{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0066833751044277, "eval_steps": 500, "global_step": 5400, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00556947925368978, "grad_norm": 19.463690451045245, "learning_rate": 7.154213036565979e-07, "loss": 8.2534, "step": 10 }, { "epoch": 0.01113895850737956, "grad_norm": 19.73356911714459, "learning_rate": 1.51033386327504e-06, "loss": 8.1778, "step": 20 }, { "epoch": 0.01670843776106934, "grad_norm": 10.029986538109684, "learning_rate": 2.3052464228934817e-06, "loss": 7.6937, "step": 30 }, { "epoch": 0.02227791701475912, "grad_norm": 7.379827099032999, "learning_rate": 3.100158982511924e-06, "loss": 6.9205, "step": 40 }, { "epoch": 0.0278473962684489, "grad_norm": 3.5728211270030625, "learning_rate": 3.895071542130366e-06, "loss": 6.3948, "step": 50 }, { "epoch": 0.03341687552213868, "grad_norm": 2.8392085477918747, "learning_rate": 4.689984101748808e-06, "loss": 6.0575, "step": 60 }, { "epoch": 0.03898635477582846, "grad_norm": 3.3563782303356495, "learning_rate": 5.48489666136725e-06, "loss": 5.7294, "step": 70 }, { "epoch": 0.04455583402951824, "grad_norm": 3.826365209612832, "learning_rate": 6.279809220985691e-06, "loss": 5.2614, "step": 80 }, { "epoch": 0.05012531328320802, "grad_norm": 3.965749199888462, "learning_rate": 7.0747217806041336e-06, "loss": 4.658, "step": 90 }, { "epoch": 0.0556947925368978, "grad_norm": 5.147891128586454, "learning_rate": 7.869634340222575e-06, "loss": 4.0477, "step": 100 }, { "epoch": 0.06126427179058758, "grad_norm": 3.695764472570234, "learning_rate": 8.664546899841019e-06, "loss": 3.5584, "step": 110 }, { "epoch": 0.06683375104427736, "grad_norm": 4.026692299224468, "learning_rate": 9.45945945945946e-06, "loss": 3.2227, "step": 120 }, { "epoch": 0.07240323029796714, "grad_norm": 5.389627945808325, "learning_rate": 1.02543720190779e-05, "loss": 3.0255, "step": 130 }, { "epoch": 0.07797270955165692, "grad_norm": 5.554941152889833, "learning_rate": 1.1049284578696344e-05, "loss": 2.8932, "step": 140 }, { "epoch": 0.0835421888053467, "grad_norm": 4.8986035107973995, "learning_rate": 1.1844197138314786e-05, "loss": 2.7959, "step": 150 }, { "epoch": 0.08911166805903648, "grad_norm": 4.56461997585786, "learning_rate": 1.2639109697933228e-05, "loss": 2.7067, "step": 160 }, { "epoch": 0.09468114731272625, "grad_norm": 9.192995633454027, "learning_rate": 1.3434022257551671e-05, "loss": 2.6194, "step": 170 }, { "epoch": 0.10025062656641603, "grad_norm": 8.520097501414543, "learning_rate": 1.4228934817170111e-05, "loss": 2.5516, "step": 180 }, { "epoch": 0.10582010582010581, "grad_norm": 14.274847317292076, "learning_rate": 1.5023847376788555e-05, "loss": 2.4789, "step": 190 }, { "epoch": 0.1113895850737956, "grad_norm": 19.214437222876512, "learning_rate": 1.5818759936406995e-05, "loss": 2.4363, "step": 200 }, { "epoch": 0.11695906432748537, "grad_norm": 16.394738746567327, "learning_rate": 1.6613672496025438e-05, "loss": 2.3902, "step": 210 }, { "epoch": 0.12252854358117515, "grad_norm": 18.247117883573495, "learning_rate": 1.740858505564388e-05, "loss": 2.3452, "step": 220 }, { "epoch": 0.12809802283486493, "grad_norm": 15.142493870476338, "learning_rate": 1.820349761526232e-05, "loss": 2.3042, "step": 230 }, { "epoch": 0.1336675020885547, "grad_norm": 8.204095855953893, "learning_rate": 1.8998410174880765e-05, "loss": 2.2554, "step": 240 }, { "epoch": 0.1392369813422445, "grad_norm": 12.061502490559853, "learning_rate": 1.9793322734499205e-05, "loss": 2.2062, "step": 250 }, { "epoch": 0.14480646059593427, "grad_norm": 13.078924954972894, "learning_rate": 2.058823529411765e-05, "loss": 2.1414, "step": 260 }, { "epoch": 0.15037593984962405, "grad_norm": 13.325114907384508, "learning_rate": 2.1383147853736092e-05, "loss": 2.059, "step": 270 }, { "epoch": 0.15594541910331383, "grad_norm": 7.4709725407687255, "learning_rate": 2.2178060413354532e-05, "loss": 1.9772, "step": 280 }, { "epoch": 0.1615148983570036, "grad_norm": 19.25846753081156, "learning_rate": 2.2972972972972976e-05, "loss": 1.9011, "step": 290 }, { "epoch": 0.1670843776106934, "grad_norm": 13.01475345365111, "learning_rate": 2.3767885532591416e-05, "loss": 1.8118, "step": 300 }, { "epoch": 0.17265385686438317, "grad_norm": 6.508219489647502, "learning_rate": 2.4562798092209856e-05, "loss": 1.7365, "step": 310 }, { "epoch": 0.17822333611807295, "grad_norm": 8.638960149014848, "learning_rate": 2.53577106518283e-05, "loss": 1.6377, "step": 320 }, { "epoch": 0.18379281537176273, "grad_norm": 5.254562295866414, "learning_rate": 2.6152623211446743e-05, "loss": 1.5737, "step": 330 }, { "epoch": 0.1893622946254525, "grad_norm": 8.547148285512923, "learning_rate": 2.6947535771065186e-05, "loss": 1.5303, "step": 340 }, { "epoch": 0.1949317738791423, "grad_norm": 6.278158376274084, "learning_rate": 2.7742448330683623e-05, "loss": 1.4716, "step": 350 }, { "epoch": 0.20050125313283207, "grad_norm": 8.26795777275924, "learning_rate": 2.8537360890302066e-05, "loss": 1.4151, "step": 360 }, { "epoch": 0.20607073238652185, "grad_norm": 4.425290147294405, "learning_rate": 2.933227344992051e-05, "loss": 1.368, "step": 370 }, { "epoch": 0.21164021164021163, "grad_norm": 6.848350616045537, "learning_rate": 3.0127186009538953e-05, "loss": 1.3383, "step": 380 }, { "epoch": 0.2172096908939014, "grad_norm": 4.869312013250087, "learning_rate": 3.092209856915739e-05, "loss": 1.2751, "step": 390 }, { "epoch": 0.2227791701475912, "grad_norm": 7.264173089824176, "learning_rate": 3.171701112877583e-05, "loss": 1.2448, "step": 400 }, { "epoch": 0.22834864940128097, "grad_norm": 7.343411468837996, "learning_rate": 3.251192368839428e-05, "loss": 1.1956, "step": 410 }, { "epoch": 0.23391812865497075, "grad_norm": 8.139260855658039, "learning_rate": 3.330683624801272e-05, "loss": 1.1626, "step": 420 }, { "epoch": 0.23948760790866053, "grad_norm": 4.441795212879553, "learning_rate": 3.410174880763116e-05, "loss": 1.1313, "step": 430 }, { "epoch": 0.2450570871623503, "grad_norm": 7.0655482262903115, "learning_rate": 3.489666136724961e-05, "loss": 1.1016, "step": 440 }, { "epoch": 0.2506265664160401, "grad_norm": 5.343797555692575, "learning_rate": 3.569157392686804e-05, "loss": 1.0682, "step": 450 }, { "epoch": 0.25619604566972987, "grad_norm": 4.837151670486665, "learning_rate": 3.648648648648649e-05, "loss": 1.0388, "step": 460 }, { "epoch": 0.26176552492341965, "grad_norm": 3.9144650767846345, "learning_rate": 3.7281399046104934e-05, "loss": 1.0, "step": 470 }, { "epoch": 0.2673350041771094, "grad_norm": 6.192111002424375, "learning_rate": 3.8076311605723374e-05, "loss": 0.9819, "step": 480 }, { "epoch": 0.2729044834307992, "grad_norm": 4.664894996941613, "learning_rate": 3.8871224165341814e-05, "loss": 0.9512, "step": 490 }, { "epoch": 0.278473962684489, "grad_norm": 5.728600939316023, "learning_rate": 3.9666136724960254e-05, "loss": 0.9292, "step": 500 }, { "epoch": 0.28404344193817876, "grad_norm": 7.358090778625861, "learning_rate": 4.04610492845787e-05, "loss": 0.9018, "step": 510 }, { "epoch": 0.28961292119186854, "grad_norm": 6.346761431472806, "learning_rate": 4.125596184419714e-05, "loss": 0.8805, "step": 520 }, { "epoch": 0.2951824004455583, "grad_norm": 3.586718245895629, "learning_rate": 4.205087440381558e-05, "loss": 0.8568, "step": 530 }, { "epoch": 0.3007518796992481, "grad_norm": 5.430039185536721, "learning_rate": 4.284578696343403e-05, "loss": 0.8284, "step": 540 }, { "epoch": 0.3063213589529379, "grad_norm": 6.430553831376545, "learning_rate": 4.364069952305246e-05, "loss": 0.8128, "step": 550 }, { "epoch": 0.31189083820662766, "grad_norm": 5.48500374978629, "learning_rate": 4.443561208267091e-05, "loss": 0.7842, "step": 560 }, { "epoch": 0.31746031746031744, "grad_norm": 5.0124287229570745, "learning_rate": 4.523052464228935e-05, "loss": 0.7803, "step": 570 }, { "epoch": 0.3230297967140072, "grad_norm": 4.960089415525465, "learning_rate": 4.6025437201907795e-05, "loss": 0.7489, "step": 580 }, { "epoch": 0.328599275967697, "grad_norm": 4.429208732066051, "learning_rate": 4.6820349761526235e-05, "loss": 0.7243, "step": 590 }, { "epoch": 0.3341687552213868, "grad_norm": 5.1268739971720905, "learning_rate": 4.7615262321144675e-05, "loss": 0.7044, "step": 600 }, { "epoch": 0.33973823447507656, "grad_norm": 4.915480885708807, "learning_rate": 4.841017488076312e-05, "loss": 0.6939, "step": 610 }, { "epoch": 0.34530771372876634, "grad_norm": 5.102518466488775, "learning_rate": 4.9205087440381556e-05, "loss": 0.6767, "step": 620 }, { "epoch": 0.3508771929824561, "grad_norm": 4.364437305811626, "learning_rate": 5e-05, "loss": 0.6467, "step": 630 }, { "epoch": 0.3564466722361459, "grad_norm": 4.372236216369743, "learning_rate": 5.079491255961845e-05, "loss": 0.6234, "step": 640 }, { "epoch": 0.3620161514898357, "grad_norm": 4.439347899189148, "learning_rate": 5.158982511923689e-05, "loss": 0.5961, "step": 650 }, { "epoch": 0.36758563074352546, "grad_norm": 3.1594832305447316, "learning_rate": 5.2384737678855336e-05, "loss": 0.5786, "step": 660 }, { "epoch": 0.37315510999721524, "grad_norm": 4.473968649061542, "learning_rate": 5.317965023847376e-05, "loss": 0.5711, "step": 670 }, { "epoch": 0.378724589250905, "grad_norm": 