287e-08, "loss": 2.4948, "step": 383670 }, { "epoch": 0.0019968, "grad_norm": 0.8482969403266907, "learning_rate": 1.5639343211549806e-08, "loss": 2.6849, "step": 383680 }, { "epoch": 0.0020224, "grad_norm": 0.9746080040931702, "learning_rate": 1.5594356104423613e-08, "loss": 2.549, "step": 383690 }, { "epoch": 0.002048, "grad_norm": 0.929966151714325, "learning_rate": 1.5549433743514696e-08, "loss": 2.6227, "step": 383700 }, { "epoch": 0.0020736, "grad_norm": 0.8616442084312439, "learning_rate": 1.5504576129115045e-08, "loss": 2.5544, "step": 383710 }, { "epoch": 0.0020992, "grad_norm": 0.9477209448814392, "learning_rate": 1.5459783261513317e-08, "loss": 2.8091, "step": 383720 }, { "epoch": 0.0021248, "grad_norm": 0.9016227722167969, "learning_rate": 1.541505514100261e-08, "loss": 2.7413, "step": 383730 }, { "epoch": 0.0021504, "grad_norm": 0.8708042502403259, "learning_rate": 1.5370391767870472e-08, "loss": 2.9225, "step": 383740 }, { "epoch": 0.002176, "grad_norm": 0.8905298709869385, "learning_rate": 1.5325793142407786e-08, "loss": 2.7795, "step": 383750 }, { "epoch": 0.0022016, "grad_norm": 0.8877900242805481, "learning_rate": 1.5281259264904313e-08, "loss": 2.5615, "step": 383760 }, { "epoch": 0.0022272, "grad_norm": 3.083221197128296, "learning_rate": 1.5236790135648715e-08, "loss": 2.8562, "step": 383770 }, { "epoch": 0.0022528, "grad_norm": 0.7691488862037659, "learning_rate": 1.519238575492854e-08, "loss": 2.4421, "step": 383780 }, { "epoch": 0.0022784, "grad_norm": 0.8486551642417908, "learning_rate": 1.5148046123031336e-08, "loss": 2.8013, "step": 383790 }, { "epoch": 0.002304, "grad_norm": 0.9868357181549072, "learning_rate": 1.510377124024576e-08, "loss": 2.6375, "step": 383800 }, { "epoch": 0.0023296, "grad_norm": 0.9195901155471802, "learning_rate": 1.5059561106859353e-08, "loss": 2.7274, "step": 383810 }, { "epoch": 0.0023552, "grad_norm": 1.121216058731079, "learning_rate": 1.5015415723157455e-08, "loss": 2.4315, "step": 383820 }, { "epoch": 0.0023808, "grad_norm": 0.8445014953613281, "learning_rate": 1.49713350894265e-08, "loss": 2.5817, "step": 383830 }, { "epoch": 0.0024064, "grad_norm": 0.7975754737854004, "learning_rate": 1.4927319205952916e-08, "loss": 2.5557, "step": 383840 }, { "epoch": 0.002432, "grad_norm": 0.8911510109901428, "learning_rate": 1.4883368073022042e-08, "loss": 2.8733, "step": 383850 }, { "epoch": 0.0024576, "grad_norm": 1.0003079175949097, "learning_rate": 1.4839481690919199e-08, "loss": 2.943, "step": 383860 }, { "epoch": 0.0024832, "grad_norm": 0.8140470385551453, "learning_rate": 1.4795660059927496e-08, "loss": 2.8285, "step": 383870 }, { "epoch": 0.0025088, "grad_norm": 0.9567468166351318, "learning_rate": 1.4751903180333372e-08, "loss": 2.7299, "step": 383880 }, { "epoch": 0.0025344, "grad_norm": 0.853996217250824, "learning_rate": 1.4708211052418819e-08, "loss": 2.697, "step": 383890 }, { "epoch": 0.00256, "grad_norm": 0.9300143122673035, "learning_rate": 1.4664583676468059e-08, "loss": 2.8269, "step": 383900 }, { "epoch": 0.0025856, "grad_norm": 1.231042742729187, "learning_rate": 1.4621021052764194e-08, "loss": 2.9816, "step": 383910 }, { "epoch": 0.0026112, "grad_norm": 0.8669976592063904, "learning_rate": 1.4577523181588116e-08, "loss": 2.927, "step": 383920 }, { "epoch": 0.0026368, "grad_norm": 1.068907618522644, "learning_rate": 1.4534090063222928e-08, "loss": 2.8162, "step": 383930 }, { "epoch": 0.0026624, "grad_norm": 1.028448462486267, "learning_rate": 1.4490721697950626e-08, "loss": 2.7808, "step": 383940 }, { "epoch": 0.002688, "grad_norm": 1.056596040725708, "learning_rate": 1.4447418086053212e-08, "loss": 3.064, "step": 383950 }, { "epoch": 0.0027136, "grad_norm": 1.0071868896484375, "learning_rate": 1.4404179227809344e-08, "loss": 2.5867, "step": 383960 }, { "epoch": 0.0027392, "grad_norm": 0.9930651187896729, "learning_rate": 1.4361005123501025e-08, "loss": 2.845, "step": 383970 }, { "epoch": 0.0027648, "grad_norm": 0.890330970287323, "learning_rate": 1.431789577340692e-08, "loss": 2.718, "step": 383980 }, { "epoch": 0.0027904, "grad_norm": 1.222462773323059, "learning_rate": 1.4274851177807913e-08, "loss": 2.8635, "step": 383990 }, { "epoch": 0.002816, "grad_norm": 1.001339077949524, "learning_rate": 1.4231871336982673e-08, "loss": 2.7123, "step": 384000 }, { "epoch": 0.0028416, "grad_norm": 0.9275880455970764, "learning_rate": 1.4188956251209861e-08, "loss": 2.8367, "step": 384010 }, { "epoch": 0.0028672, "grad_norm": 0.967477023601532, "learning_rate": 1.414610592076815e-08, "loss": 2.7503, "step": 384020 }, { "epoch": 0.0028928, "grad_norm": 0.7667969465255737, "learning_rate": 1.410332034593398e-08, "loss": 2.8544, "step": 384030 }, { "epoch": 0.0029184, "grad_norm": 0.8315072059631348, "learning_rate": 1.4060599526987129e-08, "loss": 2.7021, "step": 384040 }, { "epoch": 0.002944, "grad_norm": 1.182176947593689, "learning_rate": 1.4017943464201822e-08, "loss": 2.8427, "step": 384050 }, { "epoch": 0.0029696, "grad_norm": 0.8367744088172913, "learning_rate": 1.3975352157857835e-08, "loss": 2.7564, "step": 384060 }, { "epoch": 0.0029952, "grad_norm": 0.8356994390487671, "learning_rate": 1.3932825608228284e-08, "loss": 2.8081, "step": 384070 }, { "epoch": 0.0030208, "grad_norm": 0.9009926319122314, "learning_rate": 1.3890363815591834e-08, "loss": 2.8303, "step": 384080 }, { "epoch": 0.0030464, "grad_norm": 0.9022755026817322, "learning_rate": 1.38479667802216e-08, "loss": 2.635, "step": 384090 }, { "epoch": 0.003072, "grad_norm": 0.879457414150238, "learning_rate": 1.3805634502392917e-08, "loss": 2.8807, "step": 384100 }, { "epoch": 0.0030976, "grad_norm": 0.771063506603241, "learning_rate": 1.3763366982381121e-08, "loss": 2.8995, "step": 384110 }, { "epoch": 0.0031232, "grad_norm": 0.885650098323822, "learning_rate": 1.3721164220459327e-08, "loss": 2.8552, "step": 384120 }, { "epoch": 0.0031488, "grad_norm": 0.9743555784225464, "learning_rate": 1.3679026216901758e-08, "loss": 2.949, "step": 384130 }, { "epoch": 0.0031744, "grad_norm": 0.87956303358078, "learning_rate": 1.3636952971982641e-08, "loss": 2.7115, "step": 384140 }, { "epoch": 0.0032, "grad_norm": 1.0234094858169556, "learning_rate": 1.3594944485972872e-08, "loss": 