3.4873086420651456, "learning_rate": 5.397456279809221e-05, "loss": 0.5503, "step": 680 }, { "epoch": 0.3842940685045948, "grad_norm": 3.831386918783317, "learning_rate": 5.4769475357710656e-05, "loss": 0.529, "step": 690 }, { "epoch": 0.3898635477582846, "grad_norm": 3.653243174902386, "learning_rate": 5.5564387917329096e-05, "loss": 0.5132, "step": 700 }, { "epoch": 0.39543302701197436, "grad_norm": 3.5521167023222047, "learning_rate": 5.635930047694754e-05, "loss": 0.4877, "step": 710 }, { "epoch": 0.40100250626566414, "grad_norm": 3.8712402877933934, "learning_rate": 5.715421303656598e-05, "loss": 0.4833, "step": 720 }, { "epoch": 0.4065719855193539, "grad_norm": 3.379905907827214, "learning_rate": 5.794912559618442e-05, "loss": 0.4615, "step": 730 }, { "epoch": 0.4121414647730437, "grad_norm": 2.5936051096516417, "learning_rate": 5.8744038155802863e-05, "loss": 0.4444, "step": 740 }, { "epoch": 0.4177109440267335, "grad_norm": 3.322838857318257, "learning_rate": 5.9538950715421304e-05, "loss": 0.4318, "step": 750 }, { "epoch": 0.42328042328042326, "grad_norm": 3.3843216521985253, "learning_rate": 6.033386327503975e-05, "loss": 0.4145, "step": 760 }, { "epoch": 0.42884990253411304, "grad_norm": 3.0846350440353434, "learning_rate": 6.11287758346582e-05, "loss": 0.3998, "step": 770 }, { "epoch": 0.4344193817878028, "grad_norm": 2.7361994149983566, "learning_rate": 6.192368839427664e-05, "loss": 0.3863, "step": 780 }, { "epoch": 0.4399888610414926, "grad_norm": 3.27515261932524, "learning_rate": 6.271860095389508e-05, "loss": 0.375, "step": 790 }, { "epoch": 0.4455583402951824, "grad_norm": 3.029805585678618, "learning_rate": 6.351351351351352e-05, "loss": 0.364, "step": 800 }, { "epoch": 0.45112781954887216, "grad_norm": 2.2590750215054287, "learning_rate": 6.430842607313196e-05, "loss": 0.3453, "step": 810 }, { "epoch": 0.45669729880256194, "grad_norm": 3.1086383281500805, "learning_rate": 6.51033386327504e-05, "loss": 0.3329, "step": 820 }, { "epoch": 0.4622667780562517, "grad_norm": 2.312028278866729, "learning_rate": 6.589825119236884e-05, "loss": 0.324, "step": 830 }, { "epoch": 0.4678362573099415, "grad_norm": 2.8653110069877226, "learning_rate": 6.669316375198729e-05, "loss": 0.3113, "step": 840 }, { "epoch": 0.4734057365636313, "grad_norm": 2.0346455435163224, "learning_rate": 6.748807631160573e-05, "loss": 0.2976, "step": 850 }, { "epoch": 0.47897521581732105, "grad_norm": 2.2187482142761854, "learning_rate": 6.828298887122417e-05, "loss": 0.2838, "step": 860 }, { "epoch": 0.48454469507101083, "grad_norm": 2.5760581999823926, "learning_rate": 6.907790143084261e-05, "loss": 0.2712, "step": 870 }, { "epoch": 0.4901141743247006, "grad_norm": 2.336167318842384, "learning_rate": 6.987281399046105e-05, "loss": 0.2659, "step": 880 }, { "epoch": 0.4956836535783904, "grad_norm": 1.942101679442901, "learning_rate": 7.066772655007949e-05, "loss": 0.2524, "step": 890 }, { "epoch": 0.5012531328320802, "grad_norm": 1.9676135441330864, "learning_rate": 7.146263910969793e-05, "loss": 0.242, "step": 900 }, { "epoch": 0.50682261208577, "grad_norm": 2.42452293659568, "learning_rate": 7.225755166931639e-05, "loss": 0.2303, "step": 910 }, { "epoch": 0.5123920913394597, "grad_norm": 2.0689938731425754, "learning_rate": 7.305246422893483e-05, "loss": 0.2204, "step": 920 }, { "epoch": 0.5179615705931495, "grad_norm": 2.171275305780617, "learning_rate": 7.384737678855325e-05, "loss": 0.2108, "step": 930 }, { "epoch": 0.5235310498468393, "grad_norm": 2.018182012710717, "learning_rate": 7.46422893481717e-05, "loss": 0.1988, "step": 940 }, { "epoch": 0.5291005291005291, "grad_norm": 1.8435919023282064, "learning_rate": 7.543720190779015e-05, "loss": 0.1951, "step": 950 }, { "epoch": 0.5346700083542189, "grad_norm": 1.6966065845617362, "learning_rate": 7.623211446740859e-05, "loss": 0.1868, "step": 960 }, { "epoch": 0.5402394876079086, "grad_norm": 2.107747858389665, "learning_rate": 7.702702702702703e-05, "loss": 0.1752, "step": 970 }, { "epoch": 0.5458089668615984, "grad_norm": 1.626232566427141, "learning_rate": 7.782193958664548e-05, "loss": 0.1726, "step": 980 }, { "epoch": 0.5513784461152882, "grad_norm": 1.710543294805638, "learning_rate": 7.861685214626392e-05, "loss": 0.1625, "step": 990 }, { "epoch": 0.556947925368978, "grad_norm": 1.568201961955694, "learning_rate": 7.941176470588235e-05, "loss": 0.1552, "step": 1000 }, { "epoch": 0.5625174046226677, "grad_norm": 1.4211321412129432, "learning_rate": 8.02066772655008e-05, "loss": 0.1485, "step": 1010 }, { "epoch": 0.5680868838763575, "grad_norm": 1.4970140750834295, "learning_rate": 8.100158982511924e-05, "loss": 0.1434, "step": 1020 }, { "epoch": 0.5736563631300473, "grad_norm": 1.853797381238809, "learning_rate": 8.179650238473768e-05, "loss": 0.1356, "step": 1030 }, { "epoch": 0.5792258423837371, "grad_norm": 1.475678872579458, "learning_rate": 8.259141494435613e-05, "loss": 0.1298, "step": 1040 }, { "epoch": 0.5847953216374269, "grad_norm": 1.690324091049369, "learning_rate": 8.338632750397457e-05, "loss": 0.1258, "step": 1050 }, { "epoch": 0.5903648008911166, "grad_norm": 1.908861076416452, "learning_rate": 8.418124006359301e-05, "loss": 0.1249, "step": 1060 }, { "epoch": 0.5959342801448064, "grad_norm": 1.236244782970751, "learning_rate": 8.497615262321145e-05, "loss": 0.1171, "step": 1070 }, { "epoch": 0.6015037593984962, "grad_norm": 1.2774164058752888, "learning_rate": 8.57710651828299e-05, "loss": 0.1114, "step": 1080 }, { "epoch": 0.607073238652186, "grad_norm": 0.8951848955749875, "learning_rate": 8.656597774244833e-05, "loss": 0.1059, "step": 1090 }, { "epoch": 0.6126427179058758, "grad_norm": 1.0572189677651873, "learning_rate": 8.736089030206677e-05, "loss": 0.1012, "step": 1100 }, { "epoch": 0.6182121971595655, "grad_norm": 0.9583280956588, "learning_rate": 8.815580286168523e-05, "loss": 0.0973, "step": 1110 }, { "epoch": 0.6237816764132553, "grad_norm": 0.966004358455493, "learning_rate": 8.895071542130367e-05, "loss": 0.098, "step": 1120 }, { "epoch": 0.6293511556669451, "grad_norm": 0.8285031685005042, "learning_rate": 8.97456279809221e-05, "loss": 0.0897, "step": 1130 }, { "epoch": 0.6349206349206349, "grad_norm": 1.3954391622958595, "learning_rate": 9.054054054054055e-05, "loss": 0.087, "step": 1140 }, { "epoch": 0.6404901141743247, "grad_norm": 0.6608729564295608, "learning_rate": 9.133545310015899e-05, "loss": 0.0825, "step": 1150 }, { "epoch": 0.6460595934280144, "grad_norm": 0.8111382270197721, "learning_rate": 9.213036565977743e-05, "loss": 0.0764, "step": 1160 }, { "epoch": 0.6516290726817042, "grad_norm": 0.6891511752919594, "learning_rate": 9.292527821939587e-05, "loss": 0.0745, "step": 1170 }, { "epoch": 0.657198551935394, "grad_norm": 0.8999745942051316, "learning_rate": 9.372019077901432e-05, "loss": 0.0718, "step": 1180 }, { "epoch": 0.6627680311890838, "grad_norm": 0.7165073695545044, "learning_rate": 9.451510333863276e-05, "loss": 0.0699, "step": 1190 }, { "epoch": 0.6683375104427736, "grad_norm": 0.5691546082373657, "learning_rate": 9.531001589825119e-05, "loss": 0.0663, "step": 1200 }, { "epoch": 0.6739069896964633, "grad_norm": 0.618154476760873, "learning_rate": 9.610492845786964e-05, "loss": 0.065, "step": 1210 }, { "epoch": 0.6794764689501531, "grad_norm": 0.5460710019961567, "learning_rate": 9.689984101748808e-05, "loss": 0.0622, "step": 1220 }, { "epoch": 0.6850459482038429, "grad_norm": 0.44149906211171397, "learning_rate": 9.769475357710652e-05, "loss": 0.0582, "step": 1230 }, { "epoch": 0.6906154274575327, "grad_norm": 0.5860441580160569, "learning_rate": 9.848966613672496e-05, "loss": 0.0567, "step": 1240 }, { "epoch": 0.6961849067112225, "grad_norm": 0.8453684625264357, "learning_rate": 9.928457869634342e-05, "loss": 0.0564, "step": 1250 }, { "epoch": 0.7017543859649122, "grad_norm": 2.1379268435693555, "learning_rate": 9.999999807244224e-05, "loss": 0.0564, "step": 1260 }, { "epoch": 0.707323865218602, "grad_norm": 3.3074045726639083, "learning_rate": 9.999976676568994e-05, "loss": 0.1234, "step": 1270 }, { "epoch": 0.7128933444722918, "grad_norm": 0.6698300987807628, "learning_rate": 9.99991499494276e-05, "loss": 0.0776, "step": 1280 }, { "epoch": 0.7184628237259816, "grad_norm": 0.40986381929677007, "learning_rate": 9.999814762841102e-05, "loss": 0.0563, "step": 1290 }, { "epoch": 0.7240323029796714, "grad_norm": 0.5823587617428915, "learning_rate": 9.999675981036832e-05, "loss": 0.05, "step": 1300 }, { "epoch": 0.7296017822333611, "grad_norm": 0.36554660320089016, "learning_rate": 9.999498650599987e-05, "loss": 0.0505, "step": 1310 }, { "epoch": 0.7351712614870509, "grad_norm": 0.3490628453059406, "learning_rate": 9.999282772897828e-05, "loss": 0.0474, "step": 1320 }, { "epoch": 0.7407407407407407, "grad_norm": 0.3963213598340753, "learning_rate": 9.99902834959482e-05, "loss": 0.0462, "step": 1330 }, { "epoch": 0.7463102199944305, "grad_norm": 0.5565743717844861, "learning_rate": 9.998735382652624e-05, "loss": 0.0442, "step": 1340 }, { "epoch": 0.7518796992481203, "grad_norm": 0.29567277754356086, "learning_rate": 9.99840387433008e-05, "loss": 0.0439, "step": 1350 }, { "epoch": 0.75744917850181, "grad_norm": 0.42840154636629013, "learning_rate": 9.998033827183193e-05, "loss": 0.0432, "step": 1360 }, { "epoch": 0.7630186577554998, "grad_norm": 0.3843192755820257, "learning_rate": 9.997625244065113e-05, "loss": 0.0419, "step": 1370 }, { "epoch": 0.7685881370091896, "grad_norm": 0.2794549966082612, "learning_rate": 9.997178128126101e-05, "loss": 0.04, "step": 1380 }, { "epoch": 0.7741576162628794, "grad_norm": 0.3169631435418986, "learning_rate": 9.99669248281353e-05, "loss": 0.0395, "step": 1390 }, { "epoch": 0.7797270955165692, "grad_norm": 0.2751145242333607, "learning_rate": 9.996168311871832e-05, "loss": 0.039, "step": 1400 }, { "epoch": 0.7852965747702589, "grad_norm": 0.2873246557323452, "learning_rate": 9.995605619342484e-05, "loss": 0.0386, "step": 1410 }, { "epoch": 0.7908660540239487, "grad_norm": 0.26158401303739276, "learning_rate": 9.995004409563969e-05, "loss": 0.0364, "step": 1420 }, { "epoch": 0.7964355332776385, "grad_norm": 0.3512702948952268, "learning_rate": 9.994364687171756e-05, "loss": 0.0358, "step": 1430 }, { "epoch": 0.8020050125313283, "grad_norm": 0.30233442060518095, "learning_rate": 9.993686457098246e-05, "loss": 0.0355, "step": 1440 }, { "epoch": 0.8075744917850181, "grad_norm": 0.31354227279951397, "learning_rate": 9.992969724572746e-05, "loss": 0.0355, "step": 1450 }, { "epoch": 0.8131439710387078, "grad_norm": 0.257194967469868, "learning_rate": 9.992214495121428e-05, "loss": 0.0357, "step": 1460 }, { "epoch": 0.8187134502923976, "grad_norm": 0.3479392485987129, "learning_rate": 9.99142077456728e-05, "loss": 0.0345, "step": 1470 }, { "epoch": 0.8242829295460874, "grad_norm": 0.24684398703549518, "learning_rate": 9.990588569030068e-05, "loss": 0.0337, "step": 1480 }, { "epoch": 0.8298524087997772, "grad_norm": 0.23435105711158805, "learning_rate": 9.989717884926286e-05, "loss": 0.0328, "step": 1490 }, { "epoch": 0.835421888053467, "grad_norm": 0.226826410374761, "learning_rate": 9.988808728969104e-05, "loss": 0.0325, "step": 1500 }, { "epoch": 0.8409913673071567, "grad_norm": 0.23919961520195954, "learning_rate": 9.987861108168317e-05, "loss": 0.0319, "step": 1510 }, { "epoch": 0.8465608465608465, "grad_norm": 0.262516092950579, "learning_rate": 9.986875029830302e-05, "loss": 0.0311, "step": 1520 }, { "epoch": 0.8521303258145363, "grad_norm": 0.27505170240477905, "learning_rate": 9.985850501557941e-05, "loss": 0.0306, "step": 1530 }, { "epoch": 0.8576998050682261, "grad_norm": 0.2136353900303048, "learning_rate": 9.984787531250581e-05, "loss": 0.0306, "step": 1540 }, { "epoch": 0.8632692843219159, "grad_norm": 0.2681884993661502, "learning_rate": 9.983686127103963e-05, "loss": 0.0304, "step": 1550 }, { "epoch": 0.8688387635756056, "grad_norm": 0.27064206276631336, "learning_rate": 9.982546297610163e-05, "loss": 0.0302, "step": 1560 }, { "epoch": 0.8744082428292954, "grad_norm": 0.22756357273919053, "learning_rate": 9.981368051557522e-05, "loss": 0.0296, "step": 1570 }, { "epoch": 0.8799777220829852, "grad_norm": 0.23817203224591024, "learning_rate": 9.980151398030586e-05, "loss": 0.029, "step": 1580 }, { "epoch": 0.885547201336675, "grad_norm": 0.23612451468006537, "learning_rate": 9.978896346410025e-05, "loss": 0.0283, "step": 1590 }, { "epoch": 0.8911166805903648, "grad_norm": 0.21633179926653082, "learning_rate": 9.977602906372577e-05, "loss": 0.0284, "step": 1600 }, { "epoch": 0.8966861598440545, "grad_norm": 0.3007834433966076, "learning_rate": 9.97627108789095e-05, "loss": 0.0276, "step": 1610 }, { "epoch": 0.9022556390977443, "grad_norm": 0.22874451896032774, "learning_rate": 9.974900901233772e-05, "loss": 0.0275, "step": 1620 }, { "epoch": 0.9078251183514341, "grad_norm": 0.21229440559613427, "learning_rate": 9.973492356965488e-05, "loss": 0.0272, "step": 1630 }, { "epoch": 0.9133945976051239, "grad_norm": 0.21316799793517352, "learning_rate": 9.972045465946295e-05, "loss": 0.0266, "step": 1640 }, { "epoch": 0.9189640768588136, "grad_norm": 0.21840470782406107, "learning_rate": 9.97056023933205e-05, "loss": 0.0262, "step": 1650 }, { "epoch": 0.9245335561125034, "grad_norm": 0.2296202067200069, "learning_rate": 9.969036688574184e-05, "loss": 0.0262, "step": 1660 }, { "epoch": 0.9301030353661932, "grad_norm": 0.20035916457466751, "learning_rate": 9.967474825419619e-05, "loss": 0.0256, "step": 1670 }, { "epoch": 0.935672514619883, "grad_norm": 0.2392722746437584, "learning_rate": 9.965874661910675e-05, "loss": 0.0252, "step": 1680 }, { "epoch": 0.9412419938735728, "grad_norm": 0.2373750894769379, "learning_rate": 9.964236210384972e-05, "loss": 0.025, "step": 1690 }, { "epoch": 0.9468114731272625, "grad_norm": 0.23565128407439787, "learning_rate": 9.962559483475342e-05, "loss": 0.0245, "step": 1700 }, { "epoch": 0.9523809523809523, "grad_norm": 0.17651355447696596, "learning_rate": 9.960844494109731e-05, "loss": 0.024, "step": 1710 }, { "epoch": 0.9579504316346421, "grad_norm": 0.19685563495858988, "learning_rate": 9.959091255511093e-05, "loss": 0.0237, "step": 1720 }, { "epoch": 0.9635199108883319, "grad_norm": 0.1904961294977808, "learning_rate": 9.957299781197295e-05, "loss": 0.0235, "step": 1730 }, { "epoch": 0.9690893901420217, "grad_norm": 0.21714296341973274, "learning_rate": 9.955470084981009e-05, "loss": 0.0232, "step": 1740 }, { "epoch": 0.9746588693957114, "grad_norm": 0.21953603915286493, "learning_rate": 9.953602180969606e-05, "loss": 0.0227, "step": 1750 }, { "epoch": 0.9802283486494012, "grad_norm": 0.21497187075207344, "learning_rate": 9.951696083565048e-05, "loss": 0.0224, "step": 1760 }, { "epoch": 0.985797827903091, "grad_norm": 0.18608055976176674, "learning_rate": 9.949751807463777e-05, "loss": 0.0226, "step": 1770 }, { "epoch": 0.9913673071567808, "grad_norm": 0.20352070118783105, "learning_rate": 9.947769367656603e-05, "loss": 0.0223, "step": 1780 }, { "epoch": 0.9969367864104706, "grad_norm": 0.24791255800751252, "learning_rate": 9.945748779428585e-05, "loss": 0.0217, "step": 1790 }, { "epoch": 1.002227791701476, "grad_norm": 0.22472692732126306, "learning_rate": 9.943690058358915e-05, "loss": 0.0217, "step": 1800 }, { "epoch": 1.0077972709551657, "grad_norm": 0.2324141522417415, "learning_rate": 9.941593220320797e-05, "loss": 0.021, "step": 1810 }, { "epoch": 1.0133667502088555, "grad_norm": 0.19233986715335943, "learning_rate": 9.939458281481327e-05, "loss": 0.0212, "step": 1820 }, { "epoch": 1.0189362294625453, "grad_norm": 0.19005795401708103, "learning_rate": 9.937285258301368e-05, "loss": 0.0204, "step": 1830 }, { "epoch": 1.024505708716235, "grad_norm": 0.21691243705398805, "learning_rate": 9.935074167535419e-05, "loss": 0.0205, "step": 1840 }, { "epoch": 1.0300751879699248, "grad_norm": 0.21307344896990474, "learning_rate": 9.932825026231488e-05, "loss": 0.0197, "step": 1850 }, { "epoch": 1.0356446672236146, "grad_norm": 0.2554320699991988, "learning_rate": 9.930537851730966e-05, "loss": 0.0193, "step": 1860 }, { "epoch": 1.0412141464773044, "grad_norm": 0.204769243946618, "learning_rate": 9.928212661668483e-05, "loss": 0.0197, "step": 1870 }, { "epoch": 1.0467836257309941, "grad_norm": 0.19859219088449717, "learning_rate": 9.925849473971784e-05, "loss": 0.0196, "step": 1880 }, { "epoch": 1.052353104984684, "grad_norm": 0.20983760283366612, "learning_rate": 9.923448306861577e-05, "loss": 0.019, "step": 1890 }, { "epoch": 1.0579225842383737, "grad_norm": 0.20589867040322582, "learning_rate": 9.921009178851405e-05, "loss": 0.0193, "step": 1900 }, { "epoch": 1.0634920634920635, "grad_norm": 0.18568807424493425, "learning_rate": 9.918532108747496e-05, "loss": 0.0189, "step": 1910 }, { "epoch": 1.0690615427457533, "grad_norm": 0.24088752127036372, "learning_rate": 9.916017115648622e-05, "loss": 0.0186, "step": 1920 }, { "epoch": 1.074631021999443, "grad_norm": 0.1926030304649344, "learning_rate": 9.913464218945946e-05, "loss": 0.0185, "step": 1930 }, { "epoch": 1.0802005012531328, "grad_norm": 0.8741730355575924, "learning_rate": 9.910873438322884e-05, "loss": 0.0199, "step": 1940 }, { "epoch": 1.0857699805068226, "grad_norm": 0.2217866848455536, "learning_rate": 9.908244793754936e-05, "loss": 0.0193, "step": 1950 }, { "epoch": 1.0913394597605124, "grad_norm": 0.21213745327394049, "learning_rate": 9.905578305509546e-05, "loss": 0.0183, "step": 1960 }, { "epoch": 1.0969089390142022, "grad_norm": 0.18251989541031047, "learning_rate": 9.902873994145945e-05, "loss": 0.0178, "step": 1970 }, { "epoch": 1.102478418267892, "grad_norm": 0.17043242205288903, "learning_rate": 9.900131880514982e-05, "loss": 0.0173, "step": 1980 }, { "epoch": 1.1080478975215817, "grad_norm": 0.18442275031618716, "learning_rate": 9.897351985758974e-05, "loss": 0.0169, "step": 1990 }, { "epoch": 1.1136173767752715, "grad_norm": 0.169313148723845, "learning_rate": 9.89453433131154e-05, "loss": 0.0169, "step": 2000 }, { "epoch": 1.1191868560289613, "grad_norm": 0.1764961917184802, "learning_rate": 9.891678938897431e-05, "loss": 0.0167, "step": 2010 }, { "epoch": 1.124756335282651, "grad_norm": 0.17888846393164137, "learning_rate": 9.888785830532368e-05, "loss": 0.0165, "step": 2020 }, { "epoch": 1.1303258145363408, "grad_norm": 0.19125391501293337, "learning_rate": 9.885855028522873e-05, "loss": 0.0167, "step": 2030 }, { "epoch": 1.1358952937900306, "grad_norm": 