2.847, "step": 384150 }, { "epoch": 0.0032256, "grad_norm": 0.9022286534309387, "learning_rate": 1.3553000759146672e-08, "loss": 2.9008, "step": 384160 }, { "epoch": 0.0032512, "grad_norm": 0.93732750415802, "learning_rate": 1.351112179177494e-08, "loss": 2.7351, "step": 384170 }, { "epoch": 0.0032768, "grad_norm": 0.8988179564476013, "learning_rate": 1.3469307584128566e-08, "loss": 2.8722, "step": 384180 }, { "epoch": 0.0033024, "grad_norm": 0.9167949557304382, "learning_rate": 1.3427558136480668e-08, "loss": 2.6681, "step": 384190 }, { "epoch": 0.003328, "grad_norm": 0.8832772374153137, "learning_rate": 1.3385873449101028e-08, "loss": 2.7808, "step": 384200 }, { "epoch": 0.0033536, "grad_norm": 0.9429388046264648, "learning_rate": 1.3344253522259432e-08, "loss": 2.8, "step": 384210 }, { "epoch": 0.0033792, "grad_norm": 0.8742623329162598, "learning_rate": 1.3302698356226773e-08, "loss": 2.6404, "step": 384220 }, { "epoch": 0.0034048, "grad_norm": 1.0376635789871216, "learning_rate": 1.3261207951270615e-08, "loss": 2.7851, "step": 384230 }, { "epoch": 0.0034304, "grad_norm": 0.9460081458091736, "learning_rate": 1.3219782307662965e-08, "loss": 2.8421, "step": 384240 }, { "epoch": 0.003456, "grad_norm": 0.8454576730728149, "learning_rate": 1.3178421425669164e-08, "loss": 2.7119, "step": 384250 }, { "epoch": 0.0034816, "grad_norm": 0.8485557436943054, "learning_rate": 1.3137125305560105e-08, "loss": 2.6858, "step": 384260 }, { "epoch": 0.0035072, "grad_norm": 0.9517592787742615, "learning_rate": 1.3095893947603355e-08, "loss": 3.0387, "step": 384270 }, { "epoch": 0.0035328, "grad_norm": 0.9998790621757507, "learning_rate": 1.3054727352064256e-08, "loss": 2.7462, "step": 384280 }, { "epoch": 0.0035584, "grad_norm": 0.9561644196510315, "learning_rate": 1.3013625519211481e-08, "loss": 2.9869, "step": 384290 }, { "epoch": 0.003584, "grad_norm": 0.9801377058029175, "learning_rate": 1.2972588449311485e-08, "loss": 2.8517, "step": 384300 }, { "epoch": 0.0036096, "grad_norm": 0.8338704705238342, "learning_rate": 1.2931616142629611e-08, "loss": 2.6679, "step": 384310 }, { "epoch": 0.0036352, "grad_norm": 0.889506459236145, "learning_rate": 1.2890708599432312e-08, "loss": 2.8316, "step": 384320 }, { "epoch": 0.0036608, "grad_norm": 1.158429741859436, "learning_rate": 1.284986581998382e-08, "loss": 2.9687, "step": 384330 }, { "epoch": 0.0036864, "grad_norm": 0.902525007724762, "learning_rate": 1.280908780455059e-08, "loss": 2.8721, "step": 384340 }, { "epoch": 0.003712, "grad_norm": 1.2501028776168823, "learning_rate": 1.2768374553395745e-08, "loss": 2.8515, "step": 384350 }, { "epoch": 0.0037376, "grad_norm": 0.8937207460403442, "learning_rate": 1.2727726066783519e-08, "loss": 2.8574, "step": 384360 }, { "epoch": 0.0037632, "grad_norm": 0.8676603436470032, "learning_rate": 1.2687142344977033e-08, "loss": 2.7849, "step": 384370 }, { "epoch": 0.0037888, "grad_norm": 0.9465848207473755, "learning_rate": 1.264662338824163e-08, "loss": 2.7051, "step": 384380 }, { "epoch": 0.0038144, "grad_norm": 0.9108293652534485, "learning_rate": 1.2606169196837104e-08, "loss": 2.8679, "step": 384390 }, { "epoch": 0.00384, "grad_norm": 1.1223604679107666, "learning_rate": 1.2565779771027685e-08, "loss": 2.6976, "step": 384400 }, { "epoch": 0.0038656, "grad_norm": 1.064349889755249, "learning_rate": 1.252545511107539e-08, "loss": 2.7367, "step": 384410 }, { "epoch": 0.0038912, "grad_norm": 1.0184402465820312, "learning_rate": 1.2485195217241119e-08, "loss": 2.8207, "step": 384420 }, { "epoch": 0.0039168, "grad_norm": 0.9514756202697754, "learning_rate": 1.2445000089785775e-08, "loss": 2.6183, "step": 384430 }, { "epoch": 0.0039424, "grad_norm": 0.8922014236450195, "learning_rate": 1.240486972897026e-08, "loss": 2.759, "step": 384440 }, { "epoch": 0.003968, "grad_norm": 0.8999144434928894, "learning_rate": 1.2364804135054365e-08, "loss": 2.8731, "step": 384450 }, { "epoch": 0.0039936, "grad_norm": 0.97334223985672, "learning_rate": 1.2324803308298994e-08, "loss": 2.6849, "step": 384460 }, { "epoch": 0.0040192, "grad_norm": 0.8837193250656128, "learning_rate": 1.2284867248962828e-08, "loss": 2.9374, "step": 384470 }, { "epoch": 0.0040448, "grad_norm": 0.9662086367607117, "learning_rate": 1.224499595730455e-08, "loss": 2.8617, "step": 384480 }, { "epoch": 0.0040704, "grad_norm": 0.8336989283561707, "learning_rate": 1.2205189433582843e-08, "loss": 2.7434, "step": 384490 }, { "epoch": 0.004096, "grad_norm": 1.2944281101226807, "learning_rate": 1.2165447678056386e-08, "loss": 2.7286, "step": 384500 }, { "epoch": 0.0041216, "grad_norm": 0.8866420984268188, "learning_rate": 1.2125770690982752e-08, "loss": 2.978, "step": 384510 }, { "epoch": 0.0041472, "grad_norm": 0.8270201086997986, "learning_rate": 1.2086158472618403e-08, "loss": 2.8167, "step": 384520 }, { "epoch": 0.0041728, "grad_norm": 0.839104175567627, "learning_rate": 1.204661102322091e-08, "loss": 2.8232, "step": 384530 }, { "epoch": 0.0041984, "grad_norm": 0.833648681640625, "learning_rate": 1.2007128343046736e-08, "loss": 2.7629, "step": 384540 }, { "epoch": 0.004224, "grad_norm": 1.1359765529632568, "learning_rate": 1.1967710432352341e-08, "loss": 3.0087, "step": 384550 }, { "epoch": 0.0042496, "grad_norm": 0.8751175403594971, "learning_rate": 1.1928357291393077e-08, "loss": 2.7442, "step": 384560 }, { "epoch": 0.0042752, "grad_norm": 0.9956280589103699, "learning_rate": 1.1889068920423186e-08, "loss": 2.8046, "step": 384570 }, { "epoch": 0.0043008, "grad_norm": 0.8462393879890442, "learning_rate": 1.1849845319698016e-08, "loss": 2.9224, "step": 384580 }, { "epoch": 0.0043264, "grad_norm": 0.8924199342727661, "learning_rate": 1.1810686489472922e-08, "loss": 2.7393, "step": 384590 }, { "epoch": 0.004352, "grad_norm": 0.9488006830215454, "learning_rate": 1.1771592429999924e-08, "loss": 2.8959, "step": 384600 }, { "epoch": 0.0043776, "grad_norm": 0.9007682800292969, "learning_rate": 1.173256314153437e-08, "loss": 2.7328, "step": 384610 }, { "epoch": 0.0044032, "grad_norm": 0.8633824586868286, "learning_rate": 1.1693598624328283e-08, "loss": 2.6219, "step": 384620 }, { "epoch": 0.0044288, "grad_norm": 0.9717329144477844, "learning_rate": 