0.19194405473426313, "learning_rate": 9.882886555466091e-05, "loss": 0.0161, "step": 2040 }, { "epoch": 1.1414647730437204, "grad_norm": 0.1770624411187416, "learning_rate": 9.879880434249621e-05, "loss": 0.0161, "step": 2050 }, { "epoch": 1.1470342522974102, "grad_norm": 0.16975918892019828, "learning_rate": 9.876836688051337e-05, "loss": 0.0161, "step": 2060 }, { "epoch": 1.1526037315511, "grad_norm": 0.16854837841564632, "learning_rate": 9.873755340339211e-05, "loss": 0.0163, "step": 2070 }, { "epoch": 1.1581732108047897, "grad_norm": 0.17269429721308055, "learning_rate": 9.870636414871131e-05, "loss": 0.0159, "step": 2080 }, { "epoch": 1.1637426900584795, "grad_norm": 0.17625877797958917, "learning_rate": 9.867479935694717e-05, "loss": 0.0153, "step": 2090 }, { "epoch": 1.1693121693121693, "grad_norm": 0.15960923595525844, "learning_rate": 9.864285927147138e-05, "loss": 0.0152, "step": 2100 }, { "epoch": 1.174881648565859, "grad_norm": 0.15697254246300588, "learning_rate": 9.861054413854922e-05, "loss": 0.015, "step": 2110 }, { "epoch": 1.1804511278195489, "grad_norm": 0.16827296886706583, "learning_rate": 9.857785420733768e-05, "loss": 0.0149, "step": 2120 }, { "epoch": 1.1860206070732386, "grad_norm": 0.15126212242845452, "learning_rate": 9.854478972988351e-05, "loss": 0.0146, "step": 2130 }, { "epoch": 1.1915900863269284, "grad_norm": 0.18190859691476494, "learning_rate": 9.85113509611213e-05, "loss": 0.0147, "step": 2140 }, { "epoch": 1.1971595655806182, "grad_norm": 0.1620227168907316, "learning_rate": 9.847753815887155e-05, "loss": 0.0144, "step": 2150 }, { "epoch": 1.202729044834308, "grad_norm": 0.16437934378165805, "learning_rate": 9.84433515838386e-05, "loss": 0.0144, "step": 2160 }, { "epoch": 1.2082985240879978, "grad_norm": 0.19971461246923847, "learning_rate": 9.840879149960866e-05, "loss": 0.0144, "step": 2170 }, { "epoch": 1.2138680033416875, "grad_norm": 0.17545008123882608, "learning_rate": 9.837385817264783e-05, "loss": 0.014, "step": 2180 }, { "epoch": 1.2194374825953773, "grad_norm": 0.17002175092464197, "learning_rate": 9.833855187229993e-05, "loss": 0.0141, "step": 2190 }, { "epoch": 1.225006961849067, "grad_norm": 0.1502742530729085, "learning_rate": 9.830287287078453e-05, "loss": 0.0138, "step": 2200 }, { "epoch": 1.2305764411027569, "grad_norm": 0.1733361600396075, "learning_rate": 9.826682144319481e-05, "loss": 0.0136, "step": 2210 }, { "epoch": 1.2361459203564467, "grad_norm": 0.1703011609444242, "learning_rate": 9.823039786749542e-05, "loss": 0.0135, "step": 2220 }, { "epoch": 1.2417153996101364, "grad_norm": 0.25652792710234223, "learning_rate": 9.819360242452037e-05, "loss": 0.0137, "step": 2230 }, { "epoch": 1.2472848788638262, "grad_norm": 0.18955265882691177, "learning_rate": 9.815643539797085e-05, "loss": 0.0135, "step": 2240 }, { "epoch": 1.252854358117516, "grad_norm": 0.176670947546437, "learning_rate": 9.811889707441304e-05, "loss": 0.0133, "step": 2250 }, { "epoch": 1.2584238373712058, "grad_norm": 0.1611843791331948, "learning_rate": 9.808098774327589e-05, "loss": 0.0127, "step": 2260 }, { "epoch": 1.2639933166248956, "grad_norm": 0.17553791221607762, "learning_rate": 9.804270769684894e-05, "loss": 0.0128, "step": 2270 }, { "epoch": 1.2695627958785853, "grad_norm": 0.1713258488893283, "learning_rate": 9.800405723027998e-05, "loss": 0.0125, "step": 2280 }, { "epoch": 1.2751322751322751, "grad_norm": 0.1491188978760988, "learning_rate": 9.796503664157289e-05, "loss": 0.0125, "step": 2290 }, { "epoch": 1.280701754385965, "grad_norm": 0.16382367249207125, "learning_rate": 9.792564623158518e-05, "loss": 0.0123, "step": 2300 }, { "epoch": 1.2862712336396547, "grad_norm": 0.14990536572540042, "learning_rate": 9.788588630402586e-05, "loss": 0.012, "step": 2310 }, { "epoch": 1.2918407128933445, "grad_norm": 0.1612176660850985, "learning_rate": 9.784575716545294e-05, "loss": 0.0118, "step": 2320 }, { "epoch": 1.2974101921470342, "grad_norm": 0.13562541267893746, "learning_rate": 9.780525912527118e-05, "loss": 0.0118, "step": 2330 }, { "epoch": 1.302979671400724, "grad_norm": 0.1680412513721527, "learning_rate": 9.776439249572962e-05, "loss": 0.0114, "step": 2340 }, { "epoch": 1.3085491506544138, "grad_norm": 0.15829618023893888, "learning_rate": 9.77231575919192e-05, "loss": 0.0115, "step": 2350 }, { "epoch": 1.3141186299081036, "grad_norm": 0.16260390780626038, "learning_rate": 9.768155473177037e-05, "loss": 0.0116, "step": 2360 }, { "epoch": 1.3196881091617934, "grad_norm": 0.16275068223487438, "learning_rate": 9.763958423605057e-05, "loss": 0.0115, "step": 2370 }, { "epoch": 1.3252575884154831, "grad_norm": 0.20119872670951966, "learning_rate": 9.759724642836184e-05, "loss": 0.0109, "step": 2380 }, { "epoch": 1.330827067669173, "grad_norm": 0.172255342957574, "learning_rate": 9.755454163513824e-05, "loss": 0.0108, "step": 2390 }, { "epoch": 1.3363965469228627, "grad_norm": 0.15223068522445793, "learning_rate": 9.75114701856434e-05, "loss": 0.0109, "step": 2400 }, { "epoch": 1.3419660261765525, "grad_norm": 0.14759093427536293, "learning_rate": 9.746803241196791e-05, "loss": 0.0108, "step": 2410 }, { "epoch": 1.3475355054302423, "grad_norm": 0.14365720679126023, "learning_rate": 9.742422864902684e-05, "loss": 0.0103, "step": 2420 }, { "epoch": 1.353104984683932, "grad_norm": 0.14640480563703434, "learning_rate": 9.738005923455714e-05, "loss": 0.0104, "step": 2430 }, { "epoch": 1.3586744639376218, "grad_norm": 0.13917727110739087, "learning_rate": 9.733552450911494e-05, "loss": 0.0104, "step": 2440 }, { "epoch": 1.3642439431913116, "grad_norm": 0.14296325607296143, "learning_rate": 9.729062481607307e-05, "loss": 0.01, "step": 2450 }, { "epoch": 1.3698134224450014, "grad_norm": 0.1513321418311315, "learning_rate": 9.72453605016183e-05, "loss": 0.0099, "step": 2460 }, { "epoch": 1.3753829016986912, "grad_norm": 0.13542887753220842, "learning_rate": 9.719973191474875e-05, "loss": 0.0097, "step": 2470 }, { "epoch": 1.380952380952381, "grad_norm": 0.16834906807878625, "learning_rate": 9.715373940727112e-05, "loss": 0.0097, "step": 2480 }, { "epoch": 1.3865218602060707, "grad_norm": 0.14493297637350916, "learning_rate": 9.71073833337981e-05, "loss": 0.0099, "step": 2490 }, { "epoch": 1.3920913394597605, "grad_norm": 0.13388729788015086, "learning_rate": 9.706066405174543e-05, "loss": 0.0096, "step": 2500 }, { "epoch": 1.3976608187134503, "grad_norm": 0.12948166422313045, "learning_rate": 9.701358192132939e-05, "loss": 0.0095, "step": 2510 }, { "epoch": 1.40323029796714, "grad_norm": 0.14831134004472454, "learning_rate": 9.696613730556383e-05, "loss": 0.009, "step": 2520 }, { "epoch": 1.4087997772208298, "grad_norm": 0.13855341547369657, "learning_rate": 9.691833057025747e-05, "loss": 0.0092, "step": 2530 }, { "epoch": 1.4143692564745196, "grad_norm": 0.1333888893995881, "learning_rate": 9.687016208401107e-05, "loss": 0.0093, "step": 2540 }, { "epoch": 1.4199387357282094, "grad_norm": 0.13806477992850671, "learning_rate": 9.682163221821455e-05, "loss": 0.0091, "step": 2550 }, { "epoch": 1.4255082149818992, "grad_norm": 0.1424299673717202, "learning_rate": 9.677274134704411e-05, "loss": 0.009, "step": 2560 }, { "epoch": 1.431077694235589, "grad_norm": 0.11566085539435643, "learning_rate": 9.672348984745949e-05, "loss": 0.0088, "step": 2570 }, { "epoch": 1.4366471734892787, "grad_norm": 0.12261270767102363, "learning_rate": 9.667387809920083e-05, "loss": 0.0087, "step": 2580 }, { "epoch": 1.4422166527429685, "grad_norm": 0.16396158500026034, "learning_rate": 9.662390648478597e-05, "loss": 0.0086, "step": 2590 }, { "epoch": 1.4477861319966583, "grad_norm": 0.1532929944740895, "learning_rate": 9.657357538950733e-05, "loss": 0.0088, "step": 2600 }, { "epoch": 1.453355611250348, "grad_norm": 0.14377084311922123, "learning_rate": 9.652288520142907e-05, "loss": 0.0087, "step": 2610 }, { "epoch": 1.4589250905040378, "grad_norm": 0.13047962164454835, "learning_rate": 9.647183631138397e-05, "loss": 0.0084, "step": 2620 }, { "epoch": 1.4644945697577276, "grad_norm": 0.13602569186021218, "learning_rate": 9.642042911297055e-05, "loss": 0.0078, "step": 2630 }, { "epoch": 1.4700640490114174, "grad_norm": 0.11739120328716636, "learning_rate": 9.636866400254991e-05, "loss": 0.008, "step": 2640 }, { "epoch": 1.4756335282651072, "grad_norm": 0.13171022429436544, "learning_rate": 9.631654137924278e-05, "loss": 0.008, "step": 2650 }, { "epoch": 1.481203007518797, "grad_norm": 0.12987231380199957, "learning_rate": 9.626406164492636e-05, "loss": 0.0081, "step": 2660 }, { "epoch": 1.4867724867724867, "grad_norm": 0.11746026303181731, "learning_rate": 9.621122520423128e-05, "loss": 0.0079, "step": 2670 }, { "epoch": 1.4923419660261765, "grad_norm": 0.11452815904361804, "learning_rate": 9.615803246453843e-05, "loss": 0.0078, "step": 2680 }, { "epoch": 1.4979114452798663, "grad_norm": 0.111682722051899, "learning_rate": 9.61044838359759e-05, "loss": 0.0079, "step": 2690 }, { "epoch": 1.503480924533556, "grad_norm": 0.11111224496044826, "learning_rate": 9.60505797314157e-05, "loss": 0.0081, "step": 2700 }, { "epoch": 1.5090504037872459, "grad_norm": 0.12434768563055779, "learning_rate": 9.599632056647065e-05, "loss": 0.0075, "step": 2710 }, { "epoch": 