1.1654698878634796e-08, "loss": 3.0145, "step": 384630 }, { "epoch": 0.0044544, "grad_norm": 0.9292768836021423, "learning_rate": 1.1615863904705926e-08, "loss": 2.869, "step": 384640 }, { "epoch": 0.00448, "grad_norm": 0.8837680816650391, "learning_rate": 1.1577093702793695e-08, "loss": 2.7977, "step": 384650 }, { "epoch": 0.0045056, "grad_norm": 1.0068362951278687, "learning_rate": 1.1538388273149014e-08, "loss": 2.587, "step": 384660 }, { "epoch": 0.0045312, "grad_norm": 0.8078030347824097, "learning_rate": 1.1499747616023904e-08, "loss": 2.6713, "step": 384670 }, { "epoch": 0.0045568, "grad_norm": 0.9854329824447632, "learning_rate": 1.1461171731668164e-08, "loss": 2.8373, "step": 384680 }, { "epoch": 0.0045824, "grad_norm": 0.9262460470199585, "learning_rate": 1.1422660620331593e-08, "loss": 2.8213, "step": 384690 }, { "epoch": 0.004608, "grad_norm": 0.9324788451194763, "learning_rate": 1.1384214282263994e-08, "loss": 3.0794, "step": 384700 }, { "epoch": 0.0046336, "grad_norm": 0.8572146892547607, "learning_rate": 1.1345832717716277e-08, "loss": 2.9648, "step": 384710 }, { "epoch": 0.0046592, "grad_norm": 1.13729727268219, "learning_rate": 1.1307515926936019e-08, "loss": 2.7676, "step": 384720 }, { "epoch": 0.0046848, "grad_norm": 0.8456854224205017, "learning_rate": 1.1269263910170802e-08, "loss": 2.6339, "step": 384730 }, { "epoch": 0.0047104, "grad_norm": 1.0135647058486938, "learning_rate": 1.1231076667671536e-08, "loss": 2.8465, "step": 384740 }, { "epoch": 0.004736, "grad_norm": 0.8716503977775574, "learning_rate": 1.119295419968247e-08, "loss": 2.777, "step": 384750 }, { "epoch": 0.0047616, "grad_norm": 0.965581476688385, "learning_rate": 1.1154896506452295e-08, "loss": 2.9248, "step": 384760 }, { "epoch": 0.0047872, "grad_norm": 0.9642717838287354, "learning_rate": 1.111690358822859e-08, "loss": 2.7124, "step": 384770 }, { "epoch": 0.0048128, "grad_norm": 1.611488938331604, "learning_rate": 1.1078975445257822e-08, "loss": 3.0717, "step": 384780 }, { "epoch": 0.0048384, "grad_norm": 1.0660786628723145, "learning_rate": 1.1041112077784244e-08, "loss": 2.891, "step": 384790 }, { "epoch": 0.004864, "grad_norm": 0.9394712448120117, "learning_rate": 1.1003313486054324e-08, "loss": 2.8662, "step": 384800 }, { "epoch": 0.0048896, "grad_norm": 0.9251168370246887, "learning_rate": 1.096557967031231e-08, "loss": 2.7766, "step": 384810 }, { "epoch": 0.0049152, "grad_norm": 0.8246572613716125, "learning_rate": 1.0927910630804672e-08, "loss": 2.6203, "step": 384820 }, { "epoch": 0.0049408, "grad_norm": 0.915466845035553, "learning_rate": 1.089030636777455e-08, "loss": 2.7721, "step": 384830 }, { "epoch": 0.0049664, "grad_norm": 1.0011500120162964, "learning_rate": 1.0852766881466193e-08, "loss": 2.6622, "step": 384840 }, { "epoch": 0.004992, "grad_norm": 0.8922411799430847, "learning_rate": 1.0815292172122738e-08, "loss": 2.8075, "step": 384850 }, { "epoch": 0.0050176, "grad_norm": 0.9827182292938232, "learning_rate": 1.0777882239987325e-08, "loss": 2.9767, "step": 384860 }, { "epoch": 0.0050432, "grad_norm": 0.8841394186019897, "learning_rate": 1.0740537085303093e-08, "loss": 2.6274, "step": 384870 }, { "epoch": 0.0050688, "grad_norm": 0.8939964175224304, "learning_rate": 1.070325670831096e-08, "loss": 2.746, "step": 384880 }, { "epoch": 0.0050944, "grad_norm": 0.864689826965332, "learning_rate": 1.0666041109254066e-08, "loss": 2.9127, "step": 384890 }, { "epoch": 0.00512, "grad_norm": 0.9006081223487854, "learning_rate": 1.0628890288372218e-08, "loss": 2.8232, "step": 384900 }, { "epoch": 0.0051456, "grad_norm": 0.8631930947303772, "learning_rate": 1.0591804245907444e-08, "loss": 2.7928, "step": 384910 }, { "epoch": 0.0051712, "grad_norm": 0.8476482033729553, "learning_rate": 1.0554782982099555e-08, "loss": 2.8683, "step": 384920 }, { "epoch": 0.0051968, "grad_norm": 1.0200262069702148, "learning_rate": 1.0517826497189465e-08, "loss": 2.6779, "step": 384930 }, { "epoch": 0.0052224, "grad_norm": 0.8360037207603455, "learning_rate": 1.0480934791416986e-08, "loss": 2.6324, "step": 384940 }, { "epoch": 0.005248, "grad_norm": 0.9051554799079895, "learning_rate": 1.0444107865019704e-08, "loss": 2.7595, "step": 384950 }, { "epoch": 0.0052736, "grad_norm": 0.8581502437591553, "learning_rate": 1.0407345718237428e-08, "loss": 3.0317, "step": 384960 }, { "epoch": 0.0052992, "grad_norm": 0.8956663608551025, "learning_rate": 1.0370648351309965e-08, "loss": 2.7901, "step": 384970 }, { "epoch": 0.0053248, "grad_norm": 0.8092256188392639, "learning_rate": 1.0334015764472682e-08, "loss": 2.7944, "step": 384980 }, { "epoch": 0.0053504, "grad_norm": 0.8246195316314697, "learning_rate": 1.029744795796428e-08, "loss": 2.5259, "step": 384990 }, { "epoch": 0.005376, "grad_norm": 0.9955987930297852, "learning_rate": 1.0260944932022343e-08, "loss": 3.0063, "step": 385000 }, { "epoch": 0.0054016, "grad_norm": 0.8345599174499512, "learning_rate": 1.0224506686883351e-08, "loss": 2.5404, "step": 385010 }, { "epoch": 0.0054272, "grad_norm": 0.8448933362960815, "learning_rate": 1.0188133222783781e-08, "loss": 2.6337, "step": 385020 }, { "epoch": 0.0054528, "grad_norm": 0.8942773938179016, "learning_rate": 1.0151824539958998e-08, "loss": 2.731, "step": 385030 }, { "epoch": 0.0054784, "grad_norm": 0.8674556612968445, "learning_rate": 1.0115580638644374e-08, "loss": 2.5748, "step": 385040 }, { "epoch": 0.005504, "grad_norm": 0.8184413313865662, "learning_rate": 1.007940151907527e-08, "loss": 2.6289, "step": 385050 }, { "epoch": 0.0055296, "grad_norm": 0.9599627256393433, "learning_rate": 1.004328718148595e-08, "loss": 2.8433, "step": 385060 }, { "epoch": 0.0055552, "grad_norm": 0.9066360592842102, "learning_rate": 1.0007237626111777e-08, "loss": 2.7071, "step": 385070 }, { "epoch": 0.0055808, "grad_norm": 0.8149898648262024, "learning_rate": 9.971252853184787e-09, "loss": 2.6908, "step": 385080 }, { "epoch": 0.0056064, "grad_norm": 0.7984746098518372, "learning_rate": 9.93533286293924e-09, "loss": 2.8097, "step": 385090 }, { "epoch": 0.005632, "grad_norm": 0.8747390508651733, "learning_rate": 9.89947765560828e-09, "loss": 2.6812, "step": 385100 }, { "epoch": 0.0056576, "grad_norm": 0.8378236889839172, "learning_rate": 9.863687231423946e-09, "loss": 2.7393, "step": 385110 }, { "epoch": 0.0056832, "grad_norm": 0.8362571001052856, "learning_rate": 9.827961590619384e-09, "loss": 2.7893, "step": 385120 }, { "epoch": 0.0057088, "grad_norm": 0.9693396091461182, "learning_rate": 9.792300733424409e-09, "loss": 2.5305, "step": 385130 }, { "epoch": 0.0057344, "grad_norm": 0.8342825174331665, "learning_rate": 9.75670466007217e-09, "loss": 2.749, "step": 385140 }, { "epoch": 0.00576, "grad_norm": 0.8164143562316895, "learning_rate": 9.72117337079137e-09, "loss": 2.8302, "step": 385150 }, { "epoch": 0.0057856, "grad_norm": 0.8719869256019592, "learning_rate": 9.685706865815159e-09, "loss": 2.7136, "step": 385160 }, { "epoch": 0.0058112, "grad_norm": 0.9913022518157959, "learning_rate": 9.650305145371131e-09, "loss": 2.7921, "step": 385170 }, { "epoch": 0.0058368, "grad_norm": 0.8574216961860657, "learning_rate": 9.614968209690213e-09, "loss": 2.54, "step": 385180 }, { "epoch": 0.0058624, "grad_norm": 0.8304299712181091, "learning_rate": 9.579696058999998e-09, "loss": 2.8396, "step": 385190 }, { "epoch": 0.005888, "grad_norm": 0.9654476046562195, "learning_rate": 9.544488693531417e-09, "loss": 2.7219, "step": 385200 }, { "epoch": 0.0059136, "grad_norm": 0.8320509791374207, "learning_rate": 9.509346113510954e-09, "loss": 2.8214, "step": 385210 }, { "epoch": 0.0059392, "grad_norm": 0.8775274753570557, "learning_rate": 9.474268319168422e-09, "loss": 2.8462, "step": 385220 }, { "epoch": 0.0059648, "grad_norm": 0.869399905204773, "learning_rate": 9.4392553107292e-09, "loss": 2.7076, "step": 385230 }, { "epoch": 0.0059904, "grad_norm": 0.9603785276412964, "learning_rate": 9.404307088420884e-09, "loss": 2.8135, "step": 385240 }, { "epoch": 0.006016, "grad_norm": 0.8478602170944214, "learning_rate": 9.369423652471066e-09, "loss": 2.4858, "step": 385250 }, { "epoch": 0.0060416, "grad_norm": 0.8076793551445007, "learning_rate": 9.334605003105124e-09, "loss": 2.6577, "step": 385260 }, { "epoch": 0.0060672, "grad_norm": 0.8855184316635132, "learning_rate": 9.299851140549543e-09, "loss": 2.761, "step": 385270 }, { "epoch": 0.0060928, "grad_norm": 0.8826419115066528, "learning_rate": 9.265162065028588e-09, "loss": 2.6981, "step": 385280 }, { "epoch": 0.0061184, "grad_norm": 0.8639009594917297, "learning_rate": 9.230537776768745e-09, "loss": 2.5702, "step": 385290 }, { "epoch": 0.006144, "grad_norm": 1.1156343221664429, "learning_rate": 9.195978275992057e-09, "loss": 2.7333, "step": 385300 }, { "epoch": 0.0061696, "grad_norm": 0.8974515795707703, "learning_rate": 9.161483562925011e-09, "loss": 2.8732, "step": 385310 }, { "epoch": 0.0061952, "grad_norm": 0.8369265198707581, "learning_rate": 9.127053637790762e-09, "loss": 2.9392, "step": 385320 }, { "epoch": 0.0062208, "grad_norm": 1.008797287940979, "learning_rate": 9.092688500812462e-09, "loss": 2.87, "step": 385330 }, { "epoch": 0.0062464, "grad_norm": 0.8633074760437012, "learning_rate": 9.05838815221216e-09, "loss": 2.8039, "step": 385340 }, { "epoch": 0.006272, "grad_norm": 0.872500479221344, "learning_rate": 9.024152592214119e-09, "loss": 2.7548, "step": 385350 }, { "epoch": 0.0062976, "grad_norm": 0.9481549859046936, "learning_rate": 8.989981821038163e-09, "loss": 2.7797, "step": 385360 }, { "epoch": 0.0063232, "grad_norm": 0.9280033111572266, "learning_rate": 8.955875838906335e-09, "loss": 2.6237, "step": 385370 }, { "epoch": 0.0063488, "grad_norm": 0.8872628211975098, "learning_rate": 8.921834646041793e-09, "loss": 2.7221, "step": 385380 }, { "epoch": 0.0063744, "grad_norm": 0.8609480261802673, "learning_rate": 8.887858242663249e-09, "loss": 2.7087, "step": 385390 }, { "epoch": 0.0064, "grad_norm": 0.8436545729637146, "learning_rate": 8.853946628991639e-09, "loss": 2.7722, "step": 385400 }, { "epoch": 0.0064256, "grad_norm": 0.9122008681297302, "learning_rate": 8.820099805245674e-09, "loss": 2.7699, "step": 385410 }, { "epoch": 0.0064512, "grad_norm": 0.9391056895256042, "learning_rate": 8.7863177716474e-09, "loss": 2.8722, "step": 385420 }, { "epoch": 0.0064768, "grad_norm": 0.8564205169677734, "learning_rate": 8.752600528414424e-09, "loss": 2.7247, "step": 385430 }, { "epoch": 0.0065024, "grad_norm": 0.9338430166244507, "learning_rate": 8.718948075765455e-09, "loss": 2.8887, "step": 385440 }, { "epoch": 0.006528, "grad_norm": 0.9015585780143738, "learning_rate": 8.685360413918098e-09, "loss": 2.7977, "step": 385450 }, { "epoch": 0.0065536, "grad_norm": 0.8500043749809265, "learning_rate": 8.651837543091069e-09, "loss": 2.8473, "step": 385460 }, { "epoch": 0.0065792, "grad_norm": 0.937322199344635, "learning_rate": 8.618379463501969e-09, "loss": 2.6874, "step": 385470 }, { "epoch": 0.0066048, "grad_norm": 1.0507313013076782, "learning_rate": 8.584986175367294e-09, "loss": 2.7508, "step": 385480 }, { "epoch": 0.0066304, "grad_norm": 1.0766923427581787, "learning_rate": 8.551657678902425e-09, "loss": 3.16, "step": 385490 }, { "epoch": 0.006656, "grad_norm": 0.8749479651451111, "learning_rate": 8.518393974326078e-09, "loss": 2.7919, "step": 385500 }, { "epoch": 0.0066816, "grad_norm": 0.8517564535140991, "learning_rate": 8.485195061851415e-09, "loss": 2.9759, "step": 385510 }, { "epoch": 0.0067072, "grad_norm": 0.8509846329689026, "learning_rate": 8.452060941696039e-09, "loss": 2.624, "step": 385520 }, { "epoch": 0.0067328, "grad_norm": 0.8342828750610352, "learning_rate": 8.418991614073113e-09, "loss": 2.865, "step": 385530 }, { "epoch": 0.0067584, "grad_norm": 0.9884021878242493, "learning_rate": 8.385987079196911e-09, "loss": 2.838, "step": 385540 }, { "epoch": 0.006784, "grad_norm": 1.0739033222198486, "learning_rate": 8.353047337282816e-09, "loss": 2.8312, "step": 385550 }, { "epoch": 0.0068096, "grad_norm": 0.9345453381538391, "learning_rate": 8.32017238854288e-09, "loss": 2.8137, "step": 385560 }, { "epoch": 0.0068352, "grad_norm": 1.0096890926361084, "learning_rate": 8.287362233192486e-09, "loss": 2.8659, "step": 385570 }, { "epoch": 0.0068608, "grad_norm": 0.8419768810272217, "learning_rate": 8.254616871442577e-09, "loss": 2.8079, "step": 385580 }, { "epoch": 0.0068864, "grad_norm": 0.9172250628471375, "learning_rate": 8.221936303505206e-09, "loss": 