1.5146198830409356, "grad_norm": 0.12762658726155382, "learning_rate": 9.594170675949122e-05, "loss": 0.0077, "step": 2720 }, { "epoch": 1.5201893622946254, "grad_norm": 0.13462630987728993, "learning_rate": 9.588673873156218e-05, "loss": 0.0074, "step": 2730 }, { "epoch": 1.5257588415483152, "grad_norm": 0.11873990531246101, "learning_rate": 9.583141690649949e-05, "loss": 0.0075, "step": 2740 }, { "epoch": 1.531328320802005, "grad_norm": 0.11897341907272085, "learning_rate": 9.577574171084691e-05, "loss": 0.0072, "step": 2750 }, { "epoch": 1.5368978000556948, "grad_norm": 0.1342155424043113, "learning_rate": 9.571971357387281e-05, "loss": 0.0073, "step": 2760 }, { "epoch": 1.5424672793093845, "grad_norm": 0.11109000135852713, "learning_rate": 9.566333292756678e-05, "loss": 0.0073, "step": 2770 }, { "epoch": 1.5480367585630743, "grad_norm": 0.10749381139994726, "learning_rate": 9.560660020663636e-05, "loss": 0.0071, "step": 2780 }, { "epoch": 1.553606237816764, "grad_norm": 0.12678846265983787, "learning_rate": 9.554951584850366e-05, "loss": 0.0072, "step": 2790 }, { "epoch": 1.5591757170704539, "grad_norm": 0.11887715567893196, "learning_rate": 9.549208029330198e-05, "loss": 0.0071, "step": 2800 }, { "epoch": 1.5647451963241437, "grad_norm": 0.11528952921949762, "learning_rate": 9.543429398387246e-05, "loss": 0.0072, "step": 2810 }, { "epoch": 1.5703146755778334, "grad_norm": 0.1130261076594759, "learning_rate": 9.537615736576062e-05, "loss": 0.0068, "step": 2820 }, { "epoch": 1.5758841548315232, "grad_norm": 0.1150612035172818, "learning_rate": 9.531767088721293e-05, "loss": 0.0068, "step": 2830 }, { "epoch": 1.581453634085213, "grad_norm": 0.12119751877826486, "learning_rate": 9.525883499917334e-05, "loss": 0.0066, "step": 2840 }, { "epoch": 1.5870231133389028, "grad_norm": 0.14369604650594714, "learning_rate": 9.519965015527987e-05, "loss": 0.0067, "step": 2850 }, { "epoch": 1.5925925925925926, "grad_norm": 0.11828049051691453, "learning_rate": 9.514011681186107e-05, "loss": 0.0067, "step": 2860 }, { "epoch": 1.5981620718462823, "grad_norm": 0.1041622423129153, "learning_rate": 9.508023542793246e-05, "loss": 0.0067, "step": 2870 }, { "epoch": 1.6037315510999721, "grad_norm": 0.109441034459224, "learning_rate": 9.502000646519304e-05, "loss": 0.0064, "step": 2880 }, { "epoch": 1.609301030353662, "grad_norm": 0.11407457138432828, "learning_rate": 9.495943038802178e-05, "loss": 0.0064, "step": 2890 }, { "epoch": 1.6148705096073517, "grad_norm": 0.11169346620933446, "learning_rate": 9.48985076634739e-05, "loss": 0.0063, "step": 2900 }, { "epoch": 1.6204399888610415, "grad_norm": 0.1068634570105383, "learning_rate": 9.48372387612774e-05, "loss": 0.0063, "step": 2910 }, { "epoch": 1.6260094681147312, "grad_norm": 0.10960556348842432, "learning_rate": 9.477562415382937e-05, "loss": 0.0064, "step": 2920 }, { "epoch": 1.631578947368421, "grad_norm": 0.09964438396517761, "learning_rate": 9.471366431619236e-05, "loss": 0.0063, "step": 2930 }, { "epoch": 1.6371484266221108, "grad_norm": 0.10857713507090318, "learning_rate": 9.465135972609074e-05, "loss": 0.0061, "step": 2940 }, { "epoch": 1.6427179058758006, "grad_norm": 0.12812497053072483, "learning_rate": 9.458871086390699e-05, "loss": 0.0063, "step": 2950 }, { "epoch": 1.6482873851294904, "grad_norm": 0.12920573246968883, "learning_rate": 9.452571821267799e-05, "loss": 0.0062, "step": 2960 }, { "epoch": 1.6538568643831801, "grad_norm": 0.11378467589478443, "learning_rate": 9.446238225809137e-05, "loss": 0.0061, "step": 2970 }, { "epoch": 1.65942634363687, "grad_norm": 0.1025631830473413, "learning_rate": 9.439870348848162e-05, "loss": 0.0062, "step": 2980 }, { "epoch": 1.6649958228905597, "grad_norm": 0.09752210801488877, "learning_rate": 9.433468239482647e-05, "loss": 0.0063, "step": 2990 }, { "epoch": 1.6705653021442495, "grad_norm": 0.10280694509228655, "learning_rate": 9.427031947074305e-05, "loss": 0.006, "step": 3000 }, { "epoch": 1.6761347813979393, "grad_norm": 0.10219752893301338, "learning_rate": 9.420561521248403e-05, "loss": 0.0059, "step": 3010 }, { "epoch": 1.681704260651629, "grad_norm": 0.10744460806316994, "learning_rate": 9.414057011893389e-05, "loss": 0.0058, "step": 3020 }, { "epoch": 1.6872737399053188, "grad_norm": 0.10272875016713308, "learning_rate": 9.407518469160504e-05, "loss": 0.0057, "step": 3030 }, { "epoch": 1.6928432191590086, "grad_norm": 0.10784343296872123, "learning_rate": 9.400945943463387e-05, "loss": 0.0056, "step": 3040 }, { "epoch": 1.6984126984126984, "grad_norm": 0.1265310671727756, "learning_rate": 9.394339485477697e-05, "loss": 0.0059, "step": 3050 }, { "epoch": 1.7039821776663882, "grad_norm": 0.12380016024169994, "learning_rate": 9.387699146140724e-05, "loss": 0.0058, "step": 3060 }, { "epoch": 1.709551656920078, "grad_norm": 0.1116972787626009, "learning_rate": 9.381024976650983e-05, "loss": 0.0055, "step": 3070 }, { "epoch": 1.7151211361737677, "grad_norm": 0.10613073795533987, "learning_rate": 9.37431702846783e-05, "loss": 0.0055, "step": 3080 }, { "epoch": 1.7206906154274575, "grad_norm": 0.11702121991080053, "learning_rate": 9.367575353311064e-05, "loss": 0.0058, "step": 3090 }, { "epoch": 1.7262600946811473, "grad_norm": 0.10126745527171219, "learning_rate": 9.360800003160525e-05, "loss": 0.0057, "step": 3100 }, { "epoch": 1.731829573934837, "grad_norm": 0.10028282479858917, "learning_rate": 9.353991030255695e-05, "loss": 0.0056, "step": 3110 }, { "epoch": 1.7373990531885268, "grad_norm": 0.1102978478866508, "learning_rate": 9.347148487095294e-05, "loss": 0.0055, "step": 3120 }, { "epoch": 1.7429685324422166, "grad_norm": 0.10831564723664697, "learning_rate": 9.340272426436877e-05, "loss": 0.0056, "step": 3130 }, { "epoch": 1.7485380116959064, "grad_norm": 0.1012748795569494, "learning_rate": 9.33336290129643e-05, "loss": 0.0053, "step": 3140 }, { "epoch": 1.7541074909495962, "grad_norm": 0.11575041254493004, "learning_rate": 9.32641996494795e-05, "loss": 0.0052, "step": 3150 }, { "epoch": 1.759676970203286, "grad_norm": 0.10943195809368053, "learning_rate": 9.319443670923051e-05, "loss": 0.0056, "step": 3160 }, { "epoch": 1.7652464494569757, "grad_norm": 0.0939129237227479, "learning_rate": 9.312434073010534e-05, "loss": 0.0052, "step": 3170 }, { "epoch": 1.7708159287106655, "grad_norm": 0.1047650747357604, "learning_rate": 9.305391225255986e-05, "loss": 0.0053, "step": 3180 }, { "epoch": 1.7763854079643553, "grad_norm": 0.09958975875505627, "learning_rate": 9.298315181961356e-05, "loss": 0.0052, "step": 3190 }, { "epoch": 1.781954887218045, "grad_norm": 0.10425138001460964, "learning_rate": 9.291205997684538e-05, "loss": 0.0053, "step": 3200 }, { "epoch": 1.7875243664717348, "grad_norm": 0.09069036788388674, "learning_rate": 9.284063727238952e-05, "loss": 0.0051, "step": 3210 }, { "epoch": 1.7930938457254246, "grad_norm": 0.09833804691470377, "learning_rate": 9.276888425693115e-05, "loss": 0.0049, "step": 3220 }, { "epoch": 1.7986633249791144, "grad_norm": 0.08709400377373291, "learning_rate": 9.269680148370228e-05, "loss": 0.0048, "step": 3230 }, { "epoch": 1.8042328042328042, "grad_norm": 0.11915976087290946, "learning_rate": 9.262438950847738e-05, "loss": 0.0048, "step": 3240 }, { "epoch": 1.809802283486494, "grad_norm": 0.08988574166722815, "learning_rate": 9.255164888956918e-05, "loss": 0.005, "step": 3250 }, { "epoch": 1.8153717627401837, "grad_norm": 0.09796027881734362, "learning_rate": 9.247858018782425e-05, "loss": 0.0052, "step": 3260 }, { "epoch": 1.8209412419938735, "grad_norm": 0.09351986222567911, "learning_rate": 9.240518396661886e-05, "loss": 0.0048, "step": 3270 }, { "epoch": 1.8265107212475633, "grad_norm": 0.09530159956409921, "learning_rate": 9.233146079185444e-05, "loss": 0.0049, "step": 3280 }, { "epoch": 1.832080200501253, "grad_norm": 0.09731186711139259, "learning_rate": 9.225741123195338e-05, "loss": 0.005, "step": 3290 }, { "epoch": 1.8376496797549429, "grad_norm": 0.08614001336056387, "learning_rate": 9.218303585785448e-05, "loss": 0.0047, "step": 3300 }, { "epoch": 1.8432191590086326, "grad_norm": 0.09126829864578764, "learning_rate": 9.210833524300873e-05, "loss": 0.0048, "step": 3310 }, { "epoch": 1.8487886382623224, "grad_norm": 0.08448191010827921, "learning_rate": 9.203330996337477e-05, "loss": 0.0048, "step": 3320 }, { "epoch": 1.8543581175160122, "grad_norm": 0.08979291201416467, "learning_rate": 9.195796059741447e-05, "loss": 0.0046, "step": 3330 }, { "epoch": 1.859927596769702, "grad_norm": 0.09340579831138453, "learning_rate": 9.188228772608846e-05, "loss": 0.0046, "step": 3340 }, { "epoch": 1.8654970760233918, "grad_norm": 0.0924248400058176, "learning_rate": 9.180629193285172e-05, "loss": 0.0047, "step": 3350 }, { "epoch": 1.8710665552770815, "grad_norm": 0.09465367072304114, "learning_rate": 9.1729973803649e-05, "loss": 0.0048, "step": 3360 }, { "epoch": 1.8766360345307713, "grad_norm": 0.09532898949385789, "learning_rate": 9.165333392691032e-05, "loss": 0.0049, "step": 3370 }, { "epoch": 1.882205513784461, "grad_norm": 0.22241436561372555, "learning_rate": 9.157637289354648e-05, "loss": 0.0046, "step": 3380 }, { "epoch": 1.8877749930381509, "grad_norm": 0.09048559280760698, "learning_rate": 9.149909129694442e-05, "loss": 