2.5539, "step": 385590 }, { "epoch": 0.006912, "grad_norm": 0.921923816204071, "learning_rate": 8.189320529593536e-09, "loss": 2.7774, "step": 385600 }, { "epoch": 0.0069376, "grad_norm": 0.9287407398223877, "learning_rate": 8.156769549919618e-09, "loss": 2.8769, "step": 385610 }, { "epoch": 0.0069632, "grad_norm": 0.9460777044296265, "learning_rate": 8.124283364692176e-09, "loss": 2.5531, "step": 385620 }, { "epoch": 0.0069888, "grad_norm": 0.8769863843917847, "learning_rate": 8.09186197412326e-09, "loss": 2.8367, "step": 385630 }, { "epoch": 0.0070144, "grad_norm": 0.8659569621086121, "learning_rate": 8.059505378423814e-09, "loss": 2.6476, "step": 385640 }, { "epoch": 0.00704, "grad_norm": 0.9156970381736755, "learning_rate": 8.027213577802562e-09, "loss": 2.8312, "step": 385650 }, { "epoch": 0.0070656, "grad_norm": 0.9063670039176941, "learning_rate": 7.994986572469332e-09, "loss": 2.7513, "step": 385660 }, { "epoch": 0.0070912, "grad_norm": 0.832244873046875, "learning_rate": 7.962824362633959e-09, "loss": 3.0379, "step": 385670 }, { "epoch": 0.0071168, "grad_norm": 0.8555971384048462, "learning_rate": 7.930726948502942e-09, "loss": 2.8279, "step": 385680 }, { "epoch": 0.0071424, "grad_norm": 0.9305070042610168, "learning_rate": 7.898694330286117e-09, "loss": 2.7843, "step": 385690 }, { "epoch": 0.007168, "grad_norm": 0.8651496171951294, "learning_rate": 7.86672650818998e-09, "loss": 2.861, "step": 385700 }, { "epoch": 0.0071936, "grad_norm": 0.8726244568824768, "learning_rate": 7.83482348242326e-09, "loss": 2.8026, "step": 385710 }, { "epoch": 0.0072192, "grad_norm": 1.0259429216384888, "learning_rate": 7.802985253192452e-09, "loss": 2.7581, "step": 385720 }, { "epoch": 0.0072448, "grad_norm": 0.8651759624481201, "learning_rate": 7.771211820702951e-09, "loss": 2.8596, "step": 385730 }, { "epoch": 0.0072704, "grad_norm": 0.7996675968170166, "learning_rate": 7.739503185161257e-09, "loss": 2.7112, "step": 385740 }, { "epoch": 0.007296, "grad_norm": 0.8932444453239441, "learning_rate": 7.70785934677387e-09, "loss": 2.7973, "step": 385750 }, { "epoch": 0.0073216, "grad_norm": 0.8214940428733826, "learning_rate": 7.676280305745076e-09, "loss": 2.6512, "step": 385760 }, { "epoch": 0.0073472, "grad_norm": 0.8738315105438232, "learning_rate": 7.644766062280263e-09, "loss": 2.9408, "step": 385770 }, { "epoch": 0.0073728, "grad_norm": 0.9103987216949463, "learning_rate": 7.6133166165826e-09, "loss": 2.8409, "step": 385780 }, { "epoch": 0.0073984, "grad_norm": 0.8198259472846985, "learning_rate": 7.581931968858592e-09, "loss": 2.7544, "step": 385790 }, { "epoch": 0.007424, "grad_norm": 0.8749385476112366, "learning_rate": 7.550612119309186e-09, "loss": 2.9321, "step": 385800 }, { "epoch": 0.0074496, "grad_norm": 0.9700576663017273, "learning_rate": 7.519357068137556e-09, "loss": 3.1223, "step": 385810 }, { "epoch": 0.0074752, "grad_norm": 0.8513635993003845, "learning_rate": 7.488166815547982e-09, "loss": 2.9397, "step": 385820 }, { "epoch": 0.0075008, "grad_norm": 0.8975957632064819, "learning_rate": 7.457041361742523e-09, "loss": 2.9462, "step": 385830 }, { "epoch": 0.0075264, "grad_norm": 0.898309588432312, "learning_rate": 7.425980706922131e-09, "loss": 2.7045, "step": 385840 }, { "epoch": 0.007552, "grad_norm": 0.885250449180603, "learning_rate": 7.394984851287757e-09, "loss": 2.6797, "step": 385850 }, { "epoch": 0.0075776, "grad_norm": 0.9060420393943787, "learning_rate": 7.36405379504146e-09, "loss": 2.8487, "step": 385860 }, { "epoch": 0.0076032, "grad_norm": 1.0211125612258911, "learning_rate": 7.33318753838419e-09, "loss": 2.8568, "step": 385870 }, { "epoch": 0.0076288, "grad_norm": 0.9845327138900757, "learning_rate": 7.3023860815146785e-09, "loss": 2.8571, "step": 385880 }, { "epoch": 0.0076544, "grad_norm": 0.9556991457939148, "learning_rate": 7.271649424633875e-09, "loss": 2.8615, "step": 385890 }, { "epoch": 0.00768, "grad_norm": 0.9555646181106567, "learning_rate": 7.240977567941621e-09, "loss": 2.8224, "step": 385900 }, { "epoch": 0.0077056, "grad_norm": 0.9294506311416626, "learning_rate": 7.210370511634424e-09, "loss": 2.8863, "step": 385910 }, { "epoch": 0.0077312, "grad_norm": 1.064402461051941, "learning_rate": 7.179828255913235e-09, "loss": 2.6665, "step": 385920 }, { "epoch": 0.0077568, "grad_norm": 0.9097037315368652, "learning_rate": 7.149350800974564e-09, "loss": 2.8925, "step": 385930 }, { "epoch": 0.0077824, "grad_norm": 0.8860047459602356, "learning_rate": 7.11893814701603e-09, "loss": 2.9408, "step": 385940 }, { "epoch": 0.007808, "grad_norm": 0.8685168623924255, "learning_rate": 7.088590294236363e-09, "loss": 2.8546, "step": 385950 }, { "epoch": 0.0078336, "grad_norm": 1.1070082187652588, "learning_rate": 7.058307242830964e-09, "loss": 2.8278, "step": 385960 }, { "epoch": 0.0078592, "grad_norm": 0.9200841188430786, "learning_rate": 7.02808899299634e-09, "loss": 2.8149, "step": 385970 }, { "epoch": 0.0078848, "grad_norm": 0.8343262076377869, "learning_rate": 6.997935544929002e-09, "loss": 2.698, "step": 385980 }, { "epoch": 0.0079104, "grad_norm": 0.8580247163772583, "learning_rate": 6.967846898824349e-09, "loss": 2.731, "step": 385990 }, { "epoch": 0.007936, "grad_norm": 0.8122140169143677, "learning_rate": 6.9378230548777795e-09, "loss": 2.9657, "step": 386000 }, { "epoch": 0.0079616, "grad_norm": 0.8971158862113953, "learning_rate": 6.907864013282473e-09, "loss": 2.6805, "step": 386010 }, { "epoch": 0.0079872, "grad_norm": 0.838918149471283, "learning_rate": 6.8779697742349385e-09, "loss": 2.7895, "step": 386020 }, { "epoch": 0.0080128, "grad_norm": 0.8710182309150696, "learning_rate": 6.848140337927245e-09, "loss": 2.7942, "step": 386030 }, { "epoch": 0.0080384, "grad_norm": 0.9167349934577942, "learning_rate": 6.8183757045536814e-09, "loss": 2.7125, "step": 386040 }, { "epoch": 0.008064, "grad_norm": 0.8362902402877808, "learning_rate": 6.788675874307427e-09, "loss": 2.8172, "step": 386050 }, { "epoch": 0.0080896, "grad_norm": 0.894923746585846, "learning_rate": 6.759040847379439e-09, "loss": 2.812, "step": 386060 }, { "epoch": 0.0081152, "grad_norm": 0.8989906907081604, "learning_rate": 6.729470623965118e-09, "loss": 2.7928, "step": 386070 }, { "epoch": 0.0081408, "grad_norm": 0.8187607526779175, "learning_rate": 6.699965204253201e-09, "loss": 2.8695, "step": 386080 }, { "epoch": 0.0081664, "grad_norm": 0.8822208046913147, "learning_rate": 6.670524588435756e-09, "loss": 2.8895, "step": 386090 }, { "epoch": 0.008192, "grad_norm": 1.0125659704208374, "learning_rate": 6.6411487767048536e-09, "loss": 2.844, "step": 386100 }, { "epoch": 0.0082176, "grad_norm": 0.8912609815597534, "learning_rate": 6.611837769249229e-09, "loss": 2.8651, "step": 386110 }, { "epoch": 0.0082432, "grad_norm": 0.8563557267189026, "learning_rate": 6.5825915662609545e-09, "loss": 3.0011, "step": 386120 }, { "epoch": 0.0082688, "grad_norm": 0.9428209066390991, "learning_rate": 6.553410167928764e-09, "loss": 2.9243, "step": 386130 }, { "epoch": 0.0082944, "grad_norm": 0.8030146956443787, "learning_rate": 6.524293574440288e-09, "loss": 2.642, "step": 386140 }, { "epoch": 0.00832, "grad_norm": 1.0423529148101807, "learning_rate": 6.495241785987594e-09, "loss": 2.6446, "step": 386150 }, { "epoch": 0.0083456, "grad_norm": 0.8779008984565735, "learning_rate": 6.466254802756089e-09, "loss": 2.7824, "step": 386160 }, { "epoch": 0.0083712, "grad_norm": 0.9012126326560974, "learning_rate": 6.437332624935622e-09, "loss": 2.9641, "step": 386170 }, { "epoch": 0.0083968, "grad_norm": 0.8011431694030762, "learning_rate": 6.408475252712709e-09, "loss": 2.6587, "step": 386180 }, { "epoch": 0.0084224, "grad_norm": 0.8200453519821167, "learning_rate": 6.37968268627498e-09, "loss": 2.7427, "step": 386190 }, { "epoch": 0.008448, "grad_norm": 0.8053104877471924, "learning_rate": 6.35095492580895e-09, "loss": 2.8421, "step": 386200 }, { "epoch": 0.0084736, "grad_norm": 0.8799724578857422, "learning_rate": 6.322291971501138e-09, "loss": 2.9132, "step": 386210 }, { "epoch": 0.0084992, "grad_norm": 0.8447178602218628, "learning_rate": 6.29369382353695e-09, "loss": 2.6805, "step": 386220 }, { "epoch": 0.0085248, "grad_norm": 1.064247965812683, "learning_rate": 6.265160482101795e-09, "loss": 2.7748, "step": 386230 }, { "epoch": 0.0085504, "grad_norm": 1.0052443742752075, "learning_rate": 6.236691947382189e-09, "loss": 2.64, "step": 386240 }, { "epoch": 0.008576, "grad_norm": 0.8665882349014282, "learning_rate": 6.20828821956132e-09, "loss": 2.8524, "step": 386250 }, { "epoch": 0.0086016, "grad_norm": 1.1968636512756348, "learning_rate": 6.179949298823485e-09, "loss": 2.9833, "step": 386260 }, { "epoch": 0.0086272, "grad_norm": 0.8371992111206055, "learning_rate": 6.151675185351869e-09, "loss": 2.9702, "step": 386270 }, { "epoch": 0.0086528, "grad_norm": 1.2501473426818848, "learning_rate": 6.12346587933188e-09, "loss": 2.6401, "step": 386280 }, { "epoch": 0.0086784, "grad_norm": 0.8900216817855835, "learning_rate": 6.095321380944486e-09, "loss": 2.876, "step": 386290 }, { "epoch": 0.008704, "grad_norm": 0.8437191843986511, "learning_rate": 6.067241690372872e-09, "loss": 2.6665, "step": 386300 }, { "epoch": 0.0087296, "grad_norm": 0.8470526933670044, "learning_rate": 6.0392268077991144e-09, "loss": 2.7538, "step": 386310 }, { "epoch": 0.0087552, "grad_norm": 0.9214003682136536, "learning_rate": 6.011276733404181e-09, "loss": 2.94, "step": 386320 }, { "epoch": 0.0087808, "grad_norm": 0.8950738310813904, "learning_rate": 5.983391467371258e-09, "loss": 2.7049, "step": 386330 }, { "epoch": 0.0088064, "grad_norm": 1.002614974975586, "learning_rate": 5.955571009879091e-09, "loss": 2.8402, "step": 386340 }, { "epoch": 0.008832, "grad_norm": 0.8181468844413757, "learning_rate": 5.927815361109757e-09, "loss": 2.6851, "step": 386350 }, { "epoch": 0.0088576, "grad_norm": 0.8565642237663269, "learning_rate": 5.900124521242001e-09, "loss": 2.9283, "step": 386360 }, { "epoch": 0.0088832, "grad_norm": 0.8393332362174988, "learning_rate": 5.87249849045568e-09, "loss": 2.8434, "step": 386370 }, { "epoch": 0.0089088, "grad_norm": 1.0983866453170776, "learning_rate": 5.844937268930651e-09, "loss": 2.5073, "step": 386380 }, { "epoch": 0.0089344, "grad_norm": 0.8617362976074219, "learning_rate": 5.817440856844547e-09, "loss": 2.7499, "step": 386390 }, { "epoch": 0.00896, "grad_norm": 1.1430997848510742, "learning_rate": 5.790009254376117e-09, "loss": 2.3819, "step": 386400 }, { "epoch": 0.0089856, "grad_norm": 0.9085351228713989, "learning_rate": 5.762642461704105e-09, "loss": 2.8567, "step": 386410 }, { "epoch": 0.0090112, "grad_norm": 0.8668885231018066, "learning_rate": 5.735340479003926e-09, "loss": 2.5349, "step": 386420 }, { "epoch": 0.0090368, "grad_norm": 0.889430820941925, "learning_rate": 5.708103306455437e-09, "loss": 2.59, "step": 386430 }, { "epoch": 0.0090624, "grad_norm": 0.8331588506698608, "learning_rate": 5.680930944232943e-09, "loss": 2.7464, "step": 386440 }, { "epoch": 0.009088, "grad_norm": 1.0136723518371582, "learning_rate": 5.6538233925129696e-09, "loss": 2.8137, "step": 386450 }, { "epoch": 0.0091136, "grad_norm": 1.0179630517959595, "learning_rate": 5.626780651472042e-09, "loss": 2.8365, "step": 386460 }, { "epoch": 0.0091392, "grad_norm": 0.8828528523445129, "learning_rate": 5.5998027212855745e-09, "loss": 2.7854, "step": 386470 }, { "epoch": 0.0091648, "grad_norm": 0.8529970645904541, "learning_rate": 5.572889602127873e-09, "loss": 2.7648, "step": 386480 }, { "epoch": 0.0091904, "grad_norm": 1.0227841138839722, "learning_rate": 5.5460412941743535e-09, "loss": 2.9323, "step": 386490 }, { "epoch": 0.009216, "grad_norm": 0.8670503497123718, "learning_rate": 5.51925779759821e-09, "loss": 2.8406, "step": 386500 }, { "epoch": 0.0092416, "grad_norm": 0.9416361451148987, "learning_rate": 5.492539112573747e-09, "loss": 2.6077, "step": 386510 }, { "epoch": 0.0092672, "grad_norm": 0.9003791809082031, "learning_rate": 5.46588523927305e-09, "loss": 2.8239, "step": 386520 }, { "epoch": 0.0092928, "grad_norm": 0.8470885157585144, "learning_rate": 5.439296177870423e-09, "loss": 2.8302, "step": 386530 }, { "epoch": 0.0093184, "grad_norm": 0.7898721098899841, "learning_rate": 5.412771928537952e-09, "loss": 3.0986, "step": 386540 }, { "epoch": 0.009344, "grad_norm": 1.0238288640975952, "learning_rate": 5.386312491447721e-09, "loss": 2.6611, "step": 386550 }, { "epoch": 