0.0051, "step": 3390 }, { "epoch": 1.8933444722918407, "grad_norm": 0.09851219221221012, "learning_rate": 9.142148973296279e-05, "loss": 0.0047, "step": 3400 }, { "epoch": 1.8989139515455304, "grad_norm": 0.11728224731469955, "learning_rate": 9.134356879992716e-05, "loss": 0.0046, "step": 3410 }, { "epoch": 1.9044834307992202, "grad_norm": 0.09083216143262172, "learning_rate": 9.126532909862554e-05, "loss": 0.0045, "step": 3420 }, { "epoch": 1.91005291005291, "grad_norm": 0.09298839697175458, "learning_rate": 9.118677123230372e-05, "loss": 0.0044, "step": 3430 }, { "epoch": 1.9156223893065998, "grad_norm": 0.09340867022639449, "learning_rate": 9.110789580666066e-05, "loss": 0.0045, "step": 3440 }, { "epoch": 1.9211918685602896, "grad_norm": 0.11685852733746609, "learning_rate": 9.102870342984368e-05, "loss": 0.0045, "step": 3450 }, { "epoch": 1.9267613478139793, "grad_norm": 0.08462199664128196, "learning_rate": 9.094919471244394e-05, "loss": 0.0043, "step": 3460 }, { "epoch": 1.9323308270676691, "grad_norm": 0.09235283313911731, "learning_rate": 9.086937026749163e-05, "loss": 0.0044, "step": 3470 }, { "epoch": 1.937900306321359, "grad_norm": 0.09557316586337207, "learning_rate": 9.078923071045129e-05, "loss": 0.0044, "step": 3480 }, { "epoch": 1.9434697855750487, "grad_norm": 0.09889989513485295, "learning_rate": 9.0708776659217e-05, "loss": 0.0045, "step": 3490 }, { "epoch": 1.9490392648287385, "grad_norm": 0.1042326585110314, "learning_rate": 9.062800873410769e-05, "loss": 0.0044, "step": 3500 }, { "epoch": 1.9546087440824282, "grad_norm": 0.09129522893069832, "learning_rate": 9.054692755786236e-05, "loss": 0.0042, "step": 3510 }, { "epoch": 1.960178223336118, "grad_norm": 0.10382677134797022, "learning_rate": 9.04655337556352e-05, "loss": 0.0042, "step": 3520 }, { "epoch": 1.9657477025898078, "grad_norm": 0.08987092514326217, "learning_rate": 9.038382795499083e-05, "loss": 0.0043, "step": 3530 }, { "epoch": 1.9713171818434976, "grad_norm": 0.0849449608197563, "learning_rate": 9.030181078589947e-05, "loss": 0.0042, "step": 3540 }, { "epoch": 1.9768866610971874, "grad_norm": 0.09495769793832361, "learning_rate": 9.021948288073201e-05, "loss": 0.0042, "step": 3550 }, { "epoch": 1.9824561403508771, "grad_norm": 0.11307494496950236, "learning_rate": 9.013684487425525e-05, "loss": 0.0043, "step": 3560 }, { "epoch": 1.988025619604567, "grad_norm": 0.10184785608835481, "learning_rate": 9.005389740362691e-05, "loss": 0.004, "step": 3570 }, { "epoch": 1.9935950988582567, "grad_norm": 0.6780573883609672, "learning_rate": 8.997064110839072e-05, "loss": 0.0052, "step": 3580 }, { "epoch": 1.9991645781119465, "grad_norm": 1.913891302534067, "learning_rate": 8.988707663047156e-05, "loss": 0.0382, "step": 3590 }, { "epoch": 2.004455583402952, "grad_norm": 0.20491348477515167, "learning_rate": 8.980320461417046e-05, "loss": 0.0127, "step": 3600 }, { "epoch": 2.0100250626566414, "grad_norm": 0.11495094855882845, "learning_rate": 8.971902570615964e-05, "loss": 0.0063, "step": 3610 }, { "epoch": 2.0155945419103314, "grad_norm": 0.11775260158307438, "learning_rate": 8.963454055547751e-05, "loss": 0.0047, "step": 3620 }, { "epoch": 2.0211640211640214, "grad_norm": 0.08730294248283851, "learning_rate": 8.954974981352368e-05, "loss": 0.0041, "step": 3630 }, { "epoch": 2.026733500417711, "grad_norm": 0.08688062935334084, "learning_rate": 8.946465413405398e-05, "loss": 0.0037, "step": 3640 }, { "epoch": 2.0323029796714005, "grad_norm": 0.07712628832792173, "learning_rate": 8.93792541731753e-05, "loss": 0.0034, "step": 3650 }, { "epoch": 2.0378724589250905, "grad_norm": 0.06765602225550173, "learning_rate": 8.929355058934068e-05, "loss": 0.0033, "step": 3660 }, { "epoch": 2.0434419381787805, "grad_norm": 0.07792927728833504, "learning_rate": 8.920754404334415e-05, "loss": 0.0031, "step": 3670 }, { "epoch": 2.04901141743247, "grad_norm": 0.06508427058988357, "learning_rate": 8.912123519831559e-05, "loss": 0.003, "step": 3680 }, { "epoch": 2.0545808966861596, "grad_norm": 0.06602478817547071, "learning_rate": 8.903462471971573e-05, "loss": 0.0029, "step": 3690 }, { "epoch": 2.0601503759398496, "grad_norm": 0.07073489621873132, "learning_rate": 8.894771327533098e-05, "loss": 0.003, "step": 3700 }, { "epoch": 2.0657198551935396, "grad_norm": 0.07513785677590291, "learning_rate": 8.886050153526819e-05, "loss": 0.003, "step": 3710 }, { "epoch": 2.071289334447229, "grad_norm": 0.075200356674605, "learning_rate": 8.877299017194962e-05, "loss": 0.0029, "step": 3720 }, { "epoch": 2.0768588137009187, "grad_norm": 0.07145305892339188, "learning_rate": 8.868517986010768e-05, "loss": 0.003, "step": 3730 }, { "epoch": 2.0824282929546087, "grad_norm": 0.079246440019147, "learning_rate": 8.859707127677972e-05, "loss": 0.0027, "step": 3740 }, { "epoch": 2.0879977722082987, "grad_norm": 0.07053381470878965, "learning_rate": 8.850866510130285e-05, "loss": 0.0027, "step": 3750 }, { "epoch": 2.0935672514619883, "grad_norm": 0.07151886382294959, "learning_rate": 8.841996201530869e-05, "loss": 0.0027, "step": 3760 }, { "epoch": 2.099136730715678, "grad_norm": 0.07398921035039588, "learning_rate": 8.833096270271806e-05, "loss": 0.0028, "step": 3770 }, { "epoch": 2.104706209969368, "grad_norm": 0.06628976492907845, "learning_rate": 8.824166784973582e-05, "loss": 0.0027, "step": 3780 }, { "epoch": 2.110275689223058, "grad_norm": 0.06967783368415605, "learning_rate": 8.815207814484547e-05, "loss": 0.0026, "step": 3790 }, { "epoch": 2.1158451684767474, "grad_norm": 0.07367756743304106, "learning_rate": 8.806219427880391e-05, "loss": 0.0026, "step": 3800 }, { "epoch": 2.121414647730437, "grad_norm": 0.07016041871073152, "learning_rate": 8.797201694463606e-05, "loss": 0.0026, "step": 3810 }, { "epoch": 2.126984126984127, "grad_norm": 0.06957343148915318, "learning_rate": 8.788154683762958e-05, "loss": 0.0024, "step": 3820 }, { "epoch": 2.132553606237817, "grad_norm": 0.0698865506500755, "learning_rate": 8.779078465532942e-05, "loss": 0.0025, "step": 3830 }, { "epoch": 2.1381230854915065, "grad_norm": 0.0638845189601394, "learning_rate": 8.769973109753257e-05, "loss": 0.0025, "step": 3840 }, { "epoch": 2.143692564745196, "grad_norm": 0.06605357071535665, "learning_rate": 8.760838686628254e-05, "loss": 0.0026, "step": 3850 }, { "epoch": 2.149262043998886, "grad_norm": 0.06646085746495409, "learning_rate": 8.751675266586401e-05, "loss": 0.0027, "step": 3860 }, { "epoch": 2.154831523252576, "grad_norm": 0.08537101005349107, "learning_rate": 8.74248292027974e-05, "loss": 0.0027, "step": 3870 }, { "epoch": 2.1604010025062657, "grad_norm": 0.0674112060918515, "learning_rate": 8.733261718583339e-05, "loss": 0.0025, "step": 3880 }, { "epoch": 2.165970481759955, "grad_norm": 0.06660804464627262, "learning_rate": 8.724011732594749e-05, "loss": 0.0026, "step": 3890 }, { "epoch": 2.171539961013645, "grad_norm": 0.06849541239000093, "learning_rate": 8.714733033633454e-05, "loss": 0.0026, "step": 3900 }, { "epoch": 2.177109440267335, "grad_norm": 0.0720897091900461, "learning_rate": 8.705425693240319e-05, "loss": 0.0026, "step": 3910 }, { "epoch": 2.1826789195210248, "grad_norm": 0.06824332799424324, "learning_rate": 8.696089783177049e-05, "loss": 0.0025, "step": 3920 }, { "epoch": 2.1882483987747143, "grad_norm": 0.06717944971109528, "learning_rate": 8.686725375425617e-05, "loss": 0.0025, "step": 3930 }, { "epoch": 2.1938178780284043, "grad_norm": 0.06136685454453069, "learning_rate": 8.677332542187727e-05, "loss": 0.0025, "step": 3940 }, { "epoch": 2.1993873572820943, "grad_norm": 0.07922796621477682, "learning_rate": 8.667911355884248e-05, "loss": 0.0025, "step": 3950 }, { "epoch": 2.204956836535784, "grad_norm": 0.06748085462192983, "learning_rate": 8.658461889154654e-05, "loss": 0.0026, "step": 3960 }, { "epoch": 2.2105263157894735, "grad_norm": 0.07033503509897161, "learning_rate": 8.648984214856474e-05, "loss": 0.0026, "step": 3970 }, { "epoch": 2.2160957950431635, "grad_norm": 0.0693408764546056, "learning_rate": 8.639478406064722e-05, "loss": 0.0025, "step": 3980 }, { "epoch": 2.2216652742968535, "grad_norm": 0.06290458960875848, "learning_rate": 8.629944536071328e-05, "loss": 0.0026, "step": 3990 }, { "epoch": 2.227234753550543, "grad_norm": 0.06522173944000591, "learning_rate": 8.620382678384591e-05, "loss": 0.0025, "step": 4000 }, { "epoch": 2.2328042328042326, "grad_norm": 0.0692135716494379, "learning_rate": 8.610792906728593e-05, "loss": 0.0025, "step": 4010 }, { "epoch": 2.2383737120579226, "grad_norm": 0.06218397041619794, "learning_rate": 8.601175295042643e-05, "loss": 0.0025, "step": 4020 }, { "epoch": 2.2439431913116126, "grad_norm": 0.07425856965054406, "learning_rate": 8.591529917480705e-05, "loss": 0.0026, "step": 4030 }, { "epoch": 2.249512670565302, "grad_norm": 0.07106967477843956, "learning_rate": 8.581856848410816e-05, "loss": 0.0024, "step": 4040 }, { "epoch": 2.2550821498189917, "grad_norm": 0.07511095612847542, "learning_rate": 8.572156162414533e-05, "loss": 0.0025, "step": 4050 }, { "epoch": 2.2606516290726817, "grad_norm": 0.06682058463634225, "learning_rate": 8.562427934286334e-05, "loss": 0.0023, "step": 4060 }, { "epoch": 2.2662211083263717, "grad_norm": 0.07894516840214212, "learning_rate": 