0.0093696, "grad_norm": 0.946372389793396, "learning_rate": 5.359917866769593e-09, "loss": 3.0128, "step": 386560 }, { "epoch": 0.0093952, "grad_norm": 0.9483940601348877, "learning_rate": 5.333588054677874e-09, "loss": 2.9513, "step": 386570 }, { "epoch": 0.0094208, "grad_norm": 0.8583656549453735, "learning_rate": 5.307323055340208e-09, "loss": 2.7049, "step": 386580 }, { "epoch": 0.0094464, "grad_norm": 0.9228368401527405, "learning_rate": 5.281122868928679e-09, "loss": 2.8593, "step": 386590 }, { "epoch": 0.009472, "grad_norm": 0.8989757299423218, "learning_rate": 5.254987495613151e-09, "loss": 2.9189, "step": 386600 }, { "epoch": 0.0094976, "grad_norm": 0.8962975740432739, "learning_rate": 5.228916935562378e-09, "loss": 2.962, "step": 386610 }, { "epoch": 0.0095232, "grad_norm": 0.9401028156280518, "learning_rate": 5.202911188946224e-09, "loss": 2.6842, "step": 386620 }, { "epoch": 0.0095488, "grad_norm": 0.8630843758583069, "learning_rate": 5.1769702559323345e-09, "loss": 2.7827, "step": 386630 }, { "epoch": 0.0095744, "grad_norm": 0.8755511045455933, "learning_rate": 5.151094136689461e-09, "loss": 2.8589, "step": 386640 }, { "epoch": 0.0096, "grad_norm": 0.9144850969314575, "learning_rate": 5.125282831386358e-09, "loss": 2.9058, "step": 386650 }, { "epoch": 0.0096256, "grad_norm": 1.1520332098007202, "learning_rate": 5.099536340188449e-09, "loss": 2.8374, "step": 386660 }, { "epoch": 0.0096512, "grad_norm": 0.8791086673736572, "learning_rate": 5.0738546632644884e-09, "loss": 2.9027, "step": 386670 }, { "epoch": 0.0096768, "grad_norm": 0.8748539686203003, "learning_rate": 5.048237800779899e-09, "loss": 2.7928, "step": 386680 }, { "epoch": 0.0097024, "grad_norm": 0.9864922165870667, "learning_rate": 5.022685752901213e-09, "loss": 2.8103, "step": 386690 }, { "epoch": 0.009728, "grad_norm": 0.9455912709236145, "learning_rate": 4.997198519793855e-09, "loss": 2.769, "step": 386700 }, { "epoch": 0.0097536, "grad_norm": 0.888746440410614, "learning_rate": 4.971776101623249e-09, "loss": 2.8654, "step": 386710 }, { "epoch": 0.0097792, "grad_norm": 0.893747091293335, "learning_rate": 4.9464184985548165e-09, "loss": 2.6519, "step": 386720 }, { "epoch": 0.0098048, "grad_norm": 0.94758540391922, "learning_rate": 4.921125710751762e-09, "loss": 2.7864, "step": 386730 }, { "epoch": 0.0098304, "grad_norm": 0.8922165632247925, "learning_rate": 4.895897738379507e-09, "loss": 2.7507, "step": 386740 }, { "epoch": 0.009856, "grad_norm": 0.9372248649597168, "learning_rate": 4.870734581600145e-09, "loss": 3.05, "step": 386750 }, { "epoch": 0.0098816, "grad_norm": 0.8324198722839355, "learning_rate": 4.845636240577989e-09, "loss": 2.825, "step": 386760 }, { "epoch": 0.0099072, "grad_norm": 0.9756770133972168, "learning_rate": 4.820602715476241e-09, "loss": 2.788, "step": 386770 }, { "epoch": 0.0099328, "grad_norm": 0.9384273290634155, "learning_rate": 4.7956340064558845e-09, "loss": 2.8935, "step": 386780 }, { "epoch": 0.0099584, "grad_norm": 1.1069155931472778, "learning_rate": 4.7707301136790115e-09, "loss": 2.8348, "step": 386790 }, { "epoch": 0.009984, "grad_norm": 0.9077380299568176, "learning_rate": 4.745891037308825e-09, "loss": 2.6885, "step": 386800 }, { "epoch": 0.0100096, "grad_norm": 0.8402254581451416, "learning_rate": 4.721116777504087e-09, "loss": 2.8061, "step": 386810 }, { "epoch": 0.0100352, "grad_norm": 1.0005227327346802, "learning_rate": 4.696407334426889e-09, "loss": 2.9427, "step": 386820 }, { "epoch": 0.0100608, "grad_norm": 0.9441819190979004, "learning_rate": 4.671762708237104e-09, "loss": 2.9165, "step": 386830 }, { "epoch": 0.0100864, "grad_norm": 0.9227555990219116, "learning_rate": 4.647182899093494e-09, "loss": 2.8142, "step": 386840 }, { "epoch": 0.010112, "grad_norm": 0.9993862509727478, "learning_rate": 4.622667907158152e-09, "loss": 2.7771, "step": 386850 }, { "epoch": 0.0101376, "grad_norm": 0.8235544562339783, "learning_rate": 4.598217732586507e-09, "loss": 2.7897, "step": 386860 }, { "epoch": 0.0101632, "grad_norm": 1.116302490234375, "learning_rate": 4.573832375540654e-09, "loss": 2.8966, "step": 386870 }, { "epoch": 0.0101888, "grad_norm": 1.115843653678894, "learning_rate": 4.549511836176024e-09, "loss": 2.6773, "step": 386880 }, { "epoch": 0.0102144, "grad_norm": 0.8531378507614136, "learning_rate": 4.525256114651378e-09, "loss": 2.768, "step": 386890 }, { "epoch": 0.01024, "grad_norm": 0.8992795348167419, "learning_rate": 4.501065211124367e-09, "loss": 2.8419, "step": 386900 }, { "epoch": 0.0102656, "grad_norm": 0.854404628276825, "learning_rate": 4.4769391257515335e-09, "loss": 2.7966, "step": 386910 }, { "epoch": 0.0102912, "grad_norm": 0.9563814401626587, "learning_rate": 4.452877858688309e-09, "loss": 2.9572, "step": 386920 }, { "epoch": 0.0103168, "grad_norm": 0.9855906963348389, "learning_rate": 4.428881410092345e-09, "loss": 3.0165, "step": 386930 }, { "epoch": 0.0103424, "grad_norm": 0.9181191921234131, "learning_rate": 4.404949780119072e-09, "loss": 2.7304, "step": 386940 }, { "epoch": 0.010368, "grad_norm": 0.8448920845985413, "learning_rate": 4.381082968922812e-09, "loss": 3.3701, "step": 386950 }, { "epoch": 0.0103936, "grad_norm": 0.858128547668457, "learning_rate": 4.3572809766589955e-09, "loss": 2.7952, "step": 386960 }, { "epoch": 0.0104192, "grad_norm": 0.9867866635322571, "learning_rate": 4.333543803480833e-09, "loss": 2.6922, "step": 386970 }, { "epoch": 0.0104448, "grad_norm": 0.9108624458312988, "learning_rate": 4.309871449543757e-09, "loss": 2.7734, "step": 386980 }, { "epoch": 0.0104704, "grad_norm": 0.8694579005241394, "learning_rate": 4.286263915000977e-09, "loss": 2.6938, "step": 386990 }, { "epoch": 0.010496, "grad_norm": 1.2774171829223633, "learning_rate": 4.262721200004594e-09, "loss": 2.8745, "step": 387000 }, { "epoch": 0.0105216, "grad_norm": 0.8079336285591125, "learning_rate": 4.239243304707819e-09, "loss": 2.8396, "step": 387010 }, { "epoch": 0.0105472, "grad_norm": 0.8595162034034729, "learning_rate": 4.215830229262752e-09, "loss": 2.7019, "step": 387020 }, { "epoch": 0.0105728, "grad_norm": 0.8310526013374329, "learning_rate": 4.192481973822604e-09, "loss": 2.6945, "step": 387030 }, { "epoch": 0.0105984, "grad_norm": 0.8665131330490112, "learning_rate": 4.169198538536146e-09, "loss": 2.8103, "step": 387040 }, { "epoch": 0.010624, "grad_norm": 1.0719866752624512, "learning_rate": 4.145979923557697e-09, "loss": 2.6769, "step": 387050 }, { "epoch": 0.0106496, "grad_norm": 0.882271945476532, "learning_rate": 4.122826129034918e-09, "loss": 2.5137, "step": 387060 }, { "epoch": 0.0106752, "grad_norm": 0.9389876127243042, "learning_rate": 4.099737155118799e-09, "loss": 2.7549, "step": 387070 }, { "epoch": 0.0107008, "grad_norm": 0.9221500754356384, "learning_rate": 4.0767130019592205e-09, "loss": 2.772, "step": 387080 }, { "epoch": 0.0107264, "grad_norm": 1.0113903284072876, "learning_rate": 4.053753669706062e-09, "loss": 2.8788, "step": 387090 }, { "epoch": 0.010752, "grad_norm": 0.9402956366539001, "learning_rate": 4.030859158508094e-09, "loss": 2.9928, "step": 387100 }, { "epoch": 0.0107776, "grad_norm": 1.042495846748352, "learning_rate": 4.008029468512975e-09, "loss": 2.6864, "step": 387110 }, { "epoch": 0.0108032, "grad_norm": 0.9778586626052856, "learning_rate": 3.9852645998683655e-09, "loss": 2.8738, "step": 387120 }, { "epoch": 0.0108288, "grad_norm": 0.8743399977684021, "learning_rate": 3.962564552723036e-09, "loss": 2.8137, "step": 387130 }, { "epoch": 0.0108544, "grad_norm": 1.0120466947555542, "learning_rate": 3.9399293272246454e-09, "loss": 2.8413, "step": 387140 }, { "epoch": 0.01088, "grad_norm": 1.0799509286880493, "learning_rate": 3.917358923518632e-09, "loss": 2.8225, "step": 387150 }, { "epoch": 0.0109056, "grad_norm": 0.8523684144020081, "learning_rate": 3.894853341752658e-09, "loss": 3.0947, "step": 387160 }, { "epoch": 0.0109312, "grad_norm": 0.8532201051712036, "learning_rate": 3.87241258207105e-09, "loss": 2.7884, "step": 387170 }, { "epoch": 0.0109568, "grad_norm": 0.9230226278305054, "learning_rate": 3.850036644620359e-09, "loss": 4.316, "step": 387180 }, { "epoch": 0.0109824, "grad_norm": 0.9613077044487, "learning_rate": 3.827725529544913e-09, "loss": 2.7998, "step": 387190 }, { "epoch": 0.011008, "grad_norm": 0.8755303025245667, "learning_rate": 3.805479236991261e-09, "loss": 2.9013, "step": 387200 }, { "epoch": 0.0110336, "grad_norm": 0.9897620677947998, "learning_rate": 3.783297767101512e-09, "loss": 2.7886, "step": 387210 }, { "epoch": 0.0110592, "grad_norm": 0.8251174688339233, "learning_rate": 3.761181120021106e-09, "loss": 2.631, "step": 387220 }, { "epoch": 0.0110848, "grad_norm": 0.9176217317581177, "learning_rate": 3.739129295892152e-09, "loss": 2.9436, "step": 387230 }, { "epoch": 0.0111104, "grad_norm": 0.898712158203125, "learning_rate": 3.7171422948589775e-09, "loss": 2.8665, "step": 387240 }, { "epoch": 0.011136, "grad_norm": 0.9259202480316162, "learning_rate": 3.6952201170625812e-09, "loss": 2.9726, "step": 387250 }, { "epoch": 0.0111616, "grad_norm": 0.9974482655525208, "learning_rate": 3.673362762646182e-09, "loss": 2.8083, "step": 387260 }, { "epoch": 0.0111872, "grad_norm": 0.8985611796379089, "learning_rate": 3.6515702317518887e-09, "loss": 2.7686, "step": 387270 }, { "epoch": 0.0112128, "grad_norm": 0.8255080580711365, "learning_rate": 3.629842524519589e-09, "loss": 2.8083, "step": 387280 }, { "epoch": 0.0112384, "grad_norm": 1.1185472011566162, "learning_rate": 3.6081796410913915e-09, "loss": 2.9557, "step": 387290 }, { "epoch": 0.011264, "grad_norm": 1.0950020551681519, "learning_rate": 3.586581581607185e-09, "loss": 2.9844, "step": 387300 }, { "epoch": 0.0112896, "grad_norm": 0.9587875008583069, "learning_rate": 3.5650483462079668e-09, "loss": 2.7589, "step": 387310 }, { "epoch": 0.0113152, "grad_norm": 1.4893113374710083, "learning_rate": 3.543579935032515e-09, "loss": 2.8362, "step": 387320 }, { "epoch": 0.0113408, "grad_norm": 0.9100893139839172, "learning_rate": 3.522176348219608e-09, "loss": 2.8392, "step": 387330 }, { "epoch": 0.0113664, "grad_norm": 0.8505615592002869, "learning_rate": 3.5008375859091335e-09, "loss": 2.7204, "step": 387340 }, { "epoch": 0.011392, "grad_norm": 0.9293752908706665, "learning_rate": 3.4795636482387594e-09, "loss": 2.9837, "step": 387350 }, { "epoch": 0.0114176, "grad_norm": 0.8997573852539062, "learning_rate": 3.4583545353472634e-09, "loss": 2.9411, "step": 387360 }, { "epoch": 0.0114432, "grad_norm": 0.9138945937156677, "learning_rate": 3.437210247371203e-09, "loss": 2.849, "step": 387370 }, { "epoch": 0.0114688, "grad_norm": 0.8847377300262451, "learning_rate": 3.416130784447136e-09, "loss": 2.7314, "step": 387380 }, { "epoch": 0.0114944, "grad_norm": 0.964458167552948, "learning_rate": 3.3951161467138395e-09, "loss": 2.7469, "step": 387390 }, { "epoch": 0.01152, "grad_norm": 0.8855646848678589, "learning_rate": 3.3741663343056506e-09, "loss": 3.2469, "step": 387400 }, { "epoch": 0.0115456, "grad_norm": 0.8688324689865112, "learning_rate": 3.3532813473602378e-09, "loss": 2.8578, "step": 387410 }, { "epoch": 0.0115712, "grad_norm": 0.8910910487174988, "learning_rate": 3.332461186010827e-09, "loss": 2.77, "step": 387420 }, { "epoch": 0.0115968, "grad_norm": 0.9396799802780151, "learning_rate": 3.311705850395086e-09, "loss": 2.9254, "step": 387430 }, { "epoch": 0.0116224, "grad_norm": 0.8901739716529846, "learning_rate": 3.291015340645132e-09, "loss": 2.8419, "step": 387440 }, { "epoch": 0.011648, "grad_norm": 0.8625337481498718, "learning_rate": 3.270389656896411e-09, "loss": 2.8308, "step": 387450 }, { "epoch": 0.0116736, "grad_norm": 0.8716880679130554, "learning_rate": 3.249828799282151e-09, "loss": 3.0492, "step": 387460 }, { "epoch": 0.0116992, "grad_norm": 0.867069661617279, "learning_rate": 3.229332767935578e-09, "loss": 2.7628, "step": 387470 }, { "epoch": 0.0117248, "grad_norm": 0.9232250452041626, "learning_rate": 3.208901562991029e-09, "loss": 2.902, "step": 387480 }, { "epoch": 0.0117504, "grad_norm": 0.8574000000953674, "learning_rate": 3.1885351845784007e-09, "loss": 2.9177, "step": 387490 }, { "epoch": 0.011776, "grad_norm": 0.8821601867675781, "learning_rate": 3.1682336328331397e-09, "loss": 2.878, "step": 387500 } ], "logging_steps": 10, "max_steps": 390625, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 8.72816366276397e+19, "train_batch_size": 4, "trial_name": null, "trial_params": null }