8.55267223903306e-05, "loss": 0.0025, "step": 4070 }, { "epoch": 2.2717905875800612, "grad_norm": 0.07442523909487508, "learning_rate": 8.542889151873328e-05, "loss": 0.0023, "step": 4080 }, { "epoch": 2.277360066833751, "grad_norm": 0.07047966962425675, "learning_rate": 8.533078748236951e-05, "loss": 0.0024, "step": 4090 }, { "epoch": 2.282929546087441, "grad_norm": 0.0670309181354895, "learning_rate": 8.52324110376436e-05, "loss": 0.0024, "step": 4100 }, { "epoch": 2.288499025341131, "grad_norm": 0.05841158200173529, "learning_rate": 8.513376294306022e-05, "loss": 0.0025, "step": 4110 }, { "epoch": 2.2940685045948204, "grad_norm": 0.07235767865059842, "learning_rate": 8.503484395921847e-05, "loss": 0.0025, "step": 4120 }, { "epoch": 2.29963798384851, "grad_norm": 0.06368681101578703, "learning_rate": 8.493565484880607e-05, "loss": 0.0025, "step": 4130 }, { "epoch": 2.3052074631022, "grad_norm": 0.06306798916456337, "learning_rate": 8.483619637659352e-05, "loss": 0.0023, "step": 4140 }, { "epoch": 2.31077694235589, "grad_norm": 0.063978959789188, "learning_rate": 8.473646930942811e-05, "loss": 0.0023, "step": 4150 }, { "epoch": 2.3163464216095795, "grad_norm": 0.08113440594399092, "learning_rate": 8.463647441622811e-05, "loss": 0.0024, "step": 4160 }, { "epoch": 2.321915900863269, "grad_norm": 0.06560709654743707, "learning_rate": 8.453621246797674e-05, "loss": 0.0023, "step": 4170 }, { "epoch": 2.327485380116959, "grad_norm": 0.05878754983194171, "learning_rate": 8.443568423771632e-05, "loss": 0.0025, "step": 4180 }, { "epoch": 2.333054859370649, "grad_norm": 0.06519491241511483, "learning_rate": 8.433489050054222e-05, "loss": 0.0024, "step": 4190 }, { "epoch": 2.3386243386243386, "grad_norm": 0.06265875286229199, "learning_rate": 8.423383203359696e-05, "loss": 0.0023, "step": 4200 }, { "epoch": 2.344193817878028, "grad_norm": 0.07488402161472248, "learning_rate": 8.413250961606417e-05, "loss": 0.0023, "step": 4210 }, { "epoch": 2.349763297131718, "grad_norm": 0.06477354979976067, "learning_rate": 8.403092402916261e-05, "loss": 0.0023, "step": 4220 }, { "epoch": 2.355332776385408, "grad_norm": 0.06498384745452408, "learning_rate": 8.392907605614013e-05, "loss": 0.0024, "step": 4230 }, { "epoch": 2.3609022556390977, "grad_norm": 0.06078330248348828, "learning_rate": 8.382696648226764e-05, "loss": 0.0024, "step": 4240 }, { "epoch": 2.3664717348927873, "grad_norm": 0.0687320440446581, "learning_rate": 8.372459609483303e-05, "loss": 0.0025, "step": 4250 }, { "epoch": 2.3720412141464773, "grad_norm": 0.06587983128890958, "learning_rate": 8.362196568313516e-05, "loss": 0.0023, "step": 4260 }, { "epoch": 2.3776106934001673, "grad_norm": 0.06659212111823104, "learning_rate": 8.351907603847773e-05, "loss": 0.0023, "step": 4270 }, { "epoch": 2.383180172653857, "grad_norm": 0.06922434253253595, "learning_rate": 8.341592795416312e-05, "loss": 0.0024, "step": 4280 }, { "epoch": 2.3887496519075464, "grad_norm": 0.06994327293771829, "learning_rate": 8.331252222548645e-05, "loss": 0.0024, "step": 4290 }, { "epoch": 2.3943191311612364, "grad_norm": 0.06643409266880786, "learning_rate": 8.320885964972922e-05, "loss": 0.0025, "step": 4300 }, { "epoch": 2.3998886104149264, "grad_norm": 0.0811114126700153, "learning_rate": 8.310494102615336e-05, "loss": 0.0025, "step": 4310 }, { "epoch": 2.405458089668616, "grad_norm": 0.07211869955643185, "learning_rate": 8.300076715599496e-05, "loss": 0.0024, "step": 4320 }, { "epoch": 2.4110275689223055, "grad_norm": 0.06687148486771088, "learning_rate": 8.289633884245811e-05, "loss": 0.0024, "step": 4330 }, { "epoch": 2.4165970481759955, "grad_norm": 0.06431135751569382, "learning_rate": 8.279165689070875e-05, "loss": 0.0023, "step": 4340 }, { "epoch": 2.4221665274296855, "grad_norm": 0.07773019218992226, "learning_rate": 8.268672210786839e-05, "loss": 0.0023, "step": 4350 }, { "epoch": 2.427736006683375, "grad_norm": 0.07191534653352975, "learning_rate": 8.258153530300792e-05, "loss": 0.0025, "step": 4360 }, { "epoch": 2.4333054859370646, "grad_norm": 0.07424856314720017, "learning_rate": 8.247609728714142e-05, "loss": 0.0024, "step": 4370 }, { "epoch": 2.4388749651907546, "grad_norm": 0.06644817422012073, "learning_rate": 8.237040887321983e-05, "loss": 0.0024, "step": 4380 }, { "epoch": 2.4444444444444446, "grad_norm": 0.07935382842757754, "learning_rate": 8.226447087612472e-05, "loss": 0.0024, "step": 4390 }, { "epoch": 2.450013923698134, "grad_norm": 0.07344510881235311, "learning_rate": 8.215828411266202e-05, "loss": 0.0023, "step": 4400 }, { "epoch": 2.4555834029518238, "grad_norm": 0.06437241925673821, "learning_rate": 8.205184940155567e-05, "loss": 0.0023, "step": 4410 }, { "epoch": 2.4611528822055138, "grad_norm": 0.06191846693537428, "learning_rate": 8.19451675634414e-05, "loss": 0.0022, "step": 4420 }, { "epoch": 2.4667223614592038, "grad_norm": 0.0743929168561731, "learning_rate": 8.183823942086026e-05, "loss": 0.0022, "step": 4430 }, { "epoch": 2.4722918407128933, "grad_norm": 0.07877045476260705, "learning_rate": 8.173106579825246e-05, "loss": 0.0024, "step": 4440 }, { "epoch": 2.477861319966583, "grad_norm": 0.06063611479460526, "learning_rate": 8.162364752195084e-05, "loss": 0.0023, "step": 4450 }, { "epoch": 2.483430799220273, "grad_norm": 0.06803918605199016, "learning_rate": 8.151598542017463e-05, "loss": 0.0021, "step": 4460 }, { "epoch": 2.489000278473963, "grad_norm": 0.06684605150130163, "learning_rate": 8.140808032302296e-05, "loss": 0.0022, "step": 4470 }, { "epoch": 2.4945697577276524, "grad_norm": 0.07187232516181655, "learning_rate": 8.129993306246855e-05, "loss": 0.0022, "step": 4480 }, { "epoch": 2.500139236981342, "grad_norm": 0.06494266366065195, "learning_rate": 8.119154447235123e-05, "loss": 0.0022, "step": 4490 }, { "epoch": 2.505708716235032, "grad_norm": 0.06464060504434475, "learning_rate": 8.108291538837151e-05, "loss": 0.0022, "step": 4500 }, { "epoch": 2.511278195488722, "grad_norm": 0.06569241269740814, "learning_rate": 8.097404664808428e-05, "loss": 0.0022, "step": 4510 }, { "epoch": 2.5168476747424116, "grad_norm": 0.055242094122246455, "learning_rate": 8.086493909089208e-05, "loss": 0.0021, "step": 4520 }, { "epoch": 2.522417153996101, "grad_norm": 0.06777582514746222, "learning_rate": 8.075559355803887e-05, "loss": 0.0023, "step": 4530 }, { "epoch": 2.527986633249791, "grad_norm": 0.06478881950733577, "learning_rate": 8.064601089260344e-05, "loss": 0.0022, "step": 4540 }, { "epoch": 2.533556112503481, "grad_norm": 0.06924218417878182, "learning_rate": 8.053619193949293e-05, "loss": 0.0022, "step": 4550 }, { "epoch": 2.5391255917571707, "grad_norm": 0.06574296301329537, "learning_rate": 8.042613754543631e-05, "loss": 0.0022, "step": 4560 }, { "epoch": 2.5446950710108602, "grad_norm": 0.0652453535875289, "learning_rate": 8.031584855897782e-05, "loss": 0.0022, "step": 4570 }, { "epoch": 2.5502645502645502, "grad_norm": 0.0669710260970334, "learning_rate": 8.020532583047052e-05, "loss": 0.002, "step": 4580 }, { "epoch": 2.5558340295182402, "grad_norm": 0.06195618827884842, "learning_rate": 8.009457021206964e-05, "loss": 0.0022, "step": 4590 }, { "epoch": 2.56140350877193, "grad_norm": 0.05421163706378184, "learning_rate": 7.998358255772604e-05, "loss": 0.0022, "step": 4600 }, { "epoch": 2.5669729880256194, "grad_norm": 0.07244346720460094, "learning_rate": 7.987236372317964e-05, "loss": 0.0021, "step": 4610 }, { "epoch": 2.5725424672793094, "grad_norm": 0.07122830520202149, "learning_rate": 7.976091456595278e-05, "loss": 0.0022, "step": 4620 }, { "epoch": 2.5781119465329994, "grad_norm": 0.058368117308028544, "learning_rate": 7.964923594534372e-05, "loss": 0.002, "step": 4630 }, { "epoch": 2.583681425786689, "grad_norm": 0.06617475020782944, "learning_rate": 7.953732872241984e-05, "loss": 0.0022, "step": 4640 }, { "epoch": 2.5892509050403785, "grad_norm": 0.05757509551408226, "learning_rate": 7.942519376001116e-05, "loss": 0.0021, "step": 4650 }, { "epoch": 2.5948203842940685, "grad_norm": 0.06085258271174403, "learning_rate": 7.93128319227036e-05, "loss": 0.0021, "step": 4660 }, { "epoch": 2.6003898635477585, "grad_norm": 0.0653949688851885, "learning_rate": 7.920024407683234e-05, "loss": 0.0021, "step": 4670 }, { "epoch": 2.605959342801448, "grad_norm": 0.06266065978713677, "learning_rate": 7.908743109047512e-05, "loss": 0.0021, "step": 4680 }, { "epoch": 2.6115288220551376, "grad_norm": 0.06607862312823722, "learning_rate": 7.89743938334456e-05, "loss": 0.0021, "step": 4690 }, { "epoch": 2.6170983013088276, "grad_norm": 0.05456069896485288, "learning_rate": 7.886113317728658e-05, "loss": 0.002, "step": 4700 }, { "epoch": 2.6226677805625176, "grad_norm": 0.07060726920422762, "learning_rate": 7.874764999526334e-05, "loss": 0.0021, "step": 4710 }, { "epoch": 2.628237259816207, "grad_norm": 0.0628459007652655, "learning_rate": 7.863394516235687e-05, "loss": 0.0021, "step": 4720 }, { "epoch": 2.6338067390698967, "grad_norm": 0.07184850567413902, "learning_rate": 7.852001955525717e-05, "loss": 0.0021, "step": 4730 }, { "epoch": 2.6393762183235867, "grad_norm": 0.05931313997400148, "learning_rate": 7.840587405235642e-05, "loss": 0.0022, "step": 4740 }, { "epoch": 2.6449456975772767, "grad_norm": 0.06266716450193778, "learning_rate": 7.829150953374228e-05, "loss": 0.0021, "step": 4750 }, { "epoch": 2.6505151768309663, "grad_norm": 0.06018038870616263, "learning_rate": 7.817692688119103e-05, "loss": 0.0022, "step": 4760 }, { "epoch": 2.656084656084656, "grad_norm": 0.06516106860781504, "learning_rate": 7.806212697816085e-05, "loss": 0.0022, "step": 4770 }, { "epoch": 2.661654135338346, "grad_norm": 0.07560214888027027, "learning_rate": 7.794711070978496e-05, "loss": 0.0024, "step": 4780 }, { "epoch": 2.667223614592036, "grad_norm": 0.058287088159306756, "learning_rate": 7.78318789628648e-05, "loss": 0.002, "step": 4790 }, { "epoch": 2.6727930938457254, "grad_norm": 0.05041812827730329, "learning_rate": 7.771643262586319e-05, "loss": 0.0023, "step": 4800 }, { "epoch": 2.678362573099415, "grad_norm": 0.06769617798698746, "learning_rate": 7.76007725888975e-05, "loss": 0.0023, "step": 4810 }, { "epoch": 2.683932052353105, "grad_norm": 0.07237732703780383, "learning_rate": 7.748489974373277e-05, "loss": 0.0024, "step": 4820 }, { "epoch": 2.689501531606795, "grad_norm": 0.06774331956099881, "learning_rate": 7.736881498377488e-05, "loss": 0.0024, "step": 4830 }, { "epoch": 2.6950710108604845, "grad_norm": 0.08272693895736795, "learning_rate": 7.725251920406353e-05, "loss": 0.0023, "step": 4840 }, { "epoch": 2.700640490114174, "grad_norm": 0.0647479021182408, "learning_rate": 7.713601330126551e-05, "loss": 0.0024, "step": 4850 }, { "epoch": 2.706209969367864, "grad_norm": 0.06747122196352953, "learning_rate": 7.70192981736677e-05, "loss": 0.0023, "step": 4860 }, { "epoch": 2.711779448621554, "grad_norm": 0.06660918185561457, "learning_rate": 7.69023747211701e-05, "loss": 0.0021, "step": 4870 }, { "epoch": 2.7173489278752436, "grad_norm": 0.06014673632175607, "learning_rate": 7.678524384527897e-05, "loss": 0.0022, "step": 4880 }, { "epoch": 2.722918407128933, "grad_norm": 0.060474215458977645, "learning_rate": 7.666790644909988e-05, "loss": 0.0021, "step": 4890 }, { "epoch": 2.728487886382623, "grad_norm": 0.05475551997626191, "learning_rate": 7.655036343733068e-05, "loss": 0.0021, "step": 4900 }, { "epoch": 2.734057365636313, "grad_norm": 0.05525281945014114, "learning_rate": 7.643261571625457e-05, "loss": 0.0021, "step": 4910 }, { "epoch": 2.7396268448900027, "grad_norm": 0.06451171594706764, "learning_rate": 7.631466419373312e-05, "loss": 0.0021, "step": 4920 }, { "epoch": 2.7451963241436923, "grad_norm": 0.07105940351069889, "learning_rate": 7.619650977919923e-05, "loss": 0.0022, "step": 4930 }, { "epoch": 2.7507658033973823, "grad_norm": 0.0748324693501243, "learning_rate": 7.607815338365018e-05, "loss": 0.002, "step": 4940 }, { "epoch": 2.7563352826510723, "grad_norm": 0.0588056102878865, "learning_rate": 7.595959591964053e-05, "loss": 0.0021, "step": 4950 }, { "epoch": 2.761904761904762, "grad_norm": 0.05974979070561233, "learning_rate": 7.584083830127515e-05, "loss": 0.0021, "step": 4960 }, { "epoch": 2.7674742411584514, "grad_norm": 0.05841681850185086, "learning_rate": 7.572188144420214e-05, "loss": 0.0021, "step": 4970 }, { "epoch": 2.7730437204121414, "grad_norm": 0.06753208138557094, "learning_rate": 7.560272626560575e-05, "loss": 0.0021, "step": 4980 }, { "epoch": 2.7786131996658314, "grad_norm": 0.06433255988165418, "learning_rate": 7.548337368419937e-05, "loss": 0.002, "step": 4990 }, { "epoch": 2.784182678919521, "grad_norm": 0.06973811409366241, "learning_rate": 7.536382462021843e-05, "loss": 0.0022, "step": 5000 }, { "epoch": 2.7897521581732105, "grad_norm": 0.06230697352063296, "learning_rate": 7.52440799954132e-05, "loss": 0.002, "step": 5010 }, { "epoch": 2.7953216374269005, "grad_norm": 0.05784103057124063, "learning_rate": 7.512414073304183e-05, "loss": 0.0019, "step": 5020 }, { "epoch": 2.8008911166805905, "grad_norm": 0.055170016243759425, "learning_rate": 7.500400775786319e-05, "loss": 0.0021, "step": 5030 }, { "epoch": 2.80646059593428, "grad_norm": 0.051739538811377245, "learning_rate": 7.488368199612965e-05, "loss": 0.0019, "step": 5040 }, { "epoch": 2.8120300751879697, "grad_norm": 0.0545065555627121, "learning_rate": 7.476316437558006e-05, "loss": 0.0021, "step": 5050 }, { "epoch": 2.8175995544416597, "grad_norm": 0.051858173462886396, "learning_rate": 7.464245582543254e-05, "loss": 0.002, "step": 5060 }, { "epoch": 2.8231690336953497, "grad_norm": 0.05625922810310046, "learning_rate": 7.452155727637731e-05, "loss": 0.0019, "step": 5070 }, { "epoch": 2.828738512949039, "grad_norm": 0.06451749546980787, "learning_rate": 7.440046966056954e-05, "loss": 0.0019, "step": 5080 }, { "epoch": 2.8343079922027288, "grad_norm": 0.0592788295080246, "learning_rate": 7.42791939116221e-05, "loss": 0.0018, "step": 5090 }, { "epoch": 2.839877471456419, "grad_norm": 0.060714792093629934, "learning_rate": 7.415773096459846e-05, "loss": 0.0019, "step": 5100 }, { "epoch": 2.845446950710109, "grad_norm": 0.054065946398541405, "learning_rate": 7.40360817560054e-05, "loss": 0.0018, "step": 5110 }, { "epoch": 2.8510164299637983, "grad_norm": 0.06021220447231555, "learning_rate": 7.391424722378589e-05, "loss": 0.0019, "step": 5120 }, { "epoch": 2.856585909217488, "grad_norm": 0.05567546706138025, "learning_rate": 7.379222830731164e-05, "loss": 0.002, "step": 5130 }, { "epoch": 2.862155388471178, "grad_norm": 0.056617114000235835, "learning_rate": 7.367002594737614e-05, "loss": 0.002, "step": 5140 }, { "epoch": 2.867724867724868, "grad_norm": 0.05902717684116007, "learning_rate": 7.354764108618722e-05, "loss": 0.0019, "step": 5150 }, { "epoch": 2.8732943469785575, "grad_norm": 0.05468033114324629, "learning_rate": 7.342507466735984e-05, "loss": 0.002, "step": 5160 }, { "epoch": 2.878863826232247, "grad_norm": 0.05355173047966122, "learning_rate": 7.330232763590882e-05, "loss": 0.002, "step": 5170 }, { "epoch": 2.884433305485937, "grad_norm": 0.047606675945423234, "learning_rate": 7.317940093824151e-05, "loss": 0.0018, "step": 5180 }, { "epoch": 2.890002784739627, "grad_norm": 0.05710987607858959, "learning_rate": 7.305629552215055e-05, "loss": 0.002, "step": 5190 }, { "epoch": 2.8955722639933166, "grad_norm": 0.061919641274551386, "learning_rate": 7.293301233680656e-05, "loss": 0.0019, "step": 5200 }, { "epoch": 2.901141743247006, "grad_norm": 0.06034082866829357, "learning_rate": 7.280955233275078e-05, "loss": 0.0019, "step": 5210 }, { "epoch": 2.906711222500696, "grad_norm": 0.05055117819606763, "learning_rate": 7.268591646188776e-05, "loss": 0.0019, "step": 5220 }, { "epoch": 2.912280701754386, "grad_norm": 0.07536575724606918, "learning_rate": 7.256210567747801e-05, "loss": 0.0019, "step": 5230 }, { "epoch": 2.9178501810080757, "grad_norm": 0.05837113985627112, "learning_rate": 7.24381209341307e-05, "loss": 0.0019, "step": 5240 }, { "epoch": 2.9234196602617653, "grad_norm": 0.0834784993335112, "learning_rate": 7.231396318779623e-05, "loss": 0.0018, "step": 5250 }, { "epoch": 2.9289891395154553, "grad_norm": 0.05474102004179984, "learning_rate": 7.218963339575892e-05, "loss": 0.0019, "step": 5260 }, { "epoch": 2.9345586187691453, "grad_norm": 0.052680493771874885, "learning_rate": 7.206513251662957e-05, "loss": 0.0018, "step": 5270 }, { "epoch": 2.940128098022835, "grad_norm": 0.057180281513574904, "learning_rate": 7.194046151033813e-05, "loss": 0.0019, "step": 5280 }, { "epoch": 2.9456975772765244, "grad_norm": 0.05234359452879529, "learning_rate": 7.181562133812624e-05, "loss": 0.0018, "step": 5290 }, { "epoch": 2.9512670565302144, "grad_norm": 0.05463305761815249, "learning_rate": 7.169061296253987e-05, "loss": 0.0018, "step": 5300 }, { "epoch": 2.9568365357839044, "grad_norm": 0.06861187936883137, "learning_rate": 7.156543734742186e-05, "loss": 0.0019, "step": 5310 }, { "epoch": 2.962406015037594, "grad_norm": 0.048975954225027464, "learning_rate": 7.144009545790454e-05, "loss": 0.0017, "step": 5320 }, { "epoch": 2.9679754942912835, "grad_norm": 0.05348185471944446, "learning_rate": 7.13145882604022e-05, "loss": 0.0018, "step": 5330 }, { "epoch": 2.9735449735449735, "grad_norm": 0.060040519890833914, "learning_rate": 7.118891672260372e-05, "loss": 0.0017, "step": 5340 }, { "epoch": 2.9791144527986635, "grad_norm": 0.04694679144483152, "learning_rate": 7.10630818134651e-05, "loss": 0.0017, "step": 5350 }, { "epoch": 2.984683932052353, "grad_norm": 0.06715588212175277, "learning_rate": 7.093708450320191e-05, "loss": 0.0018, "step": 5360 }, { "epoch": 2.9902534113060426, "grad_norm": 0.05154669242107701, "learning_rate": 7.081092576328195e-05, "loss": 0.0018, "step": 5370 }, { "epoch": 2.9958228905597326, "grad_norm": 0.06543458650756455, "learning_rate": 7.068460656641761e-05, "loss": 0.0018, "step": 5380 }, { "epoch": 3.0011138958507377, "grad_norm": 0.05641584896255111, "learning_rate": 7.055812788655849e-05, "loss": 0.0015, "step": 5390 }, { "epoch": 3.0066833751044277, "grad_norm": 0.04151708786487322, "learning_rate": 7.043149069888379e-05, "loss": 0.0011, "step": 5400 } ], "logging_steps": 10, "max_steps": 12572, "num_input_tokens_seen": 0, "num_train_epochs": 7, "save_steps": 1800, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 448616760606720.0, "train_batch_size": 128, "trial_name": null